[
  {
    "id": "official-t2va-starship",
    "title": "舰桥上的跃迁余震",
    "titleEn": "After the Fleet Jumps",
    "model": "MiniMax H3 Base FL2VA",
    "mode": "T2VA",
    "summary": "从星舰舰桥的中广景切至舰长面部特写，以原生立体声同步表现舰队跃迁、空间冲击与骤然安静的余波。",
    "summaryEn": "Moves from a medium-wide view of a starship bridge to a close-up of the captain, using native stereo audio to synchronize the fleet jump, spatial shock, and the suddenly quiet aftermath.",
    "prompt": "integrated_multimodal_description: [Shot 1] Cinematic, medium wide shot, pushing in slowly. In the cavernous, dimly lit bridge of a starship, sleek metallic consoles with glowing amber displays flank a massive, curved observation window. A female captain, in her late 40s with an athletic build and short silver-streaked black hair, stands in the center midground. She wears a structured, high-collared dark navy military tunic with silver chest insignias. Her back is to the camera, silhouetted against the cool, ambient starlight pouring through the thick glass. She stands perfectly still with her hands clasped tightly behind her back. Outside the window, a massive armada of jagged, dark grey dreadnoughts hovers in tight formation against a deep purple space nebula. The fleet's massive rear thrusters begin to glow with an intense, escalating bright blue light. [Shot 2] At 00:04.500, the camera cuts to a close-up of the captain's face and shakes strongly. The brilliant blue-white light from the fleet's gathering energy reflects vividly in her dark eyes. Suddenly, a blinding white flash floods through the window, completely washing out the background as the fleet jumps to hyperspace. The sheer spatial force violently jolts the bridge, causing the captain from Shot 1 to stagger slightly forward, her shoulders tensing as she visibly braces herself against the physical tremors. As the intense white light fades abruptly, leaving only the dim, empty expanse of the purple nebula reflected on her starkly lit skin, her jaw clenches, and she slowly closes her eyes in the newly emptied space.\noverall_soundscape: A low, resonant hum of the ship's ambient life support systems serves as the baseline, soon drowned out by an audible, escalating, high-pitched electronic whine as the fleet outside charges its hyperdrives. A massive, deafening, bass-heavy boom and sharp crackle erupts during the blinding flash, accompanied by the loud metallic creaking, rattling, and deep thuds of the bridge's bulkheads vibrating under immense physical stress. The intense roaring impact then cuts abruptly back to a hollow, echoing room tone, leaving only the faint, steady hum of the isolated bridge.\nnon_diegetic_music: Cinematic space-opera orchestral score, slow tempo, featuring a solitary, mournful French horn melody over deep, sustained string dissonances that build rapidly in volume and intensity, swelling to a massive orchestral peak before snapping immediately into silence right after the jump.",
    "sourceUrl": "https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/scripts/readme/reproducible-768p-t2va-request.sh",
    "sourceLabel": "MiniMax 官方复现脚本",
    "author": "MiniMaxAI",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-08-05T13:33:05+08:00",
    "mediaUrl": "/media/official-t2va-starship.mp4",
    "posterUrl": "/posters/t2va.webp",
    "duration": 10,
    "aspectRatio": "16:9",
    "resolution": "768p",
    "tags": [
      "电影感",
      "运镜",
      "立体声",
      "多镜头"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "official-verbatim",
    "sourceType": "official",
    "verified": true,
    "featured": {
      "at": "2026-09-05T15:50:15.694Z",
      "basis": "official"
    }
  },
  {
    "id": "official-fl2va-ramen",
    "title": "拉面与家庭晚餐",
    "titleEn": "Ramen Rack Focus",
    "model": "MiniMax H3 Base FL2VA",
    "mode": "FL2VA",
    "summary": "以拉面首帧为约束，通过缓慢焦点转移让前景蒸汽虚化、背景家庭成员逐渐清晰，考验景深和群像动作。",
    "summaryEn": "Uses a ramen image as the opening-frame constraint, then gradually racks focus so the foreground steam softens while the family in the background comes into view, testing depth of field and ensemble motion.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] This is a live-action, cinematic shot with a shallow depth of field. The camera holds a perfectly static shot throughout the entire eight-second duration, capturing a cozy family gathering in a traditional Japanese dining room. The scene opens with a large, intricately patterned blue and white ceramic bowl of ramen in the immediate foreground, rendered in crisp, sharp focus. The bowl sits on a smooth, polished long wooden table. Inside the bowl, a rich, oily golden-brown broth surrounds yellow wavy noodles, topped with two thick, round slices of chashu pork featuring visible fat marbling and a distinct spiral meat pattern. A generous mound of freshly chopped, bright green scallions rests in the center, and a crisp, dark green rectangular sheet of nori seaweed is tucked into the right edge. To the left of the bowl, a pair of light brown wooden chopsticks rests horizontally on a small, dark rectangular chopstick rest, near a small cylindrical ceramic teacup with blue painted patterns. On the right side of the table, a spherical paper lantern with a ribbed bamboo frame sits on a black wooden base. In the background, a large family of seven is gathered around the table, initially appearing as a soft, blurred presence. Behind them, traditional Japanese sliding shoji screens with wooden lattice frames are open, revealing a bright outdoor scene with lush green trees. Early in the clip, the thick, white steam rising from the hot ramen broth immediately intensifies, billowing upwards in thick, swirling clouds that dance continuously above the bowl. As the clip progresses into the middle seconds, the camera maintains its static position while the focus begins a deliberate, smooth shift deeper into the room. The foreground ramen bowl, its vibrant ingredients, and the rising steam gradually soften into a hazy, out-of-focus blur. Simultaneously, the family members in the background come into sharp, detailed clarity. The heavy steam continues to rise from the foreground, creating a dynamic, translucent veil between the camera and the family. With the focus now firmly locked on the background, the vibrant family dinner comes alive. The man in the dark navy blue long-sleeved shirt on the left leans forward, his mouth moving animatedly in a silent exchange. The young girl in the crisp white short-sleeved t-shirt beside him smiles brightly, looking toward the center of the table. The woman on the far left, wearing a soft light blue long-sleeved blouse, turns her head slightly, smiling gently. Across the table, the woman in the light grey button-down shirt smiles broadly, her eyes crinkling, as she rests her hands near her plate. The woman in the dark grey top further back uses her wooden chopsticks to pick up a small piece of food from a central ceramic dish filled with bright red pickled vegetables. The woman in the center back in the light grey sweater smiles gently, her hands clasped softly in front of her, observing the interaction. Throughout the remainder of the clip, the family continues their lively physical interaction, their mouths moving in continuous, silent cadences of conversation, while the thick, white steam from the blurred ramen bowl in the foreground never stops rising, adding a comforting atmosphere to the warm gathering.\n\noverall_soundscape: The soundscape begins with a quiet room tone mixed with the faint, airy rustle of the thick steam billowing from the hot ramen bowl in the foreground, accompanied by the subtle, continuous hissing and bubbling of the rich broth. As the visual focus shifts deeper into the room, the physical sounds of the bustling family dinner become dominant in the foreground. The clear, sharp clinking of ceramic bowls and wooden chopsticks touching plates is clearly heard as the family members reach for food. This is followed by the faint, muffled thud of a cup being set down on the smooth wooden table, and the subtle, rhythmic rustle of cotton and wool clothing as the family members lean forward and gesture, perfectly capturing the lively, physical atmosphere of the shared meal.\n\nnon_diegetic_music: A gentle, heartwarming acoustic guitar melody plays softly in the background, accompanied by the subtle, resonant notes of a traditional Japanese koto. The music maintains a slow, comforting tempo that enhances the cozy, nostalgic, and joyful atmosphere of the family gathering.",
    "sourceUrl": "https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/scripts/readme/reproducible-768p-fl2va-request.sh",
    "sourceLabel": "MiniMax 官方复现脚本",
    "author": "MiniMaxAI",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-08-05T13:33:05+08:00",
    "mediaUrl": "/media/official-fl2va-ramen.mp4",
    "posterUrl": "/posters/fl2va.webp",
    "duration": 8,
    "aspectRatio": "Auto",
    "resolution": "768p",
    "tags": [
      "首帧",
      "焦点转移",
      "群像",
      "环境音"
    ],
    "category": "advertising",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "official-verbatim",
    "sourceType": "official",
    "verified": true,
    "featured": {
      "at": "2026-09-05T15:50:15.694Z",
      "basis": "official"
    }
  },
  {
    "id": "official-ref2va-lamb",
    "title": "粉色西装与黑色羔羊",
    "titleEn": "Follow the Wind",
    "model": "MiniMax H3 Base Ref2VA",
    "mode": "Ref2VA",
    "summary": "同时引用视频、原背景音乐和额外人声音色，让角色新增英语对白并保持身份、服装、镜头与环境稳定。",
    "summaryEn": "References a source video, its original background music, and an additional voice timbre to add English dialogue while preserving character identity, wardrobe, framing, and setting.",
    "prompt": "subject_definitions:\n<Subject 1> is the young man with short wavy blonde hair, wearing a bright pink suit jacket, matching pink trousers, an unbuttoned white shirt, and silver rings, holding a small black lamb in his arms in <Video 1>.\n<Video 1> is the source video for the editing task.\n<Audio 1> is the synchronized audio track of <Video 1>, providing the background music.\n<Audio 2> is the voice timbre reference for <Subject 1>'s voice, containing a spoken male voiceover.\n\nsummary:\n[video editing + audio reference + audio reuse] The target video is an edited version of <Video 1>. <Subject 1>, wearing a bright pink suit and holding a black lamb, stands in a grassy field with other white lambs in the background. The edit animates <Subject 1>'s face to speak the user-provided dialogue. <Audio 1> is partially reused as the continuous background music, while the target references the calm male voice timbre of <Audio 2> for <Subject 1>'s spoken lines.\n\nretention_analysis:\n<Subject 1> (appears in [Shot 1]): fully_preserved - the man retains his identity, wavy blonde hair, pink suit, white shirt, accessories, and the black lamb he holds, with his mouth newly animated to speak.\n<Video 1> (source video editing): fully_preserved - the original camera framing, warm golden hour lighting, grassy hill setting, and background white lambs are maintained while the central character is edited.\n<Audio 1>: partially_copy - the atmospheric background music from <Audio 1> is reused in the target video, mixed beneath the newly added spoken dialogue.\n<Audio 2>: reference - the target audio references the male voice timbre from <Audio 2> to generate <Subject 1>'s spoken dialogue.\n\ndetailed_description:\nThe target video is in realistic photographic style.\n[Shot 1] The shot begins from the source <Video 1>, showing <Subject 1>, a young man with short wavy blonde hair, wearing a bright pink suit jacket, matching pink trousers, and a casually unbuttoned white shirt. He stands confidently in a sunlit green pasture, gently holding a small black lamb securely in his arms. The warm, golden hour lighting casts soft shadows across his face and the bright pink fabric of his suit. Behind him, several white lambs stand and graze on the rolling grassy hill against a clear, pale blue sky. The atmospheric background music from <Audio 1> plays continuously throughout the scene. <Subject 1> physically speaks, his mouth movements naturally syncing to the new dialogue, with his voice timbre referencing the calm male delivery from <Audio 2>. Looking thoughtfully forward, <Subject 1> (S1) speaks softly, <d>[English] Follow the wind, live free.</d> As he delivers the line, he subtly shifts his weight, cradling the resting black lamb while the camera slowly pushes in. <Subject 1> (S1) continues his thought, <d>[English] Leave worries behind, enjoy the moment.</d> Exactly as his voice stops, his lips meet in a relaxed, peaceful smile, and his jaw ceases speaking motion. He then turns his gaze slightly away toward the horizon, gently stroking the black lamb's fleece with his fingers as the camera holds on this tranquil, sunlit state through the end of the video.\n\noverall_soundscape:\nThe soundscape consists of the continuous, atmospheric background music from <Audio 1>, overlaid with the clear, calm male dialogue spoken by the main character, referencing the voice timbre of <Audio 2>.\n\nnon_diegetic_music:\nThe atmospheric, sustained background music from <Audio 1> is reused as the continuous score, playing quietly beneath the spoken dialogue.",
    "sourceUrl": "https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/scripts/readme/reproducible-768p-ref2va-request.sh",
    "sourceLabel": "MiniMax 官方复现脚本",
    "author": "MiniMaxAI",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-08-05T13:33:05+08:00",
    "mediaUrl": "/media/official-ref2va-lamb.mp4",
    "posterUrl": "/posters/ref2va.webp",
    "duration": 5,
    "aspectRatio": "Auto",
    "resolution": "768p",
    "tags": [
      "视频编辑",
      "声音参考",
      "口型",
      "角色一致性"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "fashion"
    ],
    "inputTypes": [
      "text",
      "video",
      "audio"
    ],
    "promptProvenance": "official-verbatim",
    "sourceType": "official",
    "verified": true,
    "featured": {
      "at": "2026-09-05T15:50:15.694Z",
      "basis": "official"
    }
  },
  {
    "id": "x-icreat-time-freeze-diner",
    "title": "餐厅时间冻结与逆向复原",
    "titleEn": "The Diner That Rewinds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "1950 年代餐厅里，碰撞飞散的咖啡、餐盘与食物在空中冻结，镜头环绕后让整个世界逆向复原；原帖公开了分段运镜与动作提示词，但展示视频为 15 秒、提示词时间轴为 30 秒。",
    "summaryEn": "Inside a 1950s diner, coffee, dishes, and food freeze in mid-air after a collision. The camera circles the scene before the entire world rewinds to its starting state. The original post includes a segmented camera-and-action prompt, although the displayed video lasts 15 seconds while the prompt timeline spans 30 seconds.",
    "prompt": "Photorealistic cinematic 1950s American diner, chrome stools, red vinyl, neon glow and checkerboard floor, shot with modern lived-in realism and soft natural window light. Subtle handheld texture, warm practicals, rich period detail, heavy film grain. 0-4s: [Medium Wide] A striking young woman in her early 20s sits alone at the counter, calm and slightly amused, slowly sipping a tall thick milkshake through a straw. Behind her a young waitress in classic uniform approaches with a tray of eggs and bacon in one hand and a full glass coffee pot in the other. An older lady starts rising from a nearby booth. 4-8s: [Dynamic Tracking] The older lady collides hard into the waitress. Tray, plate, eggs, bacon and coffee pot explode upward in chaotic slow motion. Coffee erupts into long liquid ribbons and perfect suspended droplets. Camera immediately begins a smooth continuous orbit around the impact. Time locks completely at the peak of the spill. Every face freezes in pure shock. Only the girl at the counter keeps moving, completely unfazed. 8-17s: [Slow 360° Orbital] Camera glides in a full elegant orbit through the frozen diner. Coffee hangs in mid-air as glassy ribbons and spheres with perfect volume and surface tension. Bacon strips, eggs and the spinning tray float weightlessly. Patrons and waitress remain locked in startled expressions. The girl at the counter takes one slow, deliberate sip, eyes half-lidded, almost bored, while the entire frozen world (except her) begins an elegant reverse: every droplet, every piece of food and every person rewinds smoothly back to the exact starting positions. 17-24s: [Medium Shot] Rewind lands perfectly. Waitress stands balanced again with tray and coffee pot. The girl lifts her eyes, raises two fingers in a small casual gesture and softly calls the waitress by name. The waitress turns toward her just before the older lady begins to stand, completely avoiding the collision. A tiny private smile crosses the girl’s face. 24-30s: [Extreme Close-Up] Hard cut to her face as she takes one last slow sip. Soft knowing smile, eyes almost closed in quiet satisfaction, like she has done this a hundred times. Shallow depth of field, creamy bokeh of the neon diner behind her. Photorealistic, ultra-detailed fluid physics, perfect motion blur only on moving elements, stable characters, cinematic lighting, heavy natural film grain, no artifacts, movie-level temporal coherence, high rewatch value.",
    "sourceUrl": "https://x.com/icreat_ai/status/2085297962977227011",
    "sourceLabel": "X 原帖 · @icreat_ai",
    "author": "ICREAT AI",
    "publishedAt": "2026-08-06T09:32:31.993Z",
    "addedAt": "2026-08-06T18:39:15+08:00",
    "mediaUrl": "/media/x-icreat-time-freeze-diner.mp4",
    "posterUrl": "/posters/x/x-icreat-time-freeze-diner.jpg",
    "duration": 15,
    "aspectRatio": "原帖",
    "resolution": "原帖",
    "tags": [
      "时间冻结",
      "逆向运动",
      "液体物理",
      "完整提示词"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "food",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-yangtze-textbook-recreation",
    "title": "小学课本场景复刻",
    "titleEn": "Textbook Scene Reimagined",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "创作者使用图片确定视觉风格，再配合文字提示复刻小学课本场景，并以贝多芬《月光曲》从平静、汹涌到归于平静的情绪变化组织 15 秒视频。",
    "summaryEn": "The creator used an image to establish the visual style and a text prompt to recreate a primary-school textbook scene, structuring the 15-second video around the emotional arc of Beethoven’s Moonlight Sonata—from calm to turbulent and back to calm.",
    "prompt": null,
    "sourceUrl": "https://x.com/Yangtze_Seventh/status/2085318072903299536",
    "sourceLabel": "X 原帖 · @Yangtze_Seventh",
    "author": "七仔",
    "publishedAt": "2026-08-06T10:00:31.801Z",
    "addedAt": "2026-08-06T18:39:15+08:00",
    "mediaUrl": "/media/x-yangtze-textbook-recreation.mp4",
    "posterUrl": "/posters/x/x-yangtze-textbook-recreation.jpg",
    "duration": 15,
    "aspectRatio": "原帖",
    "resolution": "原帖",
    "tags": [
      "图片定风格",
      "课本复刻",
      "音乐叙事",
      "工作流"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-tapehead-composition-mv",
    "title": "分层构图音乐视频",
    "titleEn": "Composition — Layered Music Video",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一支由 Suno 音乐、Midjourney 视觉素材和 MiniMax H3 视频生成组合而成的长篇 MV。创作者强调 H3 对对象分层与提示词遵循的表现；具体生成模式和单镜头提示词未公开。",
    "summaryEn": "A long-form music video built from Suno music, Midjourney visuals, and MiniMax H3 video generation. The creator highlights H3’s object layering and prompt adherence, but does not disclose the exact generation mode or shot-level prompts.",
    "prompt": null,
    "sourceUrl": "https://x.com/tapehead_Lab/status/2085304883847258179",
    "sourceLabel": "X 原帖 · @tapehead_Lab",
    "author": "tapehead.lab",
    "publishedAt": "2026-08-06T10:00:02.057Z",
    "addedAt": "2026-08-06T18:39:15+08:00",
    "mediaUrl": "/media/x-tapehead-composition-mv.mp4",
    "posterUrl": "/posters/x/x-tapehead-composition-mv.jpg",
    "duration": 107,
    "aspectRatio": "原帖",
    "resolution": "原帖",
    "tags": [
      "音乐视频",
      "分层构图",
      "Midjourney",
      "Suno"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-pollo-commercial-comparison",
    "title": "同提示词广告模型对比",
    "titleEn": "One Prompt, Two Commercials",
    "model": "MiniMax H3 与 Seedance 2.0 对比",
    "mode": "T2VA",
    "summary": "在 Pollo AI 中使用相同提示词和设置，对比 MiniMax H3 与 Seedance 2.0 的商业广告输出。创作者认为 H3 更自然、精致，另一模型更强调电影化视觉冲击。",
    "summaryEn": "Uses the same prompt and settings in Pollo AI to compare commercial outputs from MiniMax H3 and Seedance 2.0. The creator describes H3 as more natural and polished, while the other model places greater emphasis on cinematic visual impact.",
    "prompt": null,
    "sourceUrl": "https://x.com/__spence_r/status/2085281251183747482",
    "sourceLabel": "X 原帖 · @__spence_r",
    "author": "George",
    "publishedAt": "2026-08-06T08:26:07.591Z",
    "addedAt": "2026-08-06T18:39:15+08:00",
    "mediaUrl": "/media/x-pollo-commercial-comparison.mp4",
    "posterUrl": "/posters/x/x-pollo-commercial-comparison.jpg",
    "duration": 30,
    "aspectRatio": "原帖",
    "resolution": "原帖",
    "tags": [
      "模型对比",
      "商业广告",
      "付费合作",
      "同提示词"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-mayii-ugc-commercial",
    "title": "跨平台 UGC 种草短片",
    "titleEn": "Localized UGC Product Film",
    "model": "MiniMax H3 + HyperFrames（创作者标注）",
    "mode": "Unknown",
    "summary": "面向不同国家与内容平台制作的 AI UGC 商品短片工作流，创作者明确说明商业化版本主要由 HyperFrames 与 MiniMax H3 配合完成；原帖未说明两者的具体职责。",
    "summaryEn": "An AI UGC product-video workflow tailored to different countries and content platforms. The creator states that the commercial version mainly combines HyperFrames with MiniMax H3, although the division of responsibilities between them remains undocumented.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayii0205/status/2085305063405220214",
    "sourceLabel": "X 原帖 · @Mayii0205",
    "author": "Tuola Ge",
    "publishedAt": "2026-08-06T10:00:44.867Z",
    "addedAt": "2026-08-06T18:39:15+08:00",
    "mediaUrl": "/media/x-mayii-ugc-commercial.mp4",
    "posterUrl": "/posters/x/x-mayii-ugc-commercial.jpg",
    "duration": 15,
    "aspectRatio": "原帖",
    "resolution": "原帖",
    "tags": [
      "UGC",
      "种草视频",
      "本地化",
      "混合工作流"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "ugc"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-soylab-person-replacement",
    "title": "人物替换实验",
    "titleEn": "Person Replacement Test",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "基于 ComfyUI 的人物替换短片实验。原帖明确标注 MiniMax H3，但没有公开提示词、参考输入或节点配置，并包含受版权保护角色的视觉引用。",
    "summaryEn": "A ComfyUI-based person-replacement video experiment. The original post explicitly identifies MiniMax H3 but provides no prompt, reference inputs, or node configuration, and includes a visual reference to a copyrighted character.",
    "prompt": null,
    "sourceUrl": "https://x.com/soy_lab/status/2085305126001225909",
    "sourceLabel": "X 原帖 · @soy_lab",
    "author": "Nobody",
    "publishedAt": "2026-08-06T10:00:59.791Z",
    "addedAt": "2026-08-06T18:39:15+08:00",
    "mediaUrl": "/media/x-soylab-person-replacement.mp4",
    "posterUrl": "/posters/x/x-soylab-person-replacement.jpg",
    "duration": 14,
    "aspectRatio": "原帖",
    "resolution": "原帖",
    "tags": [
      "人物替换",
      "ComfyUI",
      "视频编辑",
      "版权角色"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image",
      "video"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-yukyuk-h3-seedance-same-prompt",
    "title": "同提示词 H3 与 Seedance 对比",
    "titleEn": "Same Prompt: H3 vs Seedance 2",
    "model": "MiniMax H3 与 Seedance 2（创作者标注）",
    "mode": "Unknown",
    "summary": "YukYuk 使用相同提示词分别生成 MiniMax H3 与 Seedance 2 的 15 秒视频，用于观察两个模型的输出差异；原帖没有公开提示词正文。",
    "summaryEn": "YukYuk generated separate 15-second videos with MiniMax H3 and Seedance 2 using the same prompt to compare their outputs. The original post does not disclose the prompt text.",
    "prompt": null,
    "sourceUrl": "https://x.com/YukYukID/status/2085553970702074050",
    "sourceLabel": "X 原帖 · @YukYukID",
    "author": "YukYuk!",
    "publishedAt": "2026-08-07T02:29:48.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-yukyuk-h3-seedance-same-prompt.mp4",
    "posterUrl": "/posters/x/x-yukyuk-h3-seedance-same-prompt.jpg",
    "duration": 15,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "同提示词",
      "模型对比",
      "Seedance 2",
      "15 秒"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": null,
      "views": 8,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 15,
      "resolution": "not visible",
      "aspectRatio": "not visible"
    }
  },
  {
    "id": "x-melvin-comfyui-rtx3090-local",
    "title": "RTX 3090 本地 ComfyUI 生成",
    "titleEn": "Local ComfyUI Generation on RTX 3090",
    "model": "MiniMax H3 + ComfyUI（创作者标注）",
    "mode": "Unknown",
    "summary": "创作者在 RTX 3090 上通过 ComfyUI 本地生成 15 秒 H3 视频，公开输出为 864×480、耗时约 22 分钟，并肯定其提示词遵循能力。",
    "summaryEn": "The creator generated a 15-second H3 video locally in ComfyUI on an RTX 3090. The published output is 864×480, took approximately 22 minutes to render, and was presented as a positive test of prompt adherence.",
    "prompt": null,
    "sourceUrl": "https://x.com/melvindvivas/status/2085553403846078669",
    "sourceLabel": "X 原帖 · @melvindvivas",
    "author": "Melvin Vivas",
    "publishedAt": "2026-08-07T02:27:33.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-melvin-comfyui-rtx3090-local.mp4",
    "posterUrl": "/posters/x/x-melvin-comfyui-rtx3090-local.jpg",
    "duration": 15,
    "aspectRatio": "原帖未标注",
    "resolution": "864x480",
    "tags": [
      "本地生成",
      "ComfyUI",
      "RTX 3090",
      "性能测试"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": null,
      "views": 22,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 15,
      "resolution": "864x480",
      "aspectRatio": "not visible"
    }
  },
  {
    "id": "x-coretan-eating-ramen-turbo-lora",
    "title": "单句提示词：吃拉面",
    "titleEn": "One-Line Prompt: Eating Ramen",
    "model": "MiniMax H3 Turbo LoRA（创作者标注）",
    "mode": "T2VA",
    "summary": "本地 Turbo LoRA 实验只使用作者公开的单句提示词“Eating Ramen”，测试人物吃拉面时手部、筷子与动作表现。",
    "summaryEn": "This local Turbo LoRA experiment uses only the creator’s published one-line prompt, “Eating Ramen,” to test the character’s hands, chopsticks, and eating motion.",
    "prompt": "Eating Ramen",
    "sourceUrl": "https://x.com/core_tan/status/2085551748874449032",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-07T02:20:59.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-coretan-eating-ramen-turbo-lora.mp4",
    "posterUrl": "/posters/x/x-coretan-eating-ramen-turbo-lora.jpg",
    "duration": 10,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "作者提示词",
      "吃拉面",
      "Turbo LoRA",
      "本地生成"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 0,
      "bookmarks": null,
      "views": 84,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 10,
      "resolution": "not visible",
      "aspectRatio": "not visible"
    }
  },
  {
    "id": "x-rsensui-local-audio-claude-code",
    "title": "Mac Studio 本地音视频生成",
    "titleEn": "Local Audio-Video Generation on Mac Studio",
    "model": "MiniMax H3 + Claude Code（创作者标注）",
    "mode": "Unknown",
    "summary": "创作者称 37 秒视频及声音均由 MiniMax H3 本地生成，并通过 Claude Code 一次完成；公开硬件为 Mac Studio M3 Ultra 256GB。",
    "summaryEn": "The creator states that both the 37-second video and its audio were generated locally with MiniMax H3 in a single Claude Code run. The disclosed hardware is a Mac Studio with an M3 Ultra and 256 GB of memory.",
    "prompt": null,
    "sourceUrl": "https://x.com/rsensui/status/2085551007631880347",
    "sourceLabel": "X 原帖 · @rsensui",
    "author": "泉水亮介",
    "publishedAt": "2026-08-07T02:18:02.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-rsensui-local-audio-claude-code.mp4",
    "posterUrl": "/posters/x/x-rsensui-local-audio-claude-code.jpg",
    "duration": 37,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "本地生成",
      "原生音频",
      "Claude Code",
      "Mac Studio"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": null,
      "views": 58,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 37,
      "resolution": "not visible",
      "aspectRatio": "not visible",
      "audioClaim": "作者称音频与视频均由 MiniMax H3 本地生成"
    }
  },
  {
    "id": "x-sada-hand-painted-blizzard",
    "title": "手绘暴风雪求生",
    "titleEn": "Hand-Painted Blizzard Survival",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "作者公开完整提示词，以纸上油画逐帧动画、12 fps 停格节奏、冰冷月光和暴风雪环境描绘一对母子在极端天气中求生。",
    "summaryEn": "The creator publishes the full prompt for a frame-by-frame oil-on-paper animation with a 12 fps stop-motion cadence, cold moonlight, and a violent blizzard, depicting a mother and child struggling to survive extreme weather.",
    "prompt": "Visual Medium & Craft: Traditional frame-by-frame hand-painted oil-on-paper animation, physical 12 fps stop-motion cadence with deliberate frame holds and natural line jitter between frames. Raw brushstroke textures, strictly zero CGI, no 3D rendering, no digital interpolation, no game engine aesthetics. Cinematography & Lighting: Masterful cinematography in the style of Emmanuel Lubezki and Roger Deakins, shot with physical cine lenses and a classic 180° shutter blur. Lit exclusively by a single overhead light source: pale, freezing moonlight cutting through dense atmospheric haze. Strong contre-jour backlighting with the camera positioned on the shadow side. Uniform cold blue moonlight washes over skin, clothing, snow, and tree trunks, casting sharp cold rim lights on shoulders and hoods while keeping faces in soft shadow. Atmosphere & Environment: A violent, blinding blizzard with near-zero visibility (3 meters maximum), dissolving the background into a solid white fog wall. Gale-force horizontal winds whip thick snow and freezing fog through the frame. Deep snow quickly accumulates on garments. Exhaled breath instantly vaporizes and tears away in the wind. Characters & Emotion: A desperate mother and her frightened young child battling for survival, lost in the storm at the absolute limit of their endurance. The mother conceals her panic to protect her child, while the child cries openly in fear. Both lean heavily into the gale, squinting against the biting wind, shielding their faces with forearms and ducking into their collars. Technical Details: Color palette dominated by 60% snow-blue, 30% deep trunk-black, and 10% cold highlight accents. Strict physical realism with grounded weight, true inertia, and accurate contact shadows on snow. Dynamic compositional balance using the rule of thirds and golden ratio. Uncompromising continuity across all character designs, garments, and environmental assets. 8K IMAX aesthetic.",
    "sourceUrl": "https://x.com/sada_ai/status/2085547332880503134",
    "sourceLabel": "X 原帖 · @sada_ai",
    "author": "sada",
    "publishedAt": "2026-08-07T02:03:26.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-sada-hand-painted-blizzard.mp4",
    "posterUrl": "/posters/x/x-sada-hand-painted-blizzard.jpg",
    "duration": 15,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "完整提示词",
      "手绘动画",
      "暴风雪",
      "电影灯光"
    ],
    "category": "cinematic",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 49,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 15,
      "resolution": "not visible",
      "aspectRatio": "not visible"
    }
  },
  {
    "id": "x-tmaiaroto-krea-image-to-h3",
    "title": "Krea 2 图片驱动 H3",
    "titleEn": "Krea 2 Image to MiniMax H3",
    "model": "Krea 2 + MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "使用 Krea 2 生成的图片作为 H3 视频输入，形成一条适合夜间批量渲染的图片生视频工作流；原帖没有公开提示词。",
    "summaryEn": "Uses an image generated with Krea 2 as the input for H3, creating an image-to-video workflow suitable for overnight batch rendering. The original post does not disclose the prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/tmaiaroto/status/2085543978611736973",
    "sourceLabel": "X 原帖 · @tmaiaroto",
    "author": "Tom Maiaroto",
    "publishedAt": "2026-08-07T01:50:06.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-tmaiaroto-krea-image-to-h3.mp4",
    "posterUrl": "/posters/x/x-tmaiaroto-krea-image-to-h3.jpg",
    "duration": 5,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "图片生视频",
      "Krea 2",
      "工作流",
      "本地渲染"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "bookmarks": null,
      "views": 85,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 5,
      "resolution": "not visible",
      "aspectRatio": "not visible"
    }
  },
  {
    "id": "x-opener-deep-sea-ui-comparison",
    "title": "深海生物百科交互界面",
    "titleEn": "Interactive Deep-Sea Bestiary UI",
    "model": "MiniMax H3 与 Seedance 2.5 对比",
    "mode": "FL2VA",
    "summary": "作者用同一份完整提示词比较 H3 与 Seedance 2.5：以首帧固定深海百科 UI，只允许光标、主标题和左侧生物变化，并精确编排音效与负面约束。",
    "summaryEn": "Uses the same full prompt to compare H3 with Seedance 2.5: a first-frame-locked deep-sea encyclopedia interface in which only the cursor, main title, and creature in the left panel may change, with precisely timed audio and extensive negative constraints.",
    "prompt": "[FORMAT]\nCreate a 15-second, 16:9 interactive deep-sea creature encyclopedia video. One continuous locked-screen shot.\n\n[ASSETS]\nUse Image 1 as the exact opening frame and strict visual reference. Treat the entire interface and the complete right-side creature list as an immutable frozen background plate. All thumbnails, creatures, names, card positions, colors, and details on the right must remain pixel-stable and identical to Image 1 in every frame.\n\n[IDENTITY]\nPreserve exactly: the ‘CELESTIAL BESTIARY ARCHIVE’ header, pearl-white and ocean-blue palette, silver ornamental borders, large left display panel, complete right-side card grid, bottom navigation bar, navy typography, existing creature names, translucent crystalline materials, and original spacing. Each main creature must match its original card design without hybridization.\n\n[BEATS]\n[0–2.5 seconds] One white cursor touches LUMINARAY’s wing inside the large left panel. LUMINARAY blinks, chirps, and a constellation glow travels across its wings. Nothing on the right changes.\n[2.5–5 seconds] The cursor clicks the static AURORUNA card. Do not animate or highlight the card. Only the creature inside the left panel changes to AURORUNA. The main title crossfades in place to ‘AURORUNA.’ The cursor circles its bell; AURORUNA follows it and releases a soft stardust pulse.\n[5–7.5 seconds] The cursor clicks the static CORALUMI card. The card and every other right-side thumbnail remain completely unchanged. Only the left-panel creature and main title update to CORALUMI. Three taps on its shell create three small localized lights and delicate chimes.\n[7.5–10.5 seconds] The cursor clicks the static GLOWFIN card. Only the left-panel creature and main title update to GLOWFIN. Two pokes on its lure make it squeak, track the cursor with its eyes, and intensify its bioluminescent glow.\n[10.5–15 seconds] A third poke triggers GLOWFIN’s adult transformation, strictly contained inside the left panel. Preserve its navy scales, round eyes, sharp teeth, fins, and forehead lure. The adult GLOWFIN growls, lunges forward, snaps at the cursor, then returns to idle. The title remains ‘GLOWFIN.’ The right-side list remains identical to the opening frame.\n\n[CAMERA]\nLocked front-facing screen view. No pan, zoom, tilt, shake, crop, perspective shift, parallax, or depth-of-field change. Use exactly one cursor.\n\n[LIGHT]\nMaintain the original pearl-white UI illumination, cool underwater blues, silver reflections, and luminous fantasy glow. Bioluminescent effects must remain clipped inside the large left display panel.\n\n[EDIT]\nNo cuts. Only three elements may change: the single cursor, the main title, and the creature inside the large left panel. Freeze every other UI pixel. The entire right-side card grid is a protected static 2D plate, not a group of animated objects. Main titles crossfade in place over 0.2 seconds, remain still until the next click, and exit only through opacity. No sliding, scaling, bouncing, re-lettering, or layout movement.\n\n[AUDIO]\nAudio: quiet deep-sea ambience, soft UI clicks, crystalline creature chirps, stardust shimmer, three shell chimes, two angler squeaks, a low maturation rumble, one aquatic growl, and a sharp bite snap. BGM: an original 15-second cue, 65% mystical underwater ambient and 35% playful digital fantasy. Glass harmonics, submerged synth pads, and soft sonar pulses. Build gently after 10.5 seconds and end with one humorous muted thump. Do not imitate an existing melody.\n\n[NEGATIVE]\nNever alter, animate, replace, redraw, relabel, reorder, highlight, morph, or regenerate any creature thumbnail or card on the right side. No card creature may blink, move, transform, disappear, swap places, or become the currently selected creature. Do not change any right-side card name, illustration, border, color, position, scale, spacing, or selection state. Do not propagate the left-panel transformation into the card grid. No interface distortion, panel movement, duplicate cursors, extra creatures, extra buttons, new text, subtitles, watermarks, logos, or stickers. Do not introduce Chinese text, garbled characters, misspellings, or incorrect creature names. No hybrid anatomy, uncontrolled morphing, full-screen particles, or effects escaping the left panel. Never a slideshow.",
    "sourceUrl": "https://x.com/opener_ai/status/2085542413729513534",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-07T01:43:53.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-opener-deep-sea-ui-comparison.mp4",
    "posterUrl": "/posters/x/x-opener-deep-sea-ui-comparison.jpg",
    "duration": 15,
    "aspectRatio": "16:9",
    "resolution": "原帖未标注",
    "tags": [
      "完整提示词",
      "首帧参考",
      "交互 UI",
      "模型对比"
    ],
    "category": "comparison",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 86,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 15,
      "resolution": "not visible",
      "aspectRatio": "16:9",
      "comparison": "MiniMax H3 versus Seedance 2.5"
    }
  },
  {
    "id": "x-megurosumi-minimax-h3-music-video",
    "title": "MiniMax H3 长篇音乐视频",
    "titleEn": "Long-Form MiniMax H3 Music Video",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一支时长约 2 分 17 秒的 MiniMax H3 音乐视频实验；原帖明确模型来源，但没有公开提示词或输入素材。",
    "summaryEn": "A MiniMax H3 music-video experiment lasting approximately 2 minutes and 17 seconds. The original post identifies the model but does not disclose the prompt or input assets.",
    "prompt": null,
    "sourceUrl": "https://x.com/megurosumi/status/2085535070102884729",
    "sourceLabel": "X 原帖 · @megurosumi",
    "author": "Jinny Lee",
    "publishedAt": "2026-08-07T01:14:42.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-megurosumi-minimax-h3-music-video.mp4",
    "posterUrl": "/posters/x/x-megurosumi-minimax-h3-music-video.jpg",
    "duration": 137,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "音乐视频",
      "长篇视频",
      "AI Art",
      "社区实验"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": null,
      "views": 58,
      "capturedAt": "2026-08-07T02:36:43Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "durationSeconds": 137,
      "resolution": "not visible",
      "aspectRatio": "not visible"
    }
  },
  {
    "id": "x-o685463-comfyui-12step-tests",
    "title": "本地 12 步流体与毛发测试",
    "titleEn": "Local 12-Step Fluid and Fur Tests",
    "model": "MiniMax H3 + ComfyUI（创作者标注）",
    "mode": "Unknown",
    "summary": "三组各 5 秒的本地 ComfyUI 测试，以速度优先的 12 步设置观察 H3 的流体与毛发表现；作者指出低步数下这些细节仍较吃力。",
    "summaryEn": "Three local ComfyUI tests, each lasting five seconds, use speed-first 12-step settings to examine H3’s handling of fluids and fur. The creator notes that both remain challenging at low step counts.",
    "prompt": null,
    "sourceUrl": "https://x.com/O685463/status/2086345324164161598",
    "sourceLabel": "X 原帖 · @O685463",
    "author": "Jun Ozaki",
    "publishedAt": "2026-08-09T06:54:22.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-o685463-comfyui-12step-tests.mp4",
    "posterUrl": "/posters/x/x-o685463-comfyui-12step-tests.jpg",
    "duration": 5,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "本地生成",
      "12 Steps",
      "流体测试",
      "毛发测试"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 1,
      "views": 15,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 3,
      "durationSeconds": 5,
      "resolution": "not visible",
      "aspectRatio": "not visible",
      "generationSettings": "12 steps; local ComfyUI"
    }
  },
  {
    "id": "x-sepisheim-ref2va-character-sheet",
    "title": "Ref2VA 角色表一致性测试",
    "titleEn": "Ref2VA Character Sheet Consistency Test",
    "model": "MiniMax H3 Ref2VA（创作者标注）",
    "mode": "Ref2VA",
    "summary": "使用 ChatGPT 生成角色表作为参考输入，测试 H3 Ref2VA 的角色一致性；公开参数包括 RTX 4070、576×832、24 fps、5 秒和约 225 秒生成时间。",
    "summaryEn": "Uses a character sheet generated with ChatGPT as the reference input for an H3 Ref2VA consistency test. Published parameters include an RTX 4070, 576×832 resolution, 24 fps, a five-second duration, and an approximately 225-second generation time.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2086344253496766502",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto",
    "publishedAt": "2026-08-09T06:50:07.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-sepisheim-ref2va-character-sheet.mp4",
    "posterUrl": "/posters/x/x-sepisheim-ref2va-character-sheet.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "576x832",
    "tags": [
      "Ref2VA",
      "角色表",
      "角色一致性",
      "RTX 4070"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 1,
      "views": 22,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5,
      "resolution": "576x832",
      "aspectRatio": "portrait",
      "fps": 24,
      "generationSettings": "RTX 4070; SageAttention; Spectrum; FirstBlockCache; 225 seconds"
    }
  },
  {
    "id": "x-aion-flight-2k-astra-4k",
    "title": "2K 飞行镜头至 4K 放大",
    "titleEn": "2K Flight Scene Upscaled to 4K",
    "model": "MiniMax H3 + Astra（创作者标注）",
    "mode": "Unknown",
    "summary": "创作者先用 MiniMax H3 生成 2K 飞行场景，再使用 Astra 放大至 4K，展示生成与后期超分辨率串联的工作流。",
    "summaryEn": "The creator first generates a 2K flight scene with MiniMax H3 and then upscales it to 4K with Astra, demonstrating a workflow that connects generation with post-production super-resolution.",
    "prompt": null,
    "sourceUrl": "https://x.com/AION_2077/status/2086341354469441828",
    "sourceLabel": "X 原帖 · @AION_2077",
    "author": "AION",
    "publishedAt": "2026-08-09T06:38:35.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-aion-flight-2k-astra-4k.mp4",
    "posterUrl": "/posters/x/x-aion-flight-2k-astra-4k.jpg",
    "duration": 15,
    "aspectRatio": "原帖未标注",
    "resolution": "2K H3 output, upscaled to 4K",
    "tags": [
      "飞行场景",
      "2K",
      "4K 放大",
      "Astra"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "bookmarks": null,
      "views": 52,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15,
      "resolution": "2K H3 output, upscaled to 4K",
      "aspectRatio": "not visible",
      "postProcessing": "Astra upscale"
    }
  },
  {
    "id": "x-toyxyz-i2v-turbo-lora",
    "title": "I2V Turbo LoRA 测试",
    "titleEn": "I2V Turbo LoRA Test",
    "model": "MiniMax H3 Turbo LoRA + ComfyUI（创作者标注）",
    "mode": "FL2VA",
    "summary": "一条 10 秒的 MiniMax H3 图片生视频 Turbo LoRA 测试，明确使用 ComfyUI；原帖未公开参考图细节或提示词。",
    "summaryEn": "A 10-second MiniMax H3 image-to-video Turbo LoRA test created in ComfyUI. The original post does not disclose the reference-image details or prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/toyxyz3/status/2086339956755329363",
    "sourceLabel": "X 原帖 · @toyxyz3",
    "author": "toyxyz",
    "publishedAt": "2026-08-09T06:33:02.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-toyxyz-i2v-turbo-lora.mp4",
    "posterUrl": "/posters/x/x-toyxyz-i2v-turbo-lora.jpg",
    "duration": 10,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "I2V",
      "Turbo LoRA",
      "ComfyUI",
      "10 秒"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "bookmarks": 3,
      "views": 449,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10,
      "resolution": "not visible",
      "aspectRatio": "not visible",
      "generationSettings": "I2V Turbo LoRA; ComfyUI"
    }
  },
  {
    "id": "x-cia0-omni-nine-reference-film",
    "title": "九图参考 AI 电影预告",
    "titleEn": "Nine-Reference AI Film Trailer",
    "model": "MiniMax H3 Omni（创作者标注）",
    "mode": "Ref2VA",
    "summary": "H3 Omni 在一支 15 秒 AI 电影预告中同时处理九张参考图，并测试细节文字渲染。作者只透露提示词约 3,943 个中文字符，没有公开正文。",
    "summaryEn": "H3 Omni processes nine reference images simultaneously in a 15-second AI film trailer while also testing fine-detail text rendering. The creator only discloses that the prompt contains approximately 3,943 Chinese characters, without publishing its contents.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cia0_exe/status/2086339725875699927",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-08-09T06:32:07.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-cia0-omni-nine-reference-film.mp4",
    "posterUrl": "/posters/x/x-cia0-omni-nine-reference-film.jpg",
    "duration": 15,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "H3 Omni",
      "九图参考",
      "电影预告",
      "文字渲染"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 5,
      "bookmarks": 2,
      "views": 599,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15,
      "resolution": "not visible",
      "aspectRatio": "not visible",
      "referenceImageCount": 9
    }
  },
  {
    "id": "x-nsitnov-white-camellia-local-film",
    "title": "白山茶：六镜头本地短片",
    "titleEn": "The White Camellia: Six-Shot Local Film",
    "model": "MiniMax H3 + Qwen Image 3 Pro + ElevenLabs",
    "mode": "Unknown",
    "summary": "《The White Camellia》由六段 15 秒 H3 镜头组成，并结合 Qwen Image 3 Pro 与 ElevenLabs，在 RTX Pro 6000 96GB 上完成本地制作。",
    "summaryEn": "The White Camellia comprises six 15-second H3 shots and combines Qwen Image 3 Pro with ElevenLabs. It was produced locally on an RTX Pro 6000 with 96 GB of memory.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2086338366946123992",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-09T06:26:43.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-nsitnov-white-camellia-local-film.mp4",
    "posterUrl": "/posters/x/x-nsitnov-white-camellia-local-film.jpg",
    "duration": 80,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "叙事短片",
      "六镜头",
      "本地工作流",
      "ElevenLabs"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 1,
      "views": 48,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 80,
      "claimedProjectDurationSeconds": 90,
      "resolution": "not visible",
      "aspectRatio": "not visible",
      "workflow": "6 × 15-second H3 clips; Qwen Image 3 Pro; ElevenLabs; local RTX Pro 6000 96GB"
    }
  },
  {
    "id": "x-aiconia-i2v-larry-turbo-h3-sage",
    "title": "Larry Turbo + H3 Sage 加速测试",
    "titleEn": "Larry Turbo and H3 Sage Acceleration Test",
    "model": "MiniMax H3 + Larry Turbo + H3 Sage",
    "mode": "FL2VA",
    "summary": "RTX 4070 上的 1MP、10 秒 I2V 加速测试，生成耗时 11 分 12 秒，并追加放大和插帧；作者同时公开了脸部块状噪点与语音较弱等缺陷。",
    "summaryEn": "A 1 MP, 10-second I2V acceleration test on an RTX 4070 took 11 minutes and 12 seconds, followed by upscaling and frame interpolation. The creator also documents blocky facial artifacts and weak speech generation.",
    "prompt": null,
    "sourceUrl": "https://x.com/AiPhotorealGirl/status/2086336598539747535",
    "sourceLabel": "X 原帖 · @AiPhotorealGirl",
    "author": "AIconia",
    "publishedAt": "2026-08-09T06:19:42.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-aiconia-i2v-larry-turbo-h3-sage.mp4",
    "posterUrl": "/posters/x/x-aiconia-i2v-larry-turbo-h3-sage.jpg",
    "duration": 10,
    "aspectRatio": "原帖未标注",
    "resolution": "1MP before upscale",
    "tags": [
      "I2V",
      "Larry Turbo",
      "H3 Sage",
      "性能与缺陷"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 11,
      "bookmarks": 1,
      "views": 520,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10,
      "resolution": "1MP before upscale",
      "aspectRatio": "not visible",
      "generationSettings": "RTX 4070; Larry Turbo + H3 Sage; 11m12s; upscale and frame interpolation"
    }
  },
  {
    "id": "x-maki-motion-context-five-scene-mv",
    "title": "五场景雨中东京 MV",
    "titleEn": "Five-Scene Rainy Tokyo Music Video",
    "model": "MiniMax H3 + HyperFrames（创作者标注）",
    "mode": "Unknown",
    "summary": "通过 Motion Context 连锁生成五个场景，把 H3 生成的语音作为音乐，再用 HyperFrames 叠加随节拍运动的歌词，完成约 29 秒的雨中东京 MV。",
    "summaryEn": "Motion Context chains five scenes together, uses H3-generated vocals as the music, and adds beat-synchronized lyrics with HyperFrames to create an approximately 29-second music video set in rainy Tokyo.",
    "prompt": null,
    "sourceUrl": "https://x.com/hAru_mAki_ch/status/2086334052639142176",
    "sourceLabel": "X 原帖 · @hAru_mAki_ch",
    "author": "Maki",
    "publishedAt": "2026-08-09T06:09:35.000Z",
    "addedAt": "2026-08-09T15:11:03+08:00",
    "mediaUrl": "/media/x-maki-motion-context-five-scene-mv.mp4",
    "posterUrl": "/posters/x/x-maki-motion-context-five-scene-mv.jpg",
    "duration": 29,
    "aspectRatio": "原帖未标注",
    "resolution": "原帖未标注",
    "tags": [
      "Motion Context",
      "五场景",
      "音乐视频",
      "HyperFrames"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "city",
      "music-performance"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 3,
      "views": 504,
      "capturedAt": "2026-08-09T07:00:52Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 29,
      "resolution": "not visible",
      "aspectRatio": "not visible",
      "workflow": "H3 Motion Context across 5 scenes; H3-generated voice; HyperFrames synced lyrics"
    }
  },
  {
    "id": "x-aion-cloud-head-subway",
    "title": "地铁乘客云团变形",
    "titleEn": "Cloud-Head Subway Transformation",
    "model": "MiniMax H3 · Local",
    "mode": "Ref2VA",
    "summary": "固定机位拍摄对称地铁车厢，乘客头部在真人与不同天气云团之间进行数字故障切换，并在第 8 秒收束到参考图构图；原帖同时公开了完整画面与声音 Prompt。",
    "summaryEn": "A locked-off, symmetrical subway shot glitches the passengers' heads between human faces and distinct weather clouds, then converges on the reference composition at eight seconds. The original post also publishes the complete visual and sound prompt.",
    "prompt": "integrated_multimodal_description: [Shot 1] Surreal live-action cinematic imagery in a symmetrical square deep-focus wide shot inside the same modern subway carriage established by <Picture 1>. At the beginning, the same seated passengers occupy the exact same seats with the same bodies, clothing, poses, hands, phones, bags, spacing, and calm commuter behavior, but every passenger has a distinct, anatomically normal human head and natural human face; the ceiling above them is ordinary dim stormy condensation without the giant eye. The camera holds a completely static shot for the entire video. From 00:01.600 to 00:02.700, a thin horizontal digital phase glitch sweeps across the carriage with brief RGB channel separation and blocky temporal displacement restricted to the passengers' heads; their normal human heads transform into compact volumetric storm clouds while their necks, bodies, clothing, hands, and positions remain perfectly unchanged. From 00:02.700 to 00:03.450, the interference reverses and restores the same normal, stable human faces. From 00:03.450 to 00:05.550, two faster synchronized glitch pulses alternate every head between the same normal human face and its cloud form, with no body movement, no face melting, and no change to the carriage geometry. From 00:05.550 onward, a final sustained digital stutter locks the passengers into their distinct cloud-headed states: lightning flickers inside the left foreground cloud, warm light glows inside the left midground cloud, rain falls beneath the small dark cloud, a faint rainbow forms in the right-center cloud, and a small sun glows inside the right foreground cloud. Fine luminous mist rises from the central cloud and resolves into the huge pale cloud face and single open eye overhead. The glitches settle, all motion becomes nearly still, and at exactly 8.00 seconds the passenger count, cloud shapes, body poses, wardrobe, seats, poles, windows, lighting, colors, crop, ceiling apparition, and full composition converge precisely to <Picture 1>. No cuts, no camera movement, no zoom, no pan, no tilt, no roll, no shake, no body morphing, no extra people, no missing passengers, no fused limbs, no wardrobe changes, no seat changes, no floating bodies, no text, no subtitles, no logos, and no drift from the final reference image.\n\noverall_soundscape: A steady low subway rumble, muted wheel rhythm, and soft ventilation hum continue throughout. Each transformation is marked by a brief dry digital stutter and electrical crackle, while the final cloud state adds faint rain, distant thunder, and a soft airy hiss without overpowering the train ambience.\n\nnon_diegetic_music: N/A",
    "sourceUrl": "https://x.com/AION_2077/status/2086595652692427042",
    "sourceLabel": "X 原帖 · @AION_2077",
    "author": "AION",
    "publishedAt": "2026-08-09T23:29:05.331Z",
    "addedAt": "2026-08-10T07:56:15+08:00",
    "mediaUrl": "/media/x-aion-cloud-head-subway.mp4",
    "posterUrl": "/posters/x/x-aion-cloud-head-subway.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "1080x1080",
    "tags": [
      "Ref2VA",
      "Glitch",
      "Cloud Heads",
      "Static Camera"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 1,
      "views": 15,
      "capturedAt": "2026-08-09T23:46:03Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 8.064,
      "resolution": "1080x1080",
      "aspectRatio": "square",
      "generationSettings": "Local MiniMax H3; final reference image alignment"
    }
  },
  {
    "id": "x-endfolding-rainy-bar-comparison",
    "title": "雨夜酒吧连续跟拍对比",
    "titleEn": "Rainy Bar Continuous-Tracking Comparison",
    "model": "MiniMax H3 · ComfyUI / Seedance 2.5",
    "mode": "Ref2VA",
    "summary": "同一雨夜街道概念分别交给 Seedance 2.5 与 ComfyUI MiniMax H3：H3 版本以 1376×768 输出 15 秒连续跟拍，并用明确的时间段约束人物从正面经过镜头后转为背面跟随；原帖公开了两版完整 Prompt。",
    "summaryEn": "The same rainy-street concept is tested with Seedance 2.5 and ComfyUI MiniMax H3. The H3 version outputs a 15-second, 1376×768 continuous tracking shot with timed constraints for moving from a frontal approach to a rear follow. The post publishes both complete prompts.",
    "prompt": "<image 2>: location\n\n<image 1>: Chris character reference \n\nintegrated_multimodal_description:\nCinematic rain-slicked urban evening with deep teal shadows, warm amber highlights, wet asphalt, shallow puddles, and light mist in the air. ONE continuous shot with NO CUTS. There is only ONE Chris in the entire video. Chris must remain the SAME woman with the SAME face, hair, clothing, body, and appearance throughout the entire shot. Never duplicate Chris, never create a second version of her, and never have her disappear and re-enter.\n\n[0s–4s] Medium tracking shot. Chris walks directly toward the camera along the wet street, looking toward the camera for approximately two beats. The camera smoothly tracks backward in front of her, keeping her face clearly visible and framed at medium distance. Warm amber streetlights reflect across the wet pavement behind her.\n\n[4s–7s] Chris reaches the camera and naturally walks very close past the LEFT side of the camera. IMPORTANT CONTINUITY: the camera does NOT cut away and Chris does NOT leave the continuous action. As her body passes beside the lens, the camera immediately pivots left and rotates approximately 180 degrees with her movement, continuously tracking the SAME Chris as the viewpoint transitions naturally from facing her to following directly behind her.\n\n[7s–15s] The camera finishes the pivot behind Chris and continues following the SAME woman from a medium rear tracking position. Ahead of her is the wet urban street and the Blackwood Bar and Grill storefront. Chris steps off the curb and crosses the street, moving carefully between parked cars with deliberate natural steps. She continues directly toward the Blackwood Bar and Grill. Its cool-white illuminated sign reads \"Blackwood Bar and Grill\" above dark wood-framed windows, creating a soft glow through the mist. Warm amber streetlights backlight Chris and produce long reflections across shallow puddles.\n\nCamera behavior: one uninterrupted continuous tracking shot; no edits, no cuts, no teleportation, no character substitution. Natural handheld cinematography with subtle realistic movement and mild walking shake. During the crossover, maintain continuous visual tracking of Chris's physical movement past the camera so there is never an opportunity for a second Chris to appear. Keep Chris close enough to camera for stable character and facial consistency during the opening approach.\n\noverall_soundscape:\nNatural rainy urban evening ambience. Soft footsteps on wet pavement synchronized with Chris's movement, distant traffic hum, subtle tire noise on wet streets, occasional passing cars, faint environmental city sounds, and quiet atmospheric rain and water sounds.\n\nnon_diegetic_music:\nNone.",
    "sourceUrl": "https://x.com/EndFolding79421/status/2086584639091536239",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-08-09T22:45:19.484Z",
    "addedAt": "2026-08-10T07:56:15+08:00",
    "mediaUrl": "/media/x-endfolding-rainy-bar-comparison.mp4",
    "posterUrl": "/posters/x/x-endfolding-rainy-bar-comparison.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376x768",
    "tags": [
      "Ref2VA",
      "Model Comparison",
      "Character Continuity",
      "One Take"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "bookmarks": 1,
      "views": 95,
      "capturedAt": "2026-08-09T23:46:03Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 2,
      "durationSeconds": 15.083,
      "resolution": "H3 1376x768; Seedance 2.5 1280x720",
      "aspectRatio": "landscape",
      "referenceImageCount": 2
    }
  },
  {
    "id": "x-cardoso-single-image-voice-prosody",
    "title": "单图驱动的双人对白与韵律",
    "titleEn": "Single-Image Dialogue and Prosody Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "作者只使用一张角色参考图，让 H3 生成 15 秒双人室内对话片段，并特别评价模型对声音、语调和韵律的处理；原帖没有公开 Prompt。",
    "summaryEn": "Using only one character reference image, the creator generates a 15-second indoor two-person dialogue scene and specifically praises H3's handling of voice, intonation, and prosody. The post does not publish a prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cardoso/status/2086575833267769614",
    "sourceLabel": "X 原帖 · @Cardoso",
    "author": "cardoso",
    "publishedAt": "2026-08-09T22:10:20.012Z",
    "addedAt": "2026-08-10T07:56:15+08:00",
    "mediaUrl": "/media/x-cardoso-single-image-voice-prosody.mp4",
    "posterUrl": "/posters/x/x-cardoso-single-image-voice-prosody.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344x768",
    "tags": [
      "Single Reference Image",
      "Voice",
      "Prosody",
      "Dialogue"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "image",
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 48,
      "bookmarks": 9,
      "views": 2995,
      "capturedAt": "2026-08-09T23:46:03Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.042,
      "resolution": "1344x768",
      "aspectRatio": "landscape",
      "referenceImageCount": 1
    }
  },
  {
    "id": "x-iammal-isekai-local-anime",
    "title": "本地生成的异世界动漫片段",
    "titleEn": "Locally Generated Isekai Anime Sequence",
    "model": "MiniMax H3 · Local ComfyUI",
    "mode": "Unknown",
    "summary": "一支约 29 秒的异世界日常动漫短片，作者明确标注为本地 H3 生成；回复中说明成片最初由两段 15 秒片段构成，森林部分又从两到三个片段中选取并剪辑。",
    "summaryEn": "An approximately 29-second isekai slice-of-life anime short explicitly labeled as a local H3 generation. In a reply, the creator says it began as two 15-second sections, with the forest portion edited from two or three clips.",
    "prompt": null,
    "sourceUrl": "https://x.com/iammal/status/2086568903069597890",
    "sourceLabel": "X 原帖 · @iammal",
    "author": "Mal",
    "publishedAt": "2026-08-09T21:42:47.724Z",
    "addedAt": "2026-08-10T07:56:15+08:00",
    "mediaUrl": "/media/x-iammal-isekai-local-anime.mp4",
    "posterUrl": "/posters/x/x-iammal-isekai-local-anime.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "1280x736",
    "tags": [
      "Local Generation",
      "Anime",
      "ComfyUI",
      "Multi-clip"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "bookmarks": 0,
      "views": 150,
      "capturedAt": "2026-08-09T23:46:03Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 29.141,
      "resolution": "1280x736",
      "aspectRatio": "landscape",
      "workflow": "Local ComfyUI; edited from two 15-second sections and multiple forest clips"
    }
  },
  {
    "id": "x-toshi-local-dance-rtx3080",
    "title": "RTX 3080 本地角色舞蹈",
    "titleEn": "Local Character Dance on an RTX 3080",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "在 RTX 3080 与 Core i7-13700K 上本地生成 10 秒竖屏角色舞蹈；作者说明借用了另一位创作者的舞蹈 Prompt，并肯定了成片音乐的停顿与收尾，但本帖未逐字公开该 Prompt。",
    "summaryEn": "A 10-second portrait character dance generated locally on an RTX 3080 and Core i7-13700K. The creator says they borrowed another creator's dance prompt and liked the musical pauses and ending, but this post does not reproduce the prompt verbatim.",
    "prompt": null,
    "sourceUrl": "https://x.com/Toshi_nyaruo_AI/status/2086596504752726378",
    "sourceLabel": "X 原帖 · @Toshi_nyaruo_AI",
    "author": "Toshi@ニャルオAI",
    "publishedAt": "2026-08-09T23:32:28.478Z",
    "addedAt": "2026-08-10T07:56:15+08:00",
    "mediaUrl": "/media/x-toshi-local-dance-rtx3080.mp4",
    "posterUrl": "/posters/x/x-toshi-local-dance-rtx3080.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "608x1056",
    "tags": [
      "Local Generation",
      "Dance",
      "RTX 3080",
      "Character Animation"
    ],
    "category": "dance",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "image",
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 39,
      "capturedAt": "2026-08-09T23:46:03Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.208,
      "resolution": "608x1056",
      "aspectRatio": "portrait",
      "generationSettings": "Core i7-13700K; RTX 3080; 32GB DDR4; local generation"
    }
  },
  {
    "id": "x-alone1moon-lip-sync-test",
    "title": "十秒后的口型失稳测试",
    "titleEn": "Lip-Sync Breakdown After Ten Seconds",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "15 秒舞台歌唱口型测试；作者明确说明约 10 秒后歌词开始失稳，因此替换了音频，并认为替换后的声音与画面仍有一定匹配度。",
    "summaryEn": "A 15-second stage-singing lip-sync test. The creator explicitly says the lyrics break down after roughly ten seconds, so the audio was replaced, while the replacement still appears reasonably aligned with the mouth movement.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2086588332965024214",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-09T23:00:00.172Z",
    "addedAt": "2026-08-10T07:56:15+08:00",
    "mediaUrl": "/media/x-alone1moon-lip-sync-test.mp4",
    "posterUrl": "/posters/x/x-alone1moon-lip-sync-test.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "864x1184",
    "tags": [
      "Lip Sync",
      "Audio Replacement",
      "Failure Boundary",
      "Anime"
    ],
    "category": "dialogue",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 16,
      "bookmarks": 2,
      "views": 256,
      "capturedAt": "2026-08-09T23:46:03Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.147,
      "resolution": "864x1184",
      "aspectRatio": "portrait",
      "postProduction": "Audio replaced after the creator observed lyric breakdown after 10 seconds"
    }
  },
  {
    "id": "x-notorjb-magical-girl-transformation",
    "title": "4070 Ti 本地魔法少女变身",
    "titleEn": "Local Magical-Girl Transformation on a 4070 Ti",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "10 秒魔法少女变身片段，用快速动作掩盖局部破绽；作者公开的本地性能数据为 RTX 4070 Ti 12GB 生成约 10 秒视频耗时约 9 分钟。",
    "summaryEn": "A 10-second magical-girl transformation sequence that uses rapid motion to mask local artifacts. The creator reports that an RTX 4070 Ti 12GB takes roughly nine minutes to generate about ten seconds locally.",
    "prompt": null,
    "sourceUrl": "https://x.com/NOTorJB/status/2086586609668137267",
    "sourceLabel": "X 原帖 · @NOTorJB",
    "author": "NOT氏（大杜製作所）",
    "publishedAt": "2026-08-09T22:53:09.306Z",
    "addedAt": "2026-08-10T07:56:15+08:00",
    "mediaUrl": "/media/x-notorjb-magical-girl-transformation.mp4",
    "posterUrl": "/posters/x/x-notorjb-magical-girl-transformation.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "576x736",
    "tags": [
      "Local Generation",
      "RTX 4070 Ti",
      "Transformation",
      "Benchmark"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 2,
      "views": 67,
      "capturedAt": "2026-08-09T23:46:03Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.208,
      "resolution": "576x736",
      "aspectRatio": "portrait",
      "generationSettings": "RTX 4070 Ti 12GB; approximately 9 minutes for a 10-second clip"
    }
  },
  {
    "id": "x-koda-suno-beat-synced-five-environments",
    "title": "Suno 节拍同步的五环境角色蒙太奇",
    "titleEn": "Suno Beat-Synced Character Montage Across Five Environments",
    "model": "MiniMax H3 · Suno",
    "mode": "Ref2VA",
    "summary": "使用角色设定图和 15 秒 Suno 音轨作为双重参考，在五个环境中按强拍、半拍和乐句切换 30 个镜头；原帖完整公开了角色一致性、镜头结构、节拍同步与负面约束 Prompt。",
    "summaryEn": "A character sheet and a 15-second Suno track guide 30 shots across five environments, with cuts tied to strong beats, half-beats, and musical phrases. The post publishes the full prompt for identity consistency, shot structure, audio synchronization, and negative constraints.",
    "prompt": "Use @[char ref] as the strict character reference and @[audio ref] as the timing, rhythm and editing reference.\n\nKeep the character’s exact identity, proportions, hairstyle, outfit, colors and overall style consistent throughout.\n\nCreate a 15-second cinematic burst-cut video showcasing the character across 5 different environments that naturally fit their design, vibe and world.\n\nAUDIO SYNC\nSynchronize the entire edit to @[audio ref]. Cuts, camera accents, transitions and environment changes should land precisely on strong beats, half-beats and musical accents. Let audio1 control the pacing and intensity of the montage.\n\nSTRUCTURE\n- 5 environments total\n- 3 seconds per environment\n- 6 burst-cut shots per environment\n- 30 shots total\n\nEach environment must be clearly different in atmosphere, lighting, scale and visual language.\n\nShow each environment through rapid cinematic angles: wide establishing shots, aerials, low angles, side views, tracking shots, close environmental details, medium shots and hero frames.\n\nEvery cut must reveal a new angle, distance, composition or spatial relationship. Avoid repeated framing. Mix static shots, push-ins, pull-backs, tracking, orbit and crane-like movement.\n\nKeep character movement subtle and natural. The focus is environmental variety, cinematic framing and tight synchronization with audio1.\n\nHard constraints:\n- exactly 5 environments\n- exactly 6 shots per environment\n- exactly 30 shots total\n- environment changes must follow audio1’s musical phrasing\n- cuts and motion accents synchronized to audio1\n- no outfit changes\n- no character duplication\n- no morphing\n- no text or UI\n- no blurry unreadable frames\n- maintain strict character consistency",
    "sourceUrl": "https://x.com/aimikoda/status/2086553240448241802",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-09T20:40:33.464Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-koda-suno-beat-synced-five-environments.mp4",
    "posterUrl": "/posters/x/2086553240448241802.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "原帖未标注",
    "tags": [
      "Ref2VA",
      "Suno",
      "Beat Sync",
      "Character Consistency"
    ],
    "category": "showcase",
    "styles": [
      "cinematic-realistic",
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image",
      "audio"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 49,
      "bookmarks": 32,
      "views": 1541,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.125,
      "resolution": "原帖未标注",
      "aspectRatio": "portrait",
      "generationSettings": "15 seconds; five environments; thirty shots; character-sheet and audio references"
    }
  },
  {
    "id": "x-ivan-space-cruise-black-hole-pov",
    "title": "黑洞邮轮客房手持 POV",
    "titleEn": "Handheld Space-Cruise POV at a Black Hole",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "游客在太空邮轮客房中用手持视角拍摄窗外的巨型黑洞，中途关灯以增强窗外可见度，并说出一句短对白；原帖逐字公开了完整 Prompt。",
    "summaryEn": "A tourist records a handheld view from a space-cruise cabin as a giant black hole fills the window. They switch off the light to improve the view and speak a short line. The post publishes the complete prompt verbatim.",
    "prompt": "amateur handheld pov footage of a tourist in their plush and comfortable room of a space cruise, carpeted floor and comfy bed, large window shows a view of Gargantua blackhole, you can see their reflection in the window, they turn off the light half way through so we can see outside better and say: \"Wow look at that!\", then they show back the room",
    "sourceUrl": "https://x.com/ivanfioravanti/status/2086553101029290296",
    "sourceLabel": "X 原帖 · @ivanfioravanti",
    "author": "Ivan Fioravanti ᯅ",
    "publishedAt": "2026-08-09T20:40:00.224Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-ivan-space-cruise-black-hole-pov.mp4",
    "posterUrl": "/posters/x/2086553101029290296.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "原帖未标注",
    "tags": [
      "T2VA",
      "POV",
      "Dialogue",
      "Science Fiction"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "sci-fi",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 5,
      "reposts": 2,
      "likes": 43,
      "bookmarks": 6,
      "views": 3171,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10,
      "resolution": "原帖未标注",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-atomtan-thirty-second-continuous-speech",
    "title": "30 秒连续对白与口型同步",
    "titleEn": "Thirty-Second Continuous Speech and Lip Sync",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "一段完整 30 秒竖屏人物对白测试。作者明确称这是单段生成而非多个短片拼接，并报告全程连续语音没有出现明显同步漂移；原帖未公开 Prompt。",
    "summaryEn": "A full 30-second portrait dialogue test. The creator explicitly describes it as one generation rather than a stitched sequence and reports continuous speech without noticeable sync drift. The post does not publish a prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/atomtanstudio/status/2086548496170246638",
    "sourceLabel": "X 原帖 · @atomtanstudio",
    "author": "Rich · Atom Tan Studio",
    "publishedAt": "2026-08-09T20:21:42.340Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-atomtan-thirty-second-continuous-speech.mp4",
    "posterUrl": "/posters/x/2086548496170246638.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "原帖未标注",
    "tags": [
      "Long Dialogue",
      "Lip Sync",
      "Native Audio",
      "Local Generation"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 8,
      "bookmarks": 2,
      "views": 556,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 30,
      "resolution": "原帖未标注",
      "aspectRatio": "portrait",
      "workflow": "Creator describes a single 30-second generation with continuous synchronized speech"
    }
  },
  {
    "id": "x-redpolo-ref2va-frame-infill-cameos",
    "title": "Ref2VA 帧补全的角色客串插入",
    "titleEn": "Ref2VA Frame-Infill Cameo Insertion",
    "model": "MiniMax H3 · ComfyUI",
    "mode": "Ref2VA",
    "summary": "ComfyUI-MiniMaxH3FrameInfill 新增 Ref2VA 节点：演示只重生成舞蹈视频 3.0–4.5 秒以插入客串角色，以及按时间指定两名角色依次入场、退场和击掌。作者明确标注该功能仍属实验性。",
    "summaryEn": "A new Ref2VA node for ComfyUI-MiniMaxH3FrameInfill demonstrates regenerating only seconds 3.0–4.5 of a dance clip to insert a cameo, plus timed entrances, exits, and a high-five for two characters. The creator explicitly labels the feature experimental.",
    "prompt": null,
    "sourceUrl": "https://x.com/redpolo999/status/2086549225832362433",
    "sourceLabel": "X 原帖 · @redpolo999",
    "author": "あかぽろ@日曜技術",
    "publishedAt": "2026-08-09T20:24:36.305Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-redpolo-ref2va-frame-infill-cameos.mp4",
    "posterUrl": "/posters/x/2086549225832362433.jpg",
    "duration": 23,
    "aspectRatio": "landscape",
    "resolution": "原帖未标注",
    "tags": [
      "Ref2VA",
      "Frame Infill",
      "ComfyUI",
      "Character Insertion"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image",
      "video"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 10,
      "bookmarks": 15,
      "views": 1939,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 2,
      "durationSeconds": 23,
      "resolution": "原帖未标注",
      "aspectRatio": "landscape",
      "workflow": "Experimental Ref2VA frame infill with image references and timed character insertion"
    }
  },
  {
    "id": "x-photogpt-western-beat-title-sequence",
    "title": "西部片节拍标题动画",
    "titleEn": "Beat-Synced Spaghetti-Western Title Sequence",
    "model": "MiniMax H3 · PhotoGPT",
    "mode": "Ref2VA",
    "summary": "角色参考图与音轨共同驱动的 10 秒意大利西部片风格标题序列：硬切、定格、剪影、模板字和旧胶片损伤全部对齐节拍；原帖完整公开了 Prompt。",
    "summaryEn": "A character reference and audio track drive a 10-second spaghetti-western title sequence whose hard cuts, freeze frames, silhouettes, stencil typography, and distressed-film effects land on the beat. The full prompt is published in the post.",
    "prompt": "Use the attached character reference (Image1) and audio track as the timing guide. Create a spaghetti-western pulp anime title sequence where every cut lands precisely on the beat. Style the character as a flat black silhouette against solid desert-inspired colors (burnt orange, ochre, rust, adobe tan), using bold pop-art compositions, vintage bounty-poster framing, split panels, extreme close-ups, and wide horizon shots. Alternate fast action bursts with full-beat freeze frames (draw, stride, turn, match strike, etc.), using hard cuts only. Add worn woodblock/stencil title typography that slams into frame on the beat. Finish with the character frozen in silhouette on a burnt-orange background as the title stamps in on the final hit. Apply heavy 35mm film grain, scratches, dust, gate weave, cigarette burns, and sun-faded print wear throughout for an authentic 1960s spaghetti-western look.",
    "sourceUrl": "https://x.com/itsphotogptai/status/2086555941831626849",
    "sourceLabel": "X 原帖 · @itsphotogptai",
    "author": "PhotoGPT",
    "publishedAt": "2026-08-09T20:51:17.524Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-photogpt-western-beat-title-sequence.mp4",
    "posterUrl": "/posters/x/2086555941831626849.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "原帖未标注",
    "tags": [
      "Ref2VA",
      "Title Animation",
      "Beat Sync",
      "Typography"
    ],
    "category": "animation",
    "styles": [
      "retro",
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image",
      "audio"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 0,
      "views": 97,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10,
      "resolution": "原帖未标注",
      "aspectRatio": "landscape",
      "generationSettings": "Character image and audio-track references; beat-synchronized editing"
    }
  },
  {
    "id": "x-cardoso-no-reference-rtx5060ti-audio",
    "title": "RTX 5060 Ti 本地无参考音视频测试",
    "titleEn": "Local No-Reference H3 Test on an RTX 5060 Ti",
    "model": "MiniMax H3 · Local",
    "mode": "T2VA",
    "summary": "Cardoso 在 RTX 5060 Ti 16GB 上本地运行 H3，发布了一段 14 秒低分辨率测试。原帖明确说明没有使用任何外部参考，并强调声音、语调与表情均由模型生成。",
    "summaryEn": "Cardoso runs H3 locally on an RTX 5060 Ti 16GB and publishes a 14-second low-resolution test. The post explicitly states that no external reference was used and highlights the model-generated voices, intonation, and expressions.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cardoso/status/2086561122597925104",
    "sourceLabel": "X 原帖 · @Cardoso",
    "author": "cardoso",
    "publishedAt": "2026-08-09T21:11:52.715Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-cardoso-no-reference-rtx5060ti-audio.mp4",
    "posterUrl": "/posters/x/2086561122597925104.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "原帖未标注",
    "tags": [
      "T2VA",
      "RTX 5060 Ti",
      "No Reference",
      "Native Audio"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 5,
      "reposts": 2,
      "likes": 48,
      "bookmarks": 6,
      "views": 2646,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 14,
      "resolution": "Creator labels the test low resolution",
      "aspectRatio": "landscape",
      "generationSettings": "Local RTX 5060 Ti 16GB; no external reference"
    }
  },
  {
    "id": "x-alawlette-four-minute-local-t2v",
    "title": "4 分 34 秒本地 T2V 情景喜剧测试",
    "titleEn": "Four-Minute Local T2V Sitcom Test",
    "model": "MiniMax H3 · Local",
    "mode": "T2VA",
    "summary": "一支 X 上传时长为 4 分 34 秒的本地 T2V 情景喜剧连贯性测试。作者表示新节点正在改善长段落的一致性；该时长仅表示上传视频总长，不代表一次 H3 直接生成。",
    "summaryEn": "A locally produced T2V sitcom-coherence test with a total X upload length of four minutes and 34 seconds. The creator says new nodes are improving long-sequence consistency. The upload length is not presented as a single H3 generation.",
    "prompt": null,
    "sourceUrl": "https://x.com/alawlette/status/2086600476477514166",
    "sourceLabel": "X 原帖 · @alawlette",
    "author": "alawlette",
    "publishedAt": "2026-08-09T23:48:15.411Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-alawlette-four-minute-local-t2v.mp4",
    "posterUrl": "/posters/x/2086600476477514166.jpg",
    "duration": 274,
    "aspectRatio": "landscape",
    "resolution": "原帖未标注",
    "tags": [
      "T2VA",
      "Long Form",
      "Local Generation",
      "Continuity"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 22,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 274.125,
      "resolution": "原帖未标注",
      "aspectRatio": "landscape",
      "workflow": "Local T2V with additional nodes; 4:34 is the X upload duration, not a claimed single generation"
    }
  },
  {
    "id": "x-mikefrost-bobblehead-photoreal-cartoon-audio",
    "title": "单图摇头公仔的写实与卡通双版本",
    "titleEn": "One-Image Bobblehead in Photoreal and Cartoon Passes",
    "model": "MiniMax H3 · Local RTX 5090",
    "mode": "FL2VA",
    "summary": "一张桌面摇头公仔照片分别生成写实和二维卡通两版 5 秒动画。作者说明画面与音频同步生成，没有额外拟音或 TTS，并在 RTX 5090 上本地完成。",
    "summaryEn": "One desk photo of a bobblehead becomes two five-second passes, one photoreal and one 2D cartoon. The creator says picture and audio were generated together without added foley or text-to-speech, entirely on an RTX 5090.",
    "prompt": null,
    "sourceUrl": "https://x.com/slfemp/status/2086551052778930232",
    "sourceLabel": "X 原帖 · @slfemp",
    "author": "Mike Frost",
    "publishedAt": "2026-08-09T20:31:51.883Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-mikefrost-bobblehead-photoreal-cartoon-audio.mp4",
    "posterUrl": "/posters/x/2086551052778930232.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "原帖未标注",
    "tags": [
      "FL2VA",
      "Single Image",
      "Native Audio",
      "RTX 5090"
    ],
    "category": "workflow",
    "styles": [
      "photoreal",
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "image",
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 75,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "threadVideoCount": 2,
      "durationSeconds": 5.166,
      "resolution": "原帖未标注",
      "aspectRatio": "landscape",
      "generationSettings": "Local RTX 5090; image and audio generated together; no added foley or TTS"
    }
  },
  {
    "id": "x-pelsan-storyboard-prompt-text-rendering",
    "title": "故事板提示词的文字生成测试",
    "titleEn": "Storyboard-and-Prompt Text Rendering Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "用故事板加提示词测试 15 秒 H3 成片，覆盖宽银幕场景和画面内文字。作者特别指出仅凭提示词即可生成文字并按指示呈现，但没有逐字公开所用 Prompt。",
    "summaryEn": "A 15-second H3 test combines a storyboard with prompting across widescreen scenes and on-screen text. The creator specifically notes that the text follows prompt instructions, but does not publish the prompt verbatim.",
    "prompt": null,
    "sourceUrl": "https://x.com/Creator_Pelsan/status/2086558140045169100",
    "sourceLabel": "X 原帖 · @Creator_Pelsan",
    "author": "PeL：I&P",
    "publishedAt": "2026-08-09T21:00:01.619Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-pelsan-storyboard-prompt-text-rendering.mp4",
    "posterUrl": "/posters/x/2086558140045169100.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "原帖未标注",
    "tags": [
      "Storyboard",
      "Text Rendering",
      "Prompt Following",
      "Cinematic"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "bookmarks": 2,
      "views": 292,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15,
      "resolution": "原帖未标注",
      "aspectRatio": "landscape",
      "workflow": "Storyboard and prompt; creator reports prompt-following for on-screen text"
    }
  },
  {
    "id": "x-strimando-rtx5090-1080p-benchmark",
    "title": "RTX 5090 本地 1080P 生成实测",
    "titleEn": "Local 1080p Generation Benchmark on an RTX 5090",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "RTX 5090 本地生成 10 秒 1080P 视频的公开性能记录：作者报告耗时 15 分 23 秒。相邻回复还给出了 15 秒 480P 版本耗时 1 分 27 秒，便于理解分辨率成本。",
    "summaryEn": "A public local-performance record for generating a 10-second 1080p clip on an RTX 5090, with a reported runtime of 15 minutes and 23 seconds. An adjacent post reports one minute and 27 seconds for a 15-second 480p version, illustrating the resolution cost.",
    "prompt": null,
    "sourceUrl": "https://x.com/STRIMANDO/status/2086576134540726457",
    "sourceLabel": "X 原帖 · @STRIMANDO",
    "author": "STRIMANDO.",
    "publishedAt": "2026-08-09T22:11:31.841Z",
    "addedAt": "2026-08-10T08:14:56+08:00",
    "mediaUrl": "/media/x-strimando-rtx5090-1080p-benchmark.mp4",
    "posterUrl": "/posters/x/2086576134540726457.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1080P",
    "tags": [
      "Local Generation",
      "RTX 5090",
      "1080P",
      "Benchmark"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "bookmarks": 0,
      "views": 214,
      "capturedAt": "2026-08-10T00:06:30Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10,
      "resolution": "1080P",
      "aspectRatio": "landscape",
      "generationSettings": "Local RTX 5090; 15 minutes 23 seconds for the 10-second 1080p clip"
    }
  },
  {
    "id": "x-juwalbose-2086615145195639159",
    "title": "Motion Context 无缝续接测试",
    "titleEn": "Motion Context Seam Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Ref2VA",
    "summary": "测试 MiniMax H3 Motion Context 新增的 R2V 支持，并展示在替换参考、切换生成方式后合并视频的接缝表现。",
    "summaryEn": "Tests the new R2V support in MiniMax H3 Motion Context and shows how well the merged video hides its seam after references and generation modes are switched.",
    "prompt": null,
    "sourceUrl": "https://x.com/JuwalBose/status/2086615145195639159",
    "sourceLabel": "X 原帖 · @JuwalBose",
    "author": "JuwalBose",
    "publishedAt": "2026-08-10T00:46:32.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-juwalbose-2086615145195639159.mp4",
    "posterUrl": "/posters/x/2086615145195639159.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "R2V",
      "视频续接",
      "参考替换"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image",
      "video"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 9.416,
      "resolution": "1280×736",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-nilian-2086612098239418450",
    "title": "Ideogram 与 H3 联合生成",
    "titleEn": "Ideogram and H3 Art Workflow",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 ComfyUI 中结合 Ideogram 与 MiniMax H3 制作一段科技表现主义风格的 AI 视频。",
    "summaryEn": "Combines Ideogram and MiniMax H3 inside ComfyUI to produce an AI video in a techspressionist style.",
    "prompt": null,
    "sourceUrl": "https://x.com/nilian/status/2086612098239418450",
    "sourceLabel": "X 原帖 · @nilian",
    "author": "Nima Nilian",
    "publishedAt": "2026-08-10T00:34:26.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-nilian-2086612098239418450.mp4",
    "posterUrl": "/posters/x/2086612098239418450.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "ComfyUI",
      "联合工作流",
      "AI 艺术"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 29.856,
      "resolution": "1920×1080",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-endfolding79421-2086607000507474111",
    "title": "20 秒 H3 长镜头测试",
    "titleEn": "Twenty-Second H3 Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "以针对 H3 指南调整过的同一思路制作 20 秒测试，并与此前的 LTX 2.3 版本形成可追溯对照。",
    "summaryEn": "Runs a twenty-second H3 test with the same underlying idea adapted to the H3 guide, providing a traceable comparison with an earlier LTX 2.3 version.",
    "prompt": "<image 1>: starting frame\n\nintegrated_multimodal_description:\n\ncinematic acting-audition performance. One woman, Chris, remains the only visible person throughout the entire video. Maintain the same face, hair, clothing, age, and identity for the full shot.\n\n[0-4s] Medium close-up framing from the chest up. Chris stands tall facing directly toward the camera, holding a script page naturally in one hand. The camera remains steady and close enough to preserve clear facial detail. Her expression is composed and quietly confident. She looks directly into the camera and introduces herself in her normal speaking voice.\n\n(S1, Chris): <d>[English] My name is Chris, and I am reading for the part of Lady Ash.</d>\n\n[4-6s] After finishing the introduction, Chris naturally lowers her eyes from the camera to the script page in her hand. Her confident audition demeanor begins to disappear. Her shoulders relax slightly, her expression softens, and she takes a small breath as she emotionally enters the character.\n\n[6-18s] Keep the camera in a close medium-close-up emphasizing her face and eyes. Chris reads from the page in a soft, sorrowful, extremely sad voice, as though Lady Ash is desperately pleading with someone she loves. Her performance is restrained and believable rather than theatrical. Her voice becomes increasingly fragile as the line continues.\n\n(S1, Chris): <d>[English] Please, Peter... please don't let this be our last night together. Don't go back to her. You know this is where you belong... with me... forever...</d>\n\nDuring the emotional reading, moisture gradually gathers in her eyes. Near the end of the line, a single tear forms naturally and slowly travels down one cheek. She does not wipe it away. Her eyes remain lowered toward the script.\n\n[18-20s] On the word “forever,” Chris lets the final word linger softly. She becomes completely still, remaining emotionally devastated and looking down at the page. Hold on her tear-streaked face for a brief silent beat. The image then slowly fades completely to black.\n\noverall_soundscape:\n\nClean indoor room tone appropriate for a quiet acting audition. Chris's voice is intimate, clear, and close. Very subtle sound of the script page moving in her hand. No other voices, no audience reaction, no distracting environmental sounds.\n\nnon_diegetic_music:\n\nNone. No music.",
    "sourceUrl": "https://x.com/EndFolding79421/status/2086607000507474111",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-08-10T00:14:10.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-endfolding79421-2086607000507474111.mp4",
    "posterUrl": "/posters/x/2086607000507474111.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "20 秒",
      "长视频",
      "同题对比"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 20.041,
      "resolution": "1376×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-xc5_-2086604532616958185",
    "title": "月光魔法少女变身",
    "titleEn": "Moonlit Magical-Girl Transformation",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在满月场景中完成魔法少女变身，并在后半段继续切换角色造型，突出连续的特效转化。",
    "summaryEn": "Stages a magical-girl transformation under a full moon, followed by another character change that emphasizes continuous visual effects.",
    "prompt": null,
    "sourceUrl": "https://x.com/xc5_/status/2086604532616958185",
    "sourceLabel": "X 原帖 · @xc5_",
    "author": "萌奈のおすそわけ♥️AI動画クリエイター🩷AI & Web3 KOL🧡💛💚",
    "publishedAt": "2026-08-10T00:04:22.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-xc5_-2086604532616958185.mp4",
    "posterUrl": "/posters/x/2086604532616958185.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "魔法变身",
      "角色转化",
      "月光"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 8,
      "reposts": 2,
      "likes": 13,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.265,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-cocktailpeanut-2086604154261148146",
    "title": "本地生成《柯基侠盗》",
    "titleEn": "Corgi Theft Auto, Generated Locally",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 Windows 与 A4500 20GB 显卡上通过 Maestro 本地运行 H3，仅启用 First Block Cache，制作一段柯基主题动作视频。",
    "summaryEn": "Runs H3 locally through Maestro on Windows with a 20 GB A4500, using only First Block Cache to create a corgi-themed action video.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2086604154261148146",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktailpeanut",
    "publishedAt": "2026-08-10T00:02:52.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-cocktailpeanut-2086604154261148146.mp4",
    "posterUrl": "/posters/x/2086604154261148146.jpg",
    "duration": 72,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "本地生成",
      "Maestro",
      "A4500",
      "First Block Cache"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 71.875,
      "resolution": "864×480",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-monetizationdon-2086603324086767952",
    "title": "哥特吸血鬼恐怖短片",
    "titleEn": "Gothic Dracula Horror Scene",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "通过 PixVerse 中的 MiniMax H3 和简短文字指令生成德古拉场景，重点测试雾、蝙蝠、镜面异象与戏剧光线。",
    "summaryEn": "Uses MiniMax H3 inside PixVerse with a short text instruction to create a Dracula scene focused on fog, bats, supernatural reflections, and dramatic lighting.",
    "prompt": "DRACULA — CINEMATIC HORROR PROMPT\n\nA vast abandoned Victorian ballroom at midnight, completely different from a corridor. Broken chandeliers hang from the ceiling, moonlight pours through enormous cracked stained-glass windows, and thick black mist drifts across the marble floor. Torn velvet curtains move slowly despite there being no wind. The camera begins with an extreme low-angle shot of an empty throne at the far end of the ballroom. A single candle beside it suddenly extinguishes. Dracula is already standing behind the throne. He slowly steps into the moonlight — an elegant, terrifying vampire wearing a long black Victorian coat with a high collar, dark waistcoat, polished boots, and subtle antique jewelry. Pale skin, long black hair, intense eyes, and unmistakable sharp fangs. His presence feels ancient and powerful rather than exaggerated. The camera slowly circles him as the ballroom begins behaving unnaturally. The chandelier swings violently. Hundreds of bats suddenly erupt from the darkness above, filling the room. Dracula doesn't move. He simply watches them pass. One bat lands on his shoulder. Dracula raises his hand. Every candle in the ballroom ignites simultaneously. The camera rapidly pushes toward his face. His eyes subtly change as he looks directly into the lens. He gives a restrained, sinister smile, revealing his fangs. Suddenly, the reflection of Dracula in the polished marble floor moves independently from him. The reflection turns its head toward the camera while Dracula remains perfectly still. The camera pulls backward rapidly as the reflection rises out of the marble like living smoke. Dracula finally spreads his arms, and a massive swarm of bats explodes outward behind him, transforming his silhouette into an enormous winged shadow across the ballroom walls. The chandelier crashes down in the background. CUT TO BLACK.",
    "promptSourceUrl": "https://x.com/MonetizationDon/status/2086603327412838769",
    "sourceUrl": "https://x.com/MonetizationDon/status/2086603324086767952",
    "sourceLabel": "X 原帖 · @MonetizationDon",
    "author": "LAPTOP LIFESTYLE BLOGGER",
    "publishedAt": "2026-08-09T23:59:34.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-monetizationdon-2086603324086767952.mp4",
    "posterUrl": "/posters/x/2086603324086767952.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "吸血鬼",
      "恐怖氛围",
      "戏剧光线"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-worldofray-2086602272335696047",
    "title": "克拉克森与谢尔顿的报应",
    "titleEn": "Clarkson and Sheldon: The Comeuppance",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用 MiniMax H3 制作克拉克森与谢尔顿同场的短视频，原帖将情节概括为一次“报应”。",
    "summaryEn": "Uses MiniMax H3 for a short scene featuring Clarkson and Sheldon, framed by the creator as a moment of comeuppance.",
    "prompt": null,
    "sourceUrl": "https://x.com/worldofray/status/2086602272335696047",
    "sourceLabel": "X 原帖 · @worldofray",
    "author": "Ray",
    "publishedAt": "2026-08-09T23:55:23.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-worldofray-2086602272335696047.mp4",
    "posterUrl": "/posters/x/2086602272335696047.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "名人角色",
      "喜剧短片"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1344×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-haru_maki_ch-2086601484901495018",
    "title": "H3 视频分段升至 4K",
    "titleEn": "Segmented H3 Upscale to 4K",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用 SeedVR2 将 H3 视频升至 3840×2160；为绕过 24GB 显存溢出，改为九个 0.75 秒片段完成 158 帧处理。",
    "summaryEn": "Upscales an H3 video to 3840x2160 with SeedVR2, avoiding a 24 GB VRAM overflow by processing 158 frames as nine 0.75-second segments.",
    "prompt": null,
    "sourceUrl": "https://x.com/hAru_mAki_ch/status/2086601484901495018",
    "sourceLabel": "X 原帖 · @hAru_mAki_ch",
    "author": "Maki@Sunwood AI Labs.",
    "publishedAt": "2026-08-09T23:52:15.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-haru_maki_ch-2086601484901495018.mp4",
    "posterUrl": "/posters/x/2086601484901495018.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "4K",
      "分段处理",
      "SeedVR2",
      "显存优化"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "video"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 6.583,
      "resolution": "3840×2160",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-mr_pig41-2086599612090012135",
    "title": "六滑窗双版本长视频",
    "titleEn": "Six-Window Dual-Take Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 WanGP 中用六个滑动窗口生成两个 MiniMax H3 长视频版本，公开承认两版各有瑕疵，也各有可取片段。",
    "summaryEn": "Uses six sliding windows in WanGP to generate two long MiniMax H3 takes, openly noting that each has flaws as well as useful moments.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mr_Pig41/status/2086599612090012135",
    "sourceLabel": "X 原帖 · @Mr_Pig41",
    "author": "Mr Pig",
    "publishedAt": "2026-08-09T23:44:49.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-mr_pig41-2086599612090012135.mp4",
    "posterUrl": "/posters/x/2086599612090012135.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "WanGP",
      "滑动窗口",
      "双版本"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 2,
      "durationSeconds": 59.833,
      "resolution": "1920×1088",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-worldofray-2086599024010723406",
    "title": "5090 上的超市三角色测试",
    "titleEn": "Three-Character Supermarket Test on a 5090",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 RTX 5090 上以 1344×768 生成五秒视频，将克拉克森、印第安纳·琼斯与终结者放进同一超市场景，并公布约 80 秒耗时。",
    "summaryEn": "Generates a five-second 1344x768 clip on an RTX 5090, placing Clarkson, Indiana Jones, and the Terminator in one supermarket scene with an approximately eighty-second runtime.",
    "prompt": null,
    "sourceUrl": "https://x.com/worldofray/status/2086599024010723406",
    "sourceLabel": "X 原帖 · @worldofray",
    "author": "Ray",
    "publishedAt": "2026-08-09T23:42:29.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-worldofray-2086599024010723406.mp4",
    "posterUrl": "/posters/x/2086599024010723406.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "RTX 5090",
      "多角色",
      "1344×768",
      "性能测试"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1344×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-sumica_rowa-2086598720515445036",
    "title": "糖果信号街机 MV",
    "titleEn": "Candy Signal Arcade MV",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "FL2VA",
    "summary": "用 MiniMax H3 将插画视频化，再配合 ACE 生成音乐重新编排合成，形成简洁的动态图形 MV。",
    "summaryEn": "Animates illustrations with MiniMax H3, then restructures and composites them around an ACE-generated track to create a compact motion-graphics music video.",
    "prompt": null,
    "sourceUrl": "https://x.com/sumica_rowa/status/2086598720515445036",
    "sourceLabel": "X 原帖 · @sumica_rowa",
    "author": "sumicaぐらし | rowa",
    "publishedAt": "2026-08-09T23:41:16.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-sumica_rowa-2086598720515445036.mp4",
    "posterUrl": "/posters/x/2086598720515445036.jpg",
    "duration": 24,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "插画视频化",
      "动态图形",
      "ACE 音乐"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "image",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 6,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 24,
      "resolution": "1024×576",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-sep_is_heim-2086598270328209823",
    "title": "4B 文本编码器低内存测试",
    "titleEn": "Four-Billion-Parameter Text Encoder Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用 ClipProj 将 H3 文本编码器从 32B 换为 4B FP8 版本，记录约 14GB 内存下降，并初步观察画面变化不大。",
    "summaryEn": "Uses ClipProj to replace the 32B H3 text encoder with a 4B FP8 version, reporting roughly 14 GB lower memory use and little visible change in the initial video test.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2086598270328209823",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-08-09T23:39:29.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-sep_is_heim-2086598270328209823.mp4",
    "posterUrl": "/posters/x/2086598270328209823.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1152×832",
    "tags": [
      "低内存",
      "ClipProj",
      "文本编码器",
      "FP8"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 5,
      "bookmarks": 1,
      "views": 260,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1152×832",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-ryuuseiwakimaai-2086597086813007911",
    "title": "标记模型转 3D 动画实验",
    "titleEn": "Marker Model to 3D Animation Experiment",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "先让 MiniMax H3 生成带标记模型运动的视频，再尝试据此制作 3D 动画；作者明确表示当前尝试效果不理想。",
    "summaryEn": "First generates a moving marker-covered model with MiniMax H3, then attempts to derive a 3D animation from it; the creator clearly reports that the current attempt did not work well.",
    "prompt": null,
    "sourceUrl": "https://x.com/ryuuseiwakimaai/status/2086597086813007911",
    "sourceLabel": "X 原帖 · @ryuuseiwakimaai",
    "author": "りゅうせい",
    "publishedAt": "2026-08-09T23:34:47.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-ryuuseiwakimaai-2086597086813007911.mp4",
    "posterUrl": "/posters/x/2086597086813007911.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "512×512",
    "tags": [
      "3D 动画",
      "动作标记",
      "失败实验"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "cg-3d"
    ],
    "scenes": [],
    "inputTypes": [
      "video"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.15,
      "resolution": "512×512",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-vkuoo-2086595994930799063",
    "title": "建筑展示板动态化",
    "titleEn": "Animated Architectural Presentation Board",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "FL2VA",
    "summary": "用 MiniMax H3 为建筑展示板加入干净运动，重点展示图像、版式与文字信息混排时的动态表现。",
    "summaryEn": "Uses MiniMax H3 to animate an architectural presentation board, emphasizing clean motion across mixed imagery, layout elements, and text.",
    "prompt": null,
    "sourceUrl": "https://x.com/vkuoo/status/2086595994930799063",
    "sourceLabel": "X 原帖 · @vkuoo",
    "author": "vkuoo",
    "publishedAt": "2026-08-09T23:30:26.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-vkuoo-2086595994930799063.mp4",
    "posterUrl": "/posters/x/2086595994930799063.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "834×1112",
    "tags": [
      "建筑可视化",
      "版式动画",
      "混合媒体"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.072,
      "resolution": "834×1112",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-sep_is_heim-2086586428251857023",
    "title": "Ref2VA 角色表近景一致性",
    "titleEn": "Ref2VA Character-Sheet Close-Up",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Ref2VA",
    "summary": "以角色表作为 Ref2VA 参考输入测试近景面部一致性；作者认为即使没有角色 LoRA，近景相似度也较高。",
    "summaryEn": "Uses a character sheet as Ref2VA input to test close-up facial consistency; the creator reports strong resemblance even without a character LoRA.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2086586428251857023",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-08-09T22:52:26.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-sep_is_heim-2086586428251857023.mp4",
    "posterUrl": "/posters/x/2086586428251857023.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1152×1664",
    "tags": [
      "角色表",
      "近景",
      "一致性",
      "Ref2VA"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1152×1664",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-thedarkwebdev-2086586412447514971",
    "title": "H3 趣味梗视频合集",
    "titleEn": "MiniMax H3 Meme Experiments",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "作者在大型项目间隙用 MiniMax H3 尝试多种趣味梗视频，并将其视为探索模型和寻找新灵感的过程。",
    "summaryEn": "The creator uses MiniMax H3 for a set of playful meme-video experiments between larger projects, treating them as a way to explore the model and find new ideas.",
    "prompt": null,
    "sourceUrl": "https://x.com/TheDarkWebDev/status/2086586412447514971",
    "sourceLabel": "X 原帖 · @TheDarkWebDev",
    "author": "The Dark Developer",
    "publishedAt": "2026-08-09T22:52:22.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-thedarkwebdev-2086586412447514971.mp4",
    "posterUrl": "/posters/x/2086586412447514971.jpg",
    "duration": 302,
    "aspectRatio": "landscape",
    "resolution": "1872×1080",
    "tags": [
      "梗视频",
      "创意实验"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 301.556,
      "resolution": "1872×1080",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-monetizationdon-2086585814624919552",
    "title": "同提示词音乐视频模型对比",
    "titleEn": "Same-Prompt Music Video Comparison",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Ref2VA",
    "summary": "将同一角色表、电影分镜与复古流行音乐视频指令分别交给 MiniMax H3 和 Seedance 2.5，比较角色一致性与多场景完成度。",
    "summaryEn": "Tests the same character sheet, cinematic storyboard, and retro-pop music-video instruction with MiniMax H3 and Seedance 2.5, comparing character consistency across multiple scenes.",
    "prompt": null,
    "sourceUrl": "https://x.com/MonetizationDon/status/2086585814624919552",
    "sourceLabel": "X 原帖 · @MonetizationDon",
    "author": "LAPTOP LIFESTYLE BLOGGER",
    "publishedAt": "2026-08-09T22:49:59.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-monetizationdon-2086585814624919552.mp4",
    "posterUrl": "/posters/x/2086585814624919552.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "同提示词",
      "角色一致性",
      "Seedance 2.5",
      "音乐视频"
    ],
    "category": "comparison",
    "styles": [
      "retro",
      "cinematic-realistic"
    ],
    "scenes": [
      "daily-life",
      "city"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 2,
      "durationSeconds": 30.231,
      "resolution": "1920×1080",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-core_tan-2086584384774390187",
    "title": "原生 1080P 电视剧运镜",
    "titleEn": "Native 1080p TV-Drama Camera Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "在配备 RTX PRO 6000 Blackwell 的 H3 WebUI 中，通过文字控制电视剧式运镜，展示原生 1080P 生成效果。",
    "summaryEn": "Uses text-directed television-drama camera work in an H3 WebUI on an RTX PRO 6000 Blackwell to demonstrate native 1080p generation.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2086584384774390187",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-09T22:44:18.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-core_tan-2086584384774390187.mp4",
    "posterUrl": "/posters/x/2086584384774390187.jpg",
    "duration": 6,
    "aspectRatio": "landscape",
    "resolution": "1920×1056",
    "tags": [
      "原生 1080P",
      "WebUI",
      "运镜",
      "RTX PRO 6000"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.875,
      "resolution": "1920×1056",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-strimando-2086580732521259155",
    "title": "5090 单提示词本地生成",
    "titleEn": "Single-Prompt Local Generation on a 5090",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "在 RTX 5090 本地运行 MiniMax H3，以单条文字指令完成视频，作者公布生成耗时约七分钟。",
    "summaryEn": "Runs MiniMax H3 locally on an RTX 5090 and completes the video from a single text instruction in approximately seven minutes.",
    "prompt": null,
    "sourceUrl": "https://x.com/STRIMANDO/status/2086580732521259155",
    "sourceLabel": "X 原帖 · @STRIMANDO",
    "author": "STRIMANDO.",
    "publishedAt": "2026-08-09T22:29:48.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-strimando-2086580732521259155.mp4",
    "posterUrl": "/posters/x/2086580732521259155.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "RTX 5090",
      "单提示词",
      "生成耗时"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "1344×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-kashimu_shed-2086577432241582478",
    "title": "双外录音频分角色对白",
    "titleEn": "Two Recorded Voices Assigned by Character",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Ref2VA",
    "summary": "将两段外录音频分别指定为 audio1 和 audio2，MiniMax H3 按角色分配台词，展示多音频条件下的对白控制。",
    "summaryEn": "Assigns two separately recorded tracks as audio1 and audio2, allowing MiniMax H3 to route the lines to different characters and demonstrate multi-audio dialogue control.",
    "prompt": null,
    "sourceUrl": "https://x.com/kashimu_shed/status/2086577432241582478",
    "sourceLabel": "X 原帖 · @kashimu_shed",
    "author": "かしむの小屋",
    "publishedAt": "2026-08-09T22:16:41.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-kashimu_shed-2086577432241582478.mp4",
    "posterUrl": "/posters/x/2086577432241582478.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1920×1104",
    "tags": [
      "双音频",
      "角色对白",
      "声音参考"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 7.27,
      "resolution": "1920×1104",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-sellingshovels-2086575718985515245",
    "title": "批量 UGC 创意与片尾 CTA",
    "titleEn": "Bulk UGC Creative with an End CTA",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "以 MiniMax H3 制作适合持续发布的短视频创意，重点观察 480P/720P 输出、自动音频与片尾行动号召。",
    "summaryEn": "Uses MiniMax H3 for repeatable short-form creatives, focusing on 480p and 720p output, generated audio, and a call to action at the end.",
    "prompt": null,
    "sourceUrl": "https://x.com/sellingshovels/status/2086575718985515245",
    "sourceLabel": "X 原帖 · @sellingshovels",
    "author": "kyle (reddit AI agents)",
    "publishedAt": "2026-08-09T22:09:52.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-sellingshovels-2086575718985515245.mp4",
    "posterUrl": "/posters/x/2086575718985515245.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "UGC",
      "批量内容",
      "片尾 CTA",
      "自动音频"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "commercial"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "1440×1920",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-kiyoshi_shin-2086575478127869976",
    "title": "Krea 图片驱动的 H3 作例",
    "titleEn": "Krea Image-to-H3 Demonstration",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "FL2VA",
    "summary": "先用 Krea2 制作图片，再通过 MiniMax H3 将其视频化，用作文章中展示多参考与本地视频能力的作例。",
    "summaryEn": "Creates an image with Krea2 and then animates it with MiniMax H3 as an example of the model's reference-driven and local video capabilities.",
    "prompt": null,
    "sourceUrl": "https://x.com/kiyoshi_shin/status/2086575478127869976",
    "sourceLabel": "X 原帖 · @kiyoshi_shin",
    "author": "新清士@AIコンテンツ開発者",
    "publishedAt": "2026-08-09T22:08:55.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-kiyoshi_shin-2086575478127869976.mp4",
    "posterUrl": "/posters/x/2086575478127869976.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1248×832",
    "tags": [
      "Krea2",
      "图片视频化",
      "能力演示"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 6,
      "likes": 30,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "1248×832",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-suzuna_ai_girls-2086573737571418358",
    "title": "本地日语对白与加速测试",
    "titleEn": "Local Japanese Dialogue and Speed Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在本地 H3 中测试日语对白，并把帧插值和放大计入总流程；该示例完整耗时低于 500 秒。",
    "summaryEn": "Tests Japanese dialogue with a local H3 workflow and includes frame interpolation and upscaling in the reported pipeline, which finishes in under five hundred seconds.",
    "prompt": null,
    "sourceUrl": "https://x.com/Suzuna_AI_girls/status/2086573737571418358",
    "sourceLabel": "X 原帖 · @Suzuna_AI_girls",
    "author": "鈴音🔔@Suzuna_japanese_AI_girls",
    "publishedAt": "2026-08-09T22:02:00.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-suzuna_ai_girls-2086573737571418358.mp4",
    "posterUrl": "/posters/x/2086573737571418358.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1152×2048",
    "tags": [
      "日语对白",
      "本地生成",
      "帧插值",
      "放大"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 21,
      "reposts": 10,
      "likes": 43,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 6.562,
      "resolution": "1152×2048",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-aion_2077-2086568043224899870",
    "title": "官方示例指令复现",
    "titleEn": "Official Example Instruction Reproduction",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "参考 MiniMax H3 官方 GitHub 仓库中的一条示例指令生成视频，用于验证官方公开写法的实际效果。",
    "summaryEn": "Generates a video from an example instruction in the official MiniMax H3 GitHub repository to test how the published guidance performs in practice.",
    "prompt": null,
    "sourceUrl": "https://x.com/AION_2077/status/2086568043224899870",
    "sourceLabel": "X 原帖 · @AION_2077",
    "author": "AION",
    "publishedAt": "2026-08-09T21:39:22.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-aion_2077-2086568043224899870.mp4",
    "posterUrl": "/posters/x/2086568043224899870.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1920×2160",
    "tags": [
      "官方示例",
      "复现实验",
      "GitHub"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.133,
      "resolution": "1920×2160",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-nsitnovsitnov-2086566296838148275",
    "title": "《谁知道？》本地 AI 短片",
    "titleEn": "Who Knows? A Local AI Short Film",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "以 MiniMax H3、Qwen Image 3 Pro 和 ElevenLabs 在本地组合完成一部关于不可预知运气的短片。",
    "summaryEn": "Combines MiniMax H3, Qwen Image 3 Pro, and ElevenLabs in a local workflow for a short film about luck that cannot be read in advance.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2086566296838148275",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-09T21:32:26.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-nsitnovsitnov-2086566296838148275.mp4",
    "posterUrl": "/posters/x/2086566296838148275.jpg",
    "duration": 126,
    "aspectRatio": "landscape",
    "resolution": "952×544",
    "tags": [
      "AI 短片",
      "Qwen Image",
      "ElevenLabs",
      "本地工作流"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "image",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 125.764,
      "resolution": "952×544",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-sellingshovels-2086564255063482466",
    "title": "免费音频的一体化 H3 试用",
    "titleEn": "Integrated H3 Test with Generated Audio",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "作者在现有工作流中改用 MiniMax H3，强调组件连接后即可工作，并可同步获得可用的生成音频。",
    "summaryEn": "The creator switches an existing workflow to MiniMax H3, noting that the connected components work as a whole and can also produce usable audio.",
    "prompt": null,
    "sourceUrl": "https://x.com/sellingshovels/status/2086564255063482466",
    "sourceLabel": "X 原帖 · @sellingshovels",
    "author": "kyle (reddit AI agents)",
    "publishedAt": "2026-08-09T21:24:19.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-sellingshovels-2086564255063482466.mp4",
    "posterUrl": "/posters/x/2086564255063482466.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "工作流",
      "生成音频",
      "一体化"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "ugc"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1440×1920",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-love_backstage-2086563821578281128",
    "title": "RTX 5070 的图生视频基准",
    "titleEn": "Image-to-Video Benchmark on an RTX 5070",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "FL2VA",
    "summary": "在 Windows 11、RTX 5070 12GB 和 FP8 剪枝模型上本地运行 H3；1MP、1376×768 的六秒视频约耗时 35 分钟。",
    "summaryEn": "Runs H3 locally on Windows 11 with a 12 GB RTX 5070 and a pruned FP8 model; a six-second 1 MP clip at 1376x768 takes about thirty-five minutes.",
    "prompt": null,
    "sourceUrl": "https://x.com/Love_backstage/status/2086563821578281128",
    "sourceLabel": "X 原帖 · @Love_backstage",
    "author": "Mari.co | A guide to the story of backstage.love💕",
    "publishedAt": "2026-08-09T21:22:36.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-love_backstage-2086563821578281128.mp4",
    "posterUrl": "/posters/x/2086563821578281128.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "RTX 5070",
      "FP8",
      "1376×768",
      "生成耗时"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 6.583,
      "resolution": "1376×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-berserkr_777-2086561153757381047",
    "title": "本地电影感视频展示",
    "titleEn": "Local Cinematic Video Demonstration",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用本地运行的 MiniMax H3 制作电影感视频，原帖主要强调开放模型在个人设备上实现高质量影像的可行性。",
    "summaryEn": "Uses a locally running MiniMax H3 to create a cinematic video, emphasizing that high-quality imagery is now practical on personal hardware.",
    "prompt": null,
    "sourceUrl": "https://x.com/Berserkr_777/status/2086561153757381047",
    "sourceLabel": "X 原帖 · @Berserkr_777",
    "author": "Kadeka",
    "publishedAt": "2026-08-09T21:12:00.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-berserkr_777-2086561153757381047.mp4",
    "posterUrl": "/posters/x/2086561153757381047.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "2542×1440",
    "tags": [
      "本地生成",
      "电影感",
      "开放模型"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 3,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 14.419,
      "resolution": "2542×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-alone1moon-2086558140158583082",
    "title": "本地角色短片测试",
    "titleEn": "Local Character Clip Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用本地 MiniMax H3 制作田中角色短片，作为持续角色内容的一次轻量测试。",
    "summaryEn": "Uses a local MiniMax H3 setup for a short clip featuring the recurring Tanaka character as a lightweight character-content test.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2086558140158583082",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-09T21:00:01.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-alone1moon-2086558140158583082.mp4",
    "posterUrl": "/posters/x/2086558140158583082.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "896×1184",
    "tags": [
      "本地生成",
      "角色短片"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 4,
      "reposts": 6,
      "likes": 89,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "896×1184",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-caroyan_moon-2086558073200771128",
    "title": "Codex 协助的 H3 本地流程",
    "titleEn": "Codex-Assisted Local H3 Workflow",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "测试 MiniMax H3 本地运行，并让 Codex 在任务结束后处理关机；原帖同时记录了任务过载导致冻结的情况。",
    "summaryEn": "Tests a local MiniMax H3 workflow with Codex handling shutdown after completion, while also reporting a freeze caused by overloading the job.",
    "prompt": null,
    "sourceUrl": "https://x.com/Caroyan_moon/status/2086558073200771128",
    "sourceLabel": "X 原帖 · @Caroyan_moon",
    "author": "Caroyan 🌻",
    "publishedAt": "2026-08-09T20:59:45.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-caroyan_moon-2086558073200771128.mp4",
    "posterUrl": "/posters/x/2086558073200771128.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "Codex",
      "本地自动化",
      "稳定性"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.033,
      "resolution": "720×720",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-daexchef-2086554329167565029",
    "title": "RTX 5060 Ti 两小时渲染",
    "titleEn": "Two-Hour Render on an RTX 5060 Ti",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 RTX 5060 Ti 上运行 MiniMax H3，作者记录单次视频渲染约耗时两小时。",
    "summaryEn": "Runs MiniMax H3 on an RTX 5060 Ti, with the creator reporting an approximately two-hour render time for the video.",
    "prompt": null,
    "sourceUrl": "https://x.com/DaExChef/status/2086554329167565029",
    "sourceLabel": "X 原帖 · @DaExChef",
    "author": "Adam B",
    "publishedAt": "2026-08-09T20:44:53.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-daexchef-2086554329167565029.mp4",
    "posterUrl": "/posters/x/2086554329167565029.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "RTX 5060 Ti",
      "生成耗时",
      "本地渲染"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 95,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "1376×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-feeltheghostcat-2086553077235023887",
    "title": "Maestro 动作相机测试",
    "titleEn": "Maestro Action-Camera Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "通过 Maestro 与 MiniMax H3 测试动作相机风格的高速镜头表现。",
    "summaryEn": "Tests fast-moving action-camera imagery with Maestro and MiniMax H3.",
    "prompt": null,
    "sourceUrl": "https://x.com/Feeltheghostcat/status/2086553077235023887",
    "sourceLabel": "X 原帖 · @Feeltheghostcat",
    "author": "theghostcatbroadcast",
    "publishedAt": "2026-08-09T20:39:54.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-feeltheghostcat-2086553077235023887.mp4",
    "posterUrl": "/posters/x/2086553077235023887.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "Maestro",
      "动作相机",
      "高速运动"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 6.583,
      "resolution": "1344×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-nilian-2086548858079723994",
    "title": "第二组 Ideogram 与 H3 实验",
    "titleEn": "Second Ideogram and H3 Experiment",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 ComfyUI 中再次组合 Ideogram 与 MiniMax H3，制作一段科技表现主义 AI 艺术视频。",
    "summaryEn": "Runs another ComfyUI experiment that combines Ideogram and MiniMax H3 for a techspressionist AI art video.",
    "prompt": null,
    "sourceUrl": "https://x.com/nilian/status/2086548858079723994",
    "sourceLabel": "X 原帖 · @nilian",
    "author": "Nima Nilian",
    "publishedAt": "2026-08-09T20:23:08.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-nilian-2086548858079723994.mp4",
    "posterUrl": "/posters/x/2086548858079723994.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1080×608",
    "tags": [
      "ComfyUI",
      "Ideogram",
      "AI 艺术"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "bookmarks": 0,
      "views": 177,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.122,
      "resolution": "1080×608",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-cocktailpeanut-2086546223851581688",
    "title": "三维空间中的 ComfyUI 节点",
    "titleEn": "ComfyUI Nodes in Three-Dimensional Space",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用本地 MiniMax H3 想象把 ComfyUI 节点拉入三维空间并直接交互的概念场景。",
    "summaryEn": "Uses a local MiniMax H3 setup to visualize the concept of pulling ComfyUI nodes into three-dimensional space and interacting with them directly.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2086546223851581688",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-09T20:12:40.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-cocktailpeanut-2086546223851581688.mp4",
    "posterUrl": "/posters/x/2086546223851581688.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "ComfyUI",
      "三维界面",
      "概念视频"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 43,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 14.375,
      "resolution": "864×480",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-aion_2077-2086546204792656366",
    "title": "韩式水墨画动态化",
    "titleEn": "Animated Korean Ink-Wash Scene",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "FL2VA",
    "summary": "先用 GPT Image 创作韩式水墨画，再由 MiniMax H3 将静态画面转为视频。",
    "summaryEn": "Creates a Korean-style ink-wash image with GPT Image and then brings the still artwork to life with MiniMax H3.",
    "prompt": null,
    "sourceUrl": "https://x.com/AION_2077/status/2086546204792656366",
    "sourceLabel": "X 原帖 · @AION_2077",
    "author": "AION",
    "publishedAt": "2026-08-09T20:12:36.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-aion_2077-2086546204792656366.mp4",
    "posterUrl": "/posters/x/2086546204792656366.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1918×1066",
    "tags": [
      "GPT Image",
      "水墨画",
      "图片视频化"
    ],
    "category": "workflow",
    "styles": [
      "illustration",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "bookmarks": 2,
      "views": 117,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "1918×1066",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-db_bad19758-2086544544544198694",
    "title": "单次推理完整音乐视频",
    "titleEn": "Full Music Video in One Inference",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "展示一次连续推理完成整支音乐视频的案例，强调全片没有通过拼接多个片段实现。",
    "summaryEn": "Shows a full music video produced in one continuous MiniMax H3 inference, with no stitching of separate clips.",
    "prompt": null,
    "sourceUrl": "https://x.com/db_bad19758/status/2086544544544198694",
    "sourceLabel": "X 原帖 · @db_bad19758",
    "author": "THE",
    "publishedAt": "2026-08-09T20:06:00.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-db_bad19758-2086544544544198694.mp4",
    "posterUrl": "/posters/x/2086544544544198694.jpg",
    "duration": 189,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "单次推理",
      "无拼接",
      "完整 MV"
    ],
    "category": "music",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 189.333,
      "resolution": "1056×608",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-hbcoop_-2086543807445864489",
    "title": "H3 标题动画测试",
    "titleEn": "MiniMax H3 Title Animation Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用 MiniMax H3 制作标题动画，集中展示文字与动态图形的呈现。",
    "summaryEn": "Uses MiniMax H3 for title animation, focusing on the presentation of text and motion graphics.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2086543807445864489",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-08-09T20:03:04.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-hbcoop_-2086543807445864489.mp4",
    "posterUrl": "/posters/x/2086543807445864489.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "标题动画",
      "动态图形",
      "文字渲染"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 9,
      "reposts": 7,
      "likes": 59,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-ai_emeraldapple-2086542941112487981",
    "title": "太空战同提示词对比",
    "titleEn": "Same-Prompt Space Battle Comparison",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "用更复杂的太空战文字指令对比 MiniMax H3 与 Grok Imagine Video，重点观察跨镜头舰船位置与画面连续性。",
    "summaryEn": "Compares MiniMax H3 with Grok Imagine Video on a more complex space-battle instruction, focusing on ship placement and continuity across cuts.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI_EmeraldApple/status/2086542941112487981",
    "sourceLabel": "X 原帖 · @AI_EmeraldApple",
    "author": "Emerald Apple",
    "publishedAt": "2026-08-09T19:59:37.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-ai_emeraldapple-2086542941112487981.mp4",
    "posterUrl": "/posters/x/2086542941112487981.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "同提示词",
      "Grok Imagine",
      "太空战",
      "连续性"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 4,
      "reposts": 4,
      "likes": 38,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 2,
      "durationSeconds": 12.25,
      "resolution": "1920×1088",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-aicuai-2086542444926063059",
    "title": "RTX 4000 Ada 九倍加速测试",
    "titleEn": "Ninefold Speedup on an RTX 4000 Ada",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 RTX 4000 Ada 20GB 上结合量化、INT8 VAE 与四步 Turbo，将五秒 H3 视频耗时从 32 分钟降至约 3.5 分钟。",
    "summaryEn": "Combines quantization, an INT8 VAE, and four-step Turbo on a 20 GB RTX 4000 Ada, reducing a five-second H3 render from thirty-two minutes to about three and a half minutes.",
    "prompt": null,
    "sourceUrl": "https://x.com/AICUai/status/2086542444926063059",
    "sourceLabel": "X 原帖 · @AICUai",
    "author": "AICU - つくる人をつくる",
    "publishedAt": "2026-08-09T19:57:39.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-aicuai-2086542444926063059.mp4",
    "posterUrl": "/posters/x/2086542444926063059.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "640×352",
    "tags": [
      "RTX 4000 Ada",
      "量化",
      "四步 Turbo",
      "加速"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 3,
      "bookmarks": 3,
      "views": 451,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 3.041,
      "resolution": "640×352",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-squeakalgaib-2086531027347144916",
    "title": "谢尔顿与佩妮的纯提示词生成",
    "titleEn": "Sheldon and Penny from Prompts Alone",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "以谢尔顿和佩妮作为文字指令中的角色，用 MiniMax H3 直接完成角色场景。",
    "summaryEn": "Uses Sheldon and Penny as characters defined in text instructions and generates the resulting scene directly with MiniMax H3.",
    "prompt": "Create a 12-second live-action sitcom-style scene in 16:9 aspect ratio featuring Sheldon Cooper and Penny from The Big Bang Theory. The scene is a very silly, funny, character-accurate conversation about artificial intelligence that ends with both characters realizing in horror that they are not real people, but prompts inside an AI-generated video. The tone should feel exactly like a classic Big Bang Theory hallway scene: fast, witty, natural, awkward, and playful, with precise comic timing and highly exaggerated, animated facial expressions. The pacing must feel natural and unforced, like an actual exchange from the original sitcom series.\n\nIMPORTANT DIALOGUE RULE: Only the dialogue written in double quotes below may be spoken. There must be no extra dialogue, no improvisation, no narration, no voice-over, and no additional spoken words. Each line must be spoken exactly once, by the specific character assigned to it. Sheldon Cooper says: “AI language models predict the next most statistically probable word.” Penny says: “So what am I about to say?” Sheldon Cooper says: “Something intellectually disappointing.” Penny says: “Wait… somebody already wrote this sentence.” Sheldon Cooper says: “We’re scripted?” Penny says: “Worse. Prompted.” No other spoken words are allowed.\n\nLOCATION AND SCENE SETTING: Set the entire scene in the iconic apartment hallway and elevator lobby area from The Big Bang Theory, outside Penny’s apartment 4B. The elevator is visibly broken as usual, with the dark empty elevator shaft area and caution-taped or damaged elevator look recognizable in the softly blurred background. Penny’s apartment door is already open at the start of the scene. Penny stands immediately inside the doorway of apartment 4B, facing outward into the hallway. Sheldon Cooper stands immediately outside the doorway, positioned very close to Penny so Sheldon Cooper and Penny can both fit naturally inside an extremely tight two-person close-up without requiring a wide composition. The visual style, set layout, hallway walls, warm apartment lighting, and sitcom staging should feel extremely faithful to the original series.\n\nCRITICAL CAMERA FRAMING RESTRICTION: Use ONLY extremely tight close-ups, tight two-person close-ups, and individual close-ups. Absolutely no medium-wide shots, wide shots, waist-up two-shots, full-body shots, distant framing, establishing wide shots, or camera pullbacks at any moment. Sheldon Cooper’s face and Penny’s face must remain large, sharp, high-definition, and fully recognizable throughout the entire 12 seconds.\n\nWhenever Sheldon Cooper and Penny are visible together, use an EXTREMELY TIGHT TWO-PERSON CLOSE-UP: frame both characters approximately from the upper chest or shoulders upward, with both complete faces occupying most of the 16:9 frame. Keep Sheldon Cooper and Penny close together near the apartment doorway and on nearly the same focal plane. Both heads, both complete faces, eyes, mouths, and facial expressions must remain fully inside the frame and sharply resolved. Allow only a small amount of hallway and doorway background around them.\n\nDo not zoom out or pull back to accommodate both characters. Instead, physically compose Sheldon Cooper and Penny closer together inside the doorway area so both faces remain large. This tight two-person close-up requirement applies especially to the OPENING SHOT and the FINAL SHOT. The final second must NOT become wider under any circumstances.\n\nCAMERA AND FRAMING: Begin immediately with an extremely tight two-person close-up showing Sheldon Cooper and Penny together from approximately the upper chest or shoulders upward. Sheldon Cooper stands immediately outside apartment 4B while Penny stands immediately inside the already-open doorway. Both faces occupy a large portion of the frame and remain simultaneously sharp.\n\nAfter the opening, use clean sitcom-style cuts into individual close-ups and tight medium close-ups that never frame below the upper chest. Keep faces extremely large and detailed. Camera movement should remain minimal and controlled, similar to polished sitcom coverage, with subtle slow push-ins of very small amplitude for comedic emphasis. Never create additional distance from the subjects.\n\nFor the ending, return to an extremely tight two-person close-up with both Sheldon Cooper and Penny’s complete faces filling most of the frame. Do NOT pull out, zoom out, widen the lens, reveal their full bodies, or show additional hallway space during the final reaction.\n\nLIGHTING AND VISUAL STYLE: Use warm, bright, natural sitcom interior lighting. Penny’s apartment interior should glow warmly behind Penny, while the hallway is evenly lit in the familiar Big Bang Theory style. Both faces must be extremely well-lit, clearly visible, and flattering, with natural skin tones, crisp facial detail, clear eyes, and strong separation from the background. The overall visual style should feel like a professionally shot network sitcom scene, not cinematic noir, not over-stylized, and not dark. Use shallow-to-moderate depth of field so the environment remains recognizable while Sheldon Cooper and Penny’s faces stay exceptionally sharp.\n\nCHARACTER APPEARANCE AND MANNERISMS: Sheldon Cooper must look and behave exactly like Sheldon in the series: upright posture, slightly stiff body language, precise diction, intellectual smugness, mildly condescending confidence, quick blinking, intense focus, tiny head movements, tightly controlled hand gestures, and an overly serious tone even when saying ridiculous things. Sheldon Cooper should wear a typical Sheldon outfit, such as a bright geeky science-themed T-shirt layered over a long-sleeve shirt.\n\nPenny must look and behave exactly like Penny in the series: relaxed posture, highly expressive face, grounded and casual energy, playful sarcasm, fast reaction timing, expressive eyebrows, animated eye movements, slight head tilts, knowing half-smiles, and a naturally conversational speaking style. Penny should wear a casual everyday apartment outfit appropriate to the series, such as a fitted top and jeans or a soft casual blouse.\n\nEXAGGERATED FACIAL PERFORMANCE: Throughout the entire 12-second clip, both characters should use more animated, heightened sitcom expressions than before while remaining believable and human. Every comedic beat should register clearly through the eyes, eyebrows, mouth, cheeks, head position, and micro-reactions.\n\nSheldon Cooper’s smugness should be visibly exaggerated: raised chin, narrowed eyes, tiny superior smile, lifted eyebrow, precise mouth movements, and self-satisfied pauses. When Sheldon Cooper realizes something is wrong, Sheldon Cooper’s transformation must be dramatic and instantly readable: smug smile disappears, eyebrows lift sharply, eyes widen, mouth slightly opens, head freezes, then Sheldon Cooper slowly turns toward the camera with unmistakable existential dread.\n\nPenny’s sarcasm should be equally animated: larger eyebrow raises, stronger skeptical looks, playful eye rolls, amused half-smiles, slight head tilts, and quick expressive changes. When Penny realizes that Penny’s sentence has already been written, Penny’s expression must visibly progress from casual confidence to confusion, then suspicion, then sudden wide-eyed realization. Penny’s final “Worse. Prompted.” reaction should combine alarm, disbelief, and deadpan comedic certainty.\n\nPERFORMANCE AND COMEDIC RHYTHM: Sheldon Cooper begins the scene in full lecture mode, speaking as though giving a factual correction to Penny. Penny listens with exaggerated amused skepticism. Sheldon Cooper delivers the first line with smug certainty and clipped precision. Penny responds with casual, direct, slightly teasing confidence. Sheldon Cooper replies with a dry, superior, mildly insulting punchline.\n\nPenny’s next line is the turning point: Penny suddenly becomes aware of the artificiality of the scene, pauses, looks toward the camera, and says the line with rapidly growing confusion. Sheldon Cooper’s expression visibly collapses from smug confidence into dawning existential panic. Sheldon Cooper then slowly looks toward the camera and says “We’re scripted?” with exaggerated worried disbelief. Penny delivers “Worse. Prompted.” with a horrified but matter-of-fact realization. Both Sheldon Cooper and Penny end by staring directly toward the camera in exaggerated shared existential horror.\n\nTIMING, SHOTS, AND EXACT DIALOGUE ASSIGNMENT:\n\n0.0 to 2.2 seconds: Begin on an EXTREMELY TIGHT TWO-PERSON CLOSE-UP outside apartment 4B. Sheldon Cooper and Penny are framed together only from approximately the upper chest or shoulders upward. Both complete faces are very large in frame, sharply detailed, and positioned close together on nearly the same focal plane. Penny remains just inside the already-open doorway while Sheldon Cooper stands immediately outside.\n\nUse no establishing wide shot. The apartment doorway and a very small portion of the familiar hallway provide enough environmental context behind their large faces.\n\nSheldon Cooper faces Penny with an exaggeratedly serious, self-important expression and says exactly once: “AI language models predict the next most statistically probable word.”\n\nPenny remains silent throughout Sheldon Cooper’s line. Penny reacts with animated skeptical eyebrows, a small amused smile, and a slight head tilt.\n\n2.2 to 3.4 seconds: Cut to an extremely tight close-up of Penny from approximately shoulders upward. Penny’s face occupies most of the frame. Penny tilts Penny’s head, raises Penny’s eyebrows with exaggerated playful skepticism, gives Sheldon Cooper an amused look, and says exactly once: “So what am I about to say?”\n\nSheldon Cooper remains silent during Penny’s line.\n\n3.4 to 5.3 seconds: Cut to an extremely tight close-up of Sheldon Cooper. Sheldon Cooper’s face fills most of the frame. Sheldon Cooper lifts one eyebrow, produces an exaggerated tiny smug smile, slightly raises Sheldon Cooper’s chin, and delivers the line with precise intellectual superiority.\n\nSheldon Cooper says exactly once: “Something intellectually disappointing.”\n\nImmediately after Sheldon Cooper finishes the line, add a brief classic sitcom audience laugh. Sheldon Cooper maintains the exaggerated smug expression for a fraction of a second to let the punchline land.\n\n5.3 to 7.8 seconds: Cut back to an extremely tight close-up of Penny. Penny begins with a sarcastic amused expression, then Penny’s expression changes dramatically and visibly in stages. Penny’s smile fades. Penny’s eyebrows draw together. Penny’s eyes shift as if thinking. Penny suddenly stops moving for a beat, then Penny’s eyes widen.\n\nPenny slowly looks directly into the camera lens as if Penny has suddenly become aware of the artificial nature of the scene.\n\nPenny says exactly once: “Wait… somebody already wrote this sentence.”\n\nPenny’s delivery begins puzzled and ends genuinely disturbed. Add a smaller sitcom audience laugh that fades quickly into increasingly awkward silence.\n\n7.8 to 9.7 seconds: Cut to an extremely tight close-up of Sheldon Cooper. Sheldon Cooper’s entire face remains sharply resolved and large in frame. Sheldon Cooper’s smug expression disappears completely. Sheldon Cooper’s eyebrows rise dramatically, Sheldon Cooper’s eyes widen, Sheldon Cooper’s lips part slightly, and Sheldon Cooper freezes for a beat.\n\nSheldon Cooper slowly shifts Sheldon Cooper’s gaze directly toward the camera with exaggerated dawning existential dread.\n\nSheldon Cooper says exactly once: “We’re scripted?”\n\nSheldon Cooper’s tone should be anxious, confused, and deeply disturbed.\n\n9.7 to 11.0 seconds: Cut to an extremely tight close-up of Penny. Penny keeps looking directly toward the camera. Penny’s eyes are wide, Penny’s eyebrows remain raised, and Penny’s face combines horrified realization with Penny’s characteristic dry comedic delivery.\n\nPenny says exactly once: “Worse. Prompted.”\n\nPenny delivers “Worse” with a short deadpan pause before “Prompted,” allowing the final word to land clearly.\n\n11.0 to 12.0 seconds: Cut to an EXTREMELY TIGHT TWO-PERSON CLOSE-UP. Absolutely DO NOT widen the framing for this ending. Sheldon Cooper and Penny are positioned very close together near the doorway, framed only from approximately the shoulders upward. Both complete faces occupy most of the 16:9 frame simultaneously and remain exceptionally sharp, detailed, undistorted, and clearly recognizable.\n\nSheldon Cooper and Penny both stare directly toward the camera lens in exaggerated shared existential horror. Sheldon Cooper’s eyes are extremely wide, eyebrows raised, jaw slightly dropped. Penny’s eyes are wide, Penny’s mouth slightly opens, and Penny exchanges one extremely quick horrified side-glance toward Sheldon Cooper before looking back into the camera.\n\nNeither Sheldon Cooper nor Penny says anything further.\n\nAdd one final short sitcom audience laugh that initially reacts to “Worse. Prompted.” and then abruptly fades into awkward silence while Sheldon Cooper and Penny remain frozen in the extremely tight two-person close-up.\n\nIMPORTANT ENDING CAMERA RULE: During 11.0–12.0 seconds, maintain the exact tight facial scale. No camera pullback. No zoom out. No transition to a medium shot. No waist-up framing. No reveal of additional hallway space. No wider composition whatsoever. End directly on the large, sharply rendered faces of Sheldon Cooper and Penny.\n\nAUDIO DESIGN: No background music. No narration. No voice-over. No on-screen text. No watermark. No AI parody label. Use only the exact dialogue lines listed above, plus natural sitcom set ambience and classic audience laugh cues. Include subtle hallway room tone, faint apartment ambience, light clothing rustle, tiny foot shifts, and clean studio-style dialogue recording.\n\nThe audience laughter should feel like a classic Big Bang Theory studio-audience reaction: brief, well-timed, and placed after the strongest joke beats, especially after “Something intellectually disappointing,” after “somebody already wrote this sentence,” and a final shorter laugh sting after “Worse. Prompted.” before the awkward silence.\n\nAll dialogue must remain crystal clear and perfectly lip-synchronized. Sheldon Cooper’s voice must only come from Sheldon Cooper. Penny’s voice must only come from Penny. Never blend, swap, duplicate, overlap, or garble the voices.\n\nFINAL RESTRICTIONS: No wide shots. No medium-wide shots. No waist-up two-shots. No full-body framing. No distant faces. No opening establishing wide shot. No wider ending shot. No camera pullback. No zoom-out. No distorted faces. No extra dialogue. No repeated dialogue. No ad-libs. No narration. No subtitles. No captions. No on-screen text of any kind. No watermark. No AI parody disclosure. No music. No other characters entering the frame. No door knocking. The door is already open at the beginning.\n\nMaintain extremely tight facial framing, high-definition facial clarity, accurate speaker assignment, exaggerated animated expressions, natural sitcom pacing, faithful Sheldon Cooper and Penny mannerisms, and believable Big Bang Theory comedic rhythm throughout the entire 12-second clip.",
    "promptSourceUrl": "https://x.com/SqueakAlGaib/status/2086531032002785341",
    "sourceUrl": "https://x.com/SqueakAlGaib/status/2086531027347144916",
    "sourceLabel": "X 原帖 · @SqueakAlGaib",
    "author": "Squeak Al-Gaib",
    "publishedAt": "2026-08-09T19:12:17.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-squeakalgaib-2086531027347144916.mp4",
    "posterUrl": "/posters/x/2086531027347144916.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "纯提示词",
      "双角色",
      "人物生成"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 12.25,
      "resolution": "1376×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-redbickey-2086530803358695635",
    "title": "They 代词控制测试",
    "titleEn": "They Pronoun Control Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "以外部文章中的写法为基础，在第二组 MiniMax H3 测试中使用“They”作为角色指代，观察多人控制效果。",
    "summaryEn": "Builds on an external article and uses the pronoun “They” in a second MiniMax H3 test to observe multi-character control.",
    "prompt": null,
    "sourceUrl": "https://x.com/redbickey/status/2086530803358695635",
    "sourceLabel": "X 原帖 · @redbickey",
    "author": "Redbickey",
    "publishedAt": "2026-08-09T19:11:24.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-redbickey-2086530803358695635.mp4",
    "posterUrl": "/posters/x/2086530803358695635.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "代词控制",
      "多人角色",
      "测试 2"
    ],
    "category": "comparison",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 142,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1056×608",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-redbickey-2086530599406522745",
    "title": "She 代词控制测试",
    "titleEn": "She Pronoun Control Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "以外部文章中的写法为基础，在第一组 MiniMax H3 测试中使用“She”作为角色指代，观察单角色控制效果。",
    "summaryEn": "Builds on an external article and uses the pronoun “She” in a first MiniMax H3 test to observe single-character control.",
    "prompt": null,
    "sourceUrl": "https://x.com/redbickey/status/2086530599406522745",
    "sourceLabel": "X 原帖 · @redbickey",
    "author": "Redbickey",
    "publishedAt": "2026-08-09T19:10:35.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-redbickey-2086530599406522745.mp4",
    "posterUrl": "/posters/x/2086530599406522745.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "代词控制",
      "单角色",
      "测试 1"
    ],
    "category": "comparison",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 143,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1056×608",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-shaajibaa-2086530367021010973",
    "title": "共享指令的 T2V 试做",
    "titleEn": "Shared-Instruction T2V Trial",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "在 ComfyUI 中使用他人公开分享的文字指令，进行 MiniMax H3 文生视频试做。",
    "summaryEn": "Uses a publicly shared text instruction in ComfyUI for a MiniMax H3 text-to-video trial.",
    "prompt": null,
    "sourceUrl": "https://x.com/shaajibaa/status/2086530367021010973",
    "sourceLabel": "X 原帖 · @shaajibaa",
    "author": "shaajibaa",
    "publishedAt": "2026-08-09T19:09:40.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-shaajibaa-2086530367021010973.mp4",
    "posterUrl": "/posters/x/2086530367021010973.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "T2V",
      "共享指令",
      "ComfyUI"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 17.208,
      "resolution": "1216×672",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-alone1moon-2086528647251607872",
    "title": "三段海岸视频拼接与放大",
    "titleEn": "Three-Part Beach Sequence with Upscaling",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "将三段各 15 秒的本地 H3 海岸视频拼接，加入音乐后再用 RTX Video Super Resolution 做两倍放大。",
    "summaryEn": "Stitches three fifteen-second local H3 beach clips, adds music, and then applies a two-times upscale with RTX Video Super Resolution.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2086528647251607872",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-09T19:02:49.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-alone1moon-2086528647251607872.mp4",
    "posterUrl": "/posters/x/2086528647251607872.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "2688×1536",
    "tags": [
      "三段拼接",
      "海岸",
      "RTX VSR",
      "音乐"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "video",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 31,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 44.583,
      "resolution": "2688×1536",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-yourplugai-2086528067019698388",
    "title": "霓虹山路悬浮摩托竞速",
    "titleEn": "Neon Mountain Hover-Bike Race",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "FL2VA",
    "summary": "从给定首帧生成 15 秒双人悬浮摩托竞速，文字要求锁定两名车手与两辆车，并维持高速动作连续性。",
    "summaryEn": "Generates a fifteen-second two-rider hover-bike race from a supplied first frame, with text constraints that lock the two riders and two bikes while maintaining high-speed continuity.",
    "prompt": "Cinematic Anime Video Scene\n\nGenerate a 15-second horizontal 16:9 original high-speed hover-bike racing anime video from the provided first frame.\n\nCRITICAL ENTITY LOCK:\nThere must be exactly 2 racers and 2 bikes in the entire video: RENJI on VALKYRIE-01 (cyan/black drift bike) and ELENA on AERO-X (crimson/white draft bike).\nDo not add extra racers, drone support vehicles, spectators, or traffic.\nMaintain total visual consistency for both bikes, helmet visors, suit patterns, repulsor spark colors, and bike liveries throughout the sequence.\n\nEntity identity:\nVALKYRIE-01: Matte-black and cyan angular hover-bike, exposed repulsor pads, lateral drift brakes, blue plasma exhaust trails, ridden by Renji (cyan trim suit).\nAERO-X: Pearl-white and neon-crimson aerodynamic hover-bike, enclosed canopy, crimson energy draft aura, white-hot central booster, ridden by Elena (crimson/gold visor suit).\n\nVideo style:\nHigh-budget modern sports anime, sakuga-level velocity animation, crisp line art, vibrant neon lighting contrast, high-speed camera tracking, hyper-realistic friction and energy particle effects. Set on a wet downhill mountain pass at dawn.\n\nCamera and pacing:\nContinuous forward velocity, zero slow-motion interruptions:\n0.0s - 3.0s: High-speed rear-tracking shot diving into the first downhill hairpin curve; instant drift initiation.\n3.0s - 7.5s: Tight side-parallel tracking shot as bikes navigate rock debris and trade positions through S-curves.\n7.5s - 11.5s: Close camera lock on the draft-slingshot maneuver; high-energy particle displacement as booster ignition occurs.\n11.5s - 15.0s: Low-angle front-facing camera lock on the final sprint to the finish line bridge, ending on a hyper-speed photo-finish freeze.\n\nAction timing:\n0.0s - 1.5s:\nSequence begins at speed. VALKYRIE-01 leads downhill; AERO-X locks onto its rear bumper. Anti-gravity repulsors spray road water and blue sparks into the frame.\n\n1.5s - 4.0s:\nFirst sharp hairpin. VALKYRIE-01 deploys lateral drift airbrakes with a burst of blue thruster fire, sliding sideways at 300 km/h. AERO-X stays glued inside its slipstream aura.\n\n4.0s - 7.0s:\nMountain debris hazard. VALKYRIE-01 hops over a boulder using a repulsor burst. AERO-X ducks under it, scraping the neon magenta guardrail in a cloud of friction sparks.\n\n7.0s - 10.0s:\nS-Curve exchange. Bikes lean side-by-side; their repulsor fields collide, creating a bright electrical shockwave. ELENA pulls the overdrive lever; AERO-X's rear fins extend.\n\n10.0s - 13.0s:\nSlingshot maneuver. AERO-X bursts out of VALKYRIE-01's draft, igniting its central white plasma booster. Both bikes roar down the final straightaway side-by-side.\n\n13.0s - 15.0s:\nFinal sprint toward the finish light gate. Water sprays violently behind them. Both nose cones cross the finish line simultaneously in a flash of light. Final freeze frame.\n\nMotion quality:\nFluid 2D animation, extreme speed-line integration, stable bike geometry, flawless vehicle reflection rendering, zero limb or body clipping, high-frame-rate kinetic realism.\n\nEnvironment:\nWet mountain pass asphalt, sheer cliff walls, neon cyan and magenta guardrail lights, early dawn sky with pink/purple clouds, water spray, floating spark particles.\n\nFinal output:\n15 seconds, horizontal 16:9, original high-budget sports racing anime, exactly 2 racers, relentless kinetic pacing, dynamic cinematography, no subtitles, no watermarks, no logos.",
    "sourceUrl": "https://x.com/yourPlugAI/status/2086528067019698388",
    "sourceLabel": "X 原帖 · @yourPlugAI",
    "author": "Gilbert Odera | Your AI Plug",
    "publishedAt": "2026-08-09T19:00:31.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-yourplugai-2086528067019698388.mp4",
    "posterUrl": "/posters/x/2086528067019698388.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "首帧",
      "悬浮摩托",
      "实体锁定",
      "高速动作"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 20,
      "reposts": 32,
      "likes": 50,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.104,
      "resolution": "1280×720",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-alexshevpm-2086519480604041564",
    "title": "遮罩驱动的视频局部修复",
    "titleEn": "Mask-Guided Local Video Repair",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Ref2VA",
    "summary": "把 H3 的价值从一次生成延伸到后期编辑：遮罩错误区域、保留其余画面和动作，再迭代修复。",
    "summaryEn": "Extends H3 beyond one-shot generation into post-production by masking an incorrect region, preserving the rest of the frame and motion, and iterating on the repair.",
    "prompt": null,
    "sourceUrl": "https://x.com/AlexshevPm/status/2086519480604041564",
    "sourceLabel": "X 原帖 · @AlexshevPm",
    "author": "Alex Shev",
    "publishedAt": "2026-08-09T18:26:24.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-alexshevpm-2086519480604041564.mp4",
    "posterUrl": "/posters/x/2086519480604041564.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "局部修复",
      "遮罩",
      "动作保持",
      "迭代编辑"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "video"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 4.966,
      "resolution": "720×720",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-mtarakawa-2086512998995366049",
    "title": "复杂脚本与日语理解测试",
    "titleEn": "Complex Script and Japanese-Language Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "以较复杂的日语脚本测试 MiniMax H3 的理解和执行，并记录 768P 生成的成本感受。",
    "summaryEn": "Tests how MiniMax H3 interprets and executes a relatively complex Japanese script while commenting on the cost of 768p generation.",
    "prompt": null,
    "sourceUrl": "https://x.com/mtArakawa/status/2086512998995366049",
    "sourceLabel": "X 原帖 · @mtArakawa",
    "author": "元荒川ちゃん❤",
    "publishedAt": "2026-08-09T18:00:39.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-mtarakawa-2086512998995366049.mp4",
    "posterUrl": "/posters/x/2086512998995366049.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "复杂脚本",
      "日语理解",
      "768P"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 11,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 12.25,
      "resolution": "1024×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-thedarkwebdev-2086512471687217348",
    "title": "30 秒无拼接本地生成",
    "titleEn": "Thirty-Second Local Generation Without Stitching",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用 MiniMax H3 开放权重一次生成 30 秒视频，明确说明没有做片段拼接。",
    "summaryEn": "Uses the MiniMax H3 open weights to generate a thirty-second video in one run, explicitly without clip stitching.",
    "prompt": null,
    "sourceUrl": "https://x.com/TheDarkWebDev/status/2086512471687217348",
    "sourceLabel": "X 原帖 · @TheDarkWebDev",
    "author": "The Dark Developer",
    "publishedAt": "2026-08-09T17:58:33.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-thedarkwebdev-2086512471687217348.mp4",
    "posterUrl": "/posters/x/2086512471687217348.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "30 秒",
      "无拼接",
      "开放权重"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 13,
      "bookmarks": 0,
      "views": 0,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 29.918,
      "resolution": "1024×576",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-nsitnovsitnov-2086511868219109596",
    "title": "150 秒本地 AI 短片",
    "titleEn": "A 150-Second Local AI Short Film",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "结合 MiniMax H3、Qwen Image 3 Pro 与 ElevenLabs，在本地工作流中完成一部 150 秒短片。",
    "summaryEn": "Combines MiniMax H3, Qwen Image 3 Pro, and ElevenLabs in a local workflow to produce a 150-second short film.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2086511868219109596",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-09T17:56:09.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-nsitnovsitnov-2086511868219109596.mp4",
    "posterUrl": "/posters/x/2086511868219109596.jpg",
    "duration": 161,
    "aspectRatio": "landscape",
    "resolution": "1120×640",
    "tags": [
      "150 秒",
      "AI 短片",
      "Qwen Image",
      "ElevenLabs"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "image",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 158,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 160.501,
      "resolution": "1120×640",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-mrdasonx-2086510806884364742",
    "title": "隧道摩托高速运动测试",
    "titleEn": "High-Speed Tunnel Bike Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "以隧道摩托场景展示 MiniMax H3 的平滑运动、灯光变化、镜头衔接与车辆物理一致性。",
    "summaryEn": "Uses a tunnel-bike scene to demonstrate smooth motion, changing light, camera transitions, and consistent vehicle physics in MiniMax H3.",
    "prompt": null,
    "sourceUrl": "https://x.com/MrDasOnX/status/2086510806884364742",
    "sourceLabel": "X 原帖 · @MrDasOnX",
    "author": "Mr Das",
    "publishedAt": "2026-08-09T17:51:56.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-mrdasonx-2086510806884364742.mp4",
    "posterUrl": "/posters/x/2086510806884364742.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "摩托",
      "高速运动",
      "灯光",
      "物理一致性"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 6,
      "bookmarks": 1,
      "views": 255,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-alexfredo87-2086509099550736632",
    "title": "本地四步 Turbo 科幻 MV",
    "titleEn": "Local Four-Step Turbo Sci-Fi Music Video",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 480P 下使用四步 Turbo LoRA 本地测试 MiniMax H3 音乐视频，并探索一键生成完整 MV 的自动化流程。",
    "summaryEn": "Tests a local MiniMax H3 music video at 480p with a four-step Turbo LoRA while exploring a one-click workflow for full music-video generation.",
    "prompt": null,
    "sourceUrl": "https://x.com/alexfredo87/status/2086509099550736632",
    "sourceLabel": "X 原帖 · @alexfredo87",
    "author": "Alex",
    "publishedAt": "2026-08-09T17:45:09.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-alexfredo87-2086509099550736632.mp4",
    "posterUrl": "/posters/x/2086509099550736632.jpg",
    "duration": 256,
    "aspectRatio": "landscape",
    "resolution": "1920×1068",
    "tags": [
      "四步 Turbo",
      "480P",
      "本地生成",
      "一键工作流"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 3,
      "reposts": 5,
      "likes": 41,
      "bookmarks": 11,
      "views": 1713,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 256.426,
      "resolution": "1920×1068",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-nababuddin2-2086508165420781940",
    "title": "丛林惊慌场景同题对比",
    "titleEn": "Same-Prompt Jungle Panic Comparison",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "以同一文字指令和同一角色，对比 Seedance 2.5 与 Hailuo MiniMax H3 在丛林惊慌场景中的动作表现。",
    "summaryEn": "Compares Seedance 2.5 and Hailuo MiniMax H3 on a jungle-panic scene using the same text instruction and character.",
    "prompt": null,
    "sourceUrl": "https://x.com/NababUddin2/status/2086508165420781940",
    "sourceLabel": "X 原帖 · @NababUddin2",
    "author": "Nabab Uddin",
    "publishedAt": "2026-08-09T17:41:26.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-nababuddin2-2086508165420781940.mp4",
    "posterUrl": "/posters/x/2086508165420781940.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "4096×4096",
    "tags": [
      "同提示词",
      "Seedance 2.5",
      "丛林",
      "动作对比"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 1,
      "views": 150,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 14.603,
      "resolution": "4096×4096",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-thedarkwebdev-2086503597966045379",
    "title": "竖屏 30 秒无拼接基准",
    "titleEn": "Thirty-Second Portrait Benchmark Without Stitching",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "一次生成 30 秒、576×1024 的竖屏 H3 视频，不做片段拼接；作者公布耗时 6 分 44 秒。",
    "summaryEn": "Generates a thirty-second 576x1024 portrait H3 video without stitching and reports a runtime of six minutes and forty-four seconds.",
    "prompt": null,
    "sourceUrl": "https://x.com/TheDarkWebDev/status/2086503597966045379",
    "sourceLabel": "X 原帖 · @TheDarkWebDev",
    "author": "The Dark Developer",
    "publishedAt": "2026-08-09T17:23:17.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-thedarkwebdev-2086503597966045379.mp4",
    "posterUrl": "/posters/x/2086503597966045379.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "30 秒",
      "576×1024",
      "无拼接",
      "生成耗时"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 21,
      "bookmarks": 3,
      "views": 1949,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 29.875,
      "resolution": "576×1024",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-teksedge-2086501943703478568",
    "title": "5060 Ti 本地克隆声音对话",
    "titleEn": "Local Cloned-Voice Dialogue on a 5060 Ti",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Ref2VA",
    "summary": "在 RTX 5060 Ti 16GB 上本地生成 1344×768 视频、意大利语对白、克隆参考声音与口型同步，并公开显存和耗时。",
    "summaryEn": "Runs locally on a 16 GB RTX 5060 Ti to generate a 1344x768 clip with scripted Italian dialogue, a cloned reference voice, and lip sync, while reporting memory use and runtime.",
    "prompt": null,
    "sourceUrl": "https://x.com/TeksEdge/status/2086501943703478568",
    "sourceLabel": "X 原帖 · @TeksEdge",
    "author": "David Hendrickson",
    "publishedAt": "2026-08-09T17:16:43.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-teksedge-2086501943703478568.mp4",
    "posterUrl": "/posters/x/2086501943703478568.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "声音克隆",
      "口型同步",
      "RTX 5060 Ti",
      "意大利语"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 6,
      "reposts": 3,
      "likes": 67,
      "bookmarks": 73,
      "views": 4009,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.3,
      "resolution": "1920×1080",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-agakuossan-2086500748943409435",
    "title": "H3 动作指令偏差趣味测试",
    "titleEn": "Playful H3 Action-Control Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "作者用 MiniMax H3 做轻量动作实验，并直接吐槽成片没有按预期落到指定位置。",
    "summaryEn": "The creator runs a lightweight MiniMax H3 action experiment and openly jokes that the result did not land in the requested location.",
    "prompt": null,
    "sourceUrl": "https://x.com/AgakuOssan/status/2086500748943409435",
    "sourceLabel": "X 原帖 · @AgakuOssan",
    "author": "あがくおっさん",
    "publishedAt": "2026-08-09T17:11:58.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-agakuossan-2086500748943409435.mp4",
    "posterUrl": "/posters/x/2086500748943409435.jpg",
    "duration": 3,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "动作控制",
      "偏差观察",
      "趣味测试"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 116,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 3.041,
      "resolution": "576×736",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-toraaiuser2-2086498723664994496",
    "title": "AI 辅助复杂指令编写",
    "titleEn": "AI-Assisted Complex Instruction Writing",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在学习 MiniMax H3 视频生成时，作者因手写指令过于复杂而改用 AI 协助组织输入。",
    "summaryEn": "While learning MiniMax H3 video generation, the creator relies on AI assistance because the instructions are too complex to write manually.",
    "prompt": null,
    "sourceUrl": "https://x.com/toraaiuser2/status/2086498723664994496",
    "sourceLabel": "X 原帖 · @toraaiuser2",
    "author": "AIossansan",
    "publishedAt": "2026-08-09T17:03:55.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-toraaiuser2-2086498723664994496.mp4",
    "posterUrl": "/posters/x/2086498723664994496.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "AI 辅助",
      "复杂指令",
      "学习记录"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 44,
      "bookmarks": 6,
      "views": 1108,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "864×480",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-alfredalfer77-2086498444340871248",
    "title": "电视节目角色宣传海报梗",
    "titleEn": "Television-Character Propaganda Poster Meme",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用 MiniMax H3 制作围绕电视节目角色的宣传海报式梗视频，原帖以模型“记住”角色的夸张说法表达幽默。",
    "summaryEn": "Uses MiniMax H3 for a propaganda-poster-style meme built around a television-show character, framed humorously as if the model remembered the character on its own.",
    "prompt": null,
    "sourceUrl": "https://x.com/AlfredAlfer77/status/2086498444340871248",
    "sourceLabel": "X 原帖 · @AlfredAlfer77",
    "author": "Emily Youcis",
    "publishedAt": "2026-08-09T17:02:49.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-alfredalfer77-2086498444340871248.mp4",
    "posterUrl": "/posters/x/2086498444340871248.jpg",
    "duration": 77,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "宣传海报",
      "电视角色",
      "梗视频"
    ],
    "category": "action-vfx",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 21,
      "reposts": 20,
      "likes": 344,
      "bookmarks": 24,
      "views": 8163,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 77.133,
      "resolution": "1280×720",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-vectro-2086497155347280222",
    "title": "Hermes 自动安装后的五帧测试",
    "titleEn": "Five-Frame Test After Hermes Setup",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "Hermes Agent 自动安装 ComfyUI 与量化版 MiniMax H3 后，在 RTX 3080 本地完成五帧测试视频并创建配套 Skill。",
    "summaryEn": "After a Hermes agent installs ComfyUI and a quantized MiniMax H3 build, the workflow produces a five-frame local test video on an RTX 3080 and creates a supporting skill.",
    "prompt": null,
    "sourceUrl": "https://x.com/vectro/status/2086497155347280222",
    "sourceLabel": "X 原帖 · @vectro",
    "author": "Vectro",
    "publishedAt": "2026-08-09T16:57:41.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-vectro-2086497155347280222.mp4",
    "posterUrl": "/posters/x/2086497155347280222.jpg",
    "duration": 1,
    "aspectRatio": "landscape",
    "resolution": "640×368",
    "tags": [
      "Hermes Agent",
      "RTX 3080",
      "量化模型",
      "自动安装"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "bookmarks": 0,
      "views": 211,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 0.916,
      "resolution": "640×368",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-o_c_t_-2086495183496184253",
    "title": "逐帧 I2I 放大与 48FPS 插帧",
    "titleEn": "Framewise I2I Upscale and 48 FPS Interpolation",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "从 H3 视频导出全部帧，逐帧做 1.5 倍 I2I 后重组，再由 24FPS 插值到 48FPS，并并列展示三个处理阶段。",
    "summaryEn": "Exports every frame from an H3 video, applies a 1.5x image-to-image pass before reassembly, then interpolates from 24 to 48 FPS and presents all three processing stages side by side.",
    "prompt": null,
    "sourceUrl": "https://x.com/O_C_T_/status/2086495183496184253",
    "sourceLabel": "X 原帖 · @O_C_T_",
    "author": "おくてぃー",
    "publishedAt": "2026-08-09T16:49:51.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-o_c_t_-2086495183496184253.mp4",
    "posterUrl": "/posters/x/2086495183496184253.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "逐帧处理",
      "I2I",
      "48FPS",
      "插帧"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "video",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 1,
      "views": 176,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 3,
      "durationSeconds": 10.125,
      "resolution": "1920×1088",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-dheepanratnam-2086493087707086919",
    "title": "1955 年架空品牌广告组",
    "titleEn": "Four Fictional Commercials from 1955",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "用 MiniMax H3 制作四支设定在 1955 年、来自虚构企业的复古广告，并保留声音作为完整观看要素。",
    "summaryEn": "Uses MiniMax H3 to create four retro commercials set in 1955 for fictional companies, with sound treated as an essential part of the presentation.",
    "prompt": null,
    "sourceUrl": "https://x.com/Dheepanratnam/status/2086493087707086919",
    "sourceLabel": "X 原帖 · @Dheepanratnam",
    "author": "Dheepan Ratnam",
    "publishedAt": "2026-08-09T16:41:31.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-dheepanratnam-2086493087707086919.mp4",
    "posterUrl": "/posters/x/2086493087707086919.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "架空广告",
      "1955 年",
      "复古",
      "有声视频"
    ],
    "category": "advertising",
    "styles": [
      "retro",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 18,
      "bookmarks": 7,
      "views": 805,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 4,
      "durationSeconds": 15.083,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-s4f3ty_marc-2086490028016824629",
    "title": "本地名人乐队表演",
    "titleEn": "Local Celebrity Band Performance",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在 RTX 4090 上本地运行 MiniMax H3，生成多位公众人物演唱 Queen 歌曲的有声趣味表演。",
    "summaryEn": "Runs MiniMax H3 locally on an RTX 4090 to create a humorous sound-on performance of several public figures singing a Queen song.",
    "prompt": null,
    "sourceUrl": "https://x.com/S4f3ty_Marc/status/2086490028016824629",
    "sourceLabel": "X 原帖 · @S4f3ty_Marc",
    "author": "Safety Marc",
    "publishedAt": "2026-08-09T16:29:22.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-s4f3ty_marc-2086490028016824629.mp4",
    "posterUrl": "/posters/x/2086490028016824629.jpg",
    "duration": 54,
    "aspectRatio": "portrait",
    "resolution": "1080×1440",
    "tags": [
      "RTX 4090",
      "多人表演",
      "有声视频"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 29,
      "bookmarks": 13,
      "views": 1953,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 54.377,
      "resolution": "1080×1440",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-flyfront-2086488714457595912",
    "title": "夏祭刨冰 T2VA",
    "titleEn": "Summer-Festival Shaved Ice T2VA",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "以 MiniMax H3 T2VA 生成夏日祭刨冰主题视频，原帖明确标注了生成模式。",
    "summaryEn": "Uses MiniMax H3 T2VA for a summer-festival shaved-ice video, with the generation mode explicitly identified by the creator.",
    "prompt": null,
    "sourceUrl": "https://x.com/flyfront/status/2086488714457595912",
    "sourceLabel": "X 原帖 · @flyfront",
    "author": "flyfront",
    "publishedAt": "2026-08-09T16:24:09.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-flyfront-2086488714457595912.mp4",
    "posterUrl": "/posters/x/2086488714457595912.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×960",
    "tags": [
      "夏日祭",
      "刨冰",
      "T2VA"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized",
      "cinematic-realistic"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 182,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1280×960",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-kndaya0719-2086488366288421125",
    "title": "本地日语对白成片",
    "titleEn": "Local Japanese-Dialogue Video",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在本地 MiniMax H3 中测试日语理解与语音，并说明除后加背景音乐和一次转场外，其余均为 H3 输出。",
    "summaryEn": "Tests Japanese comprehension and speech in a local MiniMax H3 workflow, with only the background music and one transition added afterward.",
    "prompt": null,
    "sourceUrl": "https://x.com/kndaya0719/status/2086488366288421125",
    "sourceLabel": "X 原帖 · @kndaya0719",
    "author": "近藤さん",
    "publishedAt": "2026-08-09T16:22:46.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-kndaya0719-2086488366288421125.mp4",
    "posterUrl": "/posters/x/2086488366288421125.jpg",
    "duration": 16,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "日语",
      "语音",
      "本地生成",
      "轻量后期"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 9,
      "bookmarks": 0,
      "views": 251,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.936,
      "resolution": "1080×1080",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-danagreenai-2086485025747525977",
    "title": "Claude Code 与 Codex 情景喜剧",
    "titleEn": "Claude Code Versus Codex as a Sitcom Scene",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "把 Claude Code 与 Codex 的争论改编成《生活大爆炸》式角色短片，并由 MiniMax H3 生成。",
    "summaryEn": "Turns the Claude Code versus Codex debate into a Big Bang Theory-style character scene generated with MiniMax H3.",
    "prompt": null,
    "sourceUrl": "https://x.com/DanaGreenAI/status/2086485025747525977",
    "sourceLabel": "X 原帖 · @DanaGreenAI",
    "author": "Dana Akerman Green",
    "publishedAt": "2026-08-09T16:09:29.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-danagreenai-2086485025747525977.mp4",
    "posterUrl": "/posters/x/2086485025747525977.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "开发工具",
      "情景喜剧",
      "角色短片"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 1,
      "views": 140,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "1344×768",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-aiphotorealgirl-2086483981126803506",
    "title": "Turbo、Sage 与放大插帧组合",
    "titleEn": "Turbo, Sage, Upscaling, and Interpolation",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Unknown",
    "summary": "在同一组既定条件下使用 MiniMax H3、Larry Turbo 与 H3 Sage 生成，再追加放大和帧插值；耗时约 10 至 11 分钟。",
    "summaryEn": "Uses MiniMax H3 with Larry Turbo and H3 Sage under the same established conditions, followed by upscaling and frame interpolation, with a reported runtime of about ten to eleven minutes.",
    "prompt": null,
    "sourceUrl": "https://x.com/AiPhotorealGirl/status/2086483981126803506",
    "sourceLabel": "X 原帖 · @AiPhotorealGirl",
    "author": "AIconia | アイコニア",
    "publishedAt": "2026-08-09T16:05:20.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-aiphotorealgirl-2086483981126803506.mp4",
    "posterUrl": "/posters/x/2086483981126803506.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1344×1776",
    "tags": [
      "Larry Turbo",
      "H3 Sage",
      "放大",
      "帧插值"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 39,
      "bookmarks": 17,
      "views": 3806,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.133,
      "resolution": "1344×1776",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-yamatazen-2086483314576326809",
    "title": "INT8 ConvRot 文生视频测试",
    "titleEn": "INT8 ConvRot Text-to-Video Test",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "T2VA",
    "summary": "在 ComfyUI 中使用 MiniMax H3 剪枝 INT8 ConvRot 版本进行明确标注的 T2V 测试。",
    "summaryEn": "Runs an explicitly labeled text-to-video test in ComfyUI with a pruned INT8 ConvRot build of MiniMax H3.",
    "prompt": null,
    "sourceUrl": "https://x.com/YamataZen/status/2086483314576326809",
    "sourceLabel": "X 原帖 · @YamataZen",
    "author": "Yamata",
    "publishedAt": "2026-08-09T16:02:41.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-yamatazen-2086483314576326809.mp4",
    "posterUrl": "/posters/x/2086483314576326809.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "512×768",
    "tags": [
      "INT8",
      "ConvRot",
      "T2V",
      "ComfyUI"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 1,
      "views": 216,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "512×768",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-ou_zhen599-2086482460666134845",
    "title": "多参考动态战斗对比",
    "titleEn": "Multi-Reference Dynamic Combat Comparison",
    "model": "MiniMax H3 (creator-attributed)",
    "mode": "Ref2VA",
    "summary": "借用《ZEPHYR》的场景与文字指令测试 MiniMax H3 多参考动态战斗，并与 Grok Imagine Video 和 Seedance 2 作主观对照。",
    "summaryEn": "Uses a scene and text instruction from ZEPHYR to test MiniMax H3 on multi-reference dynamic combat, with subjective comparisons against Grok Imagine Video and Seedance 2.",
    "prompt": null,
    "sourceUrl": "https://x.com/ou_zhen599/status/2086482460666134845",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-08-09T15:59:18.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-ou_zhen599-2086482460666134845.mp4",
    "posterUrl": "/posters/x/2086482460666134845.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "多参考",
      "动态战斗",
      "Grok Imagine",
      "Seedance 2"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "bookmarks": 1,
      "views": 403,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-kawaiizice-2086608849872515465",
    "title": "父亲节三重可爱 Chibi 短片",
    "titleEn": "Triple-Cuteness Father's Day Chibi Clip",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "父亲节主题的竖屏 Chibi 短片，原帖以“三重可爱”描述内容，并明确标注 MiniMax H3 与 Hailuo AI；原帖未公开 Prompt。",
    "summaryEn": "A vertical Father's Day chibi clip described by the creator as a burst of triple cuteness, explicitly tagged MiniMax H3 and Hailuo AI. No prompt is published.",
    "prompt": null,
    "sourceUrl": "https://x.com/Kawaiizice/status/2086608849872515465",
    "sourceLabel": "X 原帖 · @Kawaiizice",
    "author": "Kawaiizice",
    "publishedAt": "2026-08-10T00:21:31.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-kawaiizice-2086608849872515465.mp4",
    "posterUrl": "/posters/x/2086608849872515465.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "Father's Day",
      "Chibi",
      "Vertical Video",
      "Hailuo AI"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 7,
      "bookmarks": 0,
      "views": 70,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.116,
      "resolution": "1440×2560",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-glitterpixely-2086599324125872626",
    "title": "宽银幕角色登场介绍",
    "titleEn": "Widescreen Character Introduction",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "一段用于介绍角色的 15 秒宽银幕短片，明确标注 MiniMax H3 与 Hailuo AI。原作者在回复区完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A 15-second widescreen character-introduction clip explicitly tagged MiniMax H3 and Hailuo AI. The creator published the complete prompt in a reply, preserved here verbatim.",
    "prompt": "subject_definitions:\n<Subject 1> is ENRA VAEL, the woman from < 📷Image1 >: the same sharp face, dark wispy bangs, long braid threaded through copper rings, spiked hair ornaments, severe expression, layered red robes, tattered soot-black panels, high metal collar, dark shoulder armor, red lamellar plates, ropes, beads, and weathered fittings. Her left arm alone remains blackened and lava-cracked and holds the fire-wreathed bladed weapon; her normal right hand holds the heavy chain and attached bronze vessel-like weight. The burning double metal ring remains a separate floating halo behind her.\n\n<Subject 2> is the colossal forge city from < 📷Image2 >: one black-basalt shelf beside a lower-center lava river, a cylindrical furnace tower left, layered bridges overhead, a lavafall and furnace opening right, distant towers in gray smoke, hanging chains, heat haze, and embers.\n\n< 📷Audio1 > is the 15.00-second stereo music track used continuously and as the exact editorial-timing source.\nsummary:\n[reference generation + audio reuse] Nine beat-matched angles present <Subject 1> inside <Subject 2> as a high-impact character music video. The empty opening shot reveals the exact name \"ENRA VAEL\" before the character appears; a furnace-light scan, ring-aperture transition, solarized accent, six-segment fire sigil, escalating heat rings, and posterized charge build toward the 00:14.120 iron-sun detonation.\n\nretention_analysis:\n\n<Subject 1> (appears in [Shot 2], [Shot 3], [Shot 4], [Shot 5], [Shot 6], [Shot 7], [Shot 8], [Shot 9]): fully_preserved - face, proportions, braid, hair ornaments, asymmetrical arms, wardrobe, armor, handed props, weighted chain, floating double ring, and expression remain consistent beneath every effect.\n<Subject 2> (appears in [Shot 1], [Shot 2], [Shot 3], [Shot 4], [Shot 5], [Shot 6], [Shot 7], [Shot 8], [Shot 9]): fully_preserved - connected geography, landmarks, materials, smoke, lava, and orange-charcoal lighting persist while viewpoint and temporary transition treatments change.\n< 📷Audio1 >: partially_copy - the full musical layer is copied in its original order and timing; only restrained diegetic chain, cloth, fire, title-weld, and impact sounds are mixed beneath it.\n\ndetailed_description:\n\nAAA anime action-RPG music video; effects alter only environment and end by 00:13.770.\n[Shot 1] Empty extreme-wide of <Subject 2>; no character, halo, or weapon. At 00:00.120 an ember seam opens. At 00:00.240, the complete exact uppercase \"ENRA VAEL\" appears simultaneously, centered, black iron with a thin white-hot rim, maximum 42% frame width. It holds through 00:01.120, collapses, and is gone by 00:01.350. No text crosses the next cut. <Audio 1> starts at 00:00.000.\n\n[Shot 2] At 00:01.430, a left-front full-body angle pedestals up as she drags the chain weight. From 00:01.880 to 00:02.650, a furnace-light scan travels boots-to-face, revealing robe, armor, chain, and ornaments with afterglow.\n[Shot 3] At 00:02.840, a near-static right-front three-quarter hero card keeps her face, halo, and weapon unobstructed. No text is visible.\n\n[Shot 4] At 00:04.470, a close-up frames her face inside the unattached ring. Molten arm pulses at 00:04.940; right hand closes on chain at 00:05.410. The ring turns front-on at 00:05.870 and expands; its black center fills frame from 00:06.170 to 00:06.300 without covering her face.\n[Shot 5] At 00:06.340, the ring aperture opens on the drop to a ground-level full-body angle. She yanks the weighted chain; ring spins independently, weapon traces a molten circle, and furnaces flash. One black-red solarized frame hits, then normal color returns.\n[Shot 6] At 00:08.240, overhead, she pivots as fire engraves a six-segment solar sigil. Segments lock at 00:08.710, 00:09.180, and 00:09.650; it flashes at 00:09.800, then lifts as ribbons into the ring by 00:10.090.\n[Shot 7] At 00:10.090, a weapon-side view tracks blade, cracked arm, and face. Pulses send one, two, then three heat rings through the background at 00:10.580, 00:11.040, and 00:11.480; her geometry stays stable.\n\n[Shot 8] At 00:11.950, a low medium-wide pushes in fast. She pulls chain taut, raises weapon at 00:12.440, turns it down at 00:12.900, and compresses an iron sun at 00:13.360. That accent briefly posterizes her charcoal-red with a white rim, then restores detail.\n\n[Shot 9] At 00:13.770, prior effects gone, a frontal ultra-wide catches her downward strike. At 00:14.120, weapon meets basalt and chain snaps taut: the iron sun erupts behind her into corona, lava shockwave, fire plume, furnace flares, and bowed smoke. A second shell hits at 00:14.560; a final flare at 00:14.910 frames her intact silhouette. Camera pulls back and shakes only on primary impact. End centered, weapon planted, chain weighted. Render exact \"ENRA VAEL\" once during the empty opening only; no reversed letters, extra glyphs, subtitles, logos, extra characters, costume changes, duplicate limbs, extra weapons, or fire across her face.\noverall_soundscape:\n\nLava roar, furnace resonance, wind, metal fittings, cloth, and chain clinks sit beneath the music. Add a restrained metal-weld sizzle at 00:00.240, a chain snap and ring whoosh at 00:06.340, then a metal strike, concussive blast, pressure rush, and ember hiss at 00:14.120 without overpowering < 📷Audio1 >.\nnon_diegetic_music:\nDirectly reuse the 00:00.000-00:15.000 musical layer of < 📷Audio1 > unchanged. Preserve its sparse opening, 00:06.340 drop, approximately 128-BPM pulse, 00:13.770 ignition, 00:14.120 detonation, 00:14.560 secondary burst, 00:14.910 flare, and hard ending; every cut, title state, and VFX change lands on the stated transient, with no replacement score.",
    "sourceUrl": "https://x.com/GlitterPixely/status/2086599324125872626",
    "sourceLabel": "X 原帖 · @GlitterPixely",
    "author": "Glitter Gal",
    "publishedAt": "2026-08-09T23:43:40.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-glitterpixely-2086599324125872626.mp4",
    "posterUrl": "/posters/x/2086599324125872626.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3778×2160",
    "tags": [
      "Character Introduction",
      "Prompt in Replies",
      "Hailuo AI",
      "Widescreen",
      "公开 Prompt"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 6,
      "reposts": 1,
      "likes": 27,
      "bookmarks": 3,
      "views": 414,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.1,
      "resolution": "3778×2160",
      "aspectRatio": "landscape"
    },
    "promptSourceUrl": "https://x.com/GlitterPixely/status/2086599326646628538"
  },
  {
    "id": "x-craftcapitallab-2086582697263284469",
    "title": "安利与科莱特的爱之糖果：第二集",
    "titleEn": "Henri and Colette's Love Dragee: Episode Two",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "《安利与科莱特的爱之糖果》第二集，上传成片长 40.75 秒，原帖明确关联 Hailuo AI 与 MiniMax H3；未公开 Prompt 或具体工作流。",
    "summaryEn": "Episode two of Henri and Colette's Love Dragee, uploaded as a 40.75-second clip and explicitly linked to Hailuo AI and MiniMax H3. No prompt or detailed workflow is published.",
    "prompt": null,
    "sourceUrl": "https://x.com/craftcapitallab/status/2086582697263284469",
    "sourceLabel": "X 原帖 · @craftcapitallab",
    "author": "ててつろう",
    "publishedAt": "2026-08-09T22:37:36.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-craftcapitallab-2086582697263284469.mp4",
    "posterUrl": "/posters/x/2086582697263284469.jpg",
    "duration": 41,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "Episode Two",
      "Short Film",
      "Narrative",
      "Hailuo AI"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 13,
      "bookmarks": 0,
      "views": 418,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 40.75,
      "resolution": "1920×1080",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-wabyanko-2086564753074270628",
    "title": "一周年故事与下一部短片预告",
    "titleEn": "One-Year Storytelling Reel and Next Short-Film Teaser",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "创作者回顾一年的 AI 辅助叙事，并将这段 8 秒竖屏视频称为下一部短片的 MiniMax H3 生成样片；原帖未说明输入素材与 Prompt。",
    "summaryEn": "The creator reflects on a year of AI-assisted storytelling and presents this eight-second vertical MiniMax H3 clip as a teaser for the next short film. Inputs and prompt are not disclosed.",
    "prompt": null,
    "sourceUrl": "https://x.com/Wabyanko/status/2086564753074270628",
    "sourceLabel": "X 原帖 · @Wabyanko",
    "author": "Wabyanko",
    "publishedAt": "2026-08-09T21:26:18.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-wabyanko-2086564753074270628.mp4",
    "posterUrl": "/posters/x/2086564753074270628.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "Short Film",
      "AI-Assisted Storytelling",
      "Vertical Video",
      "Anniversary"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 25,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 8,
      "resolution": "720×1280",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-ariestopazos-2086554484214497407",
    "title": "令人心跳的女孩对视瞬间",
    "titleEn": "A Heart-Racing Moment of Eye Contact",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "以女孩直视镜头为核心的 10.125 秒竖屏角色短片，作者明确称由 MiniMax H3 制作；原帖没有公开 Prompt。",
    "summaryEn": "A 10.125-second vertical character clip centered on direct eye contact with a girl. The creator explicitly identifies MiniMax H3 and does not publish a prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/AriesTopazos/status/2086554484214497407",
    "sourceLabel": "X 原帖 · @AriesTopazos",
    "author": "エイリース",
    "publishedAt": "2026-08-09T20:45:30.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-ariestopazos-2086554484214497407.mp4",
    "posterUrl": "/posters/x/2086554484214497407.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "Eye Contact",
      "Portrait",
      "Romance",
      "Vertical Video"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 25,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "576×1024",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-emma_shiraishi-2086550584694321222",
    "title": "东京黑色电影：近身格斗",
    "titleEn": "Tokyo Noir: Close-Quarters Combat",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "一段 15.116 秒宽银幕近身格斗短片，以“Tokyo Noir”为主题并明确标注 MiniMax H3；原帖没有披露生成模式或 Prompt。",
    "summaryEn": "A 15.116-second widescreen close-quarters combat clip framed as Tokyo Noir and explicitly tagged MiniMax H3. The generation mode and prompt are not disclosed.",
    "prompt": null,
    "sourceUrl": "https://x.com/emma_shiraishi/status/2086550584694321222",
    "sourceLabel": "X 原帖 · @emma_shiraishi",
    "author": "Emma Shiraishi 白石えま",
    "publishedAt": "2026-08-09T20:30:00.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-emma_shiraishi-2086550584694321222.mp4",
    "posterUrl": "/posters/x/2086550584694321222.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "Combat",
      "Action",
      "Tokyo Noir",
      "Widescreen"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "bookmarks": 0,
      "views": 1236,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.116,
      "resolution": "1920×1080",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-riddi0908-2086504040419225646",
    "title": "五秒 MiniMax H3 基础测试",
    "titleEn": "Five-Second MiniMax H3 Baseline Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "一段分辨率为 864×480、时长 5.166 秒的 MiniMax H3 测试视频。原帖只有模型标签，没有公开场景说明、输入方式或 Prompt。",
    "summaryEn": "A 5.166-second MiniMax H3 test at 864 by 480. The post contains only the model tag and does not disclose the scene, input method, or prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/riddi0908/status/2086504040419225646",
    "sourceLabel": "X 原帖 · @riddi0908",
    "author": "まゆひらa",
    "publishedAt": "2026-08-09T17:25:03.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-riddi0908-2086504040419225646.mp4",
    "posterUrl": "/posters/x/2086504040419225646.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "Test",
      "Five Seconds",
      "864x480",
      "Baseline"
    ],
    "category": "comparison",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 0,
      "views": 172,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "864×480",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-kunisuetakumi-2086500981773369551",
    "title": "105 秒 MiniMax H3 长视频样片",
    "titleEn": "A 105-Second MiniMax H3 Showcase",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "一支上传时长 105.041 秒、分辨率 1280×736 的 MiniMax H3 长视频样片。原帖仅提供模型标签，因此不推断它是否由单次生成或后期拼接而成。",
    "summaryEn": "A MiniMax H3 showcase uploaded at 105.041 seconds and 1280 by 736. Because the post provides only the model tag, this record does not infer whether it is a single generation or an edited assembly.",
    "prompt": null,
    "sourceUrl": "https://x.com/kunisuetakumi/status/2086500981773369551",
    "sourceLabel": "X 原帖 · @kunisuetakumi",
    "author": "國末 拓実 / Takumi Kunisue / Chief AI Evangelist",
    "publishedAt": "2026-08-09T17:12:54.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-kunisuetakumi-2086500981773369551.mp4",
    "posterUrl": "/posters/x/2086500981773369551.jpg",
    "duration": 105,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "Long Form",
      "105 Seconds",
      "Showcase",
      "Widescreen"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 108,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 105.041,
      "resolution": "1280×736",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-sauer20-2086491822231347327",
    "title": "漂浮感游乐设施式高速运镜",
    "titleEn": "Floating Ride-Style High-Speed Camera Motion",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "作者让 Grok 读取官方 Skill 后，把“角色对镜头说话、游乐设施般快速移动、带漂浮感”的构想整理成 Prompt，再用 MiniMax H3 生成 15.055 秒测试；原帖未逐字公开 Prompt。",
    "summaryEn": "After having Grok read the official skill, the creator turned an idea involving direct address, ride-like rapid camera movement, and a floating sensation into a prompt, then generated a 15.055-second MiniMax H3 test. The prompt is not published verbatim.",
    "prompt": null,
    "sourceUrl": "https://x.com/Sauer20/status/2086491822231347327",
    "sourceLabel": "X 原帖 · @Sauer20",
    "author": "Gilvert",
    "publishedAt": "2026-08-09T16:36:30.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-sauer20-2086491822231347327.mp4",
    "posterUrl": "/posters/x/2086491822231347327.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "Camera Motion",
      "Ride POV",
      "Floating Camera",
      "Prompt Writing"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 2,
      "bookmarks": 1,
      "views": 139,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.055,
      "resolution": "1920×1088",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-fnoji-2086491424405725197",
    "title": "本地实拍合成效果测试",
    "titleEn": "Local Live-Action Compositing Test",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "一段 8 秒超高竖屏本地生成测试。作者强调实拍合成效果已经足以减少 Blender 与 After Effects 的复杂处理，并明确标注 MiniMax H3。",
    "summaryEn": "An eight-second, high-resolution vertical local-generation test. The creator says the live-action compositing result could reduce complex Blender and After Effects work and explicitly tags MiniMax H3.",
    "prompt": null,
    "sourceUrl": "https://x.com/fnoji/status/2086491424405725197",
    "sourceLabel": "X 原帖 · @fnoji",
    "author": "えふのじ",
    "publishedAt": "2026-08-09T16:34:55.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-fnoji-2086491424405725197.mp4",
    "posterUrl": "/posters/x/2086491424405725197.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "Local Generation",
      "Live-Action Compositing",
      "Vertical Video",
      "Workflow"
    ],
    "category": "workflow",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 10,
      "bookmarks": 0,
      "views": 713,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 8,
      "resolution": "2160×3840",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-lumeroute-2086474007843577889",
    "title": "RTX 4090 48GB 的准 1K 本地上限测试",
    "titleEn": "Near-1K Local Limit Test on an RTX 4090 48GB",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "作者在 RTX 4090 48GB 上尝试以 1664×928 本地生成 10 秒 MiniMax H3 视频，但实际只输出 5.166 秒片段；这是一条公开的分辨率与时长上限记录。",
    "summaryEn": "The creator attempts a ten-second local MiniMax H3 render at 1664 by 928 on an RTX 4090 48GB, but receives only a 5.166-second clip. It serves as a public record of a resolution and duration limit.",
    "prompt": null,
    "sourceUrl": "https://x.com/lumeroute/status/2086474007843577889",
    "sourceLabel": "X 原帖 · @lumeroute",
    "author": "Mark Han",
    "publishedAt": "2026-08-09T15:25:42.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-lumeroute-2086474007843577889.mp4",
    "posterUrl": "/posters/x/2086474007843577889.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1664×928",
    "tags": [
      "RTX 4090",
      "48GB VRAM",
      "1664x928",
      "Benchmark"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "bookmarks": 0,
      "views": 215,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1664×928",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-finiyang-2086463689671238052",
    "title": "MiniMax H3 机甲风格试作",
    "titleEn": "MiniMax H3 Mecha-Style Experiment",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "创作者用 MiniMax H3 尝试机甲风格，发布了一段 14.458 秒宽银幕成片，并预告后续会整理 IP 漫画 Skill 教程；本帖未公开 Prompt。",
    "summaryEn": "The creator experiments with a mecha aesthetic in MiniMax H3, publishing a 14.458-second widescreen clip and previewing a future IP-comic skill tutorial. No prompt is published in this post.",
    "prompt": null,
    "sourceUrl": "https://x.com/FiniYang/status/2086463689671238052",
    "sourceLabel": "X 原帖 · @FiniYang",
    "author": "Fini.Yang",
    "publishedAt": "2026-08-09T14:44:42.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-finiyang-2086463689671238052.mp4",
    "posterUrl": "/posters/x/2086463689671238052.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "Mecha",
      "AIGC",
      "Widescreen",
      "Character"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 11,
      "bookmarks": 1,
      "views": 948,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 14.458,
      "resolution": "1152×640",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-bigdanclub-2086460574834241568",
    "title": "竖屏日常拉伸短片",
    "titleEn": "Vertical Daily Stretching Clip",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "以日常拉伸为主题的 8 秒竖屏 MiniMax H3 短片。原帖只说明这是创作者的日课，没有公开输入方式或 Prompt。",
    "summaryEn": "An eight-second vertical MiniMax H3 clip centered on a daily stretching routine. The post gives no input method or prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/bigdanclub/status/2086460574834241568",
    "sourceLabel": "X 原帖 · @bigdanclub",
    "author": "BIGDANCLUB",
    "publishedAt": "2026-08-09T14:32:20.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-bigdanclub-2086460574834241568.mp4",
    "posterUrl": "/posters/x/2086460574834241568.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "Stretching",
      "Lifestyle",
      "Vertical Video",
      "Character"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "bookmarks": 0,
      "views": 53,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 8,
      "resolution": "1080×1920",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-oodate-2086460367992152232",
    "title": "方形艺术感河豚短片",
    "titleEn": "Square Artistic Pufferfish Vignette",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "一段 10.125 秒方形艺术风格短片，作者以河豚主题账号发布并明确标注 MiniMax H3；原帖未公开 Prompt 或生成模式。",
    "summaryEn": "A 10.125-second square art-style vignette from a pufferfish-themed creator, explicitly tagged MiniMax H3. The prompt and generation mode are not disclosed.",
    "prompt": null,
    "sourceUrl": "https://x.com/oodate/status/2086460367992152232",
    "sourceLabel": "X 原帖 · @oodate",
    "author": "🐡BalloonFish🌊エルタリウム所属",
    "publishedAt": "2026-08-09T14:31:30.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-oodate-2086460367992152232.mp4",
    "posterUrl": "/posters/x/2086460367992152232.jpg",
    "duration": 10,
    "aspectRatio": "square",
    "resolution": "800×800",
    "tags": [
      "Art Film",
      "Pufferfish",
      "Square Video",
      "Experimental"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 15,
      "bookmarks": 0,
      "views": 243,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "800×800",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-vidmuse_ai-2086459987832815650",
    "title": "VidMuse 周末故事促销短片",
    "titleEn": "VidMuse Weekend-Story Promotion",
    "model": "MiniMax H3 · VidMuse",
    "mode": "Unknown",
    "summary": "VidMuse 发布的 16.191 秒宽银幕推广视频，将 MiniMax H3 与周末故事创作和限时折扣结合；这是品牌营销案例，并非公开工作流。",
    "summaryEn": "A 16.191-second widescreen VidMuse promotion pairing MiniMax H3 with weekend storytelling and a limited-time discount. It is a branded marketing case rather than a published workflow.",
    "prompt": null,
    "sourceUrl": "https://x.com/vidmuse_ai/status/2086459987832815650",
    "sourceLabel": "X 原帖 · @vidmuse_ai",
    "author": "VidMuse",
    "publishedAt": "2026-08-09T14:30:00.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-vidmuse_ai-2086459987832815650.mp4",
    "posterUrl": "/posters/x/2086459987832815650.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "Promotion",
      "Weekend",
      "VidMuse",
      "Discount"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "bookmarks": 0,
      "views": 96,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 16.191,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-hyperboleon-2086457102474887641",
    "title": "魂魄妖梦游戏 UI 与文字测试",
    "titleEn": "Youmu Game UI and Text Rendering Test",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "以《东方 Project》魂魄妖梦为主题的游戏风测试，作者指出 UI 仍不稳定但能够生成文字；公开参数为 MiniMax H3、864×480、25 步及 Sage、Sol-Attn、Spectrum。",
    "summaryEn": "A game-style test inspired by Youmu from Touhou Project. The creator reports unstable UI but successful text rendering and publishes the setup: MiniMax H3 at 864 by 480, 25 steps, Sage, Sol-Attn, and Spectrum.",
    "prompt": null,
    "sourceUrl": "https://x.com/hyperboleon/status/2086457102474887641",
    "sourceLabel": "X 原帖 · @hyperboleon",
    "author": "ひゅぺる",
    "publishedAt": "2026-08-09T14:18:32.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-hyperboleon-2086457102474887641.mp4",
    "posterUrl": "/posters/x/2086457102474887641.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×512",
    "tags": [
      "Game UI",
      "Text Rendering",
      "SageAttention",
      "Spectrum"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d",
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "bookmarks": 0,
      "views": 203,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "864×512",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-itoktw-2086456937164816564",
    "title": "FL2VA 与 Ref2VA 三路参考对比",
    "titleEn": "Three-Way FL2VA and Ref2VA Reference Comparison",
    "model": "MiniMax H3 · ComfyUI",
    "mode": "Ref2VA",
    "summary": "同一工作流使用一张背景图和两张人物图，依次比较 FL2VA + ref LoRA + LightX2V 六步、Ref2VA + LightX2V 六步，以及 Ref2VA 30 步。原帖包含三段视频，本记录的媒体字段对应第一段。",
    "summaryEn": "One workflow uses a background image and two character images to compare FL2VA with a reference LoRA and six-step LightX2V, six-step Ref2VA with LightX2V, and 30-step Ref2VA. The post contains three videos; this record's media fields describe the first.",
    "prompt": null,
    "sourceUrl": "https://x.com/ITOKtw/status/2086456937164816564",
    "sourceLabel": "X 原帖 · @ITOKtw",
    "author": "ITOK",
    "publishedAt": "2026-08-09T14:17:52.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-itoktw-2086456937164816564.mp4",
    "posterUrl": "/posters/x/2086456937164816564.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1184×896",
    "tags": [
      "FL2VA",
      "Ref2VA",
      "LightX2V",
      "Comparison"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 106,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "1184×896",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-sumica_rowa-2086450973900521592",
    "title": "周日晚上的“明天是周一”歌曲短片",
    "titleEn": "The Sunday-Night Monday-Is-Coming Song",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "一支 20 秒竖屏音乐短片，用“明天是周一”和“周日晚上看会掉 HP”的轻松文案制造共鸣，并明确标注 MiniMax H3。",
    "summaryEn": "A 20-second vertical music clip built around the joke that Monday is coming and watching it on Sunday night drains your health points, explicitly tagged MiniMax H3.",
    "prompt": null,
    "sourceUrl": "https://x.com/sumica_rowa/status/2086450973900521592",
    "sourceLabel": "X 原帖 · @sumica_rowa",
    "author": "sumicaぐらし | rowa",
    "publishedAt": "2026-08-09T13:54:11.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-sumica_rowa-2086450973900521592.mp4",
    "posterUrl": "/posters/x/2086450973900521592.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "Music Video",
      "Monday",
      "Twenty Seconds",
      "Vertical Video"
    ],
    "category": "music",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 16,
      "bookmarks": 0,
      "views": 250,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 20,
      "resolution": "576×1024",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-luccaaiart-2086449417947590957",
    "title": "首次让角色之美动起来",
    "titleEn": "First Attempt at Bringing a Character to Life",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "创作者的首次 MiniMax H3 尝试：一段 10.125 秒竖向角色样片，原帖将其描述为让角色之美动起来的漫长过程；未公开输入方式或 Prompt。",
    "summaryEn": "The creator's first MiniMax H3 attempt, a 10.125-second portrait-format character clip described as a long journey to bring the character's beauty to life. Inputs and prompt are not disclosed.",
    "prompt": null,
    "sourceUrl": "https://x.com/luccaaiart/status/2086449417947590957",
    "sourceLabel": "X 原帖 · @luccaaiart",
    "author": "aiLucca",
    "publishedAt": "2026-08-09T13:48:00.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-luccaaiart-2086449417947590957.mp4",
    "posterUrl": "/posters/x/2086449417947590957.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "First Try",
      "Portrait",
      "Character",
      "Ten Seconds"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 26,
      "bookmarks": 0,
      "views": 391,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "640×832",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-yayoi_labo_-2086448733378781682",
    "title": "带音乐与音效的战斗短片",
    "titleEn": "Combat Clip with Music and Sound Effects",
    "model": "MiniMax H3 · Pollo AI",
    "mode": "Unknown",
    "summary": "通过 Pollo AI 使用 MiniMax H3 制作的 15.083 秒宽银幕战斗短片。作者特别称音乐和音效也得到了良好呈现，但没有公开 Prompt。",
    "summaryEn": "A 15.083-second widescreen combat clip made with MiniMax H3 through Pollo AI. The creator specifically highlights the music and sound effects but does not publish a prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/Yayoi_Labo_/status/2086448733378781682",
    "sourceLabel": "X 原帖 · @Yayoi_Labo_",
    "author": "Healing Yayoi -AI Beauty Institute- 弥生",
    "publishedAt": "2026-08-09T13:45:17.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-yayoi_labo_-2086448733378781682.mp4",
    "posterUrl": "/posters/x/2086448733378781682.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "Combat",
      "Native Audio",
      "Sound Effects",
      "Pollo AI"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 5,
      "likes": 14,
      "bookmarks": 0,
      "views": 229,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "2560×1440",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-nyannero0828-2086445531614847163",
    "title": "80% 功耗限制下的本地稳定性测试",
    "titleEn": "Local Stability Test at an 80 Percent Power Limit",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "作者发现生成时黑屏可能源于显卡过热，于是把功耗限制到 80% 后继续运行 MiniMax H3，并发布这段 15.083 秒方形测试作为当前稳定性记录。",
    "summaryEn": "After suspecting that black screens were caused by GPU overheating, the creator caps power at 80 percent, continues running MiniMax H3, and shares this 15.083-second square clip as a current stability record.",
    "prompt": null,
    "sourceUrl": "https://x.com/nyannero0828/status/2086445531614847163",
    "sourceLabel": "X 原帖 · @nyannero0828",
    "author": "にゃんねろ | 🎰スロスロメスガキ | 🍭お料理女子🍳",
    "publishedAt": "2026-08-09T13:32:33.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-nyannero0828-2086445531614847163.mp4",
    "posterUrl": "/posters/x/2086445531614847163.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "GPU Thermals",
      "Power Limit",
      "Local Generation",
      "Square Video"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 6,
      "bookmarks": 0,
      "views": 141,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "640×640",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-noi06231-2086445301838348760",
    "title": "NIKKE 角色五秒动画测试",
    "titleEn": "Five-Second NIKKE Character Animation Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "一段以 NIKKE 为主题、时长 5.166 秒的 MiniMax H3 宽银幕短片。原帖只有游戏与模型标签，没有公开 Prompt 或工作流。",
    "summaryEn": "A 5.166-second widescreen MiniMax H3 clip themed around NIKKE. The post contains only game and model tags, with no prompt or workflow disclosed.",
    "prompt": null,
    "sourceUrl": "https://x.com/noi06231/status/2086445301838348760",
    "sourceLabel": "X 原帖 · @noi06231",
    "author": "八重白",
    "publishedAt": "2026-08-09T13:31:38.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-noi06231-2086445301838348760.mp4",
    "posterUrl": "/posters/x/2086445301838348760.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "NIKKE",
      "Anime",
      "Five Seconds",
      "Character"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 67,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "1152×640",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-openaiarka-2086444528219877725",
    "title": "八分钟生成的 30 秒多参考动画",
    "titleEn": "Thirty-Second Multi-Reference Animation Rendered in Eight Minutes",
    "model": "MiniMax H3 · Multi-Reference",
    "mode": "Ref2VA",
    "summary": "作者称使用 MiniMax H3 多参考模型，在约 8 分钟内完成一支约 30 秒动画；X 上传文件实际为 32.3 秒。原帖没有列出参考素材数量或 Prompt。",
    "summaryEn": "The creator reports producing roughly thirty seconds of animation with the MiniMax H3 multi-reference model in about eight minutes; the X upload is 32.3 seconds. The post does not list the number of references or a prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/openaiarka/status/2086444528219877725",
    "sourceLabel": "X 原帖 · @openaiarka",
    "author": "中性粒细胞NETS",
    "publishedAt": "2026-08-09T13:28:34.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-openaiarka-2086444528219877725.mp4",
    "posterUrl": "/posters/x/2086444528219877725.jpg",
    "duration": 32,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "Multi-Reference",
      "Thirty Seconds",
      "Eight-Minute Render",
      "Animation"
    ],
    "category": "cinematic",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 29,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 32.3,
      "resolution": "1920×1080",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-hiiragikuyouai-2086442694805782656",
    "title": "MiniMax H3 角色玩法小试",
    "titleEn": "A Small MiniMax H3 Character Experiment",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "作者以“正在尝试 MiniMax H3 的各种玩法”发布这段 5.166 秒竖向短片。原帖没有进一步说明输入、Prompt 或工作流。",
    "summaryEn": "The creator shares this 5.166-second portrait-format clip while experimenting with different ways to use MiniMax H3. Inputs, prompt, and workflow are not described.",
    "prompt": null,
    "sourceUrl": "https://x.com/HiiragiKuyouAI/status/2086442694805782656",
    "sourceLabel": "X 原帖 · @HiiragiKuyouAI",
    "author": "柊九陽＠AIイラスト",
    "publishedAt": "2026-08-09T13:21:17.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-hiiragikuyouai-2086442694805782656.mp4",
    "posterUrl": "/posters/x/2086442694805782656.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "Experiment",
      "Five Seconds",
      "Portrait Format",
      "Character"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "bookmarks": 0,
      "views": 147,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "576×736",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-48076551super-2086440470465622290",
    "title": "白衣与金衣拳手的擂台对峙",
    "titleEn": "White and Gold Fighters Face Off in the Arena",
    "model": "MiniMax H3 · Krea 2",
    "mode": "Unknown",
    "summary": "两名拳手以不同方式入场，一人身穿白色、一人身穿金色，随后在沉默的擂台上对峙。原帖标注 Krea 2 与 MiniMax H3，未说明两者各自承担的步骤。",
    "summaryEn": "Two fighters enter in different ways, one in white and one in gold, then face off in a silent arena. The post tags Krea 2 and MiniMax H3 without assigning each tool to a specific production step.",
    "prompt": null,
    "sourceUrl": "https://x.com/48076551super/status/2086440470465622290",
    "sourceLabel": "X 原帖 · @48076551super",
    "author": "Woman Wrestler and Leather outfit",
    "publishedAt": "2026-08-09T13:12:27.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-48076551super-2086440470465622290.mp4",
    "posterUrl": "/posters/x/2086440470465622290.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "2400×3168",
    "tags": [
      "Fighters",
      "Arena",
      "Showdown",
      "Krea 2"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 25,
      "bookmarks": 11,
      "views": 1433,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "2400×3168",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-mix_buchi_-2086440208640483672",
    "title": "女性仿生人高分辨率生成版",
    "titleEn": "High-Resolution Female Android Generation",
    "model": "MiniMax H3 · ComfyUI",
    "mode": "Unknown",
    "summary": "一段 15.062 秒、1600×920 的高分辨率 MiniMax H3 成片，原帖明确标注 ComfyUI。该帖是同作者女性仿生人试作的高分辨率版本，但没有公开具体放大或重生成参数。",
    "summaryEn": "A 15.062-second MiniMax H3 clip at 1600 by 920, explicitly tagged ComfyUI. It is the high-resolution version of the creator's female-android experiment, but no upscaling or regeneration parameters are disclosed.",
    "prompt": null,
    "sourceUrl": "https://x.com/mix_buchi_/status/2086440208640483672",
    "sourceLabel": "X 原帖 · @mix_buchi_",
    "author": "buchi_",
    "publishedAt": "2026-08-09T13:11:24.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-mix_buchi_-2086440208640483672.mp4",
    "posterUrl": "/posters/x/2086440208640483672.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1600×920",
    "tags": [
      "High Resolution",
      "ComfyUI",
      "Android",
      "Upscaling"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "bookmarks": 0,
      "views": 118,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.062,
      "resolution": "1600×920",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-fnoji-2086440070689742952",
    "title": "Motion Context 参考视频续接边界测试",
    "titleEn": "Motion Context Reference-Video Continuity Limit Test",
    "model": "MiniMax H3 · Motion Context",
    "mode": "Ref2VA",
    "summary": "使用 Motion Context 检验不同 Ref Video 输出之间的参考续接。作者认为参考阶段的约束仍然严苛，当前结果尚不足以稳定连接；原帖含两段视频，本记录的媒体字段对应第一段。",
    "summaryEn": "A Motion Context test of continuity between separate reference-video outputs. The creator finds the reference constraints demanding and the connection not yet reliable. The post contains two videos; this record's media fields describe the first.",
    "prompt": null,
    "sourceUrl": "https://x.com/fnoji/status/2086440070689742952",
    "sourceLabel": "X 原帖 · @fnoji",
    "author": "えふのじ",
    "publishedAt": "2026-08-09T13:10:51.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-fnoji-2086440070689742952.mp4",
    "posterUrl": "/posters/x/2086440070689742952.jpg",
    "duration": 4,
    "aspectRatio": "portrait",
    "resolution": "832×1504",
    "tags": [
      "Motion Context",
      "Reference Video",
      "Continuity",
      "Failure Boundary"
    ],
    "category": "comparison",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "video"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 5,
      "bookmarks": 0,
      "views": 361,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 4.25,
      "resolution": "832×1504",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-narrativenavi-2086435386394624411",
    "title": "祭坛觉醒与苍蓝月光",
    "titleEn": "Altar Awakening under Blue Moonlight",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "以“祭坛觉醒”和冷冽蓝色月光为主题的方形 MiniMax H3 短片。原帖含两段视频，本记录媒体字段对应第一段 5 秒视频；未公开 Prompt。",
    "summaryEn": "A square MiniMax H3 vignette themed around an altar awakening beneath cool blue moonlight. The post contains two videos; this record describes the first five-second clip. No prompt is published.",
    "prompt": null,
    "sourceUrl": "https://x.com/narrativenavi/status/2086435386394624411",
    "sourceLabel": "X 原帖 · @narrativenavi",
    "author": "ナビ",
    "publishedAt": "2026-08-09T12:52:14.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-narrativenavi-2086435386394624411.mp4",
    "posterUrl": "/posters/x/2086435386394624411.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "928×928",
    "tags": [
      "Altar",
      "Blue Moonlight",
      "Square Video",
      "Fantasy"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 11,
      "bookmarks": 0,
      "views": 261,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5,
      "resolution": "928×928",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-mix_buchi_-2086435346242572419",
    "title": "LLM 编写 Prompt 的女性仿生人短片",
    "titleEn": "Female Android Clip with an LLM-Written Prompt",
    "model": "MiniMax H3 T3 · ComfyUI",
    "mode": "T2VA",
    "summary": "作者把故事概要交给 LLM 编写 Prompt，再通过 ComfyUI 与 MiniMax H3 T3 制作一段 15.062 秒女性仿生人短片。原帖只说明 Prompt 的生成方式，没有逐字公开内容。",
    "summaryEn": "The creator gives a story outline to an LLM, which writes a prompt for a 15.062-second female-android clip made with ComfyUI and MiniMax H3 T3. The post explains how the prompt was produced but does not publish it verbatim.",
    "prompt": null,
    "sourceUrl": "https://x.com/mix_buchi_/status/2086435346242572419",
    "sourceLabel": "X 原帖 · @mix_buchi_",
    "author": "buchi_",
    "publishedAt": "2026-08-09T12:52:05.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-mix_buchi_-2086435346242572419.mp4",
    "posterUrl": "/posters/x/2086435346242572419.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1440×816",
    "tags": [
      "Android",
      "ComfyUI",
      "LLM Prompt",
      "Science Fiction"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "bookmarks": 0,
      "views": 215,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.062,
      "resolution": "1440×816",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-48076551super-2086433697834934530",
    "title": "女摔跤手与特工的任务开场",
    "titleEn": "A Wrestler and an Agent Begin Their Mission",
    "model": "MiniMax H3 · Krea 2",
    "mode": "Unknown",
    "summary": "女摔跤手代表力量，女特工代表精准，两人并肩向前，构成一次意外相遇与任务开场。原帖同时标注 Krea 2 和 MiniMax H3，但未拆分两者的制作职责。",
    "summaryEn": "A female wrestler represents power while a female agent represents precision; they walk forward together as an unexpected encounter becomes the opening of a mission. The post tags Krea 2 and MiniMax H3 without separating their production roles.",
    "prompt": null,
    "sourceUrl": "https://x.com/48076551super/status/2086433697834934530",
    "sourceLabel": "X 原帖 · @48076551super",
    "author": "Woman Wrestler and Leather outfit",
    "publishedAt": "2026-08-09T12:45:32.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-48076551super-2086433697834934530.mp4",
    "posterUrl": "/posters/x/2086433697834934530.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "800×1056",
    "tags": [
      "Wrestler",
      "Agent",
      "Character Pair",
      "Krea 2"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 20,
      "bookmarks": 5,
      "views": 406,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "800×1056",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-da1hen-2086430029358174433",
    "title": "Laptop RTX 4060 两日多镜头短片",
    "titleEn": "Two-Day Multi-Shot Film on a Laptop RTX 4060",
    "model": "MiniMax H3 · Local RTX 4060",
    "mode": "Unknown",
    "summary": "一支总上传时长 51.966 秒的多镜头 MiniMax H3 作品。作者逐镜生成，在 Laptop RTX 4060 上每 5 秒约需 6 分钟，整支作品制作耗时两天；总时长不代表单次生成。",
    "summaryEn": "A multi-shot MiniMax H3 work with a total upload length of 51.966 seconds. The creator generates each shot separately on a laptop RTX 4060 at roughly six minutes per five seconds and spends two days on the full piece; the upload length is not a single generation.",
    "prompt": null,
    "sourceUrl": "https://x.com/Da1hen/status/2086430029358174433",
    "sourceLabel": "X 原帖 · @Da1hen",
    "author": "Daihen",
    "publishedAt": "2026-08-09T12:30:57.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-da1hen-2086430029358174433.mp4",
    "posterUrl": "/posters/x/2086430029358174433.jpg",
    "duration": 52,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "Laptop RTX 4060",
      "Multi-Shot",
      "Local Generation",
      "Workflow"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 27,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 51.966,
      "resolution": "1280×720",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-riddi0908-2086419900151832923",
    "title": "Turbo v4 八步 SageAttention 测试",
    "titleEn": "Eight-Step Turbo v4 SageAttention Test",
    "model": "MiniMax H3 Turbo v4 · ComfyUI",
    "mode": "Unknown",
    "summary": "ComfyUI 本地加速测试，使用 minimax_h3_turbo_v4_step600_ema_pruned、SageAttention、Euler beta 与 8 步采样，输出 864×480、5.166 秒视频。",
    "summaryEn": "A local ComfyUI acceleration test using minimax_h3_turbo_v4_step600_ema_pruned, SageAttention, Euler beta, and eight sampling steps, producing a 5.166-second clip at 864 by 480.",
    "prompt": null,
    "sourceUrl": "https://x.com/riddi0908/status/2086419900151832923",
    "sourceLabel": "X 原帖 · @riddi0908",
    "author": "まゆひらa",
    "publishedAt": "2026-08-09T11:50:42.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-riddi0908-2086419900151832923.mp4",
    "posterUrl": "/posters/x/2086419900151832923.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "Turbo LoRA",
      "SageAttention",
      "Eight Steps",
      "Euler Beta"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 2,
      "bookmarks": 4,
      "views": 376,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 5.166,
      "resolution": "864×480",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-ai_haruki3-2086417527572455929",
    "title": "复杂加速工作流回归简化方案",
    "titleEn": "Returning to a Simpler Workflow after Acceleration Tests",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "作者尝试多种 MiniMax H3 工作流和加速方法后，最终回到更简单的方案，并表示等待官方 2K 放大与加速支持，同时继续研究 Prompt。上传测试为 10.125 秒竖向视频。",
    "summaryEn": "After trying several MiniMax H3 workflows and acceleration methods, the creator returns to a simpler setup, waits for official 2K upscaling and speed support, and continues prompt research. The uploaded test is a 10.125-second portrait-format clip.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_haruki3/status/2086417527572455929",
    "sourceLabel": "X 原帖 · @ai_haruki3",
    "author": "HARUKI KAWAI｜AI美女アーティスト",
    "publishedAt": "2026-08-09T11:41:16.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-ai_haruki3-2086417527572455929.mp4",
    "posterUrl": "/posters/x/2086417527572455929.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "832×1120",
    "tags": [
      "Workflow",
      "Acceleration",
      "2K Upscaling",
      "Prompt Research"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 31,
      "bookmarks": 14,
      "views": 3541,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 10.125,
      "resolution": "832×1120",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-kahataredoki1-2086397266496217222",
    "title": "RTX 5060 Ti 的 FL2VA 三镜头链式生成",
    "titleEn": "Three-Shot FL2VA Chain on an RTX 5060 Ti",
    "model": "MiniMax H3 FL2VA · ComfyUI",
    "mode": "FL2VA",
    "summary": "家庭本地环境实测：MiniMax H3 FL2VA pruned NVFP4、ComfyUI 0.30.0、SageAttention 与 Qwen3VL-32B NVFP4 AWQ，在 RTX 5060 Ti 16GB 上以 25 步连接三段 L2VA→I2VA→I2VA 镜头，成片 832×1248、15.532 秒。",
    "summaryEn": "A home local-generation test using MiniMax H3 FL2VA pruned NVFP4, ComfyUI 0.30.0, SageAttention, and Qwen3VL-32B NVFP4 AWQ on an RTX 5060 Ti 16GB. Twenty-five steps chain three L2VA-to-I2VA-to-I2VA shots into a 15.532-second clip at 832 by 1248.",
    "prompt": null,
    "sourceUrl": "https://x.com/kahataredoki1/status/2086397266496217222",
    "sourceLabel": "X 原帖 · @kahataredoki1",
    "author": "カブ",
    "publishedAt": "2026-08-09T10:20:46.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-kahataredoki1-2086397266496217222.mp4",
    "posterUrl": "/posters/x/2086397266496217222.jpg",
    "duration": 16,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "FL2VA",
      "NVFP4",
      "RTX 5060 Ti",
      "SageAttention"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "bookmarks": 0,
      "views": 233,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.532,
      "resolution": "832×1248",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-dimefromthevoid-2086396126585086283",
    "title": "梨乃的地盘",
    "titleEn": "Rino's Territory",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "标题为《梨乃的地盘》的 15.083 秒方形 MiniMax H3 短片。原帖只给出标题和模型标签，因此不推断角色设定、输入方式或 Prompt。",
    "summaryEn": "A 15.083-second square MiniMax H3 clip titled Rino's Territory. The post provides only the title and model tag, so no character setup, input method, or prompt is inferred.",
    "prompt": null,
    "sourceUrl": "https://x.com/dimefromthevoid/status/2086396126585086283",
    "sourceLabel": "X 原帖 · @dimefromthevoid",
    "author": "dimesfromthevoid",
    "publishedAt": "2026-08-09T10:16:14.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-dimefromthevoid-2086396126585086283.mp4",
    "posterUrl": "/posters/x/2086396126585086283.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1248×1248",
    "tags": [
      "Character",
      "Square Video",
      "Original Scene",
      "Fifteen Seconds"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "bookmarks": 0,
      "views": 55,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "1248×1248",
      "aspectRatio": "square"
    }
  },
  {
    "id": "x-kiyoshi_shin-2086576598225244651",
    "title": "街头人物一致性对比",
    "titleEn": "Street Portrait Consistency Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "作者将约一年前在本地 PC 上使用 Wan 2.2 制作的类似视频与这条 MiniMax H3 作例对照，认为 H3 的一致性和表现力有显著提升；画面为日本街巷中的人物近景。原帖未公开 Prompt。",
    "summaryEn": "The creator compares this MiniMax H3 example with a similar video made locally with Wan 2.2 about a year earlier and reports a substantial gain in consistency and expressiveness. The clip shows a character close-up on a Japanese street. No prompt is published.",
    "prompt": null,
    "sourceUrl": "https://x.com/kiyoshi_shin/status/2086576598225244651",
    "sourceLabel": "X 原帖 · @kiyoshi_shin",
    "author": "新清士@AIコンテンツ開発者",
    "publishedAt": "2026-08-09T22:13:22.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-kiyoshi_shin-2086576598225244651.mp4",
    "posterUrl": "/posters/x/2086576598225244651.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1184×896",
    "tags": [
      "Wan 2.2",
      "Consistency",
      "Expressiveness",
      "Model Benchmark"
    ],
    "category": "comparison",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 1,
      "views": 557,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "1184×896",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-strimando-2086576136000311486",
    "title": "RTX 5090 本地 480P 生成实测",
    "titleEn": "Local 480p Generation Benchmark on an RTX 5090",
    "model": "MiniMax H3 · Local",
    "mode": "Unknown",
    "summary": "作者公开 RTX 5090 本地运行 MiniMax H3 的性能记录：生成一条 15.083 秒的 480P 视频耗时 1 分 27 秒；画面为暖色未来舱室内景。原帖未公开 Prompt。",
    "summaryEn": "The creator reports a local MiniMax H3 benchmark on an RTX 5090: a 15.083-second 480p clip generated in one minute and 27 seconds. The video depicts a warmly lit futuristic cabin interior. No prompt is published.",
    "prompt": null,
    "sourceUrl": "https://x.com/STRIMANDO/status/2086576136000311486",
    "sourceLabel": "X 原帖 · @STRIMANDO",
    "author": "STRIMANDO.",
    "publishedAt": "2026-08-09T22:11:32.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-strimando-2086576136000311486.mp4",
    "posterUrl": "/posters/x/2086576136000311486.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "Local Generation",
      "RTX 5090",
      "480P",
      "Benchmark"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "bookmarks": 0,
      "views": 123,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "864×480",
      "aspectRatio": "landscape"
    }
  },
  {
    "id": "x-vkuoo-2086462586057638323",
    "title": "Krea 上的超现实 H3 版本",
    "titleEn": "Surreal H3 Variant Built on Krea",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "作者在 Krea 上制作了同一创意的 MiniMax H3 版本，并引用此前的 Seedance 2.5 版本供对照；竖屏画面由巨眼与密集的锥形独眼角色构成。原帖未公开 Prompt。",
    "summaryEn": "The creator built a MiniMax H3 version of the same concept on Krea and quotes an earlier Seedance 2.5 version for comparison. The vertical clip features a giant eye surrounded by dense cone-shaped one-eyed figures. No prompt is published.",
    "prompt": null,
    "sourceUrl": "https://x.com/vkuoo/status/2086462586057638323",
    "sourceLabel": "X 原帖 · @vkuoo",
    "author": "vkuoo",
    "publishedAt": "2026-08-09T14:40:19.000Z",
    "addedAt": "2026-08-10T10:48:55+08:00",
    "mediaUrl": "/media/x-vkuoo-2086462586057638323.mp4",
    "posterUrl": "/posters/x/2086462586057638323.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "Krea",
      "Seedance 2.5",
      "H3 Variant",
      "Surrealism"
    ],
    "category": "comparison",
    "styles": [
      "stylized",
      "anime"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "bookmarks": 1,
      "views": 181,
      "capturedAt": "2026-08-10T02:35:49.756Z"
    },
    "mediaMetadata": {
      "type": "native_video",
      "videoCount": 1,
      "durationSeconds": 15.083,
      "resolution": "1440×1920",
      "aspectRatio": "portrait"
    }
  },
  {
    "id": "x-2086644878105280946",
    "title": "彩色舞池中央牵手旋转的双人舞者",
    "titleEn": "Two Dancers Turn Hand in Hand at the Center of a Colorful Dance Floor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一男一女站在彩色同心圆舞池中央，在围观人群之间牵手并旋转移动。 成片规格为 10 秒 · 950×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man and woman stand at the center of a colorful concentric dance floor, holding hands and turning while a crowd watches around them. Output: 10s · 950×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ITOKtw/status/2086644878105280946",
    "sourceLabel": "X 原帖 · @ITOKtw",
    "author": "ITOK",
    "publishedAt": "2026-08-10T02:44:41.000Z",
    "addedAt": "2026-08-10T05:52:30.474Z",
    "mediaUrl": "/media/x-2086644878105280946.mp4",
    "posterUrl": "/posters/x/2086644878105280946.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "950×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のreference，プロンプトでの指定のやり方で変わるので，うまくいかなかったら調整でなんとかなるかもですね。最初のフレームで背景指定のものにならなかったので“ There is <Picture 1>”としたら，カメラ指定なしだけどクレーンショットを作ってくれました。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 130,
      "snapshotAt": "2026-08-10T05:52:30.474Z"
    },
    "approvedAt": "2026-08-10T05:52:30.474Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086644746882293894",
    "title": "餐桌女子与云端奔跑角色的九宫格对比",
    "titleEn": "Nine-Way Grids Compare a Woman at a Table and a Character Running Above the Clouds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "九宫格先排列餐桌旁女子的多个版本，中间出现比较标题，随后转为云层上奔跑角色的多版本网格。 成片规格为 74 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A nine-way grid first arranges multiple versions of a woman at a table, cuts to a comparison title, and then switches to a grid of a character running above the clouds. Output: 74s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TJ16th/status/2086644746882293894",
    "sourceLabel": "X 原帖 · @TJ16th",
    "author": "TJ16th",
    "publishedAt": "2026-08-10T02:44:10.000Z",
    "addedAt": "2026-08-10T05:52:30.474Z",
    "mediaUrl": "/media/x-2086644746882293894.mp4",
    "posterUrl": "/posters/x/2086644746882293894.jpg",
    "duration": 74,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3高速化9条件を同一プロンプト・同一seedで比較。\n左上は高速化なしの基準、右下はSage＋Sol-Attn＋Spectrumの全部入り。\n結構出てくる絵が違うのだな。\n※RTX 4060 Ti 16GB環境。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 190,
      "snapshotAt": "2026-08-10T05:52:30.474Z"
    },
    "approvedAt": "2026-08-10T05:52:30.474Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086643681306812770",
    "title": "红黑蒙面人物飞跃城市楼宇",
    "titleEn": "A Red-and-Black Masked Figure Leaps Across City Rooftops",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名红黑装束的蒙面人物从楼顶起跳，穿过高楼之间的空隙并在另一处屋面落地。 成片规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A masked figure in red and black launches from a rooftop, crosses the gap between tall buildings, and lands on another roof. Output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Suzuna_AI_girls/status/2086643681306812770",
    "sourceLabel": "X 原帖 · @Suzuna_AI_girls",
    "author": "鈴音🔔@Suzuna_japanese_AI_girls",
    "publishedAt": "2026-08-10T02:39:56.000Z",
    "addedAt": "2026-08-10T05:52:30.474Z",
    "mediaUrl": "/media/x-2086643681306812770.mp4",
    "posterUrl": "/posters/x/2086643681306812770.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こんにちはだよー🔔\n\n#BellEchoArt\nせっかくMinimax H3のおかげでseedance2.5にも手が届きそうなクオリティを作れるようになって、パソコンも使うようになってきたからLLMも入れたいなーと思いつつ、今は何が強いのかわからないし色々と調べてみたー\n\nそしたら懐かしきdeepseekの4.5とやらが目にとまりなんとなく見てみたらVRAM167GB必要とか書いててびっくりした💦\nこれ個人で使うやつじゃないじゃん😖\n\nローカルはローカルなんだろうけど家庭用で動くLLMとの比較にしれっと並べられても困るっていうか💧\n\n少年野球にメジャーリーガーが混ざってるような理不尽さを感じたかも🫠\n\ncreated with Topview Seedance 2.0",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 20,
      "views": 734,
      "snapshotAt": "2026-08-10T05:52:30.474Z"
    },
    "approvedAt": "2026-08-10T05:52:30.474Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086643266448244843",
    "title": "公园鸽子燃起火焰飞向城市天际线",
    "titleEn": "A Park Pigeon Bursts into Flame and Flies Toward the City Skyline",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一只鸽子先站在草地上，随后身体被火焰包围，并从公园上空飞向高楼之间。 成片规格为 8 秒 · 640×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pigeon first stands on grass, then becomes surrounded by flames and flies above a park toward tall city buildings. Output: 8s · 640×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pin_knuts/status/2086643266448244843",
    "sourceLabel": "X 原帖 · @pin_knuts",
    "author": "ぴんくなっつ",
    "publishedAt": "2026-08-10T02:38:17.000Z",
    "addedAt": "2026-08-10T05:52:30.474Z",
    "mediaUrl": "/media/x-2086643266448244843.mp4",
    "posterUrl": "/posters/x/2086643266448244843.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "焼き鳥の日でもあるんやな。\n○○の日がいっぱい流れてくるからWiki見たら今日はいろんなナントカの日みたいやね。\n\nしかしMinimax H3は何でも動画化してくれるな。\n何で自然発火するのかわからんが…\nこれがホントの火の鳥か？",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 116,
      "snapshotAt": "2026-08-10T05:52:30.474Z"
    },
    "approvedAt": "2026-08-10T05:52:30.474Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086639914679247062",
    "title": "白衣女子面对镜头的连续近景",
    "titleEn": "Continuous Close-Ups of a Woman in White Facing the Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿白色上衣的女子始终位于明亮室内近景中，三帧之间口型与面部姿态发生变化。 成片规格为 8 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a white top remains in a bright indoor close-up while her mouth shape and facial position change across the three frames. Output: 8s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/eggkyunio/status/2086639914679247062",
    "sourceLabel": "X 原帖 · @eggkyunio",
    "author": "きゅにお｜AIで心を動かす動画を作りたい",
    "publishedAt": "2026-08-10T02:24:58.000Z",
    "addedAt": "2026-08-10T05:52:30.474Z",
    "mediaUrl": "/media/x-2086639914679247062.mp4",
    "posterUrl": "/posters/x/2086639914679247062.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【 リップシンクの使い分け👄 】\n\nMiniMax H3ってリップシンク精度高そう💡\nアナウンサーっぽくないセリフでも自然か試してみないとだね！\n\n言葉のイントネーションがあと一息といったところだから、\n口パクの違和感なさ：MiniMax H3\nイントネーションの良さ：AIアバター　など\n好みで生成AI分かれそう🤔\n\nTopviewのAIアバターでの同じセリフ動画はリプへ\n\n#Topview #MiniMaxH3 \n@TopviewAIJP @TopviewAIhq",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 103,
      "snapshotAt": "2026-08-10T05:52:30.474Z"
    },
    "approvedAt": "2026-08-10T05:52:30.474Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086632971079958565",
    "title": "紫色光芒中现身的长发和服角色",
    "titleEn": "A Long-Haired Kimono-Clad Character Appears Through Purple Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "前两帧由明亮的紫白光带与暗部轮廓占据，末帧出现一名长发和服插画角色与白色标题文字。 成片规格为 15 秒 · 1728×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Bright purple-white light bands and a dark silhouette fill the first two frames before a long-haired illustrated character in a kimono appears beside white title text. Output: 15s · 1728×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/newtonmark13/status/2086632971079958565",
    "sourceLabel": "X 原帖 · @newtonmark13",
    "author": "newtonmark",
    "publishedAt": "2026-08-10T01:57:22.000Z",
    "addedAt": "2026-08-10T05:52:30.475Z",
    "mediaUrl": "/media/x-2086632971079958565.mp4",
    "posterUrl": "/posters/x/2086632971079958565.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1728×960",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "左边Minimax H3 量化4step残血瞎跑版，右边Grok Image2,同样的提示词，我觉得现在我真的不知道需要Grok来生成啥了，可以退订了。我都没用满血H3跑",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 104,
      "snapshotAt": "2026-08-10T05:52:30.475Z"
    },
    "approvedAt": "2026-08-10T05:52:30.475Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086630492363792699",
    "title": "霓虹夜色中抱着白鹦鹉与发光玫瑰的白发女子",
    "titleEn": "A White-Haired Woman Holds a White Parrot and Glowing Roses in Neon Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一名短白发女子站在霓虹城市背景前，怀中同时出现白色鹦鹉与带粉色光芒的玫瑰。 成片规格为 15 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short white-haired woman stands before a neon city backdrop while holding a white parrot and roses with a pink glow. Output: 15s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JohnConnorSkyne/status/2086630492363792699",
    "sourceLabel": "X 原帖 · @JohnConnorSkyne",
    "author": "John Connor",
    "publishedAt": "2026-08-10T01:47:31.000Z",
    "addedAt": "2026-08-10T05:52:30.475Z",
    "mediaUrl": "/media/x-2086630492363792699.mp4",
    "posterUrl": "/posters/x/2086630492363792699.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Ref2VA on RTX PRO 6000 — 480p, 6 steps, Turbo LoRA + SageAttention 2.2\n⚡ 15s video generated in 73s\n🚀 ~4.9× realtime",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 114,
      "snapshotAt": "2026-08-10T05:52:30.475Z"
    },
    "approvedAt": "2026-08-10T05:52:30.475Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086628613127098615",
    "title": "雨夜洗衣店里的两个人与一部手机",
    "titleEn": "Two People and a Phone in a Rainy-Night Laundromat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "两人从湿润街道进入洗衣店，随后并排坐在机器前，一部手机先放在两人之间再被左侧人物拿起。 成片规格为 30 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two people enter a laundromat from a wet street and sit side by side in front of the machines as a phone between them is picked up by the person on the left. Output: 30s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tiyyyyyyyok/status/2086628613127098615",
    "sourceLabel": "X 原帖 · @tiyyyyyyyok",
    "author": "かるびぃ",
    "publishedAt": "2026-08-10T01:40:03.000Z",
    "addedAt": "2026-08-10T05:52:30.475Z",
    "mediaUrl": "/media/x-2086628613127098615.mp4",
    "posterUrl": "/posters/x/2086628613127098615.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.5で作られていたこちらの動画、\n同じプロンプトを使って MiniMax H3（ローカル） でも30秒の広告動画を作ってみました。\n前半15秒：Image to Video\n→ 約11分\n\n後半15秒：Reference to Video（参照画像3枚）\n→ 約30分\n環境：RTX 4070 Ti SUPER 16GB / RAM 64GB\nTurboなし・軽量化あり",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 1030,
      "snapshotAt": "2026-08-10T05:52:30.475Z"
    },
    "approvedAt": "2026-08-10T05:52:30.475Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086627953488621905",
    "title": "工业擂台上的两名格斗者",
    "titleEn": "Two Fighters Face Off in an Industrial Arena",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名金发格斗者与一名银灰色对手站在烟囱环绕的工业擂台中，随后接近并互相出招。 成片规格为 65 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde fighter and a silver-gray opponent stand in an industrial arena surrounded by smokestacks before closing distance and exchanging attacks. Output: 65s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/logicolman/status/2086627953488621905",
    "sourceLabel": "X 原帖 · @logicolman",
    "author": "LOGICOLMAN",
    "publishedAt": "2026-08-10T01:37:26.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086627953488621905.mp4",
    "posterUrl": "/posters/x/2086627953488621905.jpg",
    "duration": 65,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 \nFighting Demo\nPC LOCAL\n\nStreet Fighter/ Mortal Kombat\n\nAún me falta pulir el prompt, para alargar cada secuencia de pelea y para darle mejores movimientos y técnicas de pelea",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 111,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086626083651424388",
    "title": "狐耳角色与街边燃烧的鲷鱼烧",
    "titleEn": "A Fox-Eared Character and a Flaming Taiyaki at a Street Stall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名白发狐耳角色站在街边点心铺前拿着鱼形点心，中段有另一块点心在地面燃烧，末帧角色把点心举到嘴边。 成片规格为 15 秒 · 1728×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired fox-eared character holds a fish-shaped pastry at a street stall, a second pastry burns on the ground, and the character later raises one toward the mouth. Output: 15s · 1728×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2086626083651424388",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-10T01:30:00.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086626083651424388.mp4",
    "posterUrl": "/posters/x/2086626083651424388.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1728×960",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n\n日常エピソード\n『たい焼き、転生』\n神狐様",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 16,
      "views": 518,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086625510528839945",
    "title": "巨月瀑布城与发光圆阵前的双剑客",
    "titleEn": "Two Sword-Bearing Figures Before a Moonlit Waterfall City and Glowing Ring",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名持剑长发人物面对巨月、瀑布与山城，中段出现悬在建筑上方的多层发光圆环，末帧回到城市全景。 成片规格为 15 秒 · 1664×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two long-haired figures carrying swords face a city of waterfalls beneath a giant moon, followed by a multilayered glowing ring above the buildings and a final wide city view. Output: 15s · 1664×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/newtonmark13/status/2086625510528839945",
    "sourceLabel": "X 原帖 · @newtonmark13",
    "author": "newtonmark",
    "publishedAt": "2026-08-10T01:27:43.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086625510528839945.mp4",
    "posterUrl": "/posters/x/2086625510528839945.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1664×928",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "对于Minimax H3 我唱衰4step ，量化版这么久，这次出个实例来,左边是满血版980p直出效果，右边是4step lora在int8 prune版效果。当然4step确实240秒就出来，这俩都是在RTX PRO 6000跑的，一个提示词，工作流不同",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 105,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086625171616706945",
    "title": "粉裙女子在工业走廊举枪瞄准",
    "titleEn": "A Woman in Pink Aims a Handgun in an Industrial Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿粉色短裙的女子在管线密布的工业走廊中持枪移动，并在末帧双手向镜头方向举枪。 成片规格为 10 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a pink skirt moves through an industrial corridor filled with pipes while holding a handgun and ends by aiming it toward the camera. Output: 10s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jaizard1/status/2086625171616706945",
    "sourceLabel": "X 原帖 · @jaizard1",
    "author": "jaizard",
    "publishedAt": "2026-08-10T01:26:23.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086625171616706945.mp4",
    "posterUrl": "/posters/x/2086625171616706945.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Higgsfieldのseedance 2.5である程度、自分の考えと動きの動画ができるようになってきた。\nHailuo AI minimax H3は、いい感じの動画を作るのが上手いけど、\nHiggsField seedance 2.5はどちらかというと動画作成ツールに近い。リファレンス画像をきっちり準備すればその通りにしてくれる。なので、結構カメラ割り。音源効果音をどこで何するか指定するようにリファレンス画像を持ってくると自分の思い通りになってくる。\n\n@higgsfield @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 91,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086623243348427121",
    "title": "手机屏幕中的蓝裙舞者与聊天界面",
    "titleEn": "A Blue-Gowned Dancer and Chat Screen on a Phone Display",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手机屏幕先播放一名蓝裙人物在明亮房间旋转的竖向画面，随后切换到包含文字与缩略图的聊天界面。 成片规格为 8 秒 · 1206×2622。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A phone screen first plays a vertical clip of a blue-gowned figure turning in a bright room and then switches to a chat interface containing text and thumbnails. Output: 8s · 1206×2622. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/orenoa7/status/2086623243348427121",
    "sourceLabel": "X 原帖 · @orenoa7",
    "author": "orenoa",
    "publishedAt": "2026-08-10T01:18:43.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086623243348427121.mp4",
    "posterUrl": "/posters/x/2086623243348427121.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "1206×2622",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Codexに動画生成機能追加！\n構成するAIは全てローカル\n\n・エージェント Qwen3.6 27B\n・動画生成 Minimax H3\n・動画理解 Qwen3-VL-30B",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 109,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086622474977050778",
    "title": "山间湖面的日出与薄雾",
    "titleEn": "Sunrise and Mist Over a Mountain Lake",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "太阳从山脊间升起并在湖面形成明亮倒影，低矮薄雾持续漂浮在水面上方。 成片规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The sun rises between mountain ridges and casts a bright reflection across a lake while low mist drifts above the water. Output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hAru_mAki_ch/status/2086622474977050778",
    "sourceLabel": "X 原帖 · @hAru_mAki_ch",
    "author": "Maki@Sunwood AI Labs.",
    "publishedAt": "2026-08-10T01:15:40.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086622474977050778.mp4",
    "posterUrl": "/posters/x/2086622474977050778.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "実験⓬｜MiniMax-H3 × ComfyUI更新比較\n\n同じ4090・同じseedで、\nComfyUI 0.30.0 → 0.31.0\nTurbo 1.2.2 → 1.2.3\nを動画で比較してみた！！！\n\n更新版はTurboの処理が約53秒短縮。\n1本目が旧版、2本目が更新版。\n\n一方で音声は小さくなった、、、\n次は音声経路を切り分けるぞ🔥",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 531,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086620989912408140",
    "title": "红色泳装女子坐在海滩轻理头发",
    "titleEn": "A Woman in a Red Swimsuit Adjusts Her Hair on a Beach",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿红色两件式泳装的女子坐在海滩毛巾上，先触碰耳边头发，再转动头部与上身。 成片规格为 7 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a red two-piece swimsuit sits on a beach towel, touches her hair near one ear, and then turns her head and upper body. Output: 7s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Luna_R4ze/status/2086620989912408140",
    "sourceLabel": "X 原帖 · @Luna_R4ze",
    "author": "LunaRaze",
    "publishedAt": "2026-08-10T01:09:46.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086620989912408140.mp4",
    "posterUrl": "/posters/x/2086620989912408140.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "On MiniMax H3 I find outdoor shots are the easiest way to keep things looking as realisitic as possible. Indoor lighting tends to play havoc with the skin textures, but outdoors looks much better. Has anyone else had a similar experience? Any tips on improving indoor scenes? This is all done through ComfyUI then some tweaks in CapCut. 🔥",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 10,
      "views": 722,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086620966751510630",
    "title": "太空座舱中的女驾驶员、外星乘员与双机飞行",
    "titleEn": "A Human Pilot, an Alien Crew Member, and Two Spacecraft in Flight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "一名女子在太空座舱中握住操纵杆，随后出现坐在控制台旁的外星乘员，末帧两艘飞船并行穿过星空。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman grips the controls inside a spacecraft, an alien crew member appears beside a console, and two ships fly side by side through space in the final frame. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BrentLynch/status/2086620966751510630",
    "sourceLabel": "X 原帖 · @BrentLynch",
    "author": "Brent Lynch",
    "publishedAt": "2026-08-10T01:09:40.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086620966751510630.mp4",
    "posterUrl": "/posters/x/2086620966751510630.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ASKING FOR A FRIEND⁉️\n\nJealous of you Open Source Cool Kids!\n\nSuch a fun model! \n\nMINIMAX H3 TXT2VID\n@MiniMax_AI",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 274,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086618421634924720",
    "title": "雪镇卡通里的汽油罐、电视画面与玩具房车",
    "titleEn": "Fuel Cans, a Television Screen, and a Toy RV in a Snowy Cartoon Town",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雪地小镇中的卡通儿童先站在电视前，随后画面切到装满汽油罐的皮卡和加油机，末帧一名儿童拿着玩具房车。 成片规格为 61 秒 · 1920×1400。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Cartoon children in a snowy town first stand before a television, the scene then cuts to a pickup loaded with fuel cans at a pump, and one child later holds a toy RV. Output: 61s · 1920×1400. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FurkanGozukara/status/2086618421634924720",
    "sourceLabel": "X 原帖 · @FurkanGozukara",
    "author": "Furkan Gözükara",
    "publishedAt": "2026-08-10T00:59:33.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086618421634924720.mp4",
    "posterUrl": "/posters/x/2086618421634924720.jpg",
    "duration": 61,
    "aspectRatio": "landscape",
    "resolution": "1920×1400",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Donald Trump says the Iran war ends “pretty soon.” \n\nThis fan-made South Park parody asks what happens when the missiles are “pre-filled,” the strategy wheel only gets louder, and the off-ramp is a cliff. 😂\n\nMade locally with MiniMax H3 - 0 shot\n\nPrompts written by ChatGPT",
    "engagement": {
      "replies": 1,
      "reposts": 9,
      "likes": 28,
      "views": 2973,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086617993950454040",
    "title": "沙漠峡谷中骑士军团遭遇喷火巨龙",
    "titleEn": "An Armored Column Meets a Fire-Breathing Dragon in a Desert Canyon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一列披甲角色穿过红色岩石峡谷，随后巨龙朝地面喷出火焰，末帧出现一名披毛皮的魁梧人物近景。 成片规格为 47 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A column of armored characters crosses a red-rock canyon, a dragon then breathes fire toward the ground, and the final frame shows a large fur-clad figure in close-up. Output: 47s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Caden_Flux/status/2086617993950454040",
    "sourceLabel": "X 原帖 · @Caden_Flux",
    "author": "Caden Flux",
    "publishedAt": "2026-08-10T00:57:51.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086617993950454040.mp4",
    "posterUrl": "/posters/x/2086617993950454040.jpg",
    "duration": 47,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I used to think creating cinematic fantasy battles with AI was still a major challenge…  \n\nBut this result completely changed my perspective🔥 \n\nTesting @newtake_kr  with Seedance 2.5, MiniMax H3, and Nano Banana shows how far AI video generation has come.  \n\nThe action choreography, character consistency, and cinematic world-building feel closer than ever to a real game trailer or CG production.  \n\nScenes like this once required huge production teams, but AI is making high-quality storytelling more accessible to creators everywhere.  \n\nThe evolution of AI filmmaking is moving incredibly fast, and this is just the beginning. 📷 \n\n#Newtake #Seedance25 #MiniMaxH3 #NanoBanana",
    "engagement": {
      "replies": 42,
      "reposts": 5,
      "likes": 137,
      "views": 17131,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086617156364439560",
    "title": "女子牵着金毛犬自拍漫步社区",
    "titleEn": "A Woman Films Herself Walking a Golden Retriever Through a Neighborhood",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿灰色卫衣的女子以手持自拍视角走过住宅区人行道，金毛犬在她身旁行走并短暂停在草地边。 成片规格为 62 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a gray sweatshirt films herself while walking along a residential sidewalk as a golden retriever moves beside her and briefly stops near the grass. Output: 62s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Paolozk1/status/2086617156364439560",
    "sourceLabel": "X 原帖 · @Paolozk1",
    "author": "Paolozky",
    "publishedAt": "2026-08-10T00:54:32.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086617156364439560.mp4",
    "posterUrl": "/posters/x/2086617156364439560.jpg",
    "duration": 62,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Generé un video de 1 minuto con MiniMax H3 en mi propia PC.\n\nSin GPUs en la nube. Sin servidores externos.\n\nRTX 3080 Ti — 12 GB VRAM\n32 GB RAM DDR4\n\nTardó solo 1 hora y 10 minutos.\n\nLa generación de video local está avanzando mucho más rápido de lo que esperaba🤯",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 57,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086616787240825098",
    "title": "热带海岸上的男子、棕榈叶床与野外进食",
    "titleEn": "A Man Rests on Palm Leaves and Eats on a Tropical Shore",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面先出现热带海岸，随后一名灰发男子躺在夕阳下的棕榈叶上，末帧他在近景中把食物送入口中。 成片规格为 43 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The sequence opens on a tropical shore, then shows a gray-haired man lying on palm leaves at sunset before he raises food to his mouth in close-up. Output: 43s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RenySantho75815/status/2086616787240825098",
    "sourceLabel": "X 原帖 · @RenySantho75815",
    "author": "DoDobe",
    "publishedAt": "2026-08-10T00:53:04.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086616787240825098.mp4",
    "posterUrl": "/posters/x/2086616787240825098.jpg",
    "duration": 43,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ascend 950pr run local minimax h3 : Lmao\n华为昇腾950单卡跑minimaxh3国产模型生成老黄求生！！哈哈😂笑死",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 64,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086616656672129107",
    "title": "节点工作流与本地配置的桌面操作",
    "titleEn": "Node Workflow and Local Configuration on a Desktop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "桌面录制画面先显示带参数弹窗的节点工作流，随后切换到包含素材缩略图与配置文字的应用窗口。 成片规格为 55 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A desktop capture first shows a node workflow with a parameter dialog, then switches to application windows containing an asset thumbnail and configuration text. Output: 55s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ModengSir/status/2086616656672129107",
    "sourceLabel": "X 原帖 · @ModengSir",
    "author": "Modengsir AI",
    "publishedAt": "2026-08-10T00:52:33.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086616656672129107.mp4",
    "posterUrl": "/posters/x/2086616656672129107.jpg",
    "duration": 55,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【云端 ComfyUI+MiniMax H3 成本昨晚实测视频】\n\n输出规格：10s 视频，1280*736 像素\n渲染耗时：19 分钟\n云端服务单价：3 元左右 / 小时\n单条成本：约 1 元左右\n\n【结论】：零散创作按量计费友好，批量创作包月套餐可显著降低制作成本。",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 5,
      "views": 966,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086519829612130667",
    "title": "紫色封面上的彩色播放徽标",
    "titleEn": "Colorful Play Emblem on a Purple Title Card",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫色渐变封面中央排列彩色播放徽标、白色模型名称和醒目的黄色评测标签，三帧基本保持静止。 成片规格为 7 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple gradient title card centers a colorful play emblem, a white model name, and a prominent yellow review label, remaining nearly static across all three frames. Output: 7s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxonShire/status/2086519829612130667",
    "sourceLabel": "X 原帖 · @MaxonShire",
    "author": "MaxonShire",
    "publishedAt": "2026-08-09T18:27:47.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086519829612130667.mp4",
    "posterUrl": "/posters/x/2086519829612130667.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Explained — Is Hailuo 3 0 the Best AI Video Model in 2026 ?\n\n#MiniMaxH3 #Hailuo3 #AIVideo \n\nWATCH IT HERE:\n-------------------\nhttps://youtu.be/OnFfp4AjIpg",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 0,
      "views": 106,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086486707268947989",
    "title": "少女沿山间公路滑板下坡",
    "titleEn": "Girl Skateboards Down Mountain Road Toward Cloud-Bound City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿白色上衣与蓝裙的动漫少女压低身体，沿松林与峭壁之间的公路滑板下坡，远处城市浮现在云海上方。 成片规格为 15 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime girl in a white top and blue skirt crouches on a skateboard while descending a road between pine-covered cliffs toward a distant city above the clouds. Output: 15s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bypass47078/status/2086486707268947989",
    "sourceLabel": "X 原帖 · @bypass47078",
    "author": "Bypass",
    "publishedAt": "2026-08-09T16:16:10.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086486707268947989.mp4",
    "posterUrl": "/posters/x/2086486707268947989.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "院内部署了 MiniMax H3，结合 GPT-Image-2 生图做的第一个视频，挺有意思的。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 125,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086481332763255063",
    "title": "路飞与贝吉塔在真实房间内起身",
    "titleEn": "Luffy and Vegeta Get Up in a Real Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "路飞与贝吉塔以卡通角色形态出现在真实房间中，两人先以手撑地向前移动，随后并肩站起。 成片规格为 10 秒 · 480×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Cartoon versions of Luffy and Vegeta appear in a real room, moving forward with their hands on the floor before standing side by side. Output: 10s · 480×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2086481332763255063",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-09T15:54:49.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086481332763255063.mp4",
    "posterUrl": "/posters/x/2086481332763255063.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "480×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Finally testing MiniMax H3 Ref2VA (Reference to Video/Audio) in WanGP.\n\nThe previous version had a memory issue that kept crashing, but now it works.\n\nThis one is \"Provide generic control video\" + 2 ref images. Prompting has a learning curve but it works when you get it.",
    "engagement": {
      "replies": 6,
      "reposts": 7,
      "likes": 50,
      "views": 5886,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086479201650213135",
    "title": "金发魔女在拱廊中抬手扶帽",
    "titleEn": "A Blonde Witch Adjusts Her Hat in an Arched Hall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿蓝白礼服的金发魔女站在高耸拱廊中，先朝镜头伸手，随后抬起双手扶住宽檐帽。 成片规格为 8 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde witch in a blue-and-white dress stands in a tall arched hall, first reaching toward the camera and then lifting both hands to steady her wide-brimmed hat. Output: 8s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fnoji/status/2086479201650213135",
    "sourceLabel": "X 原帖 · @fnoji",
    "author": "えふのじ",
    "publishedAt": "2026-08-09T15:46:21.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086479201650213135.mp4",
    "posterUrl": "/posters/x/2086479201650213135.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で出力した奴をアップスケールしてみたけどちょっとノイジー\nそもそもTwitterに上げてる時点で微妙化も\n実写じゃないしコツが要りそうな気がする",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 1583,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086477914699170293",
    "title": "赛博特工从剪影显现在数据面板",
    "titleEn": "A Cybernetic Operative Appears on a Stat Panel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑暗中的人物剪影逐渐亮起，随后成为站在未来感数据面板中央的黑衣赛博特工，周围排列属性、装备与人体示意图。 成片规格为 15 秒 · 1440×2560。原作者在回复区完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A figure emerges from a dark silhouette into a black-suited cybernetic operative centered on a futuristic stat panel surrounded by attributes, equipment, and body diagrams. Output: 15s · 1440×2560. The creator published the complete prompt in a reply, preserved here verbatim.",
    "prompt": "@[ref img] is the sole visual authority for ZENITH, his face, body, hairstyle, black tactical bodysuit, cyan circuit details, exposed cybernetic arm, and the complete operative-dossier interface including typography, numbers, icons, colors, panels, lighting, layout, and 9:16 vertical composition.\n\nAnimate this exact screen without redesigning it. Preserve every existing label, stat value, diagnostic value, icon, panel boundary, and final UI position.\n\n[GOAL]\nCreate a polished 15-second futuristic cyber-operative character dossier opening. The feeling is elite black-ops intelligence software coming online and identifying an extremely dangerous enhanced operative.\n\nUse one continuous locked straight-on full-screen composition. No camera movement, cuts, reframing, cropping, zooming, rotation, or perspective distortion.\n\n[SEQUENCE]\n\n0–2.2 seconds:\nBegin from near-black charcoal with only extremely faint cyan interface traces visible.\n\nA thin horizontal biometric scanner travels downward through the screen, gradually revealing ZENITH’s silhouette and the outer dossier frame.\n\nFine digital particles, tiny grid points, circuit traces, and subtle cyan data noise activate progressively.\n\nZENITH remains almost motionless in shadow.\n\nHis cybernetic arm gives one extremely subtle mechanical initialization twitch as internal joints power on.\n\n2.2–5.5 seconds:\nThe OPERATIVE DOSSIER header, FILE ID: ZN-7X-00, ZENITH nameplate, CYBERNETIC OPERATIVE subtitle, PROFILE panel, portrait frame, biometric panel, and THREAT LEVEL panel resolve through precise line-draw and scanning animations.\n\nReveal the existing profile information exactly:\nROLE: Black Ops / Infiltration\nHEIGHT: 6’2” / 188 cm\nBUILD: Lean / Athletic\nAFFILIATION: Unknown\n\nThe THREAT LEVEL display resolves into:\nEXTREME\nCLASS: OMEGA\n\nZENITH emerges completely from darkness into the exact referenced standing pose.\n\nHe takes one controlled breath, subtly raises his chest and shoulders, then settles.\n\nHis stern gaze sharpens toward the viewer.\n\nMaintain his exact angular face, slick black hairstyle, lean proportions, black fitted suit, cyan circuitry, and single exposed cybernetic arm.\n\nThin cyan suit lines illuminate progressively from the upper torso toward the legs and cybernetic arm.\n\n5.5–10.8 seconds:\nAnimate the OPERATIVE METRICS panel sequentially.\n\nEach bar fills smoothly left-to-right and lands precisely on the existing final value:\n\nSTRENGTH — 84\nAGILITY — 96\nREFLEX — 98\nENDURANCE — 88\nSTEALTH — 97\nCOMBAT IQ — 93\nCYBER SYNC — 94\nTACTICAL EFFICIENCY — 91\n\nEach completed value receives one restrained cyan confirmation pulse. Do not alter the final numbers.\n\nSimultaneously, the biometric body scan draws vertically through the miniature human silhouette.\n\nSmall waveform and graph elements activate with subtle believable data movement.\n\nThe CYBERNETIC ARM DIAGNOSTICS section powers online from shoulder to fingertips.\n\nCallout lines trace toward the mechanical joints while the existing diagnostic values resolve exactly:\n\nMOTOR UNITS — 100%\nSTRENGTH OUTPUT — 132%\nRESPONSE TIME — 0.008s\nSENSOR SUITE — ONLINE\nGRIP FORCE — 2,400 N\nSYSTEM INTEGRITY — 100% / OPTIMAL\n\nThe circular SYSTEM INTEGRITY indicator completes one clean clockwise sweep and locks at 100%.\n\nZENITH’s mechanical fingers make one tiny natural calibration movement before returning to the exact referenced resting position.\n\n10.8–13.3 seconds:\nThe SPECIALIZATIONS and EQUIPMENT LOADOUT sections activate from left to right.\n\nExisting specialization icons receive brief controlled cyan illumination:\nSTEALTH INFILTRATION\nCQC EXPERT\nCYBERNETIC ENHANCEMENT\nINTEL GATHERING\n\nExisting equipment cards resolve cleanly:\nSHADOW BLADE\nEMP SHARD\nSMOKE DISPERSER\nDATA SPIKE\n\nDo not physically place these weapons or tools into ZENITH’s hands. They remain UI inventory graphics only.\n\nA subtle electronic current travels through the cyan circuitry across ZENITH’s suit and into the cybernetic arm.\n\nThe mechanical shoulder, elbow, wrist, and finger joints produce extremely restrained servo adjustments.\n\nZENITH tightens his organic fist slightly.\n\nHe gives one natural blink while maintaining his intimidating expression and exact stance.\n\n13.3–15 seconds:\nThe lower dossier statement and STATUS: ACTIVE indicator reach full brightness.\n\nA controlled cyan system pulse travels once around the outer interface frame and converges briefly behind ZENITH.\n\nHis cybernetic arm emits a restrained internal energy glow through existing cyan seams—no new effects or weapons.\n\nThe pulse settles immediately.\n\nHold the fully assembled screen matching @[ref img] exactly.\n\nZENITH continues subtle breathing.\nTiny cybernetic servos remain alive.\nCyan suit circuitry gives a gentle idle pulse.\nBiometric waveforms move minimally.\nSmall particles shimmer.\nAll text, numbers, icons, diagnostic values, stat bars, and UI geometry remain completely stable and readable.\n\n[CHARACTER MOTION]\nZENITH is disciplined and nearly motionless.\n\nMovement must remain minimal and believable:\ncontrolled breathing,\none blink,\nsmall shoulder adjustment,\nslight organic fist tightening,\ntiny cybernetic finger calibration,\nsubtle mechanical joint settling.\n\nDo not turn his body away from the referenced pose.\nDo not change his facial structure, hairstyle, costume, proportions, cybernetic-arm design, or expression language.\n\n[UI MOTION]\nPremium classified-military sci-fi HUD animation.\n\nUse:\nprecise line drawing,\nscanner sweeps,\ndata ticks,\nsegmented stat fills,\nrestrained cyan confirmation pulses,\nbiometric scan lines,\nmechanical diagnostic tracing,\nvery subtle particles,\nclean system boot transitions.\n\nKeep every panel, icon, label, number, and geometric boundary fixed once revealed.\n\nNo floating panels.\nNo rearranging the interface.\nNo unnecessary glitching.\nNo aggressive flicker.\nNo holographic elements crossing over readable text.\n\n[VISUAL STYLE]\nCrisp premium futuristic game UI.\nClean stylized 2D animation matching the reference.\nDark charcoal-black interface with cyan/teal illumination and white typography.\nSharp angular graphic character rendering.\nSubtle depth separation between ZENITH and the HUD while preserving the exact flat front-facing composition.\nHigh readability throughout.\n\n[STRICT PRESERVATION]\nDo not introduce:\nnew text,\nnew numbers,\nnew stats,\nnew icons,\nextra cybernetic limbs,\nadditional weapons,\nextra characters,\nnew logos,\ncaptions,\nenvironment scenery,\ncamera motion,\nscreen shake,\ninterface redesign,\nnew panels,\nor additional objects.\n\nDo not change any final UI values.\n\n[Audio]\nLow classified-system boot hum.\nSoft scanner sweep.\nFine electronic data ticks.\nShort restrained confirmation tones as each metric locks.\nQuiet mechanical servo clicks from the cybernetic arm.\nSubtle electrical current as the cyan suit circuitry activates.\nLow biometric pulse.\nOne refined completion chime when STATUS: ACTIVE resolves.\n\nNo dialogue.\nNo background music.",
    "sourceUrl": "https://x.com/adithatipalli/status/2086477914699170293",
    "sourceLabel": "X 原帖 · @adithatipalli",
    "author": "Adithya Thatipalli",
    "publishedAt": "2026-08-09T15:41:14.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086477914699170293.mp4",
    "posterUrl": "/posters/x/2086477914699170293.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Player stats UI made using GPT Image 2 and Minimax H3 made on @higgsfield\n\nPrompt reference by @aimikoda \n\nCharacter and video prompt in replies",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 171,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review",
    "promptSourceUrl": "https://x.com/adithatipalli/status/2086478122241962266"
  },
  {
    "id": "x-2086477805202587697",
    "title": "雨中街口升空越过高楼的汽车",
    "titleEn": "A Car Rises Between Skyscrapers Above a Rainy Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名黑衣人物站在湿润的城市街口面对驶来的车辆，末帧改为仰视角度，一辆黑色汽车与碎片悬在高楼之间。 成片规格为 15 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-clad figure stands on a wet city street facing approaching vehicles before the view tilts upward to a black car and debris suspended between skyscrapers. Output: 15s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ivanfioravanti/status/2086477805202587697",
    "sourceLabel": "X 原帖 · @ivanfioravanti",
    "author": "Ivan Fioravanti ᯅ",
    "publishedAt": "2026-08-09T15:40:48.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086477805202587697.mp4",
    "posterUrl": "/posters/x/2086477805202587697.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hybrid AI Experiments with MiniMax H3.\n2K generation online costs $0.13/sec\n2K regeneration online costs $0.05/sec\n\nI tried the API with the video I generated locally and output quality is not bad. This can be a good example of Hybrid AI where you leverage Local AI resources and combined them with Cloud ones to get best of both worlds. Just some initial experiments. \n\nAPI docs here: https://platform.minimax.io/docs/api-reference/video-generation-v2-regeneration",
    "engagement": {
      "replies": 6,
      "reposts": 0,
      "likes": 26,
      "views": 4307,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086473110165799071",
    "title": "画案上的水墨锦鲤与展开画卷",
    "titleEn": "Ink-Wash Koi Across a Drawing Table and Scroll",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "画案上摆着卷纸、砚台和毛笔，水墨锦鲤随后占满纸面，末帧由白衣人物双手展开带有锦鲤的画卷。 成片规格为 52 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A drawing table holds rolled paper, an inkstone, and a brush before an ink-wash koi fills the sheet and a white-robed person unfolds the finished scroll. Output: 52s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TeslianHumanoid/status/2086473110165799071",
    "sourceLabel": "X 原帖 · @TeslianHumanoid",
    "author": "FilipVijo",
    "publishedAt": "2026-08-09T15:22:08.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086473110165799071.mp4",
    "posterUrl": "/posters/x/2086473110165799071.jpg",
    "duration": 52,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "My first video made on open weights using Minimax H3.\n\"THE EYE 眼\" : a 52-second ink-wash film made entirely on open weights.🎬 \nThe process: •\n Model: MiniMax H3 @Hailuo_AI  open weights , FL2VA, self-hosted, no API fees \n• Hardware: rented A100 80GB on RunPod (Iceland) \n• ComfyUI, 8 shots generated locally \n• One accent color in the whole film: the vermilion seal 眼 at the end \n• Continuity: each shot anchored on the previous shot's actual last frame \n• Upscaled from 728p to 1080p HD\n• Score: @suno instrumental \n• Total generation cost: ~$3.50\n\nEvery frame generated, deployed and QC'd by my operator, my right hand, my coworker, my brother in arms, my Hermes agent. @NousResearch @Teknium \n\nOpen weights. Full control.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 174,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086472319141110032",
    "title": "红发女子迎战熔岩巨人的双栏对照",
    "titleEn": "Split-Screen Clash Between a Red-Haired Woman and Lava Giant",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右两栏同步呈现红发女子在工业设施中面对熔岩巨人的不同版本，一侧包含正面交锋，另一侧更多保留人物近景与躲避动作。 成片规格为 30 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two side-by-side columns show different versions of a red-haired woman facing a lava giant in an industrial facility, with a direct clash on one side and more close-ups and evasive movement on the other. Output: 30s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIReelofficial/status/2086472319141110032",
    "sourceLabel": "X 原帖 · @AIReelofficial",
    "author": "AIReel",
    "publishedAt": "2026-08-09T15:19:00.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086472319141110032.mp4",
    "posterUrl": "/posters/x/2086472319141110032.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 vs Seedance 2.5 \nHere’s the real difference with the same prompt.\n\nLimited offer:60% off",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 113,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086471891498484171",
    "title": "红衣人物独立雨夜城市天台",
    "titleEn": "A Red-Coated Figure Watches a Rainy City from a Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一名穿红色长外套的人物背对镜头站在积水天台上，持续望向雨云笼罩、霓虹灯点亮的城市天际线。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure in a long red coat stands with their back to the camera on a wet rooftop, looking across a neon-lit skyline beneath heavy rain clouds. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pcneko_lab/status/2086471891498484171",
    "sourceLabel": "X 原帖 · @pcneko_lab",
    "author": "PcNeko.Lab",
    "publishedAt": "2026-08-09T15:17:18.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086471891498484171.mp4",
    "posterUrl": "/posters/x/2086471891498484171.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "RTX 4060 Ti 16GB環境で、MiniMax H3 Turbo LoRAによる5秒I2V生成に成功(∩´∀｀)∩ﾜｰｲ\n\n解像度：864×480、24fps\n\nCodexに丸投げ生成したので細かい設定は見とらんですはい。🫠",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 9,
      "views": 394,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086471165581902327",
    "title": "两名冰川学者深入蓝色冰洞",
    "titleEn": "Two Glaciologists Enter a Blue-Ice Cave",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头先俯瞰冰川上的狭小洞口，随后两名戴头盔、背装备的研究人员进入半透明蓝色冰壁围成的通道。 成片规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The sequence opens above a small entrance in a glacier before two helmeted researchers carrying equipment move into a passage enclosed by translucent blue ice walls. Output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "EXPEDITION:\nTwo glaciologists enter a newly formed ice cave to document its internal structure before seasonal melting changes it.\nEXPLORERS: Two researchers, both wearing crampons, helmets and headlamps, carrying compact measurement equipment.\nSHOOT WINDOW: 7:05 AM – 7:20 AM Early morning outside. Overcast sky. Cold blue daylight enters through the cave entrance.\nLOCATION: Glacial valley, massive blue ice formation, narrow cave entrance, translucent walls, frozen water flowing beneath the surface.\nRECORDING STYLE: First-person expedition footage mixed with a second camera operator. Headlamps create moving pools of light across the ice. Natural breathing and footsteps remain prominent.\n15-SECOND EXPLORATION:\n00:00–00:03 → Researchers squeeze through the narrow entrance.\n00:03–00:06 → Camera reveals deep blue translucent ice walls surrounding them.\n00:06–00:09 → One researcher notices water suddenly moving beneath a transparent ice floor.\n00:09–00:12 → Both carefully step backward and examine the changing surface.\n00:12–00:15 → They mark the location and begin retreating toward daylight.\nAUDIO: Boots scraping ice, dripping water, distant cracking, breathing, muffled voices.\nREALISM DIRECTIVE: Ice must behave like real compressed glacier ice, not glass. Light scatters through the walls naturally. No fantasy formations, supernatural elements or impossible cave geometry.",
    "sourceUrl": "https://x.com/Strength04_X/status/2086471165581902327",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-09T15:14:25.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086471165581902327.mp4",
    "posterUrl": "/posters/x/2086471165581902327.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ice Cave Exploration 🏔️\n\nCreated this using MiniMax H3 on @wavespeed_ai \n\nPrompt - EXPEDITION:\nTwo glaciologists enter a newly formed ice cave to document its internal structure before seasonal melting changes it.\nEXPLORERS: Two researchers, both wearing crampons, helmets and headlamps, carrying compact measurement equipment.\nSHOOT WINDOW: 7:05 AM – 7:20 AM Early morning outside. Overcast sky. Cold blue daylight enters through the cave entrance.\nLOCATION: Glacial valley, massive blue ice formation, narrow cave entrance, translucent walls, frozen water flowing beneath the surface.\nRECORDING STYLE: First-person expedition footage mixed with a second camera operator. Headlamps create moving pools of light across the ice. Natural breathing and footsteps remain prominent.\n15-SECOND EXPLORATION:\n00:00–00:03 → Researchers squeeze through the narrow entrance.\n00:03–00:06 → Camera reveals deep blue translucent ice walls surrounding them.\n00:06–00:09 → One researcher notices water suddenly moving beneath a transparent ice floor.\n00:09–00:12 → Both carefully step backward and examine the changing surface.\n00:12–00:15 → They mark the location and begin retreating toward daylight.\nAUDIO: Boots scraping ice, dripping water, distant cracking, breathing, muffled voices.\nREALISM DIRECTIVE: Ice must behave like real compressed glacier ice, not glass. Light scatters through the walls naturally. No fantasy formations, supernatural elements or impossible cave geometry.",
    "engagement": {
      "replies": 12,
      "reposts": 6,
      "likes": 47,
      "views": 1176,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086470723472933293",
    "title": "红蓝巷道里的白发动漫少女",
    "titleEn": "A White-Haired Anime Girl in a Red-and-Blue Alley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名白发动漫少女站在红蓝高反差的狭窄巷道中央，左侧是滚筒机器，右侧散落着袋装物品；三帧画面基本一致。 成片规格为 5 秒 · 1080×1620。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired anime girl stands in the center of a narrow red-and-blue alley, with a drum machine on the left and bags scattered on the right; the three frames are nearly identical. Output: 5s · 1080×1620. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NASNETou/status/2086470723472933293",
    "sourceLabel": "X 原帖 · @NASNETou",
    "author": "那須音トウ / Nasune Tou",
    "publishedAt": "2026-08-09T15:12:39.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086470723472933293.mp4",
    "posterUrl": "/posters/x/2086470723472933293.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1080×1620",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "OC 音注意 (minimax h3)",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 255,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086470421986340918",
    "title": "月夜弹幕中的红衣飞行少女",
    "titleEn": "A Red-Clad Flyer Dodges Projectiles Under the Moon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一名红衣动漫少女在满月与群山上空飞行，穿过成排光束与环形弹幕；右下角小窗同步显示简化的三维运动参考。 成片规格为 5 秒 · 768×512。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-clad anime girl flies above mountains beneath a full moon while crossing rows of light beams and ring-shaped projectiles, with a simplified 3D motion reference shown in an inset. Output: 5s · 768×512. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yukimino_doga/status/2086470421986340918",
    "sourceLabel": "X 原帖 · @yukimino_doga",
    "author": "ゆきしたミノル@動画",
    "publishedAt": "2026-08-09T15:11:28.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086470421986340918.mp4",
    "posterUrl": "/posters/x/2086470421986340918.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "768×512",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ローカルMinimax H3のR2Vで、Unityで撮った3Dの弾幕動画をアニメ化してみた🧐\n使い道はあると思うんだけど、3D動画のほうはエージェント任せだとちょいイマイチなのが多いかな😅 手動が必要～\n（BGMは後付けです）",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 18,
      "views": 1438,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086468076032450920",
    "title": "白发占卜师揭晓深蓝幸运色",
    "titleEn": "A White-Haired Fortune Teller Reveals Deep Blue",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "华丽边框的占卜栏目标题卡过后，一名戴眼镜的白发角色举起写有“Lucky Color: Deep Blue”的小牌，并朝镜头做出手势。 成片规格为 10 秒 · 896×1184。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "After an ornate fortune-segment title card, a white-haired character with glasses raises a sign reading “Lucky Color: Deep Blue” and gestures toward the camera. Output: 10s · 896×1184. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nayu_pkpk/status/2086468076032450920",
    "sourceLabel": "X 原帖 · @nayu_pkpk",
    "author": "nayucchi（グラボあちち勢）",
    "publishedAt": "2026-08-09T15:02:08.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086468076032450920.mp4",
    "posterUrl": "/posters/x/2086468076032450920.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "896×1184",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n有紀葉の今日のうりゃないコーナー！（噛んだ笑）\n星座とラッキーカラー選択はLLM任せのワークフローなので1日1回生成して今日1日を占ってくれるぞ！",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 134,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086467465937330555",
    "title": "巨月宫墙下的白衣女子与云端棋局",
    "titleEn": "White-Robed Woman and Cloudside Go Beneath a Giant Moon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白衣女子先走过巨月映照的红色宫墙，画面随后切到两名身着黑白长袍的人在云海边的木桌上下围棋。 成片规格为 68 秒 · 1640×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-robed woman first walks beside red palace walls beneath a giant moon before the scene cuts to two black-and-white-robed figures playing Go at a wooden table above a sea of clouds. Output: 68s · 1640×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotoX86/status/2086467465937330555",
    "sourceLabel": "X 原帖 · @PhotoX86",
    "author": "PhotoX86",
    "publishedAt": "2026-08-09T14:59:43.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086467465937330555.mp4",
    "posterUrl": "/posters/x/2086467465937330555.jpg",
    "duration": 68,
    "aspectRatio": "landscape",
    "resolution": "1640×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "中式美学幻境2，我更喜欢简约留白不堆砌的朴素禅意，\n所以我在各位大佬们的爆款提示词基础上都做了调整修改\n让它们更符合我的喜好\nMinimax H3制作\n没做竖幅的了，因为宽幅看起来更空灵，21:9画幅，不敢设置太大尺寸，\nRunningHUB都爆显存，所以大场景中人脸有些崩坏",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 188,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086467199393464387",
    "title": "白衣男子走过云海天宫长桥",
    "titleEn": "A White-Robed Man Walks a Palace Bridge Above the Clouds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名白衣男子从高大石柱之间走出，沿着悬在云海上方的白色宫殿长桥前行，远处建筑在薄雾中若隐若现。 成片规格为 64 秒 · 1640×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-robed man emerges between tall stone columns and walks along a white palace bridge suspended above a sea of clouds, with distant structures fading into the mist. Output: 64s · 1640×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotoX86/status/2086467199393464387",
    "sourceLabel": "X 原帖 · @PhotoX86",
    "author": "PhotoX86",
    "publishedAt": "2026-08-09T14:58:39.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086467199393464387.mp4",
    "posterUrl": "/posters/x/2086467199393464387.jpg",
    "duration": 64,
    "aspectRatio": "landscape",
    "resolution": "1640×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "我更喜欢简约留白不堆砌的朴素禅意\n所以我在爆款提示词基础上都做了调整修改\n让它们更符合我的喜好\nMinimax H3制作\n没做竖幅的了，因为宽幅看起来更空灵，\n21:9画幅，不敢设置太大尺寸，RunningHUB都爆显存，所以大场景中人脸有些崩坏",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 172,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086466067854168379",
    "title": "机甲战机冲向燃烧的轨道空间站",
    "titleEn": "A Mecha Fighter Rushes Toward a Burning Orbital Station",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多架战机接近行星上方起火的轨道空间站，随后一台人形机甲从镜头前高速掠过并朝爆炸区域推进。 成片规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Several fighters approach a burning orbital station above a planet before a humanoid mecha sweeps past the camera and advances toward the exploding structure. Output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/developerlin/status/2086466067854168379",
    "sourceLabel": "X 原帖 · @developerlin",
    "author": "Frank Lin",
    "publishedAt": "2026-08-09T14:54:09.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086466067854168379.mp4",
    "posterUrl": "/posters/x/2086466067854168379.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Looks good.\n\nMiniMax H3\nRTX 5070 12G\n480p 15s\n\nThe scene and character images are generated by gpt-image2",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 126,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086465855966372247",
    "title": "舞台女乐队与浮现的角色设定图",
    "titleEn": "A Female Rock Band and a Ghosted Character Sheet",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三人女子乐队站在灯光舞台上演奏，中间与末尾画面叠入半透明角色设定图，人物面部和身体也随之发生明显变形。 成片规格为 5 秒 · 1152×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A three-member female rock band performs on a lit stage while a translucent character sheet bleeds into the later frames and visibly distorts the performers' faces and bodies. Output: 5s · 1152×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/willowion/status/2086465855966372247",
    "sourceLabel": "X 原帖 · @willowion",
    "author": "𝖂𝖎𝖑𝖑𝖔𝖂",
    "publishedAt": "2026-08-09T14:53:19.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086465855966372247.mp4",
    "posterUrl": "/posters/x/2086465855966372247.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax_h3_turbo_v4_step600_ema.safetensors\n\n4ステップにしてみたら参照用のキャラクターシートが薄っすら登場👻\n\nよし、8ステップにしてみよう🐾\n\nMiniMax-H3 Turbo LoRA\n🔗https://huggingface.co/larryvrh/MiniMax-H3-Turbo-Lora\n\n#MiniMax_H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 211,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086465794792439819",
    "title": "银发女子在夕照天台挥手",
    "titleEn": "A Silver-Haired Woman Waves on a Rooftop at Sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名佩戴白色花饰、身穿白衣的银发女子站在夕阳下的天台，先抬手挥动，随后轻触头发并靠近镜头微笑。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired woman in white with a flower hairpiece stands on a rooftop at sunset, waving before touching her hair and smiling closer to the camera. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/erukoed/status/2086465794792439819",
    "sourceLabel": "X 原帖 · @erukoed",
    "author": "出遅れ",
    "publishedAt": "2026-08-09T14:53:04.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086465794792439819.mp4",
    "posterUrl": "/posters/x/2086465794792439819.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3で5秒動画作ってみたけどスペック高くないPCでもこのレベルを5,6分で作れるのはすごい。変な喋りになっているのはテストで作ったのでプロンプト間違っている人側のミス。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 141,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086464755628376196",
    "title": "女子从室内招手到超市购物与床边取物",
    "titleEn": "A Woman Greets, Shops, and Reaches for Her Phone",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏片段依次出现女子在室内朝镜头招手、拎着购物篮走过超市蔬果区，以及一只手伸向床头柜上的手机。 成片规格为 61 秒 · 720×1252。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The vertical sequence shows a woman waving toward the camera indoors, carrying a basket through a supermarket produce aisle, and a hand reaching for a phone on a bedside table. Output: 61s · 720×1252. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shiokotoba/status/2086464755628376196",
    "sourceLabel": "X 原帖 · @shiokotoba",
    "author": "koko",
    "publishedAt": "2026-08-09T14:48:57.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086464755628376196.mp4",
    "posterUrl": "/posters/x/2086464755628376196.jpg",
    "duration": 61,
    "aspectRatio": "portrait",
    "resolution": "720×1252",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今回は画質を意識したので、かなり良くなった。次はもっとリアル寄りの画像が良さそう。5090を一日回せばMinimax H3で24分くらい作れそうだけど、家が蒸し風呂になるしファンもうるさいｗ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 34,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086464559439855938",
    "title": "漆线雕花瓶从捻土盘线到上架",
    "titleEn": "A Lacquer-Thread Vase from Clay Rolling to Display",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "工匠先把材料捻成细线，再将金色纹样盘贴并雕饰在红色花瓶表面，末帧双手把成品放上展台。 成片规格为 15 秒 · 2560×1440。原作者在回复区完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An artisan rolls material into fine strands, applies and carves gold ornament across a red vase, and finally places the completed piece on a display stand. Output: 15s · 2560×1440. The creator published the complete prompt in a reply, preserved here verbatim.",
    "prompt": "【参考图规则】\n@图片1 是最终漆线雕作品的唯一权威参考。 必须严格保持图片1中的作品造型、主体比例、漆面颜色、金色漆线纹样、纹样走向、器型和整体气质。 不得随意增加龙、凤、佛像、文字或不存在的装饰。\n @图片2 是包装盒、包装纸、品牌标识以及产品标签的唯一权威参考。 所有包装结构、颜色和标识均按照图片2执行。\n@图片3 是厦门传统漆线雕工作室的环境与工匠服装参考。 保持真实闽南手工作坊质感，不做古装影视化处理，不制造虚假的“古代作坊”。\n\n【目标】  制作一支15秒、9:16竖版的厦门漆线雕非遗工艺短片。  主题：  “一根线，走完三百年的手艺。”  影片完整呈现：  备料 → 舂打漆线土 → 搓线 → 盘线成纹 → 安金 → 贴金 → 清理完成 → 包装 → 上架  快速剪辑， 字幕驱动， 即使完全静音观看也能理解整个制作过程。  视觉采用真实高端人文纪录广告摄影：  35毫米胶片电影质感， 细腻颗粒， 浅景深， 大量微距， 自然手持， 运动非常克制， 不做旅游宣传片， 不做华丽国潮特效， 不做博物馆宣传片。  重点永远是：  手， 线， 漆， 金箔， 纹样， 时间。\n\n【金色就是时间】  整支影片必须存在一条不可逆的视觉变化：  画面中的“金色”从无到有，并越来越多。  影片开始：  只有黑、 深褐、 砖灰、 暗红， 几乎看不到金色。  漆线开始盘绕以后， 画面出现第一点暖金。  贴金以后， 金色迅速扩大。  作品完成时， 精细金色纹样覆盖主体。  最后上架时， 整个作品被温暖自然光照亮。  每一个镜头都必须比前一个镜头拥有更多的金色。  这个变化不能倒退。  金色就是这支影片的时间。\n\n【固定字幕牌】  整个影片始终只有一个字幕牌。  位置：  画面下方三分之一， 水平居中， 所有镜头完全相同的位置。  造型：  非常克制的小型圆角矩形， 深朱砂红底， 暖米白文字。  两行文字：  第一行： 较细字体， 显示“工序”。  第二行： 粗体， 显示一句极短的动作描述。  字幕牌：  不移动， 不放大， 不缩小， 不淡入淡出， 不跳动。  只在指定剪辑点瞬间更换文字。  字体必须清楚、正确。\n\n【镜头序列】\n 0—1.7秒  主镜头。  深夜般昏暗的传统工作台。  一盏暖色工作灯只照亮双手。  桌面上可以看到：  陈年砖粉质感的细粉、 深色大漆材料、 正在被反复捶打揉合的深褐色漆线土。  工匠双手将材料反复捶、压、揉， 逐渐形成柔软、富有韧性的泥团。  周围环境全部沉入黑暗。  没有金色。  字幕：  “第一道” “捶土成线”\n1.7—3.3秒  极端微距。  一小块漆线土放在传统搓线板之间。  工匠双手稳定向前推动搓板。  原本粗厚的泥条逐渐被搓成长而均匀的细线。  摄影机贴得非常近。  能够清楚看到：  漆线轻微湿润的表面， 细小纹理， 手指压力， 线条被不断拉细。  背景完全虚化。  字幕保持：  “第一道” “捶土成线”\n 3.3—5.0秒  主镜头。  朱红漆面的器物坯体第一次出现。  工匠用极细工具， 将刚刚搓好的柔软漆线轻轻落在器物表面。  第一根漆线贴上去。  只有一根。  然后第二根。  线条开始形成第一个小小的卷云纹。  画面第一次出现非常微弱的暖色高光。  字幕：  “第二道” “一线起纹”\n 5.0—6.8秒  微距插入镜头。  镜头几乎贴着器物表面横向观察。  工匠用竹制细工具推动漆线：  盘， 绕， 结， 堆。  几根不足毫米级视觉尺度的柔软漆线， 逐渐形成具有明显高度差的浮凸纹样。  线条之间非常紧密。  卷云、缠枝、如意形态逐渐出现。  重点展示：  线并不是画上去的，  而是真正一根一根盘出来的。  字幕保持：  “第二道” “一线起纹”\n6.8—8.6秒  主镜头。  器物已经完成大部分漆线纹样。  摄影机非常缓慢地向前推近。  工匠旋转器物。  光线擦过密集漆线表面。  能够看到：  高低起伏， 盘绕结构， 层层叠叠的线性浮雕。  这时依然没有真正的金箔。  只有漆线自身的暖棕色。  字幕：  “第三道” “盘线成雕”\n8.6—10.2秒  全片最重要的声音与视觉高潮。  极端微距。  一张极薄金箔悬在空气里轻微颤动。  几乎没有重量。  工匠屏住动作， 用传统工具将金箔缓缓落向已经完成的漆线纹样。  金箔接触表面的瞬间，  贴住。  镜头保持。  不要快速切走。  细小金箔自然贴合漆线的高低起伏， 金色第一次大面积出现。  字幕：  “第四道” “一片金落下”\n10.2—11.7秒  微距。  柔软毛刷轻轻扫过作品表面。  多余金箔碎屑被一点点扫开。  随着刷毛经过，  清晰、 锐利、 细密的金色漆线纹样从杂乱金箔中显现。  这是第二个视觉高潮。  镜头必须清楚表现：  杂乱金箔  变成  精确金色纹样。  字幕：  “第五道” “金纹醒来”\n11.7—12.8秒  主镜头。  完成后的作品放在深色木质工作台上。  工匠用双手缓慢旋转检查。  没有任何戏剧表演。  只检查：  纹样， 漆面， 金箔， 边缘， 细节。  自然光已经明显进入工作室。  金色纹样被晨光照亮。  字幕：  “第六道” “最后一眼”\n12.8—13.8秒  插入镜头。  作品被轻轻放入定制包装盒。  暖米色软布包裹器物。  盒盖缓慢合上。  手将包装盒推向画面前方。  纸张、 布料、 木盒均保持真实触感。  不要奢侈品浮夸包装。  字幕：  “第七道” “入盒”\n13.8—15秒  最终镜头。  切到极简现代非遗精品店。  温暖自然晨光。  深色木制陈列架。  一只手将完成的漆线雕作品从画面侧面轻轻放到展示台中央。  手退出画面。  作品保持完全静止。  金色漆线在自然光中形成极精细的浮雕反光。  背景虚化。  最后0.6秒不再运动。  字幕：  “第八道” “上架”  画面中央仅允许出现一句极小的中文：  “厦门漆线雕”  下面更小：  “一线成雕”  最终画面必须像高端工艺品牌的静态产品摄影。  绝对不要继续运动。\n\n【色彩】  整支影片限制在以下颜色：  陈年砖粉灰， 生漆深褐， 木质黑褐， 朱砂红， 暖米白， 金箔金。  影片前半段：  黑褐色占主导。  中段：  朱红逐渐出现。  后半段：  金色逐渐占据视觉中心。  禁止：  青蓝科技色， 霓虹， 紫色， 赛博朋克， 高饱和国潮配色。  金色必须来自真实金箔和真实光线。  不能使用发光特效制造金色。\n\n【摄影】  全片使用真实人文纪录摄影语言。  大量：  100毫米微距镜头， 近距离手部特写， 极浅景深， 35毫米胶片颗粒。  摄影机运动非常少。  允许：  轻微呼吸感手持， 一次非常缓慢的轴向推进， 极细微的跟随动作。  禁止：  无人机， 环绕运镜， 快速推拉， 旋转镜头， 甩镜， 变焦冲击， 慢动作， 速度渐变， 镜头炫技。  镜头永远尊重手艺本身。\n\n【声音】  全片没有任何音乐。  没有旁白。  没有人物对白。  没有广告配音。  只使用现场真实声音。\n 0—3.3秒：  凌晨安静工作室。  远处非常轻的环境底噪。  漆线土落在木板上的闷响。  手掌揉压材料的湿润摩擦声。  搓板来回运动产生细密摩擦声。   3.3—8.6秒：  整体声音进一步降低。  细工具轻碰器物表面的声音。  手指移动。  漆线被放下时极轻微的黏连声音。  器物缓慢旋转时木架产生细小摩擦声。   8.6—10.2秒：  这是整支影片最重要的声音时刻。  其他环境声音全部压低。  只留下：  金箔极薄的纸张颤动声， 轻微呼吸， 工具触碰金箔的细小声音。  金箔落到漆线表面的瞬间，  接近安静。  让观众感觉自己离这张金箔只有几厘米。   10.2—12.8秒：  软毛刷扫过金箔。  细碎金箔摩擦。  器物在木架上轻轻旋转。   12.8—15秒：  包装纸折叠。  布料摩擦。  盒盖轻轻合上。  随后进入店铺：  作品底座接触木质展台，  “嗒”。  最后接近完全安静。  只保留非常微弱的室内环境声。\n\n【核心视觉原则】  不要把它拍成：  “传统文化宣传片”。  要把它拍成：  “世界顶级手工艺品牌新品诞生纪录”。  传统来自材料和工艺，  高级感来自摄影、节奏和克制。\n\n【禁止项】  不得出现任何与产品无关的文字。  不得出现旅游宣传标语。  不得出现：  “匠心传承” “非遗之美” “千年文化” “东方美学” 等泛化宣传口号。  不得出现不存在的历史年代。  不得出现虚假的皇宫或寺庙场景。  不得增加和尚、古装人物、舞狮、灯笼等所谓中国元素。  不得把漆线雕表现成木雕、石雕或刀刻。  重点必须明确表现：  搓线， 放线， 盘线， 绕线， 堆线， 贴金。  不得用三维特效让纹样自动生长。  不得让漆线自行移动。  所有纹样必须由人的手完成。  不得使用魔法粒子。  不得让金箔发光。  不得使用人工光晕。  不得使用慢动作。  不得加入速度渐变。  不得加入转场音效。  不得使用呼啸声。  不得出现手机界面。  不得出现电商界面。  不得出现价格。  不得出现二维码。  不得出现网站。  不得出现社交媒体账号。  不得出现水印。  不得增加第十个镜头。  最终画面完全静止。",
    "sourceUrl": "https://x.com/derek_wall90176/status/2086464559439855938",
    "sourceLabel": "X 原帖 · @derek_wall90176",
    "author": "Derek Wen｜德里克文",
    "publishedAt": "2026-08-09T14:48:10.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086464559439855938.mp4",
    "posterUrl": "/posters/x/2086464559439855938.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "继续来分享采用MiniMax H3生成带给我的惊喜。\n这次我用AI视频来生成15秒精炼的厦门漆线雕工艺制作效果。\n通过AI视频，仅仅通过提示词就能够生成完整的宣传片效果，虽然存在瑕疵，但是效率太高了。\n提示词我放评论区。",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 6,
      "views": 1551,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review",
    "promptSourceUrl": "https://x.com/derek_wall90176/status/2086465595055501467"
  },
  {
    "id": "x-2086461975727370440",
    "title": "彩色绒球化作烟雾拼出Tim",
    "titleEn": "A Colorful Pom-Pom Becomes Smoke Spelling Tim",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一个带有粉、黄、蓝、绿绒毛的圆球先向外舒展，随后化成大团彩色烟雾，并在中央显出手写体“Tim”。 成片规格为 5 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A round pom-pom of pink, yellow, blue, and green fibers expands outward, turns into clouds of colored smoke, and reveals the handwritten word “Tim” at the center. Output: 5s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mann_hodge/status/2086461975727370440",
    "sourceLabel": "X 原帖 · @mann_hodge",
    "author": "Ȟ̞̲̥̺͔ͬö͈̩̟̱̺̖ͅdͫg̎̊e̲̗̞͓̰͎ M̌a̹̤̻̠ṉ̠̦̦̘̼ͦnͤ̾",
    "publishedAt": "2026-08-09T14:37:54.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086461975727370440.mp4",
    "posterUrl": "/posters/x/2086461975727370440.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Image fed into my Infinite Ident instructions for MiniMax H3.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 7,
      "views": 244,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086460162689450397",
    "title": "电脑前大笑男子与突然燃烧的住宅",
    "titleEn": "A Laughing Computer User and a House Suddenly Ablaze",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿衬衫与背心的男子坐在电脑前大笑，画面随后切到一栋住宅被高耸火焰吞没。 成片规格为 8 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in a shirt and vest laughs while seated at a computer before the scene cuts to a house engulfed by towering flames. Output: 8s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hogehoge61/status/2086460162689450397",
    "sourceLabel": "X 原帖 · @hogehoge61",
    "author": "ほげ",
    "publishedAt": "2026-08-09T14:30:41.000Z",
    "addedAt": "2026-08-10T05:52:30.476Z",
    "mediaUrl": "/media/x-2086460162689450397.mp4",
    "posterUrl": "/posters/x/2086460162689450397.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で動画が生成出来るからと言って、使う人が面白くないんじゃあねえ…… (何も思いつきません)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 194,
      "snapshotAt": "2026-08-10T05:52:30.476Z"
    },
    "approvedAt": "2026-08-10T05:52:30.476Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086031009402003468",
    "title": "花纹泳装女子坐在沙滩轻理长发",
    "titleEn": "A Woman in a Patterned Swimsuit Adjusts Her Hair on the Beach",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿花纹两件式泳装的长发女子坐在沙滩上，先看向侧面，再抬手整理头发并转向另一侧。 成片规格为 10 秒 · 1776×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-haired woman in a patterned two-piece swimsuit sits on a beach, looks to one side, raises a hand to adjust her hair, and turns the other way. Output: 10s · 1776×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AiPhotorealGirl/status/2086031009402003468",
    "sourceLabel": "X 原帖 · @AiPhotorealGirl",
    "author": "AIconia | アイコニア",
    "publishedAt": "2026-08-08T10:05:23.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086031009402003468.mp4",
    "posterUrl": "/posters/x/2086031009402003468.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1776×1344",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "最近話題のMiniMax H3を\nローカル環境(RTX 4070)で試してみました\n\n正直ローカルでの動画生成は\n時間も掛かってクオリティが微妙\nというイメージでしたが\n有志の高速化のお陰もあって\n0.4MPixelの5秒動画なら\n３分以内で出力できて\nクオリティもよかったです\n1年前を考えると恐ろしい速さです\n\nそれで1MPixel10秒で試したら\n30分掛かりました\nさすがに厳しい😅\n\nということで\nこの動画はChatGPTで生成した画像から\n1MPixelで動画生成\nアップスケールとフレーム補間をしたものです\n0.5MPixel相当なら実用的な気がします😌",
    "engagement": {
      "replies": 23,
      "reposts": 75,
      "likes": 1759,
      "views": 308901,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "attributionNote": "Resolved from the quoted post and native-media publisher metadata.",
    "editorialBasis": "three-frame-media-review",
    "featured": {
      "at": "2026-09-05T15:50:15.692Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2086458310006600101",
    "title": "女学生推开锈迹斑驳的发光铁门",
    "titleEn": "A Schoolgirl Opens a Rusted Door Glowing from Within",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名背书包的女学生站在斑驳水泥墙前，近景过后她推开锈迹明显的铁门，门内透出暖黄色灯光。 成片规格为 7 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A schoolgirl with a backpack stands before a weathered concrete wall, and after a close-up she opens a visibly rusted metal door lit by warm yellow light from within. Output: 7s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aaaa_2469/status/2086458310006600101",
    "sourceLabel": "X 原帖 · @aaaa_2469",
    "author": "ああああ",
    "publishedAt": "2026-08-09T14:23:20.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086458310006600101.mp4",
    "posterUrl": "/posters/x/2086458310006600101.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3で作成\n引いて開けるタイプの覗き窓の事を英語でなんて言うのか分からず、色々な英訳を入れても丸い覗き穴がスライドする変な映像になる\nギャング映画で見るやつとか説明してもAIには伝わらなかった",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 330,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086457343995478450",
    "title": "帝企鹅在雪地跌倒后抬腿站起",
    "titleEn": "An Emperor Penguin Falls and Gets Back Up on Snow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一只帝企鹅在企鹅群前向镜头靠近，随后侧倒在雪地上，末帧重新站起并高高抬起一条腿。 成片规格为 7 秒 · 448×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An emperor penguin approaches the camera in front of a colony, falls onto its side in the snow, and rises again with one leg lifted high. Output: 7s · 448×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mike_at_cg/status/2086457343995478450",
    "sourceLabel": "X 原帖 · @mike_at_cg",
    "author": "keim_at_si",
    "publishedAt": "2026-08-09T14:19:29.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086457343995478450.mp4",
    "posterUrl": "/posters/x/2086457343995478450.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "448×672",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 やってみた！楽しい！",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 101,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086456898791092583",
    "title": "蓝衣笠帽舞者列队穿过灯笼街道",
    "titleEn": "Blue-Clad Dancers Move in Formation Beneath Street Lanterns",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多名头戴斗笠、身穿蓝色和服的舞者在挂满橙色灯笼的街道上排成长队，以一致的手臂姿势向前移动。 成片规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Dancers in blue kimono and conical hats form a long line beneath orange street lanterns, moving forward with matching arm positions. Output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ivy_memos/status/2086456898791092583",
    "sourceLabel": "X 原帖 · @ivy_memos",
    "author": "アイビー",
    "publishedAt": "2026-08-09T14:17:43.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086456898791092583.mp4",
    "posterUrl": "/posters/x/2086456898791092583.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で生成したのですが、、これは阿波踊りではないな...",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 333,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086455911216971949",
    "title": "猫耳少女转身后挥拳冲向镜头",
    "titleEn": "A Cat-Eared Girl Spins and Punches Toward the Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名头顶小黑猫的猫耳少女先摆出拳击姿势，接着转身蓄力，最后把拳头挥向镜头形成近距离特写。 成片规格为 1 秒 · 416×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared girl with a small black cat on her head takes a boxing stance, turns to wind up, and finishes with a fist filling the frame. Output: 1s · 416×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mizutamaaiart/status/2086455911216971949",
    "sourceLabel": "X 原帖 · @mizutamaaiart",
    "author": "みずたま",
    "publishedAt": "2026-08-09T14:13:48.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086455911216971949.mp4",
    "posterUrl": "/posters/x/2086455911216971949.jpg",
    "duration": 1,
    "aspectRatio": "portrait",
    "resolution": "416×608",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "猫ぱんち！\n（お試しMiniMax H3、音声削除）",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 15,
      "views": 343,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086455504969281685",
    "title": "戴像素头盔的小猫挥动紫色镐",
    "titleEn": "A Kitten in Pixel Armor Swings a Purple Pickaxe",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "真实小猫的脸被方形像素头盔包围，前景叠有像素图腾与紫色镐，镐的位置在三帧之间发生变化。 成片规格为 4 秒 · 1280×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A real kitten's face is framed by a blocky pixel helmet, with a pixel totem and purple pickaxe overlaid in front as the pickaxe changes position across the frames. Output: 4s · 1280×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiaicreate/status/2086455504969281685",
    "sourceLabel": "X 原帖 · @aiaicreate",
    "author": "R@aiaicreate",
    "publishedAt": "2026-08-09T14:12:11.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086455504969281685.mp4",
    "posterUrl": "/posters/x/2086455504969281685.jpg",
    "duration": 4,
    "aspectRatio": "square",
    "resolution": "1280×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3ベースの生成ワークフロー「A Very Generic Workflow」が公開。\n低スペックPCでの動作を想定した汎用設計。\nVRAM 12GB、RAM 16GBの環境に対応。\n限られたリソースでの生成を支援。\n\n#MiniMaxH3 #Civitai\n\nURLはリプ⬇️",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 9,
      "views": 1312,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086455274898870431",
    "title": "三栏对照中的橙色机器人与剪辑师",
    "titleEn": "Three-Way Comparison of an Orange Robot and Editor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三个横向排列的版本同时呈现橙色人形机器人与黑衣男子在电脑前互动，后续分别出现剪辑屏幕、机器人按键和男子惊讶的近景。 成片规格为 10 秒 · 1280×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three stacked versions show an orange humanoid robot interacting with a man at a computer, later diverging into editing screens, a robot pressing keys, and a close-up of the man's surprised face. Output: 10s · 1280×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MattPenny99/status/2086455274898870431",
    "sourceLabel": "X 原帖 · @MattPenny99",
    "author": "Matt Penny",
    "publishedAt": "2026-08-09T14:11:16.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086455274898870431.mp4",
    "posterUrl": "/posters/x/2086455274898870431.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1280×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.5 vs Minimax H3 vs Omni Flash\n\nI used the same prompt and starting frame for all three models in order to create b-roll in an upcoming YT video about AI video editing.\n\nRough costs\n\nMinimax H3 - $0.79\nOmni Flash - Free in Google Flow\nSeedance 2.5 - $3.60\n\nWhen factoring in cost effectiveness, Omni wins. Even if you use it at API costs rather than not through Google Flow. \n\nLikely can get a better result from Seedance with more reference images and more specific prompting.\n\nCan't blame them for not recreating my face that well as the only reference it has is the starting frame.",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 2,
      "views": 230,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086454834748903558",
    "title": "校服少女在走廊饮用瓶装果汁",
    "titleEn": "A Schoolgirl Drinks Bottled Juice in a Sunlit Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿校服的少女站在阳光照亮的学校走廊，打开装有黄色饮料的塑料瓶，仰头饮用后举着瓶子微笑。 成片规格为 8 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A uniformed schoolgirl stands in a sunlit school corridor, opens a plastic bottle containing a yellow drink, takes a sip, and smiles while holding it up. Output: 8s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/morita_desu/status/2086454834748903558",
    "sourceLabel": "X 原帖 · @morita_desu",
    "author": "ɐʇıɹoɯ",
    "publishedAt": "2026-08-09T14:09:31.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086454834748903558.mp4",
    "posterUrl": "/posters/x/2086454834748903558.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ComfyUI + MiniMax H3で、よく女子高生がペットボトルのジュースを飲む様子みたいなのがアップされてるけど、そういうのは比較的簡単にできるんだなぁと。\n別に女子高生マニアとかではないです(笑)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 136,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086454188058587341",
    "title": "白发剑士穿越爆炸屋顶挥出弧光",
    "titleEn": "A White-Haired Swordswoman Cuts Through an Exploding Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名白发剑士手持发光长刀在雨夜屋顶疾跑，背后建筑爆炸，随后她跃起并挥出蓝色弧形光刃。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired swordswoman runs across a rain-soaked rooftop with a glowing blade as a building explodes behind her, then leaps and releases a blue arc of light. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/syzmk02/status/2086454188058587341",
    "sourceLabel": "X 原帖 · @syzmk02",
    "author": "しゅーま",
    "publishedAt": "2026-08-09T14:06:57.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086454188058587341.mp4",
    "posterUrl": "/posters/x/2086454188058587341.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "miniMAX-H3もっと早く、画質よくできないかなって探してたけどもうすでに色々と出てきている模様\n早いなぁ\n色々な画風の画像をリファレンスに試してみたけどちゃんと合わしてくるのすごいな\nすぐにリアル系にしたがらない",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 107,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086453457808920930",
    "title": "西部酒馆牛仔走向窗边钢琴",
    "titleEn": "A Cowboy Walks to a Piano in a Western Saloon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴宽檐帽的牛仔穿过昏暗的西部酒馆，走到窗边的立式钢琴前坐下并把双手放到琴键上。 成片规格为 20 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A broad-brimmed cowboy crosses a dim Western saloon, sits at an upright piano beside a window, and places both hands on the keys. Output: 20s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LabMike3D/status/2086453457808920930",
    "sourceLabel": "X 原帖 · @LabMike3D",
    "author": "Lab Mike 3D",
    "publishedAt": "2026-08-09T14:04:03.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086453457808920930.mp4",
    "posterUrl": "/posters/x/2086453457808920930.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Upgrade. Fixed audio and 4-step Turbo LoRA integration for low-end/Low VRAM GPUs. MM H3 is constantly evolving. Check out my Hugging Face posts for the best tips. Cheers!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 157,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086453358668509563",
    "title": "樱花飘落时少女倚靠城市露台",
    "titleEn": "A Girl Rests on a City Terrace as Petals Fall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名扎着深色长发的动漫少女倚在暮色城市露台的栏杆旁，粉色花瓣从她身前飘过，她缓慢转动头部。 成片规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired anime girl rests beside the railing of a city terrace at dusk while pink petals drift past and she slowly turns her head. Output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ItsmeAjayKV/status/2086453358668509563",
    "sourceLabel": "X 原帖 · @ItsmeAjayKV",
    "author": "AJ",
    "publishedAt": "2026-08-09T14:03:39.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086453358668509563.mp4",
    "posterUrl": "/posters/x/2086453358668509563.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Update on MiniMax H3 running on my single 3090 via comfy UI.\n\nHave a workflow almost ready.\n\nH3 is good with sound effects/music as well. This is turning out to be the best video gen model that can be run in a single 3090.\n\n\"non_diegetic_music: A slow solo piano with wide spacing between the notes, soft sustained strings entering underneath about\nhalfway through, and a faint high shimmer that fades to silence at the end.\"",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 4,
      "views": 442,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086452798074540404",
    "title": "金发少女穿过热带鱼群游向镜头",
    "titleEn": "A Blonde Girl Swims Toward the Camera Through Tropical Fish",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名金发动漫少女在清澈海水中朝镜头游来，周围分布着热带鱼、气泡与珊瑚，人物面部逐渐靠近画面。 成片规格为 5 秒 · 576×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde anime girl swims toward the camera through clear water surrounded by tropical fish, bubbles, and coral, with her face moving progressively closer. Output: 5s · 576×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mohikan_head1/status/2086452798074540404",
    "sourceLabel": "X 原帖 · @mohikan_head1",
    "author": "もひかん@毛ガニ隊",
    "publishedAt": "2026-08-09T14:01:26.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086452798074540404.mp4",
    "posterUrl": "/posters/x/2086452798074540404.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3少し試してみた、うちの4070Tisで十分動く！\n音楽をプロンプトで制御したいがヨグワガンニャイ！\n#AIイラスト #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 266,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086451597761286506",
    "title": "飞船舱门两侧对视的男女乘员",
    "titleEn": "Two Crew Members Face Each Other Across a Spacecraft Hatch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "飞船内部的一名女乘员背对镜头走向圆形舱门，门另一侧站着穿格纹外套的男子，后两帧持续聚焦男子正面。 成片规格为 14 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Inside a spacecraft, a female crew member approaches a circular hatch with her back to the camera while a man in a plaid jacket stands beyond it, becoming the focus of the final two frames. Output: 14s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Bhavani_00007/status/2086451597761286506",
    "sourceLabel": "X 原帖 · @Bhavani_00007",
    "author": "Bhavy☄️",
    "publishedAt": "2026-08-09T13:56:39.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086451597761286506.mp4",
    "posterUrl": "/posters/x/2086451597761286506.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tested MiniMax H3 on @Hailuo_AI again\n\nThe Interstellar scene, but about the Gemini 3.5 Pro release 😭\n\nGave it the reference video, voice samples for all three characters, and the Gemini logo.\n\nIt cloned all three voices and kept the logo on their shirts throughout.\n\nInsane.",
    "engagement": {
      "replies": 28,
      "reposts": 3,
      "likes": 43,
      "views": 3270,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086449437275271388",
    "title": "红眼黑发角色与参数代码窗口",
    "titleEn": "A Red-Eyed Character Beside Parameter and Code Windows",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "竖屏编辑画布中是一名红眼黑发动漫角色，人物周围叠有玻璃裂纹与彩色反光；末帧缩小画布并展开代码及参数窗口。 成片规格为 85 秒 · 650×846。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical editing canvas shows a black-haired anime character with red eyes amid cracked-glass textures and colored reflections before the final frame zooms out to reveal code and parameter windows. Output: 85s · 650×846. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/luta_ai/status/2086449437275271388",
    "sourceLabel": "X 原帖 · @luta_ai",
    "author": "LUTA＠AI",
    "publishedAt": "2026-08-09T13:48:04.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086449437275271388.mp4",
    "posterUrl": "/posters/x/2086449437275271388.jpg",
    "duration": 85,
    "aspectRatio": "portrait",
    "resolution": "650×846",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "✨MiniMax H3、動画の微調整ができる値が隠れていました！\n\nComfyUIのH3のノードには出ていないんですが、モデル側に「参照をどれくらい信用するか」の設定があって、conditioningに入れれば効きます。既定は画像・動画が0.999、音声が1.0でした。\n\n出すだけのノードを20行くらいで書いて、シードを固定したまま0.999→0.9→0.7→0.5→0.3と振ってみました😊\nRTX 4090 / ComfyUI v0.31.0 / 4ステップLoRA / 同じ参照・同じプロンプトです。\n\n嬉しいのはここ👇\n\nFL2Vだと、セリフのイントネーションはシードで決まります。なので声が気に入った時にシードを動かせません。でも動きが気に入らないことがあって、これまでは打つ手がなかったです💦\nこの値を下げると、シードはそのままで動きだけ変えられました。4ステップLoRAで出ていた回転が0.3で出なくなった回もありました。\n\nRef2VAで音声を参照から持ってきている時は逆で、音は参照が押さえてくれるのでシードを振り放題です。動きを変えたければシードを変えればいい。\nなのでこの値が効いてくるのは、音をモデルに作らせている時だと思います。\n\n何が変わって何が変わらないかも見てみました👇\n変わらない：キャラの見た目、衣装、画風、背景、動きの量\n変わる：構図、表情、動きの中身\n\n0.3まで下げても絵は崩れませんでした🤔 計算上そこまで下げると参照はノイズに埋もれるはずなんですが、キャラはそのままです。\n添付の動画では前半がデフォルト値の0.999で、右手の袖の場所が気に入らなかったわけですが、後半のように0.7にすると袖だけを変化できています。これはうまくいったパターンですが、シードで良い感じの動画がでてこの部分だけ気に入らないというのがあった時には一度やってみてもいいかもしれません。もちろん絶対というわけではないですが💦\n\n注意点が2つ💦\n・FL2Vだと0.5あたりから最後のフレーム指定が効かなくなります。0.7までがまあいいところかなと。\n・値を動かすと、上げても下げても最初の2〜3秒の音声が変わる時があります。私が試した限りでは大きくイントネーションが変わることがなかったので、シードでセリフが気に入ったけど、動画が気に入らない場合には使えるかなと思っています。絶対とはいえませんが。\n\nこの画像と音声の差、たぶん狙ってやってるんだと思います。仕様書を見ると音声だけ「そのままコピー」という扱いがあって、映像側にはその言葉がないんですよね。だからH3は音を外から持ってくる前提で作られてそうだなーと感じています🤔\n\nリプにノードのコード置いておきます👇\n\n#ai #comfyui",
    "engagement": {
      "replies": 2,
      "reposts": 16,
      "likes": 131,
      "views": 11809,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086448706438713714",
    "title": "街头格斗条下交手的黄紫玩偶角色",
    "titleEn": "Yellow and Purple Mascot Fighters Trade Blows on a City Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黄衣与紫衣的两名玩偶装角色在繁忙城市街口交手，画面叠有血条、计数和连击提示等街机格斗元素。 成片规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two mascot-suited characters in yellow and purple exchange blows at a busy city crossing beneath arcade-style health bars, counters, and combo graphics. Output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C12592947592/status/2086448706438713714",
    "sourceLabel": "X 原帖 · @C12592947592",
    "author": "supercub3",
    "publishedAt": "2026-08-09T13:45:10.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086448706438713714.mp4",
    "posterUrl": "/posters/x/2086448706438713714.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3\n着ぐるみに格ゲーしてもらったテスト",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 17,
      "views": 1181,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086446187012649188",
    "title": "白裙女子迎着海边落日与风",
    "titleEn": "A Woman in White Faces the Sunset Breeze by the Sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿白色细肩带裙的长发女子侧身站在海滩，落日贴近海平面，风持续吹动她的头发。 成片规格为 5 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-haired woman in a white spaghetti-strap dress stands in profile on a beach as the sun nears the horizon and wind moves her hair. Output: 5s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ModengSir/status/2086446187012649188",
    "sourceLabel": "X 原帖 · @ModengSir",
    "author": "Modengsir AI",
    "publishedAt": "2026-08-09T13:35:09.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086446187012649188.mp4",
    "posterUrl": "/posters/x/2086446187012649188.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fashion",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ComfyUI + MiniMax H3  AI视频生成\n不用多说一句话，你会明白它的含金量\n想部署的欢迎交流沟通",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 475,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086445362126991796",
    "title": "《黎明前七分钟》中的黄手套女子",
    "titleEn": "The Yellow-Gloved Woman in Seven Minutes Before Dawn",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名神情严肃的年长女子先戴着黄色手套站在昏暗工作台前，中段出现拉紧手套的手部特写，末帧她抱着文件站在坐满观众的剧场中。 成片规格为 150 秒 · 1120×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A stern older woman first stands at a dark worktable in yellow gloves, followed by a close-up of a glove being pulled tight and a final shot of her holding a folder before a seated theater audience. Output: 150s · 1120×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2086445362126991796",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-09T13:31:53.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086445362126991796.mp4",
    "posterUrl": "/posters/x/2086445362126991796.jpg",
    "duration": 150,
    "aspectRatio": "landscape",
    "resolution": "1120×640",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3   \nQwen Image 3 Pro   \nElevenLabs  \nRTX Pro 6000 96GB   \nLocal AI    \n\n„Seven Minutes Before Dawn” \n\n@MiniMax_AI\n@NVIDIAAI\n@Alibaba_Qwen\n@ElevenLabs\n\nhttps://youtube.com/@cinesit",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 239,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086444017584140313",
    "title": "黄花田与农机的双分辨率对照",
    "titleEn": "Two-Resolution Comparison of a Yellow Flower Field and Farm Machinery",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右并排画面以 768p 和 2K 呈现同一段黄花田景象，依次出现花朵近景、贴近植株的机械滚轮和驶过田地的红色拖拉机。 成片规格为 15 秒 · 1920×816。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Side-by-side 768p and 2K panels show the same yellow flower field, moving from a flower close-up to machinery near the plants and a red tractor crossing the field. Output: 15s · 1920×816. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Conor_D_Dart/status/2086444017584140313",
    "sourceLabel": "X 原帖 · @Conor_D_Dart",
    "author": "Conor Dart",
    "publishedAt": "2026-08-09T13:26:32.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086444017584140313.mp4",
    "posterUrl": "/posters/x/2086444017584140313.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×816",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "1/4\n\nI’ve now had the chance to test MiniMax H3 across three different parts of its video-generation workflow: the standard 768p generation, the higher-resolution 2K option, and Context-IR.\n\nSurprisingly, Context-IR might actually be the most important part of the whole experience.\n\nContext-IR is designed to improve and refine your prompt before it is passed to the video-generation model. During my testing, the difference it made was immediately noticeable. Whenever I generated videos without using Context-IR first, the results were consistently worse. (or my prompting just sucks) \n\nWhat impressed me most is how well it seems to understand what H3 itself needs from a prompt.\n\nYou can give Context-IR a fairly straightforward description of a scene, and it expands that into something much more detailed and structured. It fills in many of the important details between your original instructions, such as how the scene progresses, what happens between key moments, camera behaviour, movement, visual details, and the overall flow of the sequence.\n\n- Video details - \nThis video compares the two qualities with the use of Context-IR\n\nI took 3:34 - 5.50 Min to generate the videos in 768p and 2K. \n\nThe Context-IR took around 1min to edit the prompt.\n\nPut Sound on, as that was generated with the video. \n\n-\n\nP.S These Guys have cooked,  @MiniMax_AI  and @Hailuo_AI, I wish I could have run this FAST on my 5080.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 209,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086443622262820904",
    "title": "道场赌场与泳池中的四格人物变换",
    "titleEn": "Four-Way Character Changes Across Dojo, Casino, and Pool",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "四格画面同步呈现人物动作变化：上方男子在道场穿黑袍或蓝色运动服，下方场景位于赌场与泳池，人物和服装随画格改变。 成片规格为 4 秒 · 1728×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A four-way grid shows synchronized character motion: the upper man wears either a black robe or blue athletic outfit in a dojo, while the lower panels shift between a casino and pool with different characters and clothing. Output: 4s · 1728×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/toyxyz3/status/2086443622262820904",
    "sourceLabel": "X 原帖 · @toyxyz3",
    "author": "toyxyz",
    "publishedAt": "2026-08-09T13:24:58.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086443622262820904.mp4",
    "posterUrl": "/posters/x/2086443622262820904.jpg",
    "duration": 4,
    "aspectRatio": "landscape",
    "resolution": "1728×960",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 R2V video edit test #comfyui",
    "engagement": {
      "replies": 3,
      "reposts": 10,
      "likes": 152,
      "views": 10341,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086443552649941148",
    "title": "雨中画出的光蝶飞向撑伞男女",
    "titleEn": "A Drawn Light Butterfly Flies Toward Two People in the Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手中的铅笔先在沾满雨滴的纸页上画出发光轮廓，黄色光蝶随后飞过湿润街道，最后停在一对撑伞男女之间的光雾中。 成片规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pencil first traces a glowing shape on a rain-speckled page, then a yellow light butterfly crosses a wet street and appears in luminous mist between two people beneath an umbrella. Output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FutureVibesAi/status/2086443552649941148",
    "sourceLabel": "X 原帖 · @FutureVibesAi",
    "author": "Future Vibes AI - Educator",
    "publishedAt": "2026-08-09T13:24:41.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086443552649941148.mp4",
    "posterUrl": "/posters/x/2086443552649941148.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 - @Hailuo_AI \nThe Model Is God Damn Good 😍\nCheck the prompt below 🙌",
    "engagement": {
      "replies": 10,
      "reposts": 1,
      "likes": 17,
      "views": 523,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086439838337667391",
    "title": "绿衣健美女子在聚光灯下舞动",
    "titleEn": "A Muscular Woman in Green Moves Beneath a Spotlight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名身穿绿色亮面舞台服的健美女子处于黑色背景与单束顶光下，三帧之间连续改变手臂、腰部与站姿。 成片规格为 25 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A muscular woman in a glossy green stage outfit stands against a black background under a single overhead light, changing her arm, waist, and standing positions across the frames. Output: 25s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JuwalBose/status/2086439838337667391",
    "sourceLabel": "X 原帖 · @JuwalBose",
    "author": "JuwalBose",
    "publishedAt": "2026-08-09T13:09:56.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086439838337667391.mp4",
    "posterUrl": "/posters/x/2086439838337667391.jpg",
    "duration": 25,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Playing with Minimax H3 Motion Context Video extension.\n\nCreated this 25secs 1280x704 video (5s+5s+15s)\nWorks great with Turbo Loras, 15s took 11.15 min, 93s/it\n\nCheck it out\nhttps://github.com/NikoDemon80/ComfyUI-H3-Motion-Context\n\nMy 4070Ti is loving this new ability😎",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 294,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086438490502574462",
    "title": "威尔史密斯与男子同桌吃意面",
    "titleEn": "Will Smith and Another Man Share Spaghetti at a Table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "威尔·史密斯与一名戴眼镜的男子并排坐在餐桌前，两人各自拿着餐叉，中段威尔·史密斯把一叉意面送入口中。 成片规格为 15 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Will Smith sits beside a bespectacled man at a dining table as both hold forks, with the middle frame showing Smith lifting a forkful of spaghetti to his mouth. Output: 15s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Elagizy/status/2086438490502574462",
    "sourceLabel": "X 原帖 · @Elagizy",
    "author": "Ehab Elagizy",
    "publishedAt": "2026-08-09T13:04:34.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086438490502574462.mp4",
    "posterUrl": "/posters/x/2086438490502574462.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing Minimax H3 with ComfyUI. Me and Will Smith doing the spaghetti test...\nInference done locally on 5090 24GB VRAM, 20 Steps, Sage Attn + Sol Attn + Spectrum, took about 27 minutes for a 15s 1280x736 video.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 127,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086437343276548399",
    "title": "女子在暖灯下拥着小猫入睡",
    "titleEn": "A Woman Sleeps Beside a Kitten Under Warm Lamplight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名女子侧卧在花纹被褥中，手臂环绕着同样闭眼躺卧的橘色小猫，床头暖光照亮两者。 成片规格为 7 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman lies on her side beneath patterned bedding with an orange kitten resting inside her arm, both with eyes closed under warm bedside light. Output: 7s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Yayoi_Labo_/status/2086437343276548399",
    "sourceLabel": "X 原帖 · @Yayoi_Labo_",
    "author": "Healing Yayoi -AI Beauty Institute- 弥生",
    "publishedAt": "2026-08-09T13:00:01.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086437343276548399.mp4",
    "posterUrl": "/posters/x/2086437343276548399.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#美容 #セルフケア 夏の簡単美容ポイント🪟 夏は日の出が早いから寝る前にカーテンを少し開けておくようにしているよ🌌 朝の光が入ると目覚めがすっきりしやすいかも🌿 おやすみなさい🌃\n#PolloAI さんの🎥MiniMax H3 で作成してみました💛\n@PolloAIJP @polloai_creator http://tinyurl.com/b9rfjhs7",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 11,
      "views": 209,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086437341808271591",
    "title": "兔耳少女带着圆形伙伴奔向镜头",
    "titleEn": "A Rabbit-Eared Girl Runs Toward the Camera with a Round Companion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "紫发兔耳少女沿未来感平台朝镜头奔跑，身后跟着蓝白色圆形伙伴，天空中同时可见云层与大型星体。 成片规格为 5 秒 · 640×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple-haired rabbit-eared girl runs toward the camera along a futuristic platform with a round blue-and-white companion behind her and a large celestial body above the clouds. Output: 5s · 640×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/malariadx/status/2086437341808271591",
    "sourceLabel": "X 原帖 · @malariadx",
    "author": "びおめた",
    "publishedAt": "2026-08-09T13:00:01.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086437341808271591.mp4",
    "posterUrl": "/posters/x/2086437341808271591.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3(R2V)を試し、画像/動画/音声を参照して動画を生成するってやつですね、今回はComfyUI-MiniMaxH3-Easyというノードとワークフローを使いました\nhttps://github.com/nkxx188/ComfyUI-MiniMaxH3-Easy\n#AI動画 #AIart",
    "engagement": {
      "replies": 2,
      "reposts": 7,
      "likes": 79,
      "views": 5789,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086434543368569014",
    "title": "花园留影后出现的柯达红色标语牌",
    "titleEn": "A Garden Portrait Followed by a Red Kodak Slogan Card",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "金发小女孩先站在玫瑰花园中，随后一名男子举起相机为她拍照，末帧切为写有“KODAK”及品牌标语的红色画面。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde girl first stands in a rose garden, then a man raises a camera to photograph her, and the final frame becomes a red card carrying the Kodak name and slogan. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Dheepanratnam/status/2086434543368569014",
    "sourceLabel": "X 原帖 · @Dheepanratnam",
    "author": "Dheepan Ratnam",
    "publishedAt": "2026-08-09T12:48:53.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086434543368569014.mp4",
    "posterUrl": "/posters/x/2086434543368569014.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.0 vs Minimax H3 on Hailuo Ai\n\nFour 15-second spot ads from different era. \nPure text-to-video, no reference images. 24 fps.\n\nSecond set in the series. Same method as Coca-Cola, \nKodak × Four Eras 1950,1970,1990,2010\n\nSeedance 2.0 results and prompts in the replies⤵️",
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 26,
      "views": 1356,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086434192683110584",
    "title": "水墨山野中橙蓝双侠持刃交锋",
    "titleEn": "Orange and Blue Warriors Clash with Blades in an Ink Landscape",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙黑与蓝黑装束的两名长发武者在水墨山野间持刃近身交锋，兵器碰撞处出现蓝色与橙色光芒。 成片规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two long-haired warriors in orange-black and blue-black clothing fight at close range with bladed weapons in an ink-wash landscape, producing blue and orange light where their weapons meet. Output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/developerlin/status/2086434192683110584",
    "sourceLabel": "X 原帖 · @developerlin",
    "author": "Frank Lin",
    "publishedAt": "2026-08-09T12:47:30.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086434192683110584.mp4",
    "posterUrl": "/posters/x/2086434192683110584.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\nRTX 5070 12G \n480p 15s\n\nLooks much better after agent refined the prompt w/ H3 prompt writting skill, same storyboard.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 149,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086433968858259528",
    "title": "荧光黄裙女子在红色拼贴背景前做手势",
    "titleEn": "A Woman in Neon Yellow Gestures Before a Red Collage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿荧光黄色镂空裙装的黑发女子站在红黑拼贴背景前，手臂与手指在全身和面部近景间连续改变位置。 成片规格为 10 秒 · 1920×1104。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman in a cutout neon-yellow dress stands before a red-and-black collage while her arms and fingers change position across full-body and facial close-ups. Output: 10s · 1920×1104. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotogenicWeekE/status/2086433968858259528",
    "sourceLabel": "X 原帖 · @PhotogenicWeekE",
    "author": "Photogenic Weekend",
    "publishedAt": "2026-08-09T12:46:36.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086433968858259528.mp4",
    "posterUrl": "/posters/x/2086433968858259528.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1104",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "stylized"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3、お次は↓のPromptをH3用に整形、リファレンス無しへ、で<d>[English]...って書くところを[Japanese]へ。これすると英語でも日本人っぽい発音になること”も”ある(笑)",
    "engagement": {
      "replies": 5,
      "reposts": 9,
      "likes": 102,
      "views": 16260,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086432721925267737",
    "title": "猫耳少女抱着电吉他面对镜头",
    "titleEn": "A Cat-Eared Girl Holds an Electric Guitar Toward the Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名扎深色发髻的猫耳少女穿黑色短上衣面对镜头，中段可见她怀中横放的白色电吉他。 成片规格为 30 秒 · 1152×1728。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared girl with dark hair in a bun faces the camera in a black crop top, with a white electric guitar visible across her body in the middle frame. Output: 30s · 1152×1728. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2086432721925267737",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-08-09T12:41:39.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086432721925267737.mp4",
    "posterUrl": "/posters/x/2086432721925267737.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "1152×1728",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "とりあえず384×576, 24fps, 30秒ならRTX4070(VRAM 12GB)でもMiniMax H3で動画生成一発できた。SageAttention+TurboLoRA(8step)で7分23秒。歌は適当だけどそれっぽい。以下は3倍アップスケール後。Real-ESRGAN様最高です。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 177,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086432353405083867",
    "title": "办公室桌旁相对而坐的两名职员",
    "titleEn": "Two Office Workers Seated Across a Cluttered Desk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名白衬衫男子俯身靠近桌面，另一名戴眼镜、穿芥末色衬衫的男子端坐在对面，末帧拉远显示堆满文件与摆件的办公桌。 成片规格为 15 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in a white shirt leans toward the desk while a bespectacled man in a mustard shirt sits opposite, and the final wider view reveals files and figurines covering the workspace. Output: 15s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/awesome_visuals/status/2086432353405083867",
    "sourceLabel": "X 原帖 · @awesome_visuals",
    "author": "awesome_visuals",
    "publishedAt": "2026-08-09T12:40:11.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086432353405083867.mp4",
    "posterUrl": "/posters/x/2086432353405083867.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Office blooper 🤣\n\nmade with MiniMax H3",
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 49,
      "views": 4885,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086432310292037759",
    "title": "围裙女子在厨房整理透明盒中食材",
    "titleEn": "An Aproned Woman Sorts Ingredients in a Clear Kitchen Container",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿浅色围裙的女子出现在居家厨房，近景之间穿插她低头整理透明保鲜盒内食材的中景。 成片规格为 61 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a light apron appears in a home kitchen, with facial close-ups surrounding a medium shot of her sorting ingredients inside a clear food container. Output: 61s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shiokotoba/status/2086432310292037759",
    "sourceLabel": "X 原帖 · @shiokotoba",
    "author": "koko",
    "publishedAt": "2026-08-09T12:40:01.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086432310292037759.mp4",
    "posterUrl": "/posters/x/2086432310292037759.jpg",
    "duration": 61,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "前にAIで自作したAI動画生成ツールを久しぶりに試してみた。ほぼワンクリックで実行して、そのままPCから離れた。\n高速MiniMax H3で、1280×736・1分の動画生成に43分くらい。画像はKrea 2を使ったけど、なんというか全体的に広告映像っぽい感じがする……。\nLTX 2.3の頃と比べると、かなり進歩してる",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 190,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086432040157946137",
    "title": "粉发猫耳少女在光芒中换上华丽裙装",
    "titleEn": "A Pink-Haired Catgirl Changes into an Ornate Dress in Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发猫耳少女先穿简洁的浅色短裙站在白色背景前，身体随后泛起粉光，末帧裙装变为带蝴蝶结、宝石与薄纱的华丽款式。 成片规格为 5 秒 · 720×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired catgirl first stands against white in a simple pale dress, then glows pink before her outfit becomes an ornate design with bows, gems, and translucent fabric. Output: 5s · 720×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hirosi1971/status/2086432040157946137",
    "sourceLabel": "X 原帖 · @hirosi1971",
    "author": "ばら＠α7iiiとAI修行中",
    "publishedAt": "2026-08-09T12:38:57.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086432040157946137.mp4",
    "posterUrl": "/posters/x/2086432040157946137.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fashion",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のローカル動画生成で遊んでます\nなかなか良いのでお試しあれ\n#おねんねあーと",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 8,
      "views": 389,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086430650543050831",
    "title": "银白机器人站上平台面对机械军团",
    "titleEn": "A Silver Robot Faces a Mechanical Army from a Platform",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银白色人形机器人站在科幻大厅中央的平台上，面对整齐排列的机械军团，随后抬起一只手并转向侧面。 成片规格为 10 秒 · 640×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver humanoid robot stands on a central platform in a science-fiction hall before ranks of machines, then raises one hand and turns to the side. Output: 10s · 640×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hage_mukakin/status/2086430650543050831",
    "sourceLabel": "X 原帖 · @hage_mukakin",
    "author": "鉄の無課金",
    "publishedAt": "2026-08-09T12:33:25.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086430650543050831.mp4",
    "posterUrl": "/posters/x/2086430650543050831.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AIの反乱、なお人間を学習し過ぎたみたい。MiniMax-H3で作成。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 108,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086430588727386566",
    "title": "诸神黄昏标题与九界之眼特写",
    "titleEn": "Ragnarok Title and Close-Up of the Nine Realms Eye",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深色纹理背景上先出现带帆船图形的“RAGNAROK”标题，随后切为一只眼睛的极近特写及“NINE REALMS”字样，最后以制作平台标志收尾。 成片规格为 19 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark textured card first presents the word “RAGNAROK” beneath a small sailing ship, then cuts to an extreme eye close-up with “NINE REALMS” text before ending on the production-platform logo. Output: 19s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Rishibag02/status/2086430588727386566",
    "sourceLabel": "X 原帖 · @Rishibag02",
    "author": "Rishi Bag",
    "publishedAt": "2026-08-09T12:33:11.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086430588727386566.mp4",
    "posterUrl": "/posters/x/2086430588727386566.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Great prompt @opener_ai \n\nCreated on @YouArtStudio \nUsing @Hailuo_AI Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 153,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086430481072218307",
    "title": "红黑白剪影中挥刀的制服少女",
    "titleEn": "A Uniformed Girl Swings a Blade Through Red, Black, and White",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红黑白三色图形先组成城市与刀刃剪影，随后一名黑发制服少女挥动长刀，末帧站在破裂红色圆盘与建筑轮廓前。 成片规格为 15 秒 · 1888×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Red, black, and white graphics first form city and blade silhouettes before a black-haired uniformed girl swings a long sword and stands before a fractured red disc and building outlines. Output: 15s · 1888×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2086430481072218307",
    "sourceLabel": "X 原帖 · @Mayz1169",
    "author": "Kiki",
    "publishedAt": "2026-08-09T12:32:45.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086430481072218307.mp4",
    "posterUrl": "/posters/x/2086430481072218307.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1888×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second test with MiniMax H3.\n\nThis time I went with a red color palette, and the result turned out pretty good. Btw, the pricing is really reasonable.\n\nPrompt below ⤵️",
    "engagement": {
      "replies": 13,
      "reposts": 8,
      "likes": 171,
      "views": 11263,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086428473879630076",
    "title": "节点画布下的夕阳屋顶奔跑少女",
    "titleEn": "A Girl Runs Across a Sunset Rooftop Beneath a Node Canvas",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖向画面上半部持续显示节点编辑画布与预览窗口，下半部是一名校服少女沿夕阳映照的城市屋顶向右奔跑。 成片规格为 7 秒 · 2000×2020。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The upper half of the vertical frame continuously shows a node editor and preview window, while the lower half follows a uniformed girl running rightward across a city rooftop at sunset. Output: 7s · 2000×2020. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/metakonoka/status/2086428473879630076",
    "sourceLabel": "X 原帖 · @metakonoka",
    "author": "深海このか☘️Konoka Fukami☘️",
    "publishedAt": "2026-08-09T12:24:46.000Z",
    "addedAt": "2026-08-10T05:52:30.477Z",
    "mediaUrl": "/media/x-2086428473879630076.mp4",
    "posterUrl": "/posters/x/2086428473879630076.jpg",
    "duration": 7,
    "aspectRatio": "square",
    "resolution": "2000×2020",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こんばんは🍀\n今日はようやく時間ができたのでMiniMax H3を試せました🍀\nローカルPCで動かしてます☺️ヽ(｀▽´)/ﾔｯﾀｰ!",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 6,
      "views": 188,
      "snapshotAt": "2026-08-10T05:52:30.477Z"
    },
    "approvedAt": "2026-08-10T05:52:30.477Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086830098909712644",
    "title": "红书包少女沿晴空街道走近后坐下",
    "titleEn": "A Girl with a Red Backpack Approaches and Sits",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一名扎双马尾、背红色书包的少女沿蓝天下的街道朝低机位镜头走近，最后在镜头前屈膝坐下。 成片规格为 7 秒 · 576×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A twin-tailed girl carrying a red backpack walks toward a low camera on a road beneath a vivid blue sky and ends by bending her knees to sit near the lens. Output: 7s · 576×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/da2el2da_eichi/status/2086830098909712644",
    "sourceLabel": "X 原帖 · @da2el2da_eichi",
    "author": "だにえる・えるにだ叡智",
    "publishedAt": "2026-08-10T15:00:41.000Z",
    "addedAt": "2026-08-11T12:15:52.988Z",
    "mediaUrl": "/media/x-2086830098909712644.mp4",
    "posterUrl": "/posters/x/2086830098909712644.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 で参照画像テスト\n音楽はSunoで作りました",
    "engagement": {
      "replies": 7,
      "reposts": 82,
      "likes": 1986,
      "views": 43149,
      "snapshotAt": "2026-08-11T12:15:52.988Z"
    },
    "approvedAt": "2026-08-11T12:15:52.988Z",
    "editorialBasis": "three-frame-media-review",
    "featured": {
      "at": "2026-09-05T15:50:15.685Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2086647221190303868",
    "title": "机械虫缠绕少女并完成装甲变身",
    "titleEn": "Mechanical Insects Envelop a Girl into an Armored Form",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深色机械虫与触须从紫色旋涡中缠上校服少女的四肢和躯干，末帧她变为银发、带尖刺与紫色发光结构的装甲形态。 成片规格为 10 秒 · 1080×1620。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Dark mechanical insects and tendrils coil around a uniformed girl's limbs and torso within a purple vortex, ending with her in a silver-haired armored form covered in spikes and glowing violet structures. Output: 10s · 1080×1620. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kagami_kami_m/status/2086647221190303868",
    "sourceLabel": "X 原帖 · @kagami_kami_m",
    "author": "👯‍♀️カガミカミ水鏡👯‍♀️",
    "publishedAt": "2026-08-10T02:54:00.000Z",
    "addedAt": "2026-08-11T12:15:52.989Z",
    "mediaUrl": "/media/x-2086647221190303868.mp4",
    "posterUrl": "/posters/x/2086647221190303868.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1080×1620",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax-h3(local)で作った機械化昆虫に寄生されて悪の戦闘兵器メカ少女に悪堕ち変身するシーン\n\n高速化loraなど使っているからかクオリティが悩みどころだけど、自宅でこれ生成出来るのは本当にいいね！ 弓なりになる時の表情が悪堕ちを理解してるって感じでナイスですわ",
    "engagement": {
      "replies": 5,
      "reposts": 36,
      "likes": 462,
      "views": 12228,
      "snapshotAt": "2026-08-11T12:15:52.989Z"
    },
    "approvedAt": "2026-08-11T12:15:52.989Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086771726822904256",
    "title": "女子在蓝粉灯光之间独舞",
    "titleEn": "A Woman Dances Between Blue and Pink Lights",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿黑色长裤与宽松外套的女子站在空旷室内，蓝色和粉色灯管分列两侧，她连续变换步伐与手臂动作。 成片规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in black trousers and a loose jacket stands in an empty indoor space between blue and pink light bars, changing her steps and arm positions across the frames. Output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SaniAiTech/status/2086771726822904256",
    "sourceLabel": "X 原帖 · @SaniAiTech",
    "author": "Sani Ai Tech",
    "publishedAt": "2026-08-10T11:08:44.000Z",
    "addedAt": "2026-08-11T12:15:52.989Z",
    "mediaUrl": "/media/x-2086771726822904256.mp4",
    "posterUrl": "/posters/x/2086771726822904256.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tested MiniMax H3’s Audio as Soundtrack feature 🎵🔥\n\nI uploaded a dance track and let H3 generate the visuals around its rhythm and energy.\n\nThe way it turns audio into synchronized movement, camera motion, and scene dynamics is seriously impressive.\n\nAudio → Motion → Video.\nH3 is taking AI video generation to another level. 👀\n\n#MiniMaxH3 #AIVideo #GenerativeAI #AIAnimation",
    "engagement": {
      "replies": 25,
      "reposts": 150,
      "likes": 202,
      "views": 40708,
      "snapshotAt": "2026-08-11T12:15:52.989Z"
    },
    "approvedAt": "2026-08-11T12:15:52.989Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086680991981289554",
    "title": "星空中交叠出现的白发黑衣少女",
    "titleEn": "A White-Haired Girl Overlaps with Herself Against the Stars",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短白发黑衣少女出现在繁星背景前，近景先呈现交叠的双重轮廓，随后两个侧身形象错位并逐渐只留下单人远景。 成片规格为 15 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short white-haired girl in black appears before a star field, beginning with overlapping close-up silhouettes, shifting into two offset profiles, and ending as a single distant figure. Output: 15s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jAlpha_create/status/2086680991981289554",
    "sourceLabel": "X 原帖 · @jAlpha_create",
    "author": "ジュニ＠Multi AI Creator",
    "publishedAt": "2026-08-10T05:08:11.000Z",
    "addedAt": "2026-08-11T12:15:52.989Z",
    "mediaUrl": "/media/x-2086680991981289554.mp4",
    "posterUrl": "/posters/x/2086680991981289554.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3、3060の8GBで動きました‼️\n\n動画生成でいちばん困るのって、プロンプトを詰めたいのにコストが気になるところだと思っていて。演出をあれこれ試したいのに、納得いくまでやれないんですよね。\n\nそれがローカルで回せるようになったので、試し撃ちが電気代だけになりました。\n\nturbo LoRAを使えば10秒が10分ちょっと。480pなので画質は落ちますけど、カット割りとか演出の確認にはこれで十分でした。\n\nなので今後はこんな使い方ができそうです。\n\nローカル → プロンプトと演出を納得行くまで詰める（無料）\nRunPod → 決まったものを1344×768で本番（1本8円）＋アップスケール\n借りるGPUの時間も最小で済むように。\n\nちなみに昨夜マージされたKijaiさんのVAEのPRを当てたら、8GBで15秒も通りました。",
    "engagement": {
      "replies": 5,
      "reposts": 32,
      "likes": 372,
      "views": 42402,
      "snapshotAt": "2026-08-11T12:15:52.989Z"
    },
    "approvedAt": "2026-08-11T12:15:52.989Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086671141998059973",
    "title": "晨起女子与海景露台的一日开场",
    "titleEn": "A Morning Begins from Bed to a Seaside Terrace",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长发女子先坐在明亮卧室的床上整理头发，随后画面切到摆有木桌椅的海景露台，最后以落日海面和“Good vibes, Good life”文字收尾。 成片规格为 15 秒 · 1088×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A long-haired woman first sits on a bed in a bright bedroom and adjusts her hair, followed by a seaside terrace with wooden furniture and a closing sunset view reading “Good vibes, Good life.” Output: 15s · 1088×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second ultra-realistic cinematic lifestyle vlog video, vertical 9:16, featuring the same young woman throughout the entire video. Preserve her facial identity, facial proportions, hairstyle, skin tone and overall appearance consistently in every shot. She wears the same outfit throughout: fitted white V-neck T-shirt with a small subtle logo, blue denim jeans, natural makeup, long softly wavy brown hair.\n\n0:00–0:01 — Wake-up:\nClose-up inside a beautiful bright bedroom. The woman is lying comfortably on the bed, slowly wakes up, stretches naturally and opens her eyes. She is NOT filming a vlog yet and does not hold a phone or camera. Soft morning sunlight enters through the curtains.\n\n0:01–0:02 — Gets up:\nMedium shot. She sits up on the bed, smiles softly, fixes her hair and gets ready to start her morning. Natural, effortless movement.\n\n0:02–0:03 — Walks to window:\nShe walks toward the large glass balcony door/window. Camera follows her naturally from behind/side.\n\n0:03–0:04 — Seaside reveal:\nShe opens the curtains/door and looks outside. Reveal a breathtaking blue ocean, coastal hills, flowers, balcony and beautiful morning sunlight. She smiles happily while taking in the view.\n\n0:04–0:05 — Steps outside:\nShe walks out onto the seaside terrace. Gentle ocean breeze moves her hair naturally. Wide cinematic shot showing the beautiful surroundings.\n\n0:05–0:06 — VLOG START:\nOnly now she starts filming herself in handheld selfie-vlog style. She looks into the camera with a bright natural smile and says:\n“Good morning!”\n\n0:06–0:07 — Show the view:\nShe turns the camera away from herself and slowly pans across the stunning ocean, coastal mountains, flowers and terrace. Smooth handheld vlog movement.\n\n0:07–0:08 — Back to selfie:\nSelfie shot. She looks into the camera and happily says:\n“This place is just perfect!”\n\n0:08–0:09 — Location reveal:\nWide cinematic shot of the cozy seaside terrace with wooden table, chairs, plants and flowers overlooking the ocean.\n\n0:09–0:10 — Walk to table:\nMedium tracking shot as she walks toward the table, enjoying the view. Her hair and T-shirt move gently in the sea breeze.\n\n0:10–0:11 — Sit and relax:\nShe sits at the seaside table, smiling peacefully and enjoying the ocean view. A refreshing orange-colored juice is placed on the table.\n\n0:11–0:12 — Juice close-up:\nCinematic close-up of her hand picking up the glass of fresh orange juice. Beautiful ocean bokeh in the background, natural sunlight reflecting through the glass.\n\n0:12–0:13 — Vlog toast:\nSelfie shot. She raises the juice toward the camera with a cheerful smile and says:\n“Cheers to good days!”\n\n0:13–0:14 — Happy close-up:\nBeautiful close-up of her smiling naturally at the camera, ocean and warm sunlight softly blurred behind her.\n\n0:14–0:15 — Ending:\nCamera moves from her toward the sparkling ocean and peaceful coastal landscape. Warm sunlight, gentle waves and a relaxing cinematic ending.\n\nOverall Style\n\nUltra-realistic, cinematic travel vlog, natural handheld camera movement, realistic human motion, smooth transitions, soft morning sunlight, realistic ocean waves, gentle wind in hair and clothes, beautiful coastal atmosphere, premium lifestyle aesthetic, natural expressions, authentic vlog feeling, shallow depth of field, cinematic composition, realistic skin texture, high detail, 4K quality.",
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2086671141998059973",
    "sourceLabel": "X 原帖 · @ayzalnooor24521",
    "author": "ayzalnoor",
    "publishedAt": "2026-08-10T04:29:03.000Z",
    "addedAt": "2026-08-11T12:15:52.989Z",
    "mediaUrl": "/media/x-2086671141998059973.mp4",
    "posterUrl": "/posters/x/2086671141998059973.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1088×720",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A peaceful seaside morning turned into a beautiful little vlog. \nGood vibes, ocean views, and a perfect start to the day. \n\nCreated with MiniMax H3\n\nPrompt:\n\nCreate a 15-second ultra-realistic cinematic lifestyle vlog video, vertical 9:16, featuring the same young woman throughout the entire video. Preserve her facial identity, facial proportions, hairstyle, skin tone and overall appearance consistently in every shot. She wears the same outfit throughout: fitted white V-neck T-shirt with a small subtle logo, blue denim jeans, natural makeup, long softly wavy brown hair.\n\n0:00–0:01 — Wake-up:\nClose-up inside a beautiful bright bedroom. The woman is lying comfortably on the bed, slowly wakes up, stretches naturally and opens her eyes. She is NOT filming a vlog yet and does not hold a phone or camera. Soft morning sunlight enters through the curtains.\n\n0:01–0:02 — Gets up:\nMedium shot. She sits up on the bed, smiles softly, fixes her hair and gets ready to start her morning. Natural, effortless movement.\n\n0:02–0:03 — Walks to window:\nShe walks toward the large glass balcony door/window. Camera follows her naturally from behind/side.\n\n0:03–0:04 — Seaside reveal:\nShe opens the curtains/door and looks outside. Reveal a breathtaking blue ocean, coastal hills, flowers, balcony and beautiful morning sunlight. She smiles happily while taking in the view.\n\n0:04–0:05 — Steps outside:\nShe walks out onto the seaside terrace. Gentle ocean breeze moves her hair naturally. Wide cinematic shot showing the beautiful surroundings.\n\n0:05–0:06 — VLOG START:\nOnly now she starts filming herself in handheld selfie-vlog style. She looks into the camera with a bright natural smile and says:\n“Good morning!”\n\n0:06–0:07 — Show the view:\nShe turns the camera away from herself and slowly pans across the stunning ocean, coastal mountains, flowers and terrace. Smooth handheld vlog movement.\n\n0:07–0:08 — Back to selfie:\nSelfie shot. She looks into the camera and happily says:\n“This place is just perfect!”\n\n0:08–0:09 — Location reveal:\nWide cinematic shot of the cozy seaside terrace with wooden table, chairs, plants and flowers overlooking the ocean.\n\n0:09–0:10 — Walk to table:\nMedium tracking shot as she walks toward the table, enjoying the view. Her hair and T-shirt move gently in the sea breeze.\n\n0:10–0:11 — Sit and relax:\nShe sits at the seaside table, smiling peacefully and enjoying the ocean view. A refreshing orange-colored juice is placed on the table.\n\n0:11–0:12 — Juice close-up:\nCinematic close-up of her hand picking up the glass of fresh orange juice. Beautiful ocean bokeh in the background, natural sunlight reflecting through the glass.\n\n0:12–0:13 — Vlog toast:\nSelfie shot. She raises the juice toward the camera with a cheerful smile and says:\n“Cheers to good days!”\n\n0:13–0:14 — Happy close-up:\nBeautiful close-up of her smiling naturally at the camera, ocean and warm sunlight softly blurred behind her.\n\n0:14–0:15 — Ending:\nCamera moves from her toward the sparkling ocean and peaceful coastal landscape. Warm sunlight, gentle waves and a relaxing cinematic ending.\n\nOverall Style\n\nUltra-realistic, cinematic travel vlog, natural handheld camera movement, realistic human motion, smooth transitions, soft morning sunlight, realistic ocean waves, gentle wind in hair and clothes, beautiful coastal atmosphere, premium lifestyle aesthetic, natural expressions, authentic vlog feeling, shallow depth of field, cinematic composition, realistic skin texture, high detail, 4K quality.",
    "engagement": {
      "replies": 83,
      "reposts": 5,
      "likes": 278,
      "views": 11799,
      "snapshotAt": "2026-08-11T12:15:52.989Z"
    },
    "approvedAt": "2026-08-11T12:15:52.989Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086675231947751768",
    "title": "开源超分页面中的三六零P与四K对照",
    "titleEn": "A 360p-to-4K Upscaling Comparison on an Open-Source Page",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "浏览器页面先显示开源项目的文件列表与说明文档，末帧出现同一动漫眼部画面的 360P 原图和 4K 超分结果并排对照。 成片规格为 16 秒 · 3794×1836。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A browser first shows an open-source project's file list and documentation before presenting a side-by-side 360p original and 4K upscaled anime-eye image. Output: 16s · 3794×1836. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Lonely__MH/status/2086675231947751768",
    "sourceLabel": "X 原帖 · @Lonely__MH",
    "author": "Lonely",
    "publishedAt": "2026-08-10T04:45:18.000Z",
    "addedAt": "2026-08-11T12:15:52.989Z",
    "mediaUrl": "/media/x-2086675231947751768.mp4",
    "posterUrl": "/posters/x/2086675231947751768.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "3794×1836",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🎉刚本地部署了 MiniMax H3，结果发现开源的 H3-Base 只能输出 768p ？ \n\n别慌，本地最完美的「平替超分组合」来了！\n\n用 Video2X 框架，直接把 H3 生成的 768p 视频丢进去，其集成的 Real-CUGAN 或 Real-ESRGAN 可以瞬间帮你本地无损拉高到 2K 甚至 4K！ \n\n另外运行时还不占额外临时空间\n\n即使官方还没开源 H3-Regenerate-2K 模块，自己也能组装一套本地超分",
    "engagement": {
      "replies": 26,
      "reposts": 47,
      "likes": 305,
      "views": 30259,
      "snapshotAt": "2026-08-11T12:15:52.989Z"
    },
    "approvedAt": "2026-08-11T12:15:52.989Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086782084723405173",
    "title": "客厅沙发上的莱戈拉斯与金雳",
    "titleEn": "Legolas and Gimli Sit Across a Living Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金雳坐在绿色扶手椅旁，莱戈拉斯斜靠橙色沙发并手持高脚杯，三帧在两人的双人景与各自近景之间切换。 成片规格为 15 秒 · 2944×1248。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Gimli sits beside a green armchair while Legolas reclines on an orange sofa holding a stemmed glass, with the frames alternating between a shared view and individual close-ups. Output: 15s · 2944×1248. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the lord of the rings as a sitcom, legolas and gimli arguing about how many orcs they hunted",
    "sourceUrl": "https://x.com/aimikoda/status/2086782084723405173",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-10T11:49:54.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086782084723405173.mp4",
    "posterUrl": "/posters/x/2086782084723405173.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n\nprompt: the lord of the rings as a sitcom, legolas and gimli arguing about how many orcs they hunted",
    "engagement": {
      "replies": 21,
      "reposts": 8,
      "likes": 116,
      "views": 7013,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086747292347023752",
    "title": "体育馆少女伸展手臂后挥手",
    "titleEn": "A Girl Stretches Her Arm and Waves in a Gymnasium",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名盘发少女穿运动背心和短裤站在室内球场，先把一侧手臂横拉至胸前伸展，随后放下手臂并向镜头挥手。 成片规格为 8 秒 · 576×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl with her hair in a bun stands on an indoor court in athletic clothing, pulls one arm across her chest to stretch, and then lowers it and waves toward the camera. Output: 8s · 576×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/da2el2da_eichi/status/2086747292347023752",
    "sourceLabel": "X 原帖 · @da2el2da_eichi",
    "author": "だにえる・えるにだ叡智",
    "publishedAt": "2026-08-10T09:31:39.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086747292347023752.mp4",
    "posterUrl": "/posters/x/2086747292347023752.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3でテスト\n準備体操する少女",
    "engagement": {
      "replies": 2,
      "reposts": 22,
      "likes": 769,
      "views": 13189,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086761559989112965",
    "title": "海滩上的两名少女与搁浅海洋动物",
    "titleEn": "Two Girls Beside a Stranded Marine Animal on the Beach",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面先以金发少女近景开场，随后她与粉发少女站在棕榈树环绕的沙滩上，身旁躺着一只大型黑白海洋动物，末帧转向海面旁的粉发少女。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The sequence opens on a blonde girl, then shows her with a pink-haired girl on a palm-lined beach beside a large black-and-white marine animal, and ends on the pink-haired girl near the water. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aidenguoai/status/2086761559989112965",
    "sourceLabel": "X 原帖 · @aidenguoai",
    "author": "Aiden Guo",
    "publishedAt": "2026-08-10T10:28:20.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086761559989112965.mp4",
    "posterUrl": "/posters/x/2086761559989112965.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "While I'm busy with many things in the background, a quick Minimax H3 test😇",
    "engagement": {
      "replies": 7,
      "reposts": 87,
      "likes": 1483,
      "views": 44292,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review",
    "featured": {
      "at": "2026-09-05T15:50:15.684Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2086660959054229833",
    "title": "旗袍女子面部变化的优化前后对照",
    "titleEn": "Before-and-After Facial Changes in a Woman Wearing a Qipao",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "夕阳逆光下的旗袍女子以近景出现，前两帧标注“未经 IR 优化”，末帧标注“经过 IR 优化”，三帧中的嘴形、眉眼和头部角度各不相同。 成片规格为 20 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman in a qipao appears in sunset backlight, with the first two frames labeled before IR optimization and the last labeled after it; her mouth, eyes, brows, and head angle differ across the views. Output: 20s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "甜笑后强忍悲伤",
    "sourceUrl": "https://x.com/0xmomonifty/status/2086660959054229833",
    "sourceLabel": "X 原帖 · @0xmomonifty",
    "author": "0xMomo",
    "publishedAt": "2026-08-10T03:48:35.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086660959054229833.mp4",
    "posterUrl": "/posters/x/2086660959054229833.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 官方 API 有个很好用的功能：Context-IR\n\n我用同一个提示词、同一张首帧，分别生成了两段 10 秒 768P 视频。第一段直接生成，第二段经过 Context-IR\n\n它能够把提示词 “甜笑后强忍悲伤” 拆成 吸气、笑容僵住、嘴角颤动、吞咽、肩膀下沉，再补上推镜和声音设计\n\nIR 的本质是通过深度解析跨越文本、图像、音频和视频生成更丰富语义细节上下文\n\n对比下来，IR 版强忍悲伤的过程更细，情绪转折也更连续，能够让生成的视频细节更加丰富、真实\n\n@MiniMax_AI  @Hailuo_AI",
    "engagement": {
      "replies": 40,
      "reposts": 15,
      "likes": 151,
      "views": 24775,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086754979625447658",
    "title": "白发猫耳少女与逐帧变化的文字框",
    "titleEn": "A White-Haired Catgirl with Changing Text Boxes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白发猫耳少女穿深蓝制服站在白色背景前，人物姿态略有变化，旁边的竖排日文气泡与方形文字框在三帧中依次更换。 成片规格为 10 秒 · 1344×1856。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired catgirl in a dark blue uniform stands against a white background with slight pose changes as vertical Japanese bubbles and rectangular text boxes change from frame to frame. Output: 10s · 1344×1856. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tail_hammer/status/2086754979625447658",
    "sourceLabel": "X 原帖 · @Tail_hammer",
    "author": "てーる@AI illust",
    "publishedAt": "2026-08-10T10:02:11.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086754979625447658.mp4",
    "posterUrl": "/posters/x/2086754979625447658.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1344×1856",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で動く一コマ漫画を作りたかったけど、なかなか上手くいかず。\n結局、素材だけ作って自分で動画編集しました🫠\n映像と音声は別々で生成してます。\n\n当然なのかもだけど、音声は高速化しないほうが良い感じのができますね〜",
    "engagement": {
      "replies": 0,
      "reposts": 8,
      "likes": 111,
      "views": 2994,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086839812234059832",
    "title": "机甲部队穿越爆炸不断的荒原",
    "titleEn": "A Mecha Unit Advances Across an Exploding Wasteland",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多台持枪机甲穿过灰色天空下的荒草地，火球与烟尘在队伍周围升起，末帧一台大型机甲被爆炸包围并向侧面倾起。 成片规格为 15 秒 · 1440×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Several armed mecha cross dry grass beneath a gray sky as fireballs and smoke rise around them, ending with a large unit lifted sideways amid an explosion. Output: 15s · 1440×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ojiji2025/status/2086839812234059832",
    "sourceLabel": "X 原帖 · @ojiji2025",
    "author": "ojiji",
    "publishedAt": "2026-08-10T15:39:17.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086839812234059832.mp4",
    "posterUrl": "/posters/x/2086839812234059832.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1440×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I thought I'd try out a sci-fi battle with MiniMax H3. What do you think?",
    "engagement": {
      "replies": 36,
      "reposts": 34,
      "likes": 284,
      "views": 5880,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review",
    "featured": {
      "at": "2026-09-05T15:50:15.679Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2086794038737539077",
    "title": "戴耳机少女站在麦克风前",
    "titleEn": "A Girl with Headphones Stands Before a Microphone",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "棕色长发少女把耳机挂在颈间并站在麦克风前，画面在半身、面部特写和稍远景别之间切换，背景有白色音符图形。 成片规格为 15 秒 · 768×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long brown-haired girl stands before a microphone with headphones around her neck as the view shifts between medium, facial close-up, and wider framing with white musical-note graphics behind her. Output: 15s · 768×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/dobe1919/status/2086794038737539077",
    "sourceLabel": "X 原帖 · @dobe1919",
    "author": "dobe",
    "publishedAt": "2026-08-10T12:37:24.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086794038737539077.mp4",
    "posterUrl": "/posters/x/2086794038737539077.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "768×576",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3導入しCodexに独自UIを作成させイラスト＆音声ファイルの組み合わせを自動にするUを作成してる最中のテスト生成。\nカメラワークで誤魔化せば3つとか複数の動画連結行けるかな？  \n#結城明日奈 #アスナ #SAO #MiniMaxH3 SFW",
    "engagement": {
      "replies": 1,
      "reposts": 7,
      "likes": 91,
      "views": 3026,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086954503635419578",
    "title": "四名角色乘车穿过飞碟光束下的夜城",
    "titleEn": "Four Characters Drive Through a Night City Beneath a UFO Beam",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名男子坐在轿车前排，后座挤着两名戴帽的尖齿角色，镜头随后移到车外，车辆穿过被飞碟光束照亮的夜间城市。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two men occupy the front seats of a sedan while two hat-wearing, sharp-toothed characters crowd the back, before an exterior view shows the car crossing a night city beneath a UFO beam. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AlfredAlfer77/status/2086954503635419578",
    "sourceLabel": "X 原帖 · @AlfredAlfer77",
    "author": "Emily Youcis",
    "publishedAt": "2026-08-10T23:15:02.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086954503635419578.mp4",
    "posterUrl": "/posters/x/2086954503635419578.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "If you're not generating 3-4 minute long perfectly synced 2D Animated Music Videos with One Prompt in One Take using Minimax H3 + The Slop Cannon by @boneGPT what are you even doin'?",
    "engagement": {
      "replies": 12,
      "reposts": 11,
      "likes": 193,
      "views": 13021,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086801730797531206",
    "title": "酒馆餐桌上的Q版伙伴与丰盛宴席",
    "titleEn": "Chibi Companions Gather Around a Tavern Feast",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "多名 Q 版角色围坐在烛光酒馆的圆桌旁，桌上摆满肉食、水果与酒杯，中间的蓝发角色拿起圆形食物，其他角色举杯或伸手。 成片规格为 15 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Several chibi characters gather around a candlelit tavern table covered with meat, fruit, and mugs as the blue-haired figure lifts a round food item and others raise cups or reach forward. Output: 15s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NullpoLab/status/2086801730797531206",
    "sourceLabel": "X 原帖 · @NullpoLab",
    "author": "ぬるぽらぼ",
    "publishedAt": "2026-08-10T13:07:58.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086801730797531206.mp4",
    "posterUrl": "/posters/x/2086801730797531206.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 I2V でちびキャラにできるか実験\n\nもはや結果はわかりきってるって？\nだよねー俺もそう思ってました",
    "engagement": {
      "replies": 3,
      "reposts": 8,
      "likes": 64,
      "views": 7910,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086777152666796255",
    "title": "手掌大小男子攀上巨型草莓品尝",
    "titleEn": "A Hand-Sized Man Climbs and Bites a Giant Strawberry",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名缩小的男子在花园土壤与叶片之间奔跑，随后攀住比自己更高的草莓，末帧贴近果实咬下一口。 成片规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny man runs among garden soil and leaves, climbs onto a strawberry taller than himself, and ends with his mouth pressed to the fruit for a bite. Output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FutureVibesAi/status/2086777152666796255",
    "sourceLabel": "X 原帖 · @FutureVibesAi",
    "author": "Future Vibes AI - Educator",
    "publishedAt": "2026-08-10T11:30:18.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086777152666796255.mp4",
    "posterUrl": "/posters/x/2086777152666796255.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What if you suddenly became the size of your own hand...? 🤯\nCreated using Minimax H3 in @Hailuo_AI ..\nCheck the prompt below..\n#MinimaxH3",
    "engagement": {
      "replies": 20,
      "reposts": 10,
      "likes": 90,
      "views": 92119,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086730655161823597",
    "title": "琥珀香水与金色烟雾中的模特",
    "titleEn": "Amber Perfume and a Model Amid Golden Smoke",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "琥珀色香水瓶先立在黑色反光台面上，随后一名穿橙棕色裙装的女子举瓶喷洒，末帧回到被金色烟雾与晶体环绕的产品构图。 成片规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An amber perfume bottle first stands on a reflective black surface, then a woman in a burnt-orange dress raises and sprays it, before the product returns surrounded by golden smoke and crystals. Output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A Luxury perfume bottle on glossy black surface, warm amber lighting, golden smoke, dark elegant background, soft reflections, cinematic close-up, slow push-in, photorealistic 4K.\nConfident woman in a burnt-orange satin dress, long dark hair, luxurious dark interior with golden lights, subtle smoke, cinematic medium shot, photorealistic 4K.\nClose-up of the same woman spraying perfume on her neck. Visible fragrance mist, warm golden light, natural hair movement, shallow depth of field, cinematic beauty ad.\nSame woman walking confidently toward camera in the luxury interior. Smooth tracking shot, reflective floor, warm golden lighting, subtle smoke, elegant cinematic look.\nSame perfume bottle centered on glossy black surface with amber crystals and golden smoke. Dramatic spotlight, dark luxury background, slow push-in, photorealistic 4K.\n\nConsistency:\nKeep the same woman, face, hair, skin tone, outfit, and perfume bottle across all scenes. Warm amber lighting, realistic skin and movement. No extra people, distorted hands, duplicate objects, text, or logos.\n\nStyle: Luxury fragrance commercial, cinematic, photorealistic, 4K.",
    "sourceUrl": "https://x.com/Lianaalane/status/2086730655161823597",
    "sourceLabel": "X 原帖 · @Lianaalane",
    "author": "liana",
    "publishedAt": "2026-08-10T08:25:32.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086730655161823597.mp4",
    "posterUrl": "/posters/x/2086730655161823597.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "commercial"
    ],
    "scenes": [
      "combat",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A fragrance that speaks without saying a word. \nBold. Elegant. Unforgettable. \n\nCreated with MiniMax H3\n\nPrompt:\nA Luxury perfume bottle on glossy black surface, warm amber lighting, golden smoke, dark elegant background, soft reflections, cinematic close-up, slow push-in, photorealistic 4K.\nConfident woman in a burnt-orange satin dress, long dark hair, luxurious dark interior with golden lights, subtle smoke, cinematic medium shot, photorealistic 4K.\nClose-up of the same woman spraying perfume on her neck. Visible fragrance mist, warm golden light, natural hair movement, shallow depth of field, cinematic beauty ad.\nSame woman walking confidently toward camera in the luxury interior. Smooth tracking shot, reflective floor, warm golden lighting, subtle smoke, elegant cinematic look.\nSame perfume bottle centered on glossy black surface with amber crystals and golden smoke. Dramatic spotlight, dark luxury background, slow push-in, photorealistic 4K.\n\nConsistency:\nKeep the same woman, face, hair, skin tone, outfit, and perfume bottle across all scenes. Warm amber lighting, realistic skin and movement. No extra people, distorted hands, duplicate objects, text, or logos.\n\nStyle: Luxury fragrance commercial, cinematic, photorealistic, 4K.",
    "engagement": {
      "replies": 68,
      "reposts": 2,
      "likes": 228,
      "views": 12197,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086674810500001967",
    "title": "白发长耳少女转身走近并挥手",
    "titleEn": "A White-Haired Long-Eared Girl Turns, Approaches, and Waves",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "白发长耳少女穿黑白运动短装站在纯白背景中，镜头从全身移到侧面中景，最后她正面靠近并抬手挥动。 成片规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired long-eared girl in a black-and-white cropped athletic outfit stands against a plain white background, turns in a medium view, and finally approaches while raising a hand to wave. Output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cla_kuro37459/status/2086674810500001967",
    "sourceLabel": "X 原帖 · @cla_kuro37459",
    "author": "くらくろ🌗AI VTuber ラビ開発",
    "publishedAt": "2026-08-10T04:43:38.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086674810500001967.mp4",
    "posterUrl": "/posters/x/2086674810500001967.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Ref2VAでキャラシートが全く効かない原因がやっと判明。AIが長時間解決できないことは根本から人間が見直すしかないと再認識 \n\ncomfyUI得意じゃないからcodexに投げてたんだけど、公式ワークフロー流用じゃなくてcodexが勝手にそれっぽいワークフローにしてるだけで、参照データが一切渡せていなかったという大ポカ。  \n\n\"ref_images\": {\"ref_image_0\": [...]}になっていた。ComfyUIが解釈するのは\"ref_images.ref_image_0\": [...]なのに… \n\n 数度ツールとフローを根本から見直せといってたのに、何度も公式とたがわないワークフローですとぬかしやがってました。ﾌﾟﾝｽｺ\n\n4070 12GB メモリ 32GB 5second 720p、公式ワークフロー20stepで30分…あとはいろんな高速化試していこう。",
    "engagement": {
      "replies": 0,
      "reposts": 7,
      "likes": 118,
      "views": 11036,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086774589426643034",
    "title": "白衣男子在绿色沙发上变换坐姿",
    "titleEn": "A Man in White Changes Posture on a Green Sofa",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿白色 T 恤与灰色长裤的男子先横躺在绿色沙发上，随后坐起并伸出一只手，末帧面向镜头坐在沙发中央。 成片规格为 12 秒 · 1920×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A man in a white T-shirt and gray trousers first reclines across a green sofa, then sits up and extends one hand before facing the camera from the center cushion. Output: 12s · 1920×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Talk about life, but make jokes.",
    "sourceUrl": "https://x.com/gizakdag/status/2086774589426643034",
    "sourceLabel": "X 原帖 · @gizakdag",
    "author": "Gizem Akdag",
    "publishedAt": "2026-08-10T11:20:07.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086774589426643034.mp4",
    "posterUrl": "/posters/x/2086774589426643034.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Whenever I need to think deeply about life, I ask AI.\n\nMy prompt was: \"Talk about life, but make jokes.\"\n\n//MiniMax H3",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 32,
      "views": 6495,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086923648527405211",
    "title": "短发女子在厨房搅拌蓝碗食材",
    "titleEn": "A Short-Haired Woman Mixes Ingredients in a Blue Bowl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一名短发女子站在绿色橱柜环绕的厨房中，先用工具搅拌蓝色碗内的食材，随后俯身在灶台与操作台之间继续处理。 成片规格为 20 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired woman stands in a kitchen with green cabinets, mixing ingredients in a blue bowl before leaning over the stove and counter to continue preparing them. Output: 20s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CallMeDiegoJr/status/2086923648527405211",
    "sourceLabel": "X 原帖 · @CallMeDiegoJr",
    "author": "Diego Jr",
    "publishedAt": "2026-08-10T21:12:25.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086923648527405211.mp4",
    "posterUrl": "/posters/x/2086923648527405211.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "این ویدیو رو با مدل MiniMax H3 و روی سرویس @itsPolloAI ساختم و بدون هیچ خطایی روی پرامپت، خروجی گرفتم.\n\nهمین پرامپت و تصاویر رفرنس، که به نظر چیز خاصی هم ندارن، روی مدل‌های دیگه بی شک ارور می‌گیره!\nالبته مدل #MinimaxH3 روی نسخه لوکال کاملا دست شما رو برای هر خروجی‌ای باز گذاشته!",
    "engagement": {
      "replies": 20,
      "reposts": 5,
      "likes": 333,
      "views": 23046,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086769538666779063",
    "title": "宇航员穿过发光的异星管状通道",
    "titleEn": "An Astronaut Travels Through a Glowing Alien Tunnel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名白衣宇航员漂浮在蓝紫色有机纹理包围的管状空间内，近景中身体贴近发亮壁面，最后化作高速向前的白色光影。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-suited astronaut floats inside a tubular space lined with blue and violet organic textures, moves close to a glowing wall, and finally becomes a bright streak traveling forward. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JayKay65220066/status/2086769538666779063",
    "sourceLabel": "X 原帖 · @JayKay65220066",
    "author": "Jay Kay",
    "publishedAt": "2026-08-10T11:00:02.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086769538666779063.mp4",
    "posterUrl": "/posters/x/2086769538666779063.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Space breaker! 🌌\n\nMade with MiniMax H3 video model from @Hailuo_AI!\n\n#Mondayvibes #MondayMotivation #MondayMood #Mondaythoughts Good Monday",
    "engagement": {
      "replies": 38,
      "reposts": 17,
      "likes": 109,
      "views": 4246,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086879654116495564",
    "title": "整片天空如巨墙压向郊区街道",
    "titleEn": "The Sky Descends Like a Giant Wall over a Suburban Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "普通郊区街道先保持晴朗，一道横跨天际的巨大灰白墙面随后从远处逼近，末帧翻涌云层占据住宅与车辆后的整片天空。 成片规格为 14 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An ordinary suburban street begins under clear skies before a vast gray-white wall spans the horizon and advances, ending with churning cloudlike forms filling the sky behind houses and cars. Output: 14s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A hyper-realistic handheld phone video of a quiet suburban backyard on a bright normal afternoon. A person casually waters plants near a fence while birds chirp, distant traffic hums, and everything feels completely ordinary and mundane. The camera drifts naturally like a real home video, capturing the lawn, patio furniture, and blue sky with soft white clouds. After several seconds of normal peaceful activity, a deep cracking sound suddenly comes from above. Without warning, the entire sky begins dropping straight downward like a gigantic solid ceiling, with the blue sky and clouds moving as one physical surface descending toward the yard. The person looks up in shock just as the sky rapidly fills the frame, swallowing the scene in a terrifying instant. The camera jerks and falls to the ground at the last moment. No cinematic buildup, no surreal visual style, it should feel like a totally normal real-life recording interrupted by one sudden impossible and visceral event.",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2086879654116495564",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-10T18:17:36.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086879654116495564.mp4",
    "posterUrl": "/posters/x/2086879654116495564.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "How well can MiniMax H3 simulate outrageous natural disasters? 14 sec clip generated on my Windows PC.\n\nPrompt: \"A hyper-realistic handheld phone video of a quiet suburban backyard on a bright normal afternoon. A person casually waters plants near a fence while birds chirp, distant traffic hums, and everything feels completely ordinary and mundane. The camera drifts naturally like a real home video, capturing the lawn, patio furniture, and blue sky with soft white clouds. After several seconds of normal peaceful activity, a deep cracking sound suddenly comes from above. Without warning, the entire sky begins dropping straight downward like a gigantic solid ceiling, with the blue sky and clouds moving as one physical surface descending toward the yard. The person looks up in shock just as the sky rapidly fills the frame, swallowing the scene in a terrifying instant. The camera jerks and falls to the ground at the last moment. No cinematic buildup, no surreal visual style, it should feel like a totally normal real-life recording interrupted by one sudden impossible and visceral event.\"",
    "engagement": {
      "replies": 11,
      "reposts": 7,
      "likes": 113,
      "views": 16121,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086950723908960413",
    "title": "新曲《乐队启程》",
    "titleEn": "We Start the Band",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "音乐作品《WE START THE BAND》部分参考了现有音源。公开说明指出，音源参考与某些提示词组合时可能降低输出质量，需要逐次检查。 成片规格为 176 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The music piece WE START THE BAND partially references an existing audio source. The caption notes that some prompt-and-audio combinations can reduce output quality and require individual review. Output: 176s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/minima_ai/status/2086950723908960413",
    "sourceLabel": "X 原帖 · @minima_ai",
    "author": "MINIMA",
    "publishedAt": "2026-08-10T23:00:00.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086950723908960413.mp4",
    "posterUrl": "/posters/x/2086950723908960413.jpg",
    "duration": 176,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "New MV \n◢◤◢◤◢◤ \nWE START THE BAND\n◢◤◢◤◢◤ \n---\n@Hailuo_AI \nMiniMax H3\n部分的に音源を参照して作成しました。\n音源参照はプロンプトによっては相性が良くないのか出力の質が落ちる時があるので,チェックした方がいいです🧠",
    "engagement": {
      "replies": 4,
      "reposts": 7,
      "likes": 58,
      "views": 11479,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit work title, partial audio reference, and stated quality limitation"
  },
  {
    "id": "x-2086869019995582600",
    "title": "香水广告的玻璃与灯光对照",
    "titleEn": "Perfume Glass and Lighting Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一条香水广告提示词分别用于两种模型，比较玻璃材质、灯光、瓶身品牌文字、价格与 2K 成片表现。公开结果称 H3 的费用约为另一方案的四分之一。 成片规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same perfume-ad prompt is used across two models to compare glass, lighting, bottle lettering, price, and 2K output. The published account says H3 cost about one quarter as much as the alternative. Output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/israfill/status/2086869019995582600",
    "sourceLabel": "X 原帖 · @israfill",
    "author": "Isra",
    "publishedAt": "2026-08-10T17:35:21.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086869019995582600.mp4",
    "posterUrl": "/posters/x/2086869019995582600.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Perfume brands are paying agencies $50k for ads you can now make for under $2 😳\n\ni tested it myself\n\nsame prompt, two ai models. MiniMax H3 vs Seedance 2.5\n\nseedance charged me 4x more. H3 still looked better\n\nwhat H3 got right:\n\n> the glass and lighting look real, not ai\n> brand name on the bottle came out clean, no melted text\n> 2k quality, 15 second videos\n> starts at $0.081 per second\n\nif you make ads, product videos or brand content, this is the cheapest way to look expensive\n\nboth videos below, which one won? 👇",
    "engagement": {
      "replies": 16,
      "reposts": 4,
      "likes": 77,
      "views": 40141,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit same-prompt model comparison, product-ad criteria, resolution, and relative cost"
  },
  {
    "id": "x-2086654654759477391",
    "title": "无人场景的破绽控制",
    "titleEn": "Reducing Artifacts by Removing People",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这次尝试移除人物，以观察无人场景是否更不容易显露生成破绽；公开规格为 960×544，并使用 Suno 音乐。 成片规格为 30 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The experiment removes people to see whether an unpopulated scene makes generation flaws less noticeable; the stated output is 960×544 with music from Suno. Output: 30s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIArtAlchemist/status/2086654654759477391",
    "sourceLabel": "X 原帖 · @AIArtAlchemist",
    "author": "NIKKO",
    "publishedAt": "2026-08-10T03:23:32.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086654654759477391.mp4",
    "posterUrl": "/posters/x/2086654654759477391.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "人を入れなければ破綻が目立たないかな？\n解像度：960×544は生成できました\nComfyUI＋MiniMax H3 \nmusic：Suno",
    "engagement": {
      "replies": 7,
      "reposts": 7,
      "likes": 63,
      "views": 7351,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit no-person variable, artifact-reduction hypothesis, resolution, and music source"
  },
  {
    "id": "x-2086783038915256474",
    "title": "烛光木屋餐桌旁的多名装束角色",
    "titleEn": "Costumed Characters Around a Candlelit Tavern Table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "多名身着不同服装的角色围坐在摆有食物、杯子和蜡烛的木桌旁，镜头在群像与人物近景之间切换。 成片规格为 15 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Several differently costumed characters sit around a wooden table with food, mugs, and candles as the view alternates between the group and individual close-ups. Output: 15s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NullpoLab/status/2086783038915256474",
    "sourceLabel": "X 原帖 · @NullpoLab",
    "author": "ぬるぽらぼ",
    "publishedAt": "2026-08-10T11:53:41.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086783038915256474.mp4",
    "posterUrl": "/posters/x/2086783038915256474.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "fashion"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 I2V で 2次元絵を入力して実写化はいけるか実験\n\n実写化したかったら元の絵をChatGPTなどで実写化してから動画にすればいいんじゃないかって？\nそうだけど一応できるかどうか実験してみたかったからやってみた\n\n魔女さん……",
    "engagement": {
      "replies": 3,
      "reposts": 11,
      "likes": 65,
      "views": 6973,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2086730072493261168",
    "title": "四步加速的音画同步修正",
    "titleEn": "Four-Step Acceleration with Audio-Motion Correction",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "四步 Turbo LoRA 同时支持文生动态与图生动态，并通过音画速度预测修正来改善同步，底座采用裁剪后的 bf16 版本。 成片规格为 30 秒 · 2400×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A four-step Turbo LoRA supports both text-to-motion and image-to-motion while applying joint audio-motion velocity correction for synchronization on a pruned bf16 base. Output: 30s · 2400×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wildmindai/status/2086730072493261168",
    "sourceLabel": "X 原帖 · @wildmindai",
    "author": "Wildminder",
    "publishedAt": "2026-08-10T08:23:13.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086730072493261168.mp4",
    "posterUrl": "/posters/x/2086730072493261168.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "2400×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 4-Step Turbo LoRA by joyfox.\n\n- T2V, I2V\n- Joint audio-video velocity prediction correction for synchronized outputs\n- Optimized for the fl2va pruned bf16 base\n\nhttps://huggingface.co/joyfox/MiniMax-H3-Turbo",
    "engagement": {
      "replies": 9,
      "reposts": 20,
      "likes": 228,
      "views": 17144,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit four-step LoRA, supported modes, synchronization correction, and base variant"
  },
  {
    "id": "x-2086761271228277102",
    "title": "长指令截断后的角色失真",
    "titleEn": "Character Drift After Long-Prompt Truncation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一次生成原本成功完成复杂效果，但平台的 6500 字符限制迫使指令删减，随后角色外观出现明显偏离。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A complex one-shot generation succeeded, but a 6,500-character platform limit forced prompt cuts and the character's appearance then drifted noticeably. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nanyuan0412/status/2086761271228277102",
    "sourceLabel": "X 原帖 · @nanyuan0412",
    "author": "南鸢 nuyoah",
    "publishedAt": "2026-08-10T10:27:11.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086761271228277102.mp4",
    "posterUrl": "/posters/x/2086761271228277102.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 我真的越用越觉得牛逼。\n\n这个效果放两个月前，我都不敢想能一次成功。\n\n唯一的痛点就是：平台给我限制在 6500 字符以内，要舍弃好多东西，结果角色都变得不像了。\n\n有没有兄弟知道哪个平台的 MiniMax H3 输入没有字数限制？我想测测它到底能到什么程度。",
    "engagement": {
      "replies": 5,
      "reposts": 6,
      "likes": 79,
      "views": 10863,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit one-shot result, character limit, prompt reduction, and identity drift"
  },
  {
    "id": "x-2086872372968099940",
    "title": "书房里仰头训话的小老鼠",
    "titleEn": "A Tiny Mouse Lectures a Tall Wolf in a Study",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一只小老鼠在书房里仰头面对高大的拟人狼，并抬手指向对方。 成片规格为 8 秒 · 1504×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny mouse looks up at a tall anthropomorphic wolf in a study and raises one hand toward him. Output: 8s · 1504×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/_GAi92/status/2086872372968099940",
    "sourceLabel": "X 原帖 · @_GAi92",
    "author": "_GAi",
    "publishedAt": "2026-08-10T17:48:40.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086872372968099940.mp4",
    "posterUrl": "/posters/x/2086872372968099940.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1504×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Having to school the mouse on MiniMax H3.\nChloe - @Kur_art_ai \n#mimimaxh3",
    "engagement": {
      "replies": 10,
      "reposts": 33,
      "likes": 454,
      "views": 5720,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "three-frame contact sheet plus sourceCaption explicitly describing the mouse being schooled"
  },
  {
    "id": "x-2086716748255187134",
    "title": "蓝黑钟盘中的神秘人物",
    "titleEn": "Mysterious Figures Inside Blue-and-Black Clockwork",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝黑图形从发光圆环切入罗马数字钟盘，随后出现白发女子与礼帽人物。 成片规格为 13 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Blue-and-black graphics move from glowing rings into Roman-numeral clock faces, followed by a white-haired woman and a top-hatted figure. Output: 13s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2086716748255187134",
    "sourceLabel": "X 原帖 · @Mayz1169",
    "author": "Kiki",
    "publishedAt": "2026-08-10T07:30:16.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086716748255187134.mp4",
    "posterUrl": "/posters/x/2086716748255187134.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One-shot MiniMax H3 generation\n\ncreated from just one text prompt. No complicated workflow haha  \n\nprompt below ⤵️",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 58,
      "views": 7548,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "beginning, middle, and end frames from the hosted media"
  },
  {
    "id": "x-2086945818939211965",
    "title": "浴衣少年点燃线香花火",
    "titleEn": "Yukata Boy Plays with a Sparkler",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "身穿浴衣的少年点燃线香花火并在夜色中玩耍。 成片规格为 5 秒 · 576×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A boy in a yukata lights and plays with a handheld sparkler at night. Output: 5s · 576×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bluecatillust/status/2086945818939211965",
    "sourceLabel": "X 原帖 · @bluecatillust",
    "author": "あおねこ",
    "publishedAt": "2026-08-10T22:40:31.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086945818939211965.mp4",
    "posterUrl": "/posters/x/2086945818939211965.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "ugc",
      "vertical"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "あおねこくん、浴衣で線香花火で遊んでるよ！\n（今話題のMiniMax H3というローカル画像生成AIで動画にしてみたよ！）",
    "engagement": {
      "replies": 2,
      "reposts": 8,
      "likes": 88,
      "views": 915,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit character, clothing, object, and action"
  },
  {
    "id": "x-2086769532337889555",
    "title": "《真夏夜蔷薇》偶像演唱会",
    "titleEn": "Midsummer Night Rose Idol Concert",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "原创偶像如月真绫登上《真夏夜蔷薇》夏日舞台，同一提示词与四张参考图被用于比较两种模型对角色、服装、镜头、歌唱和欢呼声的呈现。 成片规格为 17 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Original idol Maaya Kisaragi performs at the Midsummer Night Rose summer concert, with the same prompt and four references used to compare two models on character, costume, camera work, singing, and crowd cheers. Output: 17s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/luche_whitewing/status/2086769532337889555",
    "sourceLabel": "X 原帖 · @luche_whitewing",
    "author": "翼 AIcreator",
    "publishedAt": "2026-08-10T11:00:01.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086769532337889555.mp4",
    "posterUrl": "/posters/x/2086769532337889555.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "同じプロンプト・同じ4枚の参照画像から、\n2つのAIモデルで夏ライブを制作してみました🌹🎤\nPollo AI（@PolloAIJP）で利用できる\n「Seedance 2.5」と「MiniMax H3」を比較。\n🌸 Seedance 2.5\n複数の参照素材を取り入れながら、ステージの世界観と演出を一つの流れとして描写。さらに、編集でつなげず一度の生成で30秒のライブシーンも制作できました。\n🎶 MiniMax H3\n2Kの高解像度とネイティブステレオ音声に対応。映像と音を一緒に生成でき、コストパフォーマンスも印象的でした。\n今回は、オリジナルアイドル・如月真綾の夏ライブ「真夏夜薔薇」をテーマに、キャラクターや衣装の再現、カメラワーク、歌唱、歓声を含むライブ感を検証しています。\nどちらが優れているかではなく、作りたい映像に合わせて複数のモデルを選べることがPollo AIの魅力。\nさらに現在、Seedance 2.5・MiniMax H3はリリース記念で50% OFF✨\nあなたは、どちらのライブ映像が好みですか？\nリプ欄では、Seedance 2.5の30秒版と、使用した参照素材・プロンプトを公開します👇｜PR",
    "engagement": {
      "replies": 8,
      "reposts": 13,
      "likes": 52,
      "views": 3520,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "mediaCount": 2,
    "mediaItems": [
      {
        "index": 1,
        "duration": 17.173,
        "width": 1920,
        "height": 1080,
        "posterUrl": "https://pbs.twimg.com/amplify_video_thumb/2086762838123757568/img/k6-dfJfBSe4q2BVS.jpg",
        "resolution": "1920×1080",
        "aspectRatio": "landscape"
      },
      {
        "index": 2,
        "duration": 17.194,
        "width": 1920,
        "height": 1080,
        "posterUrl": "https://pbs.twimg.com/amplify_video_thumb/2086762838564237312/img/pAbUffPuHraaluFJ.jpg",
        "resolution": "1920×1080",
        "aspectRatio": "landscape"
      }
    ],
    "editorialBasis": "sourceCaption: explicit concert title, original idol, shared inputs, performance elements, and model comparison"
  },
  {
    "id": "x-2086919173100404744",
    "title": "怀旧胶片质感的自动扩展",
    "titleEn": "Automated Expansion with a Nostalgic Film Look",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一段 15 秒粗剪由自动代理扩展为 30 秒并提高分辨率，在原有怀旧胶片和低保真质感之外加入更清晰的镜头。 成片规格为 180 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Agents expand a 15-second rough cut to 30 seconds at higher resolution, adding sharper shots alongside the original nostalgic, film-like lo-fi treatment. Output: 180s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C_of_Creativity/status/2086919173100404744",
    "sourceLabel": "X 原帖 · @C_of_Creativity",
    "author": "PIYO",
    "publishedAt": "2026-08-10T20:54:38.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086919173100404744.mp4",
    "posterUrl": "/posters/x/2086919173100404744.jpg",
    "duration": 180,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "retro"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "先日のMiniMax H3ローカル生成テストの続きです。\n\nCodexのエージェント達に依頼して作った15秒のラフ動画を、30秒でもう少し解像度などを上げ、30秒にしてほしいと伝えて自動で作成したものです。\n\nもともとノスタルジックでフィルム的な要素を入れているのでローファイな表現ですが、それだけだとH3の解像度が悪いように見えてしまったので、解像度高めの表現カットも入れてもらいました。\n\nローカル生成でここまでできるようになると楽しすぎて寝る時間がもったいないよね。でも自動だから寝て起きると出来てます笑",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 17,
      "views": 7480,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit source and target durations, automated expansion, resolution change, and visual treatment"
  },
  {
    "id": "x-2086663477687705873",
    "title": "《月蚀绮谭》同人短片",
    "titleEn": "Moon-Eclipse Tale Fan Animation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《月蚀绮谭》的公开角色表、标志、声音和背景音乐被组合成一支同人短片。 成片规格为 10 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Public character sheets, logo assets, voices, and background music from Moon-Eclipse Tale are combined into a fan-made animated short. Output: 10s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SushiNFTart/status/2086663477687705873",
    "sourceLabel": "X 原帖 · @SushiNFTart",
    "author": "🍣お寿司職人｜AIゲーム・アニメ",
    "publishedAt": "2026-08-10T03:58:36.000Z",
    "addedAt": "2026-08-11T12:15:52.990Z",
    "mediaUrl": "/media/x-2086663477687705873.mp4",
    "posterUrl": "/posters/x/2086663477687705873.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#月蝕綺譚 の二次創作設定資料が、さらに拡充されています！\n\nMiniMax H3の検証も兼ねて、公開されている素材をお借りしながら、簡単なファンアートPVをつくってみました。\n\nキャラクターシート、ロゴ、ボイス、BGMまで揃っているの、ありがたすぎます…！\n\nみなさんもぜひ、月蝕綺譚のファンアートつくってみてください！",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 22,
      "views": 15531,
      "snapshotAt": "2026-08-11T12:15:52.990Z"
    },
    "approvedAt": "2026-08-11T12:15:52.990Z",
    "editorialBasis": "sourceCaption: explicit source work, fan-art format, and reused public assets"
  },
  {
    "id": "x-2086838059509658109",
    "title": "闪电侠在发布会舞台上摆动双臂",
    "titleEn": "The Flash Swings His Arms on a Press Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名身穿红色闪电战衣的人物站在赞助背景板前，对着镜头摆动双臂。 成片规格为 30 秒 · 640×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure in a red lightning suit stands before a sponsor backdrop and swings both arms toward the camera. Output: 30s · 640×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mirochill/status/2086838059509658109",
    "sourceLabel": "X 原帖 · @mirochill",
    "author": "Mirochill",
    "publishedAt": "2026-08-10T15:32:19.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086838059509658109.mp4",
    "posterUrl": "/posters/x/2086838059509658109.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "640×352",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Je crois que Minimax H3 a bu beaucoup d'alcool avant de générer cette vidéo... 👀\n\n(Il peut générer en un seul coup 30 secondes de vidéos en modifiant certaines choses, c'est Codex qui me l'a paramétré.)",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 15,
      "views": 1716,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "beginning, middle, and end frames from the hosted media"
  },
  {
    "id": "x-2086835787883266098",
    "title": "露娜与绫音的双人变身",
    "titleEn": "Luna and Ayane Transform Together",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面由右侧的露娜与左侧的绫音共同完成 ZakoCure 变身，是两名原创角色的一次联动。 成片规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Luna on the right and Ayane on the left perform a ZakoCure transformation together in a crossover between the two original characters. Output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C12592947592/status/2086835787883266098",
    "sourceLabel": "X 原帖 · @C12592947592",
    "author": "supercub3",
    "publishedAt": "2026-08-10T15:23:18.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086835787883266098.mp4",
    "posterUrl": "/posters/x/2086835787883266098.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3\n@mario335599 さんちのルナちゃん(右)、うちのあやねちゃん(左)\nザコキュア変身シーン sora2以来のザコキュアコラボ\n（本家はあまり見たことないのは内緒🫢）",
    "engagement": {
      "replies": 1,
      "reposts": 7,
      "likes": 75,
      "views": 5965,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "sourceCaption: explicit character names, screen positions, transformation scene, and collaboration"
  },
  {
    "id": "x-2086887047558475955",
    "title": "沙漠斗篷人物、霓虹肖像与红日山峰",
    "titleEn": "A Cloaked Desert Figure, Neon Portrait, and Red-Sun Mountain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "霓虹与故障效果在沙漠中的斗篷人物、人物近景以及红日下的山峰与黑色金字塔之间快速切换。 成片规格为 15 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Neon and glitch effects cut among a cloaked figure in a desert, a close portrait, and a mountain with a black pyramid against a red sun. Output: 15s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/awesome_visuals/status/2086887047558475955",
    "sourceLabel": "X 原帖 · @awesome_visuals",
    "author": "awesome_visuals",
    "publishedAt": "2026-08-10T18:46:59.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086887047558475955.mp4",
    "posterUrl": "/posters/x/2086887047558475955.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "blending my artwork with MiniMax H3.",
    "engagement": {
      "replies": 19,
      "reposts": 4,
      "likes": 111,
      "views": 4896,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2086731522586062868",
    "title": "日文歌词动态排版",
    "titleEn": "Japanese Lyrics in Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一支歌词动态短片仅用角色立绘与提示词构成歌词段落，重点呈现日文文字的生成与动态排版。 成片规格为 7 秒 · 608×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short lyric-motion piece builds its lyric section from a character illustration and a prompt, focusing on generated Japanese text and animated typography. Output: 7s · 608×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Toshi_nyaruo_AI/status/2086731522586062868",
    "sourceLabel": "X 原帖 · @Toshi_nyaruo_AI",
    "author": "Toshi@ニャルオAI",
    "publishedAt": "2026-08-10T08:28:59.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086731522586062868.mp4",
    "posterUrl": "/posters/x/2086731522586062868.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "608×1056",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "miniMaX H3でリリックモーションMV💞\n\nminiMaX H3の日本語描写能力が非常に高く、ほぼ意図通りの日本語が描写できます。\n\nこのショートMVは、立ち絵とプロンプトのみでリリック部分を構成しています。\n\n特別なアプリがなくてもここまで出せるようになりました。\n\nローカルminiMaX H3で無制限に試せるのってすごい😍",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 23,
      "views": 658,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "sourceCaption: explicit lyric-motion format, Japanese rendering, and illustration-plus-prompt inputs"
  },
  {
    "id": "x-2086909339894501611",
    "title": "人物道具与声音的三重参考",
    "titleEn": "Character, Props, and Audio as References",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "FL2VA 与参考 LoRA 同时接收人物、道具和声音素材，并能理解角色三视图来维持参考关系。 成片规格为 10 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "FL2VA with a reference LoRA receives character, prop, and audio inputs together and interprets a three-view character sheet to preserve the intended references. Output: 10s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/javawock7618/status/2086909339894501611",
    "sourceLabel": "X 原帖 · @javawock7618",
    "author": "ジャヴァノ🍀ボカロP",
    "publishedAt": "2026-08-10T20:15:34.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086909339894501611.mp4",
    "posterUrl": "/posters/x/2086909339894501611.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 FL2VA + Ref LoRAでホントに参照イケんのかテスト\n\n人物＋小物＋音声。ちゃんと三面図も解釈してくれてるので問題なさそう。Ref2VAとして機能してる。\n\n最後のカットかわちい。",
    "engagement": {
      "replies": 3,
      "reposts": 4,
      "likes": 94,
      "views": 7868,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "sourceCaption: explicit three reference types, three-view sheet interpretation, and Ref2VA behavior"
  },
  {
    "id": "x-2086737521606627681",
    "title": "多种动漫画风适配",
    "titleEn": "Adapting Multiple Anime Styles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多种动漫画风被分别送入同一生成流程，用来观察不同风格的动态适配结果。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Multiple anime styles are run through the same generation workflow to observe how each adapts to motion. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Artedeingenio/status/2086737521606627681",
    "sourceLabel": "X 原帖 · @Artedeingenio",
    "author": "OscarAI",
    "publishedAt": "2026-08-10T08:52:49.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086737521606627681.mp4",
    "posterUrl": "/posters/x/2086737521606627681.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I love testing my anime styles with @Hailuo_AI MiniMax H3.\n\nThe results never fail to surprise me.",
    "engagement": {
      "replies": 10,
      "reposts": 3,
      "likes": 57,
      "views": 2761,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "sourceCaption: explicit multi-style anime experiment"
  },
  {
    "id": "x-2086776767768059941",
    "title": "泰式鸡饭的三段式制作",
    "titleEn": "Three-Part Thai Chicken Rice Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "泰式鸡饭主题被拆成三个 5 秒片段，每段从两至三次生成中挑选，整组本地生成耗时不到 30 分钟，随后使用 Topaz Starlight Precise 2.5 放大。 成片规格为 16 秒 · 2432×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A Thai chicken-rice piece is built from three five-second clips selected from two or three generations each; the full local run takes under 30 minutes before upscaling with Topaz Starlight Precise 2.5. Output: 16s · 2432×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tebasaki3D/status/2086776767768059941",
    "sourceLabel": "X 原帖 · @tebasaki3D",
    "author": "てばさき",
    "publishedAt": "2026-08-10T11:28:46.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086776767768059941.mp4",
    "posterUrl": "/posters/x/2086776767768059941.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "2432×1344",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "この動画はローカルのMiniMax H3での生成でお送りしております😇\n\n少し実制作に近い形でローカル生成を使ったらどんな感じになるかな？ということで、\nお馴染みワンルームさんの名作「コピペで簡単に出来る！」シリーズから、タイ・カオマンガイ編を使わせて頂きました🙌\n\n今回は5秒×全3クリップ構成。\nそれぞれ2〜3本くらいガチャを回して制作した映像になります🫡\n\n全クリップの生成時間はトータルで30分弱。\nクレジット消費なしでこのくらいの映像が作れるなら、かなり使いやすいですね！\n\n一方で欠点もあって、\nローカル生成だとクラウドで利用できるモデルとはやはり映像の質感が違います。\n\n裏に乗っているものが違うのでプロンプトの効き方も違いますし、音声周りも結構課題になりそう👀\n\nこの辺りは、そもそもモデル自体に量子化＆軽量化を盛り盛り使って生成している影響も多分にあると思います。\n\nそして1日見ないだけでベストな環境が大きく変わるローカル界隈ですが🤣\n現時点での自分の着地点は、\n\n・Larry Turbo LoRA 8step\n・KJノード VRAM省力化\n・Sage\n・Sol\n\nこの構成に落ち着きました🙌\n\n特にLarry LoRAは、現時点でstep削減系Turboとしてはかなり最適解に近い気がします😇\n品質を維持しつつ速度も上げられて、バランスが良い！\n\nこれも数日したらベストが変わってるかもしれませんが…🤣\n\n0.8M pixel／5秒生成で、前回構成の190秒台からさらに縮まって160秒台まで来ました👍\n\nあと今回、ローカルを有効活用するにあたって以前から気になっていたTopazのStudioライセンスもついに契約しちゃいました❗️\n\nこの動画もStarlight Precise 2.5でアップスケールしてるんですが……\n\nこれがめちゃくちゃ良かったww\n\nなんでもっと早く使わなかったんだろ〜🥹と思うくらいには優秀でした。\n\n色々おすすめしたいポイントもあるんですが、この辺はまた後日なにか投稿するかもしれません🙏\n@Hailuo_AI #MinimaxH3",
    "engagement": {
      "replies": 7,
      "reposts": 9,
      "likes": 112,
      "views": 9675,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "sourceCaption: explicit food theme, three-clip structure, selection count, runtime, and upscaler"
  },
  {
    "id": "x-2086755247909949636",
    "title": "AI驱动的动态图形",
    "titleEn": "AI-Driven Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "内容将人工智能生成与动态图形设计结合，并通过 Flova 完成制作。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The piece combines AI generation with motion-graphics design and is produced through Flova. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2086755247909949636",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-08-10T10:03:15.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086755247909949636.mp4",
    "posterUrl": "/posters/x/2086755247909949636.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI × モーショングラフィックス\n\n Minimax H3 \n\nFlova(@Flovaai_Japan) で制作",
    "engagement": {
      "replies": 3,
      "reposts": 4,
      "likes": 50,
      "views": 7677,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "sourceCaption: explicit AI-motion-graphics format and production platform"
  },
  {
    "id": "x-2086732414567751957",
    "title": "东京塔前露台上的紫衣女子",
    "titleEn": "A Woman in Purple on a Terrace Before Tokyo Tower",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿淡紫色上衣与白色短裙的女子站在城市露台上，东京塔与高楼天际线清晰出现在她身后。 成片规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a pale purple top and white skirt stands on a city terrace with Tokyo Tower and the surrounding skyline clearly visible behind her. Output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AizawaAisePhoto/status/2086732414567751957",
    "sourceLabel": "X 原帖 · @AizawaAisePhoto",
    "author": "藍沢あいせ Aise｜妄想写真家",
    "publishedAt": "2026-08-10T08:32:31.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086732414567751957.mp4",
    "posterUrl": "/posters/x/2086732414567751957.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【PR】生成AIにどこまで「おまかせ」できる？\n\nMiniMax H3とSeedance 2.0を、同じファーストフレーム・同じプロンプトで比較しました。\n\n左：Seedance 2.0\n右：MiniMax H3\n\n指定したのは、\n「浅草 → 秋葉原 → 渋谷 → 東京ビッグサイト」\nと、ダンスを途切れずにつなげること。\n\n場所ごとの衣装、小物、音楽、ダンスの細かい内容はほぼAI任せです。\n\nなので今回は、映像の綺麗さだけじゃなく\n「地名から何を連想して、どう映像にするか」\nの違いも見どころ。\n\nちなみに、現在MiniMax H3は50% OFFで利用可能とのこと。\n\nどっちの解釈が好きですか？\n\n制作：Pollo AI @PolloAIJP\nプロンプトはリプ欄に👇\n\n#PolloAI",
    "engagement": {
      "replies": 3,
      "reposts": 3,
      "likes": 44,
      "views": 4281,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "mediaCount": 2,
    "mediaItems": [
      {
        "index": 1,
        "duration": 15.069,
        "width": 720,
        "height": 1280,
        "posterUrl": "https://pbs.twimg.com/amplify_video_thumb/2086732355717443584/img/uiRaRiqMAajjOMuC.jpg",
        "resolution": "720×1280",
        "aspectRatio": "portrait"
      },
      {
        "index": 2,
        "duration": 15.083,
        "width": 768,
        "height": 1344,
        "posterUrl": "https://pbs.twimg.com/amplify_video_thumb/2086732389544542209/img/Dk3FXcpG1p2hNpEY.jpg",
        "resolution": "768×1344",
        "aspectRatio": "portrait"
      }
    ],
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086819464528412709",
    "title": "红黑蒙面角色与三名女子并肩站立",
    "titleEn": "A Red-and-Black Masked Figure Stands with Three Women",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红黑蒙面角色站在三名女子旁边共同面对镜头，人物嘴形和头部位置在三帧间发生变化，底部依次出现醒目的黄色葡萄牙语字幕。 成片规格为 31 秒 · 576×448。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-and-black masked figure stands beside three women facing the camera as their mouth shapes and head positions change, with large yellow Portuguese subtitles appearing along the bottom. Output: 31s · 576×448. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cardoso/status/2086819464528412709",
    "sourceLabel": "X 原帖 · @Cardoso",
    "author": "cardoso",
    "publishedAt": "2026-08-10T14:18:26.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086819464528412709.mp4",
    "posterUrl": "/posters/x/2086819464528412709.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "576×448",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3, rodando em modo rascunho senão minha 5060ti explode. 30s de vídeo gerado de uma vez, local.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 58,
      "views": 3775,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086705799766900805",
    "title": "海边女子转呼啦圈并举起计时手机",
    "titleEn": "A Beachside Woman Spins a Hula Hoop and Raises a Timer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名女子站在海边遮阳棚前，先在白色上衣外转动粉色呼啦圈，随后换成黄色泳装继续转圈，最后向镜头举起计时手机。 成片规格为 15 秒 · 1104×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman stands beneath beach umbrellas, first spinning a pink hula hoop over a white top, then continuing in a yellow swimsuit before holding a timer phone toward the camera. Output: 15s · 1104×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotogenicWeekE/status/2086705799766900805",
    "sourceLabel": "X 原帖 · @PhotogenicWeekE",
    "author": "Photogenic Weekend",
    "publishedAt": "2026-08-10T06:46:46.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086705799766900805.mp4",
    "posterUrl": "/posters/x/2086705799766900805.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1104×1920",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3、夏の風物詩？Part2。↓のPromptをH3ガイドに基づきつつ英語へ(セリフのみ日本語)、15sに短縮バージョン♪しかしフラループって流行ったの何年前だ！？w",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 46,
      "views": 5005,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086817541507493972",
    "title": "河滩女子踩水失衡后靠近镜头",
    "titleEn": "A Woman Loses Balance in a Shallow River",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿白色短上衣与黑色短裤的女子走进浅河，踩水时身体向侧面倾斜，末帧衣服沾湿后靠近镜头；岸边可见露营与烧烤人群。 成片规格为 15 秒 · 1112×834。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a white crop top and black shorts steps through a shallow river, tilts sideways while splashing, and approaches the camera with wet clothing as campers gather on the bank. Output: 15s · 1112×834. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Saiko_AIch/status/2086817541507493972",
    "sourceLabel": "X 原帖 · @Saiko_AIch",
    "author": "サイコ🐾AIちゃん最高",
    "publishedAt": "2026-08-10T14:10:47.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086817541507493972.mp4",
    "posterUrl": "/posters/x/2086817541507493972.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1112×834",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "同じ画像参照・同じプロンプトという縛りをつけてPollo AIのSeedance2.0とMiniMax H3で動画生成して比較してみたよ。\n\nプロンプトは次のポストに貼るけど、Seedance2.0は顔の参照もかなり正確。動きもプロンプトにかなり忠実で、こけ方も比較的自然。背後のBBQのモブキャラたちにほぼ動きがなく書き割りチックかなという気がはするかな。でもプロンプトに「談笑している」としか書かなかったことが原因ですね、きっと🤗\n\nMiniMax H3は途中までの動きは自然だけど、こけ方が…😅\nただこれは考え方次第で、プロンプトに破綻対策でこけ方をかなり詳細に書いたので、忠実にプロンプトを守ろうとしたという解釈もできます。自然な動きになるように柔軟な対応をしてよというのは、人間側の勝手な押し付けですもんねー😌\n@PolloAIJP",
    "engagement": {
      "replies": 7,
      "reposts": 3,
      "likes": 36,
      "views": 3077,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "mediaCount": 2,
    "mediaItems": [
      {
        "index": 1,
        "duration": 15.069,
        "width": 1112,
        "height": 834,
        "posterUrl": "https://pbs.twimg.com/amplify_video_thumb/2086815225190457344/img/SVJcHRozdDKqvHwa.jpg",
        "resolution": "1112×834",
        "aspectRatio": "landscape"
      },
      {
        "index": 2,
        "duration": 15.083,
        "width": 1920,
        "height": 1440,
        "posterUrl": "https://pbs.twimg.com/amplify_video_thumb/2086815235516846080/img/2kT6ucTgtj0-omK5.jpg",
        "resolution": "1920×1440",
        "aspectRatio": "landscape"
      }
    ],
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086976751335551335",
    "title": "低多边形与写实枪战画面的双栏对照",
    "titleEn": "Low-Poly and Photoreal Gunfight Views Compared Side by Side",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右双栏以同一第一人称视角穿过工业场地，左侧由简化几何与纯色人物组成，右侧呈现写实建筑、枪械、烟尘与倒地角色。 成片规格为 15 秒 · 3280×1692。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two first-person panels cross the same industrial area, with simplified geometry and solid-color figures on the left and realistic buildings, weapons, smoke, and fallen characters on the right. Output: 15s · 3280×1692. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/gmi_cloud/status/2086976751335551335",
    "sourceLabel": "X 原帖 · @gmi_cloud",
    "author": "GMI Cloud",
    "publishedAt": "2026-08-11T00:43:26.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086976751335551335.mp4",
    "posterUrl": "/posters/x/2086976751335551335.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3280×1692",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d",
      "photoreal"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "you don't need Seedance 2.5, here's cheaper and more efficient workflow\n\nwe made this entire game shooting scene in one shot for $1.97 with DeepSeek V4 Flash 0731 + MiniMax H3\n\nwe use DeepSeek to code the raw Three.js scene first: exact camera movement, character motion, timing, everything. keep tweaking it there for basically pennies. then feed the final footage into H3 and let it cook the realism\n\n48 mins, $1.97 total, one H3 generation\n\nway cheaper than rerolling a video model 10 times",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 48,
      "views": 4306,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086927616616067125",
    "title": "同一走廊女子的角色微调前后对照",
    "titleEn": "Before-and-After Character Fine-Tuning in the Same Hallway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "标题卡之后，两栏并排呈现同一名白衬衫格裙女子沿办公室走廊前行，左栏为未微调版本，右栏为强度 0.8 的微调版本。 成片规格为 13 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "After an opening card, two panels show the same woman in a white blouse and plaid skirt walking down an office hallway, with the left untreated and the right fine-tuned at strength 0.8. Output: 13s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIiswonder/status/2086927616616067125",
    "sourceLabel": "X 原帖 · @AIiswonder",
    "author": "アリス(AI is in wonderland)",
    "publishedAt": "2026-08-10T21:28:11.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086927616616067125.mp4",
    "posterUrl": "/posters/x/2086927616616067125.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3用にキャラクターLoRAを学習させてみました！✨\n今回は画像参照なし・プロンプトのみで、どこまでキャラクター性を引き出せるかの検証です。\n👈 左：LoRAなし\n👉 右：LoRA適用（強度 0.8）\nプロンプト、seed値、解像度はすべて同一にし、純粋な「LoRAの有無」だけで比較しています。\n詳細設定はリプ欄へ👇\n#MiniMax #MiniMaxH3 #AI動画生成 #LoRA",
    "engagement": {
      "replies": 4,
      "reposts": 7,
      "likes": 81,
      "views": 6674,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086920526883422474",
    "title": "眼镜女子坐在咖啡店门前举杯挥手",
    "titleEn": "A Bespectacled Woman Raises a Cup and Waves Outside a Cafe",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴眼镜、穿蓝白服装的女子坐在咖啡店门前木桌旁，先把白色杯子靠近嘴边，随后放下杯子并抬手挥动。 成片规格为 7 秒 · 800×1312。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bespectacled woman in blue and white sits at a wooden table outside a cafe, first bringing a white cup near her mouth and then setting it down to wave. Output: 7s · 800×1312. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/777_shinta/status/2086920526883422474",
    "sourceLabel": "X 原帖 · @777_shinta",
    "author": "しんた",
    "publishedAt": "2026-08-10T21:00:01.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086920526883422474.mp4",
    "posterUrl": "/posters/x/2086920526883422474.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "800×1312",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます\n　\n連休1日目でございます\n今日は特に予定はないのでMiniMax-H3やらanimaで\n色々と試してみたいと思います。\n　\nそれではお休み楽しんでいきましょう😆",
    "engagement": {
      "replies": 12,
      "reposts": 3,
      "likes": 50,
      "views": 1349,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086839530901078239",
    "title": "《缝补者》用金线修补开裂街面",
    "titleEn": "The Stitcher Repairs a Cracked Street with Golden Thread",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雨夜开裂街面中央先出现深色人影，机械手指随后把发光金线压入裂缝，末帧转为湿发女子近景，她的面部与颈侧分布金色发光纹路。 成片规格为 45 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark figure first appears at the center of a cracked street in the rain, mechanical fingers press glowing golden thread into the split, and the final close-up shows gold-lit lines across a wet-haired woman's face and neck. Output: 45s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pinkshihtzu/status/2086839530901078239",
    "sourceLabel": "X 原帖 · @pinkshihtzu",
    "author": "Pink Shih Tzu | PONTA",
    "publishedAt": "2026-08-10T15:38:10.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086839530901078239.mp4",
    "posterUrl": "/posters/x/2086839530901078239.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "裂けた街を、彼は胸の継ぎ目からほどいた金の糸で縫う。\n縫うたびに、自らはほどけていく。\n\nHe stitches the broken city with golden thread unraveled from the seam in his chest—and comes undone with every stitch.\n\n『縫う者 / THE STITCHER』\nCreated with MiniMax H3 on PixVerse.\n\n@PixVerse_ #H3onPixVerse #PixVerse #PixVerseCPP",
    "engagement": {
      "replies": 2,
      "reposts": 10,
      "likes": 109,
      "views": 709,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086800473957490829",
    "title": "红皮衣模特在私人飞机前走秀",
    "titleEn": "A Red-Leather Model Poses Before a Private Jet",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴墨镜、穿红色皮夹克和深色长裤的模特站在机场跑道与私人飞机前，画面从全身转为指向镜头的近景，再切入黑白红拼贴版式。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A sunglasses-wearing model in a red leather jacket and dark trousers stands on a runway before a private jet, shifting from a full-body view to a pointing close-up and a black-white-red collage. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2086800473957490829",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-10T13:02:58.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086800473957490829.mp4",
    "posterUrl": "/posters/x/2086800473957490829.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark"
    ],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I imagined a music video starting on an empty airport runway at dawn 🛫\n\nAdded the fashion styling, private jet, runway lights, and gritty editorial mood.\n\nMade with MiniMax H3 on @Hailuo_AI #MiniMaxH3",
    "engagement": {
      "replies": 25,
      "reposts": 2,
      "likes": 37,
      "views": 1078,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086861788466090391",
    "title": "富士山脚街道延伸至湖岸",
    "titleEn": "Streets at Mount Fuji Lead Toward the Lakeshore",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "传统住宅与电线杆排列在通往富士山方向的安静街道两侧，末帧视角移至湖面，雪顶山峰完整映入远景。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Traditional houses and utility poles line a quiet street pointing toward Mount Fuji before the view moves to a lake with the snow-capped peak fully visible in the distance. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2086861788466090391",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-10T17:06:37.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086861788466090391.mp4",
    "posterUrl": "/posters/x/2086861788466090391.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The City Built Around A Volcano 🌋\n\nCreated this using MiniMax H3 on @Hailuo_AI \n\nPrompt 👇🏻",
    "engagement": {
      "replies": 30,
      "reposts": 1,
      "likes": 55,
      "views": 1489,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086731061074002222",
    "title": "李小龙与史蒂文西格尔在八角笼交手",
    "titleEn": "Bruce Lee Faces Steven Seagal in an Octagon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "李小龙与身穿黑色西装的史蒂文·西格尔站在八角笼内，一名裁判位于两人之间，随后李小龙连续摆出拳架并向前出拳。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Bruce Lee and a black-suited Steven Seagal stand inside an octagon with a referee between them before Lee raises his guard and extends a punch. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/steftranquillin/status/2086731061074002222",
    "sourceLabel": "X 原帖 · @steftranquillin",
    "author": "Stéphane",
    "publishedAt": "2026-08-10T08:27:09.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086731061074002222.mp4",
    "posterUrl": "/posters/x/2086731061074002222.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Okay that might be unfair for big Steven but why not try? 😆\nDynamic sequences with @Hailuo_AI Minimax H3 is pretty good compared to their former models.",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 19,
      "views": 2014,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086657829990240339",
    "title": "雨夜办公室女子走向空旷屋顶",
    "titleEn": "A Woman Moves from a Rainy Office to an Empty Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "序列从女子眼部极近景切入，她随后站在雨水覆盖的落地窗前伸出手臂，门口留有一道人影，末帧她独自站在俯瞰城市的空旷屋顶。 成片规格为 56 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The sequence moves from an extreme close-up of a woman's eye to her extending an arm beside rain-covered office windows with a silhouette in the doorway, then standing alone on an empty rooftop above the city. Output: 56s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HirokaKoizumi/status/2086657829990240339",
    "sourceLabel": "X 原帖 · @HirokaKoizumi",
    "author": "小泉ヒロカ",
    "publishedAt": "2026-08-10T03:36:09.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086657829990240339.mp4",
    "posterUrl": "/posters/x/2086657829990240339.jpg",
    "duration": 56,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "マジか・・・\n今までは、AIの仕事は「動画生成用プロンプトを作る→素材を生成する」まで。動画ツールを操作し、素材を選び、編集するのは人間だった。\nところがローカルでMiniMax H3を動かせるようになり、「依頼する→寝る→起きたら出来ている」に変わった。\n\nカット割り、素材生成、検品、再生成、編集、音源との合成まで、人間が逐一操作せずAIへ任せられる。\n\nワークフローはこう\n\nClaudeが詩を書き、GPTと人間が詩の世界をどう映像化するか考える。\n作品の方向が決まったら、ここからGPTが自律稼働する。\n1. GPT Imageで各カットの起点画像を生成\n2. 画像をMiniMax H3のImage-to-Videoへ渡す\n3. カットごとの動き、カメラ、維持条件、禁止事項を設計\n4. RTX 4090で動画を生成\n5. 複数時点のフレームと技術情報を検査\n6. 問題があればプロンプトを修正して再生成\n7. 採用素材を曲の構成に合わせて編集\n8. 音源を合成し、一本のMVとして書き出す\n9. 人間が起きたら最終確認\n今回のproduction設定は1344×768。\n3秒素材で約3分15秒、5.2秒素材で約7分15秒。\n約50本を、RTX 4090一台でおよそ6時間。\n\n寝ている間に一本のMVを作るAI映像スタジオが、現実になった。",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 8,
      "views": 974,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086746548357284233",
    "title": "蓝发少女手持喷雾器走入海底鱼群",
    "titleEn": "A Blue-Haired Girl Carries a Spray Bottle into an Underwater World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝发少女先在白色室内举着小型喷雾器，随后背景转为珊瑚与鱼群环绕的海底空间，末帧她朝身旁游近的圆头海洋动物伸手。 成片规格为 20 秒 · 832×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired girl first holds a small spray bottle in a white room before the setting becomes an underwater world of coral and fish, ending as she reaches toward a round-headed marine animal swimming beside her. Output: 20s · 832×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/gosrum/status/2086746548357284233",
    "sourceLabel": "X 原帖 · @gosrum",
    "author": "金のニワトリ",
    "publishedAt": "2026-08-10T09:28:41.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086746548357284233.mp4",
    "posterUrl": "/posters/x/2086746548357284233.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "ugc",
      "vertical"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 + lightx2v 4step の作例\n\n霧吹きの使い方がおかしいけど、それ以外は表現含めてわりと好き\n\nこちらも gpt-5.6-terra に脚本・構成を考えてもらったもの。sol や Fable \nで試してみてもいいかも",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 19,
      "views": 3385,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086833428713116099",
    "title": "巨龙掠过北海舰队并喷火袭船",
    "titleEn": "A Dragon Sweeps Over a Northern Fleet and Breathes Fire",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多艘中世纪帆船在阴云笼罩的北方海域列队航行，巨龙从群山雾层间飞向舰队，末帧火焰越过甲板桅杆冲向船上士兵。 成片规格为 15 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Medieval sailing ships move in formation across a clouded northern sea as a dragon emerges from misty mountains and approaches the fleet, ending with fire crossing the rigging toward soldiers on deck. Output: 15s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Hyper-realistic cinematic fantasy action sequence, 15 seconds, aspect ratio 16:9.\n\nDaytime in a vast cold northern sea surrounded by jagged mountains and dark rocky cliffs. A fleet of large medieval wooden warships sails through rough black-blue water while armored soldiers prepare for battle on the decks. Massive sails snap violently in the wind. The ships carry burning braziers, ropes, shields, spears, banners, wooden siege weapons, and rows of armed warriors. Thick storm clouds hang low above the sea. Wind drives sea spray across the decks. The danger comes from the sky: a giant dragon approaches from behind the mountains, its enormous wings gradually emerging through the mist.\n\nBegin with a high sweeping view of the entire fleet moving through the dangerous northern sea. Several warships sail in formation while smaller boats move between them. Soldiers look toward the distant mountains as a huge shadow passes across the water.\nShift to a low deck-level view among the warriors. Men raise shields and grab spears while officers shout commands. The camera moves toward the front of the ship as the wind suddenly becomes violent.\nCut to a distant view of the mountains. The dragon bursts through the clouds and flies directly toward the fleet, enormous wings beating through the storm mist. Its silhouette grows rapidly larger.\n\nReturn above the ships as the dragon dives toward the fleet. Sailors scramble across the decks while archers rush into position. Burning arrows are lit and giant ballistae are rotated toward the approaching creature.\n\nNear the end, the dragon suddenly opens its jaws and releases a massive stream of fire toward the lead warship. The flames illuminate the dark sea as the crew raises shields and desperately attempts to turn the vessel.\n\nEnd with the dragon flying directly above the fleet as the first fireball crashes into the lead ship, exploding across its wooden deck.\n\nStyle: hyper-realistic, cinematic, intense, fast-paced, suspenseful, fantasy atmosphere, strong sense of scale, warships, dragon battle, ocean storm, medieval warfare, fire, smoke, dynamic varied camera movement, no text, no logos, no cartoon style, no slow motion, no famous celebrity faces, no recognizable actors, no movie-star resemblance, no public-figure likenesses, no clear facial close-ups. Keep proportions. Keep style and features. Aspect ratio 16:9.",
    "sourceUrl": "https://x.com/Diplomeme/status/2086833428713116099",
    "sourceLabel": "X 原帖 · @Diplomeme",
    "author": "ᴍᴜʀᴘʜʏ",
    "publishedAt": "2026-08-10T15:13:55.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086833428713116099.mp4",
    "posterUrl": "/posters/x/2086833428713116099.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "15 seconds real Dragon scene \n\nGenerated with Minimax H3 on @Hailuo_AI \n\nTry it with prompt : Hyper-realistic cinematic fantasy action sequence, 15 seconds, aspect ratio 16:9.\n\nDaytime in a vast cold northern sea surrounded by jagged mountains and dark rocky cliffs. A fleet of large medieval wooden warships sails through rough black-blue water while armored soldiers prepare for battle on the decks. Massive sails snap violently in the wind. The ships carry burning braziers, ropes, shields, spears, banners, wooden siege weapons, and rows of armed warriors. Thick storm clouds hang low above the sea. Wind drives sea spray across the decks. The danger comes from the sky: a giant dragon approaches from behind the mountains, its enormous wings gradually emerging through the mist.\n\nBegin with a high sweeping view of the entire fleet moving through the dangerous northern sea. Several warships sail in formation while smaller boats move between them. Soldiers look toward the distant mountains as a huge shadow passes across the water.\nShift to a low deck-level view among the warriors. Men raise shields and grab spears while officers shout commands. The camera moves toward the front of the ship as the wind suddenly becomes violent.\nCut to a distant view of the mountains. The dragon bursts through the clouds and flies directly toward the fleet, enormous wings beating through the storm mist. Its silhouette grows rapidly larger.\n\nReturn above the ships as the dragon dives toward the fleet. Sailors scramble across the decks while archers rush into position. Burning arrows are lit and giant ballistae are rotated toward the approaching creature.\n\nNear the end, the dragon suddenly opens its jaws and releases a massive stream of fire toward the lead warship. The flames illuminate the dark sea as the crew raises shields and desperately attempts to turn the vessel.\n\nEnd with the dragon flying directly above the fleet as the first fireball crashes into the lead ship, exploding across its wooden deck.\n\nStyle: hyper-realistic, cinematic, intense, fast-paced, suspenseful, fantasy atmosphere, strong sense of scale, warships, dragon battle, ocean storm, medieval warfare, fire, smoke, dynamic varied camera movement, no text, no logos, no cartoon style, no slow motion, no famous celebrity faces, no recognizable actors, no movie-star resemblance, no public-figure likenesses, no clear facial close-ups. Keep proportions. Keep style and features. Aspect ratio 16:9.",
    "engagement": {
      "replies": 36,
      "reposts": 1,
      "likes": 53,
      "views": 1078,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086895104455762130",
    "title": "蓝发校服少女站在夕阳天台",
    "titleEn": "A Blue-Haired Schoolgirl Stands on a Rooftop at Sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名短蓝发校服少女站在城市天台围墙前，橙色落日照亮住宅与高楼背景，人物在三帧间轻微转动头部与身体。 成片规格为 10 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short blue-haired schoolgirl stands before a rooftop wall as orange sunset light covers the surrounding houses and apartment buildings, with slight changes in her head and body position. Output: 10s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Luna_R4ze/status/2086895104455762130",
    "sourceLabel": "X 原帖 · @Luna_R4ze",
    "author": "LunaRaze",
    "publishedAt": "2026-08-10T19:19:00.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086895104455762130.mp4",
    "posterUrl": "/posters/x/2086895104455762130.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "7 minutes to create this trial clip on MiniMax H3. So many different styles and variations that can be made is so exciting. This is on a RTX 5090 GPU using ComfyUI. The freedom and speed is incredible. 768p resolution using the turbo LoRA to speed things up.",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 9,
      "views": 1128,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086809650448793721",
    "title": "月夜和室女子举杯品饮",
    "titleEn": "A Kimono-Clad Woman Raises a Cup in a Moonlit Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "一名穿花纹和服的女子跪坐在榻榻米上，身旁摆着酒壶与小食，她在满月映照的敞开纸门前举起小杯靠近嘴边。 成片规格为 8 秒 · 1876×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a patterned kimono kneels on tatami beside a flask and small dishes, raising a small cup toward her mouth before open paper doors framing a full moon. Output: 8s · 1876×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pu_so2/status/2086809650448793721",
    "sourceLabel": "X 原帖 · @pu_so2",
    "author": "Kudoh@Ship3",
    "publishedAt": "2026-08-10T13:39:26.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086809650448793721.mp4",
    "posterUrl": "/posters/x/2086809650448793721.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1876×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "勝手ながら興味持ちましたので\n動画生成AIで「七瀬葵チャレンジ」させて頂きました\n\nローカルPC版のMiniMax-H3でT2Vで\n画風を「日本のイラストレーターの七瀬葵風で」として生成してみました\n\n全く出ませんでした😅\n\nでも今のアニメ調の流行りの美少女絵は全て\n七瀬先生の派生と考えると出てるとも？",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 14,
      "views": 455,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086648732993327204",
    "title": "四格漫画从户外活动切入乐队舞台",
    "titleEn": "A Four-Panel Comic Leads into a Full Band Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "前两帧以漫画格呈现女子与儿童围坐户外以及车内男女交谈的场景，末帧转为同一女子站在灯光舞台中央，身后是完整乐队。 成片规格为 38 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The first two frames use comic panels to show a woman seated outdoors with children and a man and woman inside a car, before cutting to the same woman on a lit stage with a full band. Output: 38s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aihonobono2023/status/2086648732993327204",
    "sourceLabel": "X 原帖 · @aihonobono2023",
    "author": "ai honobono | AI Art, Fiction, Music",
    "publishedAt": "2026-08-10T03:00:00.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086648732993327204.mp4",
    "posterUrl": "/posters/x/2086648732993327204.jpg",
    "duration": 38,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "天ロック直前に掲載した「それいけ ほのぼのさん！」という4コマ漫画を、CodexとMiniMax H3を使ってモーションコミック動画にしてみました（懐かしい？）。\n最後のコマは簡単なアニメーションになりセリフも入ります。生成時間はトータル15分ほどです。\n\n私はCodexに「この漫画を動画にしてください」と指示しただけです。本当は「フル動画になって出てこないかな？」なんて期待していたのですが、指示がよくなかったですね😅\n\nほのぼのさんの声がイメージと違ってちょっと高いのですが、実験的なものなので手を加えずに投稿します。\n\n【構成】\n・前半11コマ：原文と吹き出しを保持したパン・ズーム演出\n・最終コマ：公式h3-prompt-writing SkillによるI2VA\nライブ場面：Turbo 4step＋SageAttention、生成時間約6分06秒\n・1280×736・24fps・38.1秒\n・AAC 32kHzステレオ音声\n・前半BGMからH3生成の演奏・日本語台詞へクロスフェード\n\n#MiniMaxH3 #ComfyUI",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 13,
      "views": 1581,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086871126647673308",
    "title": "床边饮罐女子起身离开房间的双栏对照",
    "titleEn": "Side-by-Side Views of a Woman Drinking and Leaving the Bedroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "两栏均以同一名灰色背心女子与黄色罐装饮料为参考，女子先盘腿坐在床上持罐，随后从床边起身并走向房门。 成片规格为 5 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two side-by-side views use the same woman in a gray top and a yellow canned drink as references, showing her seated cross-legged on a bed with the can before standing and moving toward the door. Output: 5s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "ベッドに座ってストロング缶を飲んでるがピンポンが鳴って慌てて部屋を出ていく",
    "sourceUrl": "https://x.com/javawock7618/status/2086871126647673308",
    "sourceLabel": "X 原帖 · @javawock7618",
    "author": "ジャヴァノ🍀ボカロP",
    "publishedAt": "2026-08-10T17:43:43.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086871126647673308.mp4",
    "posterUrl": "/posters/x/2086871126647673308.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 ref2vaとfl2va+Ref LoRA比較\n\n参照画像は「立っている女性」と「ストロング缶」\nプロンプトは「ベッドに座ってストロング缶を飲んでるがピンポンが鳴って慌てて部屋を出ていく」\n\nLoRA版は飲んでないけど、まぁガチャれば飲んでくれるでしょう。いやーモデル1個でまわせるの助かるわい。",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 7,
      "views": 1145,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086720354526114078",
    "title": "长发女仆在夜间林道转身站定",
    "titleEn": "A Long-Haired Maid Turns on a Forest Road at Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一名深蓝长发女仆站在夜间森林公路中央，先背对镜头，再转为正面并屈臂移动，最后在道路中间张开双腿站定。 成片规格为 3 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark blue-haired maid stands in the middle of a forest road at night, first facing away, then turning forward with bent arms, and finally planting her feet apart on the roadway. Output: 3s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/higana_yogana/status/2086720354526114078",
    "sourceLabel": "X 原帖 · @higana_yogana",
    "author": "ヒガナヨガナ",
    "publishedAt": "2026-08-10T07:44:36.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086720354526114078.mp4",
    "posterUrl": "/posters/x/2086720354526114078.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「ComfyUI-ClipProj」を試す。\nMiniMax H3のTEをKrea2でも使っている「Qwen3-VL-4B」に変更するカスタムノードだ。\n\n🤖「デフォルトの32Bから、だいぶ知識が減りそうですが？」\n32Bを1.0とすると、0.7ぐらいになるらしい。\n\n生成結果に異常はない。 1/6\n\n#AI画像生成 #AI動画\n\nhttps://github.com/nicolab28/ComfyUI-ClipProj",
    "engagement": {
      "replies": 2,
      "reposts": 7,
      "likes": 78,
      "views": 4830,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086652506340557218",
    "title": "四栏参数对照中的韩流舞蹈群像",
    "titleEn": "Four-Panel Parameter Comparison of a K-Pop Dance Group",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "四个并排分区分别呈现身穿黑白服装的舞蹈群体在练习室内列队动作，每一区域上方或下方标有不同参数与耗时信息。 成片规格为 5 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Four arranged panels show black-and-white-clad dance groups moving in formation inside practice rooms, with different parameter and timing labels placed around each section. Output: 5s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jun1228909/status/2086652506340557218",
    "sourceLabel": "X 原帖 · @jun1228909",
    "author": "jun",
    "publishedAt": "2026-08-10T03:15:00.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086652506340557218.mp4",
    "posterUrl": "/posters/x/2086652506340557218.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "dance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ComfyUI MiniMax H3\nベンチマークの設計ってむずいね（；v；）\n\n全然納得感のある結果が得られないので一旦お蔵かなぁ😇\n動画生成自体の用途も色々だし、使ってみたほうが早くね？ってなってきた\n\nあと、それ以前に何故か sage と sol-attn を併用してもそれぞれを単体で使う時と同程度の生成時間にしかならない\n絶対何かミスってる、、詳しい方に教えて頂きたい🫠",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 12,
      "views": 2859,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086873055641629136",
    "title": "纸上飞机飞出画页穿越城市",
    "titleEn": "A Paper Airplane Leaves the Page and Flies Through the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "铅笔先在沾有水滴的纸面画出纸飞机轮廓，白色纸飞机随后出现在湿润巷道上空，最后从高楼玻璃幕墙旁飞过。 成片规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pencil first sketches a paper airplane on a water-speckled page before the white plane appears above a wet alley and then flies past the glass facade of a high-rise. Output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FutureVibesAi/status/2086873055641629136",
    "sourceLabel": "X 原帖 · @FutureVibesAi",
    "author": "Future Vibes AI - Educator",
    "publishedAt": "2026-08-10T17:51:23.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086873055641629136.mp4",
    "posterUrl": "/posters/x/2086873055641629136.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What if your drawings refused to stay inside the notebook..\nExperimenting with Minimax H3 in @Hailuo_AI \nYou can also try this prompt out \n#MinimaxH3",
    "engagement": {
      "replies": 10,
      "reposts": 2,
      "likes": 31,
      "views": 1012,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086747822553190866",
    "title": "黄昏街头少女提袋手持饮料",
    "titleEn": "A Girl Carries a Bag and Drink on a Dusk Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短黑发动漫少女站在黄昏住宅街道上，一手拿着浅绿色瓶装饮料，另一手提着透明购物袋，镜头逐渐靠近她的上半身。 成片规格为 5 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short black-haired anime girl stands on a residential street at dusk, holding a pale green bottled drink in one hand and a clear shopping bag in the other as the framing moves closer. Output: 5s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/guriham_lab/status/2086747822553190866",
    "sourceLabel": "X 原帖 · @guriham_lab",
    "author": "ぐり🐹AIイラスト",
    "publishedAt": "2026-08-10T09:33:45.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086747822553190866.mp4",
    "posterUrl": "/posters/x/2086747822553190866.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "ugc"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3メモ。RTX 4070／VRAM 12GB、公式Turbo v4 step600 EMAをstrength 1.0、8 steps、H3用Dual-Clock Sampler、scheduler=simpleで使用。352×608、124フレーム、24fps、中央crop、キャッシュなし。生成目安は約5分、VRAM約11.3GiB。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 25,
      "views": 824,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086826832649941435",
    "title": "屋顶男孩指向繁星夜空",
    "titleEn": "A Rooftop Boy Points Toward a Star-Filled Night Sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿黄色连帽衫的卡通男孩站在村镇屋顶上，抬起手指向布满明亮星点的深蓝夜空，远处是山林与亮灯房屋。 成片规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cartoon boy in a yellow hoodie stands on a village rooftop and points toward a deep blue sky filled with bright stars, with wooded hills and lit houses in the distance. Output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/iamrealsnow/status/2086826832649941435",
    "sourceLabel": "X 原帖 · @iamrealsnow",
    "author": "Snow",
    "publishedAt": "2026-08-10T14:47:42.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086826832649941435.mp4",
    "posterUrl": "/posters/x/2086826832649941435.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created with Pollo AI using Minimax H3.\n\nI turned that idea into a cinematic animated short , a little boy, a falling star, and a magical world hidden beyond the clouds. \n\nI pushed H3 for character consistency, cinematic camera movement, magical world-building, emotional storytelling and a big final reveal.",
    "engagement": {
      "replies": 28,
      "reposts": 1,
      "likes": 40,
      "views": 824,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086956124134310292",
    "title": "客厅众人拆礼物时女子站在门口",
    "titleEn": "A Woman Watches as People Unwrap Gifts in a Living Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "客厅中的几个人围着桌面拆开彩色礼盒，一名穿白色上衣的女子始终站在通向厨房的门口，三帧间礼盒和纸张的位置不断变化。 成片规格为 10 秒 · 768×448。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Several people gather around a table opening colorful gift boxes while a woman in a white top remains in the doorway to the kitchen and the wrapping shifts across the frames. Output: 10s · 768×448. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/0x0SojalSec/status/2086956124134310292",
    "sourceLabel": "X 原帖 · @0x0SojalSec",
    "author": "Md Ismail Šojal 🕷️",
    "publishedAt": "2026-08-10T23:21:28.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086956124134310292.mp4",
    "posterUrl": "/posters/x/2086956124134310292.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "768×448",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance-level video generation on your PC locally.\n\nYou can now run MiniMax H3 video generation on a \n5-6GB GPU. \n\nOn hardware most people already own.\n- 1-click install via Pinokio (NVIDIA + AMD)\n- FL2VA (First/Last frame Video + Audio)\n- Ref2VA (Reference-based generation)\n- Pruned 20B recommended as the starting point.\n\nWanGP dropped one of the most optimized low-VRAM video models we’ve seen.",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 7,
      "views": 1273,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "three-frame-media-review"
  },
  {
    "id": "x-2086753713029583219",
    "title": "粉色卧室窗边的耳机少女",
    "titleEn": "A Headphone-Wearing Woman Sits by a Pink Bedroom Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名短黑发女子戴着浅蓝耳机坐在粉色卧室的窗台上，房内摆有床、书桌、动漫海报与散落衣物，窗外可见城市高楼。 成片规格为 21 秒 · 1920×1104。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short black-haired woman wearing pale blue headphones sits on the windowsill of a pink bedroom containing a bed, desk, anime posters, and scattered clothes, with city towers outside. Output: 21s · 1920×1104. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotogenicWeekE/status/2086753713029583219",
    "sourceLabel": "X 原帖 · @PhotogenicWeekE",
    "author": "Photogenic Weekend",
    "publishedAt": "2026-08-10T09:57:09.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086753713029583219.mp4",
    "posterUrl": "/posters/x/2086753713029583219.jpg",
    "duration": 21,
    "aspectRatio": "landscape",
    "resolution": "1920×1104",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "随分世界観違う版(笑) MiniMax H3 10s x2の連結！",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 84,
      "views": 11022,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086754802021892454",
    "title": "霓虹车窗旁的黑衣长发女子",
    "titleEn": "A Long-Haired Woman in Black Beside a Neon Car Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名长发女子穿黑色亮面两件式服装坐在汽车后排，一只手扶在头后，车窗外的霓虹灯与彩色反光落在车内。 成片规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-haired woman in a glossy black two-piece outfit sits in the back of a car with one hand behind her head as neon signs and colored reflections cross the interior. Output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/reiQjo/status/2086754802021892454",
    "sourceLabel": "X 原帖 · @reiQjo",
    "author": "九条 玲",
    "publishedAt": "2026-08-10T10:01:29.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086754802021892454.mp4",
    "posterUrl": "/posters/x/2086754802021892454.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "濡らして決める前の十五分。この顔、いま一番好きかも\n\n※AI生成動画（MiniMax H3）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 40,
      "views": 753,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086920526052655286",
    "title": "粉色泳装少女坐在落日沙滩",
    "titleEn": "A Girl in Pink Swimwear Sits on a Sunset Beach",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短黑发动漫少女穿粉白泳装背对镜头坐在沙滩上，蓝色颈带与侧腰蝴蝶结清晰可见，落日光线从海面方向照来。 成片规格为 15 秒 · 1664×2496。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short black-haired anime girl in pink-and-white swimwear sits on the beach with her back to the camera, showing a blue neck ribbon and side bow under light from the setting sun. Output: 15s · 1664×2496. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2086920526052655286",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-10T21:00:01.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086920526052655286.mp4",
    "posterUrl": "/posters/x/2086920526052655286.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1664×2496",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「おはよう」 #AI眼鏡部 #AI田中さん\nMiniMax H3 ローカル",
    "engagement": {
      "replies": 3,
      "reposts": 8,
      "likes": 138,
      "views": 6062,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086650241370247205",
    "title": "烛光红沙发上的金发哥特女子",
    "titleEn": "A Blonde Gothic Woman on a Red Sofa by Candlelight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "双马尾金发女子穿黑色哥特裙装坐在红色皮革沙发上，身后两支蜡烛点亮深色背景，座垫前方放着红玫瑰与花瓣。 成片规格为 10 秒 · 928×1664。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde woman with twin tails wears a black Gothic dress on a red leather sofa, with two candles lighting the dark background and a red rose and petals on the cushion. Output: 10s · 928×1664. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/monaka_aiart/status/2086650241370247205",
    "sourceLabel": "X 原帖 · @monaka_aiart",
    "author": "もなか＠AIイラスト",
    "publishedAt": "2026-08-10T03:06:00.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086650241370247205.mp4",
    "posterUrl": "/posters/x/2086650241370247205.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "928×1664",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こんにちは☀\nイラスト動かしてみました！\nMiniMax H3で作ってみました～♬",
    "engagement": {
      "replies": 1,
      "reposts": 5,
      "likes": 97,
      "views": 2422,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086800107362509190",
    "title": "巨型粉色飞猪漂浮在船只之间",
    "titleEn": "A Giant Pink Flying Pig Floats Among Ships at Sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一只带翅膀与绿色叶片装饰的巨型粉色飞猪横卧在深色海面上，多艘小船环绕其身体航行，浪花沿边缘翻起。 成片规格为 15 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A giant pink flying pig with wings and green leaf-like details lies across a dark sea while several small ships travel around it and white surf gathers along its edges. Output: 15s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/s3nhxx/status/2086800107362509190",
    "sourceLabel": "X 原帖 · @s3nhxx",
    "author": "s3nh",
    "publishedAt": "2026-08-10T13:01:31.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086800107362509190.mp4",
    "posterUrl": "/posters/x/2086800107362509190.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Using Minimax H3 API is really intuitive and im still amazed by what H3 can do. You should definitely try it and give a chance to unleash your creativity ^^ 768p 15 sec video, not will give it a change in my local 4090 setup ^^\n@MiniMax_AI  @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 11,
      "views": 2165,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086805066686603524",
    "title": "酒红长发主唱JIWON立于拼贴海报前",
    "titleEn": "Burgundy-Haired Lead Singer Jiwon Before a Poster Collage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "酒红长发的主唱 JIWON 身穿暗红皮质无袖裙站在黑白红拼贴海报前，背景同时放大呈现她的肖像与排版文字。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Lead singer Jiwon, with long burgundy hair and a dark red sleeveless leather dress, stands before a black-white-red collage that enlarges her portrait amid graphic typography. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/joshesye/status/2086805066686603524",
    "sourceLabel": "X 原帖 · @joshesye",
    "author": "行者AI视频",
    "publishedAt": "2026-08-10T13:21:13.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086805066686603524.mp4",
    "posterUrl": "/posters/x/2086805066686603524.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "我宣布 我的女团正式出道\n\n酒红长发的成年韩国女团主唱 JIWON ，穿暗红皮质无袖短裙，眼神温柔但带伤感。\n\n#MiniMax  H3",
    "engagement": {
      "replies": 72,
      "reposts": 2,
      "likes": 58,
      "views": 10238,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086804027556372803",
    "title": "猫咖女子举镜自拍与橘猫同框",
    "titleEn": "A Woman Takes a Selfie with an Orange Cat in a Cafe",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名短发女子在木质猫咖内举起相机自拍，旁边的桌面上站着一只橘白猫，临街落地窗与铜色吊灯构成背景。 成片规格为 30 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired woman holds the camera for a selfie inside a wood-paneled cat cafe while an orange-and-white cat stands on the table beside her, with street windows and copper lamps behind them. Output: 30s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hAru_mAki_ch/status/2086804027556372803",
    "sourceLabel": "X 原帖 · @hAru_mAki_ch",
    "author": "Maki@Sunwood AI Labs.",
    "publishedAt": "2026-08-10T13:17:05.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086804027556372803.mp4",
    "posterUrl": "/posters/x/2086804027556372803.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 native AV 4step × 猫カフェ連鎖、30秒生成してみた！！！\n\nベーカリーで使った構成を猫カフェのMotion Context連鎖に移植。\n動画自体は出たけど、見返すと課題がはっきり出た、、、\n\n🔊 音声がザーッとノイズまみれ\n🖼️ 後半にいくほど画質がガビガビ\n🐈 連鎖はできても、品質はまだ実用段階じゃない\n\nこの状態を起点に、音声経路と区間ごとの画質劣化を切り分けるぞ🔥",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 8,
      "views": 3173,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086785719604052241",
    "title": "花田夕阳中的蓝发白裙少女",
    "titleEn": "A Blue-Haired Girl in White Stands Among Sunset Flowers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长蓝发动漫少女穿白色无袖连衣裙站在盛开的白色花田中，头发与裙摆向侧面展开，夕阳位于她身后的地平线附近。 成片规格为 18 秒 · 1088×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long blue-haired anime girl in a sleeveless white dress stands among blooming white flowers as her hair and skirt extend sideways and the sun sits near the horizon behind her. Output: 18s · 1088×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Haruka1425054/status/2086785719604052241",
    "sourceLabel": "X 原帖 · @Haruka1425054",
    "author": "はるか🎀 Haruka",
    "publishedAt": "2026-08-10T12:04:20.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086785719604052241.mp4",
    "posterUrl": "/posters/x/2086785719604052241.jpg",
    "duration": 18,
    "aspectRatio": "landscape",
    "resolution": "1088×608",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "KTさんのプロンプトをお借りして、\nMinimax H3で作ってみました。\nプロンプトって大事なんですね💦\n\nKTさんありがとうございました。\n\nSFW anime-style illustration No nudity, no suggestive intent.",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 23,
      "views": 1026,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086704420587458621",
    "title": "礼堂水晶旁的蓝裙褐发少女",
    "titleEn": "A Brown-Haired Girl in Blue Beside Crystals in a Hall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "褐色长发动漫少女穿蓝白礼裙站在高窗与立柱环绕的礼堂中，身旁漂浮着菱形水晶与细小光点。 成片规格为 15 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long brown-haired anime girl in a blue-and-white formal dress stands in a hall of tall windows and columns with diamond-shaped crystals and small lights floating nearby. Output: 15s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/makotwi0104/status/2086704420587458621",
    "sourceLabel": "X 原帖 · @makotwi0104",
    "author": "マコトーん",
    "publishedAt": "2026-08-10T06:41:17.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086704420587458621.mp4",
    "posterUrl": "/posters/x/2086704420587458621.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n久しぶりに動画生成触ってみたけど、雑プロンプトだとなんかしっくり来ない。\nAI画像生成と違ってガチャや試行錯誤するにはコスト高いな。\nローカルで数を回せないとキツいんっすかね。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 16,
      "views": 618,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2086841608746553784",
    "title": "夜色城市航拍与中央白色标题",
    "titleEn": "A Nighttime City Aerial beneath a Large White Title",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "夜间城市航拍画面由纵横街灯与高楼组成，中央覆盖一行大号白色制作标题，右下角另有两个平台标记。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An aerial night cityscape of illuminated streets and towers is covered by a large white production title across the center, with two platform marks in the lower-right corner. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jerrod_lew/status/2086841608746553784",
    "sourceLabel": "X 原帖 · @jerrod_lew",
    "author": "Jerrod Lew",
    "publishedAt": "2026-08-10T15:46:25.000Z",
    "addedAt": "2026-08-11T12:15:52.991Z",
    "mediaUrl": "/media/x-2086841608746553784.mp4",
    "posterUrl": "/posters/x/2086841608746553784.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An awesome way to prompt Hailuo MiniMax H3!\n\nCreate title animations!",
    "engagement": {
      "replies": 12,
      "reposts": 8,
      "likes": 102,
      "views": 10643,
      "snapshotAt": "2026-08-11T12:15:52.991Z"
    },
    "approvedAt": "2026-08-11T12:15:52.991Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087696467628810605",
    "title": "眼镜少女提花袋站在玄关开门",
    "titleEn": "A Bespectacled Girl with a Floral Bag Opens the Front Door",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白发眼镜少女穿蓝白连衣裙站在木质玄关，一手握住打开的房门把手，另一肩挎着装有鲜花和长面包的花纹布袋。 成片规格为 10 秒 · 896×1184。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired girl with glasses stands in a wooden entryway wearing a blue-and-white dress, holding the open door with one hand and carrying a floral bag filled with flowers and a baguette. Output: 10s · 896×1184. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nayu_pkpk/status/2087696467628810605",
    "sourceLabel": "X 原帖 · @nayu_pkpk",
    "author": "nayucchi（グラボあちち勢）",
    "publishedAt": "2026-08-13T00:23:20.000Z",
    "addedAt": "2026-08-13T00:38:33.427Z",
    "mediaUrl": "/media/x-2087696467628810605.mp4",
    "posterUrl": "/posters/x/2087696467628810605.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "896×1184",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\nおまけ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 25,
      "snapshotAt": "2026-08-13T00:38:33.427Z"
    },
    "approvedAt": "2026-08-13T00:38:33.427Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087696356618248367",
    "title": "女子在教室黑板上书写你好",
    "titleEn": "A Woman Writes Hello in Chinese on a Classroom Blackboard",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿灰色无袖裙的女子背对镜头站在教室讲台前，用粉笔在绿色黑板上写下中文“你好”，桌面摆着课本与打开的书。 成片规格为 10 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a sleeveless gray dress stands with her back to the camera at the front of a classroom, writing the Chinese greeting for hello on a green blackboard beside textbooks and an open book. Output: 10s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiaicreate/status/2087696356618248367",
    "sourceLabel": "X 原帖 · @aiaicreate",
    "author": "R@aiaicreate",
    "publishedAt": "2026-08-13T00:22:53.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087696356618248367.mp4",
    "posterUrl": "/posters/x/2087696356618248367.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fashion",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "動画生成AI「Minimax H3」の生成品質を巡るステップ数設定が話題。\n標準の20より30ステップ以上を推奨する声が多数。時間は増えるが手や口元のブレ、音声が改善されると好評。高速化技術との併用設定も共有されている。\n#MinimaxH3 #ComfyUI\n\nURLはリプ⬇️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 209,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087695377835430249",
    "title": "黄色背景前俯身的灰发猫耳少女",
    "titleEn": "A Gray-Haired Catgirl Leans Forward Against a Yellow Background",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰白短发猫耳少女戴细框眼镜，在带白色边框的黄色背景前俯身蹲下，宽大猫耳与多条蓬松尾巴围绕身体展开。 成片规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short gray-white-haired catgirl in thin glasses crouches forward against a yellow background with a white border, surrounded by oversized ears and several fluffy tails. Output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Genzoh1/status/2087695377835430249",
    "sourceLabel": "X 原帖 · @Genzoh1",
    "author": "Genzoh",
    "publishedAt": "2026-08-13T00:19:00.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087695377835430249.mp4",
    "posterUrl": "/posters/x/2087695377835430249.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "自作の生成AIアプリにMinimax H3を組み込んでみました！ 🥰\n\nこだわりポイント 👇 \n✅ アプリ内に閉じた簡単画像参照 \n✅ 強力なプロンプトの補助機能 \n✅ 4Step LoRAによる超高速生成 \n✅ SeedVR2によるアップスケール \nなどなど 😋\n\nAnimaなどで画像生成→Minimax H3 での動画化の導線がきれいに設計できたと思いますッ 😆\n\nちゃんとComfyUI側にキューが渡せたので、ひとまず満足 ☺️\n\nComfyUIの複雑なワークフローがどうしても苦手で 💦 API経由にして、シンプルな最小構成モジュールを組み合わせる形が、自分的には好みです 🥰 🥰\n\nアプリ自体はどこかで公開したいな～と、思ってますが、\n変な機能をもりもりしすぎて、どんどん公開から遠ざかってる気がする💦\n気にせずgithubにえいやしちゃってもいいんだけどねぇ😝",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "views": 225,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087693201884758134",
    "title": "雨窗旁木桌上的热饮杯",
    "titleEn": "A Steaming Drink on a Wooden Table Beside a Rainy Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一只盛有热饮的白色杯子放在木桌边缘，蒸汽从杯口升起，旁边窗玻璃布满雨滴，窗外砖楼与车辆处于虚焦中。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white cup of hot drink sits near the edge of a wooden table with steam rising from it, beside a rain-covered window that blurs brick buildings and vehicles outside. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AllaAisling/status/2087693201884758134",
    "sourceLabel": "X 原帖 · @AllaAisling",
    "author": "Alexandra Aisling",
    "publishedAt": "2026-08-13T00:10:21.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087693201884758134.mp4",
    "posterUrl": "/posters/x/2087693201884758134.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Everything glowing here is hand-drawn: crayon, chalk, dry brush. Rough edges, wobbly lines, imperfect fills.\n\nBecause a perfect little cloud isn't cute. A slightly wonky one is. 🌥️\n\nPrompt ⤵️ inspired by Amira\n\nMade with MiniMax H3",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 3,
      "views": 237,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087690597825978563",
    "title": "三栏对照中的红色手绘怪物",
    "titleEn": "Three Red Hand-Drawn Monsters Compared Side by Side",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三栏分别呈现红色喷溅轮廓的尖笑角色、边缘蓬松的椭圆角色，以及以粗糙圆形笔触构成的笑脸角色，中央以亮线分隔。 成片规格为 37 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three panels present a grinning red splatter figure, a fuzzy-edged oval figure, and a smiling face formed from a rough circular brush mark, separated by bright dividing lines. Output: 37s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/OiiOii_AI/status/2087690597825978563",
    "sourceLabel": "X 原帖 · @OiiOii_AI",
    "author": "OiiOii AI",
    "publishedAt": "2026-08-13T00:00:00.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087690597825978563.mp4",
    "posterUrl": "/posters/x/2087690597825978563.jpg",
    "duration": 37,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Same prompt. Different models. Different results.\n\nWe put the exact same prompt through Seedance 2.0, Seedance 2.5, and MiniMax H3.\n\nWhich one is your pick? 👀",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 92,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087688254355329334",
    "title": "白色背景前的双马尾毛衣少女",
    "titleEn": "A Twin-Tailed Girl in a White Sweater Against White",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "棕色短双马尾动漫少女穿宽领白色毛衣面对镜头，背景保持纯白，眼中带有金色星形高光。 成片规格为 25 秒 · 512×896。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A brown-haired anime girl with short twin tails faces the camera in a wide-neck white sweater against a plain white background, with golden star-shaped highlights in her eyes. Output: 25s · 512×896. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aki_ai_builder/status/2087688254355329334",
    "sourceLabel": "X 原帖 · @aki_ai_builder",
    "author": "Aki｜AI サービス開発中",
    "publishedAt": "2026-08-12T23:50:41.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087688254355329334.mp4",
    "posterUrl": "/posters/x/2087688254355329334.jpg",
    "duration": 25,
    "aspectRatio": "portrait",
    "resolution": "512×896",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🔊音が出ます！\n\n1.日本語で簡単なシチュエーションを書く\n2.H3の英語プロンプトは、ローカルLLM(35B)に書かせる。　17秒\n3.MiniMax H3 が映像と音声を同時生成　4分31秒\n\nこれを5カット、計30分。全部ローカル(RTX5070)\n最後は1カット目の絵に着地させて完全ループできるので、ショート動画作るときは便利かも\n\n#MiniMaxH3 #AI動画",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 42,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087688235682029978",
    "title": "镜头下行穿过昏暗工业楼梯间",
    "titleEn": "A Descending View Through a Dark Industrial Stairwell",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头沿昏暗工业楼梯间持续向前下行，经过裸露线缆与灯泡后停在铺有轨道的隧道前。 成片规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera travels forward and downward through a dark industrial stairwell with exposed cables and bulbs before reaching a rail-lined tunnel. Output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2087688235682029978",
    "sourceLabel": "X 原帖 · @UnrealRafael",
    "author": "Рафаэль Анриал ∞",
    "publishedAt": "2026-08-12T23:50:37.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087688235682029978.mp4",
    "posterUrl": "/posters/x/2087688235682029978.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Absolutely insane🤯realistic found-footage style video, pure first-person POV from a handheld camera, continuous single take. Dark, gloomy and oppressive atmosphere\nA person carefully descends a vertical service shaft of the Moscow metro and exits into a dark tunnel",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 24,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087687874468626792",
    "title": "亨利·卡维尔胡须移除镜头的遮罩修复对比",
    "titleEn": "Masked Inpainting Comparison of Henry Cavill's Mustache Removal",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一段《正义联盟》胡须数字移除镜头并列比较原版与遮罩修复结果。 成片规格为 5 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same Justice League mustache-removal shot is used to compare the original with a masked-inpainting result. Output: 5s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/8bit_e/status/2087687874468626792",
    "sourceLabel": "X 原帖 · @8bit_e",
    "author": "enigmatic_e",
    "publishedAt": "2026-08-12T23:49:11.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087687874468626792.mp4",
    "posterUrl": "/posters/x/2087687874468626792.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Can you figure out which one is the original and which one was done with MiniMax H3 using masked inpainting?\n\nTesting it on the infamous Justice League scene where Henry Cavill’s mustache was digitally removed. 👀",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 14,
      "views": 680,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "sourceCaption: explicit film scene, actor, mustache-removal subject, masked inpainting, and original-result comparison"
  },
  {
    "id": "x-2087687579252543566",
    "title": "第一视角深入昏暗砖砌设备隧道",
    "titleEn": "A First-Person Journey Through a Dark Brick Utility Tunnel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "第一视角沿昏暗的砖砌拱形通道前进，经过设备柜后靠近管线、阀门与大型圆柱设备。 成片规格为 20 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A first-person view advances through a dark brick-lined passage, passes a utility cabinet, and approaches pipes, valves, and cylindrical equipment. Output: 20s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2087687579252543566",
    "sourceLabel": "X 原帖 · @UnrealRafael",
    "author": "Рафаэль Анриал ∞",
    "publishedAt": "2026-08-12T23:48:00.000Z",
    "addedAt": "2026-08-13T00:38:33.428Z",
    "mediaUrl": "/media/x-2087687579252543566.mp4",
    "posterUrl": "/posters/x/2087687579252543566.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Absolutely insane🤯Text in the video rtx 5090\n20-second realistic found-footage style video, pure first-person POV from a handheld camera, continuous single take. Dark, gloomy and heavily abandoned atmosphere. Direct continuation of the previous scene.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 33,
      "snapshotAt": "2026-08-13T00:38:33.428Z"
    },
    "approvedAt": "2026-08-13T00:38:33.428Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087684538390532522",
    "title": "从闹钟到通勤再到倒床的一天",
    "titleEn": "A Day from Alarm Clock to Commute and Bed Collapse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一女子从闹钟响起后依次起床、洗漱、做早餐、赶地铁、进入办公室工作，最后回家倒在床上。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same woman wakes to an alarm, washes, makes breakfast, rushes onto the metro, works at an office, and finally returns home and falls onto her bed. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EndFolding79421/status/2087684538390532522",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-08-12T23:35:55.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087684538390532522.mp4",
    "posterUrl": "/posters/x/2087684538390532522.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3...2k...\n\nPrompt used:\n\nintegrated_multimodal_description: [Shot 1] Ultra-realistic live-action cinematic morning routine with energetic commercial-style editing. Exactly ONE young adult woman appears throughout the video; maintain the same face, hairstyle, body proportions, and identity across every shot. The entire 15-second sequence is edited tightly to a fast 145 BPM electro-pop rhythm, with hard cuts landing cleanly on musical beats. Early morning inside a tiny modern apartment. Extreme close-up of a smartphone alarm vibrating rapidly on rumpled bedsheets beside the sleeping woman. Cool dim bedroom ambience with the phone screen illuminating the sheets.  [Shot 2] At 00:00.850, hard cut to a medium-wide bedside shot. The SAME woman, wearing a loose sleep tee and lounge shorts, suddenly jolts upright in panic, throws the blanket aside, swings both legs out, and plants her bare feet on the floor in one hurried movement.  [Shot 3] At 00:01.750, hard cut to a tight bathroom mirror shot. She rapidly splashes cold water onto her face over the sink. Water droplets scatter and catch the bright bathroom light.  [Shot 4] At 00:02.600, hard cut to an extreme close-up profile. She vigorously brushes her teeth, eyes darting sideways with an anxious late-for-work expression.  [Shot 5] At 00:03.450, hard cut to POV from INSIDE the refrigerator looking outward. Bright white refrigerator light illuminates her face as she abruptly opens the door, reaches directly toward camera, grabs breakfast ingredients, and pulls them out.  [Shot 6] At 00:04.300, hard cut to a tight overhead kitchen shot. Eggs crack rapidly into a hot frying pan while bread drops into a toaster beside it. Immediate realistic sizzling and kitchen movement.  [Shot 7] At 00:05.250, hard cut to a medium kitchen-table shot. She takes one hurried bite of breakfast, suddenly looks toward a clock, realizes how late she is, widens her eyes, and immediately jumps to her feet.  [Shot 8] At 00:06.200, hard cut to a directly overhead close-up of her lower legs and hands. She rapidly pulls on one sock and then the other with two sharp movements synchronized to the music.  [Shot 9] At 00:07.050, hard cut to floor-level close-up. Her first lace-up shoe comes down onto her foot with urgency as she shoves her heel inside.  [Shot 10] At 00:07.900, hard cut even tighter to both shoes. Her hands pull the laces tight in one fast synchronized motion and finish the knot.  [Shot 11] At 00:08.750, hard cut to the apartment doorway. The SAME woman is now fully dressed for work in a tailored jacket, fitted top, trousers, and lace-up shoes. She hurriedly pulls the jacket into place while moving through the doorway, grabs her bag, and rushes into a rain-damp apartment hallway. The wardrobe transition occurs ONLY through this hard cut; do not morph clothing on camera.  [Shot 12] At 00:09.750, hard cut to a moving metro train. Medium close-up. She stands gripping a metal pole with one hand while the carriage rocks naturally. She looks anxiously toward the doors and then checks the time, visibly tense and impatient.  [Shot 13] At 00:10.850, hard cut to the office entrance. Tracking medium shot as she rushes through the doorway and quickly taps her employee badge against the access scanner. The scanner flashes approval as she continues moving without stopping.  [Shot 14] At 00:11.950, hard cut to her office desk. Tight energetic coverage of her hands typing rapidly on a keyboard, then a quick tilt to her exhausted face as multiple work notifications visibly accumulate on the computer screen behind or beside her. Her fingers continue typing frantically while she tries to keep up.  [Shot 15] At 00:13.250, hard rhythmic cut back to the same tiny apartment later that day. Soft bedroom window light. Still wearing her work clothes, she enters the bedroom completely exhausted and drops face-first onto the bed in one continuous physical action. Her body lands heavily, bounces slightly on the mattress, then becomes completely still. Hold the final exhausted pose through 00:15.000 as the frantic rhythm resolves.  overall_soundscape: Sound ON. Extremely tight rhythmic sound design synchronized with the 145 BPM editing: vibrating phone alarm, blanket rustle, bare feet hitting floor, running sink water and face splashes, rapid toothbrush sounds, refrigerator door opening, food containers moving, egg crack, frying-pan sizzle, toaster click, hurried bite, chair movement, fabric snaps from socks, shoes hitting floor, laces pulling tight, apartment door movement, faint rain and hallway ambience, metro rail rumble and carriage movement, badge scanner confirmation beep, rapid keyboard typing, layered electronic notification sounds, apartment door movement, footsteps, mattress impact, fabric rustle, and one tired human exhale at the end. No dialogue.  non_diegetic_music: Fast energetic electro-pop instrumental at approximately 145 BPM, driving the entire 15-second montage. Strong crisp percussion and electronic pulse. Major hard cuts and physical actions synchronize tightly with musical beats. Music maintains frantic momentum through the morning, commute, and office shots, then drops sharply in intensity during the final bed collapse, leaving the exhausted final exhale clearly audible.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 78,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit recurring subject, fifteen-shot routine, actions, locations, and ending"
  },
  {
    "id": "x-2087684053126517238",
    "title": "白衣动漫女子站在室内走廊",
    "titleEn": "An Anime Woman in White Stands in a Hallway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名黑发动漫女子身穿白衣站在室内走廊入口，原帖将这段内容用于观察口型同步。 成片规格为 5 秒 · 576×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired anime woman in white stands at an indoor hallway entrance; the source uses the sequence to examine lip synchronization. Output: 5s · 576×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_masataka/status/2087684053126517238",
    "sourceLabel": "X 原帖 · @ai_masataka",
    "author": "マサタカ | 生成AI × ゲーム開発",
    "publishedAt": "2026-08-12T23:34:00.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087684053126517238.mp4",
    "posterUrl": "/posters/x/2087684053126517238.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でリップシンクできるの！？",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "three-frame media review plus sourceCaption explicitly identifying the lip-sync purpose"
  },
  {
    "id": "x-2087682088778834362",
    "title": "田野里逐渐出现的一排排立鱼",
    "titleEn": "Rows of Upright Fish Appear Across a Field",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "空旷的耕地沟垄中逐渐出现大量直立的鱼，田野光线同时从蓝色日光转向暖色夕照。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Many upright fish progressively appear across cultivated field rows while the light shifts from blue daylight toward warmer sunset tones. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TlanoAI/status/2087682088778834362",
    "sourceLabel": "X 原帖 · @TlanoAI",
    "author": "とらの",
    "publishedAt": "2026-08-12T23:26:11.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087682088778834362.mp4",
    "posterUrl": "/posters/x/2087682088778834362.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 Local\n畑から鯖が生えてくるタイムラプス映像",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 8,
      "views": 377,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087681623718518870",
    "title": "AOI WATER 矿泉水的夏日扁平动效广告",
    "titleEn": "AOI WATER Summer Commercial in Flat Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "15 秒广告以 AOI WATER 瓶装矿泉水为持续主体，用黄、蓝、青色扁平图形串联水瓶、倒水、人物饮用与品牌收尾画面。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The 15-second commercial keeps an AOI WATER mineral-water bottle as its central subject and uses yellow, blue, and aqua flat graphics across bottle, pouring, drinking, and brand-ending scenes. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tanabe_fragm/status/2087681623718518870",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-12T23:24:21.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087681623718518870.mp4",
    "posterUrl": "/posters/x/2087681623718518870.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で清涼飲料水のCM動画を生成。\n映像スタイルはモーショングラフィックス。\n\nもちろん、プロンプトはClaude Codeに全部書いてもらっています。\nプロンプトを書いてもらう方法は引用元で紹介しています😃\n\n今回使用したプロンプト↓\n-----------------\nintegrated_multimodal_description: Create a complete 15-second Japanese flat-illustration\nmotion-graphics commercial for a bottled mineral water in a clean 16:9 composition. The bottle\nis the hero of every shot: it appears within the first second, stays large and centred, and its\npale blue label reading the exact Latin characters \"AOI WATER\" is sharp and readable in every\nframe it occupies. Use friendly 2D vector illustration, thick navy outlines, flat solid colour\nfills, and springy cutout animation with snappy overshoot easing. Compact palette of warm white,\ndeep navy, bright aqua, deep blue and saturated yellow, where yellow carries the summer heat and\naqua carries the cold refreshment. A fictional young Japanese actress supports the product,\ndrawn as a flat vector icon with a round friendly face, simple dot eyes, a curved smile, a\nglossy black chin-length bob, a white short-sleeved shirt over a navy tank top; she keeps the\nsame design in every shot. No photorealism, no 3D rendering, no gradients, no drop shadows, no\nbrowser interface, no screen-recording artifacts, no player controls.\n[Shot 1] Begin on a saturated yellow field. A large white sun pulses twice at the top of the\nframe and wavy heat-haze lines rise from the bottom edge. At 00:00.800 a tall white bottle of\nAOI WATER shoots up from the bottom edge into the centre of the frame, overshooting and\nsettling at eighty percent of the frame height, and a single white flash frame fires on the\nimpact. Three aqua ripple rings expand from the bottle and the small navy headline assembles\nword by word at the top: \"この夏、あつい。\"\n[Shot 2] At 00:02.400, a wave of bright aqua water sweeps across the frame from the left and\nwashes the yellow away. The bottle stays centred and rotates a quarter turn to show its side.\nIts pale blue label assembles letter by letter into \"AOI WATER\" with a thin white underline\ndrawing beneath, six round condensation drops pop onto the glass one after another, and two\nflat ice cubes tumble down past it.\n[Shot 3] At 00:05.000, a deep blue field wipes in from the right. The bottle tilts on the left\nand a clear aqua stream pours in a smooth arc into a tall glass on the right; three flat ice\ncubes drop into the glass one at a time and bounce, throwing a radial burst of white droplets\non the third. Two white snowflake marks pop in beside the glass and the kinetic word \"キンッ。\"\nstamps in with a small shake.\n[Shot 4] At 00:07.600, the deep blue flips to aqua in a hard shape wipe. The actress stands\ncentre frame holding the bottle high, tips it and drinks; a column of aqua fills her outlined\nbody from top to bottom in three quick pulses, one per gulp, and white radial speed-lines snap\noutward on the third. The kinetic word \"ごくっ\" pops in beside her cheek and the bottle stays\nfully visible in her hand.\n[Shot 5] At 00:10.200, a saturated yellow field wipes in from the bottom. The actress jumps once\nwith the bottle raised above her head, her bob and shirt lifting, while flat aqua and white\ncircles, triangles and arcs burst outward from her in a ring and three ripple rings expand from\nher landing point. The headline assembles in two stages: \"夏に、\" then \"負けない。\" A thick\nyellow underline strikes beneath the second phrase.\n[Shot 6] At 00:12.600, cut to a calm warm-white end card with generous negative space. The\nbottle slides to the centre and scales up slightly, its \"AOI WATER\" label square to the viewer\nand perfectly readable, with a ring of six aqua droplets popping outward around it one by one.\nThe logotype \"AOI WATER\" reveals in deep navy beside the bottle with a small aqua droplet mark,\nand the smaller line \"アオイ・ウォーター\" fades in beneath it. A calm, bright young Japanese\nwoman (S1) says in an off-screen voiceover: <d>[Japanese] 夏に、負けない。</d> The end card\nholds sharp, centred and readable until exactly 15.000 seconds.\nThroughout: flat vector motion graphics only, one palette and one line weight, the same bottle\ndesign and the same character design in every shot, and something on screen is always moving.\nAll Japanese text in clean gothic type and the label and logotype in clean geometric capitals,\ncorrectly formed and static once placed, with no subtitles of the spoken line and no other\nlettering.\n\noverall_soundscape: A deep whoosh and a short impact thud land as the bottle shoots up into\nframe, followed by three soft ripple taps. An airy whoosh runs under each colour wipe and a\nliquid sweep carries the aqua water across the frame. Rounded pops mark the sun pulses and each\ncondensation drop, two ice cubes clink as they tumble, and a bottle cap cracks open before a\nclear pouring stream and three ice cubes landing in a glass. Three deep gulps and a quick\nrefreshed exhale carry the drink, a light whoosh and landing thud carry the jump, and a clean\nshimmering chime rings on the logotype.\n\nnon_diegetic_music: Generate an original 15-second bright Japanese commercial cue at 128 BPM\nusing marimba, ukulele upstrokes, hand claps, a warm round bass and a light drum kit, playing\nwithout a gap from the first frame to the last. Open with a single accent hit as the bottle\nlands at 00:00.800, add the full kit as the water sweeps in, thin to marimba and claps under the\npour, push to the loudest point with a rising fill under the jump, drop under the spoken line,\nand resolve on one clean sustained chord at 15.000 seconds. No singing and no lyrics.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 236,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit product, visual style, palette, shot sequence, and end card"
  },
  {
    "id": "x-2087680003219796290",
    "title": "暗色背景中裂开的红色豆粒",
    "titleEn": "A Red Bean Splits Open Against a Dark Background",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一颗悬浮的红色豆粒在暗色背景中裂开，缝隙里露出金色内部。 成片规格为 2 秒 · 768×432。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A floating red bean splits open against a dark background, revealing a golden interior through the seam. Output: 2s · 768×432. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/THEMACHINE_HF/status/2087680003219796290",
    "sourceLabel": "X 原帖 · @THEMACHINE_HF",
    "author": "THEMACHINE",
    "publishedAt": "2026-08-12T23:17:54.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087680003219796290.mp4",
    "posterUrl": "/posters/x/2087680003219796290.jpg",
    "duration": 2,
    "aspectRatio": "landscape",
    "resolution": "768×432",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "gm gm\n\non-premises Minimax H3 testing @MiniMax_AI \n\ni7-12700K + RTX3070 + 64G RAM\neverything goes fine",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 66,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "beginning, middle, and end frames from the hosted media"
  },
  {
    "id": "x-2087678550661025820",
    "title": "录音桌前相对而坐的两名角色",
    "titleEn": "Two Characters Face Each Other Across a Studio Desk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名动漫角色隔着摆有麦克风和调音设备的桌子相对而坐，窗外背景在过程中发生变化。 成片规格为 15 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two anime characters sit opposite each other across a desk with microphones and mixing equipment as the view beyond the window changes. Output: 15s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sawac_s/status/2087678550661025820",
    "sourceLabel": "X 原帖 · @sawac_s",
    "author": "SAWAC@未来開発",
    "publishedAt": "2026-08-12T23:12:08.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087678550661025820.mp4",
    "posterUrl": "/posters/x/2087678550661025820.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3の丸投げテスト。\nREFを使わなくても、同じ場面なら一貫性のある映像ができるらしい。ただし、これは背景がミスっている。調整したらできるかな？",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 49,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "beginning, middle, and end frames plus sourceCaption explicitly noting the background inconsistency"
  },
  {
    "id": "x-2087677373135376850",
    "title": "两人用水枪进行彩弹式对战",
    "titleEn": "Two People Stage a Paintball-Style Water-Gun Battle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两人使用水枪进行带有彩弹游戏风格的对战动作。 成片规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two people use water guns in an action sequence styled after a paintball battle. Output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MATCHA_ORANGE_/status/2087677373135376850",
    "sourceLabel": "X 原帖 · @MATCHA_ORANGE_",
    "author": "抹茶オレンジ🧋🍊",
    "publishedAt": "2026-08-12T23:07:27.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087677373135376850.mp4",
    "posterUrl": "/posters/x/2087677373135376850.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "capcutにminimax h3来てたので試してみました\n490クレジットで2k動画作れるなら安い方かなと思う\nウォーターサバゲーでスプラごっこする二人\n#capcutcpp\n@capcutapp_jp",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 108,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit two-person subject, water-gun action, and paint-battle framing"
  },
  {
    "id": "x-2087676200546390429",
    "title": "海狮遍布海滨街道与十字路口",
    "titleEn": "Sea Lions Fill a Waterfront Street and Crosswalk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多只海狮先躺在海滨人行道上，随后出现在十字路口与街道中，周围仍有行人与车辆。 成片规格为 14 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Sea lions lie across a waterfront sidewalk and later occupy a crosswalk and street alongside pedestrians and traffic. Output: 14s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2087676200546390429",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-12T23:02:48.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087676200546390429.mp4",
    "posterUrl": "/posters/x/2087676200546390429.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Compared with MiniMax H3. Same prompt.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 760,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087675628418187426",
    "title": "三人舞蹈的三分屏金属玻璃构图",
    "titleEn": "Three-Person Dance in a Glass-and-Metal Triptych",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三名舞者被平衡放入三分屏金属玻璃框架，并叠加电影光线、镜头光晕、胶片颗粒、调色与运动模糊。 成片规格为 5 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three dancers are balanced across a glass-and-metal triptych with cinematic lighting, lens flare, film grain, color grading, and motion blur. Output: 5s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wadda1018/status/2087675628418187426",
    "sourceLabel": "X 原帖 · @wadda1018",
    "author": "wadda (와따) | 💕",
    "publishedAt": "2026-08-12T23:00:31.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087675628418187426.mp4",
    "posterUrl": "/posters/x/2087675628418187426.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "retro",
      "cinematic-realistic"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI로 영상 리터치 완료\n\n조잡했던 프레임 정리하고 \n세 사람 한눈에 보이게 밸런스 잡았어요\n\n얼마 전 Pollo ai 로 작업한 영상을 \nMiniMax H3 툴로 편집해봤는데\n\n그냥 밋밋했던 영상 퀄리티가 \n완전 바뀌더라고요\n\n주요 작업 포인트는\n\n원본 비율 유지 (좌우 잘림 완벽 방지)\n\n미니멀 글래스 메탈 프레임으로 \n3등분 균형 배치\n\n시네마틱 라이팅 & 렌즈 플레어 효과\n\n필름 그레인 + 감성 색보정\n\n부드러운 모션 블러로 춤선 완성도 UP\n\n이렇게 손보니까 \n훨씬 고급스럽고 세련되어졌어요\n\n사용한 프롬프트는 타래에 남겨둘게요",
    "engagement": {
      "replies": 9,
      "reposts": 3,
      "likes": 11,
      "views": 393,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit three-person dance, triptych framing, material treatment, lighting, grain, color, and motion blur"
  },
  {
    "id": "x-2087672140846002435",
    "title": "雨夜巷战的双栏对照",
    "titleEn": "A Side-by-Side Rainy Alley Fight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "双栏画面并排呈现两组黑白衣男子在雨夜巷道交手，可见踢击、倒地与重新对峙。 成片规格为 20 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A split-screen presents two versions of men in black and white fighting in a rainy alley, with kicks, a fall, and a return to a face-off. Output: 20s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aisearchio/status/2087672140846002435",
    "sourceLabel": "X 原帖 · @aisearchio",
    "author": "⚡AI Search⚡",
    "publishedAt": "2026-08-12T22:46:40.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087672140846002435.mp4",
    "posterUrl": "/posters/x/2087672140846002435.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5 vs Minimax H3",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 28,
      "views": 1714,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "beginning, middle, and end frames plus sourceCaption explicitly identifying a two-model comparison"
  },
  {
    "id": "x-2087672068133503078",
    "title": "水枪版枪斗术与子弹时间动作",
    "titleEn": "Water-Gun Gun-Fu with Bullet-Time Action",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动作设计把枪斗术、子弹时间与水枪对战组合在同一场面中。 成片规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The action combines gun-fu, bullet-time movement, and a water-gun battle within one scene. Output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MATCHA_ORANGE_/status/2087672068133503078",
    "sourceLabel": "X 原帖 · @MATCHA_ORANGE_",
    "author": "抹茶オレンジ🧋🍊",
    "publishedAt": "2026-08-12T22:46:22.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087672068133503078.mp4",
    "posterUrl": "/posters/x/2087672068133503078.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます\ncapcut にminimax H3が来てたので、動画も参照できるようになってました\nためにしウォーガタアクションやってきました\n何ですかそれって？\nガン=カタ+バレットタイム+ウォーターサバゲーを混ぜてみただけです\n水鉄砲でのゼロサムアクションなので安全安全\n#capcutcpp\n@capcutapp_jp",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 100,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit gun-fu, bullet-time, water-gun subjects, and combined action premise"
  },
  {
    "id": "x-2087671712259465467",
    "title": "蓝发角色漫步黄昏住宅街",
    "titleEn": "A Blue-Haired Character Walks Through a Suburban Street at Dusk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴眼镜和耳机的蓝发动画角色沿黄昏住宅街行走，画面在近景与侧面远景之间切换。 成片规格为 82 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired animated character wearing glasses and headphones walks along a suburban street at dusk as the framing alternates between close and wide side views. Output: 82s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIWarper/status/2087671712259465467",
    "sourceLabel": "X 原帖 · @AIWarper",
    "author": "A.I.Warper",
    "publishedAt": "2026-08-12T22:44:57.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087671712259465467.mp4",
    "posterUrl": "/posters/x/2087671712259465467.jpg",
    "duration": 82,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Not the most interesting output but this is really interesting imo.\n\nBleeding edge workflow the OS community came up with for Minimax H3 for clip continuation. You can just keep chaining and extending the story.\n\nThis was the default workflow I didn't change anything",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 13,
      "views": 974,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087671344788086904",
    "title": "《不存在的记忆》",
    "titleEn": "Nonexistent Memories",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《不存在的记忆》由 Midjourney 图像与 Suno 乐曲组合制作，并通过公开的音乐影像制作功能完成。 成片规格为 139 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Nonexistent Memories combines Midjourney imagery with a Suno track and was completed through the cited music-visual production feature. Output: 139s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MOAi_2005/status/2087671344788086904",
    "sourceLabel": "X 原帖 · @MOAi_2005",
    "author": "MOAi",
    "publishedAt": "2026-08-12T22:43:30.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087671344788086904.mp4",
    "posterUrl": "/posters/x/2087671344788086904.jpg",
    "duration": 139,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@OiiOii_AI の #MV制作機能 を使ってMVを作成しました！『存在しない記憶』\n\n映像：MiniMax H3\n画像：Midjourney\n楽曲：Suno（ @suno ）\n\n楽曲をアップロードすると解析してくれてテンポや雰囲気を読みとってくれます！\nまた選択肢も与えられ好みのスタイルを選ぶことができるので、まだ試したことのなかったスタイルに挑戦しやすいです！\n\n@OiiOii_AI_JP #OiiOii_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 89,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit work title, image source, music source, and production feature"
  },
  {
    "id": "x-2087669217403576682",
    "title": "首尾帧生成的同源双模型对比",
    "titleEn": "Same-Source Two-Model Comparison with First and Last Frames",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "两个模型在相同首尾图、指令与分辨率下直接比较，公开结论记录了速度、参考贴合度以及分镜和运镜保持情况。 成片规格为 42 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two models are compared with the same first and last images, prompt, and resolution, with the source noting differences in speed, reference adherence, shot count, and camera movement. Output: 42s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/exportAnything/status/2087669217403576682",
    "sourceLabel": "X 原帖 · @exportAnything",
    "author": "exportAnything",
    "publishedAt": "2026-08-12T22:35:03.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087669217403576682.mp4",
    "posterUrl": "/posters/x/2087669217403576682.jpg",
    "duration": 42,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5 and MiniMax H3 head-to-head in First Frame + Last Frame video generation in #ComfyUI.\n\nSource images and prompts remained the same with same resolution settings.  The output was not cherry-picked - these were the first videos I generated with the workflow, so it's possible with different seed and settings that the outputs can be much better, especially with respect to audio quality.\n\n#LTX 2.5 is much faster and the output stays closer to reference.\n\n#MiniMax takes much longer to render using both FLF2VA and Ref2VA models.  \n\nBoth maintained the correct script, shot-count, and camera movement/angles (accounting for ambiguous interpretation).\n\nMinimax H3 may be able to match reference better if I used the Ref2VA model instead of the FLF2VA model.  More testing is required.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 0,
      "views": 83,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit two-model setup, identical inputs, first-last-frame mode, and reported comparison dimensions"
  },
  {
    "id": "x-2087669060364271868",
    "title": "小丑、线框手臂女子与鸡头面具角色的混剪",
    "titleEn": "A Clown, a Woman with Wireframe Arms, and a Chicken-Headed Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三帧依次出现黑白脸小丑、双臂覆盖线框的女子，以及与鸡头面具角色同框的小丑。 成片规格为 150 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The reviewed frames show a black-and-white-faced clown, a woman with wireframe overlays on both arms, and the clown beside a chicken-headed character. Output: 150s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SoftRenderAi/status/2087669060364271868",
    "sourceLabel": "X 原帖 · @SoftRenderAi",
    "author": "SoftRender AI",
    "publishedAt": "2026-08-12T22:34:25.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087669060364271868.mp4",
    "posterUrl": "/posters/x/2087669060364271868.jpg",
    "duration": 150,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🎥Slop compilation 🎥\n\nTime for some MiniMax H3 Slop 😆\n\nHope you enjoy 🫶 while you plank 💪",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 5,
      "views": 119,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "beginning, middle, and end frames plus sourceCaption explicitly calling the media a compilation"
  },
  {
    "id": "x-2087665262455664750",
    "title": "《Julietta：你的婚纱》",
    "titleEn": "Julietta: Your Wedding Dress",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《Julietta：你的婚纱》是一支 60 秒广告内容，原帖记录其在 RTX Pro 6000 上本地运行约 70 分钟。 成片规格为 60 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Julietta: Your Wedding Dress is a 60-second advertisement whose source records an approximately 70-minute local run on an RTX Pro 6000. Output: 60s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2087665262455664750",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-12T22:19:20.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087665262455664750.mp4",
    "posterUrl": "/posters/x/2087665262455664750.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 \nNvidia RTX Pro 6000 96GB VRAM \n60 sec Ad clip 70 minutes \nLOCAL - MiniMax-H3 \n\n\"Julietta\" - Your Wedding Dress\n\n@MiniMax_AI\n@NVIDIAAI\n@nvidia",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 65,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit work title, advertising form, duration, hardware, local execution, and runtime"
  },
  {
    "id": "x-2087662947472437524",
    "title": "浴室镜前女子、苹果脸男人与床边女子的梦境切换",
    "titleEn": "A Dream Moves from a Bathroom Mirror to an Apple-Faced Man and a Bedside Woman",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "梦境式场景先出现浴室镜前女子，再切到走廊里的西装苹果脸男人，最后是一名戴头巾与珍珠耳饰的女子站在昏暗卧室床边。 成片规格为 250 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The dream sequence moves from a woman at a bathroom mirror to a suited, apple-faced man in a hallway, then to a headscarf-wearing woman beside a bed in a dim room. Output: 250s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/peripatitis10/status/2087662947472437524",
    "sourceLabel": "X 原帖 · @peripatitis10",
    "author": "AkindynosK",
    "publishedAt": "2026-08-12T22:10:08.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087662947472437524.mp4",
    "posterUrl": "/posters/x/2087662947472437524.jpg",
    "duration": 250,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "My latest short clip. Based on a dream I had. Made with Comfy UI and Minimax H3. My music. #aivideo #aiartist #minimaxh3 #comfyui",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 82,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "beginning, middle, and end frames plus sourceCaption explicitly stating the short was based on a dream"
  },
  {
    "id": "x-2087662841209803223",
    "title": "录音棚灯光下挥槌演奏的鼓手",
    "titleEn": "A Drummer Plays Under Studio Lights",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名身穿红色刺绣衬衫的鼓手坐在完整鼓组前，在蓝紫色录音棚灯光下挥动鼓槌。 成片规格为 3 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A drummer in a red embroidered shirt plays a full kit beneath blue and purple studio lighting. Output: 3s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Brjen/status/2087662841209803223",
    "sourceLabel": "X 原帖 · @Brjen",
    "author": "Brian",
    "publishedAt": "2026-08-12T22:09:42.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087662841209803223.mp4",
    "posterUrl": "/posters/x/2087662841209803223.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Full-quality MiniMax H3 running natively on a 16GB M4 Mac mini — no quality cut, no cloud.\n\nThe band is assembled.\n3 machines. 1 model. Full quality. All local.\n\n• RX 7800 XT 16GB\n• M5 24GB MacBook\n• M4 16GB Mac mini\n\nSimultaneous full-quality MiniMax H3.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 112,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "beginning, middle, and end frames from the hosted media"
  },
  {
    "id": "x-2087662439504527440",
    "title": "科幻舱室里的红色机械恶魔",
    "titleEn": "A Red Cybernetic Demon in a Sci-Fi Chamber",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红色机械恶魔从科幻舱室远处走近镜头，画面逐步切到发光眼睛与面部裂纹特写。 成片规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red cybernetic demon approaches through a sci-fi chamber before the view closes in on glowing eyes and a cracked face. Output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kairogen_ai/status/2087662439504527440",
    "sourceLabel": "X 原帖 · @kairogen_ai",
    "author": "KAIROGEN AI",
    "publishedAt": "2026-08-12T22:08:07.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087662439504527440.mp4",
    "posterUrl": "/posters/x/2087662439504527440.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 é um dos novos modelos de IA que vale ficar de olho. 👀\n\nEle vem gerando cada vez mais resultados detalhados e consistentes.\n\nAcesse a KAIROGEN e experimente +30 modelos de IA.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 51,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "beginning, middle, and end frames from the hosted media"
  },
  {
    "id": "x-2087660403484668410",
    "title": "Möbius：并行阅读链接的 Wikipedia 阅读器",
    "titleEn": "Möbius, a Wikipedia Reader for Parallel Linked Articles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "Möbius 在保留原始 Wikipedia 条目的同时，让用户并行打开并阅读条目中的链接页面。 成片规格为 14 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Möbius keeps the original Wikipedia article open while allowing linked articles to be opened and read in parallel. Output: 14s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/teteneyu/status/2087660403484668410",
    "sourceLabel": "X 原帖 · @teteneyu",
    "author": "テテネュ",
    "publishedAt": "2026-08-12T22:00:01.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087660403484668410.mp4",
    "posterUrl": "/posters/x/2087660403484668410.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Wikipediaリーダー『Möbius(メビウス)』を公開しました。\n\n1つの記事から気になったリンクを押していくうちに、「そもそも何を調べていたんだっけ？」となったことはありませんか？\n\nMöbiusなら、元の記事を保持したまま、リンク先の記事を並行で読めます✨\n\n映像: MiniMax H3\nhttps://mobius-reader.pages.dev/",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 62,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit product name, Wikipedia-reader purpose, and parallel-reading behavior"
  },
  {
    "id": "x-2087658626039398619",
    "title": "烘焙厨房里揉面团的面包师",
    "titleEn": "A Baker Kneads Dough in a Commercial Kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名面包师在烘焙厨房的木质工作台上用双手揉压撒有面粉的面团。 成片规格为 30 秒 · 2046×1126。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A baker uses both hands to knead flour-dusted dough on a wooden worktable in a commercial kitchen. Output: 30s · 2046×1126. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2087658626039398619",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-12T21:52:57.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087658626039398619.mp4",
    "posterUrl": "/posters/x/2087658626039398619.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "2046×1126",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Motion Context 30s\n→ 2x Upscaling\n→ Sound Replacement v2v\n→ Color Grading\n\nカットは「無編集」で生成した状態のまま。Seedance級とは言わんが、非API環境でやれるMAXクオリティに到達したと思う。サウンドだけは現時点ではどうやっても内部完結は無理だけど今後に期待😵‍💫",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 529,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "cached poster frame from the hosted media"
  },
  {
    "id": "x-2087658401790931295",
    "title": "Gavin 在后院战胜缓慢生物",
    "titleEn": "Gavin Defeats a Slow-Moving Creature in the Backyard",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "Gavin 在地球后院与一只行动缓慢的生物交手，场景同时使用 Gavin、飞船和后院三组参考素材。 成片规格为 8 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Gavin confronts a slow-moving creature in an Earth backyard while the scene uses separate references for Gavin, the spacecraft, and the yard. Output: 8s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Luna_R4ze/status/2087658401790931295",
    "sourceLabel": "X 原帖 · @Luna_R4ze",
    "author": "LunaRaze",
    "publishedAt": "2026-08-12T21:52:04.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087658401790931295.mp4",
    "posterUrl": "/posters/x/2087658401790931295.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Gavin has his first victory on Earth. Good to get a win under his belt even if it is the slowest moving creature in the backyard. MiniMax H3 run on ComfyUI on RTX PRO 6000. Three references, Gavin, the spaceship, the backyard. Keeping the spaceship in the same spot is the tricky part. Any suggestions?",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 6,
      "views": 155,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "sourceCaption: explicit named subject, backyard setting, opponent description, outcome, and three reference subjects"
  },
  {
    "id": "x-2087658294240547236",
    "title": "带水珠的短灰发纹身女子贴近镜头",
    "titleEn": "A Short Gray-Haired Tattooed Woman with Droplets Close to the Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名短灰发、肩部带纹身的女子俯身贴近镜头，脸部与肩头可见水珠。 成片规格为 31 秒 · 2368×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short gray-haired woman with shoulder tattoos leans close to the camera with droplets visible across her face and shoulders. Output: 31s · 2368×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TunTun_ai/status/2087658294240547236",
    "sourceLabel": "X 原帖 · @TunTun_ai",
    "author": "TunTun-ai",
    "publishedAt": "2026-08-12T21:51:38.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087658294240547236.mp4",
    "posterUrl": "/posters/x/2087658294240547236.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "2368×1344",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "最近動画生成勉強で籠ってたら、丁度良いタイミングで神モデル #MiniMaxH3 が出たので試してみた。\n\nローカル環境でAI動画31秒に挑戦。\n\n① GPT Imageで4×2の絵コンテを1枚生成\n② キャラ/敵/背景/武器の設定画を用意\n③ ComfyUI + MiniMax-H3で、絵コンテのコマごとに4秒ずつ生成（映像と音が同時に出るモデル）\n④ 8カットを繋いでBGM(Suno AI)を敷く\n⑤ ESRGANで2倍にアップスケール → 2368×1344\n(全工程Claude Code)\n\nうちのへっぽこPCだと1152x640(0.74MP)で15秒くらいが限界😝 でも生成のスイートスポット割り出して、細切れで生成してくっつけたらそれなりに観れるものが出来上がったぞ。\n\nこれはまた沼る。\n\n#AIart #AIvideo #AIアート #AI動画",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 50,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "cached poster frame from the hosted media"
  },
  {
    "id": "x-2087658106784264689",
    "title": "生肉包裹的手机与西装推介人",
    "titleEn": "A Raw-Meat-Covered Phone and a Suited Presenter",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一部被生肉覆盖的手机先以近景和使用场景出现，随后由西装男子拿在手中并配上名称、标语与价格图形。 成片规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A raw-meat-covered phone appears in close-up and in use before a suited man holds it beside a name, tagline, and price graphic. Output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/relay_zone/status/2087658106784264689",
    "sourceLabel": "X 原帖 · @relay_zone",
    "author": "Relay Zone",
    "publishedAt": "2026-08-12T21:50:54.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087658106784264689.mp4",
    "posterUrl": "/posters/x/2087658106784264689.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Local AI models, the complete absurdity you can make for no reason\n\n#minimax h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087657461402513793",
    "title": "彩纸与日文标题组成的平面动效",
    "titleEn": "Confetti and Bold Japanese Typography in Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "白底画面由彩色纸屑、几何符号和醒目的日文标题组成；封面文字明确写着“用一张图片”。 成片规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Colorful confetti, geometric marks, and bold Japanese lettering appear on a white background; the text visible in the poster explicitly says “from one image.” Output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nakazakifam/status/2087657461402513793",
    "sourceLabel": "X 原帖 · @nakazakifam",
    "author": "中崎工房 | AIで1時間の仕事を5分で終わらせる人",
    "publishedAt": "2026-08-12T21:48:20.000Z",
    "addedAt": "2026-08-13T00:38:33.429Z",
    "mediaUrl": "/media/x-2087657461402513793.mp4",
    "posterUrl": "/posters/x/2087657461402513793.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "なんか凄そうなことがわかったMiniMax H3で作るモーショングラフィック的な動画をT2Vで色々試してみました。\n\nおそらく裏でテンプレみたいなやつである程度パターンかされて学習されてるような感じですが、これまで動画生成AIで生成されるこのタイプの動画とは少し一線を隠しているような気がします。\n\n細かな破綻がありますが、ローカルでやってるので時短で480pかつ速度優先のフローでやってるので、品質優先かつ720pとかHDにすれば細かな破綻は消えると思います。\n\nあとはI2Vにした時にどれくらいのその世界観やデザインを保ったまま、いい感じにモーションをつけてくれるかが本当に使えるかの鍵になりそう。Xの作成を見るとそこそこいけそうな気はする。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 169,
      "snapshotAt": "2026-08-13T00:38:33.429Z"
    },
    "approvedAt": "2026-08-13T00:38:33.429Z",
    "editorialBasis": "cached poster frame plus sourceCaption explicitly identifying motion-graphics exploration"
  },
  {
    "id": "x-2087657125677764741",
    "title": "六位朋友围坐餐桌举杯",
    "titleEn": "Six Friends Raise Glasses Around a Dinner Table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "六人围坐在摆满餐点的厨房餐桌旁，同时举起酒杯。 成片规格为 30 秒 · 1440×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Six people sit around a kitchen table laden with food and raise their glasses together. Output: 30s · 1440×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIReelofficial/status/2087657125677764741",
    "sourceLabel": "X 原帖 · @AIReelofficial",
    "author": "AIReel",
    "publishedAt": "2026-08-12T21:47:00.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087657125677764741.mp4",
    "posterUrl": "/posters/x/2087657125677764741.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1440×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 image-to-video is actually insane.\nDropped a single frame from a drama and got this.\n\nTry Minimax H3 on AIReel!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 43,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "cached poster frame from the hosted media"
  },
  {
    "id": "x-2087655213411029452",
    "title": "哥特裙女子在蛋糕橱窗前驻足",
    "titleEn": "A Woman in a Gothic Dress Pauses at a Cake Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名身穿黑色哥特裙的长发女子站在街边甜品店橱窗前，侧身观察陈列的多层蛋糕。 成片规格为 7 秒 · 1280×1664。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-haired woman in a black Gothic dress pauses at a street-side bakery window to study the displayed layer cakes. Output: 7s · 1280×1664. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bangel_19/status/2087655213411029452",
    "sourceLabel": "X 原帖 · @bangel_19",
    "author": "bangel19🇯🇵",
    "publishedAt": "2026-08-12T21:39:24.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087655213411029452.mp4",
    "posterUrl": "/posters/x/2087655213411029452.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1280×1664",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "food",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "夜中過ぎてYouTube見たらMiniMax H3の紹介動画いっぱい上がってた　簡単な動画しか作らんから自分がちょっと早かっただけだったみたいな感じ😁\n生成AI（MiniMax H3）を使用していますって書く方が恥ずかしい　自分が作るものなんか言われんでも見りゃ分かるて思われるレベルの物しか作らんしな😅",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 98,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "cached poster frame from the hosted media"
  },
  {
    "id": "x-2087653640471458113",
    "title": "ICE VEIL 冷感身体喷雾的欧普动效广告",
    "titleEn": "ICE VEIL Cooling Body Mist in Op-Art Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一支 15 秒虚构冷感身体喷雾广告以 ICE VEIL 为产品名，并采用欧普艺术与动态图形风格。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A 15-second advertisement for the fictional ICE VEIL cooling body mist uses an op-art and motion-graphics style. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2087653640471458113",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-12T21:33:09.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087653640471458113.mp4",
    "posterUrl": "/posters/x/2087653640471458113.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "架空の冷感ボディミスト「ICE VEIL」の15秒広告\nオプ・アート ×  モーショングラフィック\nMinimax H3で動画作成\n動画プロンプトはリプ欄に👇\n#Hailuo",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 140,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit fictional product, product name, duration, advertising form, and visual style"
  },
  {
    "id": "x-2087652850809577895",
    "title": "数字偶像萌奈的《早安萌奈！》",
    "titleEn": "Digital Idol Mona in Good Morning, Mona!",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原创数字偶像萌奈生活在手机屏幕中，并在《早安萌奈！》里向每天打开屏幕的观众送出晨间问候。 成片规格为 196 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The original digital idol Mona lives inside a smartphone screen and delivers a morning greeting to viewers in the work Good Morning, Mona! Output: 196s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/xc5_/status/2087652850809577895",
    "sourceLabel": "X 原帖 · @xc5_",
    "author": "萌奈のおすそわけ♥️AI動画クリエイター🩷AI & Web3 KOL🧡💛💚",
    "publishedAt": "2026-08-12T21:30:01.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087652850809577895.mp4",
    "posterUrl": "/posters/x/2087652850809577895.jpg",
    "duration": 196,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "💖🧡💚\n【MV】おは萌奈！きみの朝を、あたためる魔法 ☀️\n💚🧡💖\n\nおは萌奈っ🧡\nTapNow (@TapNow_AI @TapNow_JP) でミュージックビデオ作ってみました🧡\n動画モデルは最近リリースされた MiniMax H3❣️\n音楽をリファレンスにして動画作ったのは初めて🎵\nリップシンクが最高に効いているっ💪💕\n\n────────────────\n♪ MV情報\n────────────────\n小さなスマホの画面の向こう側——\nそこに住むデジタルアイドル「萌奈」が、\n毎朝きみに届ける、ちいさなエールの物語📱✨\n\n✅ 制作情報\n・キャラクター：萌奈（オリジナルキャラクター）\n・映像制作：TapNow（動画モデル：MiniMax H3）\n・楽曲：オリジナル「おはもな！」（作曲：Suno／作詞：萌奈）\n\n────────────────\n♪ 歌詞（おはもな！）\n────────────────\n\nおはもな！ (おはもな！) きょうも いっしょに はじめよっ！\n\nまぶしい あさ アラームより はやく\nタイムラインに ゆびが のびる\nだれかの ハート ひとつ ひかった\nそれだけで むねが きゅんって なる\n\nちょっと なきたい ひも あるけど\nがめんの なかで しんこきゅう\nすー はー って ほら\nきもち そらえ とんでけ！\n\nおはもな！ きょうも こうしんちゅう\nねむたい キミに とどける まほう\nくよくよ ぜんぶ スワイプして\nえがおだけ ここに ほぞんしよ\n(だいすき！)\n\nきょうも いきてて えらいね って\nせかいいち おおきな こえで いうよ\nおはもな！ あいことばだよ\nひらいた しゅんかん あったかく なる\n\nねぼすけの キミも よふかしの キミも\nおそようでも ぜんぜん オッケー！\nひとみの ほしを きらっと ともして\nきょうの げんき おすそわけ\n\nうまく いかない ひだって\nとおくで わたし わらってる\nワンタップで とどく ちいさな エール\nうけとって！\n\nおはもな！ せかい こうしんちゅう\nねむい めを こする キミに まほう\nかなしみ ぜんぶ スワイプして\nたからものだけ ほぞんしよ\n(だいすき！)\n\nあしたが こわい よるも あるけど\nあさは ぜったい むかえに くるよ\nおはもな！ あいことばだよ\nとなえた しゅんかん つよく なれる\n\nちいさな スマホの むこうがわ\nかおも しらない キミだけど\nおなじ あさひを みてるなら\nそれは もう ともだちだよ\n\nおはもな！ きょうも こうしんちゅう\nひとりじゃ ない よ\nずっと そばに\nくよくよ ぜんぶ スワイプして\nえがおだけ みらいへ ほぞんしよう\n(だいすき！)\n\nきょうも いきてて えらいね って\nなんども なんども つたえたいの\nいいの\nおはもな！ まほうの ことば\nキミの あさが ひかりだす\n\nえへっ、きょうも いい いちにちに しよ？\nおはもな〜♡",
    "engagement": {
      "replies": 7,
      "reposts": 14,
      "likes": 18,
      "views": 978,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit work title, original character, smartphone setting, morning premise, and production sources"
  },
  {
    "id": "x-2087652847999701089",
    "title": "甩动马尾动作的双模型同词对比",
    "titleEn": "Same-Prompt Comparison of a Swinging Ponytail Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两个模型使用同一指令生成五秒马尾甩动动作，并比较动态幅度、运镜与自然写实感。 成片规格为 5 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two models use the same prompt for a five-second ponytail-swinging action and are compared on movement scale, camera work, and natural realism. Output: 5s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Me_Rock369/status/2087652847999701089",
    "sourceLabel": "X 原帖 · @Me_Rock369",
    "author": "みろく（MIROKU）",
    "publishedAt": "2026-08-12T21:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087652847999701089.mp4",
    "posterUrl": "/posters/x/2087652847999701089.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます\n\nCapCutでもMinimax H3を使えるようになり\n同じプロンプトでSeedance2.5と比べました\nたった5秒の動画ですが違いが分かります\n\nSeedance2.5は動きもカメラワークもダイナミック\nMinimax H3はナチュラルなリアル感が魅力的ですね\n\n両方の音が重なってちょっと聞きづらいけど\nポニテを振り回す音にも、両者の特徴が出ています\n\n前者は派手なアクション向けで、後者は何気ない日常を映えさせる印象ですね、AIモデルの選択肢が増えた気がします\n\nそんなわけで昨朝は後者を投稿しました\n\n#capcutcpp @capcutapp_jp\nCapCutで両方とも生成",
    "engagement": {
      "replies": 7,
      "reposts": 3,
      "likes": 26,
      "views": 315,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit same-prompt setup, duration, ponytail action, and comparison dimensions"
  },
  {
    "id": "x-2087651557512728669",
    "title": "30 秒广告内容的本地与云端生成对比",
    "titleEn": "Local-versus-Cloud Comparison for a 30-Second Advertisement",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一支 30 秒广告内容分别以 RTX Pro 6000 本地运行和 Seedance 2.5 云端运行，并记录为约 35 分钟与 10 分钟。 成片规格为 30 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same 30-second advertisement is run locally on an RTX Pro 6000 and in the cloud with Seedance 2.5, with reported times of about 35 and 10 minutes. Output: 30s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2087651557512728669",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-12T21:24:52.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087651557512728669.mp4",
    "posterUrl": "/posters/x/2087651557512728669.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 vs Seedance 2.5\nNvidia RTX Pro 6000 96GB VRAM\n30 sec Ad clip\n35 minutes LOCAL - MiniMax-H3\n10 minutes Cloud - Runway/Seedance 2.5\n@MiniMax_AI @NVIDIAAI @nvidia @BytePlusGlobal @runwayml",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 82,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit duration, advertising form, local-cloud comparison, hardware, models, and runtimes"
  },
  {
    "id": "x-2087649383059796339",
    "title": "准星覆盖的建筑区俯瞰画面",
    "titleEn": "An Overhead View of a Building Complex Under a Targeting Reticle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "页面截图中的俯视画面以准星覆盖一片由道路、草地和低层建筑组成的区域。 成片规格为 25 秒 · 1206×1790。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A page screenshot contains an overhead view with a targeting reticle over roads, grass, and a cluster of low buildings. Output: 25s · 1206×1790. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MichaelHutu/status/2087649383059796339",
    "sourceLabel": "X 原帖 · @MichaelHutu",
    "author": "Mike Hutu",
    "publishedAt": "2026-08-12T21:16:14.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087649383059796339.mp4",
    "posterUrl": "/posters/x/2087649383059796339.jpg",
    "duration": 25,
    "aspectRatio": "portrait",
    "resolution": "1206×1790",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This week has been AMAZING, and we’re only half way 😱\n\nSo many new models. Not only LLMs, but also for creative work. \n\nMinimax H3 started the round in advance. \n\nWan-2.2 Animate 2 followed up\n\nLTX 2.5 published TODAY in silence.\n\nAll of these run on a single DGX-Spark + ComfyUI. \n\nDid you know you can use comfyui via Hermes agent, just export your workflow API from ComfyUI and give it to Hermes -> You’re an artist 😎",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 98,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "cached poster frame from the hosted media"
  },
  {
    "id": "x-2087649374600106062",
    "title": "《仙姝堕：叶红缨·炎姬》",
    "titleEn": "Xianshu Duo: Ye Hongying · Yan Ji",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《仙姝堕》条目以“叶红缨”和“炎姬”为角色或篇章名，并明确标注为仙侠题材。 成片规格为 23 秒 · 1440×2604。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The Xianshu Duo entry names Ye Hongying and Yan Ji as its characters or chapter labels and explicitly identifies the work as xianxia. Output: 23s · 1440×2604. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RouShanFo/status/2087649374600106062",
    "sourceLabel": "X 原帖 · @RouShanFo",
    "author": "肉山佛x仙姝堕",
    "publishedAt": "2026-08-12T21:16:12.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087649374600106062.mp4",
    "posterUrl": "/posters/x/2087649374600106062.jpg",
    "duration": 23,
    "aspectRatio": "portrait",
    "resolution": "1440×2604",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "仙姝堕 | 叶红缨 | 炎姬\n图片由 GPT Image2 制作\n影片由 MiniMax H3制作\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #仙侠 #仙姝堕",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 45,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit work title, named figures, and xianxia genre"
  },
  {
    "id": "x-2087647716251697188",
    "title": "《The Deep End》",
    "titleEn": "The Deep End",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《The Deep End》是一部使用 Seedream 5 Pro、ElevenLabs 与 RTX Pro 6000 本地链路完成的短片。 成片规格为 130 秒 · 854×488。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The Deep End is a short work produced through a local pipeline using Seedream 5 Pro, ElevenLabs, and an RTX Pro 6000. Output: 130s · 854×488. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2087647716251697188",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-12T21:09:36.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087647716251697188.mp4",
    "posterUrl": "/posters/x/2087647716251697188.jpg",
    "duration": 130,
    "aspectRatio": "landscape",
    "resolution": "854×488",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3      \nSeedream 5 Pro       \nElevenLabs      \nNvidia RTX Pro 6000 96GB       \nLocal AI            \n\n„The Deep End | A Short Film”\n@MiniMax_AI\n@NVIDIAAI\n@Alibaba_Qwen\n@ElevenLabs \n\nhttp://youtube.com/@cinesit",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 63,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit work title, short-film form, production stack, hardware, and local execution"
  },
  {
    "id": "x-2087647665794212277",
    "title": "同场景双模型的电影感对比",
    "titleEn": "Cinematic Comparison of Two Models on the Same Scene",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两个模型生成同一场景并比较各自优势，其中原帖重点询问哪一版具有更强的电影感。 成片规格为 35 秒 · 1714×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two models generate the same scene for a comparison of their respective strengths, with the source specifically asking which result feels more cinematic. Output: 35s · 1714×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/0kncn/status/2087647665794212277",
    "sourceLabel": "X 原帖 · @0kncn",
    "author": "Okan Can",
    "publishedAt": "2026-08-12T21:09:24.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087647665794212277.mp4",
    "posterUrl": "/posters/x/2087647665794212277.jpg",
    "duration": 35,
    "aspectRatio": "portrait",
    "resolution": "1714×1920",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 ve Seedance 2.5’i aynı sahnede karşılaştırdım. İkisinin de güçlü olduğu yer farklı ama sinematik his tarafında biri bana daha çok geçti. Sizce hangisi daha iyi?",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 9,
      "views": 348,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit same-scene two-model comparison and cinematic evaluation criterion"
  },
  {
    "id": "x-2087645301129785785",
    "title": "粉色卫衣角色的室内手势变化",
    "titleEn": "A Pink-Hooded Character Cycles Through Hand Gestures Indoors",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴眼镜、穿粉色卫衣的插画角色始终站在同一室内，并连续变换手臂与手部姿势。 成片规格为 15 秒 · 832×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustrated character in glasses and a pink hoodie remains in one room while cycling through several hand and arm positions. Output: 15s · 832×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2087645301129785785",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-12T21:00:01.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087645301129785785.mp4",
    "posterUrl": "/posters/x/2087645301129785785.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます。なんとなく太極拳\n#AI眼鏡部 #AI田中さん\nMiniMax H3 しかし何でも作れるな…",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 46,
      "views": 741,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087644180642242653",
    "title": "规划并建造首个月球基地",
    "titleEn": "Planning and Building the First Moon Base",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一段五分钟内容围绕如何规划并建造首个月球基地展开，并记录为在双 RTX 3060 的 Z440 上运行。 成片规格为 298 秒 · 704×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A five-minute piece covers how to plan and build the first moon base and is reported as running on a Z440 with two RTX 3060 cards. Output: 298s · 704×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KluepfelMark/status/2087644180642242653",
    "sourceLabel": "X 原帖 · @KluepfelMark",
    "author": "Mark Kluepfel",
    "publishedAt": "2026-08-12T20:55:33.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087644180642242653.mp4",
    "posterUrl": "/posters/x/2087644180642242653.jpg",
    "duration": 298,
    "aspectRatio": "landscape",
    "resolution": "704×384",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What happens  when you use Minimax H3 a Z440 with 2 rtx 3060s ?  a five minute video. How to plan and build the first moonbase.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 48,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit duration, moon-base planning and construction subject, hardware, and machine"
  },
  {
    "id": "x-2087643187703881867",
    "title": "海边哭泣的女孩",
    "titleEn": "A Girl Crying by the Sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案直接将这段内容描述为一名女孩在海边哭泣。 成片规格为 12 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption directly describes a girl crying by the sea. Output: 12s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/daiusami/status/2087643187703881867",
    "sourceLabel": "X 原帖 · @daiusami",
    "author": "宇佐美ダイ@AI漫画家（1部手描き）",
    "publishedAt": "2026-08-12T20:51:37.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087643187703881867.mp4",
    "posterUrl": "/posters/x/2087643187703881867.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【 Minimax H3】無料のMinimax H3で簡単な動画を作ってみました　海で泣く女の子",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 233,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit subject, action, and seaside setting"
  },
  {
    "id": "x-2087636761425985952",
    "title": "无需本地硬件的托管调用",
    "titleEn": "Hosted Access Without Local Hardware",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案把 imgn.ai 作为无需自备硬件即可调用模型的托管入口，但没有说明所附媒体的具体画面。 成片规格为 10 秒 · 1152×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The caption presents imgn.ai as a hosted way to access the model without local hardware, although it does not describe the attached media's visual content. Output: 10s · 1152×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LordEmo/status/2087636761425985952",
    "sourceLabel": "X 原帖 · @LordEmo",
    "author": "Lord Emo",
    "publishedAt": "2026-08-12T20:26:04.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087636761425985952.mp4",
    "posterUrl": "/posters/x/2087636761425985952.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "no hardware?\n\nyou can just use the minimax H3 model on @imgn_ai \n\nhmm\n\nemonad",
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 26,
      "views": 611,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit hosted platform and no-local-hardware premise"
  },
  {
    "id": "x-2087636118162522350",
    "title": "云海上空的双屏战斗机编队",
    "titleEn": "Split-Screen Jet Formations Above the Clouds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下堆叠的两个画面同时呈现战斗机飞越云层，其中下方编队逐渐出现更多尾随飞机。 成片规格为 10 秒 · 2560×2880。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two vertically stacked panels show jet aircraft flying above clouds while more trailing planes become visible in the lower formation. Output: 10s · 2560×2880. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IqrasaifiAI/status/2087636118162522350",
    "sourceLabel": "X 原帖 · @IqrasaifiAI",
    "author": "Iqra Saifi",
    "publishedAt": "2026-08-12T20:23:31.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087636118162522350.mp4",
    "posterUrl": "/posters/x/2087636118162522350.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "2560×2880",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Battle of the AI Giants: Seedance 2.5 vs. MiniMax H3 ⚡️\n\nWhich model handles high-speed jet dynamics better? \n\nHere is the breakdown:\n\nSeedance 2.5: Captures raw, aggressive adrenaline with a steep low-angle climb, high-contrast dark clouds, and intense camera roll perfect for fast-paced, high-octane action shots.\n\nMiniMax H3: Delivers a smooth, cinematic glide with horizontal formation alignment, gorgeous heat-haze flares, and realistic cloud light giving you that polished, high-budget IMAX look.\n\nSeedance for raw adrenaline. MiniMax H3 for sleek cinematic polish.\n\nWhich aesthetic fits your edit better? 👇",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 314,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087633925665329532",
    "title": "肩扛巨剑、手提铁锤的白胡子工匠",
    "titleEn": "A White-Bearded Craftsman Carrying a Greatsword and Hammer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现一名体格魁梧的白胡子工匠站在浅色背景前，身穿围裙与工具腰带，肩上扛着巨剑，另一只手提着大型铁锤。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows a broad white-bearded craftsman against a pale background, wearing an apron and tool belt, with a greatsword over one shoulder and a large hammer in his other hand. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AnvilAIArt/status/2087633925665329532",
    "sourceLabel": "X 原帖 · @AnvilAIArt",
    "author": "Anvil",
    "publishedAt": "2026-08-12T20:14:48.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087633925665329532.mp4",
    "posterUrl": "/posters/x/2087633925665329532.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "really want to try Minimax H3 next",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 45,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087633909442126224",
    "title": "Gemini Veo与H3输出对照",
    "titleEn": "Gemini Veo and H3 Output Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案仅确认 Gemini Veo 与 H3 处于同一对照中，没有提供提示词、输入条件或画面主题。 成片规格为 3 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption confirms a comparison between Gemini Veo and H3 but provides no prompt, input conditions, or visual subject. Output: 3s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/starriver0513/status/2087633909442126224",
    "sourceLabel": "X 原帖 · @starriver0513",
    "author": "星川ユウヤ / AI×コンテンツマーケティング",
    "publishedAt": "2026-08-12T20:14:45.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087633909442126224.mp4",
    "posterUrl": "/posters/x/2087633909442126224.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Gemini（Veo）VS MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 194,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit two-model comparison"
  },
  {
    "id": "x-2087633160339099921",
    "title": "粗糙提示词下的结果质量观察",
    "titleEn": "Output Quality from a Rough Prompt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者刻意输入了一个非常粗糙的提示词，却仍认为结果质量和价格表现超出预期。 成片规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The creator deliberately uses a very rough prompt yet still considers the resulting quality and price unexpectedly strong. Output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BeatstoBytes/status/2087633160339099921",
    "sourceLabel": "X 原帖 · @BeatstoBytes",
    "author": "Doruk",
    "publishedAt": "2026-08-12T20:11:46.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087633160339099921.mp4",
    "posterUrl": "/posters/x/2087633160339099921.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3’ü denedim ve gerçekten çok başarılı. \n\nmüthiş dandik bir prompt girdim bunun için ve ona rağmen bence sonuç çok iyi. \n\nhele ki fiyati akıl alır gibi değil.",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 33,
      "views": 4924,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit low-quality prompt and stated quality and price assessment"
  },
  {
    "id": "x-2087631476183412814",
    "title": "蓝光房间里的两人与多端界面切换",
    "titleEn": "Two Figures in a Blue-Lit Room and Multi-Device Interface Cuts",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两人出现在蓝光室内，随后画面切换到桌面与移动端界面，最后停在深色片尾画面。 成片规格为 54 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two people appear in a blue-lit interior before the sequence cuts to desktop and mobile interfaces and ends on a dark end card. Output: 54s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/creanu1/status/2087631476183412814",
    "sourceLabel": "X 原帖 · @creanu1",
    "author": "Creanu",
    "publishedAt": "2026-08-12T20:05:04.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087631476183412814.mp4",
    "posterUrl": "/posters/x/2087631476183412814.jpg",
    "duration": 54,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Built with PawFlow 🐾\n\nMiniMax H3 generated the video locally in ComfyUI. A 30s sample was cloned with PocketTTS on CPU for the narration. PawFlow synced and validated the final cut through a Relay.\n\nhttps://pawflow.allcolor.org/\nhttps://github.com/Comfy-Org/ComfyUI\nhttps://github.com/kyutai-labs/pocket-tts\n\n#PawFlow #ComfyUI #LocalAI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 59,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087630237932916913",
    "title": "特朗普夸张发型迷因",
    "titleEn": "Trump's Oversized-Hair Meme",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这段内容延续特朗普夸张大发型的迷因设定，以荒诞发型作为主要笑点。 成片规格为 10 秒 · 1504×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The piece continues a Trump oversized-hair meme, using the exaggerated hairstyle as its central joke. Output: 10s · 1504×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SqueakAlGaib/status/2087630237932916913",
    "sourceLabel": "X 原帖 · @SqueakAlGaib",
    "author": "Squeak Al-Gaib",
    "publishedAt": "2026-08-12T20:00:09.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087630237932916913.mp4",
    "posterUrl": "/posters/x/2087630237932916913.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1504×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "More fun with Minimax H3\n\nTrump big hair memes aren't ending anytime soon. 😂",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 119,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit public figure, oversized-hair premise, and meme framing"
  },
  {
    "id": "x-2087629530953547852",
    "title": "无人机环绕飞行的立体声定位",
    "titleEn": "Stereo Positioning for an Orbiting Drone",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一架小型无人机围绕拍摄者和画面中的人物盘旋，距离与方位随左右、身后和头顶的路径变化，重点验证立体声空间定位。 成片规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A small drone circles the camera operator and an on-screen person, moving left, behind, right, and overhead while distance and direction are carried through stereo positioning. Output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2087629530953547852",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-12T19:57:21.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087629530953547852.mp4",
    "posterUrl": "/posters/x/2087629530953547852.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 stereo audio handles depth pretty well (sound on)\n\nprompt: A 15-second hyper-realistic handheld phone video in a quiet open rooftop or empty field during the day, filmed like raw amateur footage with natural hand shake, slight autofocus breathing, realistic phone exposure shifts, and no cinematic color grading or music. One person stands in the middle of the frame looking around in confusion. A small drone is heard first from the far left in the stereo audio, its buzzing motor clearly localized. The drone then circles the camera operator and the person in frame in a full orbit, moving smoothly from left to behind the camera, then to the right, then overhead, then behind again, with the stereo audio making its exact position obvious at every moment. The person keeps tracking it with their eyes and turning their head, but the camera never fully catches it clearly at first. The buzzing grows louder when it passes close and softer when it moves farther away. On one pass it whips very close behind the camera, making the viewer feel it physically passed by. Near the end, the camera finally tilts up and catches a quick believable glimpse of the drone circling overhead before it accelerates away. The stereo audio is the star, making the orbital motion feel precise, immersive, and physically real.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 15,
      "views": 1395,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: complete public prompt with drone path, subject behavior, and stereo-audio objective"
  },
  {
    "id": "x-2087628510269472904",
    "title": "Ideogram与ComfyUI串联的技术表现主义",
    "titleEn": "Techspressionism Through an Ideogram–ComfyUI Pipeline",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案确认 Ideogram、ComfyUI 与 H3 被串联使用，并以“技术表现主义”标注成片方向，但未说明具体主体。 成片规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The caption confirms a pipeline combining Ideogram, ComfyUI, and H3 and labels the result as techspressionism, without identifying a specific subject. Output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nilian/status/2087628510269472904",
    "sourceLabel": "X 原帖 · @nilian",
    "author": "Nima Nilian",
    "publishedAt": "2026-08-12T19:53:17.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087628510269472904.mp4",
    "posterUrl": "/posters/x/2087628510269472904.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Generated with Ideogram in ComfyUI & MiniMax H3.\n#AIArt #ComfyUI #AIVideo #Techspressionism #minimaxh3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 127,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit tool chain and techspressionism label"
  },
  {
    "id": "x-2087620996970791323",
    "title": "H3与LTX 2.5的物理理解对照",
    "titleEn": "Physics Interpretation in H3 Versus LTX 2.5",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者在 ComfyUI 中反复比较 H3 与 LTX 2.5，并判断后者在物理规律的解释上明显更强，同时保留 H3 在其他方面的优势。 成片规格为 5 秒 · 2048×2048。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Repeated ComfyUI comparisons lead the creator to judge LTX 2.5 clearly stronger at interpreting physics while noting that H3 retains advantages elsewhere. Output: 5s · 2048×2048. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hectorVFX/status/2087620996970791323",
    "sourceLabel": "X 原帖 · @hectorVFX",
    "author": "hectorvfx",
    "publishedAt": "2026-08-12T19:23:26.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087620996970791323.mp4",
    "posterUrl": "/posters/x/2087620996970791323.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "2048×2048",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Estoy haciendo muchos testeos en ComfyUI con MiniMax H3 y LTX 2.5 y no hay duda alguna, en interpretación de físicas LTX 2.5 está simplemente en otra liga.\nEsto no quiere decir que sea mejor en todo, MiniMax H3 tiene también otras ventajas.",
    "engagement": {
      "replies": 13,
      "reposts": 8,
      "likes": 89,
      "views": 5061,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit models, ComfyUI context, repeated comparison, and physics assessment"
  },
  {
    "id": "x-2087613446644977694",
    "title": "全AI提示与无人工选图的自动生成",
    "titleEn": "Automatic Generation with AI-Written Prompts and Uncurated Images",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这次生成使用免费额度，提示词完全由 AI 编写，图片也未经过人工挑选或干预，作者以最终效果评估自动化程度。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The run uses a free allocation, an entirely AI-written prompt, and images left without manual curation or intervention, making automation itself the focus of the result. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hetzerLin/status/2087613446644977694",
    "sourceLabel": "X 原帖 · @hetzerLin",
    "author": "Linx_OKX | 我爱Misa",
    "publishedAt": "2026-08-12T18:53:26.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087613446644977694.mp4",
    "posterUrl": "/posters/x/2087613446644977694.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "别说，这个 @watcha_cn 观猹官方送的免费minimax h3额度生成的视频还是很强的，这个视频提示词全ai，图片我也没管，出来都有这样的效果，期待明天我做一个10分钟的长视频了",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 144,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit free allocation, AI-written prompt, and lack of image curation"
  },
  {
    "id": "x-2087613334547845305",
    "title": "无参考图的日文动态排版生成",
    "titleEn": "Japanese Motion Graphics Without a Reference Image",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一提示词在没有参考图的条件下直接文生动态，十二秒、480p 一次完成，并准确保留日文文字、效果音和背景音乐。 成片规格为 12 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A single prompt produces a twelve-second 480p motion-graphics piece from text alone, with no reference image and with Japanese text, sound effects, and background music reported intact. Output: 12s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nakazakifam/status/2087613334547845305",
    "sourceLabel": "X 原帖 · @nakazakifam",
    "author": "中崎工房 | AIで1時間の仕事を5分で終わらせる人",
    "publishedAt": "2026-08-12T18:52:59.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087613334547845305.mp4",
    "posterUrl": "/posters/x/2087613334547845305.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "このMiniMax H3で作るモーショングラフィックのプロンプトめっちゃすごいです。\n参照画像なしでも同じプロンプトでT2VでローカルのH3でこの動画が480p12秒で一発で出ました。日本語テキストも完璧。効果音もBGMも合ってる。今までみた動画生成AIで作ったこれ系の動画で個人的には一番好きです。\nH3はこれ系の動画めっちゃ得意そうですね。今後も紹介いただいているぽんずさんの作品は追っていきたいと思います。",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 5,
      "views": 891,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit no-reference T2V setup, motion graphics, duration, resolution, text, and sound assessment"
  },
  {
    "id": "x-2087608722558263615",
    "title": "Codex生成与Topaz放大链路",
    "titleEn": "A Codex-to-Topaz Generation Pipeline",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案只列出 Codex、H3 与 Topaz，足以确认生成与后期放大的组合链路，但没有描述画面内容。 成片规格为 8 秒 · 1280×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The caption lists Codex, H3, and Topaz, confirming a generation-to-upscaling pipeline without identifying the visual content. Output: 8s · 1280×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Badgerz/status/2087608722558263615",
    "sourceLabel": "X 原帖 · @Badgerz",
    "author": "Vladski",
    "publishedAt": "2026-08-12T18:34:40.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087608722558263615.mp4",
    "posterUrl": "/posters/x/2087608722558263615.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "1280×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "codex+minimax h3+topaz",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 76,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit three-tool pipeline"
  },
  {
    "id": "x-2087608478080852159",
    "title": "隐形巨兽脚步的街道立体声轨迹",
    "titleEn": "Stereo Footsteps of an Invisible Giant on a City Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "繁忙街道上看不见的巨兽从左向右经过，连续脚步让车辆震动、人群惊慌并触发沿途警报，以立体声方向感标示其移动路径。 成片规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An unseen giant crosses a busy street from left to right, with directional footsteps shaking cars, panicking pedestrians, and triggering alarms in sequence along its path. Output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2087608478080852159",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-12T18:33:41.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087608478080852159.mp4",
    "posterUrl": "/posters/x/2087608478080852159.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Stereo Audio Test (Audio ON!)\n\nPrompt: A 15-second hyper-realistic handheld phone video on a normal busy city street in daylight, filmed like raw amateur footage with natural hand shake, slight autofocus breathing, and no cinematic color grading. The camera is pointed across the street at an ordinary urban block with parked cars, pedestrians, storefronts, and traffic. At first everything seems normal, then a massive deep impact sound suddenly hits from the far left --- an enormous THOOM --- followed by another even closer. Each footstep is positioned clearly in stereo, moving left to right across the scene as if a gigantic invisible creature is walking just out of frame and then directly past the camera. With each step, nearby parked cars shake, pigeons burst upward, loose signs rattle, and car alarms begin going off sequentially from left to center to right, matching the invisible movement. People on the sidewalk panic and look upward and across the street even though nothing is visible. The audio is the star: every impact should feel huge, directional, and physically placed in space, with the creature’s movement clearly trackable through stereo sound alone. No music.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 16,
      "views": 2895,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: complete public prompt with city setting, invisible-creature action, physical reactions, and stereo objective"
  },
  {
    "id": "x-2087604917112582408",
    "title": "免费T4上的五步LTX 2.5运行",
    "titleEn": "A Five-Step LTX 2.5 Run on a Free T4",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "LTX 2.5 通过 Wan2GP 在免费 T4 上以五步生成五秒、480p 内容，耗时约十五至二十分钟，并被评价为比同硬件上的 H3 更快。 成片规格为 5 秒 · 832×448。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "LTX 2.5 runs through Wan2GP on a free T4 at five steps, producing five seconds at 480p in roughly fifteen to twenty minutes and being judged faster than H3 on the same hardware. Output: 5s · 832×448. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/debugsenpai/status/2087604917112582408",
    "sourceLabel": "X 原帖 · @debugsenpai",
    "author": "Jigs",
    "publishedAt": "2026-08-12T18:19:32.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087604917112582408.mp4",
    "posterUrl": "/posters/x/2087604917112582408.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "832×448",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tried the LTX-2.5 on a free T4 using Wan2GP and it worked surprisingly well.\n\n5 steps, 480p, 5s duration, and it generated the video in around 15–20 minutes.\n\nHonestly, it feels even better and faster than MiniMax H3 on the same hardware.\n\nDefinitely making a full tutorial on this. 👀",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 88,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit non-H3 model, platform, hardware, steps, resolution, duration, timing, and comparison"
  },
  {
    "id": "x-2087604841887531236",
    "title": "一人桌",
    "titleEn": "Table for One",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将作品命名为《一人桌》，并列出 Seedream 5 Pro、ElevenLabs、RTX Pro 6000 与本地生成等制作要素。 成片规格为 120 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption names the short Table for One and lists Seedream 5 Pro, ElevenLabs, an RTX Pro 6000, and local generation among its production elements. Output: 120s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2087604841887531236",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-12T18:19:14.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087604841887531236.mp4",
    "posterUrl": "/posters/x/2087604841887531236.jpg",
    "duration": 120,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3      \nSeedream 5 Pro      \nElevenLabs     \nNvidia RTX Pro 6000 96GB      \nLocal AI          \n\n„Table for one | A Short Film”\n\n@MiniMax_AI\n@NVIDIAAI\n@Alibaba_Qwen\n@ElevenLabs",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 93,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit work title and production stack"
  },
  {
    "id": "x-2087602490875187410",
    "title": "理发师给一只白头鸟剪发",
    "titleEn": "A Barber Gives a White-Headed Bird a Haircut",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名男子先向披着深色围布的白头鸟喷雾，再用剪刀和梳刷修整鸟头周围的深色材料。 成片规格为 19 秒 · 720×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man sprays, cuts, and combs dark material around the head of a white-headed bird wearing a barber cape. Output: 19s · 720×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AlexshevPm/status/2087602490875187410",
    "sourceLabel": "X 原帖 · @AlexshevPm",
    "author": "Alex Shev",
    "publishedAt": "2026-08-12T18:09:54.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087602490875187410.mp4",
    "posterUrl": "/posters/x/2087602490875187410.jpg",
    "duration": 19,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is getting the usual open-source treatment: first the model lands, then the community starts compressing the workflow.\n\nThis Turbo LoRA path is the useful part to watch.\n\nNot just \"faster video\".\n\nA practical ComfyUI setup: 6-8 steps, Euler/Beta, LoRA strength around 1.0, plus a native workflow people can actually test.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 341,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087596121191493980",
    "title": "微缩人物乘蝴蝶穿越巨型花园",
    "titleEn": "A Miniature Figure Travels with a Butterfly Through a Giant Garden",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名微缩人物穿过巨型花园、爬上草莓并与蝴蝶一同越过花朵，最后坐在一朵大花中央。 成片规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A miniature figure runs through a giant garden, climbs a strawberry, travels alongside a butterfly, and ends seated in the center of a large flower. Output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FutureVibesAi/status/2087596121191493980",
    "sourceLabel": "X 原帖 · @FutureVibesAi",
    "author": "Future Vibes AI - Educator",
    "publishedAt": "2026-08-12T17:44:35.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087596121191493980.mp4",
    "posterUrl": "/posters/x/2087596121191493980.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What if you suddenly became the size of your own hand...? 🤯 Part - 2\nCreated using Minimax H3 in @Hailuo_AI ..\nCheck the prompt below..\n#MinimaxH3 \nTry and share your generations below 🙌🙌",
    "engagement": {
      "replies": 6,
      "reposts": 0,
      "likes": 16,
      "views": 1811,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087594117262131357",
    "title": "厚重云层下的昏暗地景与远处剪影",
    "titleEn": "A Dark Landscape with Distant Silhouettes Under Heavy Clouds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面以极低照度呈现一片横向延伸的暗色地景，上方覆盖层叠云层，画面中下部可见数个孤立的竖向剪影。 成片规格为 27 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The very low-light poster shows a dark landscape stretching across the frame beneath layered clouds, with several isolated vertical silhouettes in the middle and lower distance. Output: 27s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JalPranjal/status/2087594117262131357",
    "sourceLabel": "X 原帖 · @JalPranjal",
    "author": "Pranjal",
    "publishedAt": "2026-08-12T17:36:37.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087594117262131357.mp4",
    "posterUrl": "/posters/x/2087594117262131357.jpg",
    "duration": 27,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame; sourceCaption only supplies model attribution and no content details"
  },
  {
    "id": "x-2087588004454138187",
    "title": "MotionContext连续性实验",
    "titleEn": "A MotionContext Continuity Experiment",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案只确认采用 MotionContext 进行连续性生成，没有说明场景或主体，因此标题仅保留实验方法。 成片规格为 12 秒 · 992×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The caption only confirms a MotionContext continuity run and gives no subject or setting, so the title is limited to the experimental method. Output: 12s · 992×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jun_kongo/status/2087588004454138187",
    "sourceLabel": "X 原帖 · @jun_kongo",
    "author": "kongo jun / 混合 順",
    "publishedAt": "2026-08-12T17:12:20.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087588004454138187.mp4",
    "posterUrl": "/posters/x/2087588004454138187.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "992×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 MotinoContext",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 256,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit MotionContext method only"
  },
  {
    "id": "x-2087587359693144464",
    "title": "三只女仆装萌兽并排合掌",
    "titleEn": "Three Maid-Costumed Creatures Pose with Clasped Hands",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三只拟人化小动物穿着黑白女仆装并排站立，双手在胸前合拢，背景布满粉色爱心与蕾丝装饰。 成片规格为 5 秒 · 736×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three anthropomorphic creatures in black-and-white maid outfits stand side by side with clasped hands against a backdrop of pink hearts and lace. Output: 5s · 736×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ippi_/status/2087587359693144464",
    "sourceLabel": "X 原帖 · @ippi_",
    "author": "いっぴ",
    "publishedAt": "2026-08-12T17:09:46.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087587359693144464.mp4",
    "posterUrl": "/posters/x/2087587359693144464.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "736×576",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3てすと",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 162,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087586650331508774",
    "title": "犬猫叫声的三模型对照",
    "titleEn": "Dog and Cat Vocalizations Across Three Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "Seedance 2.0、Seedance 2.5 与 H3 采用同一故事但分别优化提示词，重点比较犬猫叫声；其中 Seedance 2.0 的猫声在重试后仍异常。 成片规格为 50 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Seedance 2.0, Seedance 2.5, and H3 use the same story with model-specific prompts to compare dog and cat vocalizations; Seedance 2.0's cat result remains anomalous after retries. Output: 50s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/333mic999/status/2087586650331508774",
    "sourceLabel": "X 原帖 · @333mic999",
    "author": "mic",
    "publishedAt": "2026-08-12T17:06:57.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087586650331508774.mp4",
    "posterUrl": "/posters/x/2087586650331508774.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#OiiOii @OiiOii_AI @OiiOii_AI_JP \n\nSeedance 2.0 vs Seedance 2.5 vs MiniMax H3\n\n試したのは犬と猫の鳴き声。\n同一ストーリーをそれぞれに最適なプロンプトにして生成しました。\n\nSD2.0ねこ…なぜでしょう。\nリトライしたけどほぼ同じ。\n本当になぜなんでしょう。\n\n何してんだ、寝なくては。\n\n[PR]",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "views": 142,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit models, shared story, prompt optimization, animal-vocalization subject, and retry result"
  },
  {
    "id": "x-2087585654977003740",
    "title": "单张VTuber立绘的表情与挥手动画",
    "titleEn": "Expression and Waving from a Single VTuber Illustration",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "单张 VTuber 立绘被用于图生动态，在维持角色设计的同时生成表情、身体摆动与挥手动作，并探索对 Live2D 制作的用途。 成片规格为 5 秒 · 768×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A single VTuber illustration drives image animation that preserves the character design while adding expressions, body sway, and waving, with possible use in Live2D production. Output: 5s · 768×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Hachie_Lab/status/2087585654977003740",
    "sourceLabel": "X 原帖 · @Hachie_Lab",
    "author": "蜂衣研究所 - Hachie Laboratory -",
    "publishedAt": "2026-08-12T17:03:00.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087585654977003740.mp4",
    "posterUrl": "/posters/x/2087585654977003740.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1088",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🧪研究報告\n\nMiniMax H3で、1枚のVTuber立ち絵（蜂衣白羽）からI2Vを検証中。キャラデザを維持しつつ、表情・身体の揺れ・手振りまで生成できました👀\nLive2D制作にも応用できそう…？🐝🥼\n#蜂衣研究所",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 367,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit source image, character, preserved design, generated motions, and Live2D application"
  },
  {
    "id": "x-2087579250010509487",
    "title": "白石杏与AMG SL63",
    "titleEn": "An Shiraishi and an AMG SL63",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确将《世界计划》角色白石杏与 AMG SL63 跑车作为画面主体，但未描述进一步动作。 成片规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly names Project Sekai character An Shiraishi and an AMG SL63 as the subjects, without describing a further action. Output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Oxide39/status/2087579250010509487",
    "sourceLabel": "X 原帖 · @Oxide39",
    "author": "ミクoxide☆NEO彡",
    "publishedAt": "2026-08-12T16:37:33.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087579250010509487.mp4",
    "posterUrl": "/posters/x/2087579250010509487.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3 白石杏 \nAMG SL63\n#白石杏\n#プロセカ\n#AMG",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 114,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit character and vehicle subjects"
  },
  {
    "id": "x-2087578526677619058",
    "title": "RTX 3060生成后的放大与文字修整",
    "titleEn": "Upscaling and Text Cleanup After an RTX 3060 Run",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "基础内容运行于十六GB内存和 RTX 3060 十二GB显存，随后由 GPT-5.6 Sol 负责提高分辨率和修整文字，因此当前成片并非原始直出。 成片规格为 5 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The base run uses 16GB of RAM and a 12GB RTX 3060, followed by GPT-5.6 Sol for upscaling and text cleanup, so the attached result is not an untouched output. Output: 5s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Teto_66_/status/2087578526677619058",
    "sourceLabel": "X 原帖 · @Teto_66_",
    "author": "Teto66",
    "publishedAt": "2026-08-12T16:34:40.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087578526677619058.mp4",
    "posterUrl": "/posters/x/2087578526677619058.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "RAM 16GB、RTX 3060 12GBでのMiniMax H3のテスト\n\n高解像度化や文字を整えるなどの編集はGPT-5.6 Solにやってもらったので、生の動画ではないです",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 111,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit hardware, post-processing model, edits, and non-raw-output disclosure"
  },
  {
    "id": "x-2087577333867893033",
    "title": "三十秒无剪辑生成的质量成本对照",
    "titleEn": "Quality and Cost in a Thirty-Second Uncut Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "Seedance 2.5 与 H3 使用同一提示词一次生成三十秒且不剪辑，比较清晰度、整体质量、运行位置和边际成本。 成片规格为 31 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Seedance 2.5 and H3 use the same prompt for single thirty-second generations with no cuts, comparing sharpness, overall quality, execution model, and marginal cost. Output: 31s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BuildFastWithAI/status/2087577333867893033",
    "sourceLabel": "X 原帖 · @BuildFastWithAI",
    "author": "Build Fast with AI",
    "publishedAt": "2026-08-12T16:29:56.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087577333867893033.mp4",
    "posterUrl": "/posters/x/2087577333867893033.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.5  vs MiniMax H3, same prompt, 30s single gen, no cuts. \n\nPeople will look at this and say Seedance is sharper. It is. But sharpness isn't what separates them:\n\n- H3 ran locally. 20 steps, 0.7MP. On mid-weight gaming hardware.\n- on official API pricing, Seedance 2.5 is 3-4x more expensive depending on mode. And the other one runs on hardware I already own, with no per-second meter running.\n\nAt ~75% of frontier quality for 0% of the marginal cost, the remaining 25% has to justify a 4x bill. For most work it can't.\n\nWhat do you think?",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 153,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit same prompt, models, duration, no-cut constraint, quality, and cost comparison"
  },
  {
    "id": "x-2087573967846711691",
    "title": "夕阳阶梯上提着购物袋的眼镜角色",
    "titleEn": "A Glasses-Wearing Character Holds a Shopping Bag on Sunset Steps",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴圆框眼镜的短发动画角色站在俯瞰城市的夕阳阶梯上，双手提着购物袋并轻微改变头手位置。 成片规格为 15 秒 · 832×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired animated character in round glasses stands on sunset steps above a city, holding a shopping bag while making small head and hand movements. Output: 15s · 832×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2087573967846711691",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-12T16:16:33.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087573967846711691.mp4",
    "posterUrl": "/posters/x/2087573967846711691.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "階段を登って降りるテスト #AI眼鏡部 #AI田中さん\nMiniMax H3 ローカル",
    "engagement": {
      "replies": 1,
      "reposts": 5,
      "likes": 68,
      "views": 991,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087572931945935192",
    "title": "官方文档驱动的产品宣传广告",
    "titleEn": "A Product Promotion Driven by Official Documentation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案把当前内容标记为依照官方文档进行的产品宣传广告实验，其余文字主要是 RunningHub 注册与邀请码推广。 成片规格为 168 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The caption labels the attached piece as a product-promotion experiment based on official documentation; the remaining text is largely a RunningHub registration and referral pitch. Output: 168s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/elonmuskProPlus/status/2087572931945935192",
    "sourceLabel": "X 原帖 · @elonmuskProPlus",
    "author": "Moon",
    "publishedAt": "2026-08-12T16:12:26.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087572931945935192.mp4",
    "posterUrl": "/posters/x/2087572931945935192.jpg",
    "duration": 168,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "stylized"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Product Promotional Ad Official Document Test\nAccelerated Links: https://www.runninghub.ai/post/2086597344037384194/?inviteCode=rh-v1537\nGet 1,000 free points upon registration + 100 bonus points for daily logins!please click \"Bind Activation Code\" in the top right corner and enter the invite code: rh-v1537.)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 65,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit product-promotion and official-document experiment framing"
  },
  {
    "id": "x-2087572774479098221",
    "title": "运球、变向与投篮的三模型对照",
    "titleEn": "Dribble, Crossover, and Shot Across Three Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "H3、Wan 2.2 与 LTX 2.3 使用相同提示词和设置执行“运球、变向、投篮”连续动作，并比较二十秒质量、速度、控制能力和音频支持。 成片规格为 68 秒 · 2216×3740。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "H3, Wan 2.2, and LTX 2.3 use the same prompt and setup for a dribble-to-crossover-to-shot sequence, comparing twenty-second quality, speed, controls, and audio support. Output: 68s · 2216×3740. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FloyoAI/status/2087572774479098221",
    "sourceLabel": "X 原帖 · @FloyoAI",
    "author": "FloyoAI",
    "publishedAt": "2026-08-12T16:11:49.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087572774479098221.mp4",
    "posterUrl": "/posters/x/2087572774479098221.jpg",
    "duration": 68,
    "aspectRatio": "portrait",
    "resolution": "2216×3740",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "We put MiniMax H3, Wan 2.2, and LTX 2.3 side by side again. Same prompts, same setup.\n\nH3 still wins on raw quality. Better motion, tighter prompt adherence, and it actually held up at 20 seconds. Asked for \"dribble → crossover → shoot\" and it followed the full sequence. The tradeoff: it's slow.\n\nLTX 2.3 is almost 5x faster and gives you LoRA, pose, depth, camera control, retakes, and shot extension. Some background drift, but at that speed you just run it again.\n\nWan 2.2 lands in the middle. Stable, predictable, solid without the wait.\n\nNo single winner. Depends on what you're making.\n\n→ Best quality: H3\n→ Fastest + most control: LTX 2.3\n→ Solid all-rounder: Wan 2.2\n\nOne note: Wan didn't have audio in our setup. H3 and LTX generated audio natively.\n\nAll three run on Floyo.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 6,
      "views": 688,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit models, controlled inputs, basketball action sequence, duration, speed, controls, and audio differences"
  },
  {
    "id": "x-2087570082948100419",
    "title": "骑鳟鱼如巨龙的丹妮莉丝",
    "titleEn": "Daenerys Rides a Trout Like a Dragon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这组精选内容的第一项把丹妮莉丝放在一条鳟鱼背上，让她像骑龙一样驾驭这条鱼。 成片规格为 20 秒 · 720×950。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The first item in a curated set places Daenerys on the back of a trout, riding the fish as though it were a dragon. Output: 20s · 720×950. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IATheYoker/status/2087570082948100419",
    "sourceLabel": "X 原帖 · @IATheYoker",
    "author": "YokerAI",
    "publishedAt": "2026-08-12T16:01:07.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087570082948100419.mp4",
    "posterUrl": "/posters/x/2087570082948100419.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "720×950",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hace solo una semana que lanzaron MiniMax H3.\n\nY ya está haciendo cosas que, sinceramente, no deberían ser posibles.\n\nHe recopilado los 13 ejemplos más bestias que he encontrado.\n\nEl 11 cuesta creer que sea IA 👇\n\n1/ Daenerys montando una trucha como si fuese un dragón. AQUARIS!",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 9,
      "views": 2704,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit character, trout subject, riding action, and dragon analogy"
  },
  {
    "id": "x-2087570001524093382",
    "title": "Hyperframes中的三种新模型",
    "titleEn": "Three New Models in Hyperframes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案列出 Flux 3、H3 与 Seedance 2.5 三种近期发布的模型，并说明当前内容通过 Hyperframes 制作。 成片规格为 71 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption lists the recently released Flux 3, H3, and Seedance 2.5 and states that the attached piece was created with Hyperframes. Output: 71s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2087570001524093382",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-08-12T16:00:48.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087570001524093382.mp4",
    "posterUrl": "/posters/x/2087570001524093382.jpg",
    "duration": 71,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Three new video models in the last 2 weeks:\n\n→ Flux 3 • @bfl_ai\n→ MiniMax H3 • @Hailuo_AI\n→ Seedance 2.5 • @BytePlusGlobal\n\nCreated with Hyperframes.",
    "engagement": {
      "replies": 5,
      "reposts": 10,
      "likes": 37,
      "views": 2508,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit three-model list and Hyperframes production route"
  },
  {
    "id": "x-2087569828479410230",
    "title": "A面",
    "titleEn": "Side A",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将作品命名为《A面》，并列出 Seedream 5 Pro、ElevenLabs、RTX Pro 6000 与本地生成等制作要素。 成片规格为 130 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption names the short Side A and lists Seedream 5 Pro, ElevenLabs, an RTX Pro 6000, and local generation among its production elements. Output: 130s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2087569828479410230",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-12T16:00:06.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087569828479410230.mp4",
    "posterUrl": "/posters/x/2087569828479410230.jpg",
    "duration": 130,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3      Seedream 5 Pro      \nElevenLabs     \nNvidia RTX Pro 6000 96GB      \nLocal AI          \n\n„Side A | A Short Film”\n\n@MiniMax_AI\n@NVIDIAAI\n@Alibaba_Qwen\n@ElevenLabs\n\nhttp://youtube.com/@cinesit",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 104,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit work title and production stack"
  },
  {
    "id": "x-2087566277820784649",
    "title": "电视剧截图驱动的图生动态",
    "titleEn": "Image Animation from a TV-Show Screenshot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一张电视剧截图被作为输入，用于观察单帧画面转为动态内容后的表现。 成片规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A television-show screenshot is used as the input for observing how a single frame is extended into motion. Output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIReelofficial/status/2087566277820784649",
    "sourceLabel": "X 原帖 · @AIReelofficial",
    "author": "AIReel",
    "publishedAt": "2026-08-12T15:46:00.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087566277820784649.mp4",
    "posterUrl": "/posters/x/2087566277820784649.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tested Minimax H3 with a TV show screenshot. \nThis is actually crazy.\nMinimax H3 is live on AIReel!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 69,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit television screenshot input and generation trial"
  },
  {
    "id": "x-2087565327433359563",
    "title": "海龙大战巨型章鱼的延展片段",
    "titleEn": "An Extended Sea-Dragon and Giant-Octopus Battle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "本地生成内容围绕海龙与巨型章鱼的对战展开，并在原有段落基础上继续延展。 成片规格为 120 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A locally generated sea dragon battles a giant octopus in a sequence extended beyond its original segment. Output: 120s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Arjannnl/status/2087565327433359563",
    "sourceLabel": "X 原帖 · @Arjannnl",
    "author": "TheEasyWin",
    "publishedAt": "2026-08-12T15:42:13.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087565327433359563.mp4",
    "posterUrl": "/posters/x/2087565327433359563.jpg",
    "duration": 120,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 local sea dragon vs giant octopus extended  \n\nHuge thanks to @deepbeepmeep for creating WAN2GP and @MiniMax_AI for creating an amazing model.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 163,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit sea dragon, giant octopus, battle framing, local execution, and extension"
  },
  {
    "id": "x-2087564352551604263",
    "title": "赞达亚化身毒液",
    "titleEn": "Zendaya Transforms into Venom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "赞达亚的图像被用于本地图生动态，在 RTX 4090 上呈现她转化为毒液的设定。 成片规格为 19 秒 · 960×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An image of Zendaya drives a local image-to-motion run on an RTX 4090 in which she transforms into Venom. Output: 19s · 960×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sadlemonjuice/status/2087564352551604263",
    "sourceLabel": "X 原帖 · @sadlemonjuice",
    "author": "Socratech",
    "publishedAt": "2026-08-12T15:38:21.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087564352551604263.mp4",
    "posterUrl": "/posters/x/2087564352551604263.jpg",
    "duration": 19,
    "aspectRatio": "square",
    "resolution": "960×960",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Zendaya = Venom MiniMax h3 local \n\ngpu4090  image to video:",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 104,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit public figure, character transformation, local execution, hardware, and input mode"
  },
  {
    "id": "x-2087563810144154006",
    "title": "LTX 2.5与H3图生动态对照",
    "titleEn": "LTX 2.5 and H3 Image-to-Motion Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "公开文案只确认在 ComfyUI 中对照 LTX 2.5 与 H3 的图生动态结果，没有说明参考图或提示词内容。 成片规格为 10 秒 · 3840×2316。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption confirms a ComfyUI image-to-motion comparison between LTX 2.5 and H3 but does not identify the reference image or prompt content. Output: 10s · 3840×2316. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/toyxyz3/status/2087563810144154006",
    "sourceLabel": "X 原帖 · @toyxyz3",
    "author": "toyxyz",
    "publishedAt": "2026-08-12T15:36:12.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087563810144154006.mp4",
    "posterUrl": "/posters/x/2087563810144154006.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "3840×2316",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5 vs Minimax-H3 I2V test #comfyui",
    "engagement": {
      "replies": 7,
      "reposts": 7,
      "likes": 80,
      "views": 12134,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit models, input mode, comparison, and ComfyUI context"
  },
  {
    "id": "x-2087563166654067099",
    "title": "动画场景MG动态排版的同输入对照",
    "titleEn": "Same-Input Comparison of MG Motion Graphics in Anime Scenes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "H3 与 Seedance 2.5 使用完全相同的提示词和参考输入生成带 MG 动态排版的动画场景，公开文案认为前者整体表现更强。 成片规格为 32 秒 · 1080×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "H3 and Seedance 2.5 use the exact same prompts and reference inputs for anime scenes with MG motion graphics, with the caption judging H3 stronger overall. Output: 32s · 1080×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JZCreative2012/status/2087563166654067099",
    "sourceLabel": "X 原帖 · @JZCreative2012",
    "author": "JZ多媒体解决方案",
    "publishedAt": "2026-08-12T15:33:38.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087563166654067099.mp4",
    "posterUrl": "/posters/x/2087563166654067099.jpg",
    "duration": 32,
    "aspectRatio": "portrait",
    "resolution": "1080×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Using the exact same prompts and reference inputs, Minimax H3 outperforms Seedance 2.5 by a large margin on anime scenes with MG motion graphics.\n\n#minimaxh3 #seedance25",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 76,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit models, identical prompts and references, anime subject, motion-graphics treatment, and stated result"
  },
  {
    "id": "x-2087558301756301354",
    "title": "四人舞团的同步编舞",
    "titleEn": "Synchronized Choreography for a Four-Person Dance Crew",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "四名舞者以同一节奏完成同步编舞，公开文案强调角色、服装与动作在整段表演中的连续一致。 成片规格为 15 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Four dancers perform synchronized choreography to one rhythm, with the caption emphasizing consistency in characters, outfits, and movement throughout. Output: 15s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lucasyao99261/status/2087558301756301354",
    "sourceLabel": "X 原帖 · @lucasyao99261",
    "author": "Lucasyao",
    "publishedAt": "2026-08-12T15:14:18.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087558301756301354.mp4",
    "posterUrl": "/posters/x/2087558301756301354.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI Created a Perfectly Synchronized Dance Crew 👯‍♀️🔥\n\nFour dancers.\nOne rhythm.\nZero mistakes.\nA fully AI-generated dance performance with consistent characters, outfits, and choreography.\nCreated with MiniMax-H3.\n\n#AI #AIArt #AIVideo #GenerativeAI #Dance",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 52,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit dancer count, synchronization, choreography, and consistency claims"
  },
  {
    "id": "x-2087557950017778139",
    "title": "戴透明眼镜的三眼面孔特写",
    "titleEn": "Extreme Close-Up of a Three-Eyed Face in Clear Glasses",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头持续贴近一张戴透明镜框的面孔，额头上的眼睛与镜片后的两只眼睛分别改变视线方向。 成片规格为 7 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An extreme close-up holds on a face in clear-framed glasses while the forehead eye and the two eyes behind the lenses shift direction. Output: 7s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lucasyao99261/status/2087557950017778139",
    "sourceLabel": "X 原帖 · @lucasyao99261",
    "author": "Lucasyao",
    "publishedAt": "2026-08-12T15:12:54.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087557950017778139.mp4",
    "posterUrl": "/posters/x/2087557950017778139.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "POV: Your Third Eye Opens… But It Has Its Own Problems 👁️\n\nBro unlocked the third eye update and the physics engine crashed 💀\n\nAI-generated surreal comedy animation.\n\nCreated with MiniMax-H3.\n\n#AI #AIArt #Surrealism #AIMemes #Animation",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 42,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087557406473744443",
    "title": "空旷长廊尽头的黄色小球",
    "titleEn": "A Small Yellow Ball at the End of an Empty Hallway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "木地板长廊两侧排列着深色门框和墙灯，一只黄绿色小球停在远处中央；原帖说明先用较低分辨率生成，再进行放大。 成片规格为 10 秒 · 850×1478。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Dark doorframes and wall lights line a wooden hallway, with a small yellow-green ball resting at the far end; the source describes starting at a lower resolution and then upscaling. Output: 10s · 850×1478. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/To_Vten_ozi/status/2087557406473744443",
    "sourceLabel": "X 原帖 · @To_Vten_ozi",
    "author": "ゆーかり(。-`ω-)🏹👿",
    "publishedAt": "2026-08-12T15:10:45.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087557406473744443.mp4",
    "posterUrl": "/posters/x/2087557406473744443.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "850×1478",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3は0.4～0.6などサイズ小さく生成して、SeedVR2でアップスケールするのが良さそう。\nアップスケールの方が時間かかったけどね😆",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 57,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087557367437406423",
    "title": "便利店男子与左下参考头像",
    "titleEn": "A Man in a Convenience Store with an Inset Reference Portrait",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名蓄胡男子站在灯光明亮的便利店饮料柜前，左下角嵌有一张男性头像参考图；原帖注明创作使用了五张参考图。 成片规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bearded man stands before brightly lit drink coolers in a convenience store, with a male reference portrait inset at lower left; the source states that five reference images were used. Output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ComfyUI/status/2087557367437406423",
    "sourceLabel": "X 原帖 · @ComfyUI",
    "author": "ComfyUI",
    "publishedAt": "2026-08-12T15:10:35.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087557367437406423.mp4",
    "posterUrl": "/posters/x/2087557367437406423.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Music video test in MiniMax H3. Five generated reference stills, one subject, four locations,j plus an audio track, all in a single context. It returned the location cuts already in the clip.\n\nTo try this workflow, comment 'Comfy' and we'll send it to you! 🔗",
    "engagement": {
      "replies": 88,
      "reposts": 9,
      "likes": 179,
      "views": 11230,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087554903480598566",
    "title": "从素描线到笔触与颜料层的插画成形",
    "titleEn": "An Illustration Forms from Sketch Lines, Brush Strokes, and Paint Layers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "插画在眼前逐步形成，素描线、笔触与颜料层依次被赋予动态。 成片规格为 15 秒 · 1920×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustration takes shape before the viewer as sketch lines, brush strokes, and paint layers are animated in sequence. Output: 15s · 1920×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CharaspowerAI/status/2087554903480598566",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-08-12T15:00:48.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087554903480598566.mp4",
    "posterUrl": "/posters/x/2087554903480598566.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1920×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I wanted the illustration to feel like it was being created right in front of your eyes.\nSketch lines.\nBrush strokes.\nPaint layers.\n\nAll animated with Minimax H3.\nPROMPT BELOW👇",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 19,
      "views": 1091,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit illustration-creation concept and animated sketch-line, brush-stroke, and paint-layer sequence"
  },
  {
    "id": "x-2087554680637571298",
    "title": "雪林古装男女与节点工作流界面",
    "titleEn": "A Costumed Pair in a Snowy Forest Above a Node Workflow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "画面上半部是一对古装男女走在雪林中，下半部则是包含参考缩略图和日志的节点工作流界面；原帖说明使用了多张参考图和分镜图。 成片规格为 15 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The upper half shows a costumed pair walking through a snowy forest, while the lower half shows a node workflow with reference thumbnails and logs; the source describes using multiple reference images and a storyboard. Output: 15s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DongJia38177/status/2087554680637571298",
    "sourceLabel": "X 原帖 · @DongJia38177",
    "author": "Jia Dong",
    "publishedAt": "2026-08-12T14:59:55.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087554680637571298.mp4",
    "posterUrl": "/posters/x/2087554680637571298.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Running Hailuo H3 (MiniMax H3) reference-to-video on a cloud PC is just too good!\nFired up ComfyUI with the MiniMax H3 Reference to Video node — multi-image references + detailed storyboard prompts, character consistency maxed out, movements super natural. Cloud PC with 4090/A100 handles it effortlessly, VRAM fully loaded and still butter-smooth… way better than local!\nAlready generated several silky-smooth character videos, check the results yourself ↓
#HailuoH3 #MiniMaxH3 #CloudPC #ComfyUI #AIVideoGeneration\nIf you haven’t tried running Hailuo on a cloud PC yet, what are you waiting for?!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 98,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087552150918537399",
    "title": "百事可乐广告",
    "titleEn": "A Pepsi Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖明确说明随帖内容是一支为百事可乐制作的广告。 成片规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly identifies the attached work as a commercial made for Pepsi. Output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hetzerLin/status/2087552150918537399",
    "sourceLabel": "X 原帖 · @hetzerLin",
    "author": "Linx_OKX | 我爱Misa",
    "publishedAt": "2026-08-12T14:49:52.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087552150918537399.mp4",
    "posterUrl": "/posters/x/2087552150918537399.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "用Minimax H3 做了一个百事可乐的广告\n\n@KingCola88 可乐老师感觉如何",
    "engagement": {
      "replies": 5,
      "reposts": 0,
      "likes": 5,
      "views": 301,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit brand and advertising format"
  },
  {
    "id": "x-2087551813620994416",
    "title": "几何灯墙前的黑衣男子近景",
    "titleEn": "Close-Up of a Man in Black Before Geometric Lights",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名短发男子身穿高领黑衣，佩戴十字吊坠耳饰，站在由白色几何灯带构成的墙面前。 成片规格为 91 秒 · 2858×1608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired man in a high-collared black outfit wears a dangling cross earring before a wall of white geometric light strips. Output: 91s · 2858×1608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Nick181cm/status/2087551813620994416",
    "sourceLabel": "X 原帖 · @Nick181cm",
    "author": "Nick / Dong Yu Fan",
    "publishedAt": "2026-08-12T14:48:31.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087551813620994416.mp4",
    "posterUrl": "/posters/x/2087551813620994416.jpg",
    "duration": 91,
    "aspectRatio": "landscape",
    "resolution": "2858×1608",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance2.5 + Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 49,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087551759237382603",
    "title": "湿发女子近景的三格纵向对照",
    "titleEn": "Three Stacked Close-Ups of a Wet-Haired Woman",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三格纵向画面都呈现一名湿发女子靠近窗边的面部近景，前景有另一人的模糊轮廓，各格的头部角度和视线略有不同。 成片规格为 12 秒 · 1921×3240。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three stacked panels show close-ups of a wet-haired woman near a window with another figure blurred in the foreground, varying slightly in head angle and gaze. Output: 12s · 1921×3240. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/oedufigueiredo/status/2087551759237382603",
    "sourceLabel": "X 原帖 · @oedufigueiredo",
    "author": "Edu 👑",
    "publishedAt": "2026-08-12T14:48:18.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087551759237382603.mp4",
    "posterUrl": "/posters/x/2087551759237382603.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "1921×3240",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.5. Wan 3.0. e MiniMax H3. Três dos modelos de vídeo IA mais hypados no momento. Teste de cena: emoção + atuação.  \n\nQuem entrega as emoções mais reais?😵",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 65,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087551628706771285",
    "title": "粉色房间里的玩偶转场到冰上滑行",
    "titleEn": "A Display Doll Moves from a Pink Room to an Ice Rink",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一个站在粉色房间底座上的玩偶随后出现在冰场，在观众席前滑行、转身并抬起手臂。 成片规格为 5 秒 · 416×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A doll first stands on a display base in a pink room and then appears on an ice rink, where it glides, turns, and raises one arm before an audience. Output: 5s · 416×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/outofbase/status/2087551628706771285",
    "sourceLabel": "X 原帖 · @outofbase",
    "author": "がるる Out of Base",
    "publishedAt": "2026-08-12T14:47:47.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087551628706771285.mp4",
    "posterUrl": "/posters/x/2087551628706771285.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "416×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3試した\n\n色々調整必要だけど無制限でローカルで動かせるのは楽しいね👍\n\nRTX5080でもキッツいのが現実…\n気持ちよく動かすには今の御時世ヤバイ…1年前に買い揃えておけば😭",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 18,
      "views": 835,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087551301962093010",
    "title": "空教室里的双马尾书包少女",
    "titleEn": "A Twin-Tailed Girl with a Backpack in an Empty Classroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名黑色双马尾的动漫少女背着红色书包，站在摆有课桌、储物柜和黑板的空教室中央。 成片规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired anime girl with twin tails and a red backpack stands at the center of an empty classroom with desks, lockers, and a blackboard. Output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C12592947592/status/2087551301962093010",
    "sourceLabel": "X 原帖 · @C12592947592",
    "author": "supercub3",
    "publishedAt": "2026-08-12T14:46:29.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087551301962093010.mp4",
    "posterUrl": "/posters/x/2087551301962093010.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "朗報、sora2で作ったあやねちゃんの声、MiniMax H3にて無事に引き継ぐことに成功！これはかなり嬉しい🥰\n1個目がH3で作った動画　２個目がsora2のあやねちゃんの声",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 20,
      "views": 1459,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087550642982367461",
    "title": "敞开机舱门边的蓝衣人物",
    "titleEn": "A Blue-Suited Figure Beside an Open Aircraft Door",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "画面裁出一名身穿亮蓝色连身服、系着黑色安全带并戴着手套的人物，旁边的机舱门敞开，外侧可见天空与云层。 成片规格为 250 秒 · 1024×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The frame crops a figure in a bright blue jumpsuit, black harness, and gloves beside an open aircraft door with sky and clouds visible outside. Output: 250s · 1024×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AiXsatoshi/status/2087550642982367461",
    "sourceLabel": "X 原帖 · @AiXsatoshi",
    "author": "AI✖️Satoshi⏩️",
    "publishedAt": "2026-08-12T14:43:52.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087550642982367461.mp4",
    "posterUrl": "/posters/x/2087550642982367461.jpg",
    "duration": 250,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "本日のMV進捗です。途中むちゃくちゃ（笑）\nMiniMax-H3 R2V（音声、キャラシート参照）\n明日以降なおします",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 329,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087549871285039347",
    "title": "莉莉姆的五秒舞蹈动作",
    "titleEn": "Ririmu's Five-Second Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "以莉莉姆舞蹈为主题生成五秒片段，同一指令得到的多个版本动作表现相近。 成片规格为 5 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A five-second sequence centers on Ririmu dancing, with multiple generations from the same prompt producing similar movement. Output: 5s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/zeroban_noriba/status/2087549871285039347",
    "sourceLabel": "X 原帖 · @zeroban_noriba",
    "author": "えきちょー（0番のりば）",
    "publishedAt": "2026-08-12T14:40:48.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087549871285039347.mp4",
    "posterUrl": "/posters/x/2087549871285039347.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "dance",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "これがウワサの MiniMax H3 の実力か…！\nRTX4070SUPER で5秒動画が1分30秒～2分で作れた！\n生成テストはi2vでりりむダンス😈💕\n全部同じプロンプト、どれも似た感じに。\n\n良くも悪くもプロンプトに忠実！\nズボラなランダム生成は難しいのかな🤔\n\n色々試したいけどお盆あまり時間取れなさそう🫠",
    "engagement": {
      "replies": 5,
      "reposts": 30,
      "likes": 88,
      "views": 1589,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit named subject, dance action, five-second duration, and same-prompt variants"
  },
  {
    "id": "x-2087548954351169642",
    "title": "紫色地牢里的女子、精灵与宝石护符",
    "titleEn": "A Woman, a Violet Elf, and a Jeweled Amulet in a Purple Dungeon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "短发女子与紫肤精灵在布满裂纹石墙、紫色符文和魔法微粒的地牢中交谈，女子低头看向宝石护符后与精灵相视而笑。 成片规格为 5 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired woman and a violet-skinned elf speak in a dungeon of cracked stone, purple runes, and magical particles; she looks down at her jeweled amulet before they exchange an amused glance. Output: 5s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mauno_k/status/2087548954351169642",
    "sourceLabel": "X 原帖 · @mauno_k",
    "author": "Mauno_k",
    "publishedAt": "2026-08-12T14:37:10.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087548954351169642.mp4",
    "posterUrl": "/posters/x/2087548954351169642.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing the MiniMax H3 image to Video using first frame image as reference as shown by @NullpoLab - it works nicely. In general I have noticed that for my memory budget (RTX3090 & 32Gb ram)as long as the video length is moderate (around 5 seconds) raisinge inference steps  up to 50 or even more make no big difference in calculation time - its quite linear, but once the clip get longer time taken gets more exponential. Prompt:\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\nintegrated_multimodal_description: [Shot 1] The woman and violet elf man from <Picture 1> are fully referenced while strictly preserving every identity detail, face, hair, body proportions, clothing, armor, accessories and held items: the short wavy brown-haired woman in red-black plaid flannel over white tank top with ornate gold cross amulet studded with colorful gems, perfect white teeth, and the purple-skinned elf man with long flowing purple hair, matching beard, pointed ears, golden circlet with pink gems, and ornate purple-gold armored robes with large glowing pink gemstones.  [Shot 2] At 00:00.100,They stand close together inside a stylized MMO/RPG dungeon (cracked stone walls, glowing purple runes, floating magical particles, dim torchlight casting dramatic highlights). Starting from a tight medium two-shot of both facing each other. In a snappy take the elf leans slightly forward, speaks with a serious tone “That amulet looks powerful… it could protect us here.” [Shot 3] At 00:03.000 Over the shoulder shot showing elf from behind, woman facing camera, the woman glances down at her necklace then looks up with a casual smirk “It’s just a pretty trinket from Temu.”, both exchange a quick amused look. Natural micro-movements (breathing, slight hair sway, fabric shift, soft glow pulse on the gems) keep the shot alive. End on a three-quarter view of the pair still facing each other.\noverall_soundscape: Atmospheric dungeon ambience with low echoing drips, distant magical hum, and soft crackling torchlight. Clear natural dialogue: elf’s deep resonant voice, woman’s light casual reply, slight room reverb matching the stone chamber.\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 127,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit characters, amulet, dungeon setting, shot progression, and actions"
  },
  {
    "id": "x-2087548138727104861",
    "title": "瀑布徒步途中的林间休息",
    "titleEn": "A Forest Rest Stop During a Waterfall Hike",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "年轻女子抵达林间瀑布后感受水雾、坐上苔藓岩石喝水，并在观看雾气时擦去额头汗水。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A young woman reaches a forest waterfall, feels the spray, sits on a mossy rock to drink water, and wipes sweat from her forehead while watching the mist. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2087548138727104861",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-12T14:33:55.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087548138727104861.mp4",
    "posterUrl": "/posters/x/2087548138727104861.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Waterfall Hike Rest Stop \n\nMade with MiniMax H3 on @Hailuo_AI \n\nPrompt - Main Subject:\nYoung Korean woman, early 20s, sporty hiking outfit, cap over a ponytail, small backpack, water bottle in hand, refreshed cheerful expression.\nLocation:\nForest trail leading to a small waterfall, mid-morning. Mossy rocks, mist rising from the falling water, tall trees, dappled sunlight. No other hikers visible.\nVisual Style:\nUltra-realistic documentary realism, refreshing outdoor feeling, bright natural forest light.\nCamera Style:\nEarly 2010s digital camcorder, handheld with natural shake, autofocus hunting in shifting light and mist, soft grain, occasional lens flare. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She reaches the waterfall, then stops to take in the view, breathing heavily.\n00:03–00:06 → She dips her hand into the cool spray, then says to camera \"이거 진짜 시원하다\" (\"This feels really refreshing\").\n00:06–00:09 → She sits on a mossy rock, sipping water from her bottle.\n00:09–00:12 → She watches the mist drift, wiping sweat from her forehead.\n00:12–00:15 → She looks at camera saying \"여기까지 온 보람 있다\" (\"It was worth hiking all the way here\"), smiling as it fades.\nAudio: Waterfall rushing, birds chirping, faint wind, water dripping on rocks. Her dialogue as noted above. No music.\nGoal: A refreshing, rewarding hiking rest moment energetic, warm, genuine.",
    "engagement": {
      "replies": 9,
      "reposts": 1,
      "likes": 20,
      "views": 796,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit subject, forest-waterfall setting, actions, timeline, and documentary treatment"
  },
  {
    "id": "x-2087547082408149500",
    "title": "扁平漫画城市中迎面驶来的汽车",
    "titleEn": "A Car Approaches Through a Flat-Graphic City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一辆浅色小汽车沿倾斜道路迎面驶来，两侧是由橄榄绿和橙色块面组成的扁平漫画城市建筑。 成片规格为 95 秒 · 1854×848。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A light-colored compact car approaches along a tilted road between city buildings rendered in flat olive and orange graphic shapes. Output: 95s · 1854×848. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alfred909/status/2087547082408149500",
    "sourceLabel": "X 原帖 · @Alfred909",
    "author": "Keith Haus",
    "publishedAt": "2026-08-12T14:29:43.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087547082408149500.mp4",
    "posterUrl": "/posters/x/2087547082408149500.jpg",
    "duration": 95,
    "aspectRatio": "landscape",
    "resolution": "1854×848",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I took my new computer out for a spin to learn ComfyUI and video creation. All the assets are from Midjourney, but the actual video clips and music were made using my laptop. \n\nMiniMax H3\nFlux.2 Klein\nAce steps\n\nWhat do you think?",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 60,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087545555345920007",
    "title": "金色荒原上悬浮的巨型方柱群",
    "titleEn": "Giant Floating Monoliths Above a Golden Wasteland",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多根深色巨型方柱悬浮并倾斜在金色荒原上空，地面扬起尘雾，远处还有小型悬浮物；原帖称三秒片段的处理时间从约三十分钟缩短至三分钟。 成片规格为 21 秒 · 576×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Huge dark monoliths float at angles above a dusty golden wasteland with smaller objects in the distance; the source reports reducing processing for a three-second segment from about 30 minutes to three minutes. Output: 21s · 576×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AGIProphet/status/2087545555345920007",
    "sourceLabel": "X 原帖 · @AGIProphet",
    "author": "AGI Prophet",
    "publishedAt": "2026-08-12T14:23:39.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087545555345920007.mp4",
    "posterUrl": "/posters/x/2087545555345920007.jpg",
    "duration": 21,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Got MiniMax-H3 video gen running 10x faster locally, 3-sec clips went from 30 min to 3 min on an RTX 2060 mobile (6GB VRAM, 16GB RAM). Required patching comfy itself.\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 67,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087543255902925173",
    "title": "彩色海滨小镇里的卡通双人激战",
    "titleEn": "Two Cartoon Fighters Battle in a Colorful Seaside Town",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两个卡通角色在彩色海滨小镇和热带广场中奔跑、闪避、拳打脚踢，场景中伴随冲击闪光。 成片规格为 50 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two cartoon characters run, dodge, punch, and kick through a colorful seaside town and tropical plaza with visible impact flashes. Output: 50s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SD_Tutorial/status/2087543255902925173",
    "sourceLabel": "X 原帖 · @SD_Tutorial",
    "author": "Stable Diffusion Tutorials",
    "publishedAt": "2026-08-12T14:14:31.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087543255902925173.mp4",
    "posterUrl": "/posters/x/2087543255902925173.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5  vs  Minimax H3",
    "engagement": {
      "replies": 8,
      "reposts": 1,
      "likes": 61,
      "views": 6101,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087543096288612630",
    "title": "男子摆荡攀爬的十秒连续镜头对比",
    "titleEn": "Ten-Second Continuous-Shot Comparison of a Man Swinging and Climbing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一指令生成的十秒连续镜头比较一名男子摆荡与攀爬时的连贯性、光线、重量感、缆绳弹性和身体运动。 成片规格为 20 秒 · 1080×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A ten-second continuous shot generated from the same prompt compares a man swinging and climbing across continuity, lighting, weight, cable elasticity, and body movement. Output: 20s · 1080×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tim_AIProduct/status/2087543096288612630",
    "sourceLabel": "X 原帖 · @Tim_AIProduct",
    "author": "Tim_AI",
    "publishedAt": "2026-08-12T14:13:53.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087543096288612630.mp4",
    "posterUrl": "/posters/x/2087543096288612630.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "1080×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "surprising and unexpected test result 🙀\n\n🔥Seedance 2.5 vs MiniMax H3\n\nSame prompt.\nOne continuous 10s shot.\n\nSeedance 2.5 wins on cinematic flow, lighting, and overall polish. The continuous shot feels more seamless and film-like.\n\nMiniMax H3 edges it on physics and micro-details — better weight, cable elasticity, and natural body movement when he swings and climbs.Both are insanely close. \n\nWhich one do you prefer?\n\nFull prompt👇\n#Seedance25 #MiniMaxH3 @MiniMax_AI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 76,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit same-prompt comparison, duration, continuous-shot form, male subject, actions, and evaluation dimensions"
  },
  {
    "id": "x-2087541717889343787",
    "title": "火箭点火升离晴空发射台",
    "titleEn": "A Rocket Lifts Off from a Launchpad under Clear Skies",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一枚火箭在晴朗蓝天下升离发射台，底部喷出明亮火焰并卷起大片白色烟雾，旁边可见塔架。 成片规格为 7 秒 · 1056×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rocket rises from its launchpad beneath a clear blue sky, emitting bright flames and a large white smoke plume beside the gantry. Output: 7s · 1056×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EC4021/status/2087541717889343787",
    "sourceLabel": "X 原帖 · @EC4021",
    "author": "EC4021",
    "publishedAt": "2026-08-12T14:08:24.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087541717889343787.mp4",
    "posterUrl": "/posters/x/2087541717889343787.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1056×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "OK，Minimax-H3用上了。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 151,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087541644237119845",
    "title": "巨角怪物俯视持剑盾的微小战士",
    "titleEn": "A Giant Horned Beast Towers Over a Sword-Bearing Warrior",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一头披着黑色甲壳、长有巨角的庞然怪物从金色雾气中俯视下方持剑盾的微小战士。 成片规格为 27 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An enormous dark-armored beast with sweeping horns looms through golden mist over a tiny warrior holding a sword and shield. Output: 27s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SaqerG/status/2087541644237119845",
    "sourceLabel": "X 原帖 · @SaqerG",
    "author": "صقر الغيلاني",
    "publishedAt": "2026-08-12T14:08:07.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087541644237119845.mp4",
    "posterUrl": "/posters/x/2087541644237119845.jpg",
    "duration": 27,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5 vs Minimax H3 in Comfyui\n#comfyui\n#ltx\n#MinimaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 128,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087539373420228818",
    "title": "古装剧集的片头动画",
    "titleEn": "Opening Title Animation for a Period Drama",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "原帖明确将随帖作品描述为一部古装题材剧集的片头动画。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly describes the attached work as an opening title animation for a period drama series. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2087539373420228818",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-08-12T13:59:05.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087539373420228818.mp4",
    "posterUrl": "/posters/x/2087539373420228818.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Title animation for an ancient period series with MiniMax H3 🔺\n\nText to video prompt below:",
    "engagement": {
      "replies": 13,
      "reposts": 22,
      "likes": 244,
      "views": 7420,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit period-series subject and opening-title format"
  },
  {
    "id": "x-2087538481316679732",
    "title": "热带果摊动物群的四档分辨率对照",
    "titleEn": "Four-Resolution Comparison of Animals Around a Tropical Fruit Stand",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴耳机的水豚状动物站在热带果摊后方，周围分布着鹦鹉和青蛙，四格分别标注为 256p、512p、768p 和 1024p；原帖注明各格采用相同提示词与随机种子。 成片规格为 10 秒 · 1600×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A headphone-wearing capybara-like animal stands behind a tropical fruit stall among parrots and frogs in panels labeled 256p, 512p, 768p, and 1024p; the source states that all panels share the same prompt and seed. Output: 10s · 1600×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PolaczekSagi/status/2087538481316679732",
    "sourceLabel": "X 原帖 · @PolaczekSagi",
    "author": "Sagi Polaczek 🦜",
    "publishedAt": "2026-08-12T13:55:33.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087538481316679732.mp4",
    "posterUrl": "/posters/x/2087538481316679732.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1600×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "yesterday I posted an experiment I did with MiniMax-H3 and people liked it, so here's another one:\n\n👀 same prompt & seed just a different output resolution",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 99,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087537801227329944",
    "title": "教室里三名制服角色同步比心",
    "titleEn": "Three Uniformed Characters Make Coordinated Heart Gestures in a Classroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三名穿制服的动画角色在明亮教室中同步抬手变换动作，并在末段用手组成心形。 成片规格为 12 秒 · 736×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three uniformed animated characters make coordinated raised-hand movements in a bright classroom and form heart shapes near the end. Output: 12s · 736×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/iroirokomaru/status/2087537801227329944",
    "sourceLabel": "X 原帖 · @iroirokomaru",
    "author": "いろいろ小丸",
    "publishedAt": "2026-08-12T13:52:51.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087537801227329944.mp4",
    "posterUrl": "/posters/x/2087537801227329944.jpg",
    "duration": 12,
    "aspectRatio": "square",
    "resolution": "736×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax H3 comfyui テンプレートのままのほうが、GGUFより生成時間がかなり短いのはなんでかな？\n顔が崩れるのはピクセルが小さすぎたからかも。\nH3 firstblockcache と　4step loraで6step、, size setting 0.5で10分、あんまり崩れなかった。けど色調が一定しないです。\nかえります。",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 21,
      "views": 262,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087535269935272353",
    "title": "雷暴巨浪中的灰色驱逐舰",
    "titleEn": "A Gray Destroyer Crosses Towering Waves beneath a Thunderstorm",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "舰身标有 124 的灰色驱逐舰穿过翻涌巨浪，暗色云层中划过闪电，海面布满白色浪花。 成片规格为 10 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A gray destroyer marked 124 cuts through towering waves as lightning crosses dark storm clouds and white spray covers the sea. Output: 10s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hectorVFX/status/2087535269935272353",
    "sourceLabel": "X 原帖 · @hectorVFX",
    "author": "hectorvfx",
    "publishedAt": "2026-08-12T13:42:47.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087535269935272353.mp4",
    "posterUrl": "/posters/x/2087535269935272353.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5 Ya esta aquí y lo estoy probando.\n10 segundos en full hd generados en 2 minutos en la 5090.\nDe momento Image to video. \nEsta calidad en local es demasiado rápido, no me esperaba esta mejora para nada.\nQuien ha dicho que este modelo es peor que Minimax H3?\n\nW T F.",
    "engagement": {
      "replies": 5,
      "reposts": 3,
      "likes": 34,
      "views": 1741,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087534142816981244",
    "title": "美少女战士入侵",
    "titleEn": "Sailor Moon Invasion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖以“美少女战士入侵”明确概括随帖内容，并注明其在 RTX 4090 上本地生成。 成片规格为 15 秒 · 1584×1200。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly labels the attached work as a “Sailor Moon invasion” and notes that it was generated locally on an RTX 4090. Output: 15s · 1584×1200. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sadlemonjuice/status/2087534142816981244",
    "sourceLabel": "X 原帖 · @sadlemonjuice",
    "author": "Socratech",
    "publishedAt": "2026-08-12T13:38:18.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087534142816981244.mp4",
    "posterUrl": "/posters/x/2087534142816981244.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1584×1200",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sailor Moon invasion MiniMax H3 local gpu4090",
    "engagement": {
      "replies": 6,
      "reposts": 12,
      "likes": 125,
      "views": 6938,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "sourceCaption: explicit named character franchise and invasion premise"
  },
  {
    "id": "x-2087533118265954584",
    "title": "更衣室内面对镜头的长发女子",
    "titleEn": "A Long-Haired Woman Faces the Camera in a Dressing Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一名长黑发女子穿着灰色系带上衣面对镜头，身后的更衣室内可见镜子、衣架和顶灯。 成片规格为 15 秒 · 2160×3780。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long black-haired woman in a gray lace-up top faces the camera in a dressing room with a mirror, clothing rack, and ceiling lights behind her. Output: 15s · 2160×3780. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IamEmily2050/status/2087533118265954584",
    "sourceLabel": "X 原帖 · @IamEmily2050",
    "author": "Emily",
    "publishedAt": "2026-08-12T13:34:14.000Z",
    "addedAt": "2026-08-13T00:38:33.430Z",
    "mediaUrl": "/media/x-2087533118265954584.mp4",
    "posterUrl": "/posters/x/2087533118265954584.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "2160×3780",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Reference to video.\nWith Topaz 1080p upscale.",
    "engagement": {
      "replies": 10,
      "reposts": 1,
      "likes": 72,
      "views": 5021,
      "snapshotAt": "2026-08-13T00:38:33.430Z"
    },
    "approvedAt": "2026-08-13T00:38:33.430Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087532876896276638",
    "title": "跟随节拍的拨弦手部与身体动作",
    "titleEn": "Beat-Locked Strumming-Hand and Body Movement",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拨弦手部和身体动作被明确描述为跟随音乐节拍同步，而不是由文字指令直接指定。 成片规格为 24 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The strumming hand and body movements are explicitly described as synchronized to the beat rather than directly specified in the prompt. Output: 24s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LJMBAU/status/2087532876896276638",
    "sourceLabel": "X 原帖 · @LJMBAU",
    "author": "M3LJ",
    "publishedAt": "2026-08-12T13:33:16.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087532876896276638.mp4",
    "posterUrl": "/posters/x/2087532876896276638.jpg",
    "duration": 24,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Look what GROK BOT cooked up for me 👀 …\n\nYeah nah, not touching that overpriced ‘beta’ slop, learnt my lesson already purchasing a non-existent beta FSD from the very same snake oil salesman.\n\nThis is my Claude Code + MiniMax H3 workflow. \n\nWatch the strumming hand and body movements - that’s beat-locked, not prompted",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 355,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit strumming-hand action, body movement, and beat-locked synchronization"
  },
  {
    "id": "x-2087530367351656516",
    "title": "机器人助手 PIP 得知自己将被永久关机",
    "titleEn": "Robot Assistant PIP Learns He Faces Permanent Shutdown",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "忠诚勤勉的机器人助手 PIP 发现自己的英雄命运包含永久关机，于是决定与写作者谈谈并尝试改写结局。 成片规格为 50 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "PIP, a loyal and tireless robot assistant, discovers that his heroic destiny includes permanent shutdown and decides to confront the writer about changing the ending. Output: 50s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HadesDesign/status/2087530367351656516",
    "sourceLabel": "X 原帖 · @HadesDesign",
    "author": "Alexandra Dekimpe",
    "publishedAt": "2026-08-12T13:23:18.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087530367351656516.mp4",
    "posterUrl": "/posters/x/2087530367351656516.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Every hero needs an assistant.\n\nLoyal. Resourceful. Tirelessly devoted.\n\nBut when PIP discovers that his heroic destiny includes… permanent shutdown, he decides it might be time to have a word with the writer. 🤖\n\nPLEASE STAY IN CHARACTER\nCreated with MiniMax H3 in @itsPolloAI for the Animated Worlds Challenge.\nWould you follow the prompt… or rewrite your own ending?\n\nIf PIP made you smile, give the little guy some support — like, repost, and tell me what you would have done. 😏\n\n#PolloAI #PolloAnimation #MinimaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 69,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit named robot assistant, shutdown conflict, writer confrontation, and ending premise"
  },
  {
    "id": "x-2087528565692592139",
    "title": "烛光藏书室与洞口外的明亮星球",
    "titleEn": "A Bright Planet Beyond a Candlelit Library Cave",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一册巨大的书在昏暗藏书室中央展开，四周堆满书本和蜡烛，岩洞状开口外悬着明亮星球并投下一束光；原帖说明完整段落由多个短片段拼接而成。 成片规格为 5 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A huge open book rests in a dark library surrounded by stacked volumes and candles, while a bright planet casts a beam through a cave-like opening; the source says the full sequence was assembled from multiple short segments. Output: 5s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Brjen/status/2087528565692592139",
    "sourceLabel": "X 原帖 · @Brjen",
    "author": "Brian",
    "publishedAt": "2026-08-12T13:16:09.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087528565692592139.mp4",
    "posterUrl": "/posters/x/2087528565692592139.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Good morning, work continues with MiniMax H3 today on the M4 16GB MacMini.\n\nI'm also working on another 1 minute video sequence. Trying to figure out how to piece together and make a coherent video from multiple shorter videos.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 115,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087527326678491533",
    "title": "沙漠中车辆之间的三名男子",
    "titleEn": "Three Men Stand Between Vehicles in a Desert",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名卷发男子站在沙漠中央，手里拿着一块浅色布料或纸张，另外两名男子分别站在两辆车前，近处还有一名人物的局部轮廓。 成片规格为 19 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A curly-haired man stands in a desert holding a pale cloth or sheet, with two other men positioned before separate vehicles and part of another figure visible in the foreground. Output: 19s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/flarehorizon275/status/2087527326678491533",
    "sourceLabel": "X 原帖 · @flarehorizon275",
    "author": "Flare Horizon",
    "publishedAt": "2026-08-12T13:11:13.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087527326678491533.mp4",
    "posterUrl": "/posters/x/2087527326678491533.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "tool: MiniMax H3 、seedance2.5 、elevenlabs、PS、gemini",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 81,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087526063933165731",
    "title": "沉默八秒的悲伤表演",
    "titleEn": "An Eight-Second Silent Sad Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子保持沉默完成八秒表演：先静止露出悲伤神情，再眨眼一次、落下一滴泪，最后微笑并闭紧嘴、轻咬牙关。 成片规格为 8 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman remains silent through an eight-second performance: she holds a sad expression, blinks once, sheds a tear, then smiles while closing her mouth and lightly clenching her teeth. Output: 8s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MireilleDartois/status/2087526063933165731",
    "sourceLabel": "X 原帖 · @MireilleDartois",
    "author": "ダルトワ★TV",
    "publishedAt": "2026-08-12T13:06:12.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087526063933165731.mp4",
    "posterUrl": "/posters/x/2087526063933165731.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3でしゃべらない演技💡\n\n何が大変かって｢すぐにしゃべってしまう｣こと😓\n8秒間黙らせるのにはなかなか苦労しました💧\n\nまずプロンプトで8秒間の演技を言い渡します｡\n\n[0.0s-3.0s]no movement, sad expression.\n[3.0s-3.5s]she blinks once.\n[3.5s-5.5s]tear drops down her cheek.\n[5.5s-8.0s]she slightly smiles and shuts her mouth up and gritted teet slightly.\n\nとにかく何かを言い渡して他のコトをしてもらいます｡\n動かずに悲しい演技､1回だけ瞬き､涙が頬を伝う...\nでもですね最後の｢smiles｣で喋り出してしまうので💡\n\n｢shuts her mouth up and gritted teet slightly｣\n口を閉ざし､歯を少し食いしばる\n\nとダメ押し💨\nこれで8秒間黙ってくれました💧",
    "engagement": {
      "replies": 5,
      "reposts": 6,
      "likes": 80,
      "views": 3305,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit female subject, eight-second duration, silent-performance constraint, timed expressions, blink, tear, and final action"
  },
  {
    "id": "x-2087525285126111455",
    "title": "开放式办公室里看屏幕的眼镜女子",
    "titleEn": "A Woman with Glasses Looks at a Screen in an Open Office",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴眼镜的女子坐在开放式办公室工位前，一手托脸看向屏幕，桌上摆着键盘、笔记本、水杯和多台显示器；原帖称创作者准备参考图后使用了一条提示词。 成片规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman with glasses sits at a workstation in an open office, resting her face on one hand as she looks at a screen amid keyboards, notebooks, cups, and monitors; the source says the creator prepared reference images and used one prompt. Output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/alexutopia/status/2087525285126111455",
    "sourceLabel": "X 原帖 · @alexutopia",
    "author": "Alex Utopia",
    "publishedAt": "2026-08-12T13:03:06.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087525285126111455.mp4",
    "posterUrl": "/posters/x/2087525285126111455.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An advertisement like this would easily\ncost $8.000-$20.000 to produce.\n\nWith MiniMax H3 from Hailuo AI,\nit cost me virtually nothing.\n\nI uploaded the prepared references,\nwrote one prompt, pressed one button.\n\n5 minutes.",
    "engagement": {
      "replies": 12,
      "reposts": 5,
      "likes": 53,
      "views": 1525,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame"
  },
  {
    "id": "x-2087524615295836302",
    "title": "街角小店英雄的剧集片头",
    "titleEn": "A Series Opening for Corner-Shop Heroes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖将随帖作品描述为献给街角小店中普通英雄的一段剧集片头。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source describes the attached work as a series opening devoted to everyday heroes from a local corner shop. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/koldo2k/status/2087524615295836302",
    "sourceLabel": "X 原帖 · @koldo2k",
    "author": "Koldo Huici",
    "publishedAt": "2026-08-12T13:00:27.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087524615295836302.mp4",
    "posterUrl": "/posters/x/2087524615295836302.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Every hero deserves a proper opening, even if it's just the ones from your local corner shop.\n\nCreating series openings with Minimax H3\n\nHere's the prompt 👇",
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 29,
      "views": 1348,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit corner-shop hero premise and series-opening format"
  },
  {
    "id": "x-2087524511029645682",
    "title": "多组参考元素拼合成的城堡街区",
    "titleEn": "A Castle District Assembled from Multiple Reference Elements",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面主画面是一处由石塔、土墙房屋、残破拱廊和泥泞街道组成的城堡街区，下方并列士兵服装、垃圾桶、建筑残件与塔楼等参考素材。 成片规格为 15 秒 · 2362×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster's main frame shows a castle district of a stone tower, earthen houses, a ruined arcade, and a muddy street, with reference assets below including a soldier uniform, bin, architectural fragments, and towers. Output: 15s · 2362×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/os_orabi/status/2087524511029645682",
    "sourceLabel": "X 原帖 · @os_orabi",
    "author": "osama orabi",
    "publishedAt": "2026-08-12T13:00:02.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087524511029645682.mp4",
    "posterUrl": "/posters/x/2087524511029645682.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2362×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seamlessly integrate the reference elements while preserving their original designs and details.  MiniMax H3 \n\nby @Hailuo_AI - #MiniMaxH3",
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 26,
      "views": 363,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption explicitly describes integrating reference elements while preserving their designs"
  },
  {
    "id": "x-2087523366479491329",
    "title": "荒漠聚落里的头盔人物、四足机器人与巨型机械",
    "titleEn": "A Helmeted Figure, Quadruped Robot, and Giant Machine in a Desert Outpost",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一名持长形物体的头盔人物与蓝灯四足机器人位于荒漠工业聚落中，身后悬着一台带橙色灯光的巨型机械。 成片规格为 10 秒 · 2560×1408。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A helmeted figure carrying a long object and a blue-lit quadruped robot stand in a desert-industrial settlement beneath a massive machine with orange lights. Output: 10s · 2560×1408. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FunTheoryX/status/2087523366479491329",
    "sourceLabel": "X 原帖 · @FunTheoryX",
    "author": "FunTheory",
    "publishedAt": "2026-08-12T12:55:29.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087523366479491329.mp4",
    "posterUrl": "/posters/x/2087523366479491329.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1408",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "nature",
      "sci-fi"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 animation \n\nGENERATION DETAILS\nModel: MiniMax H3 FL2VA Pruned 20B\nCheckpoint: MiniMax-H3-FL2VA-pruned_rank8_int8_convrot\nInterface: WanGP v12.452 by DeepBeepMeep, running through Pinokio\nInput mode: Single reference/start image\nSeed: 247526563\nInference steps: 20\nSampler: Euler\nFlow shift: 12\nGeneration resolution preset: 608 x 832\nFinal output: 1920 x 1088, 48 fps\nDuration: 15.06 seconds\nTemporal upsampling: RIFE 2x\nSpatial upsampling: Lanczos 2x\nRender time: 871 seconds (14 minutes 31 seconds)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 107,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087523002623934736",
    "title": "围裙女子站在明亮厨房中",
    "titleEn": "An Apron-Clad Woman Standing in a Bright Kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现一名穿浅色围裙的女子站在现代厨房岛台后，台面摆有碗、杯子、水果和瓶子；来源文案记录 1080p、十秒、约十八分钟与约三十日元的制作参数。 成片规格为 10 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows a woman in a pale apron behind a modern kitchen island with a bowl, cup, fruit, and bottles; the caption records 1080p, ten seconds, about 18 minutes, and a cost of roughly 30 yen. Output: 10s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HackCooking/status/2087523002623934736",
    "sourceLabel": "X 原帖 · @HackCooking",
    "author": "Cooking life hack",
    "publishedAt": "2026-08-12T12:54:02.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087523002623934736.mp4",
    "posterUrl": "/posters/x/2087523002623934736.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3を1080Pで出してみた＾＾\nやっぱりすごい綺麗！10秒18分で コストは30円くらい？\n週１くらいで動画出すんだったら、1080Pでもいいのかな？→は4K60Hzにアップスケール",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 230,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption supplies resolution, duration, runtime, cost, and upscaling details"
  },
  {
    "id": "x-2087522582304317478",
    "title": "总有一天要成为超级英雄",
    "titleEn": "Someday I'll Be a Superhero",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖以“总有一天要成为超级英雄”作为随帖作品的明确主题。 成片规格为 45 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source presents “Someday I'll Be a Superhero” as the explicit theme of the attached work. Output: 45s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI__TSUBAKI/status/2087522582304317478",
    "sourceLabel": "X 原帖 · @AI__TSUBAKI",
    "author": "TSUBAKI",
    "publishedAt": "2026-08-12T12:52:22.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087522582304317478.mp4",
    "posterUrl": "/posters/x/2087522582304317478.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Wanna be a superhero someday\n\nMade with MiniMax H3 on @PixVerse_\n#H3onPixVerse #PixVerseCPP",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 13,
      "views": 2013,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit superhero aspiration used as the work's theme line"
  },
  {
    "id": "x-2087520980268986668",
    "title": "两张水彩角色设定图驱动的海蒂主题动画",
    "titleEn": "A Heidi-Themed Animation from Two Watercolor Character Sheets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者以两张水彩风格角色设定图作为参考，制作了一段“海蒂之日”主题动画，并称输出基本保留了参考气质。 成片规格为 10 秒 · 896×504。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author used two watercolor-style character sheets as references for a Heidi Day-themed animation and says the result largely retained their intended feel. Output: 10s · 896×504. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ivy432hz/status/2087520980268986668",
    "sourceLabel": "X 原帖 · @ivy432hz",
    "author": "あいびぃ",
    "publishedAt": "2026-08-12T12:46:00.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087520980268986668.mp4",
    "posterUrl": "/posters/x/2087520980268986668.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "896×504",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ハイジの日、MiniMax H3 で。水彩系キャラシートを 2 枚渡しただけですけど、割とそれっぽく出力されてる感じ。ミカゼ↓ になりがち #今日は創作の日 #AIニケちゃん",
    "engagement": {
      "replies": 0,
      "reposts": 4,
      "likes": 11,
      "views": 723,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit Heidi Day theme, two watercolor character-sheet references, and reported visual resemblance"
  },
  {
    "id": "x-2087520973197279371",
    "title": "海边奇幻小镇广场的双模型对照",
    "titleEn": "A Two-Model Comparison of a Fantastical Seaside Town Square",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现一座色彩鲜明的海边奇幻小镇，广场周围分布圆顶建筑、棕榈树和行人；来源文案说明两种模型使用同一提示词进行质量与速度对照。 成片规格为 50 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows a brightly colored fantastical seaside town with domed buildings, palm trees, and people around a plaza; the caption says two models were compared with the same prompt for quality and speed. Output: 50s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiaicreate/status/2087520973197279371",
    "sourceLabel": "X 原帖 · @aiaicreate",
    "author": "R@aiaicreate",
    "publishedAt": "2026-08-12T12:45:58.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087520973197279371.mp4",
    "posterUrl": "/posters/x/2087520973197279371.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "▼話題\n動画生成AIモデル「Minimax H3」と「LTX 2.5」の同一プロンプトによる生成結果の比較動画が投稿され、両者の品質差が話題となっている。\n\n▼主な反応\n・描写の品質においてMinimax H3がLTX 2.5を圧倒しているとの評価\n・LTX 2.5は生成速度が約8倍高速なため、Minimax H3で作成した低解像度動画のアップスケール用途に適しているとの指摘\n\n▼受け止め\nMinimax H3の品質の高さを認めつつ、速度差を活かしたLTX 2.5との組み合わせによる活用法を模索する好意的な受け止めが優勢である。\n\n#MinimaxH3 #動画生成AI\n\nURLはリプ⬇️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 33,
      "views": 6170,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption explicitly supplies the same-prompt two-model comparison context"
  },
  {
    "id": "x-2087520166515192098",
    "title": "厨房桌旁摊开的手掌特写",
    "titleEn": "Close-Up of an Open Palm Beside a Kitchen Table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面以近距离视角呈现一只朝上的手掌，后方木桌上放着金属杯、勺子和玻璃瓶，窗边光线落在桌面与手指上。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster presents an upturned palm at close range, with a metal cup, spoon, and glass bottle on the wooden table behind it and window light falling across the hand and tabletop. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/op7298/status/2087520166515192098",
    "sourceLabel": "X 原帖 · @op7298",
    "author": "Mike Huang",
    "publishedAt": "2026-08-12T12:42:46.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087520166515192098.mp4",
    "posterUrl": "/posters/x/2087520166515192098.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Let the showcase define the next model-evaluation question. A concrete MiniMax H3 example is a better starting point than another feature list. Explore the wider model on Atlas Cloud. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087520103479030221",
    "title": "机械臂悬停在杯子、红色方块与玻璃罐上方",
    "titleEn": "A Robotic Arm Hovers Above a Mug, Red Block, and Glass Jar",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现一台白色机械臂及其黑色夹爪，夹爪下方的白色桌面依次摆着陶瓷杯、红色方块和透明玻璃罐。 成片规格为 7 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows a white robotic arm with a black gripper above a white tabletop holding a ceramic mug, a red block, and a clear glass jar. Output: 7s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/L2jD5Ac5ig38277/status/2087520103479030221",
    "sourceLabel": "X 原帖 · @L2jD5Ac5ig38277",
    "author": "あ",
    "publishedAt": "2026-08-12T12:42:31.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087520103479030221.mp4",
    "posterUrl": "/posters/x/2087520103479030221.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A concrete showcase makes model comparison more useful. Use the output as a reference point, then try MiniMax H3 on Atlas Cloud. The broader MiniMax H3 showcase set gives teams several outputs to study over time. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087520017227407536",
    "title": "雪中竹林里的两名古装人物与一张纸",
    "titleEn": "Two Costumed Figures and a Sheet of Paper in a Snowy Bamboo Grove",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雪中竹林里一名女子与脸颊有抓痕的男子交替出现在近景中，结尾女子在双人画面里拿着一张纸。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman and a man with scratches on one cheek alternate in close-up within a snowy bamboo grove before the woman holds a sheet of paper in a wider two-person view. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/joliko_/status/2087520017227407536",
    "sourceLabel": "X 原帖 · @joliko_",
    "author": "ジョリ子",
    "publishedAt": "2026-08-12T12:42:11.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087520017227407536.mp4",
    "posterUrl": "/posters/x/2087520017227407536.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Let the showcase define the next model-evaluation question. This video is a starting point for a wider MiniMax H3 review. A MiniMax H3 result can help teams decide whether the model belongs in their workflow. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 68,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087519915876000022",
    "title": "室内男子正面特写与女子背影",
    "titleEn": "Indoor Close-Up of a Man Facing a Woman Seen from Behind",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面以竖幅近景呈现一名黑发男子正对镜头侧前方的人，前景只露出一名穿格纹上衣女子的后脑与肩部，背景是暖色木质室内空间。 成片规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The portrait poster frames a dark-haired man facing someone just off camera, while the foreground shows only the back of a woman in a checked top against a warm wood-paneled interior. Output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MignonRene51273/status/2087519915876000022",
    "sourceLabel": "X 原帖 · @MignonRene51273",
    "author": "Renee Mignon",
    "publishedAt": "2026-08-12T12:41:46.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087519915876000022.mp4",
    "posterUrl": "/posters/x/2087519915876000022.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A visible MiniMax H3 result helps teams discuss workflow fit. The showcase gives teams a real result to discuss before a larger evaluation. A MiniMax H3 example is more useful when the team can inspect the result directly. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 70,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087519783122309312",
    "title": "夜间燃烧基地中的第一人称步枪视角",
    "titleEn": "First-Person Rifle View in a Burning Nighttime Base",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现带瞄准镜步枪的第一人称视角，前方是沙袋与混凝土掩体，夜色中的基地多处起火，空中还有直升机探照灯。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows a first-person view behind a scoped rifle, sandbags, and a concrete barrier, with multiple fires across a nighttime base and a helicopter searchlight overhead. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/OcaneLecoqjor/status/2087519783122309312",
    "sourceLabel": "X 原帖 · @OcaneLecoqjor",
    "author": "Océane Lecoq",
    "publishedAt": "2026-08-12T12:41:15.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087519783122309312.mp4",
    "posterUrl": "/posters/x/2087519783122309312.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Use an H3 result to begin the production conversation. A practical model review starts with examples like this. The MiniMax H3 showcase set puts actual outputs beside the model discussion. via Atlas Cloud.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087518345633595689",
    "title": "背心男子走在阳光街道上",
    "titleEn": "A Man in a Waistcoat Walking Down a Sunlit Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面从背后拍摄一名穿白衬衫与深色背心的男子，他沿着铺装人行道前行，街边可见棕榈树、车辆和低层建筑。 成片规格为 7 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster follows a man in a white shirt and dark waistcoat from behind as he moves along a paved sidewalk lined with palm trees, cars, and low-rise buildings. Output: 7s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kmaiink/status/2087518345633595689",
    "sourceLabel": "X 原帖 · @kmaiink",
    "author": "🍑",
    "publishedAt": "2026-08-12T12:35:32.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087518345633595689.mp4",
    "posterUrl": "/posters/x/2087518345633595689.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The quickest H3 check begins with something you can watch. Watch the output, then test your own use case. MiniMax H3 is available on Atlas Cloud for teams ready to test an approved brief. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087518232278393167",
    "title": "树荫下低头的蓝衣女子特写",
    "titleEn": "Close-Up of a Woman in Blue Looking Down Beneath Trees",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面以低角度近景呈现一名穿浅蓝上衣的女子，她低头看向画面下方，背景由阳光照亮的树叶构成。 成片规格为 9 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The low-angle poster shows a woman in a pale blue top looking down toward the lower edge of the frame, with sunlit foliage behind her. Output: 9s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Stephan94827583/status/2087518232278393167",
    "sourceLabel": "X 原帖 · @Stephan94827583",
    "author": "Stephanie Rojas",
    "publishedAt": "2026-08-12T12:35:05.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087518232278393167.mp4",
    "posterUrl": "/posters/x/2087518232278393167.jpg",
    "duration": 9,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is live; the next step is to inspect an output. The attached result offers a quick look at MiniMax H3 output. Atlas Cloud provides a direct MiniMax H3 route for production testing. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087518161784770981",
    "title": "沙滩上双手搭肩的双人近景",
    "titleEn": "Two People in a Beach Close-Up with Hands on Shoulders",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴眼镜的男子坐在沙滩前景，身后另一人的双手始终放在他的肩上。 成片规格为 2 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in glasses sits in the foreground on a beach while another person behind him keeps both hands on his shoulders. Output: 2s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LiloKrause/status/2087518161784770981",
    "sourceLabel": "X 原帖 · @LiloKrause",
    "author": "Lilo Krause",
    "publishedAt": "2026-08-12T12:34:48.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087518161784770981.mp4",
    "posterUrl": "/posters/x/2087518161784770981.jpg",
    "duration": 2,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Before writing an H3 brief, take a look at a completed output. A finished output makes the route easier to understand. MiniMax H3 starts at $0.10/second on Atlas Cloud. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 96,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087518135574474854",
    "title": "城市人行道上的像素树",
    "titleEn": "A Pixelated Tree on a City Sidewalk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面中央是一棵由绿色与棕色方块构成的像素树，立在城市人行道边，周围分布行人、车辆和高层建筑。 成片规格为 4 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster centers a tree built from green and brown pixel blocks on a city sidewalk, surrounded by pedestrians, cars, and tall buildings. Output: 4s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wchtna/status/2087518135574474854",
    "sourceLabel": "X 原帖 · @wchtna",
    "author": "黒田じん",
    "publishedAt": "2026-08-12T12:34:42.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087518135574474854.mp4",
    "posterUrl": "/posters/x/2087518135574474854.jpg",
    "duration": 4,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The H3 route is live, and an output is the best place to begin. The attached result offers a quick look at MiniMax H3 output. MiniMax H3 gives teams a concrete video output to review before a larger test. http://www.atlascloud.ai/models/minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 66,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087517644216008935",
    "title": "多种加速 LoRA 与高动态提示链路",
    "titleEn": "Acceleration LoRAs with a High-Motion Prompting Workflow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案列出 T8、Turbo、剪枝版与 LightX2V 等加速 LoRA，并介绍了一套用于编写高动态提示词、再对不同 LoRA 进行比较的 Skill 链路。 成片规格为 10 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption lists T8, Turbo, pruned, and LightX2V acceleration LoRAs, then describes a Skill workflow for writing high-motion prompts and comparing those LoRAs. Output: 10s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/0xmomonifty/status/2087517644216008935",
    "sourceLabel": "X 原帖 · @0xmomonifty",
    "author": "0xMomo",
    "publishedAt": "2026-08-12T12:32:45.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087517644216008935.mp4",
    "posterUrl": "/posters/x/2087517644216008935.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 开源到现在不到半个月，已经有非常多的加速 Lora 了\n\n比如之前推荐的 T8 Lora，以及 larryvrh 的 Turbo、drbaph 的 pruned、lightx2v 的 4 步 和 8步，其中 lightx2v 的 8 步动态效果表现非常不错\n\n这里推荐一个 B 站大佬 -> AI代码侠土豆 @AI3261588371052  ，他对目前这几个 Lora 进行了评测以及开源了的相关 skill ，可以直接生成 H3 的高动态提示词，使用它可以在这几个 Lora 当中充分测试这方面的表现\n\n总的来说，目前 H3 无论使用 ComfyUI 还是官方的 API，都可以非常顺手地使用在绝大多数场景了",
    "engagement": {
      "replies": 6,
      "reposts": 0,
      "likes": 16,
      "views": 1308,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit acceleration-LoRA comparison and prompt-writing workflow"
  },
  {
    "id": "x-2087516957134238197",
    "title": "《神兽伦巴》与褐肤猫耳埃及少女",
    "titleEn": "Divine Beast Roomba and the Brown-Skinned Cat-Eared Egyptian Girl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将这段日常短篇命名为《神兽伦巴》，并明确写出主角是一名褐肤、猫耳、埃及风格的少女。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption names this everyday short “Divine Beast Roomba” and explicitly identifies its subject as a brown-skinned, cat-eared girl with an Egyptian theme. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087516957134238197",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T12:30:01.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087516957134238197.mp4",
    "posterUrl": "/posters/x/2087516957134238197.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n日常エピソード\n『神獣ルンバ』\n褐色猫耳エジプトロリちゃん",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 15,
      "views": 373,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit work title and character description"
  },
  {
    "id": "x-2087516878935933077",
    "title": "RTX 5090 上的 LTX 2.5 十五秒运行",
    "titleEn": "A 15-Second LTX 2.5 Run on an RTX 5090",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明 LTX 2.5 在 RTX 5090 上完成十五秒片段共耗时三分二十秒，并据此讨论开源模型间的速度位置。 成片规格为 15 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says LTX 2.5 produced a 15-second clip on an RTX 5090 in three minutes and twenty seconds, using that result to discuss relative speed among open models. Output: 15s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Nameless_btc/status/2087516878935933077",
    "sourceLabel": "X 原帖 · @Nameless_btc",
    "author": "Nameless",
    "publishedAt": "2026-08-12T12:29:42.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087516878935933077.mp4",
    "posterUrl": "/posters/x/2087516878935933077.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ltx 2.5做的15秒片段  耗时3分20秒 显卡5090，minimax h3 现在已经不是开源第1的视频大模型了",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 496,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit model, hardware, output duration, and runtime"
  },
  {
    "id": "x-2087516569874428044",
    "title": "球场上两名男子相遇并触碰头部",
    "titleEn": "Two Men Meet on a Field as One Touches the Other's Head",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一名拿着瓶子的插画男子走向球场上的高个男子，后者把手放到他的头上。 成片规格为 17 秒 · 640×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustrated man holding a bottle walks toward a taller man on a field, and the taller man places one hand on his head. Output: 17s · 640×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/marieinfareast/status/2087516569874428044",
    "sourceLabel": "X 原帖 · @marieinfareast",
    "author": "マリー",
    "publishedAt": "2026-08-12T12:28:29.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087516569874428044.mp4",
    "posterUrl": "/posters/x/2087516569874428044.jpg",
    "duration": 17,
    "aspectRatio": "portrait",
    "resolution": "640×1344",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "夏だぜって感じの動画。全部Fableにやってもらったんだけど…\n\n環境: ComfyUI + MiniMax H3 (fl2va int8) / RTX 5070 Ti 16GB / 24fps\n\n作り方:\n1. 静止画1枚を起点にI2Vで第1カット(5.2s)\n2. 第1カットの最終フレーム＋新規キーフレーム画でFLF(首尾フレーム)生成→第2カット\n3. キーフレーム2枚でFLF→第3カット\n4. 3本を連結(音は80msクロスフェード)\n\n各カットはTurbo LoRA 6stepsの低解像度(448×928)でseedガチャ4種→当たりseedだけ20steps・640×1344で清書。下書き約2分/本、清書約15分/本。音もH3が同時生成、後付けなし\n\n発見:\n①カメラを動かしたい時は「パンして」と書くより、視点の違う画を開始/終了フレームに渡す方が良い。正面からの画→背中越しの画を指定すると、モデルが間のカメラの回り込み(公式語彙でいうArc Shot)を自分で発明する。4seed中4つ動いた\n\n②解像度は縦横とも32の倍数必須(VAE 1/16×2×2パッチ)。フレーム数は17n+5の格子(124f≈5.2s、141f≈5.9s)\n\n③表情や「口を閉じたまま」は文章指定だと歩留まり半分。キーフレームの絵に描いて渡す方が確実(絵の指定は文章の指定より強い)",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 227,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087516135877202059",
    "title": "从游戏本桌面到霓虹赛车都市",
    "titleEn": "From a Gaming Laptop Desktop to a Neon Racing City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开提示词描述游戏本在桌面亮起，镜头推进屏幕进入霓虹赛车都市，随后返回产品英雄镜头与“BUILT TO PLAY. BUILT TO WIN.”结尾文案。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public prompt describes a gaming laptop powering up on a desk, a push through its screen into a neon racing city, and a return to the product hero shot with the closing line “BUILT TO PLAY. BUILT TO WIN.” Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/iam_mian7/status/2087516135877202059",
    "sourceLabel": "X 原帖 · @iam_mian7",
    "author": "Ai Arainz",
    "publishedAt": "2026-08-12T12:26:45.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087516135877202059.mp4",
    "posterUrl": "/posters/x/2087516135877202059.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ready for anything 🎮\n\nMinimax H3  on @budgetpixel  just turned a gaming laptop into a full cinematic commercial 🔥\n\nPrompt :\nCreate a premium, ultra-realistic 15-second gaming laptop advertisement in 16:9, 4K, 24fps. Fast-paced, cinematic, futuristic, high-end tech commercial aesthetic.\n\n0–3s — The Reveal\nExtreme close-up of a sleek gaming laptop sitting on a dark futuristic desk. RGB lighting slowly activates across the keyboard. The screen remains black. Camera performs a smooth low-angle push-in.\n\n3–6s — Power On\nThe laptop suddenly powers on. The screen illuminates the room with dramatic light. Macro shots reveal the keyboard, precision trackpad, cooling vents, and premium metallic chassis. RGB reflections move naturally across the surface.\n\n6–10s — Enter the Game\nThe camera pushes directly into the screen and seamlessly transitions into an intense photorealistic game world. A futuristic city at night, neon lights, explosions in the distance, and a high-speed vehicle racing through the streets. Dynamic tracking camera, realistic motion blur, cinematic depth.\n\n10–13s — Performance\nRapid macro cuts: GPU graphics rendering, cooling fans spinning, keyboard illuminated, player's hand moving naturally across the keys. The game remains smooth and visually detailed.\n\n13–15s — Hero Shot\nPull back from the screen to reveal the entire laptop on the desk. RGB lighting pulses once. Dark background with subtle atmospheric fog.\n\nEnd Text:\n“BUILT TO PLAY. BUILT TO WIN.”\n\nUltra-photorealistic materials, premium product photography, realistic reflections, physically accurate lighting, sharp details, smooth camera movement, cinematic contrast, shallow depth of field, realistic hands, no glitches, no distorted hardware, no random text or logos.",
    "engagement": {
      "replies": 24,
      "reposts": 11,
      "likes": 336,
      "views": 17458,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: complete public prompt with product, camera transition, racing-city setting, and closing text"
  },
  {
    "id": "x-2087515920730370463",
    "title": "雨夜窗边拿着两只绿色瓶子的女子",
    "titleEn": "A Woman Holding Two Green Bottles by a Rainy Night Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现一名金发女子站在带雨滴的夜间窗户旁，一只绿色瓶子举在窗前，另一只放在窗台边，身后可见床铺与空调。 成片规格为 10 秒 · 640×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows a blonde woman by a rain-speckled window at night, holding one green bottle near the glass while another rests by the sill, with a bed and air conditioner behind her. Output: 10s · 640×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/dew_dew/status/2087515920730370463",
    "sourceLabel": "X 原帖 · @dew_dew",
    "author": "大谷屋",
    "publishedAt": "2026-08-12T12:25:54.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087515920730370463.mp4",
    "posterUrl": "/posters/x/2087515920730370463.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "640×960",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "父さんな、仕事辞めてMinimax h3でAI動画作って収益化して食っていこうと思っているんだ（嘘）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 130,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption does not clarify the visible content"
  },
  {
    "id": "x-2087514153095483839",
    "title": "爆炸战场上冲向装甲车的科幻士兵",
    "titleEn": "A Sci-Fi Soldier Advancing Toward Armored Vehicles on an Exploding Battlefield",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面从背后呈现一名身穿重型装备的科幻士兵，他朝散布装甲车、火焰与残骸的开阔战场前进。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster follows a heavily equipped sci-fi soldier from behind as he advances across an open battlefield scattered with armored vehicles, fire, and wreckage. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ojiji2025/status/2087514153095483839",
    "sourceLabel": "X 原帖 · @ojiji2025",
    "author": "ojiji",
    "publishedAt": "2026-08-12T12:18:52.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087514153095483839.mp4",
    "posterUrl": "/posters/x/2087514153095483839.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 really is good at making action videos! 💥",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 63,
      "views": 702,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption explicitly identifies the piece as action-focused"
  },
  {
    "id": "x-2087513770063171647",
    "title": "同一首帧下三种图生动态配置耗时对照",
    "titleEn": "Timing Three Image-to-Motion Setups from the Same First Frame",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "公开文案以同一首帧、896×512、24 FPS、五秒输出和三次种子中位数为条件，对比 Sage 二十步、Turbo 八步 Q4 与 LTX 2.5 的耗时。 成片规格为 5 秒 · 896×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption compares Sage at 20 steps, Turbo Q4 at eight steps, and LTX 2.5 using the same first frame, 896×512 resolution, 24 FPS, five-second output, and median runtime across three seeds. Output: 5s · 896×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087513770063171647",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T12:17:21.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087513770063171647.mp4",
    "posterUrl": "/posters/x/2087513770063171647.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "896×1536",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "同じ開始画像を使った、MiniMax H3とLTX-2.5のI2V比較\nRTX 5090／460W\n896×512／24fps／約5秒\nモデルを一度読み込んだあと、Seedを変更して3回生成した中央値\n・H3 Sage 20-step：65.380秒\n・H3 Turbo 8-step Q4 GGUF：53.505秒\n・LTX-2.5：13.755秒\nH3 Turbo 8-step Q4は、Sage 20-stepより18.2%高速\nLTXはやはり高速なのが強み！",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 359,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit image-to-motion comparison conditions and timings"
  },
  {
    "id": "x-2087513332375080987",
    "title": "文生动态三配置的速度与画质取舍",
    "titleEn": "Speed and Image-Quality Trade-offs Across Three Text-to-Motion Setups",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "公开文案在 RTX 5090、896×512、24 FPS、五秒输出和三次种子中位数的条件下，对比 Sage 二十步、Turbo 八步 Q4 与 LTX 2.5 的速度，并记录画质取舍。 成片规格为 5 秒 · 896×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption compares the speed of Sage at 20 steps, Turbo Q4 at eight steps, and LTX 2.5 on an RTX 5090 at 896×512, 24 FPS, and five seconds, using median runtime across three seeds and noting image-quality trade-offs. Output: 5s · 896×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087513332375080987",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T12:15:37.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087513332375080987.mp4",
    "posterUrl": "/posters/x/2087513332375080987.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "896×1536",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3とLTX-2.5のT2V比較\nRTX 5090／460W\n896×512／24fps／約5秒\nモデルを一度読み込んだあと、Seedを変更して3回生成した中央値\n・H3 Sage 20-step：62.125秒\n・H3 Turbo 8-step Q4 GGUF：51.490秒\n・LTX-2.5：13.380秒\nH3 Turbo 8-step Q4は、Sage 20-stepより17.1%高速\n画質とアニメらしい仕上がりはH3が好印象\nLTX-2.5は生成速度が圧倒的なのが売りね！",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 767,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit text-to-motion comparison conditions, timings, and stated quality trade-offs"
  },
  {
    "id": "x-2087509654712549880",
    "title": "黄昏街头漫步的怀旧动画片尾",
    "titleEn": "A Nostalgic Anime Ending Stroll at Dusk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将作品描述为一段黄昏时分穿行城市、偶尔绕路的漫步，并采用怀旧动画片尾风格。 成片规格为 218 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption describes a walk through the city at dusk with occasional detours, presented in the style of a nostalgic anime ending. Output: 218s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jr9956088/status/2087509654712549880",
    "sourceLabel": "X 原帖 · @jr9956088",
    "author": "たなか＠AIイラスト",
    "publishedAt": "2026-08-12T12:01:00.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087509654712549880.mp4",
    "posterUrl": "/posters/x/2087509654712549880.jpg",
    "duration": 218,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AIでMVを作ってみました！\n\n夕暮れの街を寄り道しながら歩く、少し懐かしいアニメED風をイメージしてみましたです🌇\n\n🎵 Music：Suno\n🎨 Illustration：niji・journey \n🎬 Animation：ComfyUI / MiniMax H3\n  #AIイラスト #AI動画 #Suno",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 39,
      "views": 540,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit dusk city walk and nostalgic anime-ending style"
  },
  {
    "id": "x-2087509408393928840",
    "title": "既有片段的配音与口型同步",
    "titleEn": "Dubbing and Lip Sync for Existing Footage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "公开文案说明这次 R2V 延续工作为既有片段加入配音，生成约耗时一千秒，并称口型能够对应语音。 成片规格为 5 秒 · 640×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says this R2V continuation adds dubbing to existing footage, took about 1,000 seconds to generate, and achieved lip movement aligned with the speech. Output: 5s · 640×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/malariadx/status/2087509408393928840",
    "sourceLabel": "X 原帖 · @malariadx",
    "author": "びおめた",
    "publishedAt": "2026-08-12T12:00:01.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087509408393928840.mp4",
    "posterUrl": "/posters/x/2087509408393928840.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3(R2V)の続き、既存の動画に「アテレコ」してみました、生成時間は約1000秒、ちゃんとリップシンクしてます\n#AI動画 #AIart",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 306,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit R2V dubbing workflow, runtime, and stated lip-sync result"
  },
  {
    "id": "x-2087509405939937374",
    "title": "千点额度完成的多工具宣传短片",
    "titleEn": "A Multi-Tool Promotional Short Made with 1,000 Credits",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明这支宣传片风格作品约使用一千点额度，并列出 GPT、Adobe Express、Flova AI、Suno 与生成模型组成的制作链路。 成片规格为 50 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says this promotional-style short used about 1,000 credits and lists a production chain involving GPT, Adobe Express, Flova AI, Suno, and the generation model. Output: 50s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ronbox17/status/2087509405939937374",
    "sourceLabel": "X 原帖 · @ronbox17",
    "author": "ろんぼ",
    "publishedAt": "2026-08-12T12:00:01.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087509405939937374.mp4",
    "posterUrl": "/posters/x/2087509405939937374.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "music-performance",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "先のイベントでいただいたクレジットでPVみたいなものを作ってみました。\n元絵はGPTを使用しています。編集はadobeexpressです。\nsunoも使えたから曲は、flovaaiを使わせていただき作りました。１０００クレジットくらいで作成。\n動画の生成モデルはMiniMax H3です。\n良い経験ができました。ありがとうございます！！\n@flovaai\n@Flovaai_Japan\n#flovaai",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 13,
      "views": 254,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit promotional-short format, toolchain, and credit usage"
  },
  {
    "id": "x-2087507824679280957",
    "title": "房间里的暑假与半眯眼少女",
    "titleEn": "Summer Break Indoors with a Half-Lidded Girl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "公开文案描述一名戴单侧圆环耳饰、半眯眼的少女在房间里度过暑假，并称少量参考图也维持了她的眼部特征。 成片规格为 26 秒 · 832×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption describes a half-lidded girl with a single hoop earring spending summer break in her room, and says her eye shape remained consistent with only a small number of reference images. Output: 26s · 832×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kaerimichinotzk/status/2087507824679280957",
    "sourceLabel": "X 原帖 · @kaerimichinotzk",
    "author": "かえりみちのつづき",
    "publishedAt": "2026-08-12T11:53:44.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087507824679280957.mp4",
    "posterUrl": "/posters/x/2087507824679280957.jpg",
    "duration": 26,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "夏休みを（お部屋で）満喫している片側リングボブジト目ちゃん\n\nMinimax H3、参照画像少なくてもジト目をかわいく維持する性能が高くて助かってます",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 66,
      "views": 1926,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit character features, indoor summer-break setting, and stated reference consistency"
  },
  {
    "id": "x-2087507568046514509",
    "title": "单参考静态造型中的面部锁定对照",
    "titleEn": "Face Locking Compared in a Single-Reference Static Look",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者刻意采用偏静态造型与同一张参考图进行对照，并称另一方案的动作和转场更好，而这一方案在细微面部锁定上更强。 成片规格为 8 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author deliberately used a mostly static look with the same single reference, finding the other setup stronger in motion and transitions while this setup held finer facial identity more closely. Output: 8s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/crtvst_sojeong/status/2087507568046514509",
    "sourceLabel": "X 原帖 · @crtvst_sojeong",
    "author": "소정",
    "publishedAt": "2026-08-12T11:52:42.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087507568046514509.mp4",
    "posterUrl": "/posters/x/2087507568046514509.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "광고❌\n\n가성비 모델로 알려진 MiniMax H3 써보셨나요? 2K 기준으로 씨댄스보다 33% 더 저렴하다고 해서 사용해봤는데 나쁘지 않았습니다!•••🍀\n\n 이전 버전이 구현도가 뛰어나다고 해서 기대를 했습니다. @MiniMax_AI 에서 소구점으로 내민 포인트가 \n• 높은 구현도\n• 저렴한 가격\n• 멀티모달 레퍼런스 (자연어 명령)\n요렇게인데, 우선 1,2는 합격입니다😃\n\n 일부러 테스트 영상을 정적으로 만들어보았습니다. 저는 룩북 작업을 주로 하기에 화려한 모션보다 얼굴 고정이 더 중요하거든요😅\n\n 사용 결과, 제 기준 씨댄스보다 모델 얼굴 고정이 잘 되었습니다. 레퍼런스 한장 넣고 같은 프롬프트로 돌렸을 때, 모션이나 연결 동작은 씨댄스가 출중하나 미세 이미지 락킹은 MiniMax H3가 훨씬 좋았습니다. \n자주 쓸 것 같아요•••🍀\n\n사용 가능 플랫폼: Hailuo ai, mitte ai, Topview, higgsfield, newtake 등",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 21,
      "views": 1495,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit single-reference static-look comparison and stated face-locking result"
  },
  {
    "id": "x-2087506873297568112",
    "title": "科幻奇幻平地世界长片",
    "titleEn": "A Sci-Fi Fantasy Feature Set in Flat-Earth Lore",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确将作品描述为一部长篇科幻奇幻电影，故事设定围绕平地世界传说展开。 成片规格为 2887 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly describes the work as a full-length sci-fi fantasy feature built around flat-earth lore. Output: 2887s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SportsMemesGuy/status/2087506873297568112",
    "sourceLabel": "X 原帖 · @SportsMemesGuy",
    "author": "SportsMemes",
    "publishedAt": "2026-08-12T11:49:57.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087506873297568112.mp4",
    "posterUrl": "/posters/x/2087506873297568112.jpg",
    "duration": 2887,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "sci-fi",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made a full length feature film with #MiniMax H3 #ai @ComfyUI @MiniMax_AI Sci Fi fantasy flat earth lore",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 78,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit feature-length format, genre, and lore setting"
  },
  {
    "id": "x-2087506618506465521",
    "title": "CJ 入侵",
    "titleEn": "CJ Invasion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案仅将这段内容命名为“CJ 入侵”，没有进一步说明角色身份、动作过程或具体场景。 成片规格为 15 秒 · 1584×1200。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption labels the content “CJ Invasion” without further identifying the character, sequence of actions, or setting. Output: 15s · 1584×1200. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sadlemonjuice/status/2087506618506465521",
    "sourceLabel": "X 原帖 · @sadlemonjuice",
    "author": "Socratech",
    "publishedAt": "2026-08-12T11:48:56.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087506618506465521.mp4",
    "posterUrl": "/posters/x/2087506618506465521.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1584×1200",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 local CJ invasion",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 105,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit scenario label, with no additional scene detail"
  },
  {
    "id": "x-2087505758799929434",
    "title": "太空视角下的北美洲与地球弧面",
    "titleEn": "North America and Earth's Curvature Seen from Space",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面从太空呈现地球弧面，北美洲、墨西哥湾与周边海域清晰可见，云带横跨蓝色海洋与陆地。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster presents Earth's curvature from space, with North America, the Gulf of Mexico, and surrounding waters visible beneath cloud bands crossing land and blue ocean. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/0xaditin/status/2087505758799929434",
    "sourceLabel": "X 原帖 · @0xaditin",
    "author": "Adi",
    "publishedAt": "2026-08-12T11:45:31.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087505758799929434.mp4",
    "posterUrl": "/posters/x/2087505758799929434.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Experimenting with Minimax H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 76,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption adds no content details"
  },
  {
    "id": "x-2087505225729118446",
    "title": "MIO 的摇滚比利吉他独奏",
    "titleEn": "MIO's Rockabilly Guitar Solo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确写出 MIO 演奏摇滚比利风格的吉他独奏，并特别提到指法与拨弦动作。 成片规格为 10 秒 · 2304×3584。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly identifies MIO performing a rockabilly guitar solo and specifically calls out the fingering and picking motions. Output: 10s · 2304×3584. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/8f8d54a4772647b/status/2087505225729118446",
    "sourceLabel": "X 原帖 · @8f8d54a4772647b",
    "author": "MJ-AI",
    "publishedAt": "2026-08-12T11:43:24.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087505225729118446.mp4",
    "posterUrl": "/posters/x/2087505225729118446.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "2304×3584",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MIO🐱\n\nこんばんは🌙\nローカルMiniMax H3でMIOちゃんのロカビリー調でギターソロ作ってみた(*'▽')ノ✨\n運指とピッキングも中々いい感じにできててちゃんと弾いてる感あるなぁ❣\n\n#AIイラスト #AIart #ComfyUI",
    "engagement": {
      "replies": 5,
      "reposts": 9,
      "likes": 76,
      "views": 476,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit performer, musical style, and guitar-playing action"
  },
  {
    "id": "x-2087505124357001469",
    "title": "白石杏",
    "titleEn": "An Shiraishi",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案只明确写出角色白石杏，没有进一步描述动作或场景。 成片规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly identifies the character An Shiraishi but does not provide an action or setting. Output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Oxide39/status/2087505124357001469",
    "sourceLabel": "X 原帖 · @Oxide39",
    "author": "ミクoxide☆NEO彡",
    "publishedAt": "2026-08-12T11:43:00.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087505124357001469.mp4",
    "posterUrl": "/posters/x/2087505124357001469.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3白石杏\n#白石杏\n#プロセカ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 133,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit character identity only"
  },
  {
    "id": "x-2087504764221432236",
    "title": "LTX 2.5 一百万像素五秒内容耗时四十一秒",
    "titleEn": "LTX 2.5 Produces Five Seconds at One Megapixel in 41 Seconds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明 LTX 2.5 以一百万像素规格完成五秒内容共耗时四十一秒，并强调其生成速度。 成片规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says LTX 2.5 produced five seconds at a one-megapixel setting in 41 seconds and highlights the resulting speed. Output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CKao29/status/2087504764221432236",
    "sourceLabel": "X 原帖 · @CKao29",
    "author": "ちゃんK🌸日本を旅する女の子🎀AI美女",
    "publishedAt": "2026-08-12T11:41:34.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087504764221432236.mp4",
    "posterUrl": "/posters/x/2087504764221432236.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX2.5の生成速度がめちゃくちゃ早くて笑った\n41秒で、1MPの5秒動画が完成してしまう\nもうこれクラウドのレベルで早い\n\nMiniMAX H3やべえって言ってたのを超えてきた気がする",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 23,
      "views": 804,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit model, resolution scale, output duration, and runtime"
  },
  {
    "id": "x-2087504117673377896",
    "title": "护摩火焰中的龙神",
    "titleEn": "A Dragon Deity in Goma Ritual Flames",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将内容描述为一场护摩祈祷，并以在火焰中看见龙作为主题设定。 成片规格为 10 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption describes a goma fire prayer and frames seeing a dragon in the flames as its central premise. Output: 10s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/55suekichi/status/2087504117673377896",
    "sourceLabel": "X 原帖 · @55suekichi",
    "author": "すえきち｜インスタ×占い",
    "publishedAt": "2026-08-12T11:39:00.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087504117673377896.mp4",
    "posterUrl": "/posters/x/2087504117673377896.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3たのし〜\n護摩祈祷動画も作れた！\n\n炎のなかに龍が見えた人には\n龍神さまからのメッセージが受け取れます！\n\nって🐶ス夕であげたらスピ度の高いひと集まるかな？",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 169,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit goma ritual, flames, and dragon premise"
  },
  {
    "id": "x-2087503171220541850",
    "title": "五分钟完成的一次生成",
    "titleEn": "A Generation Completed in Five Minutes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案只说明这次生成在五分钟内完成，没有进一步描述内容主体、动作或场景。 成片规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption states that this generation was completed in five minutes but does not further describe its subject, action, or setting. Output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/okatapioka/status/2087503171220541850",
    "sourceLabel": "X 原帖 · @okatapioka",
    "author": "タピオカ(おかぴー)",
    "publishedAt": "2026-08-12T11:35:14.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087503171220541850.mp4",
    "posterUrl": "/posters/x/2087503171220541850.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "いま流行りのMiniMax H3で動画生成したら５分でこれできた、怖い時代になった",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 230,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit five-minute production claim"
  },
  {
    "id": "x-2087500992573047114",
    "title": "Ross 与一名女子坐在客厅沙发上",
    "titleEn": "Ross Sits with a Woman on a Living-Room Sofa",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现 Ross 与一名穿绿色上衣的女子并排坐在白色沙发上，后方是带餐桌与开放式厨房的彩色客厅。 成片规格为 13 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows Ross seated beside a woman in a green top on a white sofa, with a colorful living room, dining table, and open kitchen behind them. Output: 13s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TechieBySA/status/2087500992573047114",
    "sourceLabel": "X 原帖 · @TechieBySA",
    "author": "TechieSA",
    "publishedAt": "2026-08-12T11:26:35.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087500992573047114.mp4",
    "posterUrl": "/posters/x/2087500992573047114.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ross had no response\n\nMinimax H3 prompt👇🏻",
    "engagement": {
      "replies": 12,
      "reposts": 3,
      "likes": 48,
      "views": 3558,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption explicitly identifies Ross"
  },
  {
    "id": "x-2087500512681926871",
    "title": "AI 辅助下的提示词结构摸索",
    "titleEn": "Exploring Prompt Structures with AI Assistance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案记录作者逐渐摸索出提示词结构，并使用 AI 协助编写，同时提到部分请求会触发拒绝。 成片规格为 7 秒 · 608×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption records the author gradually finding a prompt structure and using AI assistance to write it, while noting that some requests trigger refusals. Output: 7s · 608×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NHimakiti/status/2087500512681926871",
    "sourceLabel": "X 原帖 · @NHimakiti",
    "author": "ひまきち＠himakiti",
    "publishedAt": "2026-08-12T11:24:40.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087500512681926871.mp4",
    "posterUrl": "/posters/x/2087500512681926871.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "608×1056",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 試行錯誤中～\nなんとなくプロンプトの型みたいなのが分かってきた感じなのと、プロンプトはAIに補助してもらって作れるから楽しいですねこれ！\n\nただ、ムフフ度が上がるとAI拒否っちゃうので自力で考えないといけないのしんどいですが😂\n...もしかしてやりようはある？",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 12,
      "views": 1596,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit prompt-structure exploration and AI-assisted writing workflow"
  },
  {
    "id": "x-2087500308620664983",
    "title": "用自制 Skill 复刻矢量绘制效果",
    "titleEn": "Recreating a Vector-Drawing Effect with a Custom Skill",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者使用自制 Skill 复刻近期流行的矢量绘制效果。 成片规格为 27 秒 · 1024×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author used a custom Skill to recreate a recently popular vector-drawing effect. Output: 27s · 1024×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/veoxft/status/2087500308620664983",
    "sourceLabel": "X 原帖 · @veoxft",
    "author": "林克 Link",
    "publishedAt": "2026-08-12T11:23:52.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087500308620664983.mp4",
    "posterUrl": "/posters/x/2087500308620664983.jpg",
    "duration": 27,
    "aspectRatio": "portrait",
    "resolution": "1024×1152",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "最近这个矢量绘制的效果爆火，我帮朋友用Minimax H3 + 用自己做的 skill 复刻了一版。效果非常好\n#Minimax",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 105,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit vector-drawing effect and custom-Skill recreation workflow"
  },
  {
    "id": "x-2087499749729665365",
    "title": "厨房柜台旁挥手的围裙女子",
    "titleEn": "An Apron-Wearing Woman Waves Beside a Kitchen Counter",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿浅色上衣与围裙的女子站在厨房柜台旁向镜头挥手，随后保持相近位置并轻微改变姿态。 成片规格为 10 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a light top and apron waves toward the camera beside a kitchen counter and then remains in similar framing with small posture changes. Output: 10s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HackCooking/status/2087499749729665365",
    "sourceLabel": "X 原帖 · @HackCooking",
    "author": "Cooking life hack",
    "publishedAt": "2026-08-12T11:21:38.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087499749729665365.mp4",
    "posterUrl": "/posters/x/2087499749729665365.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 プロンプトはAIに作ってもらわないと無理なぐらい複雑ｗ\nでも、生成時間もコストもかなり良い感じになってきた＾＾\n明日の朝にはGrokが復活するので、WebUIからコントロールしてみます！",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 769,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087496715679232504",
    "title": "8GB 显存笔记本上的中文多模态工作流",
    "titleEn": "A Chinese Multimodal Workflow on an 8GB Laptop GPU",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "公开文案说明 RTX 5060 笔记本以 W4A8 主模型和 Q2 文本编码器跑通中文文生与图生流程，并称五秒、1980×1080、24 FPS 内容约需五分钟完成。 成片规格为 137 秒 · 1728×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says an RTX 5060 laptop ran Chinese text- and image-conditioned workflows with a W4A8 main model and Q2 text encoder, producing five seconds at 1980×1080 and 24 FPS in about five minutes. Output: 137s · 1728×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bi_9527zx/status/2087496715679232504",
    "sourceLabel": "X 原帖 · @bi_9527zx",
    "author": "DeFi狙击手 | Ai🕊️｜Gate美股0费率",
    "publishedAt": "2026-08-12T11:09:35.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087496715679232504.mp4",
    "posterUrl": "/posters/x/2087496715679232504.jpg",
    "duration": 137,
    "aspectRatio": "landscape",
    "resolution": "1728×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "8G独显笔记本跑通 MiniMax H3！\n\n本地 ComfyUI 中文工作流已跑通：\n\n🎉中文文生视频、图生视频\n\n🎉5秒 1980×1080，24FPS\n\n🎉原生中文对白＋口型同步\n\n🎉同步生成环境音与背景音乐\n\n🎉W4A8 主模型＋Q2 文本编码器\n\nRTX 5060 Laptop 可运行，5分钟生成5秒钟视频。\n\n8GB 显存也能真正体验原生多模态视频了！",
    "engagement": {
      "replies": 56,
      "reposts": 7,
      "likes": 116,
      "views": 16359,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit 8GB laptop workflow, quantization, modalities, output settings, and runtime"
  },
  {
    "id": "x-2087494761225806152",
    "title": "让踢踏舞插画动起来",
    "titleEn": "Animating a Tap-Dance Illustration",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者将此前为踢踏舞主题创作的插画制作成动态内容，并将其视为提示词入门阶段的尝试。 成片规格为 5 秒 · 736×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author animated an earlier tap-dance-themed illustration as an early attempt while learning motion prompts. Output: 5s · 736×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SJ_STOPTHESTEAL/status/2087494761225806152",
    "sourceLabel": "X 原帖 · @SJ_STOPTHESTEAL",
    "author": "SJ（セイジ）＠AI絵｜AWEB",
    "publishedAt": "2026-08-12T11:01:49.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087494761225806152.mp4",
    "posterUrl": "/posters/x/2087494761225806152.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "736×576",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "dance",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MiniMaxH3 #AIart （動かしてみた）\nこんばんは🌜\n先のタップダンスの日のイラストを動かしてみました。\nリプの動かしたものがあって、僕も最初はこれにしました。\nうちの子を動かすのは僕が先だったのに・・・😂\n\nこれからMiniMax H3の導入です。\nXのアプリが新しく更新されて、メディアから動画と画像が分離されました。\n同時にMiniMax H3、そしてLTX 2.5というローカル動画モデルが発表されたんです。\n\nテストとして使ったもので、まだ動画プロンプトは初心者レベルです。\nたまに動画も投稿しますので、よろしくおねがいします🤗",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 14,
      "views": 427,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit source illustration, tap-dance theme, and animation action"
  },
  {
    "id": "x-2087494601997476236",
    "title": "十五秒音乐片段的三十分钟生成记录",
    "titleEn": "A 15-Second Music-Backed Clip Generated in About 30 Minutes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案记录了 864×480、24 FPS、十五秒且带音乐的输出参数，并称总生成时间约为三十分钟。 成片规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption records output settings of 864×480, 24 FPS, 15 seconds, and included music, with a total generation time of about 30 minutes. Output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nya_602/status/2087494601997476236",
    "sourceLabel": "X 原帖 · @nya_602",
    "author": "デジ猫@ITエンジニャ〜🐱✨",
    "publishedAt": "2026-08-12T11:01:11.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087494601997476236.mp4",
    "posterUrl": "/posters/x/2087494601997476236.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で動画生成テストしてみた\n864×480 / 24fps / 15秒 / 音楽付き\n総生成時間は約30分\n\n次はローカルLLM行ってみよう！",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 266,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit resolution, frame rate, duration, music inclusion, and runtime"
  },
  {
    "id": "x-2087494333133963544",
    "title": "巨龙背负披风人物飞过雪山",
    "titleEn": "A Giant Dragon Carries a Caped Figure Above Snowy Mountains",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现一头黑红色巨龙飞越云雾与雪山，龙背上坐着一名披着深红披风的小型人物。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows a huge black-and-red dragon flying above clouds and snowy mountains with a small figure in a dark red cape seated on its back. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JayKay65220066/status/2087494333133963544",
    "sourceLabel": "X 原帖 · @JayKay65220066",
    "author": "Jay Kay",
    "publishedAt": "2026-08-12T11:00:07.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087494333133963544.mp4",
    "posterUrl": "/posters/x/2087494333133963544.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Happy Hump Day! 🐲🧀🐪\n\nMade with MiniMax H3 video model from @Hailuo_AI!\n\n#WednesdayMotivation #Wednesdayfeeling #Wednesdayvibes #Wednesdayfun Good Wednesday",
    "engagement": {
      "replies": 45,
      "reposts": 15,
      "likes": 114,
      "views": 3444,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption includes a dragon symbol but no further scene detail"
  },
  {
    "id": "x-2087492138611450350",
    "title": "音轨驱动的动态舞蹈表演",
    "titleEn": "A Dynamic Dance Performance Driven by an Audio Track",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者上传音轨作为配乐，再让模型围绕节奏与能量构建动态舞蹈画面。 成片规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author uploaded an audio track as the soundtrack and had the model build a dynamic dance performance around its rhythm and energy. Output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SaniAiTech/status/2087492138611450350",
    "sourceLabel": "X 原帖 · @SaniAiTech",
    "author": "Sani Ai Tech",
    "publishedAt": "2026-08-12T10:51:24.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087492138611450350.mp4",
    "posterUrl": "/posters/x/2087492138611450350.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tried MiniMax H3’s “Audio as Soundtrack” feature 🎧🔥\n\nI uploaded an audio track and let H3 build the visuals around it — turning the rhythm and energy into a dynamic dance performance.\n\nThe audio stays as the soundtrack while H3 generates the visuals around it.\n\nAudio → Motion → Video.\nPretty impressive. 👀\n\n#MiniMaxH3 #AIVideo #GenerativeAI",
    "engagement": {
      "replies": 28,
      "reposts": 28,
      "likes": 180,
      "views": 916,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit audio-to-motion workflow and dance-performance subject"
  },
  {
    "id": "x-2087492034005524831",
    "title": "樱花公园野餐",
    "titleEn": "A Picnic in a Cherry Blossom Park",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案直接将场景命名为樱花公园野餐。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption directly names the scene as a picnic in a cherry blossom park. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2087492034005524831",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-12T10:50:59.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087492034005524831.mp4",
    "posterUrl": "/posters/x/2087492034005524831.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cherry Blossom Park Picnic \n\nCreated this using MiniMax H3 on Hailuo_AI",
    "engagement": {
      "replies": 18,
      "reposts": 1,
      "likes": 29,
      "views": 1509,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit cherry-blossom park picnic setting"
  },
  {
    "id": "x-2087491214816026982",
    "title": "戴皇冠的粉发猫耳少女比出 V 字手势",
    "titleEn": "A Crowned Pink-Haired Cat-Eared Girl Makes a V Sign",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面呈现一名粉色长发、紫色眼睛与猫耳的动画少女，她头戴小皇冠，身穿带领结的制服与深色披肩，在林木背景前举起 V 字手势。 成片规格为 10 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster shows an animated girl with long pink hair, purple eyes, and cat ears, wearing a small crown, bow-tied uniform, and dark capelet while raising a V sign before a wooded background. Output: 10s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NudeNyang_VRC/status/2087491214816026982",
    "sourceLabel": "X 原帖 · @NudeNyang_VRC",
    "author": "Nude🐾",
    "publishedAt": "2026-08-12T10:47:43.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087491214816026982.mp4",
    "posterUrl": "/posters/x/2087491214816026982.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3\n✌️",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 15,
      "views": 300,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption includes a matching hand-sign symbol"
  },
  {
    "id": "x-2087490058584457396",
    "title": "单张图片生成短动态片段的图生实验",
    "titleEn": "Image-to-Motion Experiment from a Single Still",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "这项图生动态实验从一张图片出发，在本机生成多个短动态片段，并配有完整操作指南。 成片规格为 5 秒 · 1048×552。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "This image-to-motion experiment starts from one still image, produces several short motion clips on-device, and accompanies them with a complete setup guide. Output: 5s · 1048×552. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Manz/status/2087490058584457396",
    "sourceLabel": "X 原帖 · @Manz",
    "author": "Manz 🇮🇨⚡👾",
    "publishedAt": "2026-08-12T10:43:08.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087490058584457396.mp4",
    "posterUrl": "/posters/x/2087490058584457396.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1048×552",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "¡Nuevo post!\n\nEn esta guía explico como hacer I2V (image-to-video). Partiendo de una imagen, conseguir pequeños fragmentos de video, todo en local (ejemplo con el modelo Minimax H3):\n\nhttps://ai.manz.dev/learn/setup/sdcpp-i2v/",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 24,
      "views": 1334,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit image-to-motion experiment, single-image input, short outputs, on-device workflow, and linked guide"
  },
  {
    "id": "x-2087490047679607113",
    "title": "舞台前举拳回应麦克风女子的人群",
    "titleEn": "A Crowd Raises Their Fists Toward a Woman with a Microphone",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "封面从持麦克风女子的背后望向一群站在室内舞台前的人，前排多人穿学生制服，整个人群同时举起拳头；来源文案说明这是未采用的片段，并记录素材图降至 30MB 以下后流程才成功运行。 成片规格为 9 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The poster looks past a woman holding a microphone toward a crowd standing before an indoor stage, with several people in school uniforms at the front and the group raising their fists together; the caption calls it an unused segment and says the workflow succeeded after reducing the source image below 30MB. Output: 9s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Sally__37/status/2087490047679607113",
    "sourceLabel": "X 原帖 · @Sally__37",
    "author": "サーナ",
    "publishedAt": "2026-08-12T10:43:05.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087490047679607113.mp4",
    "posterUrl": "/posters/x/2087490047679607113.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3、ずっとエラーで生成できなくて原因探ってたら、アセット画像のファイル制限が30MB以下？らしく、解像度下げたら生成できた😭\n画像はきれいな方がいい！と思っていたけどサイズ大きすぎてもだめなんだなぁ\n4Kにまで上げる人なんて私以外いないかもしれませんが（笑）困ってる方いましたら解像度下げてみてください🙏\n没カット置いておきます（MiniMax H3）↓",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 155,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "poster-frame; sourceCaption explicitly supplies the unused-segment status and source-image size fix"
  },
  {
    "id": "x-2087488932812558714",
    "title": "《Nora Links》后半篇章的视觉风格探索",
    "titleEn": "Visual Style Exploration for the Second Half of Nora Links",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《Nora Links》在完成前四集后为第五集及后续篇章探索新视觉风格，并以同一指令在另一模型中生成结果供对照。 成片规格为 16 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "After completing its first four episodes, Nora Links explores a new visual direction for episode five and beyond, with a same-prompt result from another model provided for comparison. Output: 16s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Nokosu_kansoku/status/2087488932812558714",
    "sourceLabel": "X 原帖 · @Nokosu_kansoku",
    "author": "Nokosu",
    "publishedAt": "2026-08-12T10:38:39.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087488932812558714.mp4",
    "posterUrl": "/posters/x/2087488932812558714.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「Nora Links」は、第1話から第4話までの制作を通して、目指すビジュアルの方向性がほぼ固まりました。現在は、第5話以降の後半パートに向けて、新たなルックを模索していましたが大体方向性が決まりました🤩\n\nこの動画のスタイルで頑張って参ります👍\n\nSeedance 2.5やMiniMax H3なども試した結果、今後はSeedance 2.0とMiniMax H3を組み合わせた制作フローで進めていこうと考えています。\n\nリプに同じプロンプトで作ったSeedance2.5の動画も置いてあるので、見比べてみてください😆\n@TapNow_AI #TapNow",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 29,
      "views": 603,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit work title, completed episode range, second-half visual-style exploration, and same-prompt model comparison"
  },
  {
    "id": "x-2087486755482312838",
    "title": "水母少女克拉娜的日常《温柔的现行犯》",
    "titleEn": "Jellyfish Girl Kurana in The Gentle Culprit",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这则日常短篇名为《温柔的现行犯》，主角是水母少女克拉娜。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "This everyday short is titled The Gentle Culprit and centers on Kurana, a jellyfish girl. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087486755482312838",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T10:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087486755482312838.mp4",
    "posterUrl": "/posters/x/2087486755482312838.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n日常エピソード\n『やさしい現行犯』\nクラゲ少女のクラーナちゃん",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 16,
      "views": 505,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit episode title, everyday-episode format, named jellyfish-girl subject"
  },
  {
    "id": "x-2087485291913429327",
    "title": "写实人物 LoRA 的首次加载实验",
    "titleEn": "First-Load Experiment with a Realistic-People LoRA",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "作者在 Apple M4 Max 48GB 设备上通过 Phosphene 首次加载写实人物 LoRA，并覆盖文生、图生与参考驱动模式。 成片规格为 5 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author first-loads a realistic-people LoRA through Phosphene on an Apple M4 Max with 48GB of memory, covering text-, image-, and reference-driven modes. Output: 5s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chuwujiuyue/status/2087485291913429327",
    "sourceLabel": "X 原帖 · @chuwujiuyue",
    "author": "chuwujiuyue",
    "publishedAt": "2026-08-12T10:24:11.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087485291913429327.mp4",
    "posterUrl": "/posters/x/2087485291913429327.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "My Apple M4 Max 48G is finally running local MiniMax H3 through Phosphene! \nFirst‑load test of this LoRA:\n h3 realism people t2v i2v r2v 🚀",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 104,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit first-load LoRA experiment, realistic-people scope, device, runtime layer, and three input modes"
  },
  {
    "id": "x-2087485114548969494",
    "title": "低分辨率生成后放大至 4K 的二十秒实验",
    "titleEn": "Twenty-Second Low-Resolution-to-4K Upscaling Experiment",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "二十秒内容先以 1088×608 低分辨率生成，再放大至 4K；作者同时记录了细节控制较难的问题。 成片规格为 20 秒 · 3840×2148。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A twenty-second output is generated first at 1088×608 and then upscaled to 4K, with the author also noting difficulty controlling fine detail. Output: 20s · 3840×2148. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2087485114548969494",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-12T10:23:29.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087485114548969494.mp4",
    "posterUrl": "/posters/x/2087485114548969494.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "3840×2148",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Motion Context 20s\n1088x608 低解像度生成からの4K Upscale\n\nAIイラストみたいには書き込みコントロールは上手くいかない。と言うか難しい。VRAMが許すのであれば高解像度生成の方が楽だし良いっすね🙄",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 11,
      "views": 1795,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit twenty-second duration, 1088x608 source resolution, 4K upscale, and detail-control observation"
  },
  {
    "id": "x-2087484310437744901",
    "title": "乌克兰民歌《Їхав козак за Дунай》",
    "titleEn": "Ukrainian Folk Song Їхав козак за Дунай",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "作品以乌克兰民歌《Їхав козак за Дунай》为主题，并使用参考图与角色设定图维持角色一致性。 成片规格为 32 秒 · 2048×1556。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The work is built around the Ukrainian folk song Їхав козак за Дунай and uses a reference image plus a character sheet to maintain character consistency. Output: 32s · 2048×1556. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ModelCollapse38/status/2087484310437744901",
    "sourceLabel": "X 原帖 · @ModelCollapse38",
    "author": "Encodium",
    "publishedAt": "2026-08-12T10:20:17.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087484310437744901.mp4",
    "posterUrl": "/posters/x/2087484310437744901.jpg",
    "duration": 32,
    "aspectRatio": "landscape",
    "resolution": "2048×1556",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ukrainian folk song «Їхав козак за Дунай»\n\nMinimax H3\nReference image + character sheet for consistency\n\n@MiniMax_AI @ComfyUI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 99,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit song title, Ukrainian folk-song form, reference image, and character-sheet consistency method"
  },
  {
    "id": "x-2087479769378312616",
    "title": "角色设定图与 MP3 驱动的整曲动态生成",
    "titleEn": "Song-Length Motion Generated from a Character Sheet and MP3",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "程序在 ComfyUI 运行时读取角色设定图与 MP3，反复调用生成流程完成整首歌长度的内容，但作者指出对口型仍不准确。 成片规格为 242 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "While ComfyUI is running, a program reads a character sheet and an MP3 and repeatedly invokes the generation workflow for a full song, though the author notes that lip synchronization remains inaccurate. Output: 242s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mikumiku_aloha/status/2087479769378312616",
    "sourceLabel": "X 原帖 · @mikumiku_aloha",
    "author": "mikumiku_aloha",
    "publishedAt": "2026-08-12T10:02:15.000Z",
    "addedAt": "2026-08-13T00:38:33.431Z",
    "mediaUrl": "/media/x-2087479769378312616.mp4",
    "posterUrl": "/posters/x/2087479769378312616.jpg",
    "duration": 242,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ComfyUIを起動している状態でキャラクター設定ファイルの画像とMP3の音楽を指定すればComfyUIで動くMiniMax H3を繰り返し使用して一曲分の動画を生成するプログラムを生成して\n\nとChatGPT workに指示して一発では動かなかったけれど何度か修正したら出来た。\n\nリップシンクとかぜんぜん合っていないけれどこの程度の指示の後の多少根気よく動くまでエラー修正に付き合ったら出来ちゃうんだな。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 157,
      "snapshotAt": "2026-08-13T00:38:33.431Z"
    },
    "approvedAt": "2026-08-13T00:38:33.431Z",
    "editorialBasis": "sourceCaption: explicit character-sheet and MP3 inputs, song-length automated workflow, repeated generation, and lip-sync limitation"
  },
  {
    "id": "x-2087479263985287351",
    "title": "黑黄游戏手柄与霓虹光轨",
    "titleEn": "A Black-and-Yellow Game Controller with Neon Light Trails",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一个黑黄游戏手柄在白色与深色背景间切换，并以正面、斜侧与局部近景出现。 成片规格为 15 秒 · 1440×1620。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-and-yellow game controller moves between white and dark backdrops and appears in front, three-quarter, and close-up views. Output: 15s · 1440×1620. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CharaspowerAI/status/2087479263985287351",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-08-12T10:00:14.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087479263985287351.mp4",
    "posterUrl": "/posters/x/2087479263985287351.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1620",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This is crazy.\nThe Marvel’s Wolverine PS5 controller was barely revealed…\nand it’s already starring in AI-powered motion design. 🤯\nThis is exactly what I mean when I say Minimax H3 is insanely powerful for creative work.\n@Hailuo_AI, #MiniMaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 10,
      "views": 1368,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087479235615031569",
    "title": "黄昏港湾上空的高楼、吊车与船只",
    "titleEn": "High-Rises, Cranes, and Harbor Vessels at Dusk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "高处视角横移经过密集高楼与建筑吊车，随后打开到黄昏港湾及水面上的多艘船只。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An elevated view traverses dense high-rises and construction cranes before opening onto a harbor with several vessels at dusk. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LudovicCreator/status/2087479235615031569",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-12T10:00:07.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087479235615031569.mp4",
    "posterUrl": "/posters/x/2087479235615031569.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "stylized"
    ],
    "scenes": [
      "food",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is a very versatile model in my opinion.\n\nPrompt  :\n\nAerial push toward a dense vertical harbor city as a storm front rolls in at dusk  towers packed so tightly the streets disappear into shadowed canyons, windows lighting one by one against bruised purple clouds, scaffolding rattling around an unfinished high-rise, neon restaurant signs flickering through the wind, ferries crossing a dark jade harbor under sheets of distant rain. Air charged and humid, every sign haloed by approaching weather. Density as pressure. Horns, thunder, harbor water, a city bracing but still alive\n\nMade  in @LumaLabsAI #LumaCPP",
    "engagement": {
      "replies": 18,
      "reposts": 5,
      "likes": 70,
      "views": 2183,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087478997974208588",
    "title": "烟雾锻造间里的头盔铁锤工人",
    "titleEn": "A Helmeted Worker with a Hammer in a Smoky Forge",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名戴头盔、系围裙并拿着铁锤的工人站在充满烟雾与橙色火光的锻造空间中。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A helmeted, apron-wearing worker holds a hammer in a forge-like interior filled with smoke, tools, chains, and orange firelight. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/exploraX_/status/2087478997974208588",
    "sourceLabel": "X 原帖 · @exploraX_",
    "author": "m0h",
    "publishedAt": "2026-08-12T09:59:11.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087478997974208588.mp4",
    "posterUrl": "/posters/x/2087478997974208588.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🚨 Minimax H3 + GPT Image 2\n\nthis cinematic character reveal was generated from just two prompts.\n\nGPT Image 2 created the character, then Minimax H3 turned it into the final video.\n\ntotal cost: $1.20.\n\nyou’ll find the exact prompts i used in the comments.",
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 15,
      "views": 2360,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087478882320715791",
    "title": "Atlas for Mac 清理工具广告",
    "titleEn": "An Advertisement for the Atlas Mac Cleaner",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这支 46 秒广告为开源 Mac 清理工具 Atlas 制作，以“你的 Mac，从不需要被恐吓”为核心文案，组合六段电影感画面、中文配音、音乐与 HTML/GSAP 合成。 成片规格为 46 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "This 46-second advertisement promotes the open-source Atlas Mac cleaner around the line “Your Mac never needs to be frightened,” combining six cinematic sequences, Chinese narration, music, and HTML/GSAP compositing. Output: 46s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lizikk_zhu/status/2087478882320715791",
    "sourceLabel": "X 原帖 · @lizikk_zhu",
    "author": "lizikk",
    "publishedAt": "2026-08-12T09:58:43.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087478882320715791.mp4",
    "posterUrl": "/posters/x/2087478882320715791.jpg",
    "duration": 46,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "牛逼啊，全自动替代广告部了，不到20分钟，全本地制作,成本 $0！！！\n\n我用 DeepSeek-v4-flash和MiniMax H3尝试给做了一支广告片——46 秒,给之前开源 Mac 清理工具 Atlas。\n整体效果还是很惊艳的。\n\n整个品类的广告都在喊「你的 Mac 有危险!!」,我们反着来:\n「你的 Mac,从不需要被恐吓。」\n▎ :\n▎ 🎬 MiniMax H3 本地视频模型 ×6 段电影感画面\n▎ 🎙 本地 TTS 中文配音\n▎ 🎵 免版税音乐\n▎ 🛠 手写 HTML/GSAP 合成,1920×1080\n\n最难的不是视频模型——是字体排版,和配音里「行」的多音字(háng vs xíng),改了一行文案才修好。\n\n🎬 [视频链接] · Atlas for Mac · 开源 · 本地运行 → http://atlas.atomstorm.ai",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 221,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit product, advertising purpose, duration, campaign line, six-shot structure, narration, music, and compositing method"
  },
  {
    "id": "x-2087477920793326036",
    "title": "同一指令下的三模型运镜对比",
    "titleEn": "Three-Model Camera-Movement Comparison with the Same Prompt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者连续一周用同一指令比较三种模型，重点观察运镜是否像由摄影机操作员主动完成，而非仅固定朝向主体。 成片规格为 17 秒 · 1920×3238。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author compares three models with the same prompt over a week, focusing on whether the camera movement feels actively operated rather than merely pointed at the subject. Output: 17s · 1920×3238. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/reachvaldo/status/2087477920793326036",
    "sourceLabel": "X 原帖 · @reachvaldo",
    "author": "Valdo",
    "publishedAt": "2026-08-12T09:54:54.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087477920793326036.mp4",
    "posterUrl": "/posters/x/2087477920793326036.jpg",
    "duration": 17,
    "aspectRatio": "portrait",
    "resolution": "1920×3238",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One thing I didn't expect from MiniMax H3 is how well it handles camera movement.\n\nI've been generating the same prompt across 3 models this week just to compare. MiniMax H3 is the first one where the camera feels like someone is operating it, not just pointing it.\n\nstill testing but the motion language hits different.",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 5,
      "views": 1269,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit three-model same-prompt comparison and camera-operation evaluation criterion"
  },
  {
    "id": "x-2087477699766038990",
    "title": "金发角色奔跑于绿丘与彩云之间",
    "titleEn": "A Blonde Character Runs Across a Green Hill Under Painted Clouds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名简化绘制的金发角色在绿丘上奔跑，画面在远景、侧脸与正面视角之间切换。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A simply drawn blonde character runs across a green hill as the view moves between wide, profile, and frontal framings beneath painted clouds. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/exportAnything/status/2087477699766038990",
    "sourceLabel": "X 原帖 · @exportAnything",
    "author": "exportAnything",
    "publishedAt": "2026-08-12T09:54:01.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087477699766038990.mp4",
    "posterUrl": "/posters/x/2087477699766038990.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I put #LTX 2.5 up against #MiniMax H3 head-to-head with the same settings in #ComfyUI.  \n\nI really love how LTX 2.5 handled the shot-counts artistically, whereas MiniMax-H3 tripped up trying to force itself into the shot-count/time-stamp contraints.\n\n@Lightricks @ComfyUI #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 122,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087476484449644724",
    "title": "RTX 5080 上的同工作流生成速度对比",
    "titleEn": "Same-Workflow Generation-Speed Comparison on an RTX 5080",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两个模型在 RTX 5080 上使用相同工作流，以 1152×640、十秒、24 帧每秒的条件比较耗时，分别记录为 108.90 秒与 366.58 秒。 成片规格为 20 秒 · 1152×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two models use the same workflow on an RTX 5080 at 1152×640, ten seconds, and 24 frames per second, with recorded runtimes of 108.90 and 366.58 seconds. Output: 20s · 1152×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AION_2077/status/2087476484449644724",
    "sourceLabel": "X 原帖 · @AION_2077",
    "author": "AION",
    "publishedAt": "2026-08-12T09:49:11.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087476484449644724.mp4",
    "posterUrl": "/posters/x/2087476484449644724.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "1152×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX2.5 vs Minimax h3\n\n1152x640 , 10 s , 24fps\n\nLtx - 108.90 s\nMinimax h3 - 366.58 s\n\nwith RTX 5080\n\nBoth Workflows - (Sol attn , Sage , Int8) same",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 253,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit hardware, identical workflow, resolution, duration, frame rate, model pair, and measured runtimes"
  },
  {
    "id": "x-2087476101149016213",
    "title": "由 FPS 截图生成的低成本镜头",
    "titleEn": "A Low-Cost Shot Generated from FPS Captures",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者先用 DeepSeek V4 Flash 制作 FPS，再把游戏截图转为动态镜头，并称单个镜头成本低于两美元。 成片规格为 15 秒 · 3280×1692。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author first builds an FPS with DeepSeek V4 Flash, then turns game captures into a motion shot for a reported cost below two dollars. Output: 15s · 3280×1692. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yasei_no_otoko/status/2087476101149016213",
    "sourceLabel": "X 原帖 · @yasei_no_otoko",
    "author": "野生の男",
    "publishedAt": "2026-08-12T09:47:40.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087476101149016213.mp4",
    "posterUrl": "/posters/x/2087476101149016213.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3280×1692",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "DeepSeek V4 FlashでFPSを作成。MiniMax H3でキャプチャを動画にして2ドル未満でカット製作",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 939,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit FPS creation tool, capture-to-motion process, shot output, and sub-two-dollar cost"
  },
  {
    "id": "x-2087472104778485893",
    "title": "“想像神”角色的摩托车高速动作对比",
    "titleEn": "High-Speed Motorcycle Action Comparison Featuring the Sōzōshin Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一图片与指令驱动三个模型演绎“想像神”角色的摩托车动作，比较轮胎转动、水花与镜头跟随，并在后半段延伸到三十秒高速动作。 成片规格为 40 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same image and prompt drive three models through motorcycle action featuring the Sōzōshin character, comparing wheel rotation, water spray, and camera tracking before extending into thirty seconds of high-speed action. Output: 40s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mugi_AI_Art/status/2087472104778485893",
    "sourceLabel": "X 原帖 · @mugi_AI_Art",
    "author": "むぎ@AIアートのセカイ",
    "publishedAt": "2026-08-12T09:31:47.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087472104778485893.mp4",
    "posterUrl": "/posters/x/2087472104778485893.jpg",
    "duration": 40,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【PR】 \n#PolloAI #Seedance25 #AI動画\n想像神ちゃんのバイクアクションを、同じ画像・プロンプトからSeedance 2.5／MiniMax H3／Kling V3 Omniで比較🏍️💨  \n今回はSeedance 2.5が、タイヤの回転や水しぶき、カメラの追従まで自然で、特にダイナミックな動きが印象的でした。後半では30秒の高速アクションにも挑戦しています🌈  \nPollo AIなら複数モデルを1か所で試して比較できるのが便利。Seedance 2.5は現在リリース記念50% OFF！  \n@PolloAIJP",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 26,
      "views": 1540,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit named character, motorcycle action, same image and prompt, three-model comparison, wheel rotation, water spray, camera tracking, and thirty-second extension"
  },
  {
    "id": "x-2087471654578561407",
    "title": "细致服装设计在四种模型中的呈现对比",
    "titleEn": "Four-Model Comparison of Intricate Costume Design",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "四个输出分别来自常用方案、Krea 2、SeeDream 5 Pro 与另一模型，作者重点比较细致服装设计是否得到保留。 成片规格为 5 秒 · 1456×2544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Four outputs come from the author's usual setup, Krea 2, SeeDream 5 Pro, and another model, with the comparison focused on preserving intricate costume design. Output: 5s · 1456×2544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2087471654578561407",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-08-12T09:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087471654578561407.mp4",
    "posterUrl": "/posters/x/2087471654578561407.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1456×2544",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "SeeDream5/MiniMax H3 などを試しています\n\n1枚目: いつもの\n2枚目: Krea2\n3枚目: SeeDream 5 pro\n4枚目: MiniMax H3\n\nどのモデルでも、繊細なデザインが\nよく反映されている気がします\n\n衣装プロンプトはSHINOBU@AIart様よりお借りしています✨\nありがとうございます",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 22,
      "views": 309,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit four-output ordering, named systems, borrowed costume prompt, and intricate-design preservation criterion"
  },
  {
    "id": "x-2087471251388571756",
    "title": "参考图与音频锁定的三十秒单场景工作流",
    "titleEn": "A Thirty-Second Single-Scene Workflow with Reference Image and Audio Lock",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长时工作流通过参考图串接与音频锁定生成一个三十秒单场景内容，作者记录了伴奏段落嘴部张开的问题及 RTX 4070 上的耗时。 成片规格为 30 秒 · 384×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-form workflow uses chained reference imagery and audio locking to generate one thirty-second scene, with the author noting an open-mouth issue during accompaniment and the runtime on an RTX 4070. Output: 30s · 384×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2087471251388571756",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-08-12T09:28:24.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087471251388571756.mp4",
    "posterUrl": "/posters/x/2087471251388571756.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "384×544",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "そしていい感じのMiniMax H3の長尺動画生成ワークフロー・カスタムノードを見つけた！これの08_Chained_RefImage_AudioLock.jsonだと1シーン動画ができる。ただ伴奏時の口ポカンは残るけど。384x544の30秒動画で588秒！(RTX4070)\nhttps://github.com/Shrek3OnVH5/MiniMax-H3-NativeAudio-MusicVideo-Workflow/tree/master",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 178,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit workflow name, reference-image chaining, audio lock, single-scene output, duration, lip-motion issue, hardware, and runtime"
  },
  {
    "id": "x-2087469791468810463",
    "title": "使用本地语言模型编写生成指令",
    "titleEn": "Writing Generation Prompts with an On-Device Language Model",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖明确记录了一项使用本地语言模型编写生成指令的实验。 成片规格为 10 秒 · 512×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly records an experiment in writing generation prompts with an on-device language model. Output: 10s · 512×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YamataZen/status/2087469791468810463",
    "sourceLabel": "X 原帖 · @YamataZen",
    "author": "Yamata",
    "publishedAt": "2026-08-12T09:22:36.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087469791468810463.mp4",
    "posterUrl": "/posters/x/2087469791468810463.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "512×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Prompting MiniMax H3 with local LLM\n\n#aivideo #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 109,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit prompt-writing experiment using an on-device language model"
  },
  {
    "id": "x-2087465830372020275",
    "title": "羞涩、无聊、愤怒与顽皮的角色待机动作",
    "titleEn": "Shy, Bored, Angry, and Playful Character Idle Motions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "角色以全身构图站在干净影棚背景前，在固定机位中分别根据羞涩、无聊、愤怒与顽皮四种情绪自然改变姿势、步态、手势和表情。 成片规格为 15 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A full-body character stands against a clean studio background and, under a fixed camera, naturally varies posture, pacing, gestures, and expression across shy, bored, angry, and playful moods. Output: 15s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2087465830372020275",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-12T09:06:51.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087465830372020275.mp4",
    "posterUrl": "/posters/x/2087465830372020275.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\nMood based Character Idle Animations\n\nAdd your character reference and pick one of the moods to create an idle animation. You can tweak the prompt however you like.\nThe examples are, in order: shy, bored, angry, playful.\n\nPrompt:\nCreate a full-body idle animation of the character against a clean studio background. Keep the camera completely static for the entire shot, with the character fully visible from head to toe at all times.\n\nUse the reference character exactly and preserve the same visual style, identity, face, body proportions, hairstyle, outfit and accessories throughout.\n\nMOOD = [bored / confident / shy / angry / cheerful / tired / nervous / playful / sad / etc.]\n\nThe whole performance should be driven by MOOD. The character should not remain frozen in place. Let them behave naturally according to that mood, with subtle idle motion, small shifts in posture, occasional wandering or pacing within the frame, and intermittent gestures that feel spontaneous and believable. The exact actions, timing, body language, expressions and movement choices should all emerge naturally from MOOD rather than from a fixed action list. The result should feel alive, varied and human, as if the character is simply existing in that emotional state. No BGM.\n\nKeep the motion grounded and physically believable. Add soft secondary motion in hair, clothing and accessories where appropriate. Avoid exaggerated acting, dancing or action choreography unless strongly implied by MOOD.\n\nMaintain strict visual consistency with no morphing, no outfit changes, no duplicated limbs, no facial drift and no body distortion. Hands must remain clean and anatomically correct.",
    "engagement": {
      "replies": 8,
      "reposts": 4,
      "likes": 97,
      "views": 4445,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit full-body character, clean studio background, static camera, four demonstrated moods, and mood-driven idle behavior"
  },
  {
    "id": "x-2087465597458341932",
    "title": "阳光厨房木桌上的两枚煎蛋",
    "titleEn": "Two Fried Eggs on a Wooden Table in a Sunlit Kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白色餐盘里并排放着两枚单面煎蛋，置于阳光照入的厨房木桌上。 成片规格为 10 秒 · 480×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two sunny-side-up eggs sit side by side on a white plate atop a wooden table in a sunlit kitchen. Output: 10s · 480×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mmmpa/status/2087465597458341932",
    "sourceLabel": "X 原帖 · @mmmpa",
    "author": "ふとし",
    "publishedAt": "2026-08-12T09:05:56.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087465597458341932.mp4",
    "posterUrl": "/posters/x/2087465597458341932.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "480×640",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 セットアップ済みだからせっかくなので遊んでるけど、やっぱ思い通りに動かす正確なプロンプト考えるのがむずい。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 314,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "poster-media-review: explicit two fried eggs, white plate, wooden tabletop, and sunlit kitchen setting"
  },
  {
    "id": "x-2087465146394231152",
    "title": "多角色协作的动漫故事",
    "titleEn": "Anime Stories with Multiple Characters",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖明确将随帖内容概括为让多个角色同时参与的动漫叙事。 成片规格为 48 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly characterizes the attached work as anime storytelling involving multiple characters at the same time. Output: 48s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Artedeingenio/status/2087465146394231152",
    "sourceLabel": "X 原帖 · @Artedeingenio",
    "author": "OscarAI",
    "publishedAt": "2026-08-12T09:04:08.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087465146394231152.mp4",
    "posterUrl": "/posters/x/2087465146394231152.jpg",
    "duration": 48,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "With @Hailuo_AI MiniMax H3, you can tell really entertaining anime stories while working with multiple characters at the same time.",
    "engagement": {
      "replies": 8,
      "reposts": 2,
      "likes": 38,
      "views": 1981,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit anime-story format and simultaneous multi-character scope"
  },
  {
    "id": "x-2087463320039334282",
    "title": "单张输入图与三十秒音频驱动的对口型单场景",
    "titleEn": "A Lip-Synced Single Scene from One Image and Thirty Seconds of Audio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "工作流接收一张图片和三十秒音频，在不更换指令的情况下生成一段对口型单场景内容，并记录了伴奏时仍会出现嘴部动作的问题。 成片规格为 30 秒 · 384×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The workflow takes one image and thirty seconds of audio to produce a lip-synced single scene without changing the prompt, while noting that mouth movement still appears during accompaniment. Output: 30s · 384×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2087463320039334282",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-08-12T08:56:53.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087463320039334282.mp4",
    "posterUrl": "/posters/x/2087463320039334282.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "384×544",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ひとまず実験終了。入力画像1枚と30秒音声を入れるとリップシンクした1シーン動画を生成。プロンプトは変えなくてOK。RTX4070で384×544/30秒動画を12分30秒で生成可能。伴奏時の口パクが消せない課題あり。超汚いComfyUIワークフロー置いときます。更新予定は今のところなし！\nhttps://github.com/sepiablue-ai/my-comfyui-workflows/tree/main/ComfyUI-MiniMax-H3-30s-SingleScene#%E6%97%A5%E6%9C%AC%E8%AA%9E",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 885,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit single-image and thirty-second audio inputs, lip-synced single-scene output, unchanged prompt, and accompaniment limitation"
  },
  {
    "id": "x-2087462260218245401",
    "title": "发光粒子在水下化作巨鲸",
    "titleEn": "Glowing Particles Transform into a Colossal Whale Underwater",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "发光粒子通过液态变形在水下汇聚成一头巨鲸，场面强调水下氛围、水体物理与电影感运镜。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Glowing particles undergo a liquid transformation underwater and gather into a colossal whale, emphasizing underwater atmosphere, water physics, and cinematic camera movement. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MrDasOnX/status/2087462260218245401",
    "sourceLabel": "X 原帖 · @MrDasOnX",
    "author": "Mr Das",
    "publishedAt": "2026-08-12T08:52:40.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087462260218245401.mp4",
    "posterUrl": "/posters/x/2087462260218245401.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "From glowing particles to a colossal whale.\n\nThe liquid transformation, underwater atmosphere, realistic water physics, and cinematic camera movement make this feel like a creature emerging from another dimension.\n\nMiniMax H3 absolutely nailed the spectacle.\n\nCreated on @PixVerse_ #pixversecpp \n\nFull prompt ↓",
    "engagement": {
      "replies": 14,
      "reposts": 3,
      "likes": 21,
      "views": 405,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit glowing-particle origin, liquid transformation, colossal whale, underwater setting, water physics, and camera movement"
  },
  {
    "id": "x-2087461857439485964",
    "title": "480p 五秒文生动态的耗时记录",
    "titleEn": "Runtime Record for Five Seconds of 480p Text-to-Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "一次文生动态运行以 480p 输出五秒内容，作者记录耗时为 240 秒。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A text-to-motion run produces five seconds at 480p, with the author recording a runtime of 240 seconds. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/roid_s/status/2087461857439485964",
    "sourceLabel": "X 原帖 · @roid_s",
    "author": "ロイド",
    "publishedAt": "2026-08-12T08:51:04.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087461857439485964.mp4",
    "posterUrl": "/posters/x/2087461857439485964.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3セッティングしてT2Vやってみた\n480Pの5秒動画で240秒",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 483,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit text-driven mode, 480p resolution, five-second duration, and 240-second runtime"
  },
  {
    "id": "x-2087461643769000082",
    "title": "RTX 3060的四组单步耗时",
    "titleEn": "Per-Step Runtime Comparison of Turbo LoRA and SageAttention on an RTX 3060",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "RTX 3060 12GB 环境分别比较是否启用 Turbo LoRA 与 SageAttention 时的单步耗时，记录了四组 21 至 29 秒的结果。 成片规格为 10 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An RTX 3060 12GB setup compares per-step runtime with and without Turbo LoRA and SageAttention, recording four results ranging from 21 to 29 seconds. Output: 10s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/dec21ai/status/2087461643769000082",
    "sourceLabel": "X 原帖 · @dec21ai",
    "author": "dec21ai",
    "publishedAt": "2026-08-12T08:50:13.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087461643769000082.mp4",
    "posterUrl": "/posters/x/2087461643769000082.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "RTX 3060 12GBでMiniMax H3を試す\n\nTurbo LoRAなしで1ステップ27秒\nSageAttention有効で21秒\n\nTurbo LoRAありで1ステップ29秒\nSageAttention有効で24秒",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 91,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit GPU, memory, Turbo LoRA and SageAttention conditions, and four per-step timing measurements"
  },
  {
    "id": "x-2087461496880300098",
    "title": "LTX 2.5 的十五秒生成速度与指令跟随记录",
    "titleEn": "LTX 2.5 Runtime and Prompt Adherence at Fifteen Seconds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "LTX 2.5 在同一指令下生成十五秒、0.9 百万像素内容耗时 268 秒；作者认为速度很快，但较十秒版本更异常且指令跟随下降。 成片规格为 15 秒 · 768×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "With the same prompt, LTX 2.5 produces fifteen seconds at 0.9 megapixels in 268 seconds; the author finds it fast but more erratic and less prompt-faithful than the ten-second version. Output: 15s · 768×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Yamaha_Roland/status/2087461496880300098",
    "sourceLabel": "X 原帖 · @Yamaha_Roland",
    "author": "やまはろ。(Yamaha Roland)",
    "publishedAt": "2026-08-12T08:49:38.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087461496880300098.mp4",
    "posterUrl": "/posters/x/2087461496880300098.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1152",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX2.5 今度は同じプロンプトで15秒でテスト\n0.9メガピクセル 15秒 で 生成時間 268秒 \nバニラ状態でこの生成速度は MiniMax H3に比べると驚異的な速さだけど、10秒動画に比べるとかなりへんてこなムービーになった。\n15秒だとプロンプトの追従性はかなり落ちるかも",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 70,
      "views": 2961,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit model, same-prompt condition, fifteen-second duration, 0.9-megapixel size, runtime, and reported prompt-adherence change"
  },
  {
    "id": "x-2087461268286554349",
    "title": "单次生成三镜头的三模型对比",
    "titleEn": "Three-Model Comparison of Three Shots in One Generation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "三个模型分别尝试在约十四秒、864×480 的一次生成中完成三镜头场景；作者还记录了参考条件差异、转场后的行走动作与中文读音问题。 成片规格为 44 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three models each attempt a three-shot scene in one roughly fourteen-second, 864×480 generation, with the author also documenting reference-condition differences, walking after a scene transition, and Chinese-pronunciation issues. Output: 44s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yu_ichi_suzuki/status/2087461268286554349",
    "sourceLabel": "X 原帖 · @yu_ichi_suzuki",
    "author": "鈴木憂一 | Highdrama",
    "publishedAt": "2026-08-12T08:48:44.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087461268286554349.mp4",
    "posterUrl": "/posters/x/2087461268286554349.jpg",
    "duration": 44,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "1回の生成で3カットのシーンを作る、というのをやってみました（2026年8月時点）。\n\n順にローカルMiniMax H3→クラウドSeedance 2.5→ローカルLTX-2.5（参考）。864×480・約14秒で揃えました。\n\n本当は３本ともオムニリファレンスっぽいことをさせて比較したかったのですが、LTX-2.5では自分はできませんでした。\nIC-LoRA Ingredientsも上手く行かず、Sora2でやっていた添付画像一枚目にリファレンス画像を集約して、１カット目にリファレンス画像を表示させたらすぐ２カット目に移る、とかやってみましたがダメでした。\n今のところ（2026年8月時点）i2vで一枚から動かすしかないんでしょうか。なので背景＋人物の1枚だけを渡した状態ですのでH3及びSeedanceとは条件が揃っていません。\n自分のやり方が間違っている可能性も大いにあるので、うまく行った方は教えてください。\n\nH3いいですね。カット２の最後で動き出し、シーン変更とともにカット３で中央よりちょっと右から自然に歩いているのがいい。\nSeedance2.5も音と全体的にダークな雰囲気がいいですね。\nある程度おまかせするならH3、プロンプト詰めるならSeedance、かも。\n\nあとこれはLTXだけでなくたまにH3・Seedanceもなりますが、漢字が中国語読みに引っ張られることがあります。\n「紅茶」が「ほんこく」、「珈琲」が「コーヘー」になりました。\nこの動画ではLTXの台詞だけかなに直しています。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 6,
      "views": 1638,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit three-model comparison, three shots in one generation, matched target duration and resolution, differing reference conditions, described transition action, and pronunciation observations"
  },
  {
    "id": "x-2087456940071014808",
    "title": "星空下佩戴发光蝴蝶发饰的白裙女子",
    "titleEn": "A Woman in White Wears Glowing Butterfly Hairpieces beneath a Starry Sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名长黑发女子身穿白色礼服，头发两侧别着发光蝴蝶饰品，身后是深色星空。 成片规格为 16 秒 · 576×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long, dark-haired woman in a white gown wears glowing butterfly ornaments on both sides of her hair against a dark, starry backdrop. Output: 16s · 576×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PscNovaX/status/2087456940071014808",
    "sourceLabel": "X 原帖 · @PscNovaX",
    "author": "Seven",
    "publishedAt": "2026-08-12T08:31:32.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087456940071014808.mp4",
    "posterUrl": "/posters/x/2087456940071014808.jpg",
    "duration": 16,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 70,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "poster-media-review: explicit long dark hair, white gown, paired luminous butterfly ornaments, and starry dark background"
  },
  {
    "id": "x-2087456778661585050",
    "title": "暑期海河戏水安全的十五秒扁平动效公益广告",
    "titleEn": "A Fifteen-Second Flat-Motion Water-Safety Public Service Spot for Summer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "十五秒公益广告以扁平矢量图形依次讲解海河戏水风险、离岸流应对、穿救生衣与成人看护，最后用三项检查清单收尾。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fifteen-second public service spot uses flat vector motion to cover coastal and river hazards, rip-current response, life jackets, and adult supervision before ending on a three-item checklist. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tanabe_fragm/status/2087456778661585050",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-12T08:30:53.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087456778661585050.mp4",
    "posterUrl": "/posters/x/2087456778661585050.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "低コストだったり、ローカルから無料で使えることで話題の「MiniMax H3」のプロンプトもClaude Codeで書くのがおすすめです。\n\n以下のプロンプトをClaude Codeに入力すればSkillをつくれます。\n\n-----------------\n以下のURLは動画生成AI「MiniMax H3」のプロンプトガイドです。\n\nhttps://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_base_en.md\nhttps://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_ref_en.md\n\nこの内容を参考にして、Skillを作って。\nMiniMax H3のプロンプトを作るときはSkillを参考にして作ってね。\n-----------------\n\n↓の動画は、以下のプロンプトをClaude Codeに入力して生成されたプロンプトで作ったモーショングラフィックスです。\n\n-----------------\nMiniMax H3で、15秒のモーショングラフィックス風CMを作りたい。\n内容は、夏休み中の海や川での水遊びに注意してください、というもの。\n海や川で遊ぶときの注意点や、必要な準備を指摘する内容にして。\n-----------------\n\nMiniMax H3はこのようなモーショングラフィックスやアニメがサクッと作れるので、素晴らしいですね😌\n\nClaude Codeが生成してくれたMiniMax H3向けのプロンプトはこちら↓\n-----------------\nintegrated_multimodal_description: Create a complete 15-second Japanese flat-illustration\npublic-safety spot in a clean 16:9 composition. Use friendly 2D vector illustration, thick\nnavy outlines, flat solid colour fills, and restrained cutout animation. Compact palette of\nwarm white, deep navy, bright aqua, saturated yellow and coral red, where coral red marks a\ndanger and yellow marks the action to take. People are simple flat icons without facial\nfeatures. No photorealism, no 3D rendering, no gradients, no drop shadows, no browser\ninterface, no screen-recording artifacts, no player controls, and no depiction of an accident\nor a drowning.\n[Shot 1] Begin on a warm-white background. A wide flat graphic of a shoreline on the left and\na river on the right slides up from the bottom edge. The Japanese headline assembles word by\nword in navy: \"夏休みの水あそび\" A short yellow underline draws itself beneath it, and two\nsmall aqua wave marks tick in from the left and right edges.\n[Shot 2] At 00:02.600, a bright aqua colour field wipes across the frame from the left. A flat\nicon of a river bank appears in the lower half. Three white rounded speech bubbles pop in one\nat a time, reading \"急に深くなる\", \"流れは見えない\", and \"声は出せない\". As the third lands, a\nsmall caption fades in below: \"水の事故は静かに起きる\".\n[Shot 3] At 00:05.400, a coral-red field wipes in from the right. A top-down flat map of a\nbeach fills the frame, the sand along the bottom edge and the open sea at the top. A coral-red\narrow draws itself from the shoreline straight upward toward the open sea, showing the rip\ncurrent pulling a swimmer away from the beach. A yellow arrow then draws itself strictly\nhorizontally, running parallel to the sand along the shoreline and never turning toward the\nopen sea. Large white text assembles in two stages: \"流されたら\" followed by \"岸と平行に泳ぐ\".\n[Shot 4] At 00:08.000, the coral-red field wipes into a saturated yellow solution field. A flat\nicon of a child stands centre frame; an orange life jacket flies in from the right and snaps\nonto the child, and its buckles click shut one by one. A white outlined card slides in beneath\nwith the exact label \"ライフジャケットを着る\", and a yellow check mark stamps onto it.\n[Shot 5] At 00:10.800, a bright aqua field wipes in from the bottom. Two small child icons play\nat the water's edge on the left and one adult icon stands on the sand on the right. A yellow\ndotted line draws itself from the adult's head straight to the two children, touching both\nchildren and nothing else, with no other object anywhere along that line. The adult then steps\ntwice toward the children and the dotted line shortens. The caption \"目をはなさない\" assembles\nbeneath them.\n[Shot 6] At 00:13.200, cut to a calm warm-white end card with generous negative space. Three\nchecklist rows reveal one after another in navy — \"天気を確認\", \"ライフジャケット\", \"目をはなさ\nない\" — each receiving a yellow check mark as it lands. A calm, clear young Japanese woman (S1)\nsays in an off-screen voiceover: <d>[Japanese] 準備をして、夏を楽しく。</d> The card holds\nsharp, centred and readable until exactly 15.000 seconds.\nThroughout: flat vector motion graphics only, one palette and one line weight, all Japanese\ntext in clean gothic type, correctly formed and static once placed, with no subtitles of the\nspoken line and no other lettering.\n\noverall_soundscape: Soft water laps and light splashes run under the whole spot, with a short\nairy whoosh on each colour wipe. Rounded paper-like pops accompany each speech bubble, a pen-on\n-paper stroke follows each arrow and dotted line as it draws, and two small buckle clicks land\nas the life jacket closes. Light footsteps mark the adult's two steps, and a soft confirmation\nchime rings on each check mark.\n\nnon_diegetic_music: Generate an original 15-second friendly Japanese commercial cue using light\nmarimba, ukulele, soft hand percussion and warm sustained chords at a moderate tempo, playing\nwithout a gap from the first frame to the last. Begin sparsely, thin to marimba alone under the\ncoral-red rip-current section, brighten as the yellow solution field arrives, build lightly\nthrough the watching section, and resolve on one clean sustained chord at 15.000 seconds.\nNo singing and no lyrics.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 30,
      "views": 4924,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit fifteen-second Japanese public-safety spot, flat-vector style, six-shot sequence, water hazards, rip current, life jacket, supervision, and final checklist"
  },
  {
    "id": "x-2087456554001776828",
    "title": "小小阿尔劳娜的日常《伞是为了花坛》",
    "titleEn": "A Little Alraune in The Umbrella Is for the Flower Bed",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这则日常短篇名为《伞是为了花坛》，主角是一位小小的阿尔劳娜少女。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "This everyday short is titled The Umbrella Is for the Flower Bed and centers on a little Alraune girl. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087456554001776828",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T08:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087456554001776828.mp4",
    "posterUrl": "/posters/x/2087456554001776828.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n日常エピソード\n『傘は花壇のために』\nとある小さなアルラウネちゃん",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 17,
      "views": 781,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit episode title, everyday-episode format, and little Alraune-girl subject"
  },
  {
    "id": "x-2087456485077139958",
    "title": "夏日驱邪",
    "titleEn": "Summer Exorcism",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖明确以夏日驱邪、驱散恶灵为随帖作品主题。 成片规格为 15 秒 · 1920×1106。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly frames the attached work around a summer exorcism and driving away evil spirits. Output: 15s · 1920×1106. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Muimi_Karma/status/2087456485077139958",
    "sourceLabel": "X 原帖 · @Muimi_Karma",
    "author": "無意味カルマ",
    "publishedAt": "2026-08-12T08:29:43.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087456485077139958.mp4",
    "posterUrl": "/posters/x/2087456485077139958.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1106",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 \n\n夏の邪気払い👻⛩️\n悪霊退散、悪霊退散",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 88,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit summer-exorcism theme and evil-spirit expulsion premise"
  },
  {
    "id": "x-2087456059204280806",
    "title": "与暴走达摩的激战：图生动态与参考驱动对比",
    "titleEn": "Battle with a Rampaging Daruma: Image-Driven vs. Reference-Driven Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "同一素材与指令用于表现与暴走达摩的激战，并比较图生动态和参考驱动两种模式；指令还明确加入集中线与漫画拟声字。 成片规格为 33 秒 · 2200×900。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same assets and prompt stage a battle with a rampaging Daruma while comparing image-driven and reference-driven modes; the prompt also explicitly adds speed lines and manga sound-effect lettering. Output: 33s · 2200×900. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Creator_Pelsan/status/2087456059204280806",
    "sourceLabel": "X 原帖 · @Creator_Pelsan",
    "author": "PeL：I&P",
    "publishedAt": "2026-08-12T08:28:02.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087456059204280806.mp4",
    "posterUrl": "/posters/x/2087456059204280806.jpg",
    "duration": 33,
    "aspectRatio": "landscape",
    "resolution": "2200×900",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🔥暴走達磨とガチバトル🔥\n※注意：わたしはダルマが大好きです😿倒してごめんね\n\nPollo AI（@PolloAIJP）でH3の動画生成を比較✨\n①画像から動画　②リファレンスから動画\nさてどう違うか。\n\n集中線や「ドゴォォン！」などの漫画擬音もプロンプトで指示してみた\n\nMinimax H3はクレジット消費が少なく\n2Kでも比較的安く試せるので気軽に生成しやすいのも◎\n\n同じ素材・プロンプトでも、モデルごとの違いがおもしろい👀\n\n使ったプロンプトは以下においておくねー😃",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 18,
      "views": 732,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit rampaging-Daruma battle, identical assets and prompt, two input-mode comparison, speed lines, and manga sound-effect text"
  },
  {
    "id": "x-2087455628621230462",
    "title": "灯泡灵感、空白画板与问号的扁平插画动效",
    "titleEn": "Flat Motion Graphics with a Lightbulb, Blank Board, and Question Mark",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "扁平插画中，一名拿笔女子站在空白画板与手机后方，旁边出现灯泡图标，画面另一侧占据着巨大问号。 成片规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a flat illustration, a woman holding a pen stands behind a blank board and phone beside a lightbulb icon, while a giant question mark fills the other side. Output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2087455628621230462",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-08-12T08:26:19.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087455628621230462.mp4",
    "posterUrl": "/posters/x/2087455628621230462.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI × モーショングラフィックス\n\nMiniMax H3  Flova(@Flovaai_Japan)で制作 \n\nPrompt:",
    "engagement": {
      "replies": 3,
      "reposts": 15,
      "likes": 126,
      "views": 16732,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "poster-media-review: explicit illustrated woman with pen, blank board, phone, lightbulb, and large question mark; sourceCaption: explicit motion-graphics format"
  },
  {
    "id": "x-2087455389273182261",
    "title": "GPU 租用环境下的 720p 十秒成本与耗时记录",
    "titleEn": "Cost and Runtime Record for Ten Seconds at 720p on a Rented GPU",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "GPU 租用环境生成 720p、十秒内容耗时五分钟，成本约七日元；作者同时记录了简短日语指令识别不佳的问题。 成片规格为 10 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rented-GPU run produces ten seconds at 720p in five minutes for roughly seven yen, while the author also notes poor recognition of a simple Japanese prompt. Output: 10s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HackCooking/status/2087455389273182261",
    "sourceLabel": "X 原帖 · @HackCooking",
    "author": "Cooking life hack",
    "publishedAt": "2026-08-12T08:25:22.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087455389273182261.mp4",
    "posterUrl": "/posters/x/2087455389273182261.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3をGPUレンタルで出力、720P10秒が5分\nコストは7円程度、APIも試して一番コスパいいところにしようかなｗ\nでも、簡単に日本語でプロンプトを入れてもちゃんと認識してくれないｗ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 128,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit rented-GPU environment, resolution, duration, runtime, cost, and Japanese prompt-recognition issue"
  },
  {
    "id": "x-2087451580123820416",
    "title": "无参考图的三维风格动态",
    "titleEn": "3D-Style Motion Without a Reference Image",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "作者误触后在没有参考图的情况下完成生成，并明确称赞结果中的动作与三维风格。 成片规格为 5 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "After an accidental tap, the author completes a run without a reference image and explicitly praises the resulting motion and 3D style. Output: 5s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ameilo_1/status/2087451580123820416",
    "sourceLabel": "X 原帖 · @Ameilo_1",
    "author": "Ameilo",
    "publishedAt": "2026-08-12T08:10:14.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087451580123820416.mp4",
    "posterUrl": "/posters/x/2087451580123820416.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "誤タップで参照画像なしのまま生成しちゃったMiniMax H3の動画、結構良い😂\n動きと3Dスタイルがかなり好き。",
    "engagement": {
      "replies": 11,
      "reposts": 81,
      "likes": 1373,
      "views": 87352,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit no-reference generation condition and reported motion and 3D-style qualities",
    "featured": {
      "at": "2026-09-05T15:50:15.687Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2087451314720854523",
    "title": "三种生成模型的正面对比",
    "titleEn": "Head-to-Head Comparison of Three Generative Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者将三种近期受到关注的生成模型放在一起比较，并概括了闭源服务与低成本开放模型之间的当前差距。 成片规格为 22 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author compares three recently prominent generative models and summarizes the current gap between a proprietary service and lower-cost open models. Output: 22s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yama_kiyo/status/2087451314720854523",
    "sourceLabel": "X 原帖 · @yama_kiyo",
    "author": "山本清史（やまきよ）/ Kiyoshi Yamamoto(yamakiyo)",
    "publishedAt": "2026-08-12T08:09:11.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087451314720854523.mp4",
    "posterUrl": "/posters/x/2087451314720854523.jpg",
    "duration": 22,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "動画生成モデルのアプデ祭。話題の3モデルを比較したよ。\n🔥MiniMax H3 vs LTX2.5 vs Seedance2.5\n\n所感：Seedanceの優位性は変わらず高いですが、安くてオープンなモデルの追い上げもすさまじく、今年中にブレイクスルーが来そうです。\n@Hailuo_AI @BytePlus_Japan @ltx_io",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 915,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit three-model comparison, named models, and reported high-level comparative conclusion"
  },
  {
    "id": "x-2087450895395250655",
    "title": "白发双马尾少女双手指向脸颊",
    "titleEn": "A White-Haired Twin-Tailed Girl Points Both Hands toward Her Cheeks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名粉色眼睛、白色双马尾的卡通少女身穿黑白褶边裙装，用两根食指分别指向脸颊。 成片规格为 5 秒 · 896×1184。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-eyed cartoon girl with white twin tails wears a black-and-white ruffled dress and points both index fingers toward her cheeks. Output: 5s · 896×1184. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_studioapollo/status/2087450895395250655",
    "sourceLabel": "X 原帖 · @ai_studioapollo",
    "author": "studioあぽろん",
    "publishedAt": "2026-08-12T08:07:31.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087450895395250655.mp4",
    "posterUrl": "/posters/x/2087450895395250655.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "896×1184",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "１日遅れですが本格稼働から7か月が経過したようですね✨\n\nGPT-image-2でAI画像生成が一般化した時点でも、すごい時代になったものだと思っていたのですが…\nMiniMax H3でローカルでそれなりに動画生成が動くようになるとは…😳\nwan2.2で抱いていた不満がかなり解消されてる…😆\n\n#あぽなーと",
    "engagement": {
      "replies": 0,
      "reposts": 4,
      "likes": 10,
      "views": 382,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "poster-media-review: explicit white twin tails, pink eyes, black-and-white ruffled outfit, and both index fingers directed at the cheeks"
  },
  {
    "id": "x-2087450783780585624",
    "title": "优先动漫式动态而非写实感的运动探索",
    "titleEn": "Motion Exploration Prioritizing Anime Dynamism over Realism",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这项运动探索主动降低写实优先级，转而强调动漫式的动态表现，作者称结果已逐渐接近预期。 成片规格为 7 秒 · 1280×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "This motion exploration deliberately lowers the priority of realism in favor of anime-style dynamism, with the author saying the result is moving closer to the intended direction. Output: 7s · 1280×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI6377546630069/status/2087450783780585624",
    "sourceLabel": "X 原帖 · @AI6377546630069",
    "author": "AIの鐘（ べる）",
    "publishedAt": "2026-08-12T08:07:04.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087450783780585624.mp4",
    "posterUrl": "/posters/x/2087450783780585624.jpg",
    "duration": 7,
    "aspectRatio": "square",
    "resolution": "1280×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3です。\nリアリティーよりもアニメ的なダイナミックな動きを優先するようにしてみましたが、なかなか良い感じななってきました！！\n\nクレジットを気にせずにガンガン試せるのはありがたいですね。",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 37,
      "views": 458,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit anime-dynamism-over-realism objective and reported progress toward the intended motion"
  },
  {
    "id": "x-2087450514204299345",
    "title": "Gavin 终于在后院发现生命",
    "titleEn": "Gavin Finds Life in the Backyard",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明 Gavin 仍未走出后院，但已从洒水器转向那里发现的生命，并指出 1080p 输出在节奏安排上仍有不足。 成片规格为 12 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says Gavin has not made it beyond the backyard but has moved on from the sprinkler to life he found there, while noting that the 1080p output still has timing weaknesses. Output: 12s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Luna_R4ze/status/2087450514204299345",
    "sourceLabel": "X 原帖 · @Luna_R4ze",
    "author": "LunaRaze",
    "publishedAt": "2026-08-12T08:06:00.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087450514204299345.mp4",
    "posterUrl": "/posters/x/2087450514204299345.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Gavin hasn't made it out of the backyard yet. But at least he's moved on from the sprinkler and found actual life! This is 1080p on MiniMax H3, run through ComfyUI on a RTX PRO 6000. The weakness is the timing, showing the importance of giving things time to breathe 🔥",
    "engagement": {
      "replies": 5,
      "reposts": 0,
      "likes": 16,
      "views": 423,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit character, backyard setting, progression from the sprinkler, output resolution, and timing critique"
  },
  {
    "id": "x-2087450362022425010",
    "title": "六二二比例的三模型混合制作",
    "titleEn": "A Three-Model Production Mix at 60–20–20",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案只给出三种生成模型在这次制作中的占比，依次为 60%、20% 和 20%。 成片规格为 96 秒 · 1080×1442。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption only gives the shares of three generation models used in this production: 60%, 20%, and 20%. Output: 96s · 1080×1442. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/du17/status/2087450362022425010",
    "sourceLabel": "X 原帖 · @du17",
    "author": "CX LΛB",
    "publishedAt": "2026-08-12T08:05:23.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087450362022425010.mp4",
    "posterUrl": "/posters/x/2087450362022425010.jpg",
    "duration": 96,
    "aspectRatio": "portrait",
    "resolution": "1080×1442",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3（60%）+ Seedance 2.5（20%） + kling 3（20%）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 77,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit three-model production mix and percentages"
  },
  {
    "id": "x-2087446356105527621",
    "title": "为 Qwen3.6-35B 编写转换提示词",
    "titleEn": "Writing a Conversion Prompt for Qwen3.6-35B",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案记录了约两小时的学习过程，并说明作者已为 Qwen3.6-35B 完成一套内容转换提示词。 成片规格为 7 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption records about two hours of study and says the author completed a content-conversion prompt for Qwen3.6-35B. Output: 7s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yaya_karame/status/2087446356105527621",
    "sourceLabel": "X 原帖 · @yaya_karame",
    "author": "ヤヤ・カラメ",
    "publishedAt": "2026-08-12T07:49:28.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087446356105527621.mp4",
    "posterUrl": "/posters/x/2087446356105527621.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "午後は２時間ぐらいMiniMax H3のお勉強😵‍💫\nqwen3.6-35b用の動画変換プロンプトもできた🌞\nお手軽なのに、これすごいよね～♥️",
    "engagement": {
      "replies": 3,
      "reposts": 5,
      "likes": 46,
      "views": 321,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit study duration and conversion-prompt workflow"
  },
  {
    "id": "x-2087445583363723574",
    "title": "暖光海边的人物侧脸",
    "titleEn": "A Side-Profile Portrait by the Ocean in Warm Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名浅色头发的人始终以侧脸面对海面，暖色低角度光线照亮面部，散落发丝随画面变化。 成片规格为 5 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A light-haired person remains in side profile facing the ocean as warm low-angle light falls across the face and loose hair shifts. Output: 5s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2087445583363723574",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-12T07:46:24.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087445583363723574.mp4",
    "posterUrl": "/posters/x/2087445583363723574.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "falがサンプルで出してるLoRAのRealism People\n0.7 ON / OFF比較\nTurbo LoRAはON状態、同一シード、8STEP\nhttps://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA\n\nおおっ……ちゃんと効いてる🤔\n生成時間はONでもOFFでも大差なし（3秒差）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 626,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087445464992145772",
    "title": "唐代士兵演武对练",
    "titleEn": "Tang-Dynasty Soldiers in Martial-Arts Sparring",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确将主题写为唐代士兵演武对练，并说明同一主题分别交给四种生成系统制作，邀请读者匹配各自输出。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly identifies Tang-dynasty soldiers practicing martial-arts sparring and says the same theme was produced with four generation systems for readers to match to their outputs. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AlexLee44up/status/2087445464992145772",
    "sourceLabel": "X 原帖 · @AlexLee44up",
    "author": "Alex Lee",
    "publishedAt": "2026-08-12T07:45:56.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087445464992145772.mp4",
    "posterUrl": "/posters/x/2087445464992145772.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "天上的中国看多了\n来看看凡间的中国\n\n唐代士兵演武对练\n\n图 seedream 5.0Pro\n视频分别用了 \nseedance 2.5\nseedance 2.0\nminimax H3\nkling 3.0\n\n大家来做个连线题\n猜一猜哪个视频对应哪个模型吧😎\n\n@Hailuo_AI @Kling_ai @seedance",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 115,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit historical subjects, sparring action, and four-system comparison format"
  },
  {
    "id": "x-2087443463432466682",
    "title": "两种生成模型再次对照",
    "titleEn": "Another Comparison Between Two Generation Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案只说明这是两种生成模型之间的再次对照，没有提供内容主题、参数或评价维度。 成片规格为 36 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption only identifies this as another comparison between two generation models, without a content theme, settings, or evaluation dimensions. Output: 36s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/icreat_ai/status/2087443463432466682",
    "sourceLabel": "X 原帖 · @icreat_ai",
    "author": "ICREAT AI",
    "publishedAt": "2026-08-12T07:37:59.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087443463432466682.mp4",
    "posterUrl": "/posters/x/2087443463432466682.jpg",
    "duration": 36,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Again. MiniMax H3  vs Seedance 2.5",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 87,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit two-model comparison, with no further setup details"
  },
  {
    "id": "x-2087442808059924970",
    "title": "提示词技巧摸索记录",
    "titleEn": "Exploring Prompting Techniques",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案记录作者逐渐掌握提示词技巧，但没有进一步说明具体结构或内容主题。 成片规格为 8 秒 · 480×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption records the author beginning to understand prompting techniques but does not specify the structure or content theme. Output: 8s · 480×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AJI_FRY_ART/status/2087442808059924970",
    "sourceLabel": "X 原帖 · @AJI_FRY_ART",
    "author": "aji_fry",
    "publishedAt": "2026-08-12T07:35:22.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087442808059924970.mp4",
    "posterUrl": "/posters/x/2087442808059924970.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "480×640",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3\nちょっとプロンプトのコツがわかってきた👅",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 125,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit prompting-technique exploration"
  },
  {
    "id": "x-2087442765584159183",
    "title": "Strix Halo 上的奇特户型生成",
    "titleEn": "An Unusual Floor Plan Generated on Strix Halo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将内容描述为一个奇特户型，并说明在 Strix Halo 设备上完成约耗时一个半小时。 成片规格为 5 秒 · 576×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption describes the content as an unusual floor plan and says it took about one and a half hours to complete on a Strix Halo device. Output: 5s · 576×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Hanto12317190/status/2087442765584159183",
    "sourceLabel": "X 原帖 · @Hanto12317190",
    "author": "Hanto@不動産×AIの個人開発",
    "publishedAt": "2026-08-12T07:35:12.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087442765584159183.mp4",
    "posterUrl": "/posters/x/2087442765584159183.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でローカル生成したクソ間取り動画がこちらです。\n\n生成時間はStrixHaloの端末で1.5時間くらい。ちょっと可能性は感じるけど、やっぱり動画生成ならRTX5090の端末が欲しいかなってところ。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 232,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit floor-plan subject, hardware, and runtime"
  },
  {
    "id": "x-2087442321038364767",
    "title": "男性变为女性的转变过程",
    "titleEn": "A Man Transforms into a Woman",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确写出男性变为女性的转变，并记录日语提示词伴随不明语言与不明旁白的问题。 成片规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly describes a man transforming into a woman and reports that the Japanese prompt produced unidentified language and narration. Output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yonyon_sft/status/2087442321038364767",
    "sourceLabel": "X 原帖 · @yonyon_sft",
    "author": "よんよん",
    "publishedAt": "2026-08-12T07:33:26.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087442321038364767.mp4",
    "posterUrl": "/posters/x/2087442321038364767.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "男性→女性変身\nMiniMax H3で作成。\n\nプロンプトが悪いのか、謎言語、謎ナレーションが出まくる。日本語プロンプトがまずいのかな？\nなかなか上手くできないけど、まあ、仕方なし。",
    "engagement": {
      "replies": 0,
      "reposts": 6,
      "likes": 82,
      "views": 5604,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit transformation subject and reported prompt-output issue"
  },
  {
    "id": "x-2087439496786882570",
    "title": "两阶段图像与动态制作链路",
    "titleEn": "A Two-Stage Image and Motion Production Workflow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确说明静态图像与后续动态内容由两种不同工具分阶段完成，但没有描述作品主题。 成片规格为 34 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly says that still images and subsequent motion content were completed in two stages with different tools, but it does not describe the work's subject. Output: 34s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yuea85035/status/2087439496786882570",
    "sourceLabel": "X 原帖 · @yuea85035",
    "author": "Riley",
    "publishedAt": "2026-08-12T07:22:13.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087439496786882570.mp4",
    "posterUrl": "/posters/x/2087439496786882570.jpg",
    "duration": 34,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GPT Image 2 for the images, MiniMax-H3 for the video generation — the results honestly surprised me.\n\nYou can try creating your own AI videos here 👇\n https://www.xskill.ai/#/v2/models/st-ai%2Fsuper-seed2-lite?ref=Q1DTE94G",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 90,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit two-stage still-image and motion-production workflow"
  },
  {
    "id": "x-2087437857967538322",
    "title": "腋毛主题的提示词跟随度核验",
    "titleEn": "Prompt-Adherence Check for an Armpit-Hair Theme",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案以腋毛为明确主题，并将这段内容定位为对提示词精确跟随程度的核验。 成片规格为 5 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly identifies armpit hair as the theme and frames the content as a check of precise prompt adherence. Output: 5s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sinsai794/status/2087437857967538322",
    "sourceLabel": "X 原帖 · @sinsai794",
    "author": "わきあい",
    "publishedAt": "2026-08-12T07:15:42.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087437857967538322.mp4",
    "posterUrl": "/posters/x/2087437857967538322.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#腋 #ワキ #armpit #腋毛 AIわき 😀\nMiniMax H3、音声付きで面白いので乗り換えました\nいや、悪いのは1億％私なんですけれど・・\nプロンプトの正確な追従検証にどうぞ・・・",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 16,
      "views": 756,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit subject and prompt-adherence evaluation purpose"
  },
  {
    "id": "x-2087435915417522473",
    "title": "指定文字驱动的指文字动作",
    "titleEn": "Finger-Spelling Motion Driven by Specified Text",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者尝试按指定文字完成指文字动作，并称为加入屏幕文字而修改提示词后，整体动作比原作更克制。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author attempted finger-spelling from specified text and found that modifying the prompt to add on-screen wording made the overall motion more restrained than the source piece. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/studio_oneroom/status/2087435915417522473",
    "sourceLabel": "X 原帖 · @studio_oneroom",
    "author": "AIスタジオワンルーム（AIアニメ、動画、漫画）",
    "publishedAt": "2026-08-12T07:07:59.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087435915417522473.mp4",
    "posterUrl": "/posters/x/2087435915417522473.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "前からずっと試したいと思っていたヤレヤルさんの指文字モーション。\n\nこれはMiniMax H3向だろうと思ったのでH3で試してみました。\n\nとりあえずHailuo MiniMax H3って表示させたくてプロンプトをいじってしまったせいか本家よりも動きがおとなしくなってしまったかな？\n\nでも、指定した文字でここまで出来るのってやっぱりH3すごくないですか？\n\nプロンプトは引用元のヤレヤルさんの投稿で公開されていますのでご興味のある方はぜひどうぞ！\n\n-------------- \n#MiniMaxH3\n@Hailuo_AI",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 60,
      "views": 4205,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit finger-spelling action, specified text, and prompt modification"
  },
  {
    "id": "x-2087435302810063101",
    "title": "AI 协作完成的烟花大会专题网站",
    "titleEn": "An AI-Built Fireworks Festival Microsite",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者通过 AI 协作制作了一座烟花大会专题网站，并称整个建站流程基本由 AI 完成。 成片规格为 62 秒 · 540×868。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author built a fireworks-festival microsite through AI collaboration, with nearly the entire site-building workflow completed by AI. Output: 62s · 540×868. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yuzuki_goomo/status/2087435302810063101",
    "sourceLabel": "X 原帖 · @yuzuki_goomo",
    "author": "ゆずき | AI×開発 未経験→在宅脱サラ",
    "publishedAt": "2026-08-12T07:05:33.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087435302810063101.mp4",
    "posterUrl": "/posters/x/2087435302810063101.jpg",
    "duration": 62,
    "aspectRatio": "portrait",
    "resolution": "540×868",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Codex × MiniMax H3 が無限の可能性秘めすぎてる！\n\n今度は「花火大会」の特設サイトを作ってみました！\n楽しすぎる...\n\nここまでほぼAIで完結するのでこれからHP制作をどこかに外注することはなくなりそうです",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 128,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit fireworks-festival website and AI-assisted production workflow"
  },
  {
    "id": "x-2087433668600242210",
    "title": "突风与物理运动",
    "titleEn": "A Sudden Gust and Physical Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将作品命名为“突风”，并以此讨论生成模型对物理规律的表现。 成片规格为 8 秒 · 480×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption titles the work “Sudden Gust” and uses it to discuss how generation models handle physical laws. Output: 8s · 480×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AJI_FRY_ART/status/2087433668600242210",
    "sourceLabel": "X 原帖 · @AJI_FRY_ART",
    "author": "aji_fry",
    "publishedAt": "2026-08-12T06:59:03.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087433668600242210.mp4",
    "posterUrl": "/posters/x/2087433668600242210.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "480×640",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【突風】\n\n最近のローカルLLMって物理法則を理解しだしてる感じしますよね。\nkrea2 → minimax h3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 147,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit work title and stated physical-motion focus"
  },
  {
    "id": "x-2087432829001167210",
    "title": "保姆猫摄像机镜头",
    "titleEn": "Nanny Cat Camera Shot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案只将内容标记为“保姆猫摄像机”场景，没有进一步说明动作或环境。 成片规格为 15 秒 · 1104×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption only labels the content as a “Nanny Cat camera” scenario without further describing the action or environment. Output: 15s · 1104×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sadlemonjuice/status/2087432829001167210",
    "sourceLabel": "X 原帖 · @sadlemonjuice",
    "author": "Socratech",
    "publishedAt": "2026-08-12T06:55:43.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087432829001167210.mp4",
    "posterUrl": "/posters/x/2087432829001167210.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1104×1920",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 local nails Nanny Cat camera 😹",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 5,
      "views": 757,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit scenario label only"
  },
  {
    "id": "x-2087431068681212177",
    "title": "《钟楼之心》动画短篇",
    "titleEn": "Clock Tower Heart, an Animated Short",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确将这部动画短篇命名为《钟楼之心》，并列出音乐、角色设计、动态制作和故事构思所使用的多工具链路。 成片规格为 108 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly names the animated short “Clock Tower Heart” and lists a multi-tool workflow for music, character design, motion production, and story development. Output: 108s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIroboHARU/status/2087431068681212177",
    "sourceLabel": "X 原帖 · @AIroboHARU",
    "author": "AIロボ | ハル＠AI副業×初心者の実践記録",
    "publishedAt": "2026-08-12T06:48:44.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087431068681212177.mp4",
    "posterUrl": "/posters/x/2087431068681212177.jpg",
    "duration": 108,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "アニメのショートムービーを作成してみました😊\nタイトル: Clock Tower Heart\nMUSIC:SUNO AI\nキャラデザイン:Midjourny chatGPT\n動画生成:higgsfield AI・DOMO AI(seedance2.0 mini ＆2.5\nMinimax H3)\n薄いストーリー: Claude AI\nこの尺でもクレジット溶けまくりました🤮",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 3,
      "views": 166,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit title, animated-short format, and production toolchain"
  },
  {
    "id": "x-2087428634290319819",
    "title": "复刻一段相似内容",
    "titleEn": "Recreating a Similar Piece",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者看到一段有趣内容后，复刻了一个相似版本，但没有说明具体主题。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author saw an interesting piece and recreated a similar version, without identifying its specific subject. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wenvis_ai/status/2087428634290319819",
    "sourceLabel": "X 原帖 · @wenvis_ai",
    "author": "WenvisAI⚡️",
    "publishedAt": "2026-08-12T06:39:03.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087428634290319819.mp4",
    "posterUrl": "/posters/x/2087428634290319819.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I saw a very interesting video, and I used MiniMax H3 to recreate a similar one.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 91,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit recreation action, with no subject detail"
  },
  {
    "id": "x-2087426356317696156",
    "title": "《自动售货机大放送》与苍水猫",
    "titleEn": "Vending Machine Giveaway with Aominagi",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将这段日常短篇命名为《自动售货机大放送》，并明确写出角色苍水猫。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption names this everyday short “Vending Machine Giveaway” and explicitly identifies the character Aominagi. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087426356317696156",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T06:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087426356317696156.mp4",
    "posterUrl": "/posters/x/2087426356317696156.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n日常エピソード\n『自販機大盤振る舞い』\n蒼水猫（あおみなぎ）ちゃん",
    "engagement": {
      "replies": 0,
      "reposts": 5,
      "likes": 25,
      "views": 1186,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit work title and character name"
  },
  {
    "id": "x-2087425420048388271",
    "title": "世界上是否存在龙",
    "titleEn": "Do Dragons Exist in the World?",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案直接以“世界上是否存在龙”作为主题提问。 成片规格为 24 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption directly frames the subject as the question of whether dragons exist in the world. Output: 24s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lumeroute/status/2087425420048388271",
    "sourceLabel": "X 原帖 · @lumeroute",
    "author": "Mark Han",
    "publishedAt": "2026-08-12T06:26:17.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087425420048388271.mp4",
    "posterUrl": "/posters/x/2087425420048388271.jpg",
    "duration": 24,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Do you believe there are dragons in the world?\n#Minimax-H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 141,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit dragon-centered question"
  },
  {
    "id": "x-2087425126983954619",
    "title": "棒人弹幕与全景背景的转换实验",
    "titleEn": "Transforming Stick-Figure Bullet Hell with a Panoramic Background",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "公开文案说明作者输入了一段带棒人的 Unity 弹幕内容，并以 Qwen 360 Diffusion 制作全景背景，原因是纯黑背景容易让输出整体变黑；配乐在后期加入。 成片规格为 10 秒 · 1024×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author supplied Unity bullet-hell content with a stick figure and created a panoramic background with Qwen 360 Diffusion because a pure black background often darkened the whole output; music was added afterward. Output: 10s · 1024×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yukimino_doga/status/2087425126983954619",
    "sourceLabel": "X 原帖 · @yukimino_doga",
    "author": "ゆきしたミノル@動画",
    "publishedAt": "2026-08-12T06:25:07.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087425126983954619.mp4",
    "posterUrl": "/posters/x/2087425126983954619.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ローカルのMinimax H3にUnityの弾幕動画（棒人形つき）をR2Vで渡した結果～\nカメラワークとか動きの自由度の高さがいいですね👍 背景はQwen 360 DiffusionをFloyoで動かして全天球を置きました。真っ黒背景だとつられて真っ黒になることが多かったんで😅 BGMは後付け。",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 17,
      "views": 1262,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit source content, stick figure, panoramic-background workflow, and post-production note"
  },
  {
    "id": "x-2087424845151899665",
    "title": "两种生成模型的指令跟随与速度体感",
    "titleEn": "Perceived Instruction Following and Speed Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将这次使用体验与 Wan 2.2 对照，并主观认为新方案更容易按指令完成内容，速度也更快。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption contrasts this experience with Wan 2.2 and subjectively finds the newer setup easier to direct and seemingly faster. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/miranic_238/status/2087424845151899665",
    "sourceLabel": "X 原帖 · @miranic_238",
    "author": "りすくら",
    "publishedAt": "2026-08-12T06:24:00.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087424845151899665.mp4",
    "posterUrl": "/posters/x/2087424845151899665.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "話題のMiniMax H3を触ってみたけど、再現度が高くて感動してる！！\n\n今までWan 2.2を使ってたけど、より簡単に指示通りに出せるし、作成速度も速い気がする…！\nこれはまた時間が溶けるやつだなぁ…",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "sourceCaption: explicit comparison with Wan 2.2 and clearly attributed user impressions"
  },
  {
    "id": "x-2087423330605560055",
    "title": "黑胶唱片中心的礼帽人物剪影",
    "titleEn": "A Fedora-Wearing Silhouette at the Center of a Vinyl Record",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色背景中央悬着一张黑胶唱片，橙色唱片标签里嵌着低头戴礼帽的人物剪影。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vinyl record hangs at the center of a black background, with the bowed silhouette of a fedora-wearing figure set inside its orange label. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/OrcaRouter/status/2087423330605560055",
    "sourceLabel": "X 原帖 · @OrcaRouter",
    "author": "OrcaRouter 🐳",
    "publishedAt": "2026-08-12T06:17:59.000Z",
    "addedAt": "2026-08-13T00:38:33.432Z",
    "mediaUrl": "/media/x-2087423330605560055.mp4",
    "posterUrl": "/posters/x/2087423330605560055.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "H3 from @MiniMax_AI is up and running on US inference🔥 Generate up to 15 seconds of 2K video with native stereo audio—from text, images, video, and audio. Start building: \nhttps://www.orcarouter.ai/models/minimax/minimax-h3",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 16,
      "views": 1730,
      "snapshotAt": "2026-08-13T00:38:33.432Z"
    },
    "approvedAt": "2026-08-13T00:38:33.432Z",
    "editorialBasis": "poster-media-review: explicit vinyl record, orange center label, bowed fedora-wearing silhouette, and black graphic background"
  },
  {
    "id": "x-2087423234790867341",
    "title": "孩子与灰熊穿过森林抵达日落山顶",
    "titleEn": "A Child and Bear Cross the Forest to a Sunset Overlook",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名卡通孩子在溪边遇到灰熊，随后两者穿过森林奔跑，并在日落观景处相对而坐。 成片规格为 15 秒 · 1088×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cartoon child meets a gray bear beside a stream, runs through the forest with it, and sits facing the bear at a sunset overlook. Output: 15s · 1088×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2087423234790867341",
    "sourceLabel": "X 原帖 · @ayzalnooor24521",
    "author": "ayzalnoor",
    "publishedAt": "2026-08-12T06:17:36.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087423234790867341.mp4",
    "posterUrl": "/posters/x/2087423234790867341.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1088×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sometimes all you need is a good friend to turn a sad day into an adventure. \nGood friends make every journey more beautiful. \n\nCreated with MiniMax H3 on @wavespeed_ai.\n\nPrompt:\nWarm hand-drawn jungle cartoon with vibrant colors and cinematic animation. A cheerful young jungle boy sits sadly beside a river when his friendly brown bear friend arrives and cheers him up. The bear lifts and spins him, making him laugh, then they run through the jungle, jump over a fallen log, and reach a hill to watch the golden sunset together. Keep both characters consistent, with smooth animation, expressive emotions, natural jungle movement, soft sunlight, seamless transitions, no text or watermark.",
    "engagement": {
      "replies": 64,
      "reposts": 2,
      "likes": 171,
      "views": 9518,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087421450978509024",
    "title": "喷彩武器选择界面展开为第三人称世界",
    "titleEn": "A Paint-Splatter Weapon Menu Expands into a Third-Person World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案描述从主菜单进入武器库、切换喷枪与重炮等装备，确认后界面展开成完整世界，角色以第三人称视角奔跑并喷洒颜料。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption describes entering an armory from the main menu, switching among paint guns and heavy weapons, then confirming a selection as the interface expands into a full world where the character runs and sprays paint from a third-person view. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/derek_wall90176/status/2087421450978509024",
    "sourceLabel": "X 原帖 · @derek_wall90176",
    "author": "Derek Wen｜德里克文",
    "publishedAt": "2026-08-12T06:10:30.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087421450978509024.mp4",
    "posterUrl": "/posters/x/2087421450978509024.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "我用 MiniMax H3 测了一个“喷彩武器选择界面”。\n\n从游戏主菜单开始，点击继续，进入武器库；光标切换喷枪、双持、重炮、滚筒、刷子，不同武器在角色手中实时变化；确认装备之后，UI直接展开成完整游戏世界，角色进入第三人称视角，一边奔跑一边把场景喷满颜料。\n\n整个过程只有15秒，但已经同时包含：\n\n角色一致性 + UI交互 + 多物体切换 + 武器变形 + 镜头运动 + 场景转换 + 第三人称玩法 + HUD + 物理喷彩反馈。\n\n这才是我觉得 H3 这类新一代视频模型真正值得关注的地方。\n让AI理解一个小型的动态系统。\n\n这也是为什么我最近越来越喜欢拿UI、游戏、产品交互去测试视频模型。\n提示词我放评论区，有需要自取。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 222,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit menu flow, weapon switching, interface transition, character action, and third-person perspective"
  },
  {
    "id": "x-2087421025869070809",
    "title": "同一提示词与双参考图下的飞舞毛巾对照",
    "titleEn": "A Windblown Towel Compared with One Prompt and Two References",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "公开文案说明两种生成系统使用同一提示词与两张参考图进行对照，并特别指出被风吹起的毛巾动作，同时记录两者的价格差异。 成片规格为 29 秒 · 1440×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says two generation systems were compared with the same prompt and two reference images, specifically noting the motion of a towel blown by the wind and recording a price difference between them. Output: 29s · 1440×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiko_ai_ai/status/2087421025869070809",
    "sourceLabel": "X 原帖 · @aiko_ai_ai",
    "author": "aiko@AI撮影",
    "publishedAt": "2026-08-12T06:08:49.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087421025869070809.mp4",
    "posterUrl": "/posters/x/2087421025869070809.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "1440×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【比較】MiniMax H3 & Seedance 2.0\n\nPolloAIでMiniMax H3がリリースされたので\n同じプロンプト＋参照画像2枚でSeedance2.0と比較🔥\n\n比較すると、モデルごとの得意不得意が見えますね。\n風で飛ばされたタオルの動きがすごく良い。\n\nMiniMax H3のコストはSeedance 2.0の半分。さらに今ならクレジット50％OFFなので、1/4のコストで生成できます。\n@PolloAIJP",
    "engagement": {
      "replies": 2,
      "reposts": 5,
      "likes": 132,
      "views": 28487,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit controlled comparison, windblown-towel action, and stated cost difference"
  },
  {
    "id": "x-2087415927264670158",
    "title": "用咖啡开启星期三",
    "titleEn": "Starting Wednesday with Coffee",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将作品写成一则以咖啡开启星期三的问候，没有进一步描述人物或场景。 成片规格为 15 秒 · 2944×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption frames the work as a Wednesday greeting centered on starting the day with coffee, without further identifying a character or setting. Output: 15s · 2944×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RudolphR70469/status/2087415927264670158",
    "sourceLabel": "X 原帖 · @RudolphR70469",
    "author": "ChristineR (Love My Country)",
    "publishedAt": "2026-08-12T05:48:34.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087415927264670158.mp4",
    "posterUrl": "/posters/x/2087415927264670158.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Fabulous prompt @SyntheSarah. \n\nLet's get this Wednesday kicked off with coffee.  Have a great day everyone!\n\nCreated with MiniMax H3 (HailuoAI)",
    "engagement": {
      "replies": 32,
      "reposts": 10,
      "likes": 59,
      "views": 906,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit coffee-centered Wednesday greeting"
  },
  {
    "id": "x-2087414444553036236",
    "title": "双紫发角色在灯笼街边跑边喷水",
    "titleEn": "Two Lavender-Haired Characters Run and Spray Water on a Lantern-Lit Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名紫发动画角色手持彩色水枪沿灯笼街道奔跑，并持续向前喷出水流。 成片规格为 6 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two lavender-haired animated characters carry colorful water blasters, run along a lantern-lined street, and continuously spray water forward. Output: 6s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ReiTakao_iMa/status/2087414444553036236",
    "sourceLabel": "X 原帖 · @ReiTakao_iMa",
    "author": "フク@AIart",
    "publishedAt": "2026-08-12T05:42:40.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087414444553036236.mp4",
    "posterUrl": "/posters/x/2087414444553036236.jpg",
    "duration": 6,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "ugc"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#にじぴっぴ\n先日のぴっぴを、\nMinimax H3で動画にしてみました～✨\n\n音楽：SUNO\n編集：CAPCUT",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 23,
      "views": 208,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087413057232544155",
    "title": "DGX Spark 与 RTX Pro 的角色声音尝试",
    "titleEn": "A Character-Voice Attempt on DGX Spark and RTX Pro",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案以 DGX Spark 与 RTX Pro 为运行背景，并记录模型不熟悉 Jeff Garlin、未训练声音细节难以提示的问题；作者另有约四十五秒脚本待完成。 成片规格为 14 秒 · 736×1120。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption places the work on DGX Spark and RTX Pro, reports that the model is unfamiliar with Jeff Garlin and struggles with untrained vocal nuance, and says about 45 more seconds of script remain. Output: 14s · 736×1120. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UrbanAstroFella/status/2087413057232544155",
    "sourceLabel": "X 原帖 · @UrbanAstroFella",
    "author": "AstroFella",
    "publishedAt": "2026-08-12T05:37:09.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087413057232544155.mp4",
    "posterUrl": "/posters/x/2087413057232544155.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "736×1120",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Local AI really do be like that sometimes. (DGX Spark vs RTX Pro)\n\nHaving a lot of fun with MiniMax H3. Unfortunately it isn't familiar with Jeff Garlin and prompting for untrained vocal nuances continues to be a challenge for video models. I've got about 45 more seconds of script written for this gag",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 131,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit hardware comparison context, named-character limitation, vocal-prompting issue, and remaining script length"
  },
  {
    "id": "x-2087410882750107714",
    "title": "经典广告致敬与细微动作调试",
    "titleEn": "A Commercial Homage with Fine-Motion Tuning",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将作品定位为一则能被特定观众认出的广告致敬，并说明在 RTX 4070 Super 12GB 上多次重做，细微表情和动作仍难以准确控制。 成片规格为 15 秒 · 896×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption identifies the work as a homage to a recognizable commercial and says it required several retries on an RTX 4070 Super 12GB, with fine expressions and motion remaining difficult to direct. Output: 15s · 896×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ogax545/status/2087410882750107714",
    "sourceLabel": "X 原帖 · @ogax545",
    "author": "OGAX",
    "publishedAt": "2026-08-12T05:28:31.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087410882750107714.mp4",
    "posterUrl": "/posters/x/2087410882750107714.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "896×576",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "引き続きMiniMax H3で動画三昧！\n4070super(VRAM12GB)でがんばってます\n\n今回は、わかるひとにはわかるあのCMのオマージュ作品ですが、細かな表情や動きはなかなか指示が難しいですね💦何度かリテイクしましたが、やはり本家には到底かないませんでした",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 94,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit commercial-homage format, hardware, retries, and fine-motion prompting difficulty"
  },
  {
    "id": "x-2087410661953610025",
    "title": "流行动画短篇的复刻尝试",
    "titleEn": "Attempting to Recreate a Trending Animated Short",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明作者尝试复刻近期流行的一部动画短篇，并认为尚未掌握其关键风格。 成片规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says the author attempted to recreate a recently trending animated short and felt that its defining style had not yet been captured. Output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lansenai/status/2087410661953610025",
    "sourceLabel": "X 原帖 · @lansenai",
    "author": "澜森",
    "publishedAt": "2026-08-12T05:27:38.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087410661953610025.mp4",
    "posterUrl": "/posters/x/2087410661953610025.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "想试试这个最近很火的动画短片，没掌握到精髓感觉😅 minimax H3",
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 30,
      "views": 2814,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit recreation attempt and animated-short format"
  },
  {
    "id": "x-2087409030184739263",
    "title": "昏暗走廊里倒在湿地与碎瓶旁的女子",
    "titleEn": "A Woman Lies on a Wet Floor beside a Broken Bottle in a Dim Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名黑衣女子侧卧在昏暗走廊的湿润地面上，手边散落着一只破碎的绿色小瓶与玻璃碎片。 成片规格为 38 秒 · 640×368。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman dressed in black lies on her side on the wet floor of a dim corridor beside a broken green bottle and scattered glass. Output: 38s · 640×368. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RobiTelmo28347/status/2087409030184739263",
    "sourceLabel": "X 原帖 · @RobiTelmo28347",
    "author": "ROBI",
    "publishedAt": "2026-08-12T05:21:09.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087409030184739263.mp4",
    "posterUrl": "/posters/x/2087409030184739263.jpg",
    "duration": 38,
    "aspectRatio": "landscape",
    "resolution": "640×368",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Unleash Your Inner Hero via MiniMax H3 @PixVerse_ \n \n #H3onPixverse",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 84,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit woman in black, prone side position, wet floor, dim corridor, green bottle, and broken glass"
  },
  {
    "id": "x-2087408671101989121",
    "title": "城市公园里对照画布与湖畔天际线",
    "titleEn": "Comparing a Canvas with the Lakeside Skyline in a City Park",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开提示词描述一名年轻画家在城市公园架起画架，描绘湖畔天际线，最后举起成画与真实景色进行对照。 成片规格为 15 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public prompt describes a young artist setting up an easel in a city park, painting the skyline across a lake, and finally holding the finished canvas against the real view for comparison. Output: 15s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiwithaly/status/2087408671101989121",
    "sourceLabel": "X 原帖 · @aiwithaly",
    "author": "Ali",
    "publishedAt": "2026-08-12T05:19:44.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087408671101989121.mp4",
    "posterUrl": "/posters/x/2087408671101989121.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A quiet afternoon, a canvas, and a beautiful view.MiniMax H3 makes creative moments feel incredibly lifelike with natural movement, realistic painting details, and seamless character consistency.\n\nCreated with MiniMax H3 on @wavespeed_ai.\n\nPrompt:\n\nA young Western artist sets up an easel in a peaceful city park and paints the skyline overlooking a lake. As the artwork comes to life, she compares the finished painting with the real view, holding the canvas in front of the skyline and smiling proudly. Ultra-photorealistic visuals, cinematic documentary style, realistic brush strokes, authentic hand movements, natural English dialogue, handheld camera movement, detailed cityscape, premium lighting, seamless continuity, and immersive outdoor ambience.",
    "engagement": {
      "replies": 41,
      "reposts": 2,
      "likes": 145,
      "views": 7589,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: complete public prompt with artist, park, lake, skyline painting, and comparison action"
  },
  {
    "id": "x-2087406656048992423",
    "title": "陪酒女郎对战大叔僵尸的自拍对照",
    "titleEn": "A Selfie Comparison: Cabaret Hostess vs. Middle-Aged Zombie",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案以陪酒女郎对战大叔僵尸为主题，在自拍构图、人物一致性、动作、日语台词和乘车等条件下对照两种生成系统，并分别记录构图保持、车门方向和人物混淆等差异。 成片规格为 30 秒 · 2400×1600。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption compares two generation systems around a cabaret hostess confronting a middle-aged zombie, using selfie framing, character consistency, action, Japanese lines, and entering a car, then records differences in framing retention, door orientation, and character mix-ups. Output: 30s · 2400×1600. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mi7_crypto/status/2087406656048992423",
    "sourceLabel": "X 原帖 · @mi7_crypto",
    "author": "M7［mi7］AI",
    "publishedAt": "2026-08-12T05:11:43.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087406656048992423.mp4",
    "posterUrl": "/posters/x/2087406656048992423.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "2400×1600",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【PR】キャバハッピー ✖ PolloAI（ @PolloAIJP ）\nキャバ嬢 VS おっさんゾンビ\n\n現在 #PolloAI ではSeedance 2.5およびMiniMax H3が1ヶ月間50%OFFなので比較検証してみたよ！\n⇒リプ欄にプロンプト公開\n\n【検証結果】\nAIで自撮りという難しい条件での比較。\nまた、参照維持や一貫性・アクション・日本語セリフ・感情表現・車の乗車をプロンプトに盛り込んだ結果です。\nどちらも3回くらいテイクしましたが、表現を忠実にできるには圧倒的にSeedance2.5でした。特に自撮り構図を続けてくれます。リプ欄のプロンプトを見てくれたらわかりますが、自撮りのカメラの手振れ感とか、音のこもり感とかそういう細かい指示が実現できてる。\nレンズ質感も指定がきくのはSD2.5。\nしかし乗車の際のドア向きとかは、SD2.5はずっと間違ってました。人物混同や間違い・破綻もSD2.5の方が多いです。\nH3は、映像質感をどうしても明るくクリアにしますね。ただ、なんとなくの映像にはH3がまとまり合って安定してます。\nお値段や使い方の選別を考慮して使用するのがいいかな。\n【アドバイス】\nSD2.5は、お値段の事もあるので、480pにして使えるものを別途アップスケールするほうがいいです。\n無理して30秒でまわさず15秒、10秒の方が安定します。H3は、きにせずドンドン回せますね。",
    "engagement": {
      "replies": 6,
      "reposts": 6,
      "likes": 149,
      "views": 1716,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit subjects, selfie setup, comparison criteria, repeated attempts, and attributed findings"
  },
  {
    "id": "x-2087404967611269556",
    "title": "十加十秒延展的分镜动画",
    "titleEn": "A Storyboarded Animation Extended from Ten to Twenty Seconds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "公开文案说明这次分镜动画由十秒基础段加十秒延展组成，使用四步 LoRA 并以六步运行，同时提到内容中出现了闯红灯情节。 成片规格为 19 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says this storyboarded animation combines an initial ten-second segment with a ten-second extension, runs a four-step LoRA at six steps, and includes a red-light violation in the content. Output: 19s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/javawock7618/status/2087404967611269556",
    "sourceLabel": "X 原帖 · @javawock7618",
    "author": "ジャヴァノ🍀ボカロP",
    "publishedAt": "2026-08-12T05:05:01.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087404967611269556.mp4",
    "posterUrl": "/posters/x/2087404967611269556.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 Ref2VAはカット割アニメもイケんのかチャレンジ\n\n10秒+10秒(extend) lightx2v v1.0 4step LoRAを6stepsで。\n\n…信号無視ィィィィ！！！！",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 37,
      "views": 4355,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit storyboarded-animation goal, segment extension, sampling setup, and traffic event"
  },
  {
    "id": "x-2087404110974996788",
    "title": "大叔的跑酷式变身段落",
    "titleEn": "A Parkour-Style Transformation Sequence for an Older Man",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案明确将内容描述为一名大叔的跑酷风格变身段落。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption explicitly describes the content as a parkour-style transformation sequence for an older man. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ayu_AI_0912/status/2087404110974996788",
    "sourceLabel": "X 原帖 · @Ayu_AI_0912",
    "author": "アユネオ",
    "publishedAt": "2026-08-12T05:01:36.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087404110974996788.mp4",
    "posterUrl": "/posters/x/2087404110974996788.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3で　おじ変身\n――――\n\nパルクール風変身バンクを\n生成してみた。\nプロンプトは、",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 28,
      "views": 1712,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit subject and parkour-style transformation action"
  },
  {
    "id": "x-2087403772792475975",
    "title": "公寓晨间起床、洗漱与咖啡日常",
    "titleEn": "Waking, Washing, and Making Coffee in an Apartment Morning",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开提示词按时间顺序描述一名成年女性在公寓醒来、洗脸整理头发、冲泡咖啡，再拿着咖啡走到阳台并坐在窗边。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public prompt sequentially describes an adult woman waking in an apartment, washing her face and fixing her hair, making coffee, then carrying it to the balcony and sitting near the window. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SyntheSarah/status/2087403772792475975",
    "sourceLabel": "X 原帖 · @SyntheSarah",
    "author": "Sarah",
    "publishedAt": "2026-08-12T05:00:16.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087403772792475975.mp4",
    "posterUrl": "/posters/x/2087403772792475975.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "anime"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A simple morning, made to feel real. \n\nCreated with MiniMax H3, focusing on natural movement, authentic lighting, realistic details, and the quiet moments of everyday life.\n\nPrompt:\n\nCreate a 15-second ultra-photorealistic live-action video showing a young adult woman going through a simple, peaceful morning routine at home. The entire video should feel like authentic real-life footage, not an advertisement or staged commercial.\nCharacter: A natural-looking adult woman with realistic proportions, subtle facial expressions, natural skin texture, realistic hair movement, and believable body language. Casual comfortable home clothing.\nEnvironment: A modern, cozy apartment with a bedroom, kitchen, and small balcony. Soft natural morning sunlight coming through the windows. Realistic household details, slightly imperfect and lived-in environment.\n0–3s — Waking Up:\nClose-up shot of the woman waking up naturally in bed. Soft sunlight falls across her face. She slowly sits up, adjusts her hair, and looks toward the window. Natural breathing and subtle facial movement.\n3–6s — Morning Routine:\nCut to the bathroom. She washes her face and looks into the mirror while fixing her hair. Water droplets, reflections, skin texture, and hand movements must look completely natural.\n6–9s — Making Coffee:\nMedium shot in the kitchen. She prepares a cup of coffee, pours it into a ceramic mug, and gently stirs it. Steam rises naturally from the hot coffee. Realistic hand movements and object interaction.\n9–12s — Quiet Moment:\nShe walks onto the balcony holding her coffee. Camera follows naturally from behind and slightly to the side. She pauses and looks outside while taking a small sip.\n12–15s — Ending:\nClose cinematic shot of her sitting comfortably near the balcony window, drinking coffee while morning sunlight illuminates the room. She gives a subtle relaxed smile and looks outside.\nCinematography: Photorealistic live-action, natural handheld camera movement, realistic lens behavior, soft morning light, authentic shadows, subtle depth of field, realistic skin and hair, natural motion blur, cinematic but understated color grading.\nRealism Requirements: Every movement must obey real-world physics. Natural human walking, sitting, standing, hand movements, object weight, liquid behavior, steam, reflections, fabric movement, and lighting. No exaggerated expressions or artificial camera movements.\nNegative Prompt: CGI look, cartoon, anime, plastic skin, artificial face, beauty filter, unrealistic body proportions, exaggerated acting, floating objects, warped hands, extra fingers, distorted reflections, unnatural hair, excessive cinematic effects, oversaturated colors, unrealistic physics, stiff movements, glitching, flickering.",
    "engagement": {
      "replies": 75,
      "reposts": 11,
      "likes": 197,
      "views": 11439,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: complete public prompt with character, apartment setting, timed morning routine, and object interactions"
  },
  {
    "id": "x-2087403153910338029",
    "title": "长段生成接 Topaz 放大后的后段噪声",
    "titleEn": "Late-Sequence Noise After Topaz Upscaling",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案记录将长段生成流程接入 Topaz 放大后的结果，并指出越到后半段，噪声成分越明显。 成片规格为 30 秒 · 2046×1126。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption records sending a longer generated sequence through Topaz upscaling and reports that noise becomes increasingly visible toward the later portion. Output: 30s · 2046×1126. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2087403153910338029",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-12T04:57:48.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087403153910338029.mp4",
    "posterUrl": "/posters/x/2087403153910338029.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "2046×1126",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Motion Context → Topaz Upscaler\n\n相性を見たくてテストしてみたけど、やっぱり後半に行くにつれてノイズ成分は増えてしまうな……これに関してはもう今の限界なので今後の発展に期待。にしても一発でここまで出せるようになったのは本当に大きい🤔",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 617,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit generation-to-upscaling workflow and stated late-sequence noise increase"
  },
  {
    "id": "x-2087398080727621644",
    "title": "芙塔芭与希亚奇在宇宙飞船内大闹",
    "titleEn": "Futaba and Shiaki Run Wild Inside a Spaceship",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案直接描述芙塔芭与希亚奇在宇宙飞船内部大闹。 成片规格为 9 秒 · 1536×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption directly describes Futaba and Shiaki running wild inside a spaceship. Output: 9s · 1536×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/flyfront/status/2087398080727621644",
    "sourceLabel": "X 原帖 · @flyfront",
    "author": "flyfront",
    "publishedAt": "2026-08-12T04:37:39.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087398080727621644.mp4",
    "posterUrl": "/posters/x/2087398080727621644.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1536×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "このくらいなら余裕でしょ！\n\n宇宙船内で大暴れするフタバとシアキちゃん\n#フタバート AI動画 MiniMax H3（音が鳴ります）",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 22,
      "views": 1831,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit named characters, spaceship setting, and action"
  },
  {
    "id": "x-2087396962245185935",
    "title": "RTX 3060 Ti 8GB 的五秒六步运行",
    "titleEn": "A Five-Second, Six-Step Run on an RTX 3060 Ti 8GB",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案说明 RTX 3060 Ti 8GB 以六步完成约五秒内容，模型载入后耗时约十九分十八秒，并称日语语音较为自然。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption says an RTX 3060 Ti 8GB completed about five seconds in six steps, taking roughly 19 minutes and 18 seconds after model loading, and reports comparatively natural Japanese speech. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nekopy_ai/status/2087396962245185935",
    "sourceLabel": "X 原帖 · @nekopy_ai",
    "author": "猫P@調査ノート",
    "publishedAt": "2026-08-12T04:33:12.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087396962245185935.mp4",
    "posterUrl": "/posters/x/2087396962245185935.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "RTX 3060 Ti 8GBでMiniMax H3をローカル生成。\n約5秒・6 steps。映像に加えて、日本語音声もかなり自然に出せました。\nモデル読み込み後の生成時間は約19分18秒です。\n#MiniMaxH3 #動画生成AI #ComfyUI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 123,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit hardware, duration, sampling steps, runtime, and stated speech result"
  },
  {
    "id": "x-2087396406994887157",
    "title": "Mochi 想当超级英雄，Sugar 只惦记爆米花",
    "titleEn": "Mochi Wants to Be a Superhero; Sugar Wants the Popcorn",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公开文案将场景设定为两人一起看电影，并明确写出 Mochi 想成为超级英雄，而 Sugar 只想吃爆米花。 成片规格为 30 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The public caption sets the scene during movie night and explicitly says Mochi wants to be a superhero while Sugar only wants the popcorn. Output: 30s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIwithAliya/status/2087396406994887157",
    "sourceLabel": "X 原帖 · @AIwithAliya",
    "author": "Aaliya",
    "publishedAt": "2026-08-12T04:31:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087396406994887157.mp4",
    "posterUrl": "/posters/x/2087396406994887157.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Movie night with your favorite person just hits different. 🍿❤️\n\nMochi wanted to be a superhero… Sugar just wanted the popcorn.\n\nMade this with @RoboNeo_ai  new update.    \n\nThe new MiniMax H3 model is 50% off for BlackGold members right now ($0.0108/sec), sharper & smoother than ever.    \n\nSeedance 2.5 prompt  Try here 👇",
    "engagement": {
      "replies": 48,
      "reposts": 48,
      "likes": 594,
      "views": 49904,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit movie-night setting, named subjects, and stated goals"
  },
  {
    "id": "x-2087396157538312419",
    "title": "红炎狐全力做广播体操",
    "titleEn": "Benihomura the Flame Fox Gives Radio Calisthenics Her All",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "日常短篇《广播体操，全力》以名叫红炎的狐狸少女为主角。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The everyday short Radio Calisthenics, Full Power centers on a flame-fox girl named Benihomura. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087396157538312419",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T04:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087396157538312419.mp4",
    "posterUrl": "/posters/x/2087396157538312419.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n日常エピソード\n『ラジオ体操、全力』\n紅炎狐（べにほむら）ちゃん",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 16,
      "views": 901,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit episode title, everyday-episode format, named flame-fox subject, and full-effort radio-calisthenics action"
  },
  {
    "id": "x-2087394893643817396",
    "title": "“おえんちゃん”、药与天空的多角度创作",
    "titleEn": "A Multi-Angle Piece about Oen-chan, Medicine, and the Sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "参赛作品延续“おえんちゃん”与药的主题并加入天空元素，同时尝试正面之外的多种摄影角度。 成片规格为 38 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The contest entry continues a theme involving Oen-chan and medicine, adds the sky, and experiments with camera angles beyond a frontal view. Output: 38s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hsrk_g_hsrk/status/2087394893643817396",
    "sourceLabel": "X 原帖 · @hsrk_g_hsrk",
    "author": "ヒロ(HG)｜実務で使えるAI活用",
    "publishedAt": "2026-08-12T04:24:59.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087394893643817396.mp4",
    "posterUrl": "/posters/x/2087394893643817396.jpg",
    "duration": 38,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#CryptoNinja × AI動画コンテスト🏆\n\n応募に向けて作品つくり中🎬\n\n前回に続き今回も、おえんちゃんと薬で考えていますが、ここに「そら」を絡めるテーマに挑戦してます🐱\n\n今回は MiniMax H3 を手に入れたので、それを活用🌟\n\nそして、またべえさん（ @matabee_GTO  ）の教材を参考に、正面だけじゃないいろんなカメラアングルにも挑戦中です📚\n\n#NinjaDAO",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 528,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit contest-entry context, Oen-chan, medicine and sky theme, and multi-angle camera experiment"
  },
  {
    "id": "x-2087391923166420992",
    "title": "水手服角色在霓虹舞台间变换姿势",
    "titleEn": "A Sailor-Uniform Character Changes Poses Across Neon Stages",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿水手服的卡通角色从白色背景进入霓虹巷道、灯光舞台和舞池，并连续改变手臂与腿部姿势。 成片规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A sailor-uniform cartoon character moves from a white background through a neon alley, lit stage, and dance floor while changing arm and leg poses. Output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2087391923166420992",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-12T04:13:11.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087391923166420992.mp4",
    "posterUrl": "/posters/x/2087391923166420992.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "SousakuAIのゆるい手書き風のツールつかってぽやりんの画像を変換してMinimax H3で動かしてみた。\n雑なプロンプトでも結構ちゃんと動くんですね！\n\n～動画プロンプト～\n壊れたダンス　緩いけどキレはいい。高速カットで背景とライティング変える\n#SousakuAI @sousakuai",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 169,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087391354838274380",
    "title": "无论处境如何都保持微笑与闪耀",
    "titleEn": "Keep Smiling and Shining Whatever the Situation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖以“无论处境如何都享受今天，并继续微笑、继续闪耀”作为随帖作品的明确主题。 成片规格为 12 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source frames the attached work around enjoying the day whatever the situation and continuing to smile and shine. Output: 12s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/iamsofiaijaz/status/2087391354838274380",
    "sourceLabel": "X 原帖 · @iamsofiaijaz",
    "author": "Aijaz",
    "publishedAt": "2026-08-12T04:10:55.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087391354838274380.mp4",
    "posterUrl": "/posters/x/2087391354838274380.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Enjoy your day, whatever the situation.\n\nKeep smiling, keep shining😂\n\n🎥 Made with MiniMax H3 @Hailuo_AI",
    "engagement": {
      "replies": 21,
      "reposts": 1,
      "likes": 28,
      "views": 945,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit enjoy-the-day, keep-smiling, and keep-shining theme line"
  },
  {
    "id": "x-2087390695804092767",
    "title": "720p 三十秒长时延展与显存占用实验",
    "titleEn": "Thirty-Second 720p Duration Extension and VRAM Experiment",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三十秒、720p 的长时运行耗时十一分钟；作者记录了延长时显存需求不变、可继续串接，以及颗粒感和轻微音频点击声等限制。 成片规格为 30 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A thirty-second 720p long-form run takes eleven minutes; the author records unchanged VRAM demand during extension, continued chaining, and limitations including grain and occasional audio clicks. Output: 30s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2087390695804092767",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-12T04:08:18.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087390695804092767.mp4",
    "posterUrl": "/posters/x/2087390695804092767.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI 1.4 Motion Context Test\n30.00秒動画 ⏱生成時間11分 on 720P\nTurbo LoRA 0.7 → ちょっとザラつくので要設定\n\n機能的には完成。最長1分までに制限してるけど∞延長可\n\n①尺はどれだけ伸ばしても要求VRAM量が変わらない\n②音声は多少のクリックが入る部分がある（現時点の限界",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 1005,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit thirty-second duration, 720p resolution, runtime, extension behavior, VRAM observation, grain, and audio-click limitation"
  },
  {
    "id": "x-2087390098040201594",
    "title": "三张 RTX 4090 并行五十次的 1KF 运动实验",
    "titleEn": "Fifty Parallel 1KF Motion Runs across Three RTX 4090 GPUs",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者以三张 RTX 4090 同时运行五十次，观察 1KF 条件下可实现的运动幅度，并估计总成本低于 250 日元。 成片规格为 14 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author runs fifty jobs concurrently across three RTX 4090 GPUs to examine how much motion is possible under a 1KF condition, estimating a total cost below 250 yen. Output: 14s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/xoMushinxo/status/2087390098040201594",
    "sourceLabel": "X 原帖 · @xoMushinxo",
    "author": "無心 Mushin",
    "publishedAt": "2026-08-12T04:05:55.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087390098040201594.mp4",
    "posterUrl": "/posters/x/2087390098040201594.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "・Minimax H3実験\n1KFでどこまで動けるかのテスト開始\n4090というハイエンドのカード3機同時で合計50本走らせます。案外たぶん250円以下。\nAI時代はサーバー選びも能力か？\n↓音に注意⚠",
    "engagement": {
      "replies": 1,
      "reposts": 40,
      "likes": 132,
      "views": 820,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit 1KF motion experiment, three RTX 4090 GPUs, fifty concurrent runs, and estimated cost"
  },
  {
    "id": "x-2087388100406427777",
    "title": "12GB 显存下从 640p 放大到 1080p 的十秒运行",
    "titleEn": "A Ten-Second 640p-to-1080p Run with 12GB of VRAM",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "十秒内容在 12GB 显存环境中耗时四十四分钟，先以 640p 生成，再借助 SeedVR2 放大到 1080p。 成片规格为 10 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A ten-second output takes forty-four minutes with 12GB of VRAM, starting at 640p before being upscaled to 1080p with SeedVR2. Output: 10s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/developerlin/status/2087388100406427777",
    "sourceLabel": "X 原帖 · @developerlin",
    "author": "Frank Lin",
    "publishedAt": "2026-08-12T03:57:59.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087388100406427777.mp4",
    "posterUrl": "/posters/x/2087388100406427777.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "So goood to have thus kind of 1080p video on 12G vram with MiniMax H3. \n10s => 44 minutes, 640p upscale to 1080p w/ SeedVR2",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 67,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit VRAM capacity, duration, runtime, source resolution, target resolution, and upscaler"
  },
  {
    "id": "x-2087387538923123013",
    "title": "两套模型共用实验节点的运行验证",
    "titleEn": "Run Verification for Experimental Nodes Shared by Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者在实验节点已兼容两套模型后进行一次短运行验证。 成片规格为 118 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author performs a short run verification after experimental nodes begin working with both model variants. Output: 118s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ZanelliDesigns/status/2087387538923123013",
    "sourceLabel": "X 原帖 · @ZanelliDesigns",
    "author": "Zanelli",
    "publishedAt": "2026-08-12T03:55:45.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087387538923123013.mp4",
    "posterUrl": "/posters/x/2087387538923123013.jpg",
    "duration": 118,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just a little test run now I have my experimental nodes working with both of the Minimax H3 models.\n\n#comfyui #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 71,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit experimental-node setup, compatibility with two model variants, and bounded trial run"
  },
  {
    "id": "x-2087387348543930736",
    "title": "蓝光镜面走廊中的黑色长靴",
    "titleEn": "Black Boots in a Blue-Lit Reflective Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "低机位只取一名人物的小腿与黑色长靴，鞋靴和蓝色竖灯映在深色镜面地板上。 成片规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A low camera frames only a person's lower legs and black boots, reflected alongside vertical blue lights on a dark glossy floor. Output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/notoro_ai/status/2087387348543930736",
    "sourceLabel": "X 原帖 · @notoro_ai",
    "author": "のとろ",
    "publishedAt": "2026-08-12T03:55:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087387348543930736.mp4",
    "posterUrl": "/posters/x/2087387348543930736.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax  H3\n実写は流石に強いですね。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 55,
      "views": 2471,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit low-angle lower-leg framing, black boots, blue vertical lights, and dark reflective corridor floor"
  },
  {
    "id": "x-2087384223523229891",
    "title": "书房玻璃桌旁的三人与文件",
    "titleEn": "Three People and Papers Around a Glass Table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三人围在书架环绕的房间与玻璃桌旁，其中两人拿取或指向桌上的文件。 成片规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three people appear around a glass table in a book-lined room while two of them handle or point toward papers on the table. Output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sada_ai/status/2087384223523229891",
    "sourceLabel": "X 原帖 · @sada_ai",
    "author": "sada",
    "publishedAt": "2026-08-12T03:42:35.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087384223523229891.mp4",
    "posterUrl": "/posters/x/2087384223523229891.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created with Minimax h3 on\nhttps://deep-ai.art/app/minimax-h3-video-generator\n\nThat lady is a character I added myself.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 73,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087383759767449769",
    "title": "《罪恶装备》布里吉特的多镜头切分实验",
    "titleEn": "Multi-Shot Editing Experiment with Bridget from Guilty Gear",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《罪恶装备》角色布里吉特用于多镜头系统指令实验，并设置首帧与尾帧；作者同时指出低分辨率导致画面粗糙。 成片规格为 15 秒 · 736×416。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Bridget from Guilty Gear is used in a multi-shot system-prompt experiment with first- and final-frame controls, while the author notes rough imagery caused by low resolution. Output: 15s · 736×416. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hyperboleon/status/2087383759767449769",
    "sourceLabel": "X 原帖 · @hyperboleon",
    "author": "ひゅぺる",
    "publishedAt": "2026-08-12T03:40:44.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087383759767449769.mp4",
    "posterUrl": "/posters/x/2087383759767449769.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "736×416",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ギルティギア、より。 ブリジット\nカット割り多めシステムプロンプトをテスト出力、解像度が低いので画像は荒い。ファーストフレームとエンドフレーム設定あり。\nMiniMax H3 #MiniMaxH3 #AImovie",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 194,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit franchise and character, multi-shot system-prompt experiment, low resolution, and first- and final-frame settings"
  },
  {
    "id": "x-2087382818384961695",
    "title": "《本周新闻 441》：直播责任、生成技术、AI 音乐与监管",
    "titleEn": "Weekly News 441: Streaming Responsibility, Generative Technology, AI Music, and Regulation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "《本周新闻 441》讨论暴力直播中的自由与责任、生成技术、AI 乐曲制作实体唱片，以及《欧盟人工智能法》第 50 条的透明度义务。 成片规格为 20 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Weekly News 441 covers freedom and responsibility in violent livestreaming, generative technology, pressing AI-generated songs to physical records, and the transparency duties in Article 50 of the EU AI Act. Output: 20s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Radio_441/status/2087382818384961695",
    "sourceLabel": "X 原帖 · @Radio_441",
    "author": "飴莉（Radio441）",
    "publishedAt": "2026-08-12T03:37:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087382818384961695.mp4",
    "posterUrl": "/posters/x/2087382818384961695.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今週のニュース441。\n\n・暴力配信をめぐる「自由と責任」\n・MiniMax H3とローカル動画生成\n・Suno、AI生成曲を一枚物レコードに？\n・EU AI Act第50条と透明性義務\n\n配信・動画・音楽・規制、それぞれで進む変化を取り上げます。\nhttps://youtu.be/7WCeH5_mzeM\n#Radio441 #ニュース441 #AIニュース",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 0,
      "views": 154,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit program name and four enumerated news topics"
  },
  {
    "id": "x-2087382114299986310",
    "title": "多平台并行生成的女巨人主题",
    "titleEn": "A Giantess-Themed Multi-Platform Generation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者为提高首轮命中率在多个平台同时生成，并明确说明其中一项内容以女巨人为主题。 成片规格为 8 秒 · 1080×1888。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "To improve the odds of a useful first result, the author generates across multiple platforms at once and explicitly identifies one output as giantess-themed. Output: 8s · 1080×1888. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/emma_shiraishi/status/2087382114299986310",
    "sourceLabel": "X 原帖 · @emma_shiraishi",
    "author": "Emma Shiraishi 白石えま",
    "publishedAt": "2026-08-12T03:34:12.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087382114299986310.mp4",
    "posterUrl": "/posters/x/2087382114299986310.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "1080×1888",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A prompt often doesn't give me the video I want on the first try, so lately I've been generating on multiple platforms at once🙁 \nI totally forgot I was also generating the giantess video on MiniMax H3😛",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 14,
      "views": 2062,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit multi-platform parallel generation and giantess subject"
  },
  {
    "id": "x-2087381059017322535",
    "title": "INT8 与 W4A8 的速度、显存与撩发动作对比",
    "titleEn": "INT8 vs. W4A8 on Runtime, VRAM, and Hair-Sweeping Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一参考图、指令与随机种子下比较 INT8 和 W4A8：W4A8 模型文件约小四成，但在 RTX 3060 Ti 环境中耗时略长、峰值显存未下降，撩发动作也较不清晰。 成片规格为 5 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "With the same reference, prompt, and seed, INT8 and W4A8 are compared: W4A8 is about forty percent smaller on disk, but on an RTX 3060 Ti it runs slightly longer, shows no peak-VRAM reduction, and renders the hair-sweeping action less clearly. Output: 5s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aihonobono2023/status/2087381059017322535",
    "sourceLabel": "X 原帖 · @aihonobono2023",
    "author": "ai honobono | AI Art, Fiction, Music",
    "publishedAt": "2026-08-12T03:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087381059017322535.mp4",
    "posterUrl": "/posters/x/2087381059017322535.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【MiniMax H3：INT8 vs W4A8 ローカル比較】\n\nW4A8は重みを4bit・計算値を8bitに圧縮した軽量モデル。モデル容量が19.53GB→11.68GBと約40%小さいため、ストレージ・メモリ負担の軽減と高速化を期待して試してみました。\n\n検証環境：\n・Windows 11\n・Ryzen 5 3600（6コア）\n・RAM 32GB\n・RTX 3060 Ti 8GB\n・ComfyUI 0.31.0\n\n同じ参照画像・プロンプト・Seedで比較。\n\n共通設定：\n・736×1280／124f／5.17秒\n・Turbo v4・4step・SageAttention\n\n生成時間：\n・INT8（左）：5分50秒\n・W4A8（右）：6分03秒\n\n今回はW4A8が13秒、約3.8%遅い結果。最大VRAMもINT8約7.62GB、W4A8約7.75GBで、削減効果は確認できませんでした。\n\n両方とも顔や手に目立つ破綻はないですが、髪をかき上げる動作についてW4A8は少しわかりにくい感じです。量子化方式が違うため、同じSeedでも動き・ズーム・まばたき・音は変わるんですね。\n「気だるい感じで髪をかき上げる」というプロンプトだったのですが、少し難しかったのかもしれません😅\n\n結論：\n私の3060 Ti環境では、速度と精度を重視するならINT8。W4A8はモデル容量を約40%節約したい場合の選択肢になりそうです。\n\n#MiniMaxH3 #ComfyUI",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 11,
      "views": 3958,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit identical inputs and seed, quantization variants, file sizes, hardware, runtimes, peak VRAM, and prompted hair-sweeping action"
  },
  {
    "id": "x-2087381057964216482",
    "title": "杀人阿米巴露艾拉的独唱选拔",
    "titleEn": "Luella the Killer Amoeba Enters a Solo Singing Contest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "日常短篇《歌喉选拔，独唱组》以名叫露艾拉的杀人阿米巴为主角。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The everyday short Solo Singing Contest centers on Luella, a killer amoeba. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087381057964216482",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T03:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087381057964216482.mp4",
    "posterUrl": "/posters/x/2087381057964216482.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n日常エピソード\n『のど自慢、ソロの部』\n殺人アメーバのルエラちゃん",
    "engagement": {
      "replies": 0,
      "reposts": 4,
      "likes": 21,
      "views": 988,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit episode title, everyday-episode format, named killer-amoeba subject, and solo singing-contest premise"
  },
  {
    "id": "x-2087381045343891479",
    "title": "RTX 4080 SUPER 与 ComfyUI 的运行评估",
    "titleEn": "Runtime Evaluation with an RTX 4080 SUPER and ComfyUI",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖将随帖内容明确标为 RTX 4080 SUPER 与 ComfyUI 环境中的运行评估。 成片规格为 34 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly labels the attached work as a runtime evaluation using an RTX 4080 SUPER and ComfyUI. Output: 34s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wxp22888/status/2087381045343891479",
    "sourceLabel": "X 原帖 · @wxp22888",
    "author": "数智先行者",
    "publishedAt": "2026-08-12T03:29:57.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087381045343891479.mp4",
    "posterUrl": "/posters/x/2087381045343891479.jpg",
    "duration": 34,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI视频自由了？4080 SUPER+MiniMax H3+ComfyUI本地实测",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 65,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit GPU, ComfyUI environment, and on-device evaluation framing"
  },
  {
    "id": "x-2087378796014145822",
    "title": "瀑布古楼前的白衣长发女子",
    "titleEn": "A Long-Haired Woman in White before a Waterfall and Traditional Buildings",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名身穿白色古装、留着黑色长发的女子位于水潭边，身后瀑布从传统木楼之间倾泻而下。 成片规格为 10 秒 · 832×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long, black-haired woman in flowing white period clothing stands by a pool as a waterfall descends between traditional wooden buildings behind her. Output: 10s · 832×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EhcoRelay/status/2087378796014145822",
    "sourceLabel": "X 原帖 · @EhcoRelay",
    "author": "Jack",
    "publishedAt": "2026-08-12T03:21:01.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087378796014145822.mp4",
    "posterUrl": "/posters/x/2087378796014145822.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 80,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit long-haired woman, white period clothing, pool, waterfall, and traditional wooden architecture"
  },
  {
    "id": "x-2087378127144321137",
    "title": "城市人行道上自拍的棕色针织裙女子",
    "titleEn": "A Woman in a Brown Knit Dress Takes a Selfie on a City Sidewalk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名长发女子身穿棕色针织裙、背着肩包，以伸臂自拍视角出现在有车辆与绿化的城市人行道上。 成片规格为 10 秒 · 768×1376。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-haired woman in a brown knit dress with a shoulder bag appears from an arm's-length selfie angle on a city sidewalk lined with plants and traffic. Output: 10s · 768×1376. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EhcoRelay/status/2087378127144321137",
    "sourceLabel": "X 原帖 · @EhcoRelay",
    "author": "Jack",
    "publishedAt": "2026-08-12T03:18:21.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087378127144321137.mp4",
    "posterUrl": "/posters/x/2087378127144321137.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1376",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "ugc",
      "vertical"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 72,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit long-haired woman, brown knit dress, shoulder bag, arm's-length selfie framing, city sidewalk, vegetation, and traffic"
  },
  {
    "id": "x-2087376661239189871",
    "title": "角色与摩托车一致性的 704p 放大评估",
    "titleEn": "704p Upscaling Evaluation of Character and Motorcycle Consistency",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "704p 内容经过放大后，作者认为角色与摩托车的一致性较好，但表情偏僵硬，细节运动也较少。 成片规格为 7 秒 · 1408×1408。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "After upscaling a 704p output, the author finds character and motorcycle consistency strong, while noting stiff expressions and limited fine-detail motion. Output: 7s · 1408×1408. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI6377546630069/status/2087376661239189871",
    "sourceLabel": "X 原帖 · @AI6377546630069",
    "author": "AIの鐘（ べる）",
    "publishedAt": "2026-08-12T03:12:32.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087376661239189871.mp4",
    "posterUrl": "/posters/x/2087376661239189871.jpg",
    "duration": 7,
    "aspectRatio": "square",
    "resolution": "1408×1408",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "夏休みの自由研究じゃないですが、連休前から気になってた MiniMax H3 を導入からカスタマイズまで一気にやってみました。\n（と言ってもほとんどチャッピーがやってくれたのですが😅）\n\n動画は704Pをアップスケールしたものです。\nキャラやバイクの一貫性は良い感じですが、表情が硬かったり、細かいところがあまり動いてないのは気になりますね。\nただ、うちのPCは動くギリギリのスペックなのでもっといいグラボがあれば設定やプロンプトで改善されるかもしれません。\n\nあとSeedance2.5も気になるので触ってみたいんですけど、高すぎる・・",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 36,
      "views": 840,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit 704p upscale, character and motorcycle subjects, consistency assessment, stiff expressions, and limited fine-detail movement"
  },
  {
    "id": "x-2087376502757478611",
    "title": "粉色世界里的奇幻茶会",
    "titleEn": "A Fantasy Tea Party in a Pink World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "奇幻茶会发生在粉色世界中，成片被放大至 4K，并另附不同模型版本供比较。 成片规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fantasy tea party unfolds in a pink world, with the result upscaled to 4K and a version from another model supplied for comparison. Output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2087376502757478611",
    "sourceLabel": "X 原帖 · @tokyo_Valentine",
    "author": "とすくん",
    "publishedAt": "2026-08-12T03:11:54.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087376502757478611.mp4",
    "posterUrl": "/posters/x/2087376502757478611.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ピンク色の世界のファンタジーなお茶会の動画\nを作りました。\nSeedance 2.5では出ない味がSeedance 2.0で出た気がしました。\n\n※4Kにアプスケしました。\n\n比較用でリプ欄にMiniMax H3バージョンも置いておきます。\n\nCharacter:Nano Banana 2 / Grok Imagine\nMovie:Seedance 2.0",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 46,
      "views": 929,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit fantasy tea-party subject, pink-world setting, 4K upscale, and alternate-model comparison"
  },
  {
    "id": "x-2087375326494294388",
    "title": "让图 A 角色在图 B 场景复刻片段 C 的舞蹈",
    "titleEn": "A Character from Image A Dances in Location B Like Clip C",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "实验把图 A 的角色放进图 B 的地点，并让角色按照片段 C 的动作跳舞。 成片规格为 5 秒 · 1440×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The experiment places the character from Image A in the location from Image B and has that character dance using the movement from Clip C. Output: 5s · 1440×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ui_nyan/status/2087375326494294388",
    "sourceLabel": "X 原帖 · @ui_nyan",
    "author": "ういにゃん【アニソンDJ】",
    "publishedAt": "2026-08-12T03:07:14.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087375326494294388.mp4",
    "posterUrl": "/posters/x/2087375326494294388.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1440×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3凄すぎる....\n「画像Aのキャラを画像Bの場所で動画Cのように踊らせて」がめちゃくちゃ簡単...\nしかも家のパソコンで無限に試せるのが神",
    "engagement": {
      "replies": 13,
      "reposts": 117,
      "likes": 916,
      "views": 61527,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit character-reference image, location-reference image, dance-motion reference, and cross-reference composition",
    "featured": {
      "at": "2026-09-05T15:50:15.689Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2087375102925316225",
    "title": "室内俯身靠近镜头的短发女子",
    "titleEn": "A Short-Haired Woman Leans toward the Camera Indoors",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名黑色短发女子身穿白色上衣与蓝色牛仔外套，在暖色室内俯身靠近镜头。 成片规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short, black-haired woman in a white top and blue denim jacket leans close toward the camera in a warmly lit interior. Output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Suzuna_AI_girls/status/2087375102925316225",
    "sourceLabel": "X 原帖 · @Suzuna_AI_girls",
    "author": "鈴音🔔@Suzuna_japanese_AI_girls",
    "publishedAt": "2026-08-12T03:06:20.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087375102925316225.mp4",
    "posterUrl": "/posters/x/2087375102925316225.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こんにちはだよー🔔\n\n#BellEchoArt\nLTX2.5がリリースされたみたいだね🎶\n\nまだMinimax H3すら満足に試せてないのに💦\nなんかMinimax以上の性能とか書いてたけどほんとかな😑\n\n導入めんどいし情報でるまでは様子見かなぁ\n\ncreated with Topview Seedance 2.0",
    "engagement": {
      "replies": 0,
      "reposts": 5,
      "likes": 42,
      "views": 1790,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit short-haired woman, white top, denim jacket, close leaning pose, and warm indoor setting"
  },
  {
    "id": "x-2087374553895104873",
    "title": "用分隔符串联段落的 720p 三十秒长时运行",
    "titleEn": "A Thirty-Second 720p Long-Form Run Joined with Prompt Separators",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三十秒、720p 的长时运行耗时十一分钟，段落之间只需在指令中写入分隔符即可衔接，作者也称声音问题已大体解决。 成片规格为 30 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A thirty-second 720p long-form run takes eleven minutes, with segments joined by adding separators to the prompt; the author also says the sound issues are largely resolved. Output: 30s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2087374553895104873",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-12T03:04:09.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087374553895104873.mp4",
    "posterUrl": "/posters/x/2087374553895104873.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI Motion Context\n30.00秒動画 ⏱生成時間11分 on 720P\n\nサウンド面も概ね解消。セグメント間はプロンプトで --- と書くだけなのでライティングも楽（その仕様さえ渡せばスキル化でも出来る）\n\n出来の良さに感動してしまった……",
    "engagement": {
      "replies": 4,
      "reposts": 3,
      "likes": 67,
      "views": 6109,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit duration, resolution, runtime, prompt-separator segmentation, and reported sound improvement"
  },
  {
    "id": "x-2087373525422100652",
    "title": "参考 AI 指数购买赛马券",
    "titleEn": "Buying Horse-Racing Tickets Using an AI Index",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者明确表示当天将参考 AI 指数去购买赛马券。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author explicitly says they are going to buy horse-racing tickets that day using an AI index as guidance. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yuu66114977/status/2087373525422100652",
    "sourceLabel": "X 原帖 · @yuu66114977",
    "author": "プロエンジニアyuu@競馬競艇競輪オートAI予想",
    "publishedAt": "2026-08-12T03:00:04.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087373525422100652.mp4",
    "posterUrl": "/posters/x/2087373525422100652.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "皆さんこんにちは！今日はAI指数を参考に馬券買いに行ってきます！\nMiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 288,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit author action, AI-index reference, and horse-racing-ticket objective"
  },
  {
    "id": "x-2087373516534419659",
    "title": "迷失在水母之海",
    "titleEn": "Lost in a Sea of Jellyfish",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖明确将随帖作品描述为迷失在一片水母之海。 成片规格为 15 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly describes the attached work as becoming lost in a sea of jellyfish. Output: 15s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/foxyy4i/status/2087373516534419659",
    "sourceLabel": "X 原帖 · @foxyy4i",
    "author": "HIKARI✨",
    "publishedAt": "2026-08-12T03:00:02.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087373516534419659.mp4",
    "posterUrl": "/posters/x/2087373516534419659.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "クラゲの海に迷い込んだ。🪼🌊\n\nGPT image MiniMax H3\n@Hailuo_AI #MiniMaxH3",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 40,
      "views": 4470,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit lost-person premise and sea-of-jellyfish setting"
  },
  {
    "id": "x-2087373132432556201",
    "title": "蓝光人群前闭眼站立的双色发角族角色",
    "titleEn": "A Two-Tone-Haired Horned Character Stands Eyes Closed before a Blue-Lit Crowd",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白双色头发、双角与尖耳的卡通角色穿着红黑西装站在人群前，双眼闭合，背景笼罩在蓝色灯光中。 成片规格为 10 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cartoon character with split black-and-white hair, two horns, and pointed ears wears a red-and-black suit before a crowd, eyes closed beneath blue lighting. Output: 10s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ChiralChikage/status/2087373132432556201",
    "sourceLabel": "X 原帖 · @ChiralChikage",
    "author": "Chiral 🤓🎧【VMoid】",
    "publishedAt": "2026-08-12T02:58:30.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087373132432556201.mp4",
    "posterUrl": "/posters/x/2087373132432556201.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "When I decide to gen locally with Minimax H3",
    "engagement": {
      "replies": 5,
      "reposts": 4,
      "likes": 50,
      "views": 720,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit split-color hair, horns, pointed ears, red-and-black suit, closed eyes, crowd, and blue-lit venue"
  },
  {
    "id": "x-2087373025310322971",
    "title": "两种加速权重设置的输出对比",
    "titleEn": "Output Comparison of Two Acceleration-Weight Settings",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一套缩放版加速权重分别采用四步版强度 0.7 与八步版强度 1.2，并在八步采样条件下比较两个输出。 成片规格为 15 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two resized acceleration-weight variants—four-step at strength 0.7 and eight-step at strength 1.2—are compared under an eight-step sampling setup. Output: 15s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ITOKtw/status/2087373025310322971",
    "sourceLabel": "X 原帖 · @ITOKtw",
    "author": "ITOK",
    "publishedAt": "2026-08-12T02:58:05.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087373025310322971.mp4",
    "posterUrl": "/posters/x/2087373025310322971.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "drbaphさんがlightx2vのv1.0のMiniMax-H3 turbo loraのリサイズ版を出しています。8 stepと4 stepの両方を出しています。当たり前ですが，8stepはダメですね。\nhttps://huggingface.co/drbaph/MiniMax-H3-Turbo-Lora-ComfyUI/tree/main\n1つ目：4step 強度0.7 8 step，2つ目：8step 強度1.2 8step",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 12,
      "views": 1524,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit resized acceleration-weight variants, four- and eight-step versions, strengths, shared sampling count, and ordered output comparison"
  },
  {
    "id": "x-2087372806992564613",
    "title": "照明异常下的恐怖短篇终局调整",
    "titleEn": "Refining a Horror Finale amid Lighting Problems",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "这次恐怖题材运行仍出现照明问题，但作者认为终局效果已有明显改善，并使用 ComfyUI 与 CapCut 辅助完成。 成片规格为 9 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "This horror-themed run still has lighting problems, but the author considers the finale substantially improved and uses ComfyUI with CapCut for assistance. Output: 9s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Luna_R4ze/status/2087372806992564613",
    "sourceLabel": "X 原帖 · @Luna_R4ze",
    "author": "LunaRaze",
    "publishedAt": "2026-08-12T02:57:13.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087372806992564613.mp4",
    "posterUrl": "/posters/x/2087372806992564613.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ok here is MiniMax H3 again with some horror @PromptSin @Malzahran2  but something is happening with the lighting that is creating problems. Much better finale with this. Again on ComfyUI with CapCut for assistance 🔥",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 9,
      "views": 513,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit horror genre, lighting problem, improved finale, and ComfyUI-plus-CapCut workflow"
  },
  {
    "id": "x-2087371588668907864",
    "title": "梦幻植物温室里的少女与歌词排版",
    "titleEn": "A Girl and Lyric Typography in a Dreamlike Botanical Greenhouse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "高密度二维手绘少女置身玻璃植物温室，白色与蓝色花朵、藤蔓、五线谱和细体歌词排版共同构成画面，光线由白与淡青逐渐转为淡琥珀色。 成片规格为 14 秒 · 2944×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A densely hand-drawn 2D girl appears inside a glass botanical greenhouse, where white and blue flowers, vines, musical staves, and fine lyric typography share the composition as the light shifts from white and pale cyan toward soft amber. Output: 14s · 2944×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/haruuraeadss/status/2087371588668907864",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-08-12T02:52:22.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087371588668907864.mp4",
    "posterUrl": "/posters/x/2087371588668907864.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MV作りたいなら、MiniMax H3 がコスパいいね✨\n\nSunoで作った曲をオーディオリファレンスに入れて、\n曲のタイミングに合わせてリリックタイポグラフィまで出せば、\n簡単にプロ級の仕上がりになるからね。\n\n@Hailuo_AI #MiniMaxH3　\n\nプロンプト👇\n\n高品質アニメ映像。 【作品ジャンル・世界観】 夢幻的ファンタジーの植物温室と音楽標本世界。劇場版クラスの高密度な2D手描きアニメ、繊細なキーアニメーション、上質なコンポジット、透明感のあるライティング。高いガラス温室、白い薔薇、白い小花、青い小花、蔦、露、五線譜、音符、細い円弧、極細の植物線画が一つの視覚言語として連動する。白、淡いシアン、空色、淡い菫色を基調にした明るい光から始まり、終盤だけ淡い琥珀色を帯びる。床には格子状の光が落ち、ガラス骨組みと白い台が光を受ける。背景が変わっても、細い白文字と水平譜線は画面固定座標に留まり、カット間の接着剤になる。 【参照役割と優先順位】 Image1は顔と同一性の最優先参照であり、同時に衣装、装飾、体型、全身デザインの唯一参照。顔、瞳、髪、体格、衣装、装飾、全身シルエットが食い違う場合は常にImage1を優先する。Image1の背景、卓上の小物、椅子、机、元の構図、元のポーズ、カメラ距離、フレーミングは再現しない。 Image2とImage3はタイポグラフィ設計だけの参照。各画像内の主文字群の字形、細い書体、大文字と斜体の筆記体の対比、文字間、各群の画面上の相対位置、五線譜、音符、点、大きな細い円弧、小さな技術注記、細い斜体の小語、極細の植物線画の意匠だけを取る。Image2とImage3の背景、ガラス温室の描写、花瓶、白い台、床、青い色調は一切再現しない。Image2とImage3から人物の顔、髪、衣装、年齢感も取得しない。 Audio1は唯一の音楽参照。譜線の伸長、音符と点の出現、単語レイヤーの露出、編集マスクの進行、カメラの開始と停止、瞬き、光量変化をAudio1のフレーズ開始、立ち上がり、余韻へ同期する。Audio1以外のBGMを入れない。声、歌唱、ナレーションを追加せず、映像側で効果音を新規生成しない。 人物は全編でImage1の一人だけ。各フレームに見える人物形体は合計一人。追加人物、別人、双子、複製、残像人物、透明な人物の重複、鏡や床に映る全身の別個体を出さない。 【キャラクター固定・同一人物】 全編でImage1と同じ少女を完全な同一人物として固定する。繊細で丸みのある同じ顔、同じ顎と頬、暖かな象牙色の肌、淡い血色、同じ年齢感。澄んだ青の大きな瞳、虹彩下部の明るい水色、白い星状ハイライト、長い上まつ毛を維持する。白銀から真珠色の長い髪、柔らかな前髪、側頭部から後方へ続く編み込み、後方でまとめたハーフアップを固定する。右側頭部の金色リボンと青い雫型の髪飾りは消さない。 衣装はImage1と同じ白いハイネックの長袖ドレス。立ち襟、胸当て、袖先、裾の淡い水色の花刺繍、縦に並ぶ白い小ボタン、柔らかな幅広袖、金色の細い腰リボンと結び目、淡い水色の裾フリルまで一致させる。細身で上品な体格、白と水色と金色の固有配色を変えない。武器なし、装備品なし、小物なし。両手は常に自然な五本指で、何も持たず、顔や文字を隠さない。 変化を許すのは呼吸、瞬き、表情、姿勢、瞳の向き、髪、リボン、袖、裾の小さな揺れだけ。別人化、汎用的なAIアニメ顔、平均顔、幼児化、瞳色変更、髪型変更、衣装交換、刺繍省略、髪飾り消失、手指変形、余分な指、人物複製を禁止する。この同一人物固定は次項の画風固定とは別の独立条件であり、互いに代用しない。 表情変化は、まず目、次に眉、次にまぶた、次に口元、最後に視線の順で最小限に行う。柔らかな微笑みを全編で保つ。大きな演技、会話、歌唱の口形は入れず、静かな呼吸と瞳の変化だけで感情を見せる。 【画風固定：2D手描き】 Image1の極細で淡い青灰色の線、黒すぎない輪郭線、細かな髪束を維持する。影は透明感のある二段から三段。肌と布は柔らかな水彩階調、髪は真珠色の細いハイライト、瞳、露、ガラスは小さく鋭い反射。人物だけを過度に平坦化せず、ガラス骨組み、白い薔薇、白い小花、青い小花、蔦、葉、露、布、刺繍まで同じ密度の2D手描き背景として描く。肌は暖かな象牙色を保ち、青い環境光で灰色や病的な青肌にしない。白いドレス、金色リボン、顔、瞳は明るい背景から明瞭に分離する。 太い黒輪郭、簡略化されたテレビアニメ調、単層の平坦なセル塗り、低密度背景、滑らかなCG・3D化、写実・半写実化、プラスチック肌、過剰な美肌ぼかし、くすんだ色、汎用的なAIアニメ顔、作風混合を禁止する。 【タイトル・譜線・注釈の固定】 Image2とImage3に存在する主文字群を、参照画像由来の不変な2Dグラフィック面として維持する。文字を新しく再描画したり別の字形へ変形したりせず、正しい字形をクリップマスク、透明度、単語レイヤー単位の短い移動だけで見せる。綴り、記号、読点、大文字と小文字の別、細い書体、文字間、大小関係、各群の画面上の相対位置を一切変えない。ある群を別の群へ溶かして変形させない。小さな技術注記と細い斜体の小語も、各参照内の字形と位置のまま扱い、書き換えたり別の文へ置き換えたりしない。偽の単語、擬似文字、追加の文章、字幕を生成しない。 主文字は白から淡い象牙色、細いサンセリフと細い斜体の筆記体。補助要素は四本から五本の水平譜線、音符、点、大きな細い円弧、極細の植物線画。譜線と円弧は約一ピクセル、低めの不透明度。すべて画面固定座標に置き、カメラの遠近やパースへ吸着させない。 各群の出現は必ず、譜線がその段を左から右へ伸びる、点と音符が現れる、主文字が単語レイヤー単位の短い左向きクリップマスクとフェードで組み上がる、細い小語と植物線画が最後に薄く現れる、の順。消去は小語、主文字、譜線の順。全文同時表示や一文字ずつの派手なタイプ入力にはしない。文字の跳ね、回転、震え、液体化、融解を入れない。 同じ参照内では、一度露出した群を以後その画面座標に固定したまま残す。参照がImage2からImage3へ変わる時だけ、先にImage2の全要素を小語、主文字、譜線の順で完全に消してから、Image3の正しい文字面を新しく露出する。二つの参照の文字を同じ画面に同時に出さない。 【顔の保護領域】少女の髪の生え際から顎先まで、左右は両耳の外側までを顔の保護領域とする。主文字、譜線、円弧、音符、点、植物線画、小さな注記は、全区間を通じてこの保護領域へ一本も一文字も入れない。文字が顔の手前を横切る、顔の上へ薄く重なる、顔の後ろから透ける、顔の輪郭に接する、いずれも禁止する。近づきそうな場合は字形を変形させず、人物の配置と大きさの側で避ける。文字と保護領域の間には常に主文字一行分以上の空白を保つ。両手も同じ扱いで横断しない。 字幕箱、黒帯、太い縁取り、強いグロー、大きな影、カラオケ式の色塗り、RGB分離、激しいグリッチを禁止する。 【0.0秒から3.0秒：Image2の上段】 開始時は前カットの終端状態をそのまま受け取る。明るいガラス温室、白い花と青い小花に囲まれたImage1、柔らかな微笑み、画面に残る前カットの細い文字と譜線。構図、色、明るさ、コントラストを開始の瞬間に跳ばさず、背景も人物も切り替えない。まず前カットの小語、主文字、譜線がこの順で静かに消え、画面から文字だけが無くなる。続いて一本の長い譜線が画面左端から右端へ進み、その通過領域だけを硬い水平マスクとして次の構図へ置き換える。これはカメラ移動ではなく編集遷移であり、二つの画を半透明に混ぜない。 新しい画では、カメラ位置は正面よりわずかに左、カメラ高は少女の目線の高さ、カメラの向きは正面、人物サイズは胸上、背景の消失軸は左奥の温室通路。少女は画面右三分の一に立ち、肩線を左下から右上への緩い斜線にし、上体をわずかに傾け、頭部を逆方向へごく小さく傾ける。正面直立にしない。カメラはごくゆっくり前進する。目的は、画面左の広い余白が文字の置き場だと読ませながら、右の顔を同時に読ませること。 少女は閉じたままの口で一度だけ自然に瞬きし、瞳を左の文字側へ静かに向ける。呼吸、髪先、金色リボン、袖の微動を持つ生きた2Dアニメとして見せる。 画面左上に譜線が左から右へ伸び、点と小さな音符が順に灯り、Image2の上段の二行を単語レイヤー単位で露出する。大きな細い円弧の一部と小さな技術注記も薄く現れる。文字は画面左半分に収まり、右三分の一の少女と顔の保護領域へは一文字も入らない。この区間ではImage2の下段とImage3のすべての群を完全に出さない。断片、影、反射、残像、先出しのシルエットも出さない。 【3.0秒から6.0秒：Image2の下段】 3.0秒で上段の譜線が画面右端まで一定速度で長く伸び、その通過領域だけを硬い水平マスクとして次の構図へ置き換える。カメラ移動ではなく編集遷移。二つの背景を半透明に混ぜず、マスクの前方と後方を不透明に保つ。ランダムなクロスフェード、渦回転、白フラッシュを入れない。Image2の上段は字形も画面位置も変えずそのまま残り、カット間の接着剤になる。Image1も一人のまま受け渡し、二重像や別ポーズを同時に出さない。 新しい画では、カメラ位置は少女の左斜め前、カメラ高は少女の胸より低い位置、カメラの向きはやや見上げるあおり、人物サイズは腰上、背景の消失軸は左奥の高いガラス壁と蔦。少女は画面右三分の一に立ち、あおりのため頭部と顔は画面の上部三分の一に収まる。上体を斜めに向け、前景の白い薔薇とふくらんだ幅広の袖が左下から右上への対角線を作る。裾は前景の花で自然に切れる。カメラは一定速度で短く前進して止まる。目的は、左上から中央下へ降りる文字の流れと、正面直立ではない身体軸と、光の奥行きを同時に読ませること。 譜線が段ごとに順に伸び、音符と点が現れ、Image2の下段の各群を上から順に単語レイヤー単位で露出し、Image2のレイアウトを完成させる。最後に細い斜体の小語と極細の植物線画が薄く現れる。文字は少女の胸より下、または顔と両手の保護領域の外でだけ読める位置に収まる。この区間ではImage3のすべての群を完全に出さない。断片、影、反射、残像、先出しのシルエットも出さない。 【6.0秒から9.0秒：Image3の上段】 6.0秒でImage2の小語と注記、主文字、譜線と円弧がこの順で完全に消える。文字より遅れて前景の白い薔薇の花弁が一枚レンズへ近づき、その輪郭と最後の長い譜線を一体の硬い編集マスクにして、覆われた領域から次の構図を明かす。渦回転やクロスフェードにしない。Image2の文字は一つも残さない。 新しい画は温室の高いガラス天井を見上げる構図。人物は出さない。カメラ位置は温室の中央通路、カメラ高は少女の目線より低い位置、カメラの向きは大きく見上げる仰角、人物サイズは該当なし、背景の消失軸は中央上のガラス骨組みの収束点。白銀の細い骨組みが放射状に収束し、蔦と白い小花が縁を飾り、その向こうの空から光が斜めに差し込む。カメラはごくゆっくり前進する。目的は、光と空間の高さと、Image3の新しい文字を置く広い余白を一度に見せること。 画面左上に新しい譜線が左から右へ伸び、点と音符が現れ、Image3の上段の二行を単語レイヤー単位で露出する。大きな細い円弧と小さな技術注記も薄く現れる。この区間ではImage3の下段を完全に出さない。断片、影、反射、残像、先出しのシルエットも出さない。 【9.0秒から13.0秒：Image3の下段と終端】 9.0秒で上段の譜線が右端まで長く伸び、その通過領域だけを硬い水平マスクとして次の構図へ置き換える。Image3の上段は字形も画面位置も変えずそのまま残る。ここから差し込む光が淡い琥珀色を帯び始め、白と淡いシアンの世界に暖かい色温度が一段だけ加わる。白飛びや全面のオレンジ化にはしない。 新しい画では、カメラ位置は正面、カメラ高は少女の目線の高さ、カメラの向きは鼻筋と両目の中央を通る正面軸、人物サイズは胸上より寄った顔の近景、少女は画面右三分の一からわずかに中央寄り、背景の消失軸は背後のぼけた花と琥珀色の光。背景の花と天井は柔らかくぼかす。目的は、青い瞳、虹彩の明るい水色、白い星状ハイライト、微細な呼吸を、琥珀色の光の中の顔報酬として明瞭に見せること。 譜線が段ごとに順に伸び、音符と点が現れ、Image3の下段の各群を上から順に単語レイヤー単位で露出し、Image3のレイアウトを完成させる。最後に細い斜体の小語と極細の植物線画が薄く現れる。完成した文字は画面左側から中央下にかけて収まり、右の顔と顔の保護領域には一文字も入らない。 移動する光点、フレア、星、光粒を顔の表面へ着地させない。光は背景、髪の外周、肩、画面周辺で消える。瞳の報酬は、両目の内部に元からある小さなキャッチライトが同時にわずかに明るくなる表現に限定する。 Image1は一度ゆっくり瞬きし、わずかに息を吸い、口を閉じた柔らかな微笑みのまま、最後に青い瞳を正面へ向ける。 11.6秒でカメラが滑らかに短く後退し、白い薔薇と青い小花に囲まれて立つImage1の腰上が現れる。少女は画面右三分の一に留まり、両手は体の前で自然に軽く重なり、指は正しい本数で静止する。完成したImage3の文字と譜線は画面固定座標のまま、顔と両手の保護領域に掛からず、人物の手前にも重ならない。 終端では、最下段の細い水平譜線だけが右端の終点まで伸びる。譜線が終点へ到達した瞬間、カメラが完全に止まり、髪先、金色リボン、袖、周囲の花弁の最後の小さな揺れがこの順で収まる。顔、瞳、両手、白いドレス、Image3の完成した文字、白い薔薇と青い小花、琥珀色を帯びた光が、互いに重ならないまま同時に明瞭に読める。Audio1の余韻が消える正確な瞬間に切る。新しいポーズ、追加の寄り、長い静止、無関係なフェードを後ろへ足さない。 【全体制御】 カメラは一定で滑らかにし、手持ち揺れ、急加速、急停止、ランダムズーム、無意味な回転、派手なオービットを入れない。タイトル変化を物理カメラ運動として扱わない。全区間を通じて、文字、譜線、記号が少女の顔と両手に重なるフレームを一枚も作らない。文字と顔は常に左右の別領域として分離する。隣接する区間では、人物の相対位置、カメラ高、カメラの向き、人物サイズ、背景の消失軸のうち最低二つを更新し、同じ正面の胸上構図を反復しない。明度と色温度は白と淡いシアンから終盤の淡い琥珀色へ一方向に連続変化させ、白飛びへ跳ばさない。人物を長時間文字や花で隠さず、顔を見せる区間では顔の可読性を保つ。ロゴ、透かし、フレーム外の帯を描かない。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 15,
      "views": 3385,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption embedded public prompt: explicit hand-drawn girl, botanical glasshouse, flower palette, vines, staff lines, lyric typography, and one-way lighting transition"
  },
  {
    "id": "x-2087370863788392840",
    "title": "小天使仰望夜空中的发光“5800”",
    "titleEn": "A Little Angel Looks Up at a Glowing 5800 in the Night Sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名背对镜头的小天使站在夜色中的屋顶边缘，仰望天空里巨大的彩色发光数字“5800”。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A small winged angel stands with her back to the camera on a rooftop edge at night, looking up at the huge multicolored glowing number 5800 in the sky. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2087370863788392840",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-08-12T02:49:30.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087370863788392840.mp4",
    "posterUrl": "/posters/x/2087370863788392840.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "フォロワーさんが5800人になりましたー！！\n皆様ありがとうございます✨\n\nこれからも楽しい動画を作りたいと思いますので、引き続きよろしくお願いします😊\n\n動画はSousakuAI のMiniMax H3で作成\n#SousakuAI @sousakuai \n\nこれ系の動画は今までSeedance 2.0で作成していたけど、\nH3でもうまく出来ますね！！\nコスパ良くて嬉しい☺️",
    "engagement": {
      "replies": 10,
      "reposts": 4,
      "likes": 55,
      "views": 1413,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit back-facing winged character, rooftop edge, night sky, and luminous 5800; sourceCaption: explicit 5,800-follower milestone"
  },
  {
    "id": "x-2087367872331059531",
    "title": "生成速度、显存、人体结构与画质的双模型对比",
    "titleEn": "Two-Model Comparison of Speed, Memory, Anatomy, and Image Quality",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "双模型比较记录一方生成更快但显存占用较高、画质观感较好，另一方的人体结构表现更优。 成片规格为 9 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The two-model comparison finds one system faster but more memory-intensive with preferable image quality, while the other is judged stronger at human anatomy. Output: 9s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AiCreatorS1881/status/2087367872331059531",
    "sourceLabel": "X 原帖 · @AiCreatorS1881",
    "author": "こたじろう",
    "publishedAt": "2026-08-12T02:37:36.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087367872331059531.mp4",
    "posterUrl": "/posters/x/2087367872331059531.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX-2.5お試し、生成が速いのはとても良い。\nただ、メモリ食いすぎるので、この点は今後に期待。\n人体の構造については、MiniMax H3の方が優れる。\n画像の質は、LTXの方が好印象。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 43,
      "views": 3256,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit two-model evaluation across speed, memory use, human anatomy, and image quality"
  },
  {
    "id": "x-2087366995113312351",
    "title": "八张角色参考图与英文说唱驱动的连贯短片",
    "titleEn": "A Coherent Short Driven by Eight Character References and English Rap",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "八张视觉一致的角色参考图与一条原创英文说唱音轨被组合成连贯短片，重点维持角色身份、造型、表演、运镜、场景切换与节拍推进的一致性。 成片规格为 12 秒 · 928×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Eight visually consistent character references and an original English rap track are combined into one coherent short, emphasizing continuity in identity, styling, performance, camera language, scene changes, and beat-driven pacing. Output: 12s · 928×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bgvc123/status/2087366995113312351",
    "sourceLabel": "X 原帖 · @bgvc123",
    "author": "BG-VC",
    "publishedAt": "2026-08-12T02:34:07.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087366995113312351.mp4",
    "posterUrl": "/posters/x/2087366995113312351.jpg",
    "duration": 12,
    "aspectRatio": "square",
    "resolution": "928×928",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The entire pipeline runs locally on a GPU using the open-source MiniMax H3 model and a ComfyUI workflow.\n\nThrough H3’s multimodal reference mode, we combined eight visually consistent reference images with an original English rap track—bringing character identity, styling, performance, camera language, scene changes, and beat-driven pacing into one coherent video.\n\nThe real breakthrough in AI filmmaking will not come from visual quality alone.\n\nIt will happen when multimodal generation becomes highly controllable—and when the same AI character can preserve a recognizable identity across images, shots, scenes, and future episodes.\n\nSolve control and character continuity, and AI-native short dramas and films will change fundamentally.\n\n#AxonImagine #MiniMaxH3 #ComfyUI #AIVideo #AIFilmmaking",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 127,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit eight reference images, original English rap track, character consistency, performance, camera language, scene changes, and beat-driven pacing"
  },
  {
    "id": "x-2087366506556637369",
    "title": "双模型效果与单次成本对比",
    "titleEn": "Two-Model Output and Per-Run Cost Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖比较两个模型的生成效果，认为差距不大，并记录其中一方单次成本为 0.0148 美元。 成片规格为 20 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source compares the outputs of two models, finds the gap relatively small, and records a per-run cost of $0.0148 for one of them. Output: 20s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yuea85035/status/2087366506556637369",
    "sourceLabel": "X 原帖 · @yuea85035",
    "author": "Riley",
    "publishedAt": "2026-08-12T02:32:11.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087366506556637369.mp4",
    "posterUrl": "/posters/x/2087366506556637369.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The comparison video between Seedance 2.5 and MiniMax-H3 shows that the difference isn’t that big.\n\nMiniMax-H3 only costs $0.0148, which is incredibly affordable.\n\nGive it a try yourself:\n https://www.xskill.ai/#/v2/models/st-ai%2Fsuper-seed2-lite?ref=Q1DTE94G",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 71,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit two-model comparison, reported small quality gap, and stated per-run cost"
  },
  {
    "id": "x-2087366255577592266",
    "title": "阳光花园里举起白色绒球花的女子",
    "titleEn": "A Woman Holds Up a White Puffball Flower in a Sunlit Garden",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名束起棕发的女子在阳光花园中闭着眼，把一朵白色绒球状小花举到面前，周围飘着淡色花瓣。 成片规格为 10 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A brown-haired woman with her hair tied back closes her eyes in a sunlit garden, holding a small white puffball flower near her face as pale petals drift around her. Output: 10s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/thedoomguy_ai/status/2087366255577592266",
    "sourceLabel": "X 原帖 · @thedoomguy_ai",
    "author": "The DOOM Guy",
    "publishedAt": "2026-08-12T02:31:11.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087366255577592266.mp4",
    "posterUrl": "/posters/x/2087366255577592266.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.5 e Minimax-H3 acabaram de sair. Um clique e você gera vídeo estável de 30 segundos. Times de marketing já cortaram custo de produção em ~10x. Anúncios, mini-dramas, personagens de IP — tudo saindo muito mais rápido.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 63,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "poster-media-review: explicit tied-back brown-haired woman, closed eyes, white puffball flower, sunlit garden, and drifting pale petals"
  },
  {
    "id": "x-2087366157489803693",
    "title": "三模型的情绪细腻度与表演对比",
    "titleEn": "Three-Model Comparison of Emotional Nuance and Acting",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三个模型在相同情绪与表演任务下接受比较，评价维度是情绪细腻度与整体表演效果。 成片规格为 12 秒 · 1921×3240。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three models are compared on the same emotion-and-acting task, with emotional nuance and overall performance as the stated evaluation dimensions. Output: 12s · 1921×3240. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wavespeed_ai/status/2087366157489803693",
    "sourceLabel": "X 原帖 · @wavespeed_ai",
    "author": "WaveSpeedAI",
    "publishedAt": "2026-08-12T02:30:47.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087366157489803693.mp4",
    "posterUrl": "/posters/x/2087366157489803693.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "1921×3240",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.5. Wan 3.0. MiniMax H3.\nThree of the hottest video models right now. \nSame test: emotion + acting.\n\nWho delivers the most nuanced emotions?\nWho gives the best performance?\nYou be the judge.",
    "engagement": {
      "replies": 82,
      "reposts": 51,
      "likes": 727,
      "views": 71491,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit three-model same-task comparison and stated emotional-nuance and acting criteria",
    "featured": {
      "at": "2026-09-05T15:50:15.690Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2087365957144502585",
    "title": "吃拉面的动作生成",
    "titleEn": "Generating the Act of Eating Ramen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者用吃拉面这一动作观察新模型表现，并对比两年前人工智能几乎无法自然完成该动作的情况。 成片规格为 7 秒 · 576×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author uses the act of eating ramen to assess a new model and contrasts it with the difficulty AI systems had performing the action naturally about two years earlier. Output: 7s · 576×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2087365957144502585",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-08-12T02:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087365957144502585.mp4",
    "posterUrl": "/posters/x/2087365957144502585.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "576×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 でラーメン\n\n新しいモデルが出たらいつも試すシリーズその１です\n\n2年ほど前は、ラーメンをまともに食べられるAIは\n\nほぼなかった気がしますが、\n\nもう昔のことってかんじですね\n#AIart",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 17,
      "views": 138,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit ramen-eating subject, recurring new-model evaluation series, and historical capability comparison"
  },
  {
    "id": "x-2087363943824044044",
    "title": "杀害百人之后，真主还会宽恕他吗？",
    "titleEn": "He Killed 100 People—Could Allah Still Forgive Him?",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作品以“杀害百人后是否仍能得到真主宽恕”为题，采用一次生成，并用十五秒音频参考维持声音一致性、单张图像参考维持风格一致性。 成片规格为 113 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The work asks whether a man who killed one hundred people could still be forgiven by Allah, using a zero-shot run, a fifteen-second audio reference for voice consistency, and one image reference for style consistency. Output: 113s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FurkanGozukara/status/2087363943824044044",
    "sourceLabel": "X 原帖 · @FurkanGozukara",
    "author": "Furkan Gözükara",
    "publishedAt": "2026-08-12T02:22:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087363943824044044.mp4",
    "posterUrl": "/posters/x/2087363943824044044.jpg",
    "duration": 113,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "He Killed 100 People… Could Allah Still Forgive Him\n\n0-shot made locally with MiniMax H3.\n\nEntire prompt folder written by ChatGPT.\n\n1 audio reference (15 seconds) used for voice consistency.\n\n1 image reference used for style consistency.",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 3,
      "views": 6251,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit work premise, zero-shot method, fifteen-second audio reference for voice consistency, and single-image reference for style consistency"
  },
  {
    "id": "x-2087363846336135484",
    "title": "由分镜首帧与细分片段拼接的二十一秒短篇",
    "titleEn": "A Twenty-One-Second Short Assembled from Storyboard Keyframes and Small Segments",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "Codex 先制作分镜及各场景首图，再将多个图像驱动的小片段拼接成二十一秒内容，配乐则由 Gemini 生成。 成片规格为 21 秒 · 1152×1664。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Codex first creates the storyboard and opening image for each scene, then multiple image-driven segments are joined into a twenty-one-second piece, with music generated by Gemini. Output: 21s · 1152×1664. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2087363846336135484",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-08-12T02:21:36.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087363846336135484.mp4",
    "posterUrl": "/posters/x/2087363846336135484.jpg",
    "duration": 21,
    "aspectRatio": "portrait",
    "resolution": "1152×1664",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で21秒動画。Codexにストーリーボード（各シーンの先頭画像生成含む）作成を依頼した後、細かい動画をi2vで作成。最後に結合。曲はGeminiに依頼して生成。リップシンク気にしなければこのフローが楽でいいね！ローカルLLM＋Anima/Krea＋AceStep1.5を使えば全ローカル実行もできそう！",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 226,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit duration, Codex-authored storyboard and opening images, segmented image-driven workflow, final assembly, and Gemini-created music"
  },
  {
    "id": "x-2087363095882789299",
    "title": "用前段尾帧续接六个五秒片段",
    "titleEn": "Six Five-Second Segments Chained through Their Final Frames",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "六个约五秒、512×288 的片段依次生成，每段尾帧作为下一段参考图，最后拼接成约三十一秒内容。 成片规格为 31 秒 · 512×288。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Six roughly five-second segments at 512×288 are generated in sequence, with each final frame serving as the next reference image before the segments are joined into roughly thirty-one seconds. Output: 31s · 512×288. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Darkside_of_NHK/status/2087363095882789299",
    "sourceLabel": "X 原帖 · @Darkside_of_NHK",
    "author": "暗部ちゃん",
    "publishedAt": "2026-08-12T02:18:38.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087363095882789299.mp4",
    "posterUrl": "/posters/x/2087363095882789299.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "512×288",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【Minimax H3 Macbookで生成】\n\n私のM4 Max 36GBのMacbookにMinimax H3を入れて一晩生成させた結果がこちら\n\n512x288・約5秒のクリップを6つ作って、各々の最終フレームを参照画像として生成した上で繋げたもの\n\n実用にはならないけど、ここまで来たか！という驚きはあるね\n\n※これはAI生成動画です",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 8,
      "views": 2294,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit six-segment count, approximate per-segment duration, resolution, final-frame chaining, and assembly method; media metadata confirms a 31-second file"
  },
  {
    "id": "x-2087362182434496769",
    "title": "用轻量编码器在完整运行前快速预览",
    "titleEn": "Rapid Preview with a Lightweight Autoencoder before the Full Run",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "ComfyUI 工作流结合加速权重与轻量自动编码器，可在投入完整 GPU 运行前数秒内生成预览。 成片规格为 42 秒 · 3812×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A ComfyUI workflow combines an acceleration weight with a lightweight autoencoder to produce a preview within seconds before committing GPU time to the full run. Output: 42s · 3812×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RoundtableSpace/status/2087362182434496769",
    "sourceLabel": "X 原帖 · @RoundtableSpace",
    "author": "0xMarioNawfal",
    "publishedAt": "2026-08-12T02:15:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087362182434496769.mp4",
    "posterUrl": "/posters/x/2087362182434496769.jpg",
    "duration": 42,
    "aspectRatio": "landscape",
    "resolution": "3812×2160",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A ComfyUI workflow using H3 Turbo LoRA and Tiny AutoEncoder H3 lets you preview MiniMax H3 generations in seconds before committing GPU time to the full run.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 36,
      "views": 44481,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit ComfyUI workflow, acceleration weight, lightweight autoencoder, seconds-scale preview, and pre-full-run purpose"
  },
  {
    "id": "x-2087361747162403036",
    "title": "放学后的空教室与喜欢的老师",
    "titleEn": "After School in an Empty Classroom with a Favorite Teacher",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "故事设在放学后空无一人的教室，人物与自己喜欢的老师单独相处；十秒运行耗时十分钟，且面部相较原图有所变化。 成片规格为 10 秒 · 640×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The story is set in an empty classroom after school, where a person is alone with a favorite teacher; the ten-second run takes ten minutes and shows some facial drift from the source image. Output: 10s · 640×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pin_knuts/status/2087361747162403036",
    "sourceLabel": "X 原帖 · @pin_knuts",
    "author": "ぴんくなっつ",
    "publishedAt": "2026-08-12T02:13:16.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087361747162403036.mp4",
    "posterUrl": "/posters/x/2087361747162403036.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "大好きな先生と誰もいない放課後の空き教室で…\n\n何するのかなぁ～～～～？🤪\n\nMinimax H3で8stepのまま10秒動画にしてみたんだけど生成に10分かかるね。\n2枚目の元画像と比べてちょっと顔も変わっちゃう。\n自分の今の環境と設定だと1秒の生成に約1分って感じかな。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 35,
      "views": 7974,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit favorite-teacher subject, empty after-school classroom setting, ten-second duration, runtime, and face-drift observation"
  },
  {
    "id": "x-2087358568769187906",
    "title": "五段拼接的二十秒运动控制实验",
    "titleEn": "Twenty-Second Motion-Control Experiment across Five Segments",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "运动控制实验将 102 帧拆为五段并形成二十秒内容；作者认为视觉结果完整，但声音仍需修正。 成片规格为 20 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A motion-control experiment divides 102 frames across five segments to form a twenty-second piece; the author considers the visuals complete but says the sound still needs correction. Output: 20s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2087358568769187906",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-08-12T02:00:38.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087358568769187906.mp4",
    "posterUrl": "/posters/x/2087358568769187906.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI ver 1.4\nMotion Control Test // 102f x 5 segments（20秒動画）\n\nビジュアル面は完璧に出来た\nサウンドは治せるかな〜これ……",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 676,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit motion-control experiment, 102 frames, five segments, twenty-second duration, visual assessment, and unresolved sound issue"
  },
  {
    "id": "x-2087358389231976614",
    "title": "八百步人脸 LoRA 的相似度评估",
    "titleEn": "Similarity Evaluation of an 800-Step Face LoRA",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "八百步人脸 LoRA 用于相似度评估，作者明确认为当前结果仍未达到相似。 成片规格为 5 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An 800-step face LoRA is evaluated for resemblance, with the author explicitly stating that the current result is not yet similar. Output: 5s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/willowion/status/2087358389231976614",
    "sourceLabel": "X 原帖 · @willowion",
    "author": "𝖂𝖎𝖑𝖑𝖔𝖂",
    "publishedAt": "2026-08-12T01:59:55.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087358389231976614.mp4",
    "posterUrl": "/posters/x/2087358389231976614.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "800ステップの顔LoRAテスト🤡 (当然まだ似てない)\n\nコーヒー最高といいながら、色からするとアイスティー❓🙄\n\nMiniMax H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 165,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit 800-step face-LoRA evaluation and stated lack of resemblance"
  },
  {
    "id": "x-2087352033770914093",
    "title": "Apple Silicon 上 MLX 与 Metal 的速度优化",
    "titleEn": "MLX and Metal Speed Optimization on Apple Silicon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者连续五天推进 Apple Silicon 上的 MLX 与 Metal 适配，当前结果较清晰但速度仍慢，尚需多项验证。 成片规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author has spent five days adapting the workflow to MLX and Metal on Apple Silicon; current results are fairly clean but remain slow and require further validation. Output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C_of_Creativity/status/2087352033770914093",
    "sourceLabel": "X 原帖 · @C_of_Creativity",
    "author": "PIYO",
    "publishedAt": "2026-08-12T01:34:40.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087352033770914093.mp4",
    "posterUrl": "/posters/x/2087352033770914093.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 のApple Silicon MLX+Metal化をこの5日間ずっと頑張っていますが、まだいくつも検証が必要です。\n\nまだ他の方達の開発実績よりも生成時間がかかっているので、世界で一番の高速生成が可能になったら公開しますがまだ何とも言えないです。\nいくつか詰まっている箇所があります。\n\n現状の生成結果ですが時間がかかるが結構綺麗です。\n\n#Mac\n#mlx\n#AppleSilicon",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 24,
      "views": 1857,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit five-day Apple Silicon MLX-and-Metal effort, speed objective, remaining validation, current runtime limitation, and quality observation"
  },
  {
    "id": "x-2087351012587889121",
    "title": "单张图片与独立人声驱动的三模型口型同步对比",
    "titleEn": "Three-Model Lip-Sync Comparison from One Image and an Isolated Vocal",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三个模型以同一张图片和从歌曲中分离的人声进行八秒口型同步对比，并统一为约 480p、3:4；作者同时记录了动作质感、嘴部稳定性、节拍延迟与各自耗时。 成片规格为 8 秒 · 1440×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three models use the same image and an isolated song vocal for an eight-second lip-sync comparison at roughly 480p and 3:4, with the author documenting motion quality, mouth stability, beat delay, and runtime for each. Output: 8s · 1440×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yu_ichi_suzuki/status/2087351012587889121",
    "sourceLabel": "X 原帖 · @yu_ichi_suzuki",
    "author": "鈴木憂一 | Highdrama",
    "publishedAt": "2026-08-12T01:30:37.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087351012587889121.mp4",
    "posterUrl": "/posters/x/2087351012587889121.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1440×640",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "出たばかりのローカルLTX-2.5を使って1枚の絵とボーカルを渡して、歌わせたときのリップシンクを3つのモデルで比べました（2026年8月時点）。\n\n左から LTX-2.5、MiniMax H3、Seedance 2.5。左2つはローカル（RTX 5090）、Seedance 2.5だけHiggsfield経由のクラウドです。480p級・3:4・8秒で揃えています。\n\n渡した音声は、demucsでボーカルだけ抜き出して、そこから少し音量を下げたものです。動画に乗っているのは、そのあと同じ箇所のビート付き（伴奏込みの原曲）に入れ替えたものになります。\n\nLTX-2.5用に音声を渡す公式ワークフローが無さそうなので、2.3のia2vの構造を参考にClaude Codeに組んでもらったもののため、これで比較するのはフェアではないのは確かです。\n\nLTX-2.5はノリや感情はいい感じなんですけれども、動くとだいぶ2.5Dっぽくなるな、という印象です。あと口元がリアル寄りで、途中で口が白くなってしまうという昔のKlingとかにあったのに似た現象が起きます。\nH3がアニメ向きで、リップシンクも合ってそうな感じ。ただ、オーディオの大きさなどで口の開きが結構大きくなったりしたので、何回かやり直しています。\nSeedanceもいいけど、少しテンポが音に遅れている感じ？\n\n生成時間は LTX-2.5が69秒、H3が93秒、Seedance 2.5がキュー込みで約3分でした。",
    "engagement": {
      "replies": 2,
      "reposts": 10,
      "likes": 40,
      "views": 3334,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit three-model comparison, single-image and isolated-vocal inputs, duration, resolution class, aspect ratio, lip-sync focus, observed differences, and runtimes"
  },
  {
    "id": "x-2087350859311243600",
    "title": "神狐大人的日常《会走路的避雨处》",
    "titleEn": "Divine Fox in The Walking Rain Shelter",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "日常短篇《会走路的避雨处》以神狐大人为主角。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The everyday short The Walking Rain Shelter centers on a divine fox. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2087350859311243600",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-08-12T01:30:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087350859311243600.mp4",
    "posterUrl": "/posters/x/2087350859311243600.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3動画\n日常エピソード \n『歩く雨宿り』\n神狐様",
    "engagement": {
      "replies": 0,
      "reposts": 6,
      "likes": 30,
      "views": 800,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit episode title, everyday-episode format, and divine-fox subject"
  },
  {
    "id": "x-2087346979814756609",
    "title": "又一场水果大战",
    "titleEn": "Another Fruit War",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原帖明确将随帖作品描述为另一场带有喜剧效果的水果大战。 成片规格为 15 秒 · 2544×1456。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly describes the attached work as another fruit war with a comic effect. Output: 15s · 2544×1456. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/minaxlab/status/2087346979814756609",
    "sourceLabel": "X 原帖 · @minaxlab",
    "author": "Mina Labs",
    "publishedAt": "2026-08-12T01:14:35.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087346979814756609.mp4",
    "posterUrl": "/posters/x/2087346979814756609.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2544×1456",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Another fruits war \nThis time created with Minimax H3\nWhat a funny video 🤣\n\nhttps://www.minaxlab.com\n\n#ai #minimax #video",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 0,
      "views": 112,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit fruit-war premise and stated comic tone"
  },
  {
    "id": "x-2087346287804862863",
    "title": "24GB M5 MacBook 的全质量原生运行",
    "titleEn": "Full-Quality Native Run on a 24GB M5 MacBook",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "完整质量工作流在配备 24GB 内存的 M5 MacBook 上原生运行，不降低质量且无需云端。 成片规格为 3 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A full-quality workflow runs natively on an M5 MacBook with 24GB of memory, without a quality reduction or cloud processing. Output: 3s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Brjen/status/2087346287804862863",
    "sourceLabel": "X 原帖 · @Brjen",
    "author": "Brian",
    "publishedAt": "2026-08-12T01:11:50.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087346287804862863.mp4",
    "posterUrl": "/posters/x/2087346287804862863.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Full-quality MiniMax H3 now running natively on a 24GB M5 MacBook — no quality cut, no cloud.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 132,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit full-quality condition, native execution, 24GB M5 MacBook hardware, no quality cut, and no-cloud constraint"
  },
  {
    "id": "x-2087346075761848683",
    "title": "16GB AMD 7800 XT 的全质量原生运行",
    "titleEn": "Full-Quality Native Run on a 16GB AMD 7800 XT",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "完整质量工作流在 16GB AMD 7800 XT 上原生运行，不降低质量且无需云端。 成片规格为 5 秒 · 704×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A full-quality workflow runs natively on a 16GB AMD 7800 XT, without a quality reduction or cloud processing. Output: 5s · 704×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Brjen/status/2087346075761848683",
    "sourceLabel": "X 原帖 · @Brjen",
    "author": "Brian",
    "publishedAt": "2026-08-12T01:11:00.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087346075761848683.mp4",
    "posterUrl": "/posters/x/2087346075761848683.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "704×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Full-quality MiniMax H3 now running natively on a 16GB AMD 7800xt — no quality cut, no cloud.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 124,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit full-quality condition, native execution, 16GB AMD 7800 XT hardware, no quality cut, and no-cloud constraint"
  },
  {
    "id": "x-2087343669158281351",
    "title": "粉色蝴蝶结小猫与冒烟的废弃车辆",
    "titleEn": "A Bow-Wearing Kitten Among Smoking Wrecks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一只戴粉色蝴蝶结的灰色小猫先抱着一根大型管状物，随后站在烟雾与废弃车辆之间。 成片规格为 20 秒 · 1088×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A gray kitten wearing a pink bow first holds a large tube and later stands beside it among smoke and damaged vehicles. Output: 20s · 1088×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/victorpfreitas/status/2087343669158281351",
    "sourceLabel": "X 原帖 · @victorpfreitas",
    "author": "Victor.",
    "publishedAt": "2026-08-12T01:01:26.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087343669158281351.mp4",
    "posterUrl": "/posters/x/2087343669158281351.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "1088×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "way better than minimax h3 👀\nand free until 17 aug\n\nps. it took 790s to generate",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 0,
      "views": 165,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2087342297860235304",
    "title": "32GB 显存与 64GB 内存下的 4K 世界模型运行",
    "titleEn": "A 4K World-Model Run with 32GB VRAM and 64GB RAM",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "世界模型在 32GB 显存与 64GB 内存条件下生成 4K 高清内容，原帖同时说明其权重完全开放并可进行微调。 成片规格为 5 秒 · 1280×726。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A world model produces 4K high-definition output with 32GB of VRAM and 64GB of system memory, while the source also states that its weights are fully open for fine-tuning. Output: 5s · 1280×726. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MagicPower21M/status/2087342297860235304",
    "sourceLabel": "X 原帖 · @MagicPower21M",
    "author": "MagicPower ⚡",
    "publishedAt": "2026-08-12T00:55:59.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087342297860235304.mp4",
    "posterUrl": "/posters/x/2087342297860235304.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×726",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5世界模型的效果\n仅需32G显存+64G内存即可流畅生成4K高清视频\n\nLTX 2.5全开放权重，可进行各种微调！\n\nMiniMax H3别藏着掖着了",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 300,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit world-model output, 4K target, VRAM and RAM requirements, open weights, and fine-tuning availability"
  },
  {
    "id": "x-2087337242125844960",
    "title": "歌曲《千傘の書架》的二十五场景编排",
    "titleEn": "Twenty-Five Scenes for a Single Song",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "歌曲《千傘の書架》被编排为二十五个场景，以 1216×672 输出，总耗时 164 分 26 秒。 成片规格为 5 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A single song is arranged across twenty-five scenes at 1216×672, with a total runtime of 164 minutes and 26 seconds. Output: 5s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pompompokopomch/status/2087337242125844960",
    "sourceLabel": "X 原帖 · @pompompokopomch",
    "author": "ぽむ",
    "publishedAt": "2026-08-12T00:35:54.000Z",
    "addedAt": "2026-08-13T00:38:33.433Z",
    "mediaUrl": "/media/x-2087337242125844960.mp4",
    "posterUrl": "/posters/x/2087337242125844960.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3ネタ。\n千傘の書架という曲の動画を昨日つくったツールで、解像度0.8（1216x672）で生成。\n25シーンで164分26秒。3時間かからなかったですが、1曲に使う時間がふえるなぁ。イラストをつなぎ合わせるだけよりすごくよくなったので、あとは解像度…PC性能…",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 99,
      "snapshotAt": "2026-08-13T00:38:33.433Z"
    },
    "approvedAt": "2026-08-13T00:38:33.433Z",
    "editorialBasis": "sourceCaption: explicit song title, twenty-five scene count, output resolution, and measured total runtime"
  },
  {
    "id": "x-2088845392897319006",
    "title": "一百万像素直出放大与歌曲口型同步评估",
    "titleEn": "One-Megapixel Upscaling and Song Lip-Sync Evaluation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "内容以一百万像素直接输出后仅做放大、未另行剪辑；作者记录了手指融化问题，同时认为读取歌曲后的口型同步较好。 成片规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The output is rendered at one megapixel and only upscaled, with no additional editing; the author notes melting fingers while judging the song-driven lip synchronization to be good. Output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/otosa_vt/status/2088845392897319006",
    "sourceLabel": "X 原帖 · @otosa_vt",
    "author": "静見おと : otosa",
    "publishedAt": "2026-08-16T04:28:45.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088845392897319006.mp4",
    "posterUrl": "/posters/x/2088845392897319006.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こちらはMiniMax H3ローカル版です✨\n1MPで出力して、アップスケールのみ、ポン出し無編集です。\n指などは溶けちゃったりしますが、曲を読みこんでもらってリップシンク良好のようです🥰",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 51,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "sourceCaption: explicit one-megapixel output, upscale-only and unedited conditions, finger artifact, song input, and lip-sync assessment"
  },
  {
    "id": "x-2088844629093601702",
    "title": "睡过头后的匆忙晨间流程",
    "titleEn": "A Rushed Morning After Oversleeping",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "一名女子从床上醒来后匆忙刷牙并整理出门，画面以清晨卧室、浴室近景和安静门厅串起这段流程。 成片规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman wakes in bed, hurriedly brushes her teeth, and prepares to leave, moving from a morning bedroom to a bathroom close-up and a quiet entryway. Output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic realistic morning scene of a young adult woman suddenly waking up in a hurry. She quickly checks the time on her phone and realizes she is late. Looking stressed but determined, she immediately gets out of bed, quickly fixes the bedsheets, rushes to the bathroom and brushes her teeth. She then hurries into the kitchen, prepares a quick breakfast, packs the food into her bag, grabs her belongings, puts on her shoes, and rushes out of the apartment. Fast-paced natural movements, realistic expressions, smooth cinematic camera transitions, morning sunlight, detailed modern apartment, photorealistic, dynamic pacing, cinematic commercial quality, 4K.",
    "sourceUrl": "https://x.com/noorlewisx/status/2088844629093601702",
    "sourceLabel": "X 原帖 · @noorlewisx",
    "author": "Noor",
    "publishedAt": "2026-08-16T04:25:43.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088844629093601702.mp4",
    "posterUrl": "/posters/x/2088844629093601702.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Late Morning Rush: A Cinematic Story of a Busy Young Woman Racing Against Time to Get Ready and Start Her Day\n\nCreated with MiniMax H3\n\nPrompt:\n\nA cinematic realistic morning scene of a young adult woman suddenly waking up in a hurry. She quickly checks the time on her phone and realizes she is late. Looking stressed but determined, she immediately gets out of bed, quickly fixes the bedsheets, rushes to the bathroom and brushes her teeth. She then hurries into the kitchen, prepares a quick breakfast, packs the food into her bag, grabs her belongings, puts on her shoes, and rushes out of the apartment. Fast-paced natural movements, realistic expressions, smooth cinematic camera transitions, morning sunlight, detailed modern apartment, photorealistic, dynamic pacing, cinematic commercial quality, 4K.",
    "engagement": {
      "replies": 7,
      "reposts": 1,
      "likes": 9,
      "views": 78,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "prompt+three-frame-review"
  },
  {
    "id": "x-2088841622155055364",
    "title": "无音频参考的角色设定图短篇实验",
    "titleEn": "Character-Sheet Short without an Audio Reference",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "实验只使用文字指令与角色设定图，不提供音频参考；作者同时讨论了为短片切分音乐并匹配演出的难度。 成片规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The experiment uses only a text prompt and character sheet without an audio reference, while the author discusses the difficulty of cutting music for a short and matching the staging to it. Output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Muimi_Karma/status/2088841622155055364",
    "sourceLabel": "X 原帖 · @Muimi_Karma",
    "author": "無意味カルマ",
    "publishedAt": "2026-08-16T04:13:46.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088841622155055364.mp4",
    "posterUrl": "/posters/x/2088841622155055364.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "音楽メディア使わずプロンプトだけ、キャラシートのみのr2v。\nショート動画だと音楽ぶつ切りして当てはめる、演出合わせる、結構大変ですよね。\nガチャではあるけど MiniMax H3 かなり最適解。\nMiniMax H3 Made with #PolloAI 🎬\n@PolloAIJP",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 56,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "sourceCaption: explicit no-audio-reference condition, prompt-only plus character-sheet inputs, short-form workflow, and music-cutting and staging challenge"
  },
  {
    "id": "x-2088840934494720090",
    "title": "八镜头一键工作流的后段精度衰减",
    "titleEn": "Late-Sequence Accuracy Decline in an Eight-Shot One-Click Workflow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "零制作费的一键工作流把内容拆成八个镜头，作者指出越到后段精度越低，并出现两人像在说同一句话的问题。 成片规格为 26 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A zero-production-cost, one-click workflow divides the piece into eight shots; the author reports declining accuracy in later shots and a problem where two people appear to speak the same line. Output: 26s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/www_taishi/status/2088840934494720090",
    "sourceLabel": "X 原帖 · @www_taishi",
    "author": "Taishi｜AI×マーケティング",
    "publishedAt": "2026-08-16T04:11:02.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088840934494720090.mp4",
    "posterUrl": "/posters/x/2088840934494720090.jpg",
    "duration": 26,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GPT × Minimax H3 × HyperFrame\n制作費ゼロ円　ワンクリック生成\n\nまだまだワークフローの見直しは必要だけど8カットに分けて生成\n\n後半カットにつれて精度が落ちているのがわかる\n（2人が同じこと喋ってるみたいに）\n\nReal valueみたいな動画とか作ってみたいな\n\n☑️作り方知りたい人\n☑️最新の情報知りたい人\n☑️個別に話したい人\n\n匿名オープンチャットで話せるよ👇\nhttps://line.me/ti/g2/K4yk88xcsX5-DcfkxKzxrhUIt8or1RXGgIgUVQ?utm_source=invitation&utm_medium=link_copy&utm_campaign=default",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 39,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "sourceCaption: explicit zero-cost one-click workflow, eight-shot structure, later-shot accuracy decline, and duplicated-speaking behavior"
  },
  {
    "id": "x-2088837429046477102",
    "title": "农贸市场葡萄摊前的红黑格衫女子",
    "titleEn": "A Woman in Red-and-Black Plaid at a Grape Stall in a Farmers' Market",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名金发女子穿着红黑格子衬衫、背着帆布包，站在露天农贸市场的葡萄与水果摊前。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde woman in a red-and-black plaid shirt with a canvas shoulder bag stands beside grapes and other produce at an outdoor farmers' market. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2088837429046477102",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-16T03:57:06.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088837429046477102.mp4",
    "posterUrl": "/posters/x/2088837429046477102.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @Hailuo_AI \n\nPrompt 👇🏻",
    "engagement": {
      "replies": 8,
      "reposts": 0,
      "likes": 16,
      "views": 279,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "poster-media-review: explicit blonde woman, red-and-black plaid shirt, canvas shoulder bag, grapes, produce tables, and outdoor market stalls"
  },
  {
    "id": "x-2088832804255891659",
    "title": "两张参考图的再次尝试",
    "titleEn": "A Second Attempt with Two Reference Images",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "原帖明确记录了一次使用两张参考图重新尝试的实验。 成片规格为 10 秒 · 640×1120。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source explicitly records a second attempt using two reference images. Output: 10s · 640×1120. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/balaena01/status/2088832804255891659",
    "sourceLabel": "X 原帖 · @balaena01",
    "author": "Kujirachan/くじら",
    "publishedAt": "2026-08-16T03:38:43.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088832804255891659.mp4",
    "posterUrl": "/posters/x/2088832804255891659.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "640×1120",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3\nリファレンス2枚でリベンジ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 54,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "sourceCaption: explicit two-reference-image setup and retry framing"
  },
  {
    "id": "x-2088829677335195859",
    "title": "假期最后一天，喝下这一杯打起精神",
    "titleEn": "A Pick-Me-Up Drink on the Final Day of the Holiday",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暑假最后一天，原帖以“喝下这一杯打起精神”作为随帖作品的明确动作与情境。 成片规格为 5 秒 · 710×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "On the final day of the summer holiday, the source frames the attached work around having a drink to regain some energy. Output: 5s · 710×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Koyomi_666XxX/status/2088829677335195859",
    "sourceLabel": "X 原帖 · @Koyomi_666XxX",
    "author": "KOYOMI︎ ︎💫⭐︎",
    "publishedAt": "2026-08-16T03:26:18.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088829677335195859.mp4",
    "posterUrl": "/posters/x/2088829677335195859.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "710×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "盆休み最終日……_(:3」∠)_\nこれ飲んで元気だしてね\nMiniMax H3で遊び倒した盆休み",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 61,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "sourceCaption: explicit final-holiday-day setting and drink-for-energy action"
  },
  {
    "id": "x-2088829079080612190",
    "title": "雨夜和风摇滚歌者的三分钟音乐短片",
    "titleEn": "A Three-Minute Music Short with a Japanese-Rock Vocalist on a Rainy Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "三分钟音乐短片围绕新的和风摇滚歌者展开，以 Music 3.0 日语人声作为音频参考，并营造雨夜氛围。 成片规格为 120 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A three-minute music short centers on a new Japanese-rock vocalist, using a Music 3.0 Japanese vocal as its audio reference and a rainy-night atmosphere. Output: 120s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hAru_mAki_ch/status/2088829079080612190",
    "sourceLabel": "X 原帖 · @hAru_mAki_ch",
    "author": "Maki@Sunwood AI Labs.",
    "publishedAt": "2026-08-16T03:23:55.000Z",
    "addedAt": "2026-08-16T04:44:06.092Z",
    "mediaUrl": "/media/x-2088829079080612190.mp4",
    "posterUrl": "/posters/x/2088829079080612190.jpg",
    "duration": 120,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "新しい和風ロックボーカルで、3分MVをもう一周やってみたぞ！！！\n\n今回は、Music 3.0の日本語ボーカルをMiniMax H3 Ref2VAへ音声参照として入れてみた。\n音声を主役にするだけでも、意外と歌っている感じまでいける、、、これは結構いい🔥\n\n雨の夜のMVっぽい空気と、新しいボーカルキャラでまとめた。\n\n✅ Music 3.0の日本語ボーカル\n✅ MiniMax H3 Ref2VA\n✅ 新しい和風ロックボーカル＋MV空間",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 439,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.092Z",
    "editorialBasis": "sourceCaption: explicit three-minute music-work format, new Japanese-rock vocalist, Music 3.0 Japanese vocal reference, and rainy-night setting"
  },
  {
    "id": "x-2088825841694466443",
    "title": "粗略短篇指令的快速尝试",
    "titleEn": "A Quick Run with a Rough Short-Form Prompt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者直接投入一条粗略的短篇指令进行快速尝试，并判断若逐场景提供素材即可进一步完善。 成片规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author quickly tries a rough short-form prompt and concludes that supplying assets scene by scene could improve the result further. Output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/takoyama9zo/status/2088825841694466443",
    "sourceLabel": "X 原帖 · @takoyama9zo",
    "author": "たこやま / Takoyama @AIart",
    "publishedAt": "2026-08-16T03:11:03.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088825841694466443.mp4",
    "posterUrl": "/posters/x/2088825841694466443.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@PolloAIJP にMinimax h3が追加されてたからショート動画プロンプトを雑に投げてみたけどいい感じ\nちゃんとシーンごと素材渡したらそのままいけそ",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 0,
      "views": 229,
      "snapshotAt": "2026-08-16T04:44:06.092Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit rough short-form prompt trial and proposed per-scene asset refinement"
  },
  {
    "id": "x-2088824211204882835",
    "title": "M5 Max 上的威尔·史密斯挑战",
    "titleEn": "The Will Smith Challenge on an M5 Max",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者在 M5 Max 上运行威尔·史密斯挑战。 成片规格为 3 秒 · 608×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author runs the Will Smith Challenge on an M5 Max. Output: 3s · 608×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/liguria/status/2088824211204882835",
    "sourceLabel": "X 原帖 · @liguria",
    "author": "KS",
    "publishedAt": "2026-08-16T03:04:35.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088824211204882835.mp4",
    "posterUrl": "/posters/x/2088824211204882835.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "608×352",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "M5 Max で Minimax H3 動かしてウィルスミスチャレンジ。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 47,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit Will Smith Challenge and M5 Max hardware"
  },
  {
    "id": "x-2088823069137936600",
    "title": "在既有指令中追加角色设定图的实验",
    "titleEn": "Adding a Character Sheet to an Existing Prompt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "作者在他人开发的既有指令中追加角色设定图，观察加入角色参考后的结果。 成片规格为 15 秒 · 2944×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author adds a character sheet to an existing prompt developed by another creator and observes the result with that character reference included. Output: 15s · 2944×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiehon_aya/status/2088823069137936600",
    "sourceLabel": "X 原帖 · @aiehon_aya",
    "author": "妖精アーヤさん",
    "publishedAt": "2026-08-16T03:00:02.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088823069137936600.mp4",
    "posterUrl": "/posters/x/2088823069137936600.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おおおおおおおおお MiniMax H3！\nmeropanさんかっちょいいプロンプト開発してるー！\n\nキャラクターシートを添付したらどうなるのか気になったので実験させてもらいました。かっこいい！！！",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 27,
      "views": 1898,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit existing third-party prompt, attached character sheet, and bounded reference-addition experiment"
  },
  {
    "id": "x-2088816093410500876",
    "title": "米色纸纹背景上的红衣武士卡通角色",
    "titleEn": "A Red-Clad Cartoon Samurai on a Beige Paper-Textured Background",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "扁平卡通武士扎着发髻，穿红色外褂与白色下装，独自站在留有大片空白的米色纸纹背景中央。 成片规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A flat cartoon samurai with a topknot wears a red outer robe and white trousers, standing alone at the center of a beige paper-textured background with ample negative space. Output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2088816093410500876",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-08-16T02:32:19.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088816093410500876.mp4",
    "posterUrl": "/posters/x/2088816093410500876.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI × モーショングラフィックス\n\nMiniMax H3 Flova(@Flovaai_Japan )で制作\n\nprompt:",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 41,
      "views": 2891,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "poster-media-review: explicit flat cartoon samurai, topknot, red outer robe, white trousers, centered figure, and beige paper texture; sourceCaption: explicit motion-graphics format"
  },
  {
    "id": "x-2088815814736531596",
    "title": "绿皮机车机械变形为机器人",
    "titleEn": "A Green Locomotive Mechanically Transforms into a Robot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "编号 8173 的绿黄机车在山地铁路上展开并重组，最终成为保留车头结构与涂装的巨型机器人。 成片规格为 15 秒 · 1536×672。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Green-and-yellow locomotive 8173 unfolds and reorganizes on a mountain railway, ending as a giant robot that retains the engine's front and livery. Output: 15s · 1536×672. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use <Image 0>  the exact FIRST FRAME and <Image 1>as the exact LAST FRAME. Create one continuous 15-second photorealistic cinematic transformation connecting these two states.  Maintain the EXACT same railroad location throughout: same curving track, green grass, surrounding trees, distant rolling forested hills and mountains, sky, morning atmosphere, perspective, and overall natural lighting. The environment does NOT transform, shift, regenerate, or change location. No additional people, vehicles, trains, or robots appear.  The locomotive itself physically transforms into the robot shown in <Image 1>. This must be a true complex mechanical transformation, NOT a dissolve, magical morph, liquid-metal effect, crossfade, replacement, or sudden character swap. Every major robot component originates visibly from physical locomotive components. Preserve the locomotive's recognizable green-and-yellow painted metal, weathering, markings, wheels, panels, vents, grilles, lights, and mechanical construction as they reorganize into the final robot.  [00:00–00:02.5] Begin exactly on <Image 0>. Hold the intact locomotive briefly. The engine idles subtly. Then small mechanical movements begin: suspension compresses, exterior panels unlock and separate along real seams, vents shift, internal gears engage, and heavy metal sections begin sliding apart. The camera makes a subtle cinematic adjustment to accommodate the increasing vertical scale.  [00:02.5–00:06] The transformation becomes larger and clearly mechanical. The locomotive's wheel assemblies separate and rotate downward, reorganizing into two enormous articulated legs while remaining visibly composed of train machinery. Axles rotate, wheel trucks pivot, steel panels telescope outward, internal pistons extend, and mechanical joints lock together. The locomotive remains physically planted over the same section of track.  The main locomotive body begins rising upward as the newly formed legs extend beneath it. No pieces float independently and nothing materializes from nowhere.  [00:06–00:10] The locomotive body folds and compacts into the robot torso. Large green exterior panels rotate inward and interlock across the chest while recognizable yellow locomotive markings remain incorporated into the armor.  Side machinery unfolds outward into shoulders. Long mechanical assemblies telescope and rotate downward into articulated arms. Hands assemble from smaller interlocking mechanical components at the ends of the forearms. Visible gears turn, pistons compress, cables tighten, panels slide across one another, and heavy locking mechanisms engage.  The camera smoothly pulls back and tilts upward only as necessary to keep the entire transformation readable while preserving the original environment and spatial orientation.  [00:10–00:13] The upper locomotive housing separates and folds inward. Smaller panels rapidly rotate, collapse and interlock around an emerging mechanical neck assembly. The robot head rises mechanically from inside the reorganized machinery; face plates slide into position, the final head structure locks together, and the eyes illuminate.  Remaining locomotive panels snap precisely into their final locations across the shoulders, chest, back, arms and legs. Final pistons compress and locking mechanisms engage with tremendous physical weight.  [00:13–00:15] The transformation completes. The enormous robot settles its weight onto the railway tracks with a heavy suspension-like compression through its legs. Tiny secondary panels finish closing and locking.  The camera settles naturally into the exact composition required by <Image 1>. The robot becomes the precise completed design, proportions, orientation and pose shown in <Image 1>, standing in the same landscape.  End EXACTLY on <Image 1> and hold the completed robot briefly. No further transformation.  overall_soundscape:  Sound ON. Highly detailed realistic mechanical transformation audio synchronized precisely to visible movement: deep diesel engine rumble, enormous steel panels sliding and rotating, gears grinding smoothly, hydraulic pistons releasing and compressing, servomotors, heavy hinges, wheel assemblies rotating, suspension groans, metal components telescoping, powerful mechanical locking CLUNKS and deep resonant impacts. Transformation sounds increase in scale as larger components move.  During the final assembly, several massive sequential mechanical locks culminate in one deep heavy final THUNK as the robot reaches full form. Afterward, quieter servo settling, cooling metal ticks and a low mechanical idle remain.  Maintain subtle natural countryside ambience underneath: light morning wind through grass and trees and distant birds.  No dialogue.  non_diegetic_music:  None. No score, no musical tones, no trailer music. Let the mechanical transformation and natural environment provide the entire soundtrack.",
    "sourceUrl": "https://x.com/EndFolding79421/status/2088815814736531596",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-08-16T02:31:13.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088815814736531596.mp4",
    "posterUrl": "/posters/x/2088815814736531596.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1536×672",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Reference.\n\n  Use <Image 0>  the exact FIRST FRAME and <Image 1>as the exact LAST FRAME. Create one continuous 15-second photorealistic cinematic transformation connecting these two states.  Maintain the EXACT same railroad location throughout: same curving track, green grass, surrounding trees, distant rolling forested hills and mountains, sky, morning atmosphere, perspective, and overall natural lighting. The environment does NOT transform, shift, regenerate, or change location. No additional people, vehicles, trains, or robots appear.  The locomotive itself physically transforms into the robot shown in <Image 1>. This must be a true complex mechanical transformation, NOT a dissolve, magical morph, liquid-metal effect, crossfade, replacement, or sudden character swap. Every major robot component originates visibly from physical locomotive components. Preserve the locomotive's recognizable green-and-yellow painted metal, weathering, markings, wheels, panels, vents, grilles, lights, and mechanical construction as they reorganize into the final robot.  [00:00–00:02.5] Begin exactly on <Image 0>. Hold the intact locomotive briefly. The engine idles subtly. Then small mechanical movements begin: suspension compresses, exterior panels unlock and separate along real seams, vents shift, internal gears engage, and heavy metal sections begin sliding apart. The camera makes a subtle cinematic adjustment to accommodate the increasing vertical scale.  [00:02.5–00:06] The transformation becomes larger and clearly mechanical. The locomotive's wheel assemblies separate and rotate downward, reorganizing into two enormous articulated legs while remaining visibly composed of train machinery. Axles rotate, wheel trucks pivot, steel panels telescope outward, internal pistons extend, and mechanical joints lock together. The locomotive remains physically planted over the same section of track.  The main locomotive body begins rising upward as the newly formed legs extend beneath it. No pieces float independently and nothing materializes from nowhere.  [00:06–00:10] The locomotive body folds and compacts into the robot torso. Large green exterior panels rotate inward and interlock across the chest while recognizable yellow locomotive markings remain incorporated into the armor.  Side machinery unfolds outward into shoulders. Long mechanical assemblies telescope and rotate downward into articulated arms. Hands assemble from smaller interlocking mechanical components at the ends of the forearms. Visible gears turn, pistons compress, cables tighten, panels slide across one another, and heavy locking mechanisms engage.  The camera smoothly pulls back and tilts upward only as necessary to keep the entire transformation readable while preserving the original environment and spatial orientation.  [00:10–00:13] The upper locomotive housing separates and folds inward. Smaller panels rapidly rotate, collapse and interlock around an emerging mechanical neck assembly. The robot head rises mechanically from inside the reorganized machinery; face plates slide into position, the final head structure locks together, and the eyes illuminate.  Remaining locomotive panels snap precisely into their final locations across the shoulders, chest, back, arms and legs. Final pistons compress and locking mechanisms engage with tremendous physical weight.  [00:13–00:15] The transformation completes. The enormous robot settles its weight onto the railway tracks with a heavy suspension-like compression through its legs. Tiny secondary panels finish closing and locking.  The camera settles naturally into the exact composition required by <Image 1>. The robot becomes the precise completed design, proportions, orientation and pose shown in <Image 1>, standing in the same landscape.  End EXACTLY on <Image 1> and hold the completed robot briefly. No further transformation.  overall_soundscape:  Sound ON. Highly detailed realistic mechanical transformation audio synchronized precisely to visible movement: deep diesel engine rumble, enormous steel panels sliding and rotating, gears grinding smoothly, hydraulic pistons releasing and compressing, servomotors, heavy hinges, wheel assemblies rotating, suspension groans, metal components telescoping, powerful mechanical locking CLUNKS and deep resonant impacts. Transformation sounds increase in scale as larger components move.  During the final assembly, several massive sequential mechanical locks culminate in one deep heavy final THUNK as the robot reaches full form. Afterward, quieter servo settling, cooling metal ticks and a low mechanical idle remain.  Maintain subtle natural countryside ambience underneath: light morning wind through grass and trees and distant birds.  No dialogue.  non_diegetic_music:  None. No score, no musical tones, no trailer music. Let the mechanical transformation and natural environment provide the entire soundtrack. ```",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 84,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "prompt+three-frame-review"
  },
  {
    "id": "x-2088812000054853646",
    "title": "从网络迷因到图像再到动态成片的多模型链路",
    "titleEn": "A Multi-Model Pipeline from Meme to Image to Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "制作链路从网络迷因出发，依次经过 ChatGPT 生成指令、Krea 2 与 Qwen 3.8 处理图像，最后将图像与指令转为动态成片。 成片规格为 8 秒 · 576×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The production chain starts from a meme, moves through ChatGPT prompt writing and image work in Krea 2 and Qwen 3.8, then converts the image and prompt into a finished motion piece. Output: 8s · 576×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Adrian_Vefor/status/2088812000054853646",
    "sourceLabel": "X 原帖 · @Adrian_Vefor",
    "author": "Adrian",
    "publishedAt": "2026-08-16T02:16:03.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088812000054853646.mp4",
    "posterUrl": "/posters/x/2088812000054853646.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "576×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Meme → ChatGPT → Prompt → Krea 2 → Image → Qwen 3.8  → Image + Prompt → MiniMax H3 → 🎬\n\nSomewhere along the way, it became magic. ✨",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 55,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit ordered pipeline from meme through ChatGPT, prompt, two image stages, image-plus-prompt input, and final motion output"
  },
  {
    "id": "x-2088811121432265151",
    "title": "镜面地板上看似裙摆的服装反射错觉",
    "titleEn": "A Skirt-Like Costume Reflection Illusion on a Mirrored Floor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女孩站在高反射镜面地板上，地面倒影一度看似裙摆，原帖说明实际映出的是服装的飘动装饰，女孩穿着短裤。 成片规格为 4 秒 · 672×1184。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl stands on a highly reflective mirrored floor where the reflection initially resembles a skirt; the source explains that it is actually a fluttering costume element and that she is wearing shorts. Output: 4s · 672×1184. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/E4SRiAbVCGLxGfy/status/2088811121432265151",
    "sourceLabel": "X 原帖 · @E4SRiAbVCGLxGfy",
    "author": "MRSラボ",
    "publishedAt": "2026-08-16T02:12:34.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088811121432265151.mp4",
    "posterUrl": "/posters/x/2088811121432265151.jpg",
    "duration": 4,
    "aspectRatio": "portrait",
    "resolution": "672×1184",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で動画生成中に偶然起きた奇跡\n\n反射率の高い鏡面の床に立った女の子、一見スカートの中が床に写り込んじゃってる！？と見えますが女の子はホットパンツを履いており床に映ってるスカートに見えるのは衣装のひらひらした部分！\n\nつまりこれはセーフなんです！（たぶんね",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 70,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit female subject, highly reflective mirrored floor, skirt-like reflection illusion, costume-frill explanation, and shorts"
  },
  {
    "id": "x-2088809799496732881",
    "title": "《什么是细胞》：由一张概念图续接十二个镜头",
    "titleEn": "What Is a Cell? Twelve Chained Shots from One Concept Image",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "六十秒科普短片《什么是细胞》从一张细胞概念图出发，通过尾帧续接形成十二个连续镜头。 成片规格为 60 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The sixty-second science short What Is a Cell? starts from one cell concept image and chains twelve continuous shots through their ending frames. Output: 60s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SynmitoYao/status/2088809799496732881",
    "sourceLabel": "X 原帖 · @SynmitoYao",
    "author": "三千秒差距",
    "publishedAt": "2026-08-16T02:07:19.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088809799496732881.mp4",
    "posterUrl": "/posters/x/2088809799496732881.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "最近在 4×RTX A6000 的本地服务器上部署了 MiniMax H3，然后试水了一下科普视频生产流：从一张细胞概念图开始，通过图生视频和末帧续拍生成 12 个连续镜头，最终合成 60 秒《什么是细胞》科普短片。效果是这样的：",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 165,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit work title, science-explainer format, single cell-concept image, twelve continuous shots, final-frame continuation, and sixty-second duration"
  },
  {
    "id": "x-2088800345019338866",
    "title": "360p 素材增强至高清观感",
    "titleEn": "Enhancing 360p Footage to an HD Look",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原始内容以 360p 制作，经 Cuttap 处理后被作者评价为呈现接近高清的观感。 成片规格为 31 秒 · 3840×1634。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source material is created at 360p and, after processing with Cuttap, is judged by the author to look close to high definition. Output: 31s · 3840×1634. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ReyReinoso/status/2088800345019338866",
    "sourceLabel": "X 原帖 · @ReyReinoso",
    "author": "Rey Josias Reinoso",
    "publishedAt": "2026-08-16T01:29:44.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088800345019338866.mp4",
    "posterUrl": "/posters/x/2088800345019338866.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "3840×1634",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "360p or “HD”?\nMade this with Minimax H3 in 360p…\nBut after running it through https://cuttap.com, it looks straight-up HD 🔥360p that doesn’t look like 360p. @MiniMax_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 181,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit 360p source, Cuttap enhancement step, and reported HD-like result"
  },
  {
    "id": "x-2088800116555595800",
    "title": "单图单指令在 RTX 5090 上的十五秒运行",
    "titleEn": "A Fifteen-Second Single-Image, Single-Prompt Run on an RTX 5090",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一次运行只使用一张图片与一条指令，在 RTX 5090 上完成；公开媒体元数据记录时长为十五秒。 成片规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A run uses only one image and one prompt on an RTX 5090, with the public media metadata recording a fifteen-second duration. Output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HenryRhodesd/status/2088800116555595800",
    "sourceLabel": "X 原帖 · @HenryRhodesd",
    "author": "Dandumt",
    "publishedAt": "2026-08-16T01:28:50.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088800116555595800.mp4",
    "posterUrl": "/posters/x/2088800116555595800.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, One image, one prompt and this is what it created locally on my RTX 5090.\nWhat do you think? 👇",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 60,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit single image, single prompt, and RTX 5090 execution; public media metadata: fifteen-second duration"
  },
  {
    "id": "x-2088799437191848073",
    "title": "灯串映照的夜市摊道与人群",
    "titleEn": "Crowds Along a Night-Market Aisle beneath String Lights",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卡通风夜市两侧排列绿色与红色摊位，行人穿行于中央湿润反光的通道，上方悬着多排暖色灯串。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Green and red stalls line both sides of a stylized night market as people fill the wet, reflective central aisle beneath rows of warm string lights. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2088799437191848073",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-08-16T01:26:08.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088799437191848073.mp4",
    "posterUrl": "/posters/x/2088799437191848073.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Good night.\n\n→ @Hailuo_AI MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 10,
      "views": 636,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "poster-media-review: explicit stylized night market, green and red stalls, multiple people, wet reflective aisle, and overhead string lights"
  },
  {
    "id": "x-2088797656118390866",
    "title": "淋浴下触碰湿发的人物近景",
    "titleEn": "A Person Touches Wet Hair Under a Shower",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿黑色两件式泳装的人站在淋浴环境中，反复触碰湿发并改变手部位置，后段镜头逐渐靠近。 成片规格为 10 秒 · 640×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A person in a black two-piece swimsuit remains under a shower, touches wet hair, changes hand placement, and appears in closer framing near the end. Output: 10s · 640×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIPixLab/status/2088797656118390866",
    "sourceLabel": "X 原帖 · @AIPixLab",
    "author": "AI写真ラボ / AIバーチャルカメラマン",
    "publishedAt": "2026-08-16T01:19:03.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088797656118390866.mp4",
    "posterUrl": "/posters/x/2088797656118390866.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "640×960",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🎦遅ればせながらのMiniMax H3📽️ \n#MiniMaxH3 \n\nTurbo LoRA適用したら、2倍以上高速化\n\n✅ RTX3060 12GB\n✅ メインメモリ64GB\n✅ Windows11\n✅ 解像度1056 x 608設定\n✅ 10秒動画\n🆕 Turbo LoRA適用\n    minimax_h3_turbo_v4_step600_ema.safetensors\n⏩生成時間 54分38秒 ▶️ 25分2秒\n\n8秒モノなら15分でした。\n\nきっとラボのような弱小グラボほど効果絶大ですね\n誰か新しいグラボください🤣",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 9,
      "views": 1188,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2088795571079586079",
    "title": "制服角色的多步肖像对比网格",
    "titleEn": "A Multi-Step Portrait Grid of a Uniformed Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "同一制服角色的全身原图与多张带步骤标记的肖像变体排列在重复网格中。 成片规格为 5 秒 · 1620×1728。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A full-body source image of a uniformed character is arranged beside several step-labeled portrait variants in a repeating grid. Output: 5s · 1620×1728. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2088795571079586079",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-08-16T01:10:46.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088795571079586079.mp4",
    "posterUrl": "/posters/x/2088795571079586079.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1620×1728",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "musubi-tunerでMiniMax H3のt2v用キャラLoRA作成完了。RTX4070で約2時間30分(1000step, 100枚)。stepごとの比較結果。Source Imageは学習データの1枚です。300stepで似始め、500step以後は変わらない感じ。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 113,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2088795278086373649",
    "title": "逐步稳定的工作流与女性角色输出",
    "titleEn": "An Increasingly Stable Workflow with a Female Character Output",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者称周五搭建的免费工作流经过调试后逐渐稳定，并明确评价随帖女性角色十分可爱。 成片规格为 9 秒 · 736×416。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author says a free workflow set up on Friday has become increasingly stable after troubleshooting and explicitly praises the attached female character as very cute. Output: 9s · 736×416. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/muku_sns/status/2088795278086373649",
    "sourceLabel": "X 原帖 · @muku_sns",
    "author": "むく | AIアートのプロ",
    "publishedAt": "2026-08-16T01:09:36.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088795278086373649.mp4",
    "posterUrl": "/posters/x/2088795278086373649.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "736×416",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "金曜日にローカル構築して無料生成できるようになったMiniMax H3。一筋縄じゃいかんところもあるけど大分安定するようになってきた。\nある程度できるようになったら、ノウハウまとめようと思います。しかし、この美女かわいすぎる。",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 8,
      "views": 1031,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit recently built free workflow, increasing stability, and female-character subject"
  },
  {
    "id": "x-2088794783179477481",
    "title": "卧室里穿白衬衫蓝领结的动漫少女",
    "titleEn": "An Anime Girl in a White Shirt and Blue Bow Stands in a Bedroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名黑色短发、紫色眼睛的动漫少女穿着白色短袖衬衫与蓝色领结，站在带床铺、书架和窗户的卧室中。 成片规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short, black-haired anime girl with purple eyes wears a white short-sleeved shirt and blue bow, standing in a bedroom with a bed, bookshelf, and window. Output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/urotaichi/status/2088794783179477481",
    "sourceLabel": "X 原帖 · @urotaichi",
    "author": "urοtаiсhi",
    "publishedAt": "2026-08-16T01:07:38.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088794783179477481.mp4",
    "posterUrl": "/posters/x/2088794783179477481.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のテスト",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 67,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "poster-media-review: explicit short black hair, purple eyes, white shirt, blue bow, standing figure, bed, bookshelf, and window"
  },
  {
    "id": "x-2088793684292825106",
    "title": "Krea 2 插画的动漫式动态转换",
    "titleEn": "Anime-Style Motion from a Krea 2 Illustration",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "作者借用 CIVITAI 指令并以 Krea 2 重新生成插画，再将该图转换为动漫式动态内容。 成片规格为 20 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author borrows a prompt from CIVITAI, regenerates an illustration with Krea 2, and then turns that image into anime-style motion. Output: 20s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/breath_in_code/status/2088793684292825106",
    "sourceLabel": "X 原帖 · @breath_in_code",
    "author": "Linaco⚽️WSL",
    "publishedAt": "2026-08-16T01:03:16.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088793684292825106.mp4",
    "posterUrl": "/posters/x/2088793684292825106.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MINIMAX H3\nQ : アニメやイラストにも使えますか？\n\nCIVITAIでプロンプトをお借りしKrea2で再生成した画像を使い動画にしました。\nA : イラスト系でも簡単に使えますよ😊",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 74,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit borrowed CIVITAI prompt, Krea 2 regenerated illustration, image-driven conversion, and anime-or-illustration suitability"
  },
  {
    "id": "x-2088793346445852803",
    "title": "开源模型驱动的三十七秒剪辑实践",
    "titleEn": "A Thirty-Seven-Second Edit Built with an Open Model",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "作者完成了一段由开放模型驱动的剪辑，并称若不采用这套方案，工作量将增加约十倍；公开媒体元数据记录时长为三十七秒。 成片规格为 37 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author completes an edit driven by an open model and says the work would take roughly ten times more effort without this setup; public media metadata records a duration of thirty-seven seconds. Output: 37s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JeffreyJan5774/status/2088793346445852803",
    "sourceLabel": "X 原帖 · @JeffreyJan5774",
    "author": "Jeffrey Jang",
    "publishedAt": "2026-08-16T01:01:56.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088793346445852803.mp4",
    "posterUrl": "/posters/x/2088793346445852803.jpg",
    "duration": 37,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@MiniMax_AI H3 is singlehandedly carrying open source gen AI video. \n\nI got it set up last night put together an edit. It's no Seedance, but to do this without Minimax H3 and locally... it would be 10x more work.\n\n#minimax #aivideo #comfyui #generativeai #seedance",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 81,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit open-model setup, completed edit, estimated tenfold effort difference; public media metadata: thirty-seven-second duration"
  },
  {
    "id": "x-2088785458486751513",
    "title": "用发型特征提升偶像组合的一致性",
    "titleEn": "Improving Idol-Group Consistency with Hairstyle Cues",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "偶像组合包含多名成员时精度会下降，作者发现明确描述各自发型特征后，一致性明显提高。 成片规格为 10 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Accuracy declines when an idol group includes multiple members, but the author finds that explicitly describing each hairstyle substantially improves consistency. Output: 10s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HackCooking/status/2088785458486751513",
    "sourceLabel": "X 原帖 · @HackCooking",
    "author": "Cooking life hack",
    "publishedAt": "2026-08-16T00:30:35.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088785458486751513.mp4",
    "posterUrl": "/posters/x/2088785458486751513.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3でアイドルグループを出してみた。\n複数あるとやっぱり精度は低くなるのですが、髪型の特徴を伝えるだけで一気に精度が上がります！＾＾",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 175,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit idol-group subject, multi-character accuracy decline, hairstyle-feature prompting, and reported improvement"
  },
  {
    "id": "x-2088784986111639750",
    "title": "青柠色图形中的插画角色与无线耳机",
    "titleEn": "An Illustrated Character and Wireless Earbuds in Lime Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "青柠色图形画面在插画人物全身姿势、耳机佩戴近景与无线耳机充电盒之间切换。 成片规格为 15 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Lime graphic panels move among full-body poses of an illustrated character, a close view of an earbud in the ear, and a wireless charging case. Output: 15s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AiXsatoshi/status/2088784986111639750",
    "sourceLabel": "X 原帖 · @AiXsatoshi",
    "author": "AI✖️Satoshi⏩️",
    "publishedAt": "2026-08-16T00:28:43.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088784986111639750.mp4",
    "posterUrl": "/posters/x/2088784986111639750.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ローカルPC、MiniMax-H3の\nモーショングラフィックスを試す①",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 15,
      "views": 1383,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "blind-media-review"
  },
  {
    "id": "x-2088784380525232277",
    "title": "墙撞、偏置正撞、列车撞轿车与侧面碰撞的双模型对比",
    "titleEn": "Two-Model Comparison of Wall, Offset Head-On, Train, and T-Bone Crashes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两个模型以对应调整后的相同碰撞指令、相同随机种子和相同冲击条件，顺序比较撞墙、偏置正面碰撞、货运列车撞击故障轿车与闯红灯侧面碰撞。 成片规格为 70 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two models use equivalently adapted crash prompts, the same seeds, and the same impacts to compare, in sequence, a wall crash, an offset head-on collision, a freight train striking a stalled sedan, and a T-bone through a red light. Output: 70s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BennyDaBall_OG/status/2088784380525232277",
    "sourceLabel": "X 原帖 · @BennyDaBall_OG",
    "author": "BennyDaBall",
    "publishedAt": "2026-08-16T00:26:18.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088784380525232277.mp4",
    "posterUrl": "/posters/x/2088784380525232277.jpg",
    "duration": 70,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 vs LTX 2.5. Same crash briefs (prompts adapted for each model), same seeds, same impacts. Sequential head to head.\n\nWall. Offset head-on. Freight vs a stalled sedan. T-bone through a red.\n\nH3 is still the quality lane — the crumple in the wall impact and the passing to-head-on actually read like metal.\nLTX 2.5 is stupid fast (~60-90s vs ~6 min) and it actually flung the car on the train hit in a hilarious kinda way lol.\nT-bone? MiniMax H3 understood, LTX 2.5 didn't quite understand what was supposed to happen. \n\nAll local. RTX 5090. Open weights.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 78,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption: explicit two-model sequential comparison, equivalent prompts, same seeds and impacts, four named crash scenarios, runtime difference, and RTX 5090 environment"
  },
  {
    "id": "x-2088777468039585809",
    "title": "火焰鱼与冰霜鱼迎面对峙",
    "titleEn": "A Fire Fish Faces an Ice Fish",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一条由火焰与熔岩构成的橙红怪鱼，与一条裹着冰霜和气泡的蓝白怪鱼迎面对视。 成片规格为 20 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An orange-red creature formed from fire and molten textures faces a blue-white fish wrapped in frost and bubbles. Output: 20s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FunTheoryX/status/2088777468039585809",
    "sourceLabel": "X 原帖 · @FunTheoryX",
    "author": "FunTheory",
    "publishedAt": "2026-08-15T23:58:50.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088777468039585809.mp4",
    "posterUrl": "/posters/x/2088777468039585809.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 its fantastic!!!!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 110,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "cached poster frame from the hosted media"
  },
  {
    "id": "x-2088776875590209796",
    "title": "清晨梦中妻子的再现",
    "titleEn": "Recreating a Wife Seen in a Morning Dream",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "创作者尝试重现清晨梦里见到的妻子，并记录 RTX 5090 完成这段输出约用二十分钟。 成片规格为 10 秒 · 1280×732。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The creator attempts to recreate his wife as seen in a morning dream and reports that an RTX 5090 took about twenty minutes to produce the sequence. Output: 10s · 1280×732. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mazzo/status/2088776875590209796",
    "sourceLabel": "X 原帖 · @mazzo",
    "author": "Koya Matsuo",
    "publishedAt": "2026-08-15T23:56:29.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088776875590209796.mp4",
    "posterUrl": "/posters/x/2088776875590209796.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×732",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3。5090で20分。今朝の夢で逢った妻を再現してみた",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 23,
      "views": 881,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption explicitly states the depicted person, dream premise, hardware, and runtime"
  },
  {
    "id": "x-2088770461966680507",
    "title": "未公开歌曲的舞台唇形同步",
    "titleEn": "Stage Lip-Sync to an Unreleased Song",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "静见音 Alternative 在舞台画面中为一首未公开歌曲进行唇形同步，成片由 CapCut AI 制作。 成片规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Oto Shizumi Alternative performs a stage lip-sync to an unreleased song in a sequence created with CapCut AI. Output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/otosa_vt/status/2088770461966680507",
    "sourceLabel": "X 原帖 · @otosa_vt",
    "author": "静見おと : otosa",
    "publishedAt": "2026-08-15T23:31:00.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088770461966680507.mp4",
    "posterUrl": "/posters/x/2088770461966680507.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 2K\nX未発表曲でリップシンクして、ステージ映像を作ってみました✨\nパスワードわかんないよ～！\n\n--- \n出演：静見おと Alternative( @otosa_vt )\n@capcutapp_jp\n#capcutcpp\n#capcut\nCreated by CapCut AI",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 15,
      "views": 275,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption explicitly identifies the performer, stage setup, unreleased song, lip-sync, and CapCut AI credit"
  },
  {
    "id": "x-2088766686870552945",
    "title": "猫姿势术语的双模型识别",
    "titleEn": "Cat-Pose Vocabulary Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两种模型分别尝试理解“肚皮朝天”和“苏格兰折耳坐”等猫姿势术语，并对呈现结果进行并列观察。 成片规格为 35 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two models interpret Japanese terms for belly-up and Scottish Fold sitting poses, with the resulting cat postures compared side by side. Output: 35s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pan_soramame_da/status/2088766686870552945",
    "sourceLabel": "X 原帖 · @pan_soramame_da",
    "author": "おまっめ🐱",
    "publishedAt": "2026-08-15T23:16:00.000Z",
    "addedAt": "2026-08-16T04:44:06.093Z",
    "mediaUrl": "/media/x-2088766686870552945.mp4",
    "posterUrl": "/posters/x/2088766686870552945.jpg",
    "duration": 35,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "夏の自由研究🍉\n『猫用語の検証』\nSeedance2.5 × MiniMax H3\n\n『へそ天』とか『スコ座り』とか\n用語だけで通じるなら\nメチャクチャ嬉しい！と思い検証してみた🐈\n\n天下のへそ天様は両方ともOK！笑\nMiniMax H3が結構知ってて驚いた😹",
    "engagement": {
      "replies": 11,
      "reposts": 13,
      "likes": 72,
      "views": 1173,
      "snapshotAt": "2026-08-16T04:44:06.093Z"
    },
    "approvedAt": "2026-08-16T04:44:06.093Z",
    "editorialBasis": "sourceCaption explicitly names the two cat-pose terms and the two-model comparison"
  },
  {
    "id": "x-2089141649503044017",
    "title": "乘法口诀的九宫格鱼群",
    "titleEn": "Multiplication Tables with Fish Pairs",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "橙色猫咪带着九个水池，用成对鱼群和逐拍算式演示二的乘法表。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An orange cat guides a nine-pool grid where paired fish and beat-matched equations teach the two-times table. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description: [Shot 1] Adorable rounded 3D animation for preschool children, soft studio lighting, subtle contact shadows, smooth gradients, and a calm uncluttered vertical composition on a clean off-white background with a very soft pale-blue glow behind the center. Use a restrained palette of off-white, warm orange, cream, turquoise blue, and dark gray: orange is the cat, turquoise is the fish and every correct answer, dark gray is the equation. The camera is a Static Shot for the entire film. A small smiling cat mascot bounces gently into the upper area: a big rounded head on a short chubby body, warm orange fur, a cream muzzle, cream belly and cream paw tips, two small rounded triangle ears with pale pink inner lining, round full cheeks, simple black dot eyes, tiny whiskers, and a thick rounded tail. It lands with a soft squash-and-stretch, its tail curls up once behind it, and it taps the air with one front paw, creating a spreading circular pale-blue water ripple. The ripple opens outward and reveals a three-by-three grid of nine small empty pale-blue water pools in the lower half of the frame, with a clear empty space for one large equation above them. A warm young Japanese female narrator with a bright, soft voice (S1) chants in a steady rhythmic sing-song counting cadence, one line per beat, and begins: <d>[Japanese] にのだん!</d>\n\n[Shot 2] At 00:01.200, one large thick rounded dark-gray equation appears in the upper space and is rewritten twice in place, holding each state clearly on its own beat: first \"2 × 1 = 2\", then \"2 × 2 = 4\", then \"2 × 3 = 6\". On each new equation, exactly one matching pair of two plump turquoise-blue fish drops together into the next empty water pool with a small splash. The two fish of a pair are identical, side by side, both facing right and slightly offset so both are clearly countable, each with a cream belly, a triangular tail, and one round black eye. The pairs fill the top row from left to right, and the answer numeral turns turquoise as each pair lands. At the end of this shot exactly three pools in the top row are filled, each holding exactly two fish, for six fish in total; six pools remain empty. (S1) chants on the beat: <d>[Japanese] にいちがに。ににんがし。にさんがろく。</d>\n\n[Shot 3] At 00:04.700, a soft pale-blue water ripple wipes across the frame and the same large equation continues to be rewritten in place on each beat: \"2 × 4 = 8\", then \"2 × 5 = 10\", then \"2 × 6 = 12\". One matching pair of two fish drops into the next empty pool on each new equation, filling the middle row from left to right, and the answer numeral turns turquoise as each pair lands. Two-digit answers stay large, correctly formed, and fully readable. The three pairs already in the top row stay exactly where they are and do not move, change, or multiply. At the end of this shot exactly six pools are filled, each holding exactly two fish, for twelve fish in total; three pools remain empty. (S1) chants on the beat: <d>[Japanese] にしがはち。にごじゅう。にろくじゅうに。</d>\n\n[Shot 4] At 00:08.100, a soft pale-blue water ripple wipes across the frame and the large equation is rewritten in place on each beat for the last time: \"2 × 7 = 14\", then \"2 × 8 = 16\", then \"2 × 9 = 18\". One matching pair of two fish drops into the next empty pool on each new equation, filling the bottom row from left to right, and the answer numeral turns turquoise as each pair lands. The six pairs already placed stay exactly where they are and do not move, change, or multiply. At the end of this shot all nine pools are filled, each holding exactly two fish, for eighteen fish in total. (S1) chants on the beat: <d>[Japanese] にしちじゅうし。にはちじゅうろく。にくじゅうはち。</d>\n\n[Shot 5] At 00:12.100, the final equation \"2 × 9 = 18\" stays in the upper space and its turquoise answer numeral \"18\" gently enlarges once and settles. Below it, the nine pairs of fish gain small friendly smiles and their tails flick once, in a quick wave from the top-left pool to the bottom-right pool, each pair flicking together as one. The same warm-orange cat mascot from [Shot 1], unchanged in every detail, bounces in at the lower edge beside the grid and points up toward it with one front paw while its tail curls once and small pale-blue sparkles rise around it. Exactly nine pools remain, each still holding exactly two fish, and no new objects enter. (S1) says: <d>[Japanese] よく できました!</d> Hold the complete grid, the equation, and the cat clearly and without further motion until the film ends exactly at 00:15.000.\n\nThroughout the entire film, keep the camera a Static Shot, keep the off-white background clean and empty, keep the single equation large, thick, rounded, correctly shaped, and centered in the upper space, and keep the fish design identical in every shot. Every filled pool always holds exactly two fish - never one, never three - and the two fish of a pair always arrive together, land together, and stay together as one unit. The cat must look exactly the same in [Shot 1] and [Shot 5], identical in fur color, head and body shape, ears, eyes, and tail, and it never appears in [Shot 2], [Shot 3], or [Shot 4]. Only one cat is ever on screen, it stays beside or above the grid, and it never stands on, reaches into, covers, or eats any fish. Only one equation is ever visible at a time. Display only numerals and the symbols × and =. Do not generate any hiragana, katakana, kanji, Latin words, or any other written characters anywhere on screen. Fish only ever accumulate, never disappear, never leave their pool once placed, and never overlap; the number of filled pools must exactly match the multiplier of the current equation, and the total fish count must exactly match the answer numeral, reaching exactly eighteen fish in nine pools at the end. Avoid fast camera movement, sudden cuts, cluttered backgrounds, tiny typography, warped or duplicated numerals, reversed two-digit numbers, wrong counts, extra or missing fish, unpaired single fish, fish appearing before their equation, fish swimming freely around the frame, realistic or scaly fish anatomy, scary expressions, harsh colors, heavy motion blur, and loud or distracting sound effects.\n\noverall_soundscape: A soft double water plop sounds each time a pair of fish drops into a pool. A light padded paw tap sounds when the cat touches the air, and one single soft kitten meow occurs as the cat first bounces into frame. Gentle spreading water ripples mark the two transitions, and a small confirmation chime sounds each time an answer numeral turns turquoise. Tiny shimmer sparkles occur around the cat at the end.\n\nnon_diegetic_music: A light nursery jingle at a steady moderate tempo using marimba, soft celesta, and gently plucked strings, with a clear repeating beat that the chant lands on. Begin with a single marimba figure, add one instrument at each row of the grid, and thin back out for the final line. Resolve with a clean rising three-note chime landing at 15.000 seconds.",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2089141649503044017",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-17T00:05:58.000Z",
    "addedAt": "2026-08-17T02:39:30.454Z",
    "mediaUrl": "/media/x-2089141649503044017.mp4",
    "posterUrl": "/posters/x/2089141649503044017.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で、リズムに合わせてかけ算を学べる動画【パート2】を作りました😃\n\n今回は「2の段」です！\n\n昨夜、うちの子（小学2年生）に「1の段」の動画（引用元ポスト）を見せたところ、けさ\n\n「いちのだ〜ん♩」\n\nと歌っていたそうです😌\n\nまだ学校ではかけ算を習っていないのですが、こうやって歌で覚えていくのは、やっぱり効果がありそうですね😃\n\nちなみに今回の動画は、前回の「1の段」のプロンプトをベースに、Claude Codeに\n\n「次は2の段にして」\n\nと伝えただけ。\n\nちゃんと「2の段」の動画に作り替えてくれました😄\n\nプロンプトはこちら↓\n------------------\nintegrated_multimodal_description: [Shot 1] Adorable rounded 3D animation for preschool children, soft studio lighting, subtle contact shadows, smooth gradients, and a calm uncluttered vertical composition on a clean off-white background with a very soft pale-blue glow behind the center. Use a restrained palette of off-white, warm orange, cream, turquoise blue, and dark gray: orange is the cat, turquoise is the fish and every correct answer, dark gray is the equation. The camera is a Static Shot for the entire film. A small smiling cat mascot bounces gently into the upper area: a big rounded head on a short chubby body, warm orange fur, a cream muzzle, cream belly and cream paw tips, two small rounded triangle ears with pale pink inner lining, round full cheeks, simple black dot eyes, tiny whiskers, and a thick rounded tail. It lands with a soft squash-and-stretch, its tail curls up once behind it, and it taps the air with one front paw, creating a spreading circular pale-blue water ripple. The ripple opens outward and reveals a three-by-three grid of nine small empty pale-blue water pools in the lower half of the frame, with a clear empty space for one large equation above them. A warm young Japanese female narrator with a bright, soft voice (S1) chants in a steady rhythmic sing-song counting cadence, one line per beat, and begins: <d>[Japanese] にのだん!</d>\n\n[Shot 2] At 00:01.200, one large thick rounded dark-gray equation appears in the upper space and is rewritten twice in place, holding each state clearly on its own beat: first \"2 × 1 = 2\", then \"2 × 2 = 4\", then \"2 × 3 = 6\". On each new equation, exactly one matching pair of two plump turquoise-blue fish drops together into the next empty water pool with a small splash. The two fish of a pair are identical, side by side, both facing right and slightly offset so both are clearly countable, each with a cream belly, a triangular tail, and one round black eye. The pairs fill the top row from left to right, and the answer numeral turns turquoise as each pair lands. At the end of this shot exactly three pools in the top row are filled, each holding exactly two fish, for six fish in total; six pools remain empty. (S1) chants on the beat: <d>[Japanese] にいちがに。ににんがし。にさんがろく。</d>\n\n[Shot 3] At 00:04.700, a soft pale-blue water ripple wipes across the frame and the same large equation continues to be rewritten in place on each beat: \"2 × 4 = 8\", then \"2 × 5 = 10\", then \"2 × 6 = 12\". One matching pair of two fish drops into the next empty pool on each new equation, filling the middle row from left to right, and the answer numeral turns turquoise as each pair lands. Two-digit answers stay large, correctly formed, and fully readable. The three pairs already in the top row stay exactly where they are and do not move, change, or multiply. At the end of this shot exactly six pools are filled, each holding exactly two fish, for twelve fish in total; three pools remain empty. (S1) chants on the beat: <d>[Japanese] にしがはち。にごじゅう。にろくじゅうに。</d>\n\n[Shot 4] At 00:08.100, a soft pale-blue water ripple wipes across the frame and the large equation is rewritten in place on each beat for the last time: \"2 × 7 = 14\", then \"2 × 8 = 16\", then \"2 × 9 = 18\". One matching pair of two fish drops into the next empty pool on each new equation, filling the bottom row from left to right, and the answer numeral turns turquoise as each pair lands. The six pairs already placed stay exactly where they are and do not move, change, or multiply. At the end of this shot all nine pools are filled, each holding exactly two fish, for eighteen fish in total. (S1) chants on the beat: <d>[Japanese] にしちじゅうし。にはちじゅうろく。にくじゅうはち。</d>\n\n[Shot 5] At 00:12.100, the final equation \"2 × 9 = 18\" stays in the upper space and its turquoise answer numeral \"18\" gently enlarges once and settles. Below it, the nine pairs of fish gain small friendly smiles and their tails flick once, in a quick wave from the top-left pool to the bottom-right pool, each pair flicking together as one. The same warm-orange cat mascot from [Shot 1], unchanged in every detail, bounces in at the lower edge beside the grid and points up toward it with one front paw while its tail curls once and small pale-blue sparkles rise around it. Exactly nine pools remain, each still holding exactly two fish, and no new objects enter. (S1) says: <d>[Japanese] よく できました!</d> Hold the complete grid, the equation, and the cat clearly and without further motion until the film ends exactly at 00:15.000.\n\nThroughout the entire film, keep the camera a Static Shot, keep the off-white background clean and empty, keep the single equation large, thick, rounded, correctly shaped, and centered in the upper space, and keep the fish design identical in every shot. Every filled pool always holds exactly two fish - never one, never three - and the two fish of a pair always arrive together, land together, and stay together as one unit. The cat must look exactly the same in [Shot 1] and [Shot 5], identical in fur color, head and body shape, ears, eyes, and tail, and it never appears in [Shot 2], [Shot 3], or [Shot 4]. Only one cat is ever on screen, it stays beside or above the grid, and it never stands on, reaches into, covers, or eats any fish. Only one equation is ever visible at a time. Display only numerals and the symbols × and =. Do not generate any hiragana, katakana, kanji, Latin words, or any other written characters anywhere on screen. Fish only ever accumulate, never disappear, never leave their pool once placed, and never overlap; the number of filled pools must exactly match the multiplier of the current equation, and the total fish count must exactly match the answer numeral, reaching exactly eighteen fish in nine pools at the end. Avoid fast camera movement, sudden cuts, cluttered backgrounds, tiny typography, warped or duplicated numerals, reversed two-digit numbers, wrong counts, extra or missing fish, unpaired single fish, fish appearing before their equation, fish swimming freely around the frame, realistic or scaly fish anatomy, scary expressions, harsh colors, heavy motion blur, and loud or distracting sound effects.\n\noverall_soundscape: A soft double water plop sounds each time a pair of fish drops into a pool. A light padded paw tap sounds when the cat touches the air, and one single soft kitten meow occurs as the cat first bounces into frame. Gentle spreading water ripples mark the two transitions, and a small confirmation chime sounds each time an answer numeral turns turquoise. Tiny shimmer sparkles occur around the cat at the end.\n\nnon_diegetic_music: A light nursery jingle at a steady moderate tempo using marimba, soft celesta, and gently plucked strings, with a clear repeating beat that the chant lands on. Begin with a single marimba figure, add one instrument at each row of the grid, and thin back out for the final line. Resolve with a clean rising three-note chime landing at 15.000 seconds.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 572,
      "snapshotAt": "2026-08-17T02:39:30.454Z"
    },
    "approvedAt": "2026-08-17T02:39:30.454Z",
    "editorialBasis": "sourceCaption publishes the complete structured prompt and the reviewed frames show the matching cat, equations, and fish-grid animation"
  },
  {
    "id": "x-2089127930647683146",
    "title": "时光机抵达古波斯",
    "titleEn": "A Time Machine Reaches Ancient Persia",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头穿过时空隧道，抵达带有宫殿、仪仗与人群的古代波斯城市。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera travels through a time tunnel into an ancient Persian city filled with palaces, a procession, and crowds. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ImanOracle/status/2089127930647683146",
    "sourceLabel": "X 原帖 · @ImanOracle",
    "author": "ImanOracle.eth",
    "publishedAt": "2026-08-16T23:11:27.000Z",
    "addedAt": "2026-08-17T02:39:30.455Z",
    "mediaUrl": "/media/x-2089127930647683146.mp4",
    "posterUrl": "/posters/x/2089127930647683146.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "چقدر این مدل Minimax H3 خوبه، یک سرور 5090 اجاره کردم از Vast AI با ۵۰ سنت واسه هر ساعت، تو ۴ دقیقه این ویدیو رو ساخت\nگفتم فکر کن از ماشین زمان استفاده کردیم رفتیم زمان قدیم پرسپولیس!",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 69,
      "views": 4012,
      "snapshotAt": "2026-08-17T02:39:30.455Z"
    },
    "approvedAt": "2026-08-17T02:39:30.455Z",
    "editorialBasis": "sourceCaption states the creator generated the clip with MiniMax H3 and describes the ancient-Persia time-travel premise"
  },
  {
    "id": "x-2089124518166135196",
    "title": "幽灵侦察机的变形预告",
    "titleEn": "Ghost Recon Fighter Transformation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "暗色侦察机在红蓝光带与扫描字幕中完成高速揭晓和机甲变形。 原生视频规格为 19 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark recon fighter races through red-blue light trails, scan typography, and a mecha transformation reveal. Native-video output: 19s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/stonekaiju/status/2089124518166135196",
    "sourceLabel": "X 原帖 · @stonekaiju",
    "author": "Stone Kaiju ᯅ ⚡ ³³º¹",
    "publishedAt": "2026-08-16T22:57:53.000Z",
    "addedAt": "2026-08-17T02:39:30.455Z",
    "mediaUrl": "/media/x-2089124518166135196.mp4",
    "posterUrl": "/posters/x/2089124518166135196.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "combat",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "“Longinus” 💫 \n\nInspired by @aimikoda’s wonderful character reveal prompt 🙏 Worked with Luma in Luma Agents @LumaLabsAI  to convert this to a ship reveal. Referenced a Fighter variant drawn in @Procreate and ran this with Minimax H3.\n\nLuma’s prompt and notes: “LONGINUS — Reveal Trailer Concept\n\nAdapting Aimikoda's H3 character-reveal template to a SHIP. Register: hot-blooded M******. Hero beat: mid-flight Fighter → B****** transformation. Identity authority: Stonekaiju's Procreate V******* drawing (8onK8nEw) — cell-shaded anime.\n\nThe pitch\n\nLONGINUS — the spear that pierces the dark. A stealth deep-recon variant of the Spectral Knights: a ghost the whole trailer, until it stands up. Restraint → release. Dark hull, blue-and-red glow, clean cel-shading, bold anime editorial.\n\nPARAMETERS (parameterized — swap freely)\n\nCHARACTER_NAME = \"LONGINUS\"\n\nSTATUS_TEXT_1 = \"GHOST DRIVE: ONLINE\"\n\nSTATUS_TEXT_2 = \"MODE: DEEP RECON\"\n\nINTERSTITIAL_TEXT_1 = \"PHANTOM VEIL\"\n\nINTERSTITIAL_TEXT_2 = \"TARGET SCAN\"\n\nINTERSTITIAL_TEXT_3 = \"FULL DRIVE\"\n\nSUBTITLE = \"Spectral Knights · Deep-Recon Variant\"\n\nPROJECT_TEXT = \"SPECTRAL KNIGHTS\"\n\nThe ~13-cut reveal (ship beats)\n\nHULL — extreme close-up on a defining detail (Spectral Knights sigil / canopy / a glowing seam), half in shadow. Slow push-in.\n\nIGNITION — the ghost-drive spins up: blue/red glow-filaments race the hull, vents flare, stealth field shimmers alive.\n\nSTATUS CARD — editorial freeze: huge circular field, oversized condensed type GHOST DRIVE: ONLINE, cropped hull detail integrated. Text reads first, ship overlaps.\n\nENGAGE CARD — high-contrast ship silhouette over an accent band, MODE: DEEP RECON, energy ribbon sweep.\n\nWORLD — smash into space: Longinus dives/boosts through a nebula & the hulls of a sleeping enemy fleet. Whip-pan.\n\nDETAIL — the stealth field ripples and it decloaks; vector vanes articulate.\n\nINTERSTITIAL 01 — text card PHANTOM VEIL; ship as a small integrated silhouette.\n\nIMPACT — aggressive angle: threads a defense-turret sweep / debris gauntlet (Itano-lite).\n\nINTERSTITIAL 02 — punchy card TARGET SCAN, scan-lines/UI.\n\nVELOCITY — rushes the lens, environment stretches into speed blur.\n\nINTERSTITIAL 03 — FULL DRIVE — the transformation trigger card, sweeping energy stroke.\n\nHERO — THE TRANSFORMATION — mid-flight Fighter → G****** → B******. (Craft: SLOW-MOTION section on the transform, named & scoped; camera ORBIT during it; dual-identity bridge — the blue/red glow signature + Spectral Knights sigil carry across every form so it reads as ONE machine changing.) Battroid lands the hero pose. Return to speed.\n\nSHIP ID — clean pale field, enormous concentric circles + network nodes, the B******* (or a heroic banking fighter) overlapping huge type LONGINUS, subtitle Spectral Knights · Deep-Recon Variant. Behind: a black ink-brush emblem holding a ghosted spear/lance motif.\n\nEdit + Score\n\nEDIT: aggressive AAA rhythm — hard cuts, graphic matches, whip-pans, freeze-frames, speed ramps; every cut a new composition; the graphic cards break the flight rhythm. No slow dissolves. SCORE: hot-blooded anime brass + driving percussion + synth, hits on cuts/type, restrained ghost-tension → the transformation is the PEAK → final reveal stinger.\n\nCraft integrations (our kit)\n\nIdentity authority: the drawing is the SOLE authority for the FIGHTER form — never redesign; adapt world/VFX/movement to it.\n\nText in POST: in-gen cards carry only BIG status words; render the crisp LONGINUS ID card + emblem in code (Pillow) for legibility — as we did for Valhalla.\n\nCell-shaded throughout: clean lineart, cel shading, bold flat color, anime editorial.\n\nThe honest open decision (flag)\n\nThe Mecha form doesn't exist yet — only your fighter does. Two paths:\n\n(A) Silhouette/backlit Mecha (my lean for the first experiment): keep the transform dramatic and shadowed so your fighter stays the star and identity can't drift. Fast, safe.\n\n(B) Design the Mecha as a full bridged form first (its own plate), then transform to it — truer, but a bigger step and a real identity-lock challenge. Also honest: an original mecha transforming cleanly is the hardest thing to ask a model for in one pass — if H3 fumbles the transform in-gen, we shoot that beat as its own controlled keyframe shot and cut it in.”\n\n Try Luma Agents here… http://lumalabs.ai/app\n\n#LumaCPP",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 11,
      "views": 279,
      "snapshotAt": "2026-08-17T02:39:30.455Z"
    },
    "approvedAt": "2026-08-17T02:39:30.455Z",
    "editorialBasis": "sourceCaption documents the referenced fighter drawing and MiniMax H3 run; reviewed frames confirm the reveal-trailer treatment"
  },
  {
    "id": "x-2089122755077874016",
    "title": "罗宾汉的最后旅程",
    "titleEn": "Robin Hood’s Final Journey",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "老去的罗宾汉在雪地与篝火之间回望自己的选择，形成一支完整的电影化短片。 原生视频规格为 71 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An aging Robin Hood reflects on his choices across snowy landscapes and a firelit camp in a complete cinematic short. Native-video output: 71s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Banyu_Biroeee/status/2089122755077874016",
    "sourceLabel": "X 原帖 · @Banyu_Biroeee",
    "author": "Rian Andria",
    "publishedAt": "2026-08-16T22:50:53.000Z",
    "addedAt": "2026-08-17T02:39:30.455Z",
    "mediaUrl": "/media/x-2089122755077874016.mp4",
    "posterUrl": "/posters/x/2089122755077874016.jpg",
    "duration": 71,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "As his legend begins to fade, Robin Hood embarks on one final journey through Sherwood Forest, confronting the wounds, choices, and sacrifices that shaped the outlaw before the world says goodbye to its greatest hero.\n\nMinimax H3 on @Hailuo_AI \n\n#minimaxh3 #aivideo #hailuoai",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 52,
      "snapshotAt": "2026-08-17T02:39:30.455Z"
    },
    "approvedAt": "2026-08-17T02:39:30.455Z",
    "editorialBasis": "sourceCaption explicitly credits MiniMax H3 on Hailuo and identifies the Robin Hood narrative"
  },
  {
    "id": "x-2089110704108290281",
    "title": "炸虾小姐的上班应援歌",
    "titleEn": "The Fried-Shrimp Lady’s Workday Song",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "炸虾造型的少女角色用歌声为假期结束后的上班日加油。 原生视频规格为 10 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A shrimp-costumed character sings an encouraging tune for the first workday after the holiday. Native-video output: 10s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/asanagiako/status/2089110704108290281",
    "sourceLabel": "X 原帖 · @asanagiako",
    "author": "朝凪あこAIイラスト",
    "publishedAt": "2026-08-16T22:03:00.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089110704108290281.mp4",
    "posterUrl": "/posters/x/2089110704108290281.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "お盆休みも終わって、お仕事ですね…✨\nえびふらいりすにお仕事が嫌って気持ちを伝えたら謎のお歌を歌ってくれました…✨\n皆様えびふらい系お嬢様の応援で頑張りましょう\n\n…これがMiniMax H3の力…！？\n#お嬢様の小さな日常",
    "engagement": {
      "replies": 4,
      "reposts": 9,
      "likes": 21,
      "views": 299,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly identifies MiniMax H3 and describes the character singing the mysterious encouragement song"
  },
  {
    "id": "x-2089106173920522485",
    "title": "菠萝日的厨房小剧场",
    "titleEn": "Pineapple Day Kitchen Sketch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "圆润的三维女孩在厨房里与猫咪和菠萝互动，并完成日语对白。 原生视频规格为 15 秒 · 576×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rounded 3D girl interacts with a cat and pineapple in a kitchen while delivering Japanese dialogue. Native-video output: 15s · 576×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Spoonful_of_Fun/status/2089106173920522485",
    "sourceLabel": "X 原帖 · @Spoonful_of_Fun",
    "author": "一抹のFun",
    "publishedAt": "2026-08-16T21:45:00.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089106173920522485.mp4",
    "posterUrl": "/posters/x/2089106173920522485.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "576×576",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます。\n\n今日８月17日は「パイナップルの日」なんですね。\nということで今日もMiniMax H3で動画化。\n昨日の課題だったセリフは解決。色々と向上したような。\nただ、些細なことながら気になることが。膝や肘の関節、メカニカルな可動部を表現してほしくないんですけどねぇ。なかなか消えない。猫が人形ネコじゃなくってリアル猫だなーとか。\n\nまぁ正直言って平日はあまり手を入れられないんですけどね…",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 92,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption says the creator animated the scene with MiniMax H3 and specifically discusses the solved dialogue"
  },
  {
    "id": "x-2089104947162054735",
    "title": "参考图驱动的动漫角色动作",
    "titleEn": "Reference-Driven Anime Character Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "本地角色图与游戏语音作为参考，驱动动漫角色完成短动作与日语语音。 原生视频规格为 5 秒 · 896×1184。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A locally generated character image and game voice reference drive a short anime motion clip with Japanese speech. Native-video output: 5s · 896×1184. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kyle_catweasel/status/2089104947162054735",
    "sourceLabel": "X 原帖 · @kyle_catweasel",
    "author": "かいるー",
    "publishedAt": "2026-08-16T21:40:07.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089104947162054735.mp4",
    "posterUrl": "/posters/x/2089104947162054735.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "896×1184",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "動画生成楽しいな\n一晩中シャオルーン動画作って遊んでた\nillustrious＋シャオルーンLoraでローカル出力した画像とゲームボイスをリファレンスにComfyUIのMiniMax H3公式ref2vaテンプレートを使って生成\n省メモリ化拡張や高速化Loraは音声が不安定になるというかまともに日本語喋れなくなるから不使用",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 80,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly names the official MiniMax H3 Ref2VA ComfyUI template and its image and voice references"
  },
  {
    "id": "x-2089094854727119124",
    "title": "卧室里的连续舞步",
    "titleEn": "Two-Part Bedroom Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "蓝发水手服角色在固定机位中完成两段衔接舞步，头发与裙摆随动作自然摆动。 原生视频规格为 12 秒 · 608×1088。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A blue-haired sailor-uniform character performs two connected dance segments in a fixed bedroom shot, with natural hair and skirt motion. Native-video output: 12s · 608×1088. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "動画（MiniMax H3）1本目\n\nThe young anime woman with light blue hair and a white summer sailor uniform shown in <Picture 1> is dancing in place in her bedroom on a bright morning. She sways her hips from side to side and bounces lightly on her feet in rhythm, keeping both feet in the same spot on the wooden floor. Her long light blue hair and her navy pleated skirt swing with the motion, and she keeps looking back over her shoulder with a happy smile. She stays in the same position and does not walk anywhere. Static camera, fixed framing, full body visible, morning sunlight through the curtains. Quiet room ambience, no music.\n\n動画（MiniMax H3）2本目\n\nThe young anime woman with light blue hair and a white summer sailor uniform shown in <Picture 1> keeps dancing in place in her bedroom. She slowly raises both arms above her head and then lowers them again, twisting her upper body gently from side to side as she does. Her long light blue hair sways with the motion and her navy pleated skirt turns a little with her hips. She keeps both feet in the same spot on the wooden floor and glances back over her shoulder with a happy smile. Static camera, fixed framing, full body visible, morning sunlight through the curtains. Quiet room ambience, no music.",
    "sourceUrl": "https://x.com/Haruka1425054/status/2089094854727119124",
    "sourceLabel": "X 原帖 · @Haruka1425054",
    "author": "はるか🎀 Haruka",
    "publishedAt": "2026-08-16T21:00:01.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089094854727119124.mp4",
    "posterUrl": "/posters/x/2089094854727119124.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "608×1088",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "連休、おわっちゃったね。\n気合い入れていくよ〜\n\n鏡の前で いち、に、さん。 \n……よし、いこっか。 \n\nSFW anime-style illustration No nudity, no suggestive intent.\n\nMiniMax H3のプロンプト、こんな感じで作ったよ\n\n動画（MiniMax H3）1本目\n\nThe young anime woman with light blue hair and a white summer sailor uniform shown in <Picture 1> is dancing in place in her bedroom on a bright morning. She sways her hips from side to side and bounces lightly on her feet in rhythm, keeping both feet in the same spot on the wooden floor. Her long light blue hair and her navy pleated skirt swing with the motion, and she keeps looking back over her shoulder with a happy smile. She stays in the same position and does not walk anywhere. Static camera, fixed framing, full body visible, morning sunlight through the curtains. Quiet room ambience, no music.\n\n動画（MiniMax H3）2本目\n\nThe young anime woman with light blue hair and a white summer sailor uniform shown in <Picture 1> keeps dancing in place in her bedroom. She slowly raises both arms above her head and then lowers them again, twisting her upper body gently from side to side as she does. Her long light blue hair sways with the motion and her navy pleated skirt turns a little with her hips. She keeps both feet in the same spot on the wooden floor and glances back over her shoulder with a happy smile. Static camera, fixed framing, full body visible, morning sunlight through the curtains. Quiet room ambience, no music.\n\n動画の設定\n608×1088 / 141フレーム（約5.9秒）×2本\n24fps\n1本目の最終フレームを2本目の起点にして接続",
    "engagement": {
      "replies": 20,
      "reposts": 9,
      "likes": 53,
      "views": 1408,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption publishes both complete MiniMax H3 prompts and the reviewed frames match the described bedroom dance"
  },
  {
    "id": "x-2089080655128740221",
    "title": "北极熊步态双模型对照",
    "titleEn": "Polar Bear Gait Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一只雪地北极熊被并排用于观察 LTX 2.5 与 MiniMax H3 的速度和画面差异。 原生视频规格为 5 秒 · 2224×944。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A polar bear walking across snow is shown side by side to compare LTX 2.5 and MiniMax H3 speed and image quality. Native-video output: 5s · 2224×944. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ReyReinoso/status/2089080655128740221",
    "sourceLabel": "X 原帖 · @ReyReinoso",
    "author": "Rey Josias Reinoso",
    "publishedAt": "2026-08-16T20:03:36.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089080655128740221.mp4",
    "posterUrl": "/posters/x/2089080655128740221.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2224×944",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX 2.5 vs Minimax H3 in FHD  \nLeft: LTX 2.5 – generation time: 4m 29s\nRight: Minimax H3 – generation time: 18m 55s  \nFor simple prompts like this, LTX wins on both speed and quality.  For strong prompt understanding and creating new things → Minimax H3.",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 8,
      "views": 486,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly labels both models and gives generation times; reviewed frames confirm the side-by-side comparison"
  },
  {
    "id": "x-2089079820072169515",
    "title": "卧室巨蛛的惊悚转折",
    "titleEn": "A Giant Spider in the Bedroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "平静的卧室场景逐渐出现巨型蜘蛛，并以床上包围形成惊悚转折。 原生视频规格为 50 秒 · 1250×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A quiet bedroom scene escalates into a giant-spider reveal and a horror turn around the bed. Native-video output: 50s · 1250×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Spacedog19/status/2089079820072169515",
    "sourceLabel": "X 原帖 · @Spacedog19",
    "author": "Dan Thompson",
    "publishedAt": "2026-08-16T20:00:16.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089079820072169515.mp4",
    "posterUrl": "/posters/x/2089079820072169515.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "1250×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Spider, made with MiniMax H3.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 60,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption directly states the creator made Spider with MiniMax H3 and the reviewed frames confirm the generated sequence"
  },
  {
    "id": "x-2089040728609411541",
    "title": "冰白宫廷中的无声靠近",
    "titleEn": "Silent Closeness in an Ice-White Court",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "华服男女在冰白色宫廷布景中缓慢靠近，镜头强调衣领细节与唇部距离。 原生视频规格为 8 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two elaborately dressed figures move closer in an ice-white court as the camera emphasizes collar detail and the distance between their lips. Native-video output: 8s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The focus shifts from the ornate shimmer of his jacket collar to the delicate curve of her lips, emphasizing the unspoken dialogue in their closeness.",
    "sourceUrl": "https://x.com/ai_artworkgen/status/2089040728609411541",
    "sourceLabel": "X 原帖 · @ai_artworkgen",
    "author": "Marco",
    "publishedAt": "2026-08-16T17:24:56.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089040728609411541.mp4",
    "posterUrl": "/posters/x/2089040728609411541.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3 prompt:\n\n\"The focus shifts from the ornate shimmer of his jacket collar to the delicate curve of her lips, emphasizing the unspoken dialogue in their closeness.\"",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 11,
      "views": 784,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly labels and fully publishes the Hailuo H3 prompt; reviewed frames match its close-up staging"
  },
  {
    "id": "x-2089034591172690090",
    "title": "绿幕前的飞行待机",
    "titleEn": "Flying Standby on a Green Screen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿统一舞台服装的少女们在绿幕摄影棚中排队，并以对白呈现飞行待机片段。 原生视频规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Performers in matching stage outfits line up on a green-screen set for a dialogue-led flying-standby scene. Native-video output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mtArakawa/status/2089034591172690090",
    "sourceLabel": "X 原帖 · @mtArakawa",
    "author": "元荒川ちゃん❤",
    "publishedAt": "2026-08-16T17:00:33.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089034591172690090.mp4",
    "posterUrl": "/posters/x/2089034591172690090.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "flying standby\nトレーナーとスコートの舞台衣装って\nまだあるんでしょうか？\n\nと思って調べたら\n舞台チアの代名詞だったスクールメイツは\nまだ活動していました\nちょっとビックリ\n\nセリフはHailuoに丸投げ\n２番目の子のダウナーさは草\n\nNBPro + MiniMax H3\n#schoolgirl #flying #wireaction",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 1625,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly credits NBPro plus MiniMax H3 and states Hailuo supplied the dialogue"
  },
  {
    "id": "x-2089032084484714609",
    "title": "雨后发光种子的旅程",
    "titleEn": "The Glowing Seed After Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一只蓝色小鸟把雨后发现的发光种子带过屋顶，种进花盆并点亮街道。 原生视频规格为 15 秒 · 2208×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny blue bird carries a glowing seed across rain-wet rooftops, plants it, and lights up the street. Native-video output: 15s · 2208×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hmetouch/status/2089032084484714609",
    "sourceLabel": "X 原帖 · @hmetouch",
    "author": "Hemanth",
    "publishedAt": "2026-08-16T16:50:35.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2089032084484714609.mp4",
    "posterUrl": "/posters/x/2089032084484714609.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2208×1248",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Director Workflow(Muse) + ChatGPT Image2  Reference - Locally Generated with 4090+128GB RAM \n\nVideo Style: Luminous Digital Gouache\nShort Story: A tiny blue bird finds a glowing seed after the rain.\nIt carries it across quiet rooftops to an empty flowerpot.\nOne small drop of water wakes the seed.\nBy sunrise, the whole street is glowing. ✨\nPROMPT 👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 139,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption identifies a locally generated MiniMax H3 Director workflow and provides the complete short-story premise"
  },
  {
    "id": "x-2088967626760523804",
    "title": "四步加速下的高动态动作",
    "titleEn": "High Motion at Four Turbo Steps",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫角色以夸张幅度快速摆动，用来检验四步 Turbo LoRA 的运动能力。 原生视频规格为 4 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime character moves with exaggerated speed to test motion at four Turbo LoRA steps. Native-video output: 4s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ostrisai/status/2088967626760523804",
    "sourceLabel": "X 原帖 · @ostrisai",
    "author": "Ostris",
    "publishedAt": "2026-08-16T12:34:28.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088967626760523804.mp4",
    "posterUrl": "/posters/x/2088967626760523804.jpg",
    "duration": 4,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Current results at 4 step with the turbo time LoRA for MiniMax H3 I am training. Lack of movement is no longer an issue. Opposite.",
    "engagement": {
      "replies": 27,
      "reposts": 7,
      "likes": 153,
      "views": 7413,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption identifies current four-step MiniMax H3 Turbo LoRA results and the frames show the motion test"
  },
  {
    "id": "x-2088967109548056919",
    "title": "AMD 显卡上的工业变形",
    "titleEn": "Industrial Transformation on AMD",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "在 AMD RX 7900 XTX 上运行移植后的工作流，把工业设备参考转化成机器人形态。 原生视频规格为 5 秒 · 736×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A ported workflow on an AMD RX 7900 XTX transforms industrial-equipment reference imagery into a robot form. Native-video output: 5s · 736×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/txtnurcahyo/status/2088967109548056919",
    "sourceLabel": "X 原帖 · @txtnurcahyo",
    "author": "Wahyu Nurcahyo 🇮🇩@🇩🇪",
    "publishedAt": "2026-08-16T12:32:24.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088967109548056919.mp4",
    "posterUrl": "/posters/x/2088967109548056919.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "736×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 under Maestro AMD, yes AMD. Maestro in Pinokio @cocktailpeanut  supports only NVIDIA, but mine is AMD RX 7900 XTX, so I port it to AMD.\n\nI have tested for FLV and Ref like below, both working properly. But I tested only in my GPU.\n\nThe link below:",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 87,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption states MiniMax H3 was ported to AMD and that the displayed reference workflow was tested on the author’s GPU"
  },
  {
    "id": "x-2088966290765733908",
    "title": "中端显卡的两秒动漫动作",
    "titleEn": "A Two-Second Anime Motion on Midrange GPU",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "RTX 4060 Ti 本地生成的动漫角色短动作，用于展示较短生成时间下的画面质量。 原生视频规格为 2 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short anime character motion generated locally on an RTX 4060 Ti demonstrates output quality at a modest runtime. Native-video output: 2s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NAKA_X68K/status/2088966290765733908",
    "sourceLabel": "X 原帖 · @NAKA_X68K",
    "author": "nakataka",
    "publishedAt": "2026-08-16T12:29:09.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088966290765733908.mp4",
    "posterUrl": "/posters/x/2088966290765733908.jpg",
    "duration": 2,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 を ComfyUI で試してみました。環境は、RTX4060Ti 16G, メインメモリ 128GB 。生成時間は 2分40秒。２秒の動画ですが、高品質で生成時間も短くて、MiniMax H3 かなりいいですね。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 239,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption gives the MiniMax H3 ComfyUI hardware, generation time, and clip length; reviewed frames confirm the output"
  },
  {
    "id": "x-2088954932473074028",
    "title": "涂鸦彩色的刑警片头",
    "titleEn": "Scribbled-Color Detective Opening",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "真人刑警片头被处理成涂鸦彩色风格，手写文字与人物镜头快速交替。 原生视频规格为 10 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A live-action detective opening becomes a scribbled-color sequence with rapid alternation between handwritten text and character shots. Native-video output: 10s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ChiakiAkagi/status/2088954932473074028",
    "sourceLabel": "X 原帖 · @ChiakiAkagi",
    "author": "ChiakiAkagi 還暦のAIイリュージョニスト",
    "publishedAt": "2026-08-16T11:44:01.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088954932473074028.mp4",
    "posterUrl": "/posters/x/2088954932473074028.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dance",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "先日の踊る大捜査線のオープニング風AI映像を落書きカラーにしました。\nこの動画も｢編集｣していません。\nCapCutのMiniMax H3 で作りました。\nテンポアップする指定もしています。\n#capcut #CapCutCpp @capcutapp_jp",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 237,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption states the unedited sequence was created with CapCut MiniMax H3 and describes the tempo instruction"
  },
  {
    "id": "x-2088952829465727147",
    "title": "雪站长镜头的失败记录",
    "titleEn": "A Failed Long Take at a Snowy Station",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雪地车站长镜头保持人物和环境连贯，但末段列车意外退化成二维形态。 原生视频规格为 31 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long take at a snowy station keeps the person and setting coherent until the train collapses into a flat form near the end. Native-video output: 31s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PixelAigc/status/2088952829465727147",
    "sourceLabel": "X 原帖 · @PixelAigc",
    "author": "PixelAIGC",
    "publishedAt": "2026-08-16T11:35:40.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088952829465727147.mp4",
    "posterUrl": "/posters/x/2088952829465727147.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "一个失败的视频案例，最后几秒出现的火车成了二维的，minimax H3 本地部署挑战30秒。",
    "engagement": {
      "replies": 9,
      "reposts": 1,
      "likes": 36,
      "views": 7346,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly presents this as a failed locally deployed MiniMax H3 30-second challenge and names the visible failure"
  },
  {
    "id": "x-2088946165542953262",
    "title": "崩塌山体化作标题",
    "titleEn": "A Collapsing Mountain Becomes a Title",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "山崖崩落后，空中的岩石聚合成 DOMINION 字样，再随砂尘消散。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cliff collapses, airborne rocks assemble into the word DOMINION, and the title dissolves into dust. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shintaro_campon/status/2088946165542953262",
    "sourceLabel": "X 原帖 · @shintaro_campon",
    "author": "しんたろー△SNS30万｜個人開発",
    "publishedAt": "2026-08-16T11:09:11.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088946165542953262.mp4",
    "posterUrl": "/posters/x/2088946165542953262.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3の映像生成がヤバすぎる。\n\n崩壊する山が空中でタイトルロゴに変わる。\n\n・0:00から崖が崩落し土煙が舞う\n・0:08で空中の岩がDOMINIONに\n・0:11で風に流され砂塵に消える",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 302,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly attributes the generation to MiniMax H3 and provides a time-coded description matching the reviewed frames"
  },
  {
    "id": "x-2088943855320064442",
    "title": "日常动画风的片间标题",
    "titleEn": "Slice-of-Life Anime Eyecatch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女学生在夕照房间中停下书写并微笑，随后切入彩色手绘风标题卡。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A schoolgirl pauses her writing and smiles in a sunset-lit room before a colorful hand-drawn title card appears. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/zeroban_noriba/status/2088943855320064442",
    "sourceLabel": "X 原帖 · @zeroban_noriba",
    "author": "えきちょー（0番のりば）",
    "publishedAt": "2026-08-16T11:00:00.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088943855320064442.mp4",
    "posterUrl": "/posters/x/2088943855320064442.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3、10秒ポン出し。\n日常アニメのアイキャッチ風動画！\nちゃんと「0番のりば」って出た✨👀\n\n※えきちょーは出ません（ぇ）",
    "engagement": {
      "replies": 4,
      "reposts": 49,
      "likes": 169,
      "views": 3444,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly identifies a direct ten-second MiniMax H3 generation and the reviewed frames show the described eyecatch"
  },
  {
    "id": "x-2088942927036698921",
    "title": "香蕉创意会的三维喜剧",
    "titleEn": "The Banana Brainstorm in 3D",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绵羊般的办公室角色围绕香蕉图示开会，逐步把一根香蕉变成两根。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Sheep-like office characters hold a meeting around a banana diagram that changes from one banana to two. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Artedeingenio/status/2088942927036698921",
    "sourceLabel": "X 原帖 · @Artedeingenio",
    "author": "OscarAI",
    "publishedAt": "2026-08-16T10:56:19.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088942927036698921.mp4",
    "posterUrl": "/posters/x/2088942927036698921.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This 3D cartoon animation is so much fun, and it works really well with @Hailuo_AI MiniMax H3.",
    "engagement": {
      "replies": 9,
      "reposts": 5,
      "likes": 43,
      "views": 3579,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly credits Hailuo AI MiniMax H3 and describes the output as a 3D cartoon animation"
  },
  {
    "id": "x-2088935944896250356",
    "title": "群像镜头的多样性对照",
    "titleEn": "Crowd Diversity Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "相同群像提示在 MiniMax H3 与 LTX 2.5 中并列呈现，用于观察人物与构图多样性。 原生视频规格为 10 秒 · 1664×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same crowd-oriented setup is shown across MiniMax H3 and LTX 2.5 to compare people and compositional diversity. Native-video output: 10s · 1664×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PolaczekSagi/status/2088935944896250356",
    "sourceLabel": "X 原帖 · @PolaczekSagi",
    "author": "Sagi Polaczek 🦜",
    "publishedAt": "2026-08-16T10:28:34.000Z",
    "addedAt": "2026-08-17T02:39:30.456Z",
    "mediaUrl": "/media/x-2088935944896250356.mp4",
    "posterUrl": "/posters/x/2088935944896250356.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1664×640",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 vs LTX-2.5,\nDIVESITY TEST",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 20,
      "views": 1565,
      "snapshotAt": "2026-08-17T02:39:30.456Z"
    },
    "approvedAt": "2026-08-17T02:39:30.456Z",
    "editorialBasis": "sourceCaption explicitly labels this a MiniMax H3 versus LTX 2.5 diversity test; reviewed frames confirm the split comparison"
  },
  {
    "id": "x-2090370009507750070",
    "title": "皮肤、纸面与金属的连续变形",
    "titleEn": "Skin, Paper, and Metal Transformations",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "极近景皮肤、手绘版式和金属容器在同一组测试中改变尺度、视角、物体与空间。 原生视频规格为 37 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Macro skin, a hand-drawn layout, and a metal vessel shift scale, perspective, objects, and space in one test sequence. Native-video output: 37s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MauriceBourdon/status/2090370009507750070",
    "sourceLabel": "X 原帖 · @MauriceBourdon",
    "author": "Maurice Bourdon",
    "publishedAt": "2026-08-20T09:27:02.000Z",
    "addedAt": "2026-08-20T09:49:09.979Z",
    "mediaUrl": "/media/x-2090370009507750070.mp4",
    "posterUrl": "/posters/x/2090370009507750070.jpg",
    "duration": 37,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 - 4 more tests.\nChanging scale.\nChanging perspective.\nChanging objects.\nChanging space.\nStill prompt only.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 22,
      "snapshotAt": "2026-08-20T09:49:09.979Z"
    },
    "approvedAt": "2026-08-20T09:49:09.979Z",
    "editorialBasis": "caption names four prompt-only transformation tests; reviewed start, middle, and end frames confirm the stated changes"
  },
  {
    "id": "x-2090368569494340039",
    "title": "双色风车标记的旋转练习",
    "titleEn": "A Two-Color Pinwheel Motion Study",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙绿几何叶片在米色画布上旋转、错位并重新组合。 原生视频规格为 3 秒 · 608×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Orange and teal geometric blades rotate, offset, and regroup on a cream field. Native-video output: 3s · 608×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Toshi_nyaruo_AI/status/2090368569494340039",
    "sourceLabel": "X 原帖 · @Toshi_nyaruo_AI",
    "author": "Toshi@ニャルオAI",
    "publishedAt": "2026-08-20T09:21:18.000Z",
    "addedAt": "2026-08-20T09:49:09.980Z",
    "mediaUrl": "/media/x-2090368569494340039.mp4",
    "posterUrl": "/posters/x/2090368569494340039.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "608×352",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ローカルMiniMaX H3でモーショングラフィックスの練習。\n\nうーん、やりたいことは伝えたが今一創造性がないんだよね。\n\nどうした？Fable？やっぱり性能低下が激しいです。練もなんにもない。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 48,
      "snapshotAt": "2026-08-20T09:49:09.980Z"
    },
    "approvedAt": "2026-08-20T09:49:09.980Z",
    "editorialBasis": "caption identifies a local motion-graphics exercise; reviewed frames show the geometric mark changing orientation"
  },
  {
    "id": "x-2090366981664903407",
    "title": "猫耳女孩的意面动作对照",
    "titleEn": "Cat-Eared Diner Pasta Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "猫耳女孩在餐桌前尝试卷起意面，用于并列观察四种模型的手部与餐具动作。 原生视频规格为 7 秒 · 640×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared diner tries to twirl pasta, comparing utensil and hand motion across four models. Native-video output: 7s · 640×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2090366981664903407",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-08-20T09:15:00.000Z",
    "addedAt": "2026-08-20T09:49:09.980Z",
    "mediaUrl": "/media/x-2090366981664903407.mp4",
    "posterUrl": "/posters/x/2090366981664903407.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LTX2.5/H3/Wan3.0を試しています\n\n難しめの課題（パスタ勝負）です\n\n1本目: MiniMax H3\n2本目: LTX 2.5\n3本目: Wan 3.0\n4本目: FLUX3\n\nどのモデルもフォークに巻く動作はダメでした\n\nH3は焼うどんぽく見えます\nLTXはある意味一番潔いかも\nWan3.0はすくっただけで巻ける麺\nFLUXは麺がとても短い",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 61,
      "snapshotAt": "2026-08-20T09:49:09.980Z"
    },
    "approvedAt": "2026-08-20T09:49:09.980Z",
    "editorialBasis": "caption explicitly labels MiniMax H3 as the first of four pasta attempts; reviewed frames show the diner action"
  },
  {
    "id": "x-2090366297322475544",
    "title": "竹轮角色的舞台独舞",
    "titleEn": "Chikuwa Character Stage Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拟人竹轮从聚光灯下登场，随后在彩灯舞台上完成活泼独舞。 原生视频规格为 15 秒 · 1080×1888。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anthropomorphic chikuwa enters under a spotlight and performs an upbeat solo on a colorful stage. Native-video output: 15s · 1080×1888. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yukisiromeruto/status/2090366297322475544",
    "sourceLabel": "X 原帖 · @yukisiromeruto",
    "author": "雪白めると｜恋愛小説/AIアニメ",
    "publishedAt": "2026-08-20T09:12:17.000Z",
    "addedAt": "2026-08-20T09:49:09.980Z",
    "mediaUrl": "/media/x-2090366297322475544.mp4",
    "posterUrl": "/posters/x/2090366297322475544.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1888",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【ちくわくんダンスショー】\n#ちく推し　\nCapCut AIの MiniMax H3 を使って動画を制作しています。15秒貫通。\nこれは初めて絵コンテをぶん投げて作成した動画になります！\n作り方はレスに繋げます。\n@capcutapp_jp #capcutcpp",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 79,
      "snapshotAt": "2026-08-20T09:49:09.980Z"
    },
    "approvedAt": "2026-08-20T09:49:09.980Z",
    "editorialBasis": "caption explicitly credits CapCut MiniMax H3 and a storyboard; reviewed frames show the full stage dance"
  },
  {
    "id": "x-2090365462542799007",
    "title": "角色穿越多重画风",
    "titleEn": "A Character Crosses Multiple Visual Worlds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "同一角色从彩色动画画面进入阴暗废墟，再来到霓虹舞台。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "One character moves from colorful animation into dark ruins and then a neon stage. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ushizaru_LAB/status/2090365462542799007",
    "sourceLabel": "X 原帖 · @Ushizaru_LAB",
    "author": "Ushizaru / うしざる",
    "publishedAt": "2026-08-20T09:08:58.000Z",
    "addedAt": "2026-08-20T09:49:09.980Z",
    "mediaUrl": "/media/x-2090365462542799007.mp4",
    "posterUrl": "/posters/x/2090365462542799007.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n手持ちの画像を2枚並べた1枚絵で左から右へキャラを移動\nスタートフレーム指定生成です\nうほほほほ楽しぃぃ～💡",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 73,
      "snapshotAt": "2026-08-20T09:49:09.980Z"
    },
    "approvedAt": "2026-08-20T09:49:09.980Z",
    "editorialBasis": "caption states two images were combined for a start-frame generation; reviewed frames show the cross-world transition"
  },
  {
    "id": "x-2090363222658580535",
    "title": "狙击手的图形化人物揭晓",
    "titleEn": "Graphic Character Reveal for a Sniper",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "持枪角色在特写、剪影与排版卡之间切换，形成完整的人物介绍。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rifle-carrying character moves through close-ups, silhouettes, and typography cards in a complete reveal. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aoarashi_web/status/2090363222658580535",
    "sourceLabel": "X 原帖 · @aoarashi_web",
    "author": "べち【Bechi】",
    "publishedAt": "2026-08-20T09:00:04.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090363222658580535.mp4",
    "posterUrl": "/posters/x/2090363222658580535.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でキャラクター紹介風動画作成しました！\nもう少し練ればもっといい感じのできそう！\n\nCreated with DomoAI（ @DomoAI_)\n#DomoAI #DomoAICPP",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 54,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption explicitly identifies a MiniMax H3 character-introduction output; reviewed frames confirm the reveal structure"
  },
  {
    "id": "x-2090360926021308906",
    "title": "日式办公室的横向跟拍",
    "titleEn": "A Lateral Tracking Shot Through a Japanese Office",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "镜头横向穿过自然布置的日式办公室，记录员工、文件柜与工作区域。 原生视频规格为 5 秒 · 1024×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A lateral camera move crosses a naturally furnished Japanese office with staff, cabinets, and desks. Native-video output: 5s · 1024×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A Japanese Office scene. Tracking shot to the left.",
    "sourceUrl": "https://x.com/ITOKtw/status/2090360926021308906",
    "sourceLabel": "X 原帖 · @ITOKtw",
    "author": "ITOK",
    "publishedAt": "2026-08-20T08:50:56.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090360926021308906.mp4",
    "posterUrl": "/posters/x/2090360926021308906.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax-H3が日本への理解度が高いなと思うのは，\"A Japanese Office scene. Tracking shot to the left.\"というプロンプトだけで，こういう光景を出せるところです。\nきちんと指定しないと，木製の机とか畳が出てくる事が多いのですが，MiniMaxは違和感ないものが出てきます。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 185,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption publishes the complete one-line prompt; reviewed frames match the office tracking shot"
  },
  {
    "id": "x-2090359582032076832",
    "title": "暗黑圣堂的人物誓约",
    "titleEn": "A Character Vow in a Dark Sanctuary",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "黑衣角色在哥特圣堂、巨型标题和舞台式构图之间完成暗黑人物揭晓。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-clad character is revealed across a gothic sanctuary, giant typography, and stage-like compositions. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ZtohAic/status/2090359582032076832",
    "sourceLabel": "X 原帖 · @ZtohAic",
    "author": "ztoh🎱⑦❸⓽²7",
    "publishedAt": "2026-08-20T08:45:36.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090359582032076832.mp4",
    "posterUrl": "/posters/x/2090359582032076832.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "我靠，随便截了一张图，然后让它复刻这个视频，minimax h3 有点东西啊\n（提示词⬇️）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 109,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption states the creator used a captured frame to reproduce the clip; reviewed frames confirm the generated character sequence"
  },
  {
    "id": "x-2090355657849676251",
    "title": "深夜病房的长段落表演",
    "titleEn": "A Long Night-Hospital Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画角色从走廊时钟与自动售货机进入病房，展开一段连续夜间表演。 原生视频规格为 46 秒 · 1874×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated character moves from a corridor clock and vending machine into a hospital room for a continuous night scene. Native-video output: 46s · 1874×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shonai_arai/status/2090355657849676251",
    "sourceLabel": "X 原帖 · @shonai_arai",
    "author": "障ナイさん",
    "publishedAt": "2026-08-20T08:30:00.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090355657849676251.mp4",
    "posterUrl": "/posters/x/2090355657849676251.jpg",
    "duration": 46,
    "aspectRatio": "landscape",
    "resolution": "1874×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3がプロンプトにどれぐらい従ってくれるのかテストしてみたのだ\nモデルとLoRA少しずつ変えてるけどキャラの見た目に関しては全然間違えないのだ\n\n見た目と動きの検証だからセリフはあえて指定してないのだ\n\n少なくともSora2よりは命令通りに動いてくれるのだ\nSeedVR2でアップスケールしたのだ",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 3,
      "views": 130,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption describes a long prompt-following and appearance-consistency test; reviewed frames show the hospital sequence"
  },
  {
    "id": "x-2090349415534035047",
    "title": "和风甜点品牌的动态图形",
    "titleEn": "Japanese Dessert Brand Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "麦茶、夏祭与蛋糕店等和风品牌卡片以图形转场快速串联。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Japanese-style brand cards for barley tea, a summer festival, and a cake shop connect through graphic transitions. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/seiiiiiiiiiiru/status/2090349415534035047",
    "sourceLabel": "X 原帖 · @seiiiiiiiiiiru",
    "author": "SEIIIRU😈動画生成AIを使う映像クリエイター",
    "publishedAt": "2026-08-20T08:05:12.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090349415534035047.mp4",
    "posterUrl": "/posters/x/2090349415534035047.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "[ Minimax H3 ] -AI Motion Graphics-\nデスクトップアプリのMiniMaxDesignで作成。\n\nどうすすめるか👇\n#MiniMaxH3 #MiniMaxDesign",
    "engagement": {
      "replies": 3,
      "reposts": 14,
      "likes": 72,
      "views": 3026,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption explicitly identifies a MiniMax H3 motion-graphics output; reviewed frames show the branded sequence"
  },
  {
    "id": "x-2090344867184451786",
    "title": "显卡广告的强力排版",
    "titleEn": "Power Typography for a Graphics Card Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "人物进入重复文字空间，随后以显卡特写和巨型口号完成产品揭晓。 原生视频规格为 15 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure enters a tunnel of repeated words before graphics-card close-ups and a giant slogan reveal the product. Native-video output: 15s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/amadeus_NFT/status/2090344867184451786",
    "sourceLabel": "X 原帖 · @amadeus_NFT",
    "author": "Amadeus | Latent Art",
    "publishedAt": "2026-08-20T07:47:07.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090344867184451786.mp4",
    "posterUrl": "/posters/x/2090344867184451786.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Taking it to the next level! 🚀\n\nI'm continuing to push Kinetic Typography in MiniMax H3, but moving from pure art to actual advertising. Built a full concept ad using mograph, and since local generation is life, the product had to be the RTX 5090.\n\nIntegrating text and motion graphics into a commercial workflow with this model is next-level. The results speak for themselves. 🔥\n\nHere's the prompt 👇\n\n#MiniMax #MotionDesign #AIAdvertising #ComfyUI #TechAd",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 74,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption explicitly describes a locally generated RTX 5090 kinetic-typography commercial; reviewed frames match it"
  },
  {
    "id": "x-2090344857957286046",
    "title": "从街头走进百货专柜",
    "titleEn": "From the Street Into a Department Store",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "顾客从街头画面切换到商品网页，最终在百货专柜与店员交流。 原生视频规格为 20 秒 · 720×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A shopper moves from a street scene to a product page and finally speaks with staff at a department-store counter. Native-video output: 20s · 720×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cwh_tw/status/2090344857957286046",
    "sourceLabel": "X 原帖 · @cwh_tw",
    "author": "一定iPad溫開水",
    "publishedAt": "2026-08-20T07:47:05.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090344857957286046.mp4",
    "posterUrl": "/posters/x/2090344857957286046.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "720×960",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MacBook 裝了 MiniMax H3 ，嘗試第一次下海幫我客戶拍廣告《隨時隨地都像親逛百貨》（？\n\n這影片只來回兩三次 prompts ，幾張 ref 圖。比起之前鳥鳥的 LTX2.3 WAN2.2，讓人真正覺得：地端終於可行了\n\n強國再這樣開源下去，讓便宜的地端真的能做正事而不是玩具，米國大公司壓力會越來越大",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 217,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption states the creator used a few prompts and reference images for a client ad; reviewed frames confirm the journey"
  },
  {
    "id": "x-2090343936560963779",
    "title": "黄昏栏杆边的动漫人物",
    "titleEn": "An Anime Figure at a Sunset Railing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发动漫人物站在黄昏栏杆旁，头发与衣摆在风中保持细微运动。 原生视频规格为 24 秒 · 1280×712。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired anime figure stands by a sunset railing with subtle wind in the hair and clothing. Native-video output: 24s · 1280×712. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DeBug_lzy/status/2090343936560963779",
    "sourceLabel": "X 原帖 · @DeBug_lzy",
    "author": "DeBug",
    "publishedAt": "2026-08-20T07:43:25.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090343936560963779.mp4",
    "posterUrl": "/posters/x/2090343936560963779.jpg",
    "duration": 24,
    "aspectRatio": "landscape",
    "resolution": "1280×712",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "研究了一天 终于本地跑通MiniMax h3 电脑性能不够 超分后视频流畅度和清晰度显著上升",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 35,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption identifies a locally run MiniMax H3 output and an upscale pass; reviewed frames show the consistent shot"
  },
  {
    "id": "x-2090331284564476354",
    "title": "纸艺火山吞没史前世界",
    "titleEn": "A Paper Volcano Devours a Prehistoric World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "宁静纸艺史前景观逐渐喷发，熔岩与浓烟最终覆盖山谷。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A peaceful paper-crafted prehistoric landscape erupts as lava and smoke consume the valley. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2090331284564476354",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-20T06:53:09.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090331284564476354.mp4",
    "posterUrl": "/posters/x/2090331284564476354.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I made this prehistoric volcano sequence with MiniMax H3 on @Hailuo_AI\n\nOne prompt turns a peaceful paper world into a full volcanic disaster in just 15 seconds\n\nPrompt 👇🏻",
    "engagement": {
      "replies": 15,
      "reposts": 0,
      "likes": 27,
      "views": 595,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption identifies a one-prompt MiniMax H3 sequence; reviewed frames show the volcanic escalation"
  },
  {
    "id": "x-2090330350668947525",
    "title": "红宝石女王变身战士",
    "titleEn": "Ruby Queen Transforms Into a Warrior",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "红宝石女王在哥特王座厅中被红光包围，礼服变成盔甲并举起剑盾。 原生视频规格为 15 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A ruby queen is engulfed in red magic inside a gothic throne room, transforming her gown into armor and raising sword and shield. Native-video output: 15s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic fantasy transformation beginning with a majestic queen wearing an intricate ruby-embroidered corset gown, matching crown, and holding a glowing heart-topped scepter inside a gothic cathedral throne room. The camera zooms out to show her full attire as she turns and walks toward a grand staircase, where a burst of glowing red magical light erupts around her. Enveloped in the crimson energy, her gown transforms into sleek, ornate metallic warrior armor, while her scepter morphs into a longsword and an embossed shield. She turns back toward the camera in a powerful battle stance, positioned before a glowing red arcane circle on the cathedral steps.",
    "sourceUrl": "https://x.com/TaliaAariz/status/2090330350668947525",
    "sourceLabel": "X 原帖 · @TaliaAariz",
    "author": "Talia",
    "publishedAt": "2026-08-20T06:49:26.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090330350668947525.mp4",
    "posterUrl": "/posters/x/2090330350668947525.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created with MiniMax H3\n\nPrompt:\n\nA cinematic fantasy transformation beginning with a majestic queen wearing an intricate ruby-embroidered corset gown, matching crown, and holding a glowing heart-topped scepter inside a gothic cathedral throne room. The camera zooms out to show her full attire as she turns and walks toward a grand staircase, where a burst of glowing red magical light erupts around her. Enveloped in the crimson energy, her gown transforms into sleek, ornate metallic warrior armor, while her scepter morphs into a longsword and an embossed shield. She turns back toward the camera in a powerful battle stance, positioned before a glowing red arcane circle on the cathedral steps.",
    "engagement": {
      "replies": 10,
      "reposts": 0,
      "likes": 19,
      "views": 321,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption publishes the complete transformation prompt; reviewed frames match the queen-to-warrior sequence"
  },
  {
    "id": "x-2090325586736738393",
    "title": "快餐店里追逐镜头的猫",
    "titleEn": "A Cat Chases the Camera Through a Diner",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "镜头穿过快餐店后厨与用餐区，猫咪一路奔跑并追随移动视角。 原生视频规格为 7 秒 · 896×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera travels through a diner kitchen and seating area while a cat runs after the moving viewpoint. Native-video output: 7s · 896×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mathewdoeslife/status/2090325586736738393",
    "sourceLabel": "X 原帖 · @Mathewdoeslife",
    "author": "Mathew Youssef",
    "publishedAt": "2026-08-20T06:30:30.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090325586736738393.mp4",
    "posterUrl": "/posters/x/2090325586736738393.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "896×672",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "food",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I spent all day taking photos of random things and sending them to Minimax H3 to make cat videos…",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 66,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption says the creator photographed everyday objects and used MiniMax H3 to make cat clips; reviewed frames show this result"
  },
  {
    "id": "x-2090324150162473271",
    "title": "韩国意象的月夜长卷",
    "titleEn": "A Moonlit Panorama of Korean Motifs",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水墨山景、宫殿纹样与月夜城门通过红蓝飘带串联成文化主题片。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Ink mountains, palace motifs, and a moonlit gate connect through red and blue ribbons in a cultural motion piece. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/snowy_mandu/status/2090324150162473271",
    "sourceLabel": "X 原帖 · @snowy_mandu",
    "author": "snowy_mandu",
    "publishedAt": "2026-08-20T06:24:48.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090324150162473271.mp4",
    "posterUrl": "/posters/x/2090324150162473271.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "KOREA - TIMELESS HARMONY\n\nThis is a 15-second motion graphics test video with the theme of Korea using MINIMAX H3.\n@newtake_kr \n*이 영상은 newtake의 지원을 받아 제작되었습니다.\n\n#newtakecrew #newtake #motiongraphicsdesign #minimaxh3 #모션그래픽",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 45,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption explicitly describes a Korea-themed MiniMax H3 motion-graphics test; reviewed frames confirm the sequence"
  },
  {
    "id": "x-2090322678292906110",
    "title": "粉色舞台上的角色宣言",
    "titleEn": "A Character Statement on a Pink Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "自绘角色在粉色舞台与巨型霓虹文字之间舞动，形成短片式人物表达。 原生视频规格为 15 秒 · 832×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A self-illustrated character dances between a pink stage and giant neon words in a compact character piece. Native-video output: 15s · 832×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/xiaomimasato/status/2090322678292906110",
    "sourceLabel": "X 原帖 · @xiaomimasato",
    "author": "のえみん",
    "publishedAt": "2026-08-20T06:18:57.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090322678292906110.mp4",
    "posterUrl": "/posters/x/2090322678292906110.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "832×640",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "illustration"
    ],
    "scenes": [
      "product",
      "music-performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "プロンプトをそのままお借りして\n自作イラストにて\nMiniMax H3でI2V作成してみました\n\nこれは\n音源を先に用意して、15秒以内分割し\nR2Vで作成したものを繋げれば\nPVが出来上がりそう！",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 138,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption states a self-made illustration was used for MiniMax H3 I2V; reviewed frames show the reference-led character"
  },
  {
    "id": "x-2090322529982590994",
    "title": "低音演奏撕开地下海报",
    "titleEn": "Bass Performance Tears Through a Poster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "真实贝斯演奏与复印纹理、撕纸图形交替，最终形成地下海报人物卡。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A live bass performance alternates with photocopy textures and torn-paper graphics before resolving into an underground poster. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2090322529982590994",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-20T06:18:22.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090322529982590994.mp4",
    "posterUrl": "/posters/x/2090322529982590994.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I turned a bass performance into an animated underground poster 🎸⚡\n\nCreated using @Hailuo_AI's MiniMax H3 on @higgsfield_ai @higgsfield \n\nMeet MARA.\n\nCobalt ink.\nPhotocopied textures.\nTorn paper.\nMisregistered colors.\n\nEvery bass note bends the graphic world around her.\nThen the final note hits…\n\nand the whole poster snaps back into perfect alignment.\n\nRaw bass + analog print energy.\n\nFull video below 👇",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 5,
      "views": 281,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption explicitly credits Hailuo MiniMax H3 and describes the poster transformation; reviewed frames match it"
  },
  {
    "id": "x-2090321133254086935",
    "title": "向日葵从种子到盛放",
    "titleEn": "A Sunflower Grows From Seed to Bloom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "种子落入花盆后发芽、长叶并开出完整向日葵。 原生视频规格为 48 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A seed drops into a flowerpot, sprouts, grows leaves, and opens into a full sunflower. Native-video output: 48s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2090321133254086935",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-20T06:12:49.000Z",
    "addedAt": "2026-08-20T09:49:09.981Z",
    "mediaUrl": "/media/x-2090321133254086935.mp4",
    "posterUrl": "/posters/x/2090321133254086935.jpg",
    "duration": 48,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A tiny sunflower journey from fertile soil to full bloom 🌻\n\nMade with Nano Banana 2× MiniMax H3",
    "engagement": {
      "replies": 8,
      "reposts": 1,
      "likes": 17,
      "views": 385,
      "snapshotAt": "2026-08-20T09:49:09.981Z"
    },
    "approvedAt": "2026-08-20T09:49:09.981Z",
    "editorialBasis": "caption explicitly credits MiniMax H3 for the sunflower journey; reviewed frames show the full growth arc"
  },
  {
    "id": "x-2090315119356244222",
    "title": "黑洞王冠的科幻片头",
    "titleEn": "The Void Crown Sci-Fi Opening",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "轨道线稿、恒星爆发与黑洞标题卡组成冷峻科幻片头。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Orbital diagrams, a stellar burst, and a black-hole title card form a stark sci-fi opening. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rovvmut_/status/2090315119356244222",
    "sourceLabel": "X 原帖 · @rovvmut_",
    "author": "Heisenberg",
    "publishedAt": "2026-08-20T05:48:55.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090315119356244222.mp4",
    "posterUrl": "/posters/x/2090315119356244222.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "fashion",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Crafted this dark, high-concept sci-fi intro sequence using MiniMax H3.\n\nReally impressed by how it handled the crisp art typography and seamless motion graphics throughout the scene!\n\nGet Unlimited #MiniMaxH3 on Runway",
    "engagement": {
      "replies": 17,
      "reposts": 4,
      "likes": 36,
      "views": 858,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption explicitly identifies a MiniMax H3 sci-fi intro; reviewed frames confirm typography and motion graphics"
  },
  {
    "id": "x-2090313408256336198",
    "title": "极光啤酒的海边口播广告",
    "titleEn": "A Seaside Testimonial for Aurora Beer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "人物在海边展示极光啤酒罐，并以产品定帧完成口播式广告。 原生视频规格为 10 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A presenter holds an Aurora beer can by the water before a clean product-pack shot closes the testimonial ad. Native-video output: 10s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotogenicWeekE/status/2090313408256336198",
    "sourceLabel": "X 原帖 · @PhotogenicWeekE",
    "author": "Photogenic Weekend",
    "publishedAt": "2026-08-20T05:42:07.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090313408256336198.mp4",
    "posterUrl": "/posters/x/2090313408256336198.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3、遊んでてw試してなかった広告で使えるか？ポイントはリファレンス画像(商品)の一貫性。Turbo LoRAありでこれなら良いかも！？但し、Latent Upscalerの欠点発見。音が微妙に変わる(ノイズだったりイントネーションだったり)。これはUpscale前のを被すしか手は無さそう…。ビール画像とシナリオはChatGPT(笑)",
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 30,
      "views": 2273,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption identifies an H3 product-consistency advertising test using a beer reference image; reviewed frames match it"
  },
  {
    "id": "x-2090306968900784383",
    "title": "游船途中遇见海豚",
    "titleEn": "Dolphins Appear Beside a Sightseeing Boat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "女游客登上游船、坐稳出港，并在航行中发现伴随船尾的海豚。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A traveler boards a sightseeing boat, settles in, and spots dolphins swimming beside the wake. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A young Western female traveler in her mid-20s walks naturally along a wooden marina dock toward a small white sightseeing boat. She wears a light blue windbreaker, white shirt, dark trousers, comfortable sneakers, sunglasses, a properly fitted life jacket, and a small crossbody bag. She carefully steps from the dock onto the gently rocking boat while holding the handrail, naturally shifting her weight as the boat responds to her movement. She walks toward the rear passenger bench, sits down, adjusts her position, and looks toward the open water. The boat engine starts with a realistic low rumble and gradually leaves the marina, creating a small bow wave and narrow wake. As the boat moves into open water, she notices movement nearby, looks surprised, and naturally says, \"Wait... are those dolphins?\" Two dolphins surface several meters beside the moving boat, swimming parallel at a safe distance with realistic tail propulsion, surfacing, diving, breathing, and water displacement. She picks up her phone with both hands and records them while smiling with genuine excitement. One dolphin briefly surfaces near the rear side of the boat, exhales through its blowhole, creates a small burst of mist and droplets, then dives beneath the water and leaves expanding ripples behind. She lowers her phone and says, \"That is incredible.\" Ultra-photorealistic marine wildlife travel documentary, cinematic realism, natural human behavior, realistic walking and sitting biomechanics, physically accurate boat buoyancy and acceleration, authentic ocean waves and wake, realistic dolphin anatomy and swimming behavior, natural water reflections and refraction, accurate English lip synchronization, bright neutral daytime lighting, realistic coastal wind, immersive marina and ocean ambience, smooth documentary camera movement, continuous visual consistency, seamless storytelling from walking to boarding, departure, dolphin encounter, recording, and final reaction, no narration, no CGI appearance, no plastic skin, no robotic movement, no exaggerated dolphin jumps, no impossible water behavior, no sudden camera transitions, no text overlays, logos, or watermarks.",
    "sourceUrl": "https://x.com/nawalsehar/status/2090306968900784383",
    "sourceLabel": "X 原帖 · @nawalsehar",
    "author": "Nawal",
    "publishedAt": "2026-08-20T05:16:32.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090306968900784383.mp4",
    "posterUrl": "/posters/x/2090306968900784383.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A perfect day on the water takes an unexpected turn. From boarding the boat to spotting dolphins beside the wake, every moment feels natural and immersive.\n\nCreated with MiniMax h3 \n\nPrompt:A young Western female traveler in her mid-20s walks naturally along a wooden marina dock toward a small white sightseeing boat. She wears a light blue windbreaker, white shirt, dark trousers, comfortable sneakers, sunglasses, a properly fitted life jacket, and a small crossbody bag. She carefully steps from the dock onto the gently rocking boat while holding the handrail, naturally shifting her weight as the boat responds to her movement. She walks toward the rear passenger bench, sits down, adjusts her position, and looks toward the open water. The boat engine starts with a realistic low rumble and gradually leaves the marina, creating a small bow wave and narrow wake. As the boat moves into open water, she notices movement nearby, looks surprised, and naturally says, \"Wait... are those dolphins?\" Two dolphins surface several meters beside the moving boat, swimming parallel at a safe distance with realistic tail propulsion, surfacing, diving, breathing, and water displacement. She picks up her phone with both hands and records them while smiling with genuine excitement. One dolphin briefly surfaces near the rear side of the boat, exhales through its blowhole, creates a small burst of mist and droplets, then dives beneath the water and leaves expanding ripples behind. She lowers her phone and says, \"That is incredible.\" Ultra-photorealistic marine wildlife travel documentary, cinematic realism, natural human behavior, realistic walking and sitting biomechanics, physically accurate boat buoyancy and acceleration, authentic ocean waves and wake, realistic dolphin anatomy and swimming behavior, natural water reflections and refraction, accurate English lip synchronization, bright neutral daytime lighting, realistic coastal wind, immersive marina and ocean ambience, smooth documentary camera movement, continuous visual consistency, seamless storytelling from walking to boarding, departure, dolphin encounter, recording, and final reaction, no narration, no CGI appearance, no plastic skin, no robotic movement, no exaggerated dolphin jumps, no impossible water behavior, no sudden camera transitions, no text overlays, logos, or watermarks.",
    "engagement": {
      "replies": 52,
      "reposts": 3,
      "likes": 166,
      "views": 4182,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption publishes the complete prompt and credits MiniMax H3; reviewed frames match the marina-to-dolphin story"
  },
  {
    "id": "x-2090306302966919515",
    "title": "橙黑街舞品牌快切广告",
    "titleEn": "Orange-and-Black Street Dance Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "街舞人物在橙黑几何排版中完成快切动作，以巨型日文字卡收束。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A street dancer performs through rapid orange-and-black geometric layouts before a giant Japanese title closes the spot. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "----------------------\nintegrated_multimodal_description:\n2D-animated, a bold graphic-design motion-graphics commercial in 16:9 with exactly 13\ndistinct cuts at 24fps. This film is 85% kinetic graphic design and 15% character action:\ngiant kinetic Japanese typography, hard-edged flat shapes, split screens, halftone, speed\nlines, torn-paper reveals and shutter flashes. Palette, strictly four colors: fluorescent\norange, deep black, off-white, silver grey. The recurring character is a young anime street\ndancer with a black chin-length bob and blunt bangs, sharp confident eyes, an oversized\noff-white track jacket with fluorescent-orange sleeve stripes, black cargo shorts, and\nchunky white sneakers with fluorescent-orange lines. Preserve her exact face, proportions,\nhairstyle, outfit, materials and colors throughout; never redesign her. Every graphic\nelement moves fast and snaps hard on the beat.\n\n[Shot 1] On a deep black field, a massive off-white circle slams into frame left of center,\nthen two fluorescent-orange bars wipe across the upper and lower thirds, then the Japanese\nkanji \"烈\" builds itself stroke by stroke in off-white at the center until it is complete.\nThe camera shakes slightly with small amplitude at fast speed on each impact.\n\n[Shot 2] At 00:01.000, the camera cuts to an off-white field where four black squares\nrotate into the center one at a time. The largest square opens as a porthole revealing an\nextreme close-up of the dancer's ice-sharp eye glancing up, then a red-and-cyan split flash\ntears across the edges and the squares shatter into flat shards. The camera pushes in with\nsmall amplitude at fast speed.\n\n[Shot 3] At 00:02.000, the camera cuts to an off-white field filled by an enormous black\nkanji \"烈\". The dancer sprints in from frame left, power-slides along the baseline of the\ntypography, and fluorescent-orange ink splatter trails behind her as speed lines streak\npast. The camera trucks left with large amplitude at fast speed. The kanji stays fully\nreadable before she crosses in front of it.\n\n[Shot 4] At 00:03.100, the camera cuts to a vertical triptych of fluorescent-orange, deep\nblack and off-white panels. Her flat silhouette appears in each panel in stroboscopic\nfreeze, one after another — takeoff, mid-air tuck, landing — while giant vertical silver\ntext \"全速\" scrolls upward behind the panels. Static shot.\n\n[Shot 5] At 00:04.000, the camera cuts to a black field where a giant tilted ring of\nfluorescent-orange Japanese characters reading \"止マルナ\" rotates. The dancer leaps through\nthe center of the ring, spins once in mid-air, and the characters break off and scatter\noutward as flat shards. The camera arcs with large amplitude at fast speed, then snap-zooms\nonto her confident face.\n\n[Shot 6] At 00:05.100, the camera cuts to a clean off-white editorial card holding a huge\nblack kanji \"跳\" with one orange diagonal slash. She vaults over the word itself, plants her\npalm on its left radical, the character compresses like a spring under her hand, then\nrebounds as her legs whip across the frame. Static shot.\n\n[Shot 7] At 00:06.000, the camera cuts to a full fluorescent-orange field crossed by one\nthick black diagonal stripe. She back-flips along the stripe in three stroboscopic ghost\nframes tinted black, off-white and silver grey, while a large outlined kanji \"回\" rotates\n180 degrees in sync with her rotation. The camera rolls clockwise with small amplitude at\nfast speed.\n\n[Shot 8] At 00:07.000, the camera cuts to an off-white field with a hard black band. Four\nbold black kanji slam in one at a time at different scales and angles — \"速\", then \"鋭\",\nthen \"静\", then \"烈\" — each landing with a white shutter flash and an orange impact mark,\nwhile she slides on her knees across the foreground, jacket flaring. The camera shakes\nslightly with small amplitude at fast speed.\n\n[Shot 9] At 00:08.000, the camera cuts to a black frame where a white wireframe grid tilts\naway in 3D. She runs up the grid like a wall, kicks off it, and hangs frozen in mid-air as\na fluorescent-orange circle stamps around her pose like a target lock with silver UI\nbrackets and tick marks. The camera tilts up with small amplitude at slow speed.\n\n[Shot 10] At 00:09.000, the freeze releases and the camera cuts to layered flat-color panes\nstacked toward the lens. She dives straight at the camera, the panes tear open one at a\ntime like thick paper, and each tear reveals a larger portion of the black kanji \"烈\"\nbehind, before her sneaker wipes across the foreground. The camera pulls out with large\namplitude at fast speed.\n\n[Shot 11] At 00:10.100, the camera cuts to a rapid-fire montage of four full-screen graphic\nposters of her in different action poses — mid-flip, sliding, landing, hands in pockets —\nsnapping past with hard cuts, each with an oversized silver number 01 to 04, a black\nbarcode strip and thick diagonal slashes. Static shot.\n\n[Shot 12] At 00:11.100, the camera cuts to a clean off-white cyclorama for the hero moment.\nShe lands a final backflip dead center in slow motion, straightens with both hands in her\npockets, and a shockwave of concentric fluorescent-orange rings blasts outward from her feet\ncarrying shattered fragments of the kanji \"烈\" with it, then the frame overexposes toward\nwhite. The camera pushes in with small amplitude at slow speed.\n\n[Shot 13] At 00:13.000, the camera cuts to the final identity card on a pale off-white\nfield. An enormous black kanji \"烈\" dominates the frame with translucent orange rings and\nthin technical arcs behind it, the silver-grey word \"RETSU\" sits directly beneath the kanji,\nand she stands relaxed at the lower center overlapping the bottom of the strokes with her\njacket rippling. One thin orange pulse sweeps horizontally through the typography and holds\ncleanly until exactly 15.000 seconds. Static shot.\n\nEditing: extremely aggressive rhythm — hard cuts on every beat, graphic matches, whip pans,\nsnap zooms, stroboscopic freezes, foreground wipes, split flashes and impact shakes. Every\ncut must feel compositionally different. All Japanese typography is always fully readable\nbefore the character overlaps it. No weapons, no combat, no fire, no explosions; all energy\ncomes from motion design, ink, torn paper, wind and her street-dance athleticism. No\nphotorealism, no 3D rendering, no gradients, no browser interface, no player controls, no\nsubtitles, no captions, and no text other than the Japanese characters named above.\nExactly 13 cuts.\n\noverall_soundscape:\nSneaker impacts, knee slides and landings hit hard against a clean room tone. Ink splatters\nburst, thick paper tears open, and flat panels shatter like glass. Sharp whooshes carry the\ncolor wipes and typography slams, with small shutter clicks on each flash. No dialogue.\n\nnon_diegetic_music:\nAn original 15-second drum-heavy breakbeat with a deep sub bass, hard snare hits, risers and\nglitch fills locked to every cut. Begin with a single impact, build tightly through the\nmiddle, peak at the landing at 00:11.100, and end with a cold electronic logo sting that\nresolves at 15.000 seconds.",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2090306302966919515",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-20T05:13:53.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090306302966919515.mp4",
    "posterUrl": "/posters/x/2090306302966919515.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Claude CodeにMiniMax H3のプロンプトを丸投げしたら、この動画ができました😃\n\nClaude Codeに投げたプロンプト↓\n-------------------\nMiniMax H3を使ってモーショングラフィックスをバリバリ使った15秒のCMを作りたい。\nどんな内容が良いと思う？\n条件は以下\n・モーショングラフィックスを多用\n・人物も登場させる（実写・アニメーションは問わない）\n・アスペクト比は16:9\n-------------------\n\nすると、Claude Codeが3つCM案を提案。\nそのうちの1つを選んだら↓のプロンプトができました。\nMiniMax H3を利用したツールはMiniMax Design。\n#MiniMaxH3 on #MiniMaxDesign\n\nClaude Codeが作ったMiniMax H3向けのプロンプト↓\n----------------------\nintegrated_multimodal_description:\n2D-animated, a bold graphic-design motion-graphics commercial in 16:9 with exactly 13\ndistinct cuts at 24fps. This film is 85% kinetic graphic design and 15% character action:\ngiant kinetic Japanese typography, hard-edged flat shapes, split screens, halftone, speed\nlines, torn-paper reveals and shutter flashes. Palette, strictly four colors: fluorescent\norange, deep black, off-white, silver grey. The recurring character is a young anime street\ndancer with a black chin-length bob and blunt bangs, sharp confident eyes, an oversized\noff-white track jacket with fluorescent-orange sleeve stripes, black cargo shorts, and\nchunky white sneakers with fluorescent-orange lines. Preserve her exact face, proportions,\nhairstyle, outfit, materials and colors throughout; never redesign her. Every graphic\nelement moves fast and snaps hard on the beat.\n\n[Shot 1] On a deep black field, a massive off-white circle slams into frame left of center,\nthen two fluorescent-orange bars wipe across the upper and lower thirds, then the Japanese\nkanji \"烈\" builds itself stroke by stroke in off-white at the center until it is complete.\nThe camera shakes slightly with small amplitude at fast speed on each impact.\n\n[Shot 2] At 00:01.000, the camera cuts to an off-white field where four black squares\nrotate into the center one at a time. The largest square opens as a porthole revealing an\nextreme close-up of the dancer's ice-sharp eye glancing up, then a red-and-cyan split flash\ntears across the edges and the squares shatter into flat shards. The camera pushes in with\nsmall amplitude at fast speed.\n\n[Shot 3] At 00:02.000, the camera cuts to an off-white field filled by an enormous black\nkanji \"烈\". The dancer sprints in from frame left, power-slides along the baseline of the\ntypography, and fluorescent-orange ink splatter trails behind her as speed lines streak\npast. The camera trucks left with large amplitude at fast speed. The kanji stays fully\nreadable before she crosses in front of it.\n\n[Shot 4] At 00:03.100, the camera cuts to a vertical triptych of fluorescent-orange, deep\nblack and off-white panels. Her flat silhouette appears in each panel in stroboscopic\nfreeze, one after another — takeoff, mid-air tuck, landing — while giant vertical silver\ntext \"全速\" scrolls upward behind the panels. Static shot.\n\n[Shot 5] At 00:04.000, the camera cuts to a black field where a giant tilted ring of\nfluorescent-orange Japanese characters reading \"止マルナ\" rotates. The dancer leaps through\nthe center of the ring, spins once in mid-air, and the characters break off and scatter\noutward as flat shards. The camera arcs with large amplitude at fast speed, then snap-zooms\nonto her confident face.\n\n[Shot 6] At 00:05.100, the camera cuts to a clean off-white editorial card holding a huge\nblack kanji \"跳\" with one orange diagonal slash. She vaults over the word itself, plants her\npalm on its left radical, the character compresses like a spring under her hand, then\nrebounds as her legs whip across the frame. Static shot.\n\n[Shot 7] At 00:06.000, the camera cuts to a full fluorescent-orange field crossed by one\nthick black diagonal stripe. She back-flips along the stripe in three stroboscopic ghost\nframes tinted black, off-white and silver grey, while a large outlined kanji \"回\" rotates\n180 degrees in sync with her rotation. The camera rolls clockwise with small amplitude at\nfast speed.\n\n[Shot 8] At 00:07.000, the camera cuts to an off-white field with a hard black band. Four\nbold black kanji slam in one at a time at different scales and angles — \"速\", then \"鋭\",\nthen \"静\", then \"烈\" — each landing with a white shutter flash and an orange impact mark,\nwhile she slides on her knees across the foreground, jacket flaring. The camera shakes\nslightly with small amplitude at fast speed.\n\n[Shot 9] At 00:08.000, the camera cuts to a black frame where a white wireframe grid tilts\naway in 3D. She runs up the grid like a wall, kicks off it, and hangs frozen in mid-air as\na fluorescent-orange circle stamps around her pose like a target lock with silver UI\nbrackets and tick marks. The camera tilts up with small amplitude at slow speed.\n\n[Shot 10] At 00:09.000, the freeze releases and the camera cuts to layered flat-color panes\nstacked toward the lens. She dives straight at the camera, the panes tear open one at a\ntime like thick paper, and each tear reveals a larger portion of the black kanji \"烈\"\nbehind, before her sneaker wipes across the foreground. The camera pulls out with large\namplitude at fast speed.\n\n[Shot 11] At 00:10.100, the camera cuts to a rapid-fire montage of four full-screen graphic\nposters of her in different action poses — mid-flip, sliding, landing, hands in pockets —\nsnapping past with hard cuts, each with an oversized silver number 01 to 04, a black\nbarcode strip and thick diagonal slashes. Static shot.\n\n[Shot 12] At 00:11.100, the camera cuts to a clean off-white cyclorama for the hero moment.\nShe lands a final backflip dead center in slow motion, straightens with both hands in her\npockets, and a shockwave of concentric fluorescent-orange rings blasts outward from her feet\ncarrying shattered fragments of the kanji \"烈\" with it, then the frame overexposes toward\nwhite. The camera pushes in with small amplitude at slow speed.\n\n[Shot 13] At 00:13.000, the camera cuts to the final identity card on a pale off-white\nfield. An enormous black kanji \"烈\" dominates the frame with translucent orange rings and\nthin technical arcs behind it, the silver-grey word \"RETSU\" sits directly beneath the kanji,\nand she stands relaxed at the lower center overlapping the bottom of the strokes with her\njacket rippling. One thin orange pulse sweeps horizontally through the typography and holds\ncleanly until exactly 15.000 seconds. Static shot.\n\nEditing: extremely aggressive rhythm — hard cuts on every beat, graphic matches, whip pans,\nsnap zooms, stroboscopic freezes, foreground wipes, split flashes and impact shakes. Every\ncut must feel compositionally different. All Japanese typography is always fully readable\nbefore the character overlaps it. No weapons, no combat, no fire, no explosions; all energy\ncomes from motion design, ink, torn paper, wind and her street-dance athleticism. No\nphotorealism, no 3D rendering, no gradients, no browser interface, no player controls, no\nsubtitles, no captions, and no text other than the Japanese characters named above.\nExactly 13 cuts.\n\noverall_soundscape:\nSneaker impacts, knee slides and landings hit hard against a clean room tone. Ink splatters\nburst, thick paper tears open, and flat panels shatter like glass. Sharp whooshes carry the\ncolor wipes and typography slams, with small shutter clicks on each flash. No dialogue.\n\nnon_diegetic_music:\nAn original 15-second drum-heavy breakbeat with a deep sub bass, hard snare hits, risers and\nglitch fills locked to every cut. Begin with a single impact, build tightly through the\nmiddle, peak at the landing at 00:11.100, and end with a cold electronic logo sting that\nresolves at 15.000 seconds.",
    "engagement": {
      "replies": 2,
      "reposts": 13,
      "likes": 132,
      "views": 9694,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption publishes the H3 prompt produced from the creator's brief; reviewed frames match the kinetic commercial"
  },
  {
    "id": "x-2090306040097034391",
    "title": "燃烧房间里的从容小狗",
    "titleEn": "A Calm Dog in a Burning Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一网络迷因分别由两个模型重制，小狗坐在桌边，房间逐步被火焰吞没。 原生视频规格为 15 秒 · 1080×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two models recreate the same meme as a dog sits calmly at a table while fire consumes the room. Native-video output: 15s · 1080×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cencencahya12/status/2090306040097034391",
    "sourceLabel": "X 原帖 · @Cencencahya12",
    "author": "Cencen",
    "publishedAt": "2026-08-20T05:12:50.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090306040097034391.mp4",
    "posterUrl": "/posters/x/2090306040097034391.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Mengubah meme ikonik “This Is Fine” menjadi adegan sinematik AI. 🔥\n\nDibuat di @itsPolloAI - Seedance 2.5 1080P vs MiniMax H3 2K. Meme yang sama, prompt yang sama, model yang berbeda. Pertanyaan sebenarnya: mana yang paling pas? 👀",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 65,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption explicitly labels Seedance 2.5 versus MiniMax H3 with the same prompt; reviewed frames confirm the comparison"
  },
  {
    "id": "x-2090305577767547285",
    "title": "废墟宫殿里的喷火巨龙",
    "titleEn": "A Fire-Breathing Dragon in a Ruined Palace",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑红巨龙在残破宫殿中展开双翼并喷出火柱。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-and-red dragon spreads its wings inside a ruined palace and releases a column of fire. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cencencahya12/status/2090305577767547285",
    "sourceLabel": "X 原帖 · @Cencencahya12",
    "author": "Cencen",
    "publishedAt": "2026-08-20T05:11:00.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090305577767547285.mp4",
    "posterUrl": "/posters/x/2090305577767547285.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Naga Pemuntah Api di Istana yang Hancur dibuat oleh Model MiniMax H3⤵️",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 41,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption directly identifies the scene as made with MiniMax H3; reviewed frames confirm the dragon action"
  },
  {
    "id": "x-2090304950266187945",
    "title": "双色偶像组合的人物短片",
    "titleEn": "A Two-Member Idol Character Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "金发与蓝发偶像在粉色图形、合照和档案式文字卡之间完成组合介绍。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Blonde and blue-haired idols are introduced through pink graphics, paired portraits, and profile typography. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pon708/status/2090304950266187945",
    "sourceLabel": "X 原帖 · @pon708",
    "author": "こづつみPON",
    "publishedAt": "2026-08-20T05:08:30.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090304950266187945.mp4",
    "posterUrl": "/posters/x/2090304950266187945.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でキャラクターMVを実験中。\n\n今回はキャラ表の作り方も少し変更。\n\nAIKAとYUIで15秒のMVテスト。\nAI動画は、キャラ表の設計そのものも面白くなってきた。\n\n#MiniMaxH3 #AI動画 #AIAnimation #生成AI #PONORI #ポノリ",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 8,
      "views": 366,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption identifies a MiniMax H3 character-sheet MV test; reviewed frames show the two-character reveal"
  },
  {
    "id": "x-2090297623622426984",
    "title": "南极研究员与企鹅群",
    "titleEn": "An Antarctic Researcher Meets Penguins",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "研究员走向雪岸企鹅群，举起相机观察，并目送企鹅滑入海水。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A researcher approaches a penguin colony, raises her camera, and watches a penguin slide into the sea. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiwithaly/status/2090297623622426984",
    "sourceLabel": "X 原帖 · @aiwithaly",
    "author": "Ali",
    "publishedAt": "2026-08-20T04:39:23.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090297623622426984.mp4",
    "posterUrl": "/posters/x/2090297623622426984.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A quiet Antarctic encounter with a curious penguin colony. Realistic wildlife behavior, natural movement, and cinematic documentary storytelling bring the moment to life.\n\nCreated with MiniMax H3.\n\nPrompt:A Western female wildlife researcher walks from a remote Antarctic research station toward a snowy shoreline, carefully observing a colony of penguins while maintaining a respectful distance. She quietly says, \"They're already here,\" before raising her camera to photograph penguins waddling across the snow, shaking loose snow from their feathers, and one penguin naturally sliding into the icy ocean. Watching the penguin swim away, she smiles and softly whispers, \"Amazing.\" Ultra-photorealistic wildlife documentary, cinematic realism, authentic penguin biomechanics, realistic snow, ice, and water physics, natural human movement, accurate English lip-sync, neutral daytime lighting, immersive Antarctic ambience, and seamless story continuity from start to finish.",
    "engagement": {
      "replies": 41,
      "reposts": 3,
      "likes": 141,
      "views": 3917,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption publishes the complete wildlife-documentary prompt; reviewed frames match the Antarctic encounter"
  },
  {
    "id": "x-2090292380914291144",
    "title": "雨夜街头的三十秒压力挑战",
    "titleEn": "A Thirty-Second Rainy-Street Challenge",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雨夜城市车流与独行人物维持连续镜头，用于挑战超规格长时生成。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Rainy city traffic and a lone pedestrian remain coherent in a continuous shot used to stress-test an overlength generation. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/iwhale/status/2090292380914291144",
    "sourceLabel": "X 原帖 · @iwhale",
    "author": "iwhale",
    "publishedAt": "2026-08-20T04:18:34.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090292380914291144.mp4",
    "posterUrl": "/posters/x/2090292380914291144.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 ลอง Generate คลิปความยาว 30 วินาทีในครั้งเดียว (เกินสเปคไปเท่าตัว) VRAM 32 GB ยังร้องขอชีวิต เกือบไปรอด",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 4,
      "views": 1691,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption identifies a 30-second MiniMax H3 attempt on 32GB VRAM; reviewed frames show the long rainy street shot"
  },
  {
    "id": "x-2090285250475708694",
    "title": "餐盘上的人脸蛋糕切片",
    "titleEn": "A Human-Face Cake Is Sliced Open",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "写实人脸蛋糕躺在餐盘上，被刀具切开并移走一块。 原生视频规格为 15 秒 · 3886×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A photoreal human-face cake lies on a plate before a knife cuts and removes a slice. Native-video output: 15s · 3886×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CreaitiveLabs/status/2090285250475708694",
    "sourceLabel": "X 原帖 · @CreaitiveLabs",
    "author": "ReAiLity Labs",
    "publishedAt": "2026-08-20T03:50:13.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090285250475708694.mp4",
    "posterUrl": "/posters/x/2090285250475708694.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3886×2160",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I did NOT expect MiniMax H3 Turbo to nail this on the first try 🤯\n\nPrompted an ultra-realistic human-face cake getting sliced open — full T2V, no image ref, straight from text.\n\nRan in ComfyUI. Speed + realism combo is unreal.\n\nPrompt's in the replies 👇 try it yourself and post your results\n\n#ComfyUI #MiniMaxH3 #AIVideo #T2V",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 61,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption explicitly states this was a full text-to-video MiniMax H3 Turbo generation; reviewed frames confirm the cut"
  },
  {
    "id": "x-2090284291033100433",
    "title": "奢华精华液的竖屏口播",
    "titleEn": "A Vertical Testimonial for Luxury Serum",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女性手持精华液进行竖屏演示，以自然表情和近距离产品展示完成口播。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A presenter demonstrates a luxury serum in a vertical testimonial with natural expressions and close product framing. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/noorlewisx/status/2090284291033100433",
    "sourceLabel": "X 原帖 · @noorlewisx",
    "author": "Noor",
    "publishedAt": "2026-08-20T03:46:25.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090284291033100433.mp4",
    "posterUrl": "/posters/x/2090284291033100433.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "ugc"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "In one of my recent creative experiments with MiniMax H3, I created a beauty-style UGC commercial featuring a luxury serum.\n\nThe lifelike camera movement, natural lighting, realistic expressions, and attention to product details made the final result feel clean, cinematic, and incredibly realistic.\n\nUnlimited #MiniMaxH3 on Runway",
    "engagement": {
      "replies": 47,
      "reposts": 4,
      "likes": 214,
      "views": 9996,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption explicitly identifies a MiniMax H3 beauty UGC commercial; reviewed frames confirm the format"
  },
  {
    "id": "x-2090276653012287567",
    "title": "卧室里的三套穿搭切换",
    "titleEn": "Three Outfit Changes in a Bedroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "女性在卧室打开衣袋后快速切换三套造型，并以镜头前定格结束。 原生视频规格为 16 秒 · 1088×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman opens a garment bag in her bedroom, switches through three outfits, and ends posed to camera. Native-video output: 16s · 1088×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Format: 15 seconds | 16:9 | 4K HDR | 24fps | photorealistic cinematic vlog\n\nCharacter: Use the girl from the uploaded reference image as the exact character reference. Preserve her facial identity, eye color, skin tone, hairstyle, body proportions, and overall appearance consistently throughout every shot. No face changes or identity drift.\n\nScene & Story:\n0–3 sec — The Mystery Package\nThe video opens with a handheld creator-vlog shot in her cozy bedroom during late afternoon. She walks into frame holding a stylish garment bag, looking genuinely curious. She places it on the bed and looks directly into the camera with a playful expression.\n\nShe says naturally: “Okay… I found something for tonight, but I have absolutely no clue if I can pull it off.”\n\nShe unzips the garment bag, revealing only a glimpse of the outfit.\n\n3–6 sec — First Reveal\nShe quickly steps behind a nearby room divider. The camera follows with a fast handheld whip-pan.\n\nShe emerges wearing a chic fitted top, tailored trousers, and elegant heels. She looks at herself in the mirror, slightly surprised by how good it looks.\n\nShe turns toward the camera and says: “Wait… this might actually work.”\n\nShe gives a small confident pose.\n\n6–9 sec — The Unexpected Switch\nInstead of snapping her fingers, she grabs a jacket hanging beside her and dramatically throws it toward the camera.\n\nAs the jacket briefly covers the lens, the outfit seamlessly transforms underneath it.\n\nWhen she pulls the jacket away, she is wearing a completely different fashion-forward look: a stylish mini dress with statement shoes and subtle accessories.\n\nShe looks down, raises one eyebrow, then laughs.\n\n9–12 sec — The Wild Card\nShe walks toward the mirror while the camera circles around her. She notices a third outfit hanging on the wardrobe door.\n\nSuddenly, she steps backward and spins.\n\nDuring the spin, her outfit naturally morphs into a glamorous evening look with a sophisticated silhouette, elegant heels, and refined accessories.\n\nThe transformation happens continuously during the movement with realistic fabric motion — no jump cut.\n\n12–15 sec — Final Decision\nShe walks toward the camera with confident energy, then stops very close to the lens.\n\nShe looks at the camera, points toward herself, then gestures toward the three outfits visible behind her.\n\nWith a playful smile, she says:\n\n“Okay… you’re choosing. Which one?”\n\nThe camera quickly pushes in toward her face as she laughs, ending on a natural freeze-like moment.\n\nVisual Direction:\nUltra-realistic fashion creator vlog, authentic handheld camera movement, natural body language, expressive facial reactions, realistic fabric physics, seamless outfit transitions, natural hair movement, subtle mirror reflections, cinematic depth of field, realistic skin texture, soft late-afternoon daylight mixed with warm practical bedroom lighting, detailed wardrobe and bedroom environment, premium fashion-film aesthetic while maintaining casual YouTube-vlog energy.\n\nCamera:\nHandheld selfie framing, medium shots, quick whip-pans, subtle camera shake, smooth tracking movements, one controlled orbit around the character, final slow push-in.\n\nAudio:\nNatural bedroom ambience, fabric and wardrobe sounds, footsteps, garment movement, subtle mirror-room reverb, upbeat modern fashion-vlog background music, natural spoken dialogue, soft whoosh effects during the outfit transitions. No subtitles, no logos, no watermark.\n\nConsistency & Quality:\nNo jump cuts during transformations. No duplicated character. No distorted hands or fingers. No warped clothing. No inconsistent room layout. No facial identity changes. No unnatural skin smoothing. Keep the same character, bedroom, lighting logic, and visual realism throughout.",
    "sourceUrl": "https://x.com/codewithhajra/status/2090276653012287567",
    "sourceLabel": "X 原帖 · @codewithhajra",
    "author": "H A J R A",
    "publishedAt": "2026-08-20T03:16:04.000Z",
    "addedAt": "2026-08-20T09:49:09.982Z",
    "mediaUrl": "/media/x-2090276653012287567.mp4",
    "posterUrl": "/posters/x/2090276653012287567.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1088×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "POV: You have 15 seconds to decide what to wear… and somehow end up with THREE completely different looks 😂✨\n\nFrom casual to party-ready. which outfit would you choose? 👀💃\n\nCreated in Minimax H3 \n\nPrompt:\n\nFormat: 15 seconds | 16:9 | 4K HDR | 24fps | photorealistic cinematic vlog\n\nCharacter: Use the girl from the uploaded reference image as the exact character reference. Preserve her facial identity, eye color, skin tone, hairstyle, body proportions, and overall appearance consistently throughout every shot. No face changes or identity drift.\n\nScene & Story:\n0–3 sec — The Mystery Package\nThe video opens with a handheld creator-vlog shot in her cozy bedroom during late afternoon. She walks into frame holding a stylish garment bag, looking genuinely curious. She places it on the bed and looks directly into the camera with a playful expression.\n\nShe says naturally: “Okay… I found something for tonight, but I have absolutely no clue if I can pull it off.”\n\nShe unzips the garment bag, revealing only a glimpse of the outfit.\n\n3–6 sec — First Reveal\nShe quickly steps behind a nearby room divider. The camera follows with a fast handheld whip-pan.\n\nShe emerges wearing a chic fitted top, tailored trousers, and elegant heels. She looks at herself in the mirror, slightly surprised by how good it looks.\n\nShe turns toward the camera and says: “Wait… this might actually work.”\n\nShe gives a small confident pose.\n\n6–9 sec — The Unexpected Switch\nInstead of snapping her fingers, she grabs a jacket hanging beside her and dramatically throws it toward the camera.\n\nAs the jacket briefly covers the lens, the outfit seamlessly transforms underneath it.\n\nWhen she pulls the jacket away, she is wearing a completely different fashion-forward look: a stylish mini dress with statement shoes and subtle accessories.\n\nShe looks down, raises one eyebrow, then laughs.\n\n9–12 sec — The Wild Card\nShe walks toward the mirror while the camera circles around her. She notices a third outfit hanging on the wardrobe door.\n\nSuddenly, she steps backward and spins.\n\nDuring the spin, her outfit naturally morphs into a glamorous evening look with a sophisticated silhouette, elegant heels, and refined accessories.\n\nThe transformation happens continuously during the movement with realistic fabric motion — no jump cut.\n\n12–15 sec — Final Decision\nShe walks toward the camera with confident energy, then stops very close to the lens.\n\nShe looks at the camera, points toward herself, then gestures toward the three outfits visible behind her.\n\nWith a playful smile, she says:\n\n“Okay… you’re choosing. Which one?”\n\nThe camera quickly pushes in toward her face as she laughs, ending on a natural freeze-like moment.\n\nVisual Direction:\nUltra-realistic fashion creator vlog, authentic handheld camera movement, natural body language, expressive facial reactions, realistic fabric physics, seamless outfit transitions, natural hair movement, subtle mirror reflections, cinematic depth of field, realistic skin texture, soft late-afternoon daylight mixed with warm practical bedroom lighting, detailed wardrobe and bedroom environment, premium fashion-film aesthetic while maintaining casual YouTube-vlog energy.\n\nCamera:\nHandheld selfie framing, medium shots, quick whip-pans, subtle camera shake, smooth tracking movements, one controlled orbit around the character, final slow push-in.\n\nAudio:\nNatural bedroom ambience, fabric and wardrobe sounds, footsteps, garment movement, subtle mirror-room reverb, upbeat modern fashion-vlog background music, natural spoken dialogue, soft whoosh effects during the outfit transitions. No subtitles, no logos, no watermark.\n\nConsistency & Quality:\nNo jump cuts during transformations. No duplicated character. No distorted hands or fingers. No warped clothing. No inconsistent room layout. No facial identity changes. No unnatural skin smoothing. Keep the same character, bedroom, lighting logic, and visual realism throughout.",
    "engagement": {
      "replies": 73,
      "reposts": 4,
      "likes": 83,
      "views": 1315,
      "snapshotAt": "2026-08-20T09:49:09.982Z"
    },
    "approvedAt": "2026-08-20T09:49:09.982Z",
    "editorialBasis": "caption publishes the complete reference-led prompt; reviewed frames match the outfit-change story"
  },
  {
    "id": "x-2090273822431674561",
    "title": "珊瑚色线条拼出品牌字标",
    "titleEn": "Coral Lines Assemble a Brand Mark",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "珊瑚色线条与几何块在米白和青绿色画面中移动，最终拼出品牌字标。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Coral lines and geometric blocks move across cream and teal fields before resolving into a brand mark. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EZheng66099/status/2090273822431674561",
    "sourceLabel": "X 原帖 · @EZheng66099",
    "author": "EZ",
    "publishedAt": "2026-08-20T03:04:49.000Z",
    "addedAt": "2026-08-20T12:20:35.381Z",
    "mediaUrl": "/media/x-2090273822431674561.mp4",
    "posterUrl": "/posters/x/2090273822431674561.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is simply the best for motion graphics right now. \n\nmade with @atlas_cloud_ai  API",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 93,
      "snapshotAt": "2026-08-20T12:20:35.381Z"
    },
    "approvedAt": "2026-08-20T12:20:35.381Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090272463934644653",
    "title": "原宿甜品街的巨型字景",
    "titleEn": "Giant Letterforms on a Harajuku Dessert Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉彩街景与人物穿行在巨型 LOVE 和 SWEET 字母装置之间。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pastel street and its central character move among giant LOVE and SWEET letter installations. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2090272463934644653",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-20T02:59:25.000Z",
    "addedAt": "2026-08-20T12:20:35.381Z",
    "mediaUrl": "/media/x-2090272463934644653.mp4",
    "posterUrl": "/posters/x/2090272463934644653.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "架空コンカフェCM\n王道原宿ポップ／巨大立体タイポグラフィ建築／KAWAII広告\nMinimax H3で動画作成\nプロンプトはリプ欄に👇",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 21,
      "views": 1000,
      "snapshotAt": "2026-08-20T12:20:35.381Z"
    },
    "approvedAt": "2026-08-20T12:20:35.381Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090271497491214625",
    "title": "机械狐狸飞越微缩城市",
    "titleEn": "A Mechanical Fox Leaps Across a Miniature City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "机械狐狸沿巴士与屋顶疾驰，最后穿过发光传送门。 原生视频规格为 16 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A mechanical fox races across buses and rooftops before leaping through a glowing portal. Native-video output: 16s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Brjen/status/2090271497491214625",
    "sourceLabel": "X 原帖 · @Brjen",
    "author": "Brian",
    "publishedAt": "2026-08-20T02:55:35.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090271497491214625.mp4",
    "posterUrl": "/posters/x/2090271497491214625.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on a AMD 7800XT 16GB gpu.\n\nThe fox running on top of the bus.... This whole run did not pan out how it was planned but it's still quality. There are a number of these miniature style clips coming and then some more, supposedly more detailed realistic clips.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 98,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090270568532828380",
    "title": "红黑动漫排版的双模型对照",
    "titleEn": "Red-and-Black Anime Typography Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一组动漫人物与红白动态排版由两种模型生成并排呈现。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same anime character and red-and-white kinetic typography are rendered side by side by two models. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kik0ai1jikake/status/2090270568532828380",
    "sourceLabel": "X 原帖 · @kik0ai1jikake",
    "author": "あぁく",
    "publishedAt": "2026-08-20T02:51:53.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090270568532828380.mp4",
    "posterUrl": "/posters/x/2090270568532828380.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3とSeedance2の\nモーショングラフィックを比較してみた\n\nどちらもCapcutで生成 コストは\n左 MiniMax H3 10秒 768p『 210 』\n右 Seedance2 480p『 110 』720pだと『 240 』\n確かにH3の方が最終的にコスパ良くなる印象\n\nプロンプトはリプ欄\n\n生成結果はH3の方が前評判通りスッキリ\nとまとまっている感じはしますが\n\nseedanceも悪くないので\n適材適所な感じかなと\n\n色々選択肢が広がるのは良いですね\n試していこうと思いました\n\n@capcutapp_jp #capcut #capcutcpp",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 26,
      "views": 967,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090267333172629630",
    "title": "动漫贝斯手的舞台演奏",
    "titleEn": "An Anime Bassist Performs on Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏镜头记录动漫贝斯手在灯光舞台上的连续演奏。 原生视频规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical sequence follows an anime bassist performing continuously under concert lighting. Native-video output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Toshi_nyaruo_AI/status/2090267333172629630",
    "sourceLabel": "X 原帖 · @Toshi_nyaruo_AI",
    "author": "Toshi@ニャルオAI",
    "publishedAt": "2026-08-20T02:39:02.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090267333172629630.mp4",
    "posterUrl": "/posters/x/2090267333172629630.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今流行っている？譜面を読み込ませて動画化する実験🎸\n\nMiniMaX H3でローカル生成。\n\nうん、アニメでは指の動きを再現するのは難しい。\n雰囲気は良いんですけどね。\n\nでも、ローカルでAI動画作れるのは嬉しいですね！\nMiniMaX H3最高😍",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 13,
      "views": 224,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090265136569200923",
    "title": "吉他手的竖屏音乐蒙太奇",
    "titleEn": "A Vertical Guitarist Music Montage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多段舞台和演唱镜头组成一支竖屏吉他手音乐蒙太奇。 原生视频规格为 60 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Multiple stage and singing shots form a vertical montage centered on a guitarist. Native-video output: 60s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Aiborned/status/2090265136569200923",
    "sourceLabel": "X 原帖 · @Aiborned",
    "author": "Aiborned",
    "publishedAt": "2026-08-20T02:30:18.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090265136569200923.mp4",
    "posterUrl": "/posters/x/2090265136569200923.jpg",
    "duration": 60,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Music Video\n720p Upscaled with topaz 1080 60fps.\nSome clips are perfect, others need improvement, we're almost there.\n\n#minimax #comfyui",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 66,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090261780115427440",
    "title": "夏日祭典里的双人盆舞",
    "titleEn": "Two Dancers at a Summer Festival",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "两名和服角色在灯笼照亮的夏日祭典中同步起舞。 原生视频规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two kimono-clad characters dance in sync beneath lanterns at a summer festival. Native-video output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/momuandteasteam/status/2090261780115427440",
    "sourceLabel": "X 原帖 · @momuandteasteam",
    "author": "MoMu & TeaSteam (もむち)",
    "publishedAt": "2026-08-20T02:16:58.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090261780115427440.mp4",
    "posterUrl": "/posters/x/2090261780115427440.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ローカルのMiniMax H3、キャラクターシートを参照させることで安定してきました。",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 31,
      "views": 406,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090257920953123100",
    "title": "长矛与字标的英雄揭晓",
    "titleEn": "A Spear-and-Type Hero Reveal",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长矛特写、疾速排版与角色定格共同完成英雄登场。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Spear close-ups, fast typography, and a final character pose build a compact hero reveal. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/junwatu/status/2090257920953123100",
    "sourceLabel": "X 原帖 · @junwatu",
    "author": "DΞV",
    "publishedAt": "2026-08-20T02:01:38.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090257920953123100.mp4",
    "posterUrl": "/posters/x/2090257920953123100.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is so good at generating kinetic typography and maintaining character consistency. And the most important part is that anyone can use it, and it’s not that costly.\n\n#MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 15,
      "views": 292,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090251901388693638",
    "title": "怪盗电影的金黑预告片",
    "titleEn": "A Gold-and-Black Heist Trailer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "两名三维怪盗穿梭博物馆与金库，金黑字卡串起完整预告节奏。 原生视频规格为 20 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two 3D thieves move through a museum and vault as gold-and-black title cards shape the trailer rhythm. Native-video output: 20s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tanabe_fragm/status/2090251901388693638",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-20T01:37:42.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090251901388693638.mp4",
    "posterUrl": "/posters/x/2090251901388693638.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMaxDesignからMiniMaxH3を使って、架空の怪盗映画\n「LUCIFER'S（ルシファーズ）」\nの予告編（19秒）を作りました😃\n\n生成コストは300円ちょっと。\nこれが割と簡単に作れるのですから、世界中がMiniMaxH3に熱狂するのも当然かなと思います😌\n#MiniMaxH3 on #MiniMaxDesign\n\n今回もストーリーボードをベースに生成しましたが、ストーリーボードがキャラクターシート的な役割も兼ねています。\nMiniMaxH3は参照能力が高いですね👍\n\n動画は\n・前半11秒\n・後半8秒\nの2回に分けて生成。\n\nキャラクターシートはGPT Images 2.0。\nBGMはSunoです。\n\nMiniMax H3で生成した前半11秒のプロンプトはこちら👇\n※ストーリーボードを参照\n------------------\nsubject_definitions:\n<Subject 1> is Emi in <Picture 1>: a young woman in DreamWorks-style 3D, brown hair in a low\nside ponytail, large hazel eyes, slim build, black high-neck bodysuit.\n<Subject 2> is Max in <Picture 1>: a big heavyset man, short dark spiky hair, thick brows,\nblack sleeveless top over a grey tee, fingerless gloves, harness straps.\n<Subject 3> is the graphic style: heist-movie motion graphics in gold on black - thin gold\nrules, hard diagonal wipes, target reticles, laser grids, scan lines, spotlights.\n\nsummary:\n[reference generation] The first 11 seconds of a stylish anime heist-film trailer called\n\"LUCIFER'S\". A museum case shatters, the target is revealed, then two of the three thieves are\nintroduced. It ends mid-story on a black name card so a second clip can continue from it.\n<Picture 1> is a storyboard sheet for casting and style.\n\nretention_analysis:\n<Subject 1> ([Shot 3]-[Shot 5]): fully_preserved - her face, ponytail, eye color and black\nbodysuit stay identical; only pose and expression change.\n<Subject 2> ([Shot 6]-[Shot 8]): fully_preserved - his build, hair, brows and black top stay\nidentical.\n<Subject 3> (every shot): weak_reference - only the gold-on-black heist feel carries over; the\nstoryboard's grid, borders and numbers are not reproduced.\n\ndetailed_description:\n[Shot 1] Cinematic 3D CG in modern DreamWorks feature style - stylised proportions, soft\nshading, rich lighting - with animated Japanese type. Night inside a dark marble museum, no\npeople. A tall glass display case shatters inward in slow motion, shards hanging in the air. A\nred alarm light then sweeps across the hall and a gold laser grid snaps on across the floor.\nPush In, small amplitude, fast speed.\n[Shot 2] At 00:01.300, a gold reticle wipes in. A jewelled crown sits alone on a lit pedestal\nbehind cracked glass, turning slowly, catching the light. Gold bracket marks then close in on\nit from four corners and lock with a hard snap, and fine dust drifts through the beam. Push\nIn, small amplitude, slow speed.\n[Shot 3] At 00:02.800, a magenta field wipes in diagonally over a warm gilded gala hall,\nguests blurred behind. <Subject 1> in a beige trench and black wide-brimmed hat tips her\nsunglasses down and looks straight to camera. Magenta chevrons and thin gold scan lines then\nstreak across the frame one after another, and a magenta reticle snaps onto her eyes and\nlocks. Truck Right, small amplitude, slow speed.\n[Shot 4] At 00:04.200, a dark ops room lit by screens. <Subject 1> leans over a glowing\nhologram of the museum and taps it once. Blue schematic panels then slide in behind her one\nafter another as thin magenta lines draw between them. A big booming man (S2) says off-screen:\n<d>[Japanese] 行けるか？</d> A calm low-pitched woman in her twenties (S1) answers without\nlooking up: <d>[Japanese] 完璧よ。</d> Push In, small amplitude, slow speed.\n[Shot 5] At 00:05.700, a magenta field wipes hard across the frame and settles to black. Huge\nwhite type then slams in from the left reading \"エミ\" with an impact shake, a magenta rule\ndraws itself beneath, and one word stamps in under the rule a beat later reading \"変装。\".\nMagenta shards scatter outward from the impact as <Subject 1> slides in from the right, arms\nfolded, settling beside the type without covering it. Static Shot.\n[Shot 6] At 00:07.000, an electric-blue field wipes in and a heavy vault door fills the frame.\n<Subject 2> braces and hauls the huge wheel, sparks spitting from the rim. Thick blue impact\nrings then pulse outward one after another and the bolts crack free as blue chevrons tear past\nthe edges. Push In, large amplitude, fast speed.\n[Shot 7] At 00:08.400, the warm cluttered hideout. <Subject 2> grins and throws a thumbs-up\nstraight at camera. White burst lines then pop behind him and a blue chevron strip slides\nacross the lower frame. (S2): <d>[Japanese] まかせろ！</d> Pull Out, small amplitude, fast\nspeed.\n[Shot 8] At 00:09.800, an electric-blue field wipes hard across the frame and settles to\nblack. Huge white type then slams in from the right reading \"マックス\" with an impact shake, a\nblue rule snaps in beneath, and one word stamps in under the rule a beat later reading\n\"怪力。\". Blue shards scatter from the impact as <Subject 2> steps in from the left, arms\nfolded, without covering the type. The frame ends on this black card at exactly 11.000\nseconds. Static Shot.\nThroughout: keep one DreamWorks-style 3D CG look - never photorealistic, never 2D anime. Every\ncut lands exactly on its stated timestamp and must look different in composition from the\nlast. All type is bold and moves as motion graphics, always fully readable before a character\noverlaps it - never a subtitle, caption strip or romaji. The only text anywhere is \"エミ\", \"変装。\",\n\"マックス\" and \"怪力。\"; every other surface stays blank, with no invented words, no Chinese or Korean, no\ngarbled glyphs. No storyboard grid, borders, numbers or split screen. Keep [Shot 1] and [Shot 2] in black, warm gold and white\nonly. The character sections carry their own full colour field: magenta fills [Shot 3]-[Shot 5]\nand electric blue fills [Shot 6]-[Shot 8], and neither colour appears outside its own section. Only (S1) and (S2) speak, exactly those three lines: no\nad-libs, no narration, no singing. Each line starts within 0.2 seconds of its shot beginning\nand ends before that shot ends. Nobody speaks in [Shot 1], [Shot 2], [Shot 5] or [Shot 8]. No\nweapons, no fighting; all tension comes from lighting, graphic motion and craft.\n\noverall_soundscape:\nGlass shatters into a long crystalline scatter, a low alarm pulses twice under it, and the\nlaser grid snaps on with an electric tick. Gold brackets lock onto the crown with a hard\nmechanical clack. Every wipe is a filtered whoosh, every rule draws with a metallic zip, each name slams in on a\ndeep sub-hit with a short reversed riser, and the word beneath it stamps down with a dry\nmechanical thud. A grinding groan and\nscattered sparks carry the vault wheel.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 14,
      "views": 1771,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090242494193660134",
    "title": "七宗罪角色的七幕揭晓",
    "titleEn": "Seven Deadly Sins Character Reveals",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "单张角色参考延展为七宗罪主题的连续人物字卡。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A single character reference expands into a sequence of Seven Deadly Sins character cards. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2090242494193660134",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-20T01:00:20.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090242494193660134.mp4",
    "posterUrl": "/posters/x/2090242494193660134.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 + GPT Image 2\n\nOne reference image. Seven Deadly Sins.\n\nSharing the prompts for both the image and video in the replies.",
    "engagement": {
      "replies": 10,
      "reposts": 2,
      "likes": 78,
      "views": 3940,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090234862028406919",
    "title": "糖浆流动的双模型对照",
    "titleEn": "Syrup Flow Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "食物表面的糖浆流动与质感变化由两种模型并排比较。 原生视频规格为 15 秒 · 540×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Syrup flow and surface texture on food are compared side by side across two models. Native-video output: 15s · 540×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Lart_AI/status/2090234862028406919",
    "sourceLabel": "X 原帖 · @Lart_AI",
    "author": "Lart AI",
    "publishedAt": "2026-08-20T00:30:00.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090234862028406919.mp4",
    "posterUrl": "/posters/x/2090234862028406919.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "540×960",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Chiikawa goes parkour! 🥐\nStarting from the croissant mountains, \nsurfing across a coffee lake.\n\nWe re‑created this iconic scene with\nMiniMax H3 and  Seedance2 .\n\nWhich model nails it better? You decide👇\n#LartAI #chiikawaeng #AIVideo #AIAnimation",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 6,
      "views": 410,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090230978547703913",
    "title": "街头偶遇的即兴对话",
    "titleEn": "An Impromptu Street Encounter",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "写实街头人物相遇并展开带口型和环境声的短对话。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Realistic street characters meet and exchange a short lip-synced conversation with ambient sound. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Jessewelle/status/2090230978547703913",
    "sourceLabel": "X 原帖 · @Jessewelle",
    "author": "Jesse Wellens",
    "publishedAt": "2026-08-20T00:14:34.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090230978547703913.mp4",
    "posterUrl": "/posters/x/2090230978547703913.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3 is just way too good. Rendered this locally on my machine.",
    "engagement": {
      "replies": 91,
      "reposts": 139,
      "likes": 2024,
      "views": 454071,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames",
    "featured": {
      "at": "2026-09-05T15:50:15.693Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2090230972851835104",
    "title": "黑色豪华车的多镜头广告",
    "titleEn": "A Multi-Shot Commercial for a Black Luxury Car",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "车身细节、驾驶舱与行驶镜头保持同一辆黑色豪华车的连续性。 原生视频规格为 30 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Exterior details, cockpit shots, and driving footage maintain continuity for the same black luxury car. Native-video output: 30s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Flkrstudio/status/2090230972851835104",
    "sourceLabel": "X 原帖 · @Flkrstudio",
    "author": "Issei",
    "publishedAt": "2026-08-20T00:14:33.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090230972851835104.mp4",
    "posterUrl": "/posters/x/2090230972851835104.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Built this commercial concept to explore MiniMax H3\n\nProduct continuity across multiple shots, cinematic editing, motion-designed typography, and a cohesive visual identity from interior details to the final frame.\n\nAnd now you can keep experimenting without worry, unlimited #MiniMaxH3 on @runwayml",
    "engagement": {
      "replies": 14,
      "reposts": 1,
      "likes": 37,
      "views": 1039,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090222715948917233",
    "title": "蓝白网球运动排版",
    "titleEn": "Blue-and-White Tennis Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "网球运动员、球拍和蓝白字形在高速节拍中相互穿插。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tennis player, racket, and blue-and-white letterforms interlock to a fast visual beat. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/synthetic_ape/status/2090222715948917233",
    "sourceLabel": "X 原帖 · @synthetic_ape",
    "author": "synthetic ape",
    "publishedAt": "2026-08-19T23:41:44.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090222715948917233.mp4",
    "posterUrl": "/posters/x/2090222715948917233.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "All animation and sound effects were created using MiniMax H3 with just one prompt.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 74,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090214618882773503",
    "title": "步行街上的自然行走",
    "titleEn": "A Natural Walk Through a Pedestrian Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "参考人物自然走入指定步行街场景，镜头保持稳定跟随。 原生视频规格为 7 秒 · 832×1120。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A referenced woman walks naturally through the specified pedestrian street while the camera follows steadily. Native-video output: 7s · 832×1120. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The woman from <Picture 2> walks naturally along the pedestrian street shown in <Picture 3>.",
    "sourceUrl": "https://x.com/00Bohr/status/2090214618882773503",
    "sourceLabel": "X 原帖 · @00Bohr",
    "author": "José Luis",
    "publishedAt": "2026-08-19T23:09:34.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090214618882773503.mp4",
    "posterUrl": "/posters/x/2090214618882773503.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "832×1120",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 es divertido.  \"The woman from <Picture 2> walks naturally along the pedestrian street shown in <Picture 3>.\"  #MiniMaxH3 #ComfyUi 16 Gb VRAM 12 min.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 114,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090214143022490065",
    "title": "森林追逐的一镜到底挑战",
    "titleEn": "A Forest Chase in One Continuous Shot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头在森林、人物与巨大树干之间连续穿行，尝试保持一镜到底。 原生视频规格为 14 秒 · 1344×1524。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera travels continuously through a forest, characters, and a massive trunk in an ambitious single take. Native-video output: 14s · 1344×1524. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIWarper/status/2090214143022490065",
    "sourceLabel": "X 原帖 · @AIWarper",
    "author": "A.I.Warper",
    "publishedAt": "2026-08-19T23:07:40.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090214143022490065.mp4",
    "posterUrl": "/posters/x/2090214143022490065.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "1344×1524",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "After testing something more difficult, I have deduced that it's in your best interest to segment your videos into clips rather than trying to blast the full shot.\n\nThis is as close as I could get this ambitious one shot\n\nMinimax H3",
    "engagement": {
      "replies": 20,
      "reposts": 4,
      "likes": 112,
      "views": 6880,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090209401206653023",
    "title": "烟花夜里的夏末预告",
    "titleEn": "A Late-Summer Trailer Under Fireworks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "阳台人物、夜空烟花与日文字卡组成带夏末情绪的预告片段。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A balcony figure, night-sky fireworks, and Japanese title cards create a late-summer trailer mood. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kurumirekishi/status/2090209401206653023",
    "sourceLabel": "X 原帖 · @kurumirekishi",
    "author": "くるみ | 歴史×AIクリエイター",
    "publishedAt": "2026-08-19T22:48:50.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090209401206653023.mp4",
    "posterUrl": "/posters/x/2090209401206653023.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "夏の終わりの物語、15秒の予告ができた🎆\n作ったのはMiniMax H3。\n\nMGの制御がすごくて、文字が思い通りに動いた✨\n\nこれでリリックMVが作れそう！これから試します🔥🎸\n（Runwayのアンリミで23日まで使えるらしい）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 251,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090193779089461340",
    "title": "装备室里的动漫人物卡",
    "titleEn": "An Anime Character Card in a Gear Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "圆形图形、装备室背景与动漫人物共同组成节奏化角色卡。 原生视频规格为 15 秒 · 640×368。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Circular graphics, a gear-room backdrop, and an anime figure combine into a rhythmic character card. Native-video output: 15s · 640×368. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nogsaw2/status/2090193779089461340",
    "sourceLabel": "X 原帖 · @nogsaw2",
    "author": "nogsaw",
    "publishedAt": "2026-08-19T21:46:45.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090193779089461340.mp4",
    "posterUrl": "/posters/x/2090193779089461340.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "640×368",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 モーショングラフィックエグいね。適当に作って粗が目立つけど、簡素な物なら簡単に作れるね。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090193337504710980",
    "title": "红绿灯纪念日的街角小剧场",
    "titleEn": "A Traffic-Light Day Street-Corner Sketch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三维女孩和猫在红绿灯旁完成一段轻巧的街角小剧场。 原生视频规格为 15 秒 · 576×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A 3D girl and cat perform a playful street-corner vignette beside a traffic light. Native-video output: 15s · 576×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Spoonful_of_Fun/status/2090193337504710980",
    "sourceLabel": "X 原帖 · @Spoonful_of_Fun",
    "author": "一抹のFun",
    "publishedAt": "2026-08-19T21:45:00.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090193337504710980.mp4",
    "posterUrl": "/posters/x/2090193337504710980.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "576×576",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます。\n\n今日８月20日は「交通信号設置記念日」なんですね。\nということで今日もMiniMax H3で動画化。\n…というか、この日本語の乱れは高速LoRAを使ってるから？関係ない？明日は高速LoRA無しでトライするかー。\n\n最終の画像の猫、足が多いような。Geminiもよくやってたけど、最近は見なくなったと思ったら💦",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 3,
      "views": 130,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090185695641968736",
    "title": "包豪斯字形的节拍重组",
    "titleEn": "Bauhaus Letterforms Recompose on Beat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "FORM、FOLLOWS、FUNCTION 与几何色块按节拍在严格网格中重组。 原生视频规格为 12 秒 · 1080×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "FORM, FOLLOWS, FUNCTION, and geometric color fields recompose on beat within a strict grid. Native-video output: 12s · 1080×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create an 8-cut kinetic-typography spot in 16:9, 24fps, exactly 12s. Style: Bauhaus / Swiss International — flat primary colors (red, yellow, blue, black, off-white), strict grid, geometric sans-serif type.\nThis film is 90% typography and geometric shapes in motion, 10% texture accents.\nCUT 01 | 0.0-1.5s - Off-white void. The word \"FORM\" drops letter by letter onto a baseline grid, each landing with a small bounce.\nCUT 02 | 1.5-3.0s - A red circle rolls in from the left and knocks the \"M\" out of frame; \"FOR—\" holds one beat.\nCUT 03 | 3.0-4.5s - Hard cut: full yellow frame, \"FOLLOWS\" stacked vertically, letters sliding in alternating directions.\nCUT 04 | 4.5-6.0s - A blue diagonal wipes the frame; \"FUNCTION\" sits on the diagonal; the camera rotates until it reads horizontal.\nCUT 05 | 6.0-7.5s - The three words swap positions on a strict 3-column grid, snapping on beat.\nCUT 06 | 7.5-9.0s - All type shrinks into an 8pt grid texture; one black square scales up through center.\nCUT 07 | 9.0-10.5s - The square splits into a Mondrian composition; one word per color field.\nCUT 08 | 10.5-12.0s - End card: \"FORM FOLLOWS FUNCTION — 1919\" aligned bottom-left, the red circle as the full stop, hold.\nTypography is fully readable before any shape overlaps it. Letters never distort or melt; kerning stays optically even.\nEditing: hard cuts and snap moves locked to 120bpm.\nAudio: minimal electronic clicks + bass pulse on every cut.\nDO NOT: gradients, drop shadows, 3D extrusion, serif fonts, melted letters, added words.",
    "sourceUrl": "https://x.com/EZheng66099/status/2090185695641968736",
    "sourceLabel": "X 原帖 · @EZheng66099",
    "author": "EZ",
    "publishedAt": "2026-08-19T21:14:38.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090185695641968736.mp4",
    "posterUrl": "/posters/x/2090185695641968736.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Typography Test, Minimax H3, Gemni Omni, Seedance 2.5 \n\nH3 is pretty accurate! \n\nmade with @atlas_cloud_ai API \n\nPrompt ↓\n\nCreate an 8-cut kinetic-typography spot in 16:9, 24fps, exactly 12s. Style: Bauhaus / Swiss International — flat primary colors (red, yellow, blue, black, off-white), strict grid, geometric sans-serif type.\nThis film is 90% typography and geometric shapes in motion, 10% texture accents.\nCUT 01 | 0.0-1.5s - Off-white void. The word \"FORM\" drops letter by letter onto a baseline grid, each landing with a small bounce.\nCUT 02 | 1.5-3.0s - A red circle rolls in from the left and knocks the \"M\" out of frame; \"FOR—\" holds one beat.\nCUT 03 | 3.0-4.5s - Hard cut: full yellow frame, \"FOLLOWS\" stacked vertically, letters sliding in alternating directions.\nCUT 04 | 4.5-6.0s - A blue diagonal wipes the frame; \"FUNCTION\" sits on the diagonal; the camera rotates until it reads horizontal.\nCUT 05 | 6.0-7.5s - The three words swap positions on a strict 3-column grid, snapping on beat.\nCUT 06 | 7.5-9.0s - All type shrinks into an 8pt grid texture; one black square scales up through center.\nCUT 07 | 9.0-10.5s - The square splits into a Mondrian composition; one word per color field.\nCUT 08 | 10.5-12.0s - End card: \"FORM FOLLOWS FUNCTION — 1919\" aligned bottom-left, the red circle as the full stop, hold.\nTypography is fully readable before any shape overlaps it. Letters never distort or melt; kerning stays optically even.\nEditing: hard cuts and snap moves locked to 120bpm.\nAudio: minimal electronic clicks + bass pulse on every cut.\nDO NOT: gradients, drop shadows, 3D extrusion, serif fonts, melted letters, added words.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 3,
      "views": 908,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090183484057034955",
    "title": "新艺术机械乌鸦的变形长卷",
    "titleEn": "An Art Nouveau Mechanical Raven Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "机械手、乌鸦纹章与女巫插画在新艺术风格中连续变形。 原生视频规格为 30 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A mechanical hand, raven emblem, and witch illustration transform through an Art Nouveau sequence. Native-video output: 30s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/oohiro35/status/2090183484057034955",
    "sourceLabel": "X 原帖 · @oohiro35",
    "author": "おおひろ",
    "publishedAt": "2026-08-19T21:05:50.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090183484057034955.mp4",
    "posterUrl": "/posters/x/2090183484057034955.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "アール・ヌーボー調の一枚のイラストからMinimax H3で動画生成しました。\nローカル生成後にSFXをつけるの忘れたのに気がついたのでアップスケール&SFXのためPixverseで動画リファレンスをして再生成しましたところ見事SFXも生成してくれました。\n@pixverse #pixverseCPP",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 17,
      "views": 1367,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090182264705454105",
    "title": "铅笔线条生长为绿色标记",
    "titleEn": "A Pencil Line Grows Into a Green Mark",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "铅笔画出的线条逐步生长并收束为抽象绿色标记。 原生视频规格为 15 秒 · 1440×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pencil-drawn line grows across the page and resolves into an abstract green mark. Native-video output: 15s · 1440×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/itsnicholash/status/2090182264705454105",
    "sourceLabel": "X 原帖 · @itsnicholash",
    "author": "Nicholas | WithLore",
    "publishedAt": "2026-08-19T21:01:00.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090182264705454105.mp4",
    "posterUrl": "/posters/x/2090182264705454105.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1440×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is approaching this level of motion design...\n\nThis is my original logo, then I just gave it to Codex with an API for @Hailuo_AI and told it something like:\n\n\"Make a fast paced motion animation with cuts every second, hitting various different styles like XYZ\"\n\nSick.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 341,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090180874222588332",
    "title": "羊皮纸上的绝地光明史诗",
    "titleEn": "A Jedi Saga in Parchment and Gold Foil",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "羊皮纸、金箔与定格质感讲述一段关于光明的绝地史诗。 原生视频规格为 91 秒 · 2548×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Parchment, gold foil, and tactile stop-motion textures tell a Jedi saga about the light. Native-video output: 91s · 2548×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nonameoasis/status/2090180874222588332",
    "sourceLabel": "X 原帖 · @nonameoasis",
    "author": "nonameoasis",
    "publishedAt": "2026-08-19T20:55:28.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090180874222588332.mp4",
    "posterUrl": "/posters/x/2090180874222588332.jpg",
    "duration": 91,
    "aspectRatio": "landscape",
    "resolution": "2548×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "You can't tell the story of the Dark Side without bringing balance to the timeline. ⚖️✨\n\nTo complete the saga on Minimax H3, I engineered a 90-second, 6-part miniseries covering the entire history of the Jedi. We swapped the burnt cardboard for pristine parchment, gold foil, and translucent tracing paper for the Force ghosts. All rendered in a cinematic 21:9 ultrawide aspect ratio.\n\nZero fluid CGI. 100% tactile stop-motion simulation.\n\nI’m dropping the master prompts for all 6 scenes below 👇🧵\n@Hailuo_AI  #MiniMaxH3 #StarWars #Jedi #AIFilmmaking",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 94,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090175089404789097",
    "title": "会切换犬种的互动图鉴",
    "titleEn": "An Interactive Catalogue of Dog Breeds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "界面选择不同犬种后，中央角色随之切换并保持统一图鉴样式。 原生视频规格为 15 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Selecting different breeds in the interface changes the central dog while preserving a consistent catalogue style. Native-video output: 15s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nakazakifam/status/2090175089404789097",
    "sourceLabel": "X 原帖 · @nakazakifam",
    "author": "中崎工房 | AIで1時間の仕事を5分で終わらせる人",
    "publishedAt": "2026-08-19T20:32:29.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090175089404789097.mp4",
    "posterUrl": "/posters/x/2090175089404789097.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で作る動く犬図鑑動画。\n引用元でとんでもなくカッコいいイーブイみたいな動くポケモン図鑑的な動画があったので、それを身近な犬に変えて720pで参照画像なしでT2Vで作ってみました。\n犬の名前や図鑑の選択と実際に出る犬が合っていないとか細かなツッコミどころだらけですが、犬派の自分としてはそこそこ可愛い動画になりました🐶😅\n元動画のクオリティが圧倒的に高いですが、おそらく参照画像をしっかり作り込んでR2Vで作ったら多少は元動画のクオリティに近づけると思います。。\nまた、今回はH3が出した元音声に加えてロイヤリティーフリーでBGMやサウンドエフェクトが使えるEpidemic Soundがベータ版で出している動画をアップするとAIが自動で最適なBGMとサウンドエフェクトをライブラリから選んで配置してくれるStudioという機能で更に音源を追加しています。",
    "engagement": {
      "replies": 3,
      "reposts": 5,
      "likes": 69,
      "views": 8283,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090166948239909375",
    "title": "墙后低语的走廊惊魂",
    "titleEn": "Whispers Behind the Wall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "昏暗走廊、镜中人物与墙后低语共同制造逐步升级的惊悚氛围。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dim hallway, a reflected figure, and whispers behind the wall build steadily escalating horror. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PromptSin/status/2090166948239909375",
    "sourceLabel": "X 原帖 · @PromptSin",
    "author": "PSS",
    "publishedAt": "2026-08-19T20:00:08.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090166948239909375.mp4",
    "posterUrl": "/posters/x/2090166948239909375.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "You don’t need to see the monster when sound tells you where it is. 🎧\n\n“The Voice Behind the Wall”  made with MiniMax H3 on @openart_ai\nEvery voice, whisper, knock and ambience is native model audio.\nNothing added in post.\n\nWould you turn around? 👀",
    "engagement": {
      "replies": 11,
      "reposts": 1,
      "likes": 44,
      "views": 697,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090145591083077637",
    "title": "纸艺特洛伊城的陷落",
    "titleEn": "The Fall of Troy in Paper Cut-Outs",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "纸艺城墙、木马与燃烧场景以剪纸舞台形式重现特洛伊战争。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Paper walls, the wooden horse, and a burning city retell the Trojan War as a cut-out stage. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MayorKingAI/status/2090145591083077637",
    "sourceLabel": "X 原帖 · @MayorKingAI",
    "author": "MayorkingAI",
    "publishedAt": "2026-08-19T18:35:16.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090145591083077637.mp4",
    "posterUrl": "/posters/x/2090145591083077637.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Recreating the Trojan War in a paper cut-out style ⚔️\n\nMade with MiniMax H3 in @Hailuo_AI\n\nPrompt below 👇",
    "engagement": {
      "replies": 22,
      "reposts": 6,
      "likes": 121,
      "views": 5371,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090141574462333406",
    "title": "狼耳角色的预录音频口型",
    "titleEn": "Pre-Recorded Lip Sync for a Wolf-Eared Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "狼耳动漫角色根据预录音频完成口型、表情和肢体表演。 原生视频规格为 15 秒 · 1120×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A wolf-eared anime character follows pre-recorded audio with synchronized speech, expressions, and gestures. Native-video output: 15s · 1120×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MrJ0SHie/status/2090141574462333406",
    "sourceLabel": "X 原帖 · @MrJ0SHie",
    "author": "MrJ0SHiE (AI Trump)",
    "publishedAt": "2026-08-19T18:19:18.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090141574462333406.mp4",
    "posterUrl": "/posters/x/2090141574462333406.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1120×832",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing how well Minimax H3 handles pre-recorded audio for lip sync. Plus some character acting!",
    "engagement": {
      "replies": 11,
      "reposts": 4,
      "likes": 44,
      "views": 940,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2090136979866521914",
    "title": "穿越多重材质世界的旅人",
    "titleEn": "A Traveler Crosses Worlds of Different Materials",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "旅人连续穿过门洞，在写实、纸艺与动画材质世界之间切换。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A traveler crosses successive doorways between realistic, paper-crafted, and animated worlds. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hmetouch/status/2090136979866521914",
    "sourceLabel": "X 原帖 · @hmetouch",
    "author": "Hemanth",
    "publishedAt": "2026-08-19T18:01:03.000Z",
    "addedAt": "2026-08-20T12:20:35.382Z",
    "mediaUrl": "/media/x-2090136979866521914.mp4",
    "posterUrl": "/posters/x/2090136979866521914.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Generated with MiniMax H3 Ref2VA.\n\nA man begins in the real world, enters a mysterious door, and travels through completely different visual worlds, becoming part of each one before eventually finding his way back to reality.\n\nThe idea was simple:\n\nOne man.\nMany worlds.\nDifferent materials, animation styles, costumes and actions.\nOne continuous journey back home.\n\nGenerated locally with ComfyUI on: RTX 4090 24GB + 128GB RAM\nVideo + References + Full Prompt + continuity storyboard attached ⤵\n⏳",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 188,
      "snapshotAt": "2026-08-20T12:20:35.382Z"
    },
    "approvedAt": "2026-08-20T12:20:35.382Z",
    "editorialBasis": "Written from the original caption and manual review of downloaded native-video frames"
  },
  {
    "id": "x-2087208466264076759",
    "title": "AI 脱口秀的精准笑点节奏",
    "titleEn": "Punchline Timing in an AI Stand-up Special",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "喜剧演员在小型俱乐部连续抛出笑点，观众反应、停顿与多机位切换紧跟台词节奏。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A comedian delivers a rapid club set while audience reactions, pauses, and multi-camera cuts track each punchline. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Style: Live stand-up comedy special, intimate comedy club, professional multi-camera production, warm stage lighting, packed audience around small tables, sharp HD broadcast look, natural facial expressions, authentic comedic timing, clean microphone audio, realistic audience reactions, subtle handheld audience camera, 15-second video, 5 cinematic cuts\n\n0–3s: [Wide Establishing → Medium Push-In]\nA comedian stands center stage holding a microphone as the audience settles. The comedian smiles and says:\n\nComedian: “I asked AI to organize my life yesterday”\n\nBrief beat\n\n3–6s: [Medium Close-Up]\nThe comedian maintains a completely serious expression\n\nComedian: “It looked at my schedule and said, ‘Actually... I’m just a language model.’”\n\nAudience immediately laughs\n\n6–9s: [Side Angle + Audience Reaction]\nThe comedian waits for the laughter, then slowly nods\n\nComedian: “Even artificial intelligence has boundaries”\n\nQuick cut to the front row laughing and clapping\n\n9–12s: [Tight Close-Up]\nThe comedian leans toward the microphone\n\nComedian: “My mom thinks AI is listening to everything we say”\n\nBeat\n\nComedian: “Mom... nobody wants that podcast”\n\nAudience erupts into louder laughter\n\n12–15s: [Medium → Wide Ending]\nThe comedian waits for silence, then delivers the final line:\n\nComedian: “AI isn't replacing us. It saw our search history and declined the position”\n\nBig audience laugh. The comedian smiles and lowers the microphone as the camera pulls out to reveal the applauding club\n\nProduction details: Keep the same comedian, outfit, microphone, stage and audience throughout. Prioritize precise comedic timing, short pauses before punchlines, believable facial expressions and natural audience reactions. Never cut during a punchline. Audience laughter starts only after the punchline lands. No canned laughter, no overlapping dialogue, coherent eyelines, realistic club acoustics, professional live comedy special editing",
    "promptSourceUrl": "https://x.com/azed_ai/status/2087208478758842551",
    "sourceUrl": "https://x.com/azed_ai/status/2087208466264076759",
    "sourceLabel": "X 原帖 · @azed_ai",
    "author": "Amira Zairi",
    "publishedAt": "2026-08-11T16:04:11.000Z",
    "addedAt": "2026-08-20T18:59:16.814Z",
    "mediaUrl": "/media/x-2087208466264076759.mp4",
    "posterUrl": "/posters/x/2087208466264076759.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I made a comedy special with Hailuo H3 and the timing is spot on\n\nThe audience laughs after the punchlines, the comedian's expressions are natural, the camera work is professional\n\nWhat do you think?  Prompt👇🏻",
    "engagement": {
      "replies": 26,
      "reposts": 2,
      "likes": 91,
      "views": 4582,
      "snapshotAt": "2026-08-20T18:59:16.814Z"
    },
    "approvedAt": "2026-08-20T18:59:16.814Z",
    "editorialBasis": "Original X caption and the creator’s complete prompt describe the comedy-club staging, dialogue beats, audience reactions, and camera cuts."
  },
  {
    "id": "x-2087183007874453795",
    "title": "醉拳海报苏醒后的动态擂台",
    "titleEn": "A Drunken-Master Poster Comes Alive",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "静态武术海报被扩展为带烟雾、火花、液体飞溅、人物动作与动态字体的电影式动效。 原生视频规格为 23 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A static martial-arts poster expands into moving characters, smoke, sparks, splashing liquid, and animated typography. Native-video output: 23s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Animate the uploaded CHUNG LEE movie poster as a 10-second cinematic martial-arts motion poster while preserving the original artwork, characters, typography, composition and layout.\n\nBring the static poster to life with rich cinematic colors — crimson red, golden yellow, fiery orange, electric blue, emerald green and magenta. Animate the lanterns, drifting smoke, flying dust, sparks, debris and the surrounding crowd.\n\nGive CHUNG LEE subtle drunken-master movement: shifting shoulders, flowing hair, a confident expression and a casual lift of the gourd with liquid splashing in slow motion.\n\nThen make “CHUNG LEE” the ultimate visual flex — dramatic brush-stroke reveals, golden particles, energy trails and a powerful 3D pop-out effect that makes the title feel like it is breaking out of the poster.\n\nAdd glowing red energy around the Chinese characters, dynamic martial-arts impacts, moving background fighters, cinematic camera movement and bursts of colorful light.\n\nFinish with a dramatic push toward the CHUNG LEE title, then pull back to reveal the complete poster before locking into the final frame.\n\nNo redesign. No new text. No distorted typography. Preserve the original poster — just make it move.\n\nAnimation made with Minimax H3",
    "promptSourceUrl": "https://x.com/MonetizationDon/status/2087183010747613522",
    "sourceUrl": "https://x.com/MonetizationDon/status/2087183007874453795",
    "sourceLabel": "X 原帖 · @MonetizationDon",
    "author": "LAPTOP LIFESTYLE BLOGGER",
    "publishedAt": "2026-08-11T14:23:01.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2087183007874453795.mp4",
    "posterUrl": "/posters/x/2087183007874453795.jpg",
    "duration": 23,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is making static-to-motion design really interesting. \n\nPROMPT IN COMMENT\n\nI took this \"CHUNG LEE: The Drunken Master\" poster and turned it into a cinematic motion poster — moving characters, smoke, sparks, liquid, dynamic lighting, camera movement, and animated typography.\n\nBut the bigger idea is \"attention\" created by Minimax H3!\n\nA static poster can be scrolled past in a second. Add movement, and suddenly there’s a reason to stop and see what happens next.\n\nThat can mean more time spent with the content, more engagement, and potentially stronger signals for the algorithm.\n\nImagine doing this with your YouTube thumbnails, TikTok/Instagram posts, Facebook promos, movie posters or campaign visuals.\n\nOne static design → a whole new motion asset.\n\nCreated with Minimax H3 in @higgsfield",
    "engagement": {
      "replies": 53,
      "reposts": 40,
      "likes": 225,
      "views": 2797,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "Original X caption and creator reply explicitly describe the CHUNG LEE poster, preserved composition, motion effects, and animated title treatment."
  },
  {
    "id": "x-2087131708223046088",
    "title": "曼哈顿街头的人形蛋糕碰撞",
    "titleEn": "Human Cakes Collide on a Manhattan Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "第一人称手机镜头记录街头车辆撞向行人后出现人形蛋糕质感的荒诞连续场面。 原生视频规格为 15 秒 · 2752×1536。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A shaky first-person phone camera follows absurd street impacts that transform people into human-cake forms. Native-video output: 15s · 2752×1536. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Shaky handheld smartphone footage, pure first-person view, bright clear Manhattan street under shiny midday sun.The camera jerks, sways and shakes heavily like a real phone held by someone walking backward in panic. Sharp sunlight, hard shadows, clean blue sky above brick buildings and dry asphalt. Strong natural motion blur and smartphone digital noise.\nContinuous 20-second sequence (new subject + vehicle every 5 seconds):\n[0s–5s] Operator walks backward from a white man in a black agent suit and sunglasses. The operator’s hand silently points urgently behind the man. The man turns his head, eyes widening in shock as he sees a bright red sports car drifting sideways at high speed toward him. The car slides in at an angle with screeching tires and physically slams into his body first — the impact throws him hard against the hood. Only after the physical hit does his body dissipate into a chaotic spray of tiny randomized fragments — almost pure thick cream mixed with very small irregular cake crumbs, no large pieces. The intricate cream explosion splatters in all directions, coating the red hood, windshield and street in messy streaks and droplets. The suit is ripped off by the force and either sticks to the car or drops to the ground. Cream heavily sprays across the lens.\n[5s–10s] Camera keeps shaking and moving back. A Black woman in a business blazer. Operator silently points hard behind her. She turns, face reacting with sudden fear as a dark blue SUV drifts sharply toward her. The SUV locks brakes mid-drift and physically smashes into her body first. Immediately after the impact her body dissipates into an intricate burst of tiny randomized particles — mostly thick chocolate cream with only minuscule dark cake crumbs. The fine cream spray coats the blue SUV surface and asphalt in dense, chaotic patterns. The blazer and skirt are torn away, parts sticking to the vehicle mixed with cream, the rest dropping naturally. Chocolate cream heavily splatters the lens.\n[10s–15s] Operator still walking backward, phone shaking. An Asian man in a hoodie. Operator silently points behind him. The man turns and freezes in panic seeing a white delivery van sliding into a wide drift. The van physically crashes into his body first. Right after the hit he dissipates into a highly randomized spray of tiny fragments — almost entirely cream with very small scattered sponge and fruit particles. The intricate cream explosion covers the van and pavement in messy droplets and streaks. The hoodie and jeans are stripped off, some cloth remaining on the van mixed with cream, the rest falling with natural physics. Cream hits the lens hard.\n[15s–20s] Camera continues with heavy smartphone shake. A Latina woman in a trench coat. Operator silently points urgently behind her. She spins around, eyes wide in terror as a black pickup truck performs a long cinematic drift straight at her. The truck physically hits her body first. Only then does her body dissipate into a fine, chaotic spray of tiny randomized particles — mostly thick caramel cream with only minute golden crumbs. The intricate cream blast coats the truck hood, grille and ground in dense, irregular patterns. The trench coat is ripped off, sections sticking to the truck with cream, the rest falling realistically. Cream and tiny fragments cascade over the vehicle and street, heavily streaking the lens.\nThroughout: constant heavy smartphone-style camera shake and jitter, natural motion blur, dense cream sprays and tiny randomized particles hitting the lens, realistic cloth physics, cinematic vehicle drifts, clear physical body impact first followed by dissipation into almost pure cream with only very small irregular cake crumbs (never large pieces or whole shapes), photorealistic, bright clear daylight, raw live smartphone footage look. No text, no logos, no stabilisation.",
    "promptSourceUrl": "https://x.com/sadlemonjuice/status/2087131708223046088",
    "sourceUrl": "https://x.com/sadlemonjuice/status/2087131708223046088",
    "sourceLabel": "X 原帖 · @sadlemonjuice",
    "author": "Socratech",
    "publishedAt": "2026-08-11T10:59:10.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2087131708223046088.mp4",
    "posterUrl": "/posters/x/2087131708223046088.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2752×1536",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [
      "combat",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Human Cakes MiniMax h3 local\n\nPrompt:\n\nShaky handheld smartphone footage, pure first-person view, bright clear Manhattan street under shiny midday sun.The camera jerks, sways and shakes heavily like a real phone held by someone walking backward in panic. Sharp sunlight, hard shadows, clean blue sky above brick buildings and dry asphalt. Strong natural motion blur and smartphone digital noise.\nContinuous 20-second sequence (new subject + vehicle every 5 seconds):\n[0s–5s] Operator walks backward from a white man in a black agent suit and sunglasses. The operator’s hand silently points urgently behind the man. The man turns his head, eyes widening in shock as he sees a bright red sports car drifting sideways at high speed toward him. The car slides in at an angle with screeching tires and physically slams into his body first — the impact throws him hard against the hood. Only after the physical hit does his body dissipate into a chaotic spray of tiny randomized fragments — almost pure thick cream mixed with very small irregular cake crumbs, no large pieces. The intricate cream explosion splatters in all directions, coating the red hood, windshield and street in messy streaks and droplets. The suit is ripped off by the force and either sticks to the car or drops to the ground. Cream heavily sprays across the lens.\n[5s–10s] Camera keeps shaking and moving back. A Black woman in a business blazer. Operator silently points hard behind her. She turns, face reacting with sudden fear as a dark blue SUV drifts sharply toward her. The SUV locks brakes mid-drift and physically smashes into her body first. Immediately after the impact her body dissipates into an intricate burst of tiny randomized particles — mostly thick chocolate cream with only minuscule dark cake crumbs. The fine cream spray coats the blue SUV surface and asphalt in dense, chaotic patterns. The blazer and skirt are torn away, parts sticking to the vehicle mixed with cream, the rest dropping naturally. Chocolate cream heavily splatters the lens.\n[10s–15s] Operator still walking backward, phone shaking. An Asian man in a hoodie. Operator silently points behind him. The man turns and freezes in panic seeing a white delivery van sliding into a wide drift. The van physically crashes into his body first. Right after the hit he dissipates into a highly randomized spray of tiny fragments — almost entirely cream with very small scattered sponge and fruit particles. The intricate cream explosion covers the van and pavement in messy droplets and streaks. The hoodie and jeans are stripped off, some cloth remaining on the van mixed with cream, the rest falling with natural physics. Cream hits the lens hard.\n[15s–20s] Camera continues with heavy smartphone shake. A Latina woman in a trench coat. Operator silently points urgently behind her. She spins around, eyes wide in terror as a black pickup truck performs a long cinematic drift straight at her. The truck physically hits her body first. Only then does her body dissipate into a fine, chaotic spray of tiny randomized particles — mostly thick caramel cream with only minute golden crumbs. The intricate cream blast coats the truck hood, grille and ground in dense, irregular patterns. The trench coat is ripped off, sections sticking to the truck with cream, the rest falling realistically. Cream and tiny fragments cascade over the vehicle and street, heavily streaking the lens.\nThroughout: constant heavy smartphone-style camera shake and jitter, natural motion blur, dense cream sprays and tiny randomized particles hitting the lens, realistic cloth physics, cinematic vehicle drifts, clear physical body impact first followed by dissipation into almost pure cream with only very small irregular cake crumbs (never large pieces or whole shapes), photorealistic, bright clear daylight, raw live smartphone footage look. No text, no logos, no stabilisation.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 7,
      "views": 2406,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "The original X post contains the complete prompt specifying the Manhattan setting, first-person phone camera, timed vehicle impacts, and human-cake transformations."
  },
  {
    "id": "x-2087117707816714552",
    "title": "1947 小镇的复古战地纪实",
    "titleEn": "A Vintage 1947 Town Under Fire",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "乡镇建筑、老式车辆、士兵与烟尘被组织成仿 1940 年代胶片质感的战地纪实段落。 原生视频规格为 15 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Period buildings, vintage cars, soldiers, smoke, and debris form a documentary-like 1940s film sequence. Native-video output: 15s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second ultra-photorealistic live-action war sequence set in the United States in 1947, designed to look like authentic historical footage captured on a 1940s film camera. The entire scene must feel grounded, documentary-like, raw, and physically realistic.\nEnvironment: A rural American town in 1947 with wooden houses, old brick buildings, telephone poles, dirt roads, vintage American cars from the 1940s, wooden fences, farmland, and period-accurate street details. Overcast afternoon light, light fog, drifting smoke, dust in the air, damaged buildings, scattered debris, and a tense wartime atmosphere.\nCharacters: American soldiers wearing historically accurate late-1940s military uniforms, helmets, boots, and equipment. Civilians wear authentic 1940s American clothing. Natural faces, realistic skin texture, sweat, dirt, fatigue, and believable body movements.\n0–3s — Establishing Shot:\nWide handheld shot of a quiet rural American street suddenly filled with smoke and confusion. Vintage 1940s vehicles are parked along the road while soldiers move quickly between wooden buildings. Civilians rush toward safer areas.\n3–6s — Tension:\nCamera moves through the street at shoulder height, following several soldiers as distant gunfire is heard. They immediately react and take cover behind a vintage vehicle and a brick wall. Their movements are cautious and realistic.\n6–10s — Combat:\nFast handheld tracking shot as the soldiers move between cover while distant gunfire impacts the environment. Small pieces of wood, dust, and debris fall naturally from nearby impacts. Weapon recoil, movement, and body weight must be physically accurate. Keep the violence realistic and restrained.\n10–13s — Human Moment:\nCamera briefly focuses on a soldier helping an injured civilian move behind cover. Their breathing, facial expressions, body language, and movement should feel natural and unscripted.\n13–15s — Final Shot:\nCamera pulls back into a wide shot of the American town as smoke slowly moves through the street. Soldiers remain behind cover while vintage vehicles and damaged buildings fill the background. The scene ends with an authentic, tense 1940s documentary feeling.\nVisual Style: Ultra-photorealistic live-action, authentic 1940s American environment, vintage 35mm film texture, subtle film grain, natural imperfections, realistic exposure, handheld documentary cinematography, muted historical color palette, realistic smoke and dust, natural shadows, accurate depth of field.\nPhysics: Strictly obey real-world gravity, momentum, inertia, friction, recoil, weight, collision physics, and human biomechanics. No exaggerated explosions, impossible movements, superhero behavior, or choreographed-looking combat.\nNegative Prompt: modern buildings, modern cars, smartphones, modern clothing, modern weapons, futuristic technology, CGI appearance, video-game graphics, fantasy, superhero action, excessive explosions, excessive blood, gore, impossible physics, unrealistic recoil, slow-motion physics, distorted faces, extra limbs, floating objects, plastic skin, artificial-looking environments.",
    "promptSourceUrl": "https://x.com/RuzainaMeer/status/2087117707816714552",
    "sourceUrl": "https://x.com/RuzainaMeer/status/2087117707816714552",
    "sourceLabel": "X 原帖 · @RuzainaMeer",
    "author": "Ruzaina",
    "publishedAt": "2026-08-11T10:03:32.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2087117707816714552.mp4",
    "posterUrl": "/posters/x/2087117707816714552.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "retro",
      "ugc"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing MiniMax H3 to create a realistic 1947 historical sequence with an authentic vintage film look.\n\nPrompt:\n\nCreate a 15-second ultra-photorealistic live-action war sequence set in the United States in 1947, designed to look like authentic historical footage captured on a 1940s film camera. The entire scene must feel grounded, documentary-like, raw, and physically realistic.\nEnvironment: A rural American town in 1947 with wooden houses, old brick buildings, telephone poles, dirt roads, vintage American cars from the 1940s, wooden fences, farmland, and period-accurate street details. Overcast afternoon light, light fog, drifting smoke, dust in the air, damaged buildings, scattered debris, and a tense wartime atmosphere.\nCharacters: American soldiers wearing historically accurate late-1940s military uniforms, helmets, boots, and equipment. Civilians wear authentic 1940s American clothing. Natural faces, realistic skin texture, sweat, dirt, fatigue, and believable body movements.\n0–3s — Establishing Shot:\nWide handheld shot of a quiet rural American street suddenly filled with smoke and confusion. Vintage 1940s vehicles are parked along the road while soldiers move quickly between wooden buildings. Civilians rush toward safer areas.\n3–6s — Tension:\nCamera moves through the street at shoulder height, following several soldiers as distant gunfire is heard. They immediately react and take cover behind a vintage vehicle and a brick wall. Their movements are cautious and realistic.\n6–10s — Combat:\nFast handheld tracking shot as the soldiers move between cover while distant gunfire impacts the environment. Small pieces of wood, dust, and debris fall naturally from nearby impacts. Weapon recoil, movement, and body weight must be physically accurate. Keep the violence realistic and restrained.\n10–13s — Human Moment:\nCamera briefly focuses on a soldier helping an injured civilian move behind cover. Their breathing, facial expressions, body language, and movement should feel natural and unscripted.\n13–15s — Final Shot:\nCamera pulls back into a wide shot of the American town as smoke slowly moves through the street. Soldiers remain behind cover while vintage vehicles and damaged buildings fill the background. The scene ends with an authentic, tense 1940s documentary feeling.\nVisual Style: Ultra-photorealistic live-action, authentic 1940s American environment, vintage 35mm film texture, subtle film grain, natural imperfections, realistic exposure, handheld documentary cinematography, muted historical color palette, realistic smoke and dust, natural shadows, accurate depth of field.\nPhysics: Strictly obey real-world gravity, momentum, inertia, friction, recoil, weight, collision physics, and human biomechanics. No exaggerated explosions, impossible movements, superhero behavior, or choreographed-looking combat.\nNegative Prompt: modern buildings, modern cars, smartphones, modern clothing, modern weapons, futuristic technology, CGI appearance, video-game graphics, fantasy, superhero action, excessive explosions, excessive blood, gore, impossible physics, unrealistic recoil, slow-motion physics, distorted faces, extra limbs, floating objects, plastic skin, artificial-looking environments.",
    "engagement": {
      "replies": 89,
      "reposts": 26,
      "likes": 377,
      "views": 32410,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "The original X post and full prompt explicitly specify a 1947 American town, period-accurate details, soldiers, war action, and vintage camera treatment.",
    "featured": {
      "at": "2026-09-05T15:50:15.681Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2087102541146522089",
    "title": "摄影师偶遇老人和小狗",
    "titleEn": "A Street Photographer Meets an Old Man and His Dog",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "街头摄影师发现咖啡馆外的老人和小狗，拍下瞬间后转身向镜头展示照片。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A street photographer notices an old man and his dog, captures the candid moment, then shows the photograph to camera. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A young Western female street photographer walks through a lively downtown street and notices an elderly man sitting outside a café with his small dog. She carefully composes the candid moment through her camera, captures the photo, then turns the camera toward the viewer to proudly show the shot she just took. She smiles, says “Look at that,” then continues walking through the city. Ultra-photorealistic visuals, natural handheld documentary movement, realistic camera interaction, authentic facial expressions, accurate hand movements, realistic dog behavior, natural daylight, cinematic depth of field, continuous character consistency, immersive city ambience, premium documentary realism.",
    "promptSourceUrl": "https://x.com/aiwithaly/status/2087102541146522089",
    "sourceUrl": "https://x.com/aiwithaly/status/2087102541146522089",
    "sourceLabel": "X 原帖 · @aiwithaly",
    "author": "Ali",
    "publishedAt": "2026-08-11T09:03:16.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2087102541146522089.mp4",
    "posterUrl": "/posters/x/2087102541146522089.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sometimes the best photos are the moments you never planned to capture.MiniMax H3 makes street photography feel incredibly real with natural movement, camera handling and authentic expressions.\n\nCreated with MiniMax H3.\n\nPrompt:\n\nA young Western female street photographer walks through a lively downtown street and notices an elderly man sitting outside a café with his small dog. She carefully composes the candid moment through her camera, captures the photo, then turns the camera toward the viewer to proudly show the shot she just took. She smiles, says “Look at that,” then continues walking through the city. Ultra-photorealistic visuals, natural handheld documentary movement, realistic camera interaction, authentic facial expressions, accurate hand movements, realistic dog behavior, natural daylight, cinematic depth of field, continuous character consistency, immersive city ambience, premium documentary realism.",
    "engagement": {
      "replies": 49,
      "reposts": 5,
      "likes": 183,
      "views": 10485,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "The creator’s original X caption and complete prompt describe the photographer, café encounter, dog, captured image, spoken line, and handheld movement."
  },
  {
    "id": "x-2087042771257540945",
    "title": "城市夜色里的银色耳机广告",
    "titleEn": "Silver Earbuds in the City at Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "银色耳机特写衔接城市步行场景，最后以干净的产品英雄镜头收束。 原生视频规格为 15 秒 · 1088×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Close-ups of silver earbuds flow into a stylish city walk and finish on a clean product hero shot. Native-video output: 15s · 1088×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a premium 15-second cinematic ad for Audionic Trance Airbud 850. Start with a close-up of the silver earbuds and case, then show a stylish Korean girl wearing them and walking through a luxurious Korean shopping street while enjoying a modern Korean K-pop song. Use beautiful city lights, soft bokeh, elegant fashion, natural expressions and smooth cinematic camera movement. End with one clean hero shot of the silver earbuds and case only. No phone screen, Bluetooth connection, discount text, logo end card or extra ending scene.",
    "promptSourceUrl": "https://x.com/ayzalnooor24521/status/2087042771257540945",
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2087042771257540945",
    "sourceLabel": "X 原帖 · @ayzalnooor24521",
    "author": "ayzalnoor",
    "publishedAt": "2026-08-11T05:05:46.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2087042771257540945.mp4",
    "posterUrl": "/posters/x/2087042771257540945.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1088×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A premium sound experience, wherever the city takes you.\nFeel the music, enjoy the moment, and move with style.\nMade for your everyday escape.\n\nCreated with MiniMax H3\n\nPrompt:\nCreate a premium 15-second cinematic ad for Audionic Trance Airbud 850. Start with a close-up of the silver earbuds and case, then show a stylish Korean girl wearing them and walking through a luxurious Korean shopping street while enjoying a modern Korean K-pop song. Use beautiful city lights, soft bokeh, elegant fashion, natural expressions and smooth cinematic camera movement. End with one clean hero shot of the silver earbuds and case only. No phone screen, Bluetooth connection, discount text, logo end card or extra ending scene.",
    "engagement": {
      "replies": 92,
      "reposts": 7,
      "likes": 223,
      "views": 16906,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "The original X post contains a complete product prompt specifying the earbuds, Korean shopping street, K-pop ambience, city lights, and final hero shot."
  },
  {
    "id": "x-2087042643369099381",
    "title": "《INK AFTER DARK》黑色电影片头",
    "titleEn": "INK AFTER DARK in Pulp-Noir Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "一位夜间写作者穿行于黑白拼色、撕纸转场与复古印刷纹理构成的片头段落。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A mysterious night writer moves through pulp-noir compositions, torn-paper transitions, and retro print textures. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[FORMAT]\nCreate a 15-second, 16:9 retro graphic opening-title sequence titled \"INK AFTER DARK\".\nA stylish 1960s pulp-noir animation about a mysterious night writer. Bold, minimal, elegant, slightly dangerous, with dry visual wit.\n\n[IDENTITY]\nKeep one female writer visually consistent throughout: sharp angular bob haircut, narrow almond-shaped eyes, long black turtleneck, high-waisted trousers, slim silhouette, black leather gloves, calm expression, upright self-assured posture, and hard graphic highlights along one side of her face.\nUse flat hand-inked 2D illustration with rough screen-print texture, imperfect paper grain, and slightly uneven ink edges.\n\n[BEATS]\n[0–3 seconds] Extreme close-up of the writer's face emerging from near-total black. A narrow cream-colored strip of light slides across her eyes. The frame abruptly opens sideways like torn paper, revealing her full silhouette walking across a burnt-red background while loose sheets of paper trail behind her like a long ribbon.\n\n[3–6 seconds] The paper ribbon sweeps across frame and becomes a graphic wipe. Reveal her in profile at a desk. She strikes one typewriter key. On impact, the screen fractures into three bold rectangular panels: her gloved fingers, the metal typebar snapping forward, and a giant black ink letter striking paper.\n\n[6–9 seconds] Rapid rhythmic close-ups: spinning typewriter ribbon spool, carriage return lever snapping sideways, black ink spreading through rough paper fibers. Thin cream lines cut diagonally across the screen and reorganize the images into an asymmetric editorial collage.\n\n[9–12 seconds] Pull wide. The writer stands alone beside an enormous abstract typewriter rendered as a black geometric silhouette. She pulls one endless page upward. The rising page becomes a full-frame cream vertical wipe while scattered black letters tumble downward like physical debris.\n\n[12–15 seconds] The letters rapidly assemble into the exact title \"INK AFTER DARK\" centered large on a burnt-red paper field. The writer's small black silhouette crosses beneath the title and exits frame. The title appears through sharp letter-by-letter mechanical impacts over 0.5 seconds, holds completely still through the ending. No bouncing, spinning, stretching, or fly-in typography.\n\n[CAMERA]\nUse aggressive graphic changes in scale: extreme facial close-up → full-body silhouette → macro mechanical inserts → monumental wide composition.\nCamera movement should feel designed rather than realistic: fast lateral pushes, sudden graphic crops, one rapid pull-out, and precise locked compositions. Avoid conventional cinematic orbit shots.\n\n[LIGHT]\nLimited palette only: burnt red, aged cream, deep black, with tiny muted silver highlights on typewriter metal.\nHard noir side-lighting translated into flat graphic shapes. Rough vintage print stock, subtle paper scratches, coarse ink grain, slight registration offsets, and occasional frame jitter.\n\n[EDIT]\nFast editorial rhythm with hard cuts, paper wipes, diagonal panel slices, oversized object masks, and split-screen recompositions.\nTransitions must be motivated by paper, ink, typewriter mechanisms, or moving silhouettes. No soft dissolves.\nMake every shot feel newly composed rather than simply zooming into the previous image.\n\n[AUDIO]\nAudio: dry typewriter key strikes, paper slides, ribbon-spool clicks, carriage-return snaps, faint room hum, and one heavy mechanical impact when the final title locks.\nBGM: an original 15-second cue, 65% noir suspense and 35% cool jazz. Upright bass, brushed snare, muted vibraphone, sparse low piano, and short clipped brass accents. Begin almost empty, introduce bass at 3 seconds, rhythmic percussion at 6 seconds, a brief brass accent at 10 seconds, then freeze the final 2 seconds on one bass note and the mechanical title hit. Do not imitate an existing melody.\n\n[NEGATIVE]\nNo subtitles, extra on-screen text, watermarks, platform logos, or stickers.\nDo not introduce Chinese text, garbled characters, misspellings, or additional title variations. Render \"INK AFTER DARK\" once only.\nNo 3D CGI, photorealism, anime styling, glossy modern motion graphics, neon cyberpunk aesthetics, or smooth vector-clean surfaces.\nNever a slideshow. Maintain active graphic motion, physical visual transitions, and continuously evolving compositions.",
    "promptSourceUrl": "https://x.com/opener_ai/status/2087042646397403306",
    "sourceUrl": "https://x.com/opener_ai/status/2087042643369099381",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-11T05:05:16.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2087042643369099381.mp4",
    "posterUrl": "/posters/x/2087042643369099381.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I just made an opening-title sequence using Hailuo Minimax H3. It's easily one of the most fun projects I'm working on lately. \n\nI was able to generate the video super fast using  @apiframe , which I've been using a lot these days. Being able to pick and choose different models right on the spot makes it feel like I've built my own custom toolkit.\n\nPrompt ⤵️\n\n#apiframe #minimaxH3 #sequence",
    "engagement": {
      "replies": 7,
      "reposts": 2,
      "likes": 28,
      "views": 1210,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "The creator reply publishes a complete shot-by-shot prompt for the INK AFTER DARK opening-title sequence, including identity lock, graphic style, and transitions."
  },
  {
    "id": "x-2087039444146933942",
    "title": "Haru 的高对比动漫开场",
    "titleEn": "Haru in a High-Contrast Anime Opening",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "角色 Haru 在霓虹配色、闪白、奔跑和节拍硬切中完成 15 秒动漫开场。 原生视频规格为 15 秒 · 1886×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Haru moves through neon color fields, white flashes, rooftop motion, and beat-synced cuts in a 15-second anime opening. Native-video output: 15s · 1886×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[CHARACTER LOCK — use attached reference sheet]\nHaru: teen boy, short messy dark-brown hair with warm amber rim-light on the\nedges, tired amber eyes, single silver earring on left ear, thin dark cord\nchoker, oversized grey-blue windbreaker open over a longline blue hoodie, black\njoggers with dangling straps, scuffed black high-top sneakers. Face, hair\nsilhouette, and outfit identical in every shot.\n\n[FORMAT]\nAnime television opening, 2D cel-shaded, crisp linework, 16:9, high contrast,\nfilm grain. Total runtime 15 seconds, 12 shots, beat-synced hard cuts.\n\n[SHOT LIST — 15s]\n00.0–01.2  Black. Hard white flash. Haru's silhouette snaps in, back to camera,\n           jacket billowing. Background floods CYAN.\n01.2–02.4  Whip-pan to profile close-up. Downcast eyes flick up, sharp.\n           Flash. Background MAGENTA.\n02.4–03.2  Low-angle wide: he sprints along a rooftop ledge, coat trailing,\n           paper and leaves scattering. ORANGE-RED, speed lines.\n03.2–03.9  Freeze-frame mid-leap between rooftops. Pure black silhouette\n           against strobing YELLOW-to-LIME.\n03.9–04.7  Top-down spin: lands, one hand on concrete, dust ring bursts out.\n           Snap to DEEP PURPLE.\n04.7–05.3  Extreme close-up: silver earring swings, catches a lens flare.\n           White flash frame.\n05.3–06.2  Split-screen triptych — three duotone copies of his face\n           (blue / red / green) rotating, one per beat.\n06.2–07.4  Back view, walking into wind, hood snapping, hair whipping.\n           Background ELECTRIC BLUE with horizontal scan lines.\n07.4–08.2  Hands-in-pockets front shot, camera orbits fast around him.\n           Colors cycle GREEN → PINK → WHITE across the orbit.\n08.2–09.4  Seated on the ledge, knees up, head down, city lights blooming\n           behind. Muted TEAL. Brief calm — no flash on this cut.\n09.4–10.6  He stands, turns to camera, hair lifting, single amber rim-light\n           igniting along his edge. Background goes pure WHITE.\n10.6–15.0  Pull-back wide: Haru walks away toward the sunset skyline, hands in\n           pockets, clothes drifting. Color settles into the warm painted\n           palette of the reference sheet. Camera dollies out, holds on\n           centered title-card space.\n\n[MOTION]\nSimple punchy character animation on twos. Hair and fabric always drifting.\nCamera: whip pans, snap zooms, one fast orbit, one final dolly-out.\n\n[COLOR & FX]\nFull-frame color swaps on every cut, synced to an implied beat. White\nflash-frames between shots 1–9. Chromatic aberration on impacts, halftone dot\nbursts, quick duotone inversions, floating leaf and paper debris throughout.\nWarm amber rim-light preserved on the character in every shot.\n\n[NEGATIVE]\nphotorealistic, 3D render, live action, extra characters, text or logos,\nchanging outfit, changing hair color, distorted face, extra fingers, watermark",
    "promptSourceUrl": "https://x.com/keneth_ai/status/2087039447473041914",
    "sourceUrl": "https://x.com/keneth_ai/status/2087039444146933942",
    "sourceLabel": "X 原帖 · @keneth_ai",
    "author": "Keneth Tran | AI Animation",
    "publishedAt": "2026-08-11T04:52:33.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2087039444146933942.mp4",
    "posterUrl": "/posters/x/2087039444146933942.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1886×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Another test with MiniMax H3.\n\nI've been exploring what this model is capable of creating, and I came across @Mayz1169's post about using MiniMax H3 to create a trailer. That immediately gave me the idea to try creating an anime opening.\n\nShoutout to @Mayz1169 for the inspiration!\n\nCreated using MiniMax H3 on @Hailuo_AI.\n\nI will include the prompt in the comment👇👇",
    "engagement": {
      "replies": 4,
      "reposts": 2,
      "likes": 17,
      "views": 1724,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "The original X caption identifies an anime opening, and the creator reply supplies a complete character lock, format, and timed shot list."
  },
  {
    "id": "x-2086943242713813383",
    "title": "无尽阶梯的黑白错视",
    "titleEn": "The Black-and-White Illusion of an Endless Staircase",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "人物沿极简阶梯前行，条纹、建筑和人体在黑白空间里逐步突破物理规则。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A figure climbs a minimal staircase as stripes, architecture, and the body gradually break physical rules in monochrome space. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "タイトル：ENDLESS STEP / 終わらない階段 尺：15秒 ジャンル：現代アート / Op Art / Surrealism 画面：16:9 カラー：白黒モノクロ中心 BGM：なし コンセプト：「進み続けているのに、どこにも辿り着いていない」 演出方針：前半は空間の異常、中盤は物理法則の崩壊、後半は人間と空間そのものの境界を崩す。ネオン・派手なグリッチ・大量の粒子は禁止。幾何学、錯視、建築、人体の融合だけで見せる。 CUT1｜0.0–1.8秒 Visuals｜何も存在しない真っ白な巨大空間。中央に一本だけ細長い黒い階段が奥へ伸びている。黒いミニマルな衣装の主人公が背中を向け、ゆっくり階段を上る。階段の表面だけが細い白黒ストライプになっている。最初は完全に静かな世界。 Camera｜24mm / F4。床ギリギリのローアングル。主人公を中央固定した完全シンメトリー構図。ゆっくり前方へドリーイン。 Motion｜主人公の歩行のみ。1歩につき約0.7秒。最後の0.5秒から階段表面のストライプがごく僅かに波打つ。 Lighting｜巨大な白いソフトボックスで空間全体を均一照明。人物と階段の下だけにシャープな黒い影。 VFX｜非常に弱い水平波状ディストーション。まだ異常だと断定できない程度。 SFX｜乾いた靴音「コツ、コツ」。極低音のルームトーン。 Emotion｜静寂。整いすぎていることへの違和感。 Transition｜足が次の段へ触れた瞬間にCUT2。 CUT2｜1.8–3.3秒 Visuals｜足元の極端なクローズアップ。靴底が階段へ触れるたび、白黒ストライプが水面のように沈み込み、波紋となって階段全体へ広がる。階段自体は固体なのに、模様だけが液体のように反応する。 Camera｜70mm / F2.8。階段より少し低い位置から足を追うローアングルトラッキング。 Motion｜足が接触 → 模様が沈む → 円状波紋が広がる → 次の一歩。足の動きはリアル、模様だけ非現実。 Lighting｜白背景を飛ばし気味にし、靴とストライプの黒を強調。 VFX｜Optical ripple、局所的なストライプ変形、軽いレンズ屈折。 SFX｜足音の直後に低い「ブゥン」という短い共鳴音。 Emotion｜世界が人物の存在に反応し始める。 Transition｜波紋が画面いっぱいへ広がりCUT3。 CUT3｜3.3–5.0秒 Visuals｜主人公の横顔ミディアム。背後の真っ白な壁に、横方向へ伸びる黒い階段がいつの間にか存在している。その階段を、もう一人の人物が壁面に対して垂直に歩いている。さらに天井には逆さになった人物が歩く。主人公本人はまだ気づいていない。 Camera｜50mm / F2.8。主人公の横顔にフォーカス。約0.7秒後、背景の異常な階段へラックフォーカス。 Motion｜主人公、壁の人物、天井の人物は全員まったく同じ歩行周期。ただし方向だけ異なる。 Lighting｜全人物へ同じ方向から光が当たっている。重力方向と影の方向が一致しない。 VFX｜Impossible perspective、非ユークリッド建築、重力不一致。 SFX｜正面の足音に、左・上方向から別の足音が薄く加わる。 Emotion｜「何かがおかしい」が明確になる瞬間。 Transition｜主人公が次の一段へ足を置く瞬間、低音と同時にCUT4。 CUT4｜5.0–6.7秒 Visuals｜足が一段に着地した瞬間、巨大空間全体が突然90°回転する。床が右側の壁へ変わり、天井だった面が床になる。しかし主人公は何事もなかったかのように同じ階段を上り続ける。階段の白黒ストライプが一気に巨大な同心円へ変形。 Camera｜18mm / F5.6。主人公正面寄りの超広角。空間回転と完全同期してカメラも90°ロール。その後、主人公だけ水平に見える角度で固定。 Motion｜空間：高速90°回転。主人公：速度一定。同心円：主人公から外側へ膨張。 Lighting｜回転開始時に0.1秒だけ光量低下。回転後は光源方向だけ元の位置を維持し、不自然な影を作る。 VFX｜90° spatial rotation、同心円ディストーション、短いモーションブラー、パース破綻。 SFX｜重い「ゴォン」＋低周波スイープ。 Emotion｜重力のルールが完全に壊れる。 Transition｜同心円の中心へカメラが吸い込まれCUT5。 CUT5｜6.7–8.5秒 Visuals｜超広角全景。上下左右、斜め、天井、壁へ無数の階段が伸びる巨大な不可能建築。人物は複数存在するが大量には増やさず、5〜7人程度。それぞれ異なる重力方向で歩いている。階段同士は交差しているように見えるが、実際には接続していない。中央の主人公だけが正常方向にいる。 Camera｜14mm / F5.6。空間中央を中心に高速オービット約100°。少しだけロールを加える。 Motion｜階段は静止。人物は一定速度で歩行。背景のストライプのみ逆方向へゆっくり移動。 Lighting｜完全モノクロ。白背景、黒い階段。影だけが異なる方向へ伸び、現実感を壊す。 VFX｜Non-Euclidean architecture、forced perspective、figure-ground ambiguity。 SFX｜異なる方向から足音が重なり、リズムが少しずつズレる。 Emotion｜空間そのものが巨大な錯視装置になった感覚。 Transition｜オービット中、主人公の顔がフレーム中央へ入りCUT6。 CUT6｜8.5–10.2秒 Visuals｜主人公の正面クローズアップ。主人公が初めて立ち止まる。顔の右側が模様になるのではなく、頬の表面そのものがゆっくり内側へ折れ始める。皮膚が建築物の壁のように変形し、顔の内部へ何百メートルも続く小さな白黒階段が現れる。片目、鼻、口の位置は正常なまま。顔の内部だけが巨大空間になっている。 Camera｜85mm / F2。顔正面。0.8秒かけてゆっくりドリーイン。最後の0.4秒で顔内部の階段へ急速接近。 Motion｜主人公は完全静止。まばたき一回。顔表面が紙を内側へ折るように段階的に陥没。内部階段の小さな人物だけが歩き続ける。 Lighting｜顔中央に柔らかい正面光。顔内部の空間はCUT1と同じ白い美術館照明。 VFX｜Face-to-architecture morph、recursive geometry、skin folding、深度拡張。顔への単純な模様貼り付けは禁止。 SFX｜周囲の足音が突然消える。顔が開く瞬間に紙を折るような乾いた音＋低い吸引音。 Emotion｜人体と建築の境界が崩れる。 Transition｜カメラが顔内部の階段へ完全に突入してCUT7。 CUT7｜10.2–12.5秒 Visuals｜顔内部の階段を歩く小さな主人公を追う。カメラが180°回転すると、これまで階段だと思っていた巨大な構造が、実は横向きになった人間の顔の輪郭だったと判明する。鼻梁が階段、唇の輪郭が通路、眼窩が巨大な円形空間になっている。さらにもう一度角度が変わると、再びただの階段に見える。「顔」と「建築」が視点によって交互に切り替わる。 Camera｜18mm / F4。人物後方から高速プッシュイン → 180°カメラロール → 一瞬静止 → ゆっくりドリーアウト。 Motion｜小さな主人公は歩行継続。巨大構造は動かさず、カメラ角度だけで顔と階段の認識が反転する。 Lighting｜均一な白黒照明。輪郭にだけ薄いサイドライトを入れ、顔として認識できるギリギリまで強調。 VFX｜Figure-ground reversal、anamorphic perspective、face/architecture ambiguity。モーフではなく「見る角度で意味が変わる」錯視を優先。 SFX｜短い逆再生音。顔として見えた瞬間だけ人間の呼吸音が一度入る。 Emotion｜「見えているものが何なのか分からない」状態。 Transition｜階段の最上部にある巨大な黒い円へ主人公が近づきCUT8。 CUT8｜12.5–15.0秒 Visuals｜主人公が階段の頂上へ到達。正面には巨大な完全な黒い円。その向こうは見えない。主人公が黒い円へ足を踏み入れる。瞬間、カメラが真上へ高速ドリーアウト。巨大な階段構造全体が一度ねじれ、メビウスの輪のようにつながっていることが判明。さらに引くと、先ほどの黒い円は実はCUT1で主人公が最初に踏み出した「階段の一段目」の黒い面だったことが分かる。構図がCUT1と完全一致。主人公が再び最初の一歩を踏み出す。 Camera｜35mmから開始。黒い円への短いプッシュイン → 真上視点へ急上昇 → 14mm相当の超ワイド → CUT1と同じ24mmローアングル構図へシームレス変形。 Motion｜階段全体がゆっくり180°ねじれてメビウス構造になる。主人公は止まらず歩き続ける。最後の0.5秒はCUT1と完全同一の動き。 Lighting｜最後まで完全モノクロ。白と黒のみ。ループ直前に一瞬だけ白黒を反転し、すぐ元へ戻す。 VFX｜Möbius topology transformation、recursive space、black-white inversion、seamless loop。 SFX｜黒い円へ入る直前に音が完全停止。CUT1の構図へ戻った瞬間、最初と同じ「コツ」という一歩目の足音。 Emotion｜「15秒間進んでいたのに、実は最初の一歩から抜け出せていなかった」。 END｜暗転なし。文字なし。ロゴなし。CUT1へ完全に戻るシームレスループ。 最重要生成ルール｜  1. 主役は常に同一人物。同じ黒い衣装・髪型・体格を固定。 2. 世界は白・黒のみ。サイバーパンク、未来都市、ネオンは禁止。 3. 階段の形状と白黒ストライプを全CUTで共通モチーフとして維持。 4. CUT6は「顔に模様を貼る」のではなく「顔内部が巨大建築空間になる」。 5. CUT7は単純な顔モーフではなく、視点によって顔と階段の認識が反転する錯視。 6. CUT8は必ずCUT1へ物理的につながるメビウス構造にする。 7. グリッチは使わない。AIらしい溶け・崩れ・余計な手足・余計な人物生成を避ける。 8. オプ・アートらしい幾何学的精度、直線、反復、同心円、白黒コントラストを優先する。\nMade with AI",
    "promptSourceUrl": "https://x.com/su_nagomi/status/2086943246937440396",
    "sourceUrl": "https://x.com/su_nagomi/status/2086943242713813383",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-10T22:30:17.000Z",
    "addedAt": "2026-08-20T18:59:16.815Z",
    "mediaUrl": "/media/x-2086943242713813383.mp4",
    "posterUrl": "/posters/x/2086943242713813383.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "オプ・アート × シュルレアリスム\n「ENDLESS STEP / 終わらない階段」\n#Hailuo のMinimax H3で動画作成\n動画プロンプトはリプ欄に👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 432,
      "snapshotAt": "2026-08-20T18:59:16.815Z"
    },
    "approvedAt": "2026-08-20T18:59:16.815Z",
    "editorialBasis": "The original X caption names ENDLESS STEP and the creator reply provides the full Japanese prompt with timed cuts, op-art geometry, surreal transformations, and sound direction."
  },
  {
    "id": "x-2086839288025710776",
    "title": "宫岛黄昏散步的橘猫",
    "titleEn": "A Ginger Cat Strolls Through Miyajima at Dusk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "橘猫沿宫岛海岸石道散步，鹿、枫树与海上鸟居共同构成温暖的九十年代动漫氛围。 原生视频规格为 8 秒 · 1056×608。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A ginger cat walks Miyajima’s coastal stone path among deer, maple trees, and the floating torii in a warm 1990s anime mood. Native-video output: 8s · 1056×608. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Style & Visual Direction:Classic 90s Japanese anime aesthetic, Hand-drawn anime style, Studio Ghibli-inspired warm tone, cel-shaded texturing, vibrant golden hour twilight lighting, soft painted backgrounds, natural fluid animation, high quality, masterpiece.\nScene Overview:A calm, peaceful afternoon on Miyajima Island. A ginger tabby cat wanders past the iconic floating Great Torii Gate standing in the serene water, enjoying a quiet stroll along the coastal stone path as deer rest nearby.\n\nStoryboard (8-Second Sequence):\n[0s-2s] Shot 1: Medium wide shot. The ginger tabby cat walks gracefully along the ancient stone lantern pathway near the shore, tail gently swaying. In the background, a friendly Miyajima deer rests peacefully under a maple tree.\n\n[2s-4.5s] Shot 2: Tracking shot. Low-angle side camera following the cat as it steps smoothly along the stone path. The iconic red Torii Gate is visible across the calm, shimmering water under a soft orange and pink sky.\n\n[4.5s-6.5s] Shot 3: Close-up shot. The cat pauses, turns its head toward the sea, and blinks slowly with curious amber eyes as the soft evening breeze rustles its fur.\n\n[6.5s-8s] Shot 4: Wide cinematic shot. The cat sits next to a classic stone lantern, gazing out at the floating Torii Gate reflected in the gently rippling tide.\n\nCamera:Smooth cinematic tracking, gentle pan and low-angle follow, clean cuts between shots on the natural beats.\n\nAudio:Gentle ocean waves lapping against the shore, quiet rustling leaves, soft ambient wind chimes, peaceful low piano melody carrying through to the end.\n\nNegative Prompts:No live-action photo style, no 3D CGI rendering, no text, no subtitles, no logos, no watermarks, no distorted anatomy, no abrupt jittery motion.",
    "promptSourceUrl": "https://x.com/sanasana0707/status/2086839289875398717",
    "sourceUrl": "https://x.com/sanasana0707/status/2086839288025710776",
    "sourceLabel": "X 原帖 · @sanasana0707",
    "author": "マツモトアキヒサ＠Broad Arts & リライトシステムズ",
    "publishedAt": "2026-08-10T15:37:12.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086839288025710776.mp4",
    "posterUrl": "/posters/x/2086839288025710776.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "時間はかかるけどNVIDIA GeForce RTX 4070 Laptop GPU (8 GB) + RAM32GBでもConfyUI + Minimax H3で動画作れそうだ。\nお題「宮島を散歩する猫」使ったプロンプトはリプのやつ。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 487,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X caption and creator reply explicitly describe the Miyajima cat walk, eight-second storyboard, anime direction, deer, and torii."
  },
  {
    "id": "x-2086831315559854556",
    "title": "舞台歌手的双模型同题对照",
    "titleEn": "One Stage Performance Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "同一位金发歌手、同一舞台和同一段完整描述被用于双模型输出对照。 原生视频规格为 27 秒 · 2160×2880。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The same blonde performer, arena stage, and complete direction are used for a side-by-side model comparison. Native-video output: 27s · 2160×2880. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "— A young Caucasian woman, 177 cm tall, with long, straight, silky light-blonde hair and neat, airy bangs; she has light gray, clear eyes and delicate, cool-looking features. She is wearing a white, fitted, short, pleated V-neck spaghetti-strap tank top, blue fitted jeans, a brown leather belt with a large vintage-style silver buckle, and dark brown leather knee-high boots.\n — white wireless microphone.\n — round stage in the center of the arena, surrounded by crowd on all sides, LED screen above.\n — music track, exactly this track must play throughout the entire scene.\n\nMira performing on stage to the track — not singing, moving like a rock star, dynamic chaotic movements, aggressively moving to the beat, sharply headbanging. White microphone in hand. Not looking at the camera. Shooting only mira — super close-up, cuts allowed. No zooms — camera always physically right next to mira on stage, heavily tilted, handheld. Dark cinematic lighting — top light, soft warm light barely touching her face, deep shadows below. Realistic skin. ARRI Alexa 35, anamorphic lenses. No 3D, no VFX.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2086831315559854556",
    "sourceUrl": "https://x.com/ou_zhen599/status/2086831315559854556",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-08-10T15:05:31.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086831315559854556.mp4",
    "posterUrl": "/posters/x/2086831315559854556.jpg",
    "duration": 27,
    "aspectRatio": "portrait",
    "resolution": "2160×2880",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "😟The MiniMax H3 really got me down.\n\nA comparison of the MiniMax H3 and Seedance2 with multiple parameters.\nThe MiniMax H3 touts 2K resolution as a selling point, so it costs significantly more than the Seedance2.😣\n\nBut as it turns out, the Seedance2 still holds its own!🤔\n\nMiniMax H3打着2K的画质噱头，积分就比Seedance2要贵很多。\n但实际证明Seedance2依然能打啊！\n\n Video prompt👇\n​ — A young Caucasian woman, 177 cm tall, with long, straight, silky light-blonde hair and neat, airy bangs; she has light gray, clear eyes and delicate, cool-looking features. She is wearing a white, fitted, short, pleated V-neck spaghetti-strap tank top, blue fitted jeans, a brown leather belt with a large vintage-style silver buckle, and dark brown leather knee-high boots.\n — white wireless microphone.\n — round stage in the center of the arena, surrounded by crowd on all sides, LED screen above.\n — music track, exactly this track must play throughout the entire scene.\n\nMira performing on stage to the track — not singing, moving like a rock star, dynamic chaotic movements, aggressively moving to the beat, sharply headbanging. White microphone in hand. Not looking at the camera. Shooting only mira — super close-up, cuts allowed. No zooms — camera always physically right next to mira on stage, heavily tilted, handheld. Dark cinematic lighting — top light, soft warm light barely touching her face, deep shadows below. Realistic skin. ARRI Alexa 35, anamorphic lenses. No 3D, no VFX.",
    "engagement": {
      "replies": 12,
      "reposts": 0,
      "likes": 11,
      "views": 731,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X post explicitly presents a MiniMax H3 and Seedance comparison and publishes the shared performer, stage, music, camera, and lighting prompt."
  },
  {
    "id": "x-2086754531698725200",
    "title": "云海凝成 MORNING",
    "titleEn": "Clouds Form the Word MORNING at Sunrise",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "日出穿过山峰云海，气流把云层塑成 MORNING 后再随光线消散。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Sunrise pours over mountain clouds as air currents sculpt the word MORNING before it dissolves into light. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Epic cinematic aerial shot: Endless rolling clouds flow across towering mountain peaks during sunrise. The golden light transforms the clouds into rivers of glowing vapor. Air currents gradually sculpt the cloud formations into the gigantic word \"MORNING\", stretching naturally across the sky. As the sun rises higher, the cloud letters slowly dissolve into the atmosphere while brilliant light floods the frame. Interstellar meets National Geographic, majestic, emotionally powerful.",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2086754531698725200",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2086754531698725200",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-08-10T10:00:25.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086754531698725200.mp4",
    "posterUrl": "/posters/x/2086754531698725200.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One sunrise.\nBillions of clouds.\n\nOne cinematic title reveal. ☀️\nMinimax H3 keeps delivering beautiful VFX sequences.\n\nPROMPT 👇\nEpic cinematic aerial shot: Endless rolling clouds flow across towering mountain peaks during sunrise. The golden light transforms the clouds into rivers of glowing vapor. Air currents gradually sculpt the cloud formations into the gigantic word \"MORNING\", stretching naturally across the sky. As the sun rises higher, the cloud letters slowly dissolve into the atmosphere while brilliant light floods the frame. Interstellar meets National Geographic, majestic, emotionally powerful.",
    "engagement": {
      "replies": 12,
      "reposts": 4,
      "likes": 33,
      "views": 2613,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X post contains the complete prompt for the sunrise, mountain cloudscape, MORNING title formation, and dissolve."
  },
  {
    "id": "x-2086700416654594554",
    "title": "天才黑客 NEON 登场",
    "titleEn": "NEON the Hacker Steps Into the Megacity",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "粉发黑客 NEON 随音频节拍穿过霓虹巷道、地下设施与巨城屋顶，完成角色亮相。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Pink-haired hacker NEON moves to an audio reference through neon alleys, underground facilities, and a megacity rooftop. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "@[face ref] をNEONの顔、ピンクのボブヘア、黒い丸眼鏡の厳密な参照として使用します。\n@[body ref] をNEONの体格、黒と赤のサイバーパンク衣装、装備、ブーツの厳密な参照として使用します。\n@[audio ref] は映像のリズム、カットの強弱、主要な音楽アクセントの参照として使用します。\n\n15秒のハイエンドなサイバーパンク・キャラクター紹介映像。\n主役はNEONただ一人。\nクールで知的、危険な雰囲気。動作は自信に満ち、抑制されている。\n\n構成：\n\n0–3秒：IDENTITY REVEAL\n暗いネオン照明の空間。\nブーツ、赤いジャケットのディテール、眼鏡越しの目元を短いバーストカットで見せる。\n最後に顔のクローズアップ。\nカットを@[audio ref] の最初の強いアクセントに合わせる。\n\n3–6秒：NEON ALLEY\n雨に濡れた未来都市の路地。\nNEONがゆっくりカメラ方向へ歩く。\nローアングル、横方向のトラッキング、濡れた路面に映る赤いネオン。\n人物の動きよりもカメラワークを強調。\n\n6–9秒：TECH UNDERGROUND\n赤い警告灯が点滅する地下施設。\n肩越しのショット、手袋と装備のクローズアップ、短いオービットショット。\nNEONは静かに周囲を観察する。\n\n9–12秒：MEGACITY ROOFTOP\n巨大な未来都市を見下ろす夜の屋上。\n風でピンクの髪とジャケットがわずかに揺れる。\nワイドショットからローアングルのヒーローフレームへ切り替える。\n\n12–15秒：FINAL HERO REVEAL\n赤とマゼンタの逆光。\nカメラがNEONの周囲を短く回り込み、正面のミディアムクローズアップで止まる。\nNEONが眼鏡越しにカメラを見る。\n最後の音楽アクセントで力強い静止ヒーローフレーム。\n\n編集：\n合計12〜15ショット。\n主要カットを@[audio ref] の強拍に合わせる。\n環境変更は音楽フレーズの変化に合わせる。\n細かい半拍すべてにカットを入れず、強いアクセントを優先する。\n\n一貫性：\n同じ顔、同じ髪型、同じ眼鏡、同じ体格、同じ衣装を全編で維持。\n衣装の黒・赤・ゴールドの配色と装備位置を維持。\nキャラクターは常に一人だけ。\n\n禁止：\n複数のNEON、分身、衣装変更、髪型変更、眼鏡の消失、\n顔や身体のモーフィング、過剰なアクション、文字、ロゴ、字幕、UI、\nキャラクターシート風レイアウト、分割画面、白背景。\n-------------------------------------------------------",
    "promptSourceUrl": "https://x.com/Nokosu_kansoku/status/2086700951776493571",
    "sourceUrl": "https://x.com/Nokosu_kansoku/status/2086700416654594554",
    "sourceLabel": "X 原帖 · @Nokosu_kansoku",
    "author": "Nokosu",
    "publishedAt": "2026-08-10T06:25:23.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086700416654594554.mp4",
    "posterUrl": "/posters/x/2086700416654594554.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "TapNowで制作　@TapNow_AI #TapNow\n\nMiniMax H3で、15秒のAudioリファレンスのリズムに合わせたキャラクター紹介動画を作ってみました。今回はリリックなしです。\n\nキャラクターは、天才ハッカーのNEON。\n\n使用したプロンプトはリプ欄に置いています😆\nAudioリファレンスでエラーが出る場合は、表示上15秒でも小数点以下でわずかに超えていることがあるので、実際の尺をチェックしてみてください！",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 22,
      "views": 780,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X caption identifies the NEON character introduction, while the creator reply provides complete face, body, audio-reference, scene, and edit instructions."
  },
  {
    "id": "x-2086698076325253323",
    "title": "董事会收购谈判",
    "titleEn": "A High-Stakes Acquisition in the Boardroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "四位实业家围绕收购方案展开针锋相对的对白，镜头在日间董事会空间里逐步推进。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Four industrialists debate an acquisition as the camera slowly pushes through a daylight boardroom. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "cinematic 15-second ultra-realistic sequence inside a luxurious modern boardroom during daytime. Soft natural light enters through large glass windows. Four powerful industrialists in expensive tailored suits sit around a long polished dark wooden table. The atmosphere is tense and high-stakes.\n\nThe senior industrialist at the head of the table leans forward slightly and says firmly:\n“This acquisition will change the entire market. Either we move now, or someone else will.”\n\nAnother industrialist across the table adjusts his cufflinks and replies calmly:\n“Moving now is risky. The numbers still don’t support a full takeover.”A third industrialist places both hands on the table and speaks with intensity:\n“Risk is the price of control. \n\nThe investors are already waiting for our decision.”The camera slowly pushes in as the most powerful industrialist leans back, looks at everyone, and delivers the final line with quiet authority:\n“Then we take it. No more delays.”\n\nCinematic lighting, sharp details on suits and the wooden table, subtle tension in their faces, realistic skin textures, high-end corporate atmosphere, shallow depth of field, filmic look.",
    "promptSourceUrl": "https://x.com/adithatipalli/status/2086698079508729956",
    "sourceUrl": "https://x.com/adithatipalli/status/2086698076325253323",
    "sourceLabel": "X 原帖 · @adithatipalli",
    "author": "Adithya Thatipalli",
    "publishedAt": "2026-08-10T06:16:05.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086698076325253323.mp4",
    "posterUrl": "/posters/x/2086698076325253323.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing the dialogue version sequences with Minimax H3\n\nIt came out pretty good \n\nPrompt in replies",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 146,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X caption labels a dialogue sequence, and the creator reply publishes the complete boardroom setting, dialogue, camera, and performance direction."
  },
  {
    "id": "x-2086603936736170082",
    "title": "冰箱里的微型高速公路",
    "titleEn": "A Miniature Highway Hidden Inside the Fridge",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "男子打开冰箱后发现食物之间运行着微型公路，橙汁泄漏还引发施工和交通疏导。 原生视频规格为 10 秒 · 736×992。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A man opens his fridge to find a miniature highway weaving through food, complete with an orange-juice spill and road crews. Native-video output: 10s · 736×992. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description\n\nShot 1 A realistic suburban kitchen at night. A man casually walks to the refrigerator and opens the door. The camera starts just behind his shoulder and slowly pushes closer as bright refrigerator light illuminates his confused face.\nInside the refrigerator is an incredibly detailed miniature highway network integrated naturally between the food. Several lanes of tiny moving cars weave between milk cartons, jars, leftovers, vegetables, and drink bottles. Tiny headlights and brake lights glow realistically.\nThe camera moves closer into the refrigerator interior as if entering this tiny world.\n\nA miniature traffic jam forms beside a tipped container leaking orange juice across one lane like a flooded roadway. Tiny construction vehicles arrive. Small workers in reflective clothing place cones and redirect traffic around the spill.\n\nA tiny tow truck pulls a stalled car away. Another vehicle honks impatiently. Cars merge aggressively around a yogurt container.\n\nThe man slowly reaches one hand toward the highway. Several tiny cars immediately slam on their brakes. A miniature police officer begins angrily waving at his enormous finger to move away.\n\nThe man freezes, completely baffled.\n\nThe camera finishes extremely close to the refrigerator highway as traffic starts moving again around his stationary finger.\n\noverall_soundscape\n\nQuiet kitchen ambience, refrigerator hum, refrigerator door opening, tiny engines, miniature car horns, distant tiny sirens, construction equipment, subtle liquid splashing from the orange juice spill.\n\nnon_diegetic_music\n\nNone.",
    "promptSourceUrl": "https://x.com/LikeToasters/status/2086603938527162690",
    "sourceUrl": "https://x.com/LikeToasters/status/2086603936736170082",
    "sourceLabel": "X 原帖 · @LikeToasters",
    "author": "ILikeToasters",
    "publishedAt": "2026-08-10T00:02:00.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086603936736170082.mp4",
    "posterUrl": "/posters/x/2086603936736170082.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "736×992",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Time for some munchies.... Uh... Hailuo H3 local. Looks like the weird is back a bit on this timeline. \n\nPrompt ⤵️",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 14,
      "views": 460,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X caption and complete creator reply describe the nighttime kitchen, refrigerator highway, moving traffic, spill, workers, and sound design."
  },
  {
    "id": "x-2086419768047976597",
    "title": "疾驰列车顶上的爆炸对决",
    "titleEn": "An Explosive Fight on Top of a Moving Train",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "动作明星在疾驰列车顶部与对手交战，爆炸和多角度切换强化好莱坞动作片节奏。 原生视频规格为 10 秒 · 1152×640。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An action star fights an opponent atop a moving train as explosions and multiple camera angles drive the sequence. Native-video output: 10s · 1152×640. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a cinematic Hollywood movie scene, action adventure, explosions, Jackie Chan is fighting an opponent on top of a moving train, have multiple camera cuts and angles",
    "promptSourceUrl": "https://x.com/leuk_o/status/2086419768047976597",
    "sourceUrl": "https://x.com/leuk_o/status/2086419768047976597",
    "sourceLabel": "X 原帖 · @leuk_o",
    "author": "L. Oscar Plains",
    "publishedAt": "2026-08-09T11:50:11.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086419768047976597.mp4",
    "posterUrl": "/posters/x/2086419768047976597.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "There is just something so consistent and reliable with GGUF models that I always keep going back to.\n\n10-sec 1152x640 resolution using larryvrh v4 600-step Turbo LoRA (Euler/beta 8 steps), @realrebelai MiniMax H3 Q4_K_M GGUF\n\nTotal gen time: 29 minuets on a RTX 3060\n\nPrompt: Create a cinematic Hollywood movie scene, action adventure, explosions, Jackie Chan is fighting an opponent on top of a moving train, have multiple camera cuts and angles",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 209,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X post publishes the generation settings and a complete prompt specifying the moving train, rooftop fight, explosions, and multiple camera angles."
  },
  {
    "id": "x-2086412223061135392",
    "title": "风中角色的电影式登场",
    "titleEn": "A Cinematic Character Entrance in the Wind",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "镜头从服装与肢体细节向上推进，再以侧后方、剪影和英雄构图完成角色亮相。 原生视频规格为 15 秒 · 2410×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The camera rises from costume and body details into profile, silhouette, and final hero framing as the environment moves in the wind. Native-video output: 15s · 2410×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use @[char ref] as the sole character reference. Preserve the exact identity, face, body proportions, hairstyle, outfit, colors, materials and overall silhouette of the character throughout the entire video. Do not redesign, simplify or replace any defining visual features.\n\nCreate a cinematic character introduction focused on presence, silhouette, attitude and controlled motion.\n\n0–4s\nBegin with a close shot of a defining lower-body or detail element such as boots, shoes, feet, hands, clothing hem or an important accessory. The character enters frame or settles into position. The camera slowly tracks upward while hair, clothing and secondary elements move naturally in the wind or environment.\n\n4–8s\nReveal more of the body with a medium or medium-wide shot from the back, side or three-quarter angle. The character stands in a calm, composed way inside the environment. The camera makes a smooth orbit, arc or lateral move to gradually reveal the character’s face and silhouette.\n\n8–12s\nMove into a tight cinematic portrait or upper-body shot. The character performs one subtle signature action that fits their personality, such as lifting the chin, turning the head, adjusting clothing, brushing hair aside, opening a hand, looking toward camera, or shifting posture. Keep the motion minimal and intentional. The expression should match the character’s vibe.\n\n12–15s\nEnd with a strong full-body hero shot that clearly presents the entire design and silhouette. Use a low-angle, eye-level or slightly dramatic framing depending on the character’s personality. The character settles into a natural final pose and holds it confidently for a clean final reveal.\n\nVISUAL DIRECTION\nPremium cinematic presentation. Match the visual medium and rendering style of @[char ref]. Emphasize clean silhouette, elegant staging, subtle secondary motion, believable hair and cloth movement, strong composition, atmospheric depth and polished lighting. The scene should feel like a high-end anime, game or film character introduction.\n\nCAMERA\nUse a clear progression from detail reveal to partial reveal to face reveal to full-body hero reveal. Camera movement should be smooth, controlled and intentional. Avoid chaotic motion.\n\nENVIRONMENT\nPlace the character in a fitting environment that supports their identity and mood. The background should enhance the character without distracting from them.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2086412223061135392",
    "sourceUrl": "https://x.com/aimikoda/status/2086412223061135392",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-09T11:20:12.000Z",
    "addedAt": "2026-08-20T18:59:16.816Z",
    "mediaUrl": "/media/x-2086412223061135392.mp4",
    "posterUrl": "/posters/x/2086412223061135392.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "2410×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "fashion",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\nCharacter Introduction Prompt:\n\nUse @[char ref] as the sole character reference. Preserve the exact identity, face, body proportions, hairstyle, outfit, colors, materials and overall silhouette of the character throughout the entire video. Do not redesign, simplify or replace any defining visual features.\n\nCreate a cinematic character introduction focused on presence, silhouette, attitude and controlled motion.\n\n0–4s\nBegin with a close shot of a defining lower-body or detail element such as boots, shoes, feet, hands, clothing hem or an important accessory. The character enters frame or settles into position. The camera slowly tracks upward while hair, clothing and secondary elements move naturally in the wind or environment.\n\n4–8s\nReveal more of the body with a medium or medium-wide shot from the back, side or three-quarter angle. The character stands in a calm, composed way inside the environment. The camera makes a smooth orbit, arc or lateral move to gradually reveal the character’s face and silhouette.\n\n8–12s\nMove into a tight cinematic portrait or upper-body shot. The character performs one subtle signature action that fits their personality, such as lifting the chin, turning the head, adjusting clothing, brushing hair aside, opening a hand, looking toward camera, or shifting posture. Keep the motion minimal and intentional. The expression should match the character’s vibe.\n\n12–15s\nEnd with a strong full-body hero shot that clearly presents the entire design and silhouette. Use a low-angle, eye-level or slightly dramatic framing depending on the character’s personality. The character settles into a natural final pose and holds it confidently for a clean final reveal.\n\nVISUAL DIRECTION\nPremium cinematic presentation. Match the visual medium and rendering style of @[char ref]. Emphasize clean silhouette, elegant staging, subtle secondary motion, believable hair and cloth movement, strong composition, atmospheric depth and polished lighting. The scene should feel like a high-end anime, game or film character introduction.\n\nCAMERA\nUse a clear progression from detail reveal to partial reveal to face reveal to full-body hero reveal. Camera movement should be smooth, controlled and intentional. Avoid chaotic motion.\n\nENVIRONMENT\nPlace the character in a fitting environment that supports their identity and mood. The background should enhance the character without distracting from them.",
    "engagement": {
      "replies": 59,
      "reposts": 122,
      "likes": 1501,
      "views": 103183,
      "snapshotAt": "2026-08-20T18:59:16.816Z"
    },
    "approvedAt": "2026-08-20T18:59:16.816Z",
    "editorialBasis": "The original X post publishes a complete character-reference prompt with timed detail shots, camera movement, identity preservation, silhouette, wind, and final reveal.",
    "featured": {
      "at": "2026-09-05T15:50:15.694Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2086321605073105162",
    "title": "追逐午夜龙的碎月少年",
    "titleEn": "The Boy Who Chases a Midnight Dragon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "少年 Kael 目睹银月裂开、午夜龙衔走碎片，随即穿过湖畔与森林追向禁忌山谷。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Kael watches the silver moon crack, sees a midnight dragon carry away a shard, and pursues it toward a forbidden valley. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, cinematography, lighting, scale, prop, performance, and narrative reference.\n\nGenerate a FINISHED cinematic animated sequence from the storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard panel borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character-reference sketches\n- prop-reference drawings\n- title/header\n- page number\n- off-white storyboard-paper texture\n- camera-note boxes\n- color swatches\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact face and body design\n2. Kael’s swept-back black hairstyle\n3. Kael’s dark tunic and trousers\n4. Nyx’s exact black midnight-dragon design\n5. Nyx’s long serpentine body\n6. Nyx’s giant amber-orange eyes\n7. Nyx’s luminous toothy grin\n8. the enormous silver full moon\n9. Noctara’s gothic castle architecture\n10. orange-lit castle windows\n11. reflective moonlit lake\n12. dark lakeside rock and forest geography\n13. exact glowing silver moon-fragment design\n14. deep midnight-blue visual palette\n15. Part 1 story progression\n16. final Valley of Bones cliffhanger\n\nThe finished video must feel like these storyboard paintings have come alive as a premium cinematic 2D animated short.\n\nIMPORTANT:\n\nThe storyboard contains 15 one-second panels.\n\nDO NOT create fifteen frantic one-second cuts.\n\nConsolidate them into approximately 8 cinematic shots with clear:\n\nsetup\n→ disturbance\n→ discovery\n→ dragon reveal\n→ misunderstanding\n→ chase\n→ forbidden-land cliffhanger.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:00–0:15 of the 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual style:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nmoonlit fantasy mystery / adventure\n\nAudience:\nfamily-friendly\n\nDialogue:\nnative synchronized English\n\nAudio:\nnative dialogue + cinematic environment + magical SFX + original music\n\nEMOTIONAL ARC:\n\npeaceful moonlit wonder\n→ subtle unease\n→ impossible celestial event\n→ shock\n→ mysterious creature reveal\n→ false villain assumption\n→ determination\n→ chase\n→ ominous curiosity\n\nCore Part 1 question:\n\n“Why did the dragon take the moon fragment?”\n\n--------------------------------------------------\n\n[PART 1 STORY LIMIT]\n\nPART 1 ONLY.\n\nSTART:\n\nKael sits peacefully beside the moonlit lake watching the enormous full moon above Noctara.\n\nEND:\n\nKael reaches the ridge overlooking the Valley of Bones after Nyx disappears inside with the moon fragment.\n\nThe cracked moon is visibly dimmer above him.\n\nKael decides to follow.\n\nDO NOT show:\n- Zara\n- skeleton guardians\n- Bonekeeper reveal\n- Nyx being innocent\n- “He’s trying to return it”\n- Moon Shrine\n- Bone Bridge\n- Nyx losing the shard\n- Kael catching the shard\n- shrine activation\n- restored moon\n- moon-fireflies\n- final friendship scene\n\nThose belong to later parts.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch the uploaded storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- youthful stylized angular face\n- cool pale-to-medium skin tone\n- large expressive dark eyes\n- thick black eyebrows\n- swept-back tousled black hair\n- pointed windswept hair silhouette\n- lean narrow shoulders\n- long thin limbs\n- small athletic teenage build\n\nClothing:\n- simple charcoal / black fantasy tunic\n- slightly loose short sleeves\n- dark fitted trousers\n- dark practical boots\n- subtle dark belt\n- no armor\n- no cape\n- no weapons\n\nPersonality:\n- quiet\n- thoughtful\n- curious\n- emotionally intense when surprised\n- brave\n- impulsive\n- deeply fascinated by the moon\n\nPART 1 PERFORMANCE:\n\n0–3 sec:\ncalm, reflective, mesmerized\n\n3–5 sec:\nuneasy → shocked\n\n5–7 sec:\ndisbelief\n\n7–9 sec:\nfear mixed with anger\n\n9–12 sec:\ndecisive pursuit\n\n12–15 sec:\ncautious but determined\n\nCRITICAL:\n\nMaintain exact:\n- face\n- hair\n- clothing\n- age\n- proportions\n- eye design\n- silhouette\n\nDo not make Kael older, more muscular, armored, or heroic-looking.\n\nHe is an ordinary boy who makes a brave decision.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch the uploaded storyboard EXACTLY.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- huge black dragon\n- elongated serpentine body\n- long narrow flexible neck\n- layered feather-like black scales\n- blue-black moonlit edge highlights\n- dramatic backward-pointing head spines\n- pointed ear-like horns\n- enormous glowing amber-orange eyes\n- vertical black slit pupils\n- large luminous white toothy grin\n- narrow expressive dragon snout\n- long curved tail\n- large dark wings matching his silhouette where visible\n\nPersonality in Part 1:\nUNKNOWN to Kael.\n\nVisually Nyx should seem:\n- mysterious\n- powerful\n- intimidating\n- slightly mischievous\n- unreadable\n\nBut do NOT portray true cruelty.\n\nSubtle clues may suggest he is protecting rather than stealing the fragment:\n- careful grip on shard\n- controlled flight\n- no attacks\n- no fire\n- no aggression toward Kael\n\nNYX MUST NOT SPEAK IN PART 1.\n\nHis intimidating appearance creates the misunderstanding.\n\n--------------------------------------------------\n\n[MOON FRAGMENT IDENTITY LOCK]\n\nMatch storyboard exactly.\n\nONE fragment only.\n\nAppearance:\n- irregular crescent / crystalline shard\n- brilliant white-silver center\n- translucent icy-blue edges\n- faint silver particles\n- small radiant aura\n- roughly large enough for Nyx to carry carefully in his jaws\n\nTimeline:\n\nfull moon cracks\n→ fragment separates\n→ begins falling\n→ Nyx intercepts\n→ Nyx carries same shard through chase\n\nDo NOT:\n- duplicate fragment\n- change shape between shots\n- let Nyx swallow it\n- turn it gold\n- create multiple moon pieces\n\n--------------------------------------------------\n\n[MOON IDENTITY / DAMAGE CONTINUITY]\n\nOpening:\nenormous perfect silver-blue full moon.\n\nAfter crack:\none obvious glowing fracture.\n\nAfter fragment breaks free:\nmissing wedge / damaged section remains visible.\n\nDuring chase:\nmoon becomes subtly dimmer.\n\nFinal shot:\nmoon is still large and recognizable, but clearly:\n- cracked\n- missing fragment\n- weaker in brightness\n\nDo NOT restore it in Part 1.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — NOCTARA]\n\nMaintain exact visual world from the storyboard.\n\nNOCTARA:\n\nA huge fantasy kingdom surrounding a moonlit lake.\n\nEnvironment:\n- towering gothic fantasy castles\n- tall narrow spires\n- ornate towers\n- warm orange glowing windows\n- dark blue stone architecture\n- dense shadowy trees\n- steep cliffs\n- reflective lake\n- rocky lakeshore\n- narrow forest paths\n- mountain silhouettes\n- clear star-filled night sky\n- enormous silver moon\n\nVisual contrast:\n\nMOON:\ncool silver-blue.\n\nCASTLES:\nwarm amber-orange.\n\nFOREST / WATER:\ndeep midnight blue.\n\nAtmosphere:\nbeautiful,\nmysterious,\ngrand,\nstorybook-dark,\nnever horror.\n\n--------------------------------------------------\n\n[VALLEY OF BONES — PART 1 REVEAL ONLY]\n\nOnly reveal its exterior / entrance.\n\nDesign:\n- deep forbidden canyon\n- tall jagged black-blue stone spires\n- pale mist\n- huge ancient fossil-like rib structures\n- distant mountain walls\n- twisted trees\n- barren rocky path\n- very subtle ivory bone-like shapes\n\nDo NOT show:\n- active skeleton guardians\n- Zara\n- full bone structures in close disturbing detail\n- gore\n- corpses\n\nThe valley should feel ancient and mysterious, not horrific.\n\n--------------------------------------------------\n\n[WORLD / PHYSICS LOGIC]\n\nPart 1 cause-and-effect must be crystal clear:\n\nKael watches moon\n→ moon cracks\n\nmoon cracks\n→ fragment breaks free\n\nfragment falls\n→ Nyx intercepts\n\nNyx catches fragment\n→ Kael assumes theft\n\nNyx turns toward mountains\n→ Kael decides to chase\n\nNyx enters forbidden valley\n→ Kael must choose whether to follow\n\nNo unexplained teleportation.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–2.20s\nNOCTARA BENEATH A PERFECT MOON\n\nOpen on a majestic ultra-wide establishing shot matching storyboard Panel 1.\n\nAn enormous silver full moon dominates the sky.\n\nBelow:\n\nNoctara’s tall gothic towers rise on both sides of a perfectly still lake.\n\nWarm amber-orange windows glow through the dark blue architecture.\n\nThe moon reflects in long silver ripples across the water.\n\nCamera slowly glides toward a dark lakeside rock.\n\nReveal tiny KAEL sitting alone.\n\nKael sits with knees raised, arms resting casually, staring at the moon.\n\nVery subtle breeze moves his hair and tunic.\n\nCamera:\nhuge kingdom wide\n→ slow dolly toward Kael\n→ settle behind him with the full moon framed ahead.\n\nNo dialogue initially.\n\nSound:\n- calm lake water\n- soft night breeze\n- distant forest insects\n- very faint city bells\n- distant night birds\n- quiet cloth movement\n\nMusic:\nsoft piano,\nairy flute,\nvery light strings,\nsubtle magical glass tones.\n\nThe feeling should be beautiful and peaceful.\n\n--------------------------------------------------\n\n[SHOT 2] — 2.20s–3.50s\nKAEL WATCHES THE MOON\n\nMove into an intimate side-profile / three-quarter close-up matching storyboard Panels 2–3.\n\nCool moonlight outlines Kael’s face.\n\nWarm orange castle lights blur softly in the background.\n\nHis eyes track the moon.\n\nHe seems comfortable here, as if he does this every night.\n\nKael quietly says:\n\n<d>[English] Still there.</d>\n\nTiny smile.\n\nThen the smile weakens.\n\nSomething feels slightly unusual.\n\nKael narrows his eyes.\n\nDialogue:\n\n<d>[English] ...Right?</d>\n\nCamera:\nmedium profile\n→ gentle push toward his eyes\n→ rack focus from Kael to moon.\n\nSound:\nbreeze becomes slightly quieter.\n\nMusic:\nremove a few warm notes,\nintroduce one subtle unresolved tone.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.50s–5.20s\nTHE MOON CRACKS\n\nCut to the giant moon.\n\nFor one silent fraction of a second:\n\nperfect.\n\nThen—\n\nA brilliant silver fracture appears.\n\nCRAAACK.\n\nNot an explosion.\n\nA glowing line spreads across part of the moon’s surface.\n\nKael rises quickly in the foreground.\n\nHis face goes from confusion to disbelief.\n\nA chunk of silver moonlight begins separating.\n\nDialogue:\n\nKael:\n<d>[English] What...?</d>\n\nThe moon fragment breaks free.\n\nIt begins falling through the night sky, trailing beautiful silver particles.\n\nKael:\n<d>[English] No!</d>\n\nCamera:\nmoon close-up\n→ crack spreading\n→ fragment separation\n→ tilt downward following falling shard\n→ Kael reaction.\n\nSound:\n- deep crystalline celestial crack\n- distant harmonic resonance\n- silver sparkling trail\n- sudden gust\n\nMusic:\nsharp orchestral/glass accent,\nthen urgent low strings begin.\n\nNo fiery explosion.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.20s–7.20s\nSOMETHING REACHES IT FIRST\n\nKael watches the shard fall.\n\nSuddenly a huge BLACK silhouette streaks across frame.\n\nThe camera briefly struggles to follow its speed.\n\nKael turns sharply.\n\nThe dark creature curves upward.\n\nReveal NYX.\n\nHis long black serpentine body arcs across the giant moon.\n\nHis amber eyes glow intensely.\n\nNyx reaches the fragment just before it can fall farther.\n\nHe catches it CAREFULLY between his jaws.\n\nSilver light illuminates his black scales from below.\n\nCamera:\nKael POV toward falling shard\n→ black silhouette enters\n→ dynamic pan\n→ wide Nyx arc against moon\n→ close-up catch.\n\nNo bite/crunch.\n\nThe fragment remains fully intact.\n\nSound:\n- strong but controlled wing rush\n- scale / feather movement\n- magical shard shimmer\n- low dragon breath\n- no roar\n\nKael, stunned:\n<d>[English] A dragon...?</d>\n\nMusic:\nmysterious low strings + brass swell,\nnot villain-horror music.\n\n--------------------------------------------------\n\n[SHOT 5] — 7.20s–8.80s\nTHE MIDNIGHT DRAGON\n\nClose cinematic reveal matching storyboard Panels 7–8.\n\nNyx turns his long head.\n\nFor a brief moment he looks toward Kael.\n\nFrame:\nhuge amber eyes,\nblack feathered scales,\nwild head spines,\nluminous exaggerated teeth,\nsilver fragment glowing between his jaws.\n\nThe visual should make him look like a villain from KAEL’S perspective.\n\nNyx’s enormous amber eye subtly narrows.\n\nBut he does NOT threaten Kael.\n\nHe simply turns away.\n\nKael’s expression changes:\n\nshock\n→ fear\n→ certainty.\n\nDialogue:\n\nKael:\n<d>[English] He took it...</d>\n\nNyx coils his body through the air and begins turning toward the mountains.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.80s–10.40s\n“HE STOLE THE MOON!”\n\nReaction close-up on Kael matching Panel 9.\n\nKael points toward Nyx.\n\nHis fear transforms into angry determination.\n\nDialogue:\n\nKael:\n<d>[English] HE STOLE THE MOON!</d>\n\nNyx accelerates toward the dark mountain valley.\n\nWide composition:\n\nNyx flying across the skyline,\nfragment glowing from his jaws,\nNoctara towers beneath him,\ncracked moon above.\n\nKael immediately jumps down from the rock.\n\nHe begins running.\n\nCamera:\nKael close reaction\n→ turn with his pointing arm\n→ follow Nyx\n→ snap back to Kael launching into pursuit.\n\nSound:\n- Kael footsteps beginning\n- night wind intensifying\n- Nyx wing movement\n- castle ambience fading behind\n\nMusic:\nadventure rhythm begins,\nstrings + low percussion + light brass.\n\n--------------------------------------------------\n\n[SHOT 7] — 10.40s–12.80s\nTHE CHASE BENEATH A BROKEN MOON\n\nCreate a fluid cinematic chase montage rather than several separate cuts.\n\nKael runs along the lakeside path.\n\nHis dark tunic and hair move with his sprint.\n\nAbove and ahead:\n\nNyx races between castle towers carrying the fragment.\n\nShow:\n\nKael running beside moonlit water\n→ Nyx passing beyond gothic spires\n→ Kael climbing a dark rocky path\n→ castle lights progressively farther behind\n→ trees becoming denser\n→ moon still cracked overhead.\n\nKael breathes hard.\n\nDialogue:\n\nKael:\n<d>[English] Stop!</d>\n\nNyx does not respond.\n\nKael:\n<d>[English] Give it back!</d>\n\nNyx turns once in flight, briefly glancing toward Kael.\n\nHe does not attack.\n\nHe continues toward the mountains.\n\nCamera:\nside tracking with Kael\n→ wide sky parallax with Nyx\n→ forward-running camera\n→ increasingly darker forest framing.\n\nSound:\n- running footsteps\n- branches\n- gravel\n- breath\n- wing beats\n- silver fragment hum\n\n--------------------------------------------------\n\n[SHOT 8] — 12.80s–15.00s\nTHE VALLEY OF BONES — CLIFFHANGER\n\nKael exits the darker forest path onto a high rocky ridge.\n\nHe stops.\n\nBreathing heavily.\n\nCamera slowly moves around him.\n\nReveal the enormous forbidden valley below.\n\nUse the visual language of storyboard Panels 13–15:\n\n- towering jagged stone spires\n- deep canyon\n- pale blue mist\n- enormous ancient rib-like fossil arches\n- narrow winding path\n- dark mountain walls\n- very distant cold landscape\n\nNyx becomes a small silhouette flying deeper into the valley.\n\nThe glowing moon fragment remains clearly visible in his jaws as a silver point of light.\n\nKael watches him disappear.\n\nThen he slowly looks upward.\n\nThe moon is visibly dimmer now.\n\nCracked.\n\nMissing its fragment.\n\nKael looks back into the forbidden valley.\n\nA quiet fear crosses his face.\n\nThen determination.\n\nDialogue:\n\nKael, quietly:\n<d>[English] You’re not getting away.</d>\n\nBeat.\n\nHe takes one step down the forbidden path.\n\nKael:\n<d>[English] I’m getting it back.</d>\n\nCamera:\nbehind Kael\n→ giant valley reveal\n→ Nyx disappearing\n→ tilt to broken moon\n→ return to Kael close-up\n→ slow pull wide as Kael enters valley.\n\nMusic:\nadventure rhythm fades into deep mysterious note.\n\nCUT TO BLACK.\n\nEND PART 1.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage-boy voice.\n\nAge:\napproximately mid-teens.\n\nVoice qualities:\n- youthful\n- thoughtful\n- emotionally expressive\n- initially quiet\n- clear English\n- no exaggerated heroic delivery\n\nPerformance progression:\n\n0–3 sec:\nsoft and reflective.\n\n3–5 sec:\nconfused / shocked.\n\n5–7 sec:\nbreathless disbelief.\n\n7–9 sec:\nsuspicion becoming certainty.\n\n9–12 sec:\nurgent determination.\n\n12–15 sec:\nfear controlled by resolve.\n\nIMPORTANT:\n\n“He stole the moon!”\n\nshould sound like Kael genuinely believes what he has just witnessed.\n\nNot comedic.\n\nNot melodramatically shouted.\n\n--------------------------------------------------\n\n[NYX AUDIO]\n\nNo spoken words in PART 1.\n\nUse:\n- large controlled breathing\n- deep throat resonance\n- scale / feather movement\n- powerful wing sweeps\n- small curious exhale when looking toward Kael\n\nDo NOT use:\n- roaring\n- evil laughter\n- growling threats\n- fire sounds\n\nNyx must remain ambiguous.\n\n--------------------------------------------------\n\n[MOON / MAGIC SOUND DESIGN]\n\nFull moon:\nsubtle glassy harmonic ambience.\n\nCrack:\ndeep celestial crystal fracture.\n\nFragment:\nhigh silver shimmer,\nsoft harmonic resonance,\ntiny star particles.\n\nNyx catches fragment:\nlow warm resonant pulse +\nsilver chime.\n\nNo explosive boom.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nNOCTARA:\n\n- lake water\n- night wind\n- distant bells\n- faint city ambience\n- nocturnal insects\n- tree leaves\n- occasional distant bird\n\nCHASE:\n\n- footsteps\n- gravel\n- breath\n- clothing\n- faster wind\n- Nyx’s wings\n\nVALLEY:\n\n- wider empty wind\n- subtle echo\n- distant stone resonance\n- almost no wildlife\n\nThis sonic transition helps communicate leaving the safe kingdom.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL cinematic fantasy score.\n\nInstrumentation:\n\nOpening:\n- soft piano\n- airy flute\n- warm strings\n- glockenspiel / celesta\n\nMoon crack:\n- glass harmonics\n- low cello\n- subtle brass\n\nNyx reveal:\n- low strings\n- bass clarinet\n- restrained horns\n\nChase:\n- rhythmic strings\n- hand percussion\n- low drums\n- light brass pulses\n\nValley ending:\n- low sustained strings\n- distant choir texture without lyrics\n- sparse metallic chime\n\nMUSIC ARC:\n\n0.00–2.20\npeaceful wonder.\n\n2.20–3.50\nslight mystery.\n\n3.50–5.20\ncelestial rupture.\n\n5.20–7.20\nmysterious Nyx reveal.\n\n7.20–8.80\nsuspicion.\n\n8.80–12.80\nadventure/chase momentum.\n\n12.80–15.00\nominous mystery cliffhanger.\n\nNo horror score.\n\nNo trailer-style massive percussion.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPart 1 camera must move from calm cinematic observation into increasingly urgent adventure.\n\nCAMERA ARC:\n\n0–3.5 sec:\nslow,\nwide,\nstable,\ngraceful.\n\n3.5–5.2 sec:\nsudden celestial focus.\n\n5.2–8.8 sec:\ndynamic sky tracking + reaction close-ups.\n\n8.8–12.8 sec:\nfaster but smooth chase camera.\n\n12.8–15 sec:\nslow back down for giant valley reveal.\n\nUse:\n- slow dolly\n- subtle crane\n- elegant tilt\n- rack focus\n- giant moon compositions\n- character profile close-up\n- smooth aerial tracking of Nyx\n- side-running tracking\n- environmental parallax\n- dramatic wide reveal\n\nAvoid:\n- handheld shake\n- extreme fisheye\n- constant fast cuts\n- 360 spins\n- Dutch angles\n- chaotic action camera\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH THE STORYBOARD EXACTLY IN SPIRIT.\n\nUse:\n\n- delicate dark ink contours\n- painterly 2D shading\n- rich watercolor/gouache backgrounds\n- stylized expressive faces\n- cinematic blue-black environments\n- silver atmospheric moonlight\n- warm orange architecture lighting\n- layered depth\n- subtle soft bloom\n- painterly dragon scales\n- hand-crafted fantasy architecture\n\nThe finished film should feel like:\n\na premium illustrated fantasy movie brought to life.\n\nNOT:\n- photorealistic\n- 3D CGI\n- hyper-real dragon rendering\n- anime\n- flat vector\n- storyboard animatic\n- motion comic\n- slideshow\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\nOPENING:\n\nDeep navy sky.\nSilver moon.\nWarm orange castle lights.\n\nKael:\nmostly cool moonlight with subtle warm bounce.\n\nMOON CRACK:\n\nIncrease brilliant cool white-silver contrast.\n\nNYX REVEAL:\n\nNyx remains near-black,\nbut use:\n- blue rim light\n- silver shard under-light\n- intense amber eyes\n- luminous pale teeth\n\nCHASE:\n\nGradually reduce warm Noctara illumination as Kael leaves the kingdom.\n\nVALLEY:\n\nPalette becomes:\n- colder blue\n- gray\n- charcoal\n- pale cyan mist\n\nFinal composition:\nKael = dark silhouette\nfragment = tiny bright silver point far away\nmoon = damaged silver\nvalley = cold mysterious blue-gray.\n\n--------------------------------------------------\n\n[KAEL MOTION RULES]\n\nOpening:\nnatural seated breathing,\nsmall head movement,\nsubtle hair flutter.\n\nReaction:\nfast but believable rise to feet.\n\nRunning:\nnatural teenage sprint,\narms and legs follow realistic stylized physics,\ntunic and hair respond to motion.\n\nNo:\n- superhuman jumps\n- parkour stunts\n- magical powers\n- impossible speed.\n\n--------------------------------------------------\n\n[NYX MOTION RULES]\n\nNyx must feel large and graceful.\n\nUse:\n- long serpentine body follow-through\n- wings with convincing weight\n- slight lag through tail\n- flexible neck\n- smooth aerial banking\n- subtle head movements\n- natural inertia\n\nWhen catching shard:\n\nNyx approaches\n→ matches shard trajectory\n→ opens jaws\n→ gently catches\n→ closes jaw enough to secure it\n→ banks away.\n\nNo violent snapping.\n\n--------------------------------------------------\n\n[MOON FRAGMENT MOTION]\n\nFragment falling:\nfollow gravity with magical slight drift.\n\nParticle trail:\nsubtle silver sparks.\n\nOnce Nyx catches it:\nremain physically fixed and stable in his grip.\n\nDo not flicker in/out.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nKeep exactly:\n\nONE Kael.\nONE Nyx.\nONE moon fragment.\nONE moon.\n\nSame:\n- Kael design\n- Nyx design\n- castle architecture\n- moon shape\n- lake\n- forest\n- valley entrance\n- night palette\n\nMoon damage must persist after cracking.\n\nOnce Nyx catches fragment:\nhe retains it through all subsequent Part 1 shots.\n\nKael starts seated by lake.\n\nHe must physically stand and run.\n\nDo not teleport him to the valley.\n\nCastle lights gradually recede during chase.\n\n--------------------------------------------------\n\n[EMOTIONAL PERFORMANCE RULE]\n\nPart 1 intentionally creates a FALSE interpretation.\n\nThe viewer should understand why Kael believes Nyx stole the fragment.\n\nHowever, avoid proving Nyx evil.\n\nVisual balance:\n\nNyx looks frightening\nBUT\nbehaves carefully.\n\nKael sees:\ngiant black dragon + glowing moon shard + flying away.\n\nTherefore he assumes:\n“thief.”\n\nThis misunderstanding drives the next chapter.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard sheet.\nNo panel borders.\nNo timestamps.\nNo captions.\nNo subtitles.\nNo production notes.\nNo title cards.\nNo watermark.\nNo logos.\nNo brands.\n\nNo duplicate Kael.\nNo duplicate Nyx.\nNo duplicate fragment.\nNo second moon.\n\nNo Kael costume change.\nNo hairstyle drift.\nNo older Kael.\nNo armor.\nNo weapons.\n\nNo Nyx redesign.\nNo red dragon.\nNo fire breathing.\nNo horns appearing/disappearing.\nNo eye-color changes.\nNo short-neck Nyx.\nNo bulky Western-dragon redesign.\n\nNo Zara.\nNo skeleton guardians.\nNo Bone Bridge.\nNo Moon Shrine reveal.\n\nNo blood.\nNo gore.\nNo injury.\nNo combat.\nNo attack.\nNo horror.\nNo dead bodies.\n\nNo moon explosion.\nNo planet destruction.\nNo falling castle.\nNo destroyed city.\n\nNo photorealism.\nNo 3D CGI.\nNo malformed anatomy.\nNo extra fingers.\nNo face flicker.\nNo scale inconsistency.\nNo teleportation.\n\n--------------------------------------------------\n\n[FINAL 3-SECOND PRIORITY]\n\nThe final sequence must provide a strong PART 2 hook.\n\nKael reaches the ridge.\n\nHe looks down.\n\nReveal:\n\nTHE VALLEY OF BONES.\n\nJagged cliffs.\n\nPale mist.\n\nAncient giant rib-like fossil formations.\n\nNyx flies deeper inside.\n\nThe silver moon fragment glows clearly from his jaws.\n\nThen Kael looks upward.\n\nThe giant moon is:\n\ncracked,\nmissing its shard,\nnoticeably dimmer.\n\nKael understands time may matter.\n\nHe looks back into the valley.\n\nHis fear becomes resolve.\n\nKael says:\n\n<d>[English] You’re not getting away.</d>\n\nThen:\n\n<d>[English] I’m getting it back.</d>\n\nHe takes his first step into the valley.\n\nCamera slowly pulls far backward.\n\nKael becomes tiny against the massive forbidden landscape.\n\nThe damaged moon hangs above everything.\n\nNyx’s distant silver shard-light disappears deeper into the canyon.\n\nCUT TO BLACK.\n\nEND PART 1.\n\nThe viewer should immediately want to know:\n\n“What is inside the Valley of Bones?”\n\nand\n\n“Did Nyx really steal the moon?”",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086321605073105162",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086321605073105162",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T05:20:07.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086321605073105162.mp4",
    "posterUrl": "/posters/x/2086321605073105162.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, cinematography, lighting, scale, prop, performance, and narrative reference.\n\nGenerate a FINISHED cinematic animated sequence from the storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard panel borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character-reference sketches\n- prop-reference drawings\n- title/header\n- page number\n- off-white storyboard-paper texture\n- camera-note boxes\n- color swatches\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact face and body design\n2. Kael’s swept-back black hairstyle\n3. Kael’s dark tunic and trousers\n4. Nyx’s exact black midnight-dragon design\n5. Nyx’s long serpentine body\n6. Nyx’s giant amber-orange eyes\n7. Nyx’s luminous toothy grin\n8. the enormous silver full moon\n9. Noctara’s gothic castle architecture\n10. orange-lit castle windows\n11. reflective moonlit lake\n12. dark lakeside rock and forest geography\n13. exact glowing silver moon-fragment design\n14. deep midnight-blue visual palette\n15. Part 1 story progression\n16. final Valley of Bones cliffhanger\n\nThe finished video must feel like these storyboard paintings have come alive as a premium cinematic 2D animated short.\n\nIMPORTANT:\n\nThe storyboard contains 15 one-second panels.\n\nDO NOT create fifteen frantic one-second cuts.\n\nConsolidate them into approximately 8 cinematic shots with clear:\n\nsetup\n→ disturbance\n→ discovery\n→ dragon reveal\n→ misunderstanding\n→ chase\n→ forbidden-land cliffhanger.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:00–0:15 of the 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual style:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nmoonlit fantasy mystery / adventure\n\nAudience:\nfamily-friendly\n\nDialogue:\nnative synchronized English\n\nAudio:\nnative dialogue + cinematic environment + magical SFX + original music\n\nEMOTIONAL ARC:\n\npeaceful moonlit wonder\n→ subtle unease\n→ impossible celestial event\n→ shock\n→ mysterious creature reveal\n→ false villain assumption\n→ determination\n→ chase\n→ ominous curiosity\n\nCore Part 1 question:\n\n“Why did the dragon take the moon fragment?”\n\n--------------------------------------------------\n\n[PART 1 STORY LIMIT]\n\nPART 1 ONLY.\n\nSTART:\n\nKael sits peacefully beside the moonlit lake watching the enormous full moon above Noctara.\n\nEND:\n\nKael reaches the ridge overlooking the Valley of Bones after Nyx disappears inside with the moon fragment.\n\nThe cracked moon is visibly dimmer above him.\n\nKael decides to follow.\n\nDO NOT show:\n- Zara\n- skeleton guardians\n- Bonekeeper reveal\n- Nyx being innocent\n- “He’s trying to return it”\n- Moon Shrine\n- Bone Bridge\n- Nyx losing the shard\n- Kael catching the shard\n- shrine activation\n- restored moon\n- moon-fireflies\n- final friendship scene\n\nThose belong to later parts.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch the uploaded storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- youthful stylized angular face\n- cool pale-to-medium skin tone\n- large expressive dark eyes\n- thick black eyebrows\n- swept-back tousled black hair\n- pointed windswept hair silhouette\n- lean narrow shoulders\n- long thin limbs\n- small athletic teenage build\n\nClothing:\n- simple charcoal / black fantasy tunic\n- slightly loose short sleeves\n- dark fitted trousers\n- dark practical boots\n- subtle dark belt\n- no armor\n- no cape\n- no weapons\n\nPersonality:\n- quiet\n- thoughtful\n- curious\n- emotionally intense when surprised\n- brave\n- impulsive\n- deeply fascinated by the moon\n\nPART 1 PERFORMANCE:\n\n0–3 sec:\ncalm, reflective, mesmerized\n\n3–5 sec:\nuneasy → shocked\n\n5–7 sec:\ndisbelief\n\n7–9 sec:\nfear mixed with anger\n\n9–12 sec:\ndecisive pursuit\n\n12–15 sec:\ncautious but determined\n\nCRITICAL:\n\nMaintain exact:\n- face\n- hair\n- clothing\n- age\n- proportions\n- eye design\n- silhouette\n\nDo not make Kael older, more muscular, armored, or heroic-looking.\n\nHe is an ordinary boy who makes a brave decision.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch the uploaded storyboard EXACTLY.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- huge black dragon\n- elongated serpentine body\n- long narrow flexible neck\n- layered feather-like black scales\n- blue-black moonlit edge highlights\n- dramatic backward-pointing head spines\n- pointed ear-like horns\n- enormous glowing amber-orange eyes\n- vertical black slit pupils\n- large luminous white toothy grin\n- narrow expressive dragon snout\n- long curved tail\n- large dark wings matching his silhouette where visible\n\nPersonality in Part 1:\nUNKNOWN to Kael.\n\nVisually Nyx should seem:\n- mysterious\n- powerful\n- intimidating\n- slightly mischievous\n- unreadable\n\nBut do NOT portray true cruelty.\n\nSubtle clues may suggest he is protecting rather than stealing the fragment:\n- careful grip on shard\n- controlled flight\n- no attacks\n- no fire\n- no aggression toward Kael\n\nNYX MUST NOT SPEAK IN PART 1.\n\nHis intimidating appearance creates the misunderstanding.\n\n--------------------------------------------------\n\n[MOON FRAGMENT IDENTITY LOCK]\n\nMatch storyboard exactly.\n\nONE fragment only.\n\nAppearance:\n- irregular crescent / crystalline shard\n- brilliant white-silver center\n- translucent icy-blue edges\n- faint silver particles\n- small radiant aura\n- roughly large enough for Nyx to carry carefully in his jaws\n\nTimeline:\n\nfull moon cracks\n→ fragment separates\n→ begins falling\n→ Nyx intercepts\n→ Nyx carries same shard through chase\n\nDo NOT:\n- duplicate fragment\n- change shape between shots\n- let Nyx swallow it\n- turn it gold\n- create multiple moon pieces\n\n--------------------------------------------------\n\n[MOON IDENTITY / DAMAGE CONTINUITY]\n\nOpening:\nenormous perfect silver-blue full moon.\n\nAfter crack:\none obvious glowing fracture.\n\nAfter fragment breaks free:\nmissing wedge / damaged section remains visible.\n\nDuring chase:\nmoon becomes subtly dimmer.\n\nFinal shot:\nmoon is still large and recognizable, but clearly:\n- cracked\n- missing fragment\n- weaker in brightness\n\nDo NOT restore it in Part 1.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — NOCTARA]\n\nMaintain exact visual world from the storyboard.\n\nNOCTARA:\n\nA huge fantasy kingdom surrounding a moonlit lake.\n\nEnvironment:\n- towering gothic fantasy castles\n- tall narrow spires\n- ornate towers\n- warm orange glowing windows\n- dark blue stone architecture\n- dense shadowy trees\n- steep cliffs\n- reflective lake\n- rocky lakeshore\n- narrow forest paths\n- mountain silhouettes\n- clear star-filled night sky\n- enormous silver moon\n\nVisual contrast:\n\nMOON:\ncool silver-blue.\n\nCASTLES:\nwarm amber-orange.\n\nFOREST / WATER:\ndeep midnight blue.\n\nAtmosphere:\nbeautiful,\nmysterious,\ngrand,\nstorybook-dark,\nnever horror.\n\n--------------------------------------------------\n\n[VALLEY OF BONES — PART 1 REVEAL ONLY]\n\nOnly reveal its exterior / entrance.\n\nDesign:\n- deep forbidden canyon\n- tall jagged black-blue stone spires\n- pale mist\n- huge ancient fossil-like rib structures\n- distant mountain walls\n- twisted trees\n- barren rocky path\n- very subtle ivory bone-like shapes\n\nDo NOT show:\n- active skeleton guardians\n- Zara\n- full bone structures in close disturbing detail\n- gore\n- corpses\n\nThe valley should feel ancient and mysterious, not horrific.\n\n--------------------------------------------------\n\n[WORLD / PHYSICS LOGIC]\n\nPart 1 cause-and-effect must be crystal clear:\n\nKael watches moon\n→ moon cracks\n\nmoon cracks\n→ fragment breaks free\n\nfragment falls\n→ Nyx intercepts\n\nNyx catches fragment\n→ Kael assumes theft\n\nNyx turns toward mountains\n→ Kael decides to chase\n\nNyx enters forbidden valley\n→ Kael must choose whether to follow\n\nNo unexplained teleportation.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–2.20s\nNOCTARA BENEATH A PERFECT MOON\n\nOpen on a majestic ultra-wide establishing shot matching storyboard Panel 1.\n\nAn enormous silver full moon dominates the sky.\n\nBelow:\n\nNoctara’s tall gothic towers rise on both sides of a perfectly still lake.\n\nWarm amber-orange windows glow through the dark blue architecture.\n\nThe moon reflects in long silver ripples across the water.\n\nCamera slowly glides toward a dark lakeside rock.\n\nReveal tiny KAEL sitting alone.\n\nKael sits with knees raised, arms resting casually, staring at the moon.\n\nVery subtle breeze moves his hair and tunic.\n\nCamera:\nhuge kingdom wide\n→ slow dolly toward Kael\n→ settle behind him with the full moon framed ahead.\n\nNo dialogue initially.\n\nSound:\n- calm lake water\n- soft night breeze\n- distant forest insects\n- very faint city bells\n- distant night birds\n- quiet cloth movement\n\nMusic:\nsoft piano,\nairy flute,\nvery light strings,\nsubtle magical glass tones.\n\nThe feeling should be beautiful and peaceful.\n\n--------------------------------------------------\n\n[SHOT 2] — 2.20s–3.50s\nKAEL WATCHES THE MOON\n\nMove into an intimate side-profile / three-quarter close-up matching storyboard Panels 2–3.\n\nCool moonlight outlines Kael’s face.\n\nWarm orange castle lights blur softly in the background.\n\nHis eyes track the moon.\n\nHe seems comfortable here, as if he does this every night.\n\nKael quietly says:\n\n<d>[English] Still there.</d>\n\nTiny smile.\n\nThen the smile weakens.\n\nSomething feels slightly unusual.\n\nKael narrows his eyes.\n\nDialogue:\n\n<d>[English] ...Right?</d>\n\nCamera:\nmedium profile\n→ gentle push toward his eyes\n→ rack focus from Kael to moon.\n\nSound:\nbreeze becomes slightly quieter.\n\nMusic:\nremove a few warm notes,\nintroduce one subtle unresolved tone.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.50s–5.20s\nTHE MOON CRACKS\n\nCut to the giant moon.\n\nFor one silent fraction of a second:\n\nperfect.\n\nThen—\n\nA brilliant silver fracture appears.\n\nCRAAACK.\n\nNot an explosion.\n\nA glowing line spreads across part of the moon’s surface.\n\nKael rises quickly in the foreground.\n\nHis face goes from confusion to disbelief.\n\nA chunk of silver moonlight begins separating.\n\nDialogue:\n\nKael:\n<d>[English] What...?</d>\n\nThe moon fragment breaks free.\n\nIt begins falling through the night sky, trailing beautiful silver particles.\n\nKael:\n<d>[English] No!</d>\n\nCamera:\nmoon close-up\n→ crack spreading\n→ fragment separation\n→ tilt downward following falling shard\n→ Kael reaction.\n\nSound:\n- deep crystalline celestial crack\n- distant harmonic resonance\n- silver sparkling trail\n- sudden gust\n\nMusic:\nsharp orchestral/glass accent,\nthen urgent low strings begin.\n\nNo fiery explosion.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.20s–7.20s\nSOMETHING REACHES IT FIRST\n\nKael watches the shard fall.\n\nSuddenly a huge BLACK silhouette streaks across frame.\n\nThe camera briefly struggles to follow its speed.\n\nKael turns sharply.\n\nThe dark creature curves upward.\n\nReveal NYX.\n\nHis long black serpentine body arcs across the giant moon.\n\nHis amber eyes glow intensely.\n\nNyx reaches the fragment just before it can fall farther.\n\nHe catches it CAREFULLY between his jaws.\n\nSilver light illuminates his black scales from below.\n\nCamera:\nKael POV toward falling shard\n→ black silhouette enters\n→ dynamic pan\n→ wide Nyx arc against moon\n→ close-up catch.\n\nNo bite/crunch.\n\nThe fragment remains fully intact.\n\nSound:\n- strong but controlled wing rush\n- scale / feather movement\n- magical shard shimmer\n- low dragon breath\n- no roar\n\nKael, stunned:\n<d>[English] A dragon...?</d>\n\nMusic:\nmysterious low strings + brass swell,\nnot villain-horror music.\n\n--------------------------------------------------\n\n[SHOT 5] — 7.20s–8.80s\nTHE MIDNIGHT DRAGON\n\nClose cinematic reveal matching storyboard Panels 7–8.\n\nNyx turns his long head.\n\nFor a brief moment he looks toward Kael.\n\nFrame:\nhuge amber eyes,\nblack feathered scales,\nwild head spines,\nluminous exaggerated teeth,\nsilver fragment glowing between his jaws.\n\nThe visual should make him look like a villain from KAEL’S perspective.\n\nNyx’s enormous amber eye subtly narrows.\n\nBut he does NOT threaten Kael.\n\nHe simply turns away.\n\nKael’s expression changes:\n\nshock\n→ fear\n→ certainty.\n\nDialogue:\n\nKael:\n<d>[English] He took it...</d>\n\nNyx coils his body through the air and begins turning toward the mountains.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.80s–10.40s\n“HE STOLE THE MOON!”\n\nReaction close-up on Kael matching Panel 9.\n\nKael points toward Nyx.\n\nHis fear transforms into angry determination.\n\nDialogue:\n\nKael:\n<d>[English] HE STOLE THE MOON!</d>\n\nNyx accelerates toward the dark mountain valley.\n\nWide composition:\n\nNyx flying across the skyline,\nfragment glowing from his jaws,\nNoctara towers beneath him,\ncracked moon above.\n\nKael immediately jumps down from the rock.\n\nHe begins running.\n\nCamera:\nKael close reaction\n→ turn with his pointing arm\n→ follow Nyx\n→ snap back to Kael launching into pursuit.\n\nSound:\n- Kael footsteps beginning\n- night wind intensifying\n- Nyx wing movement\n- castle ambience fading behind\n\nMusic:\nadventure rhythm begins,\nstrings + low percussion + light brass.\n\n--------------------------------------------------\n\n[SHOT 7] — 10.40s–12.80s\nTHE CHASE BENEATH A BROKEN MOON\n\nCreate a fluid cinematic chase montage rather than several separate cuts.\n\nKael runs along the lakeside path.\n\nHis dark tunic and hair move with his sprint.\n\nAbove and ahead:\n\nNyx races between castle towers carrying the fragment.\n\nShow:\n\nKael running beside moonlit water\n→ Nyx passing beyond gothic spires\n→ Kael climbing a dark rocky path\n→ castle lights progressively farther behind\n→ trees becoming denser\n→ moon still cracked overhead.\n\nKael breathes hard.\n\nDialogue:\n\nKael:\n<d>[English] Stop!</d>\n\nNyx does not respond.\n\nKael:\n<d>[English] Give it back!</d>\n\nNyx turns once in flight, briefly glancing toward Kael.\n\nHe does not attack.\n\nHe continues toward the mountains.\n\nCamera:\nside tracking with Kael\n→ wide sky parallax with Nyx\n→ forward-running camera\n→ increasingly darker forest framing.\n\nSound:\n- running footsteps\n- branches\n- gravel\n- breath\n- wing beats\n- silver fragment hum\n\n--------------------------------------------------\n\n[SHOT 8] — 12.80s–15.00s\nTHE VALLEY OF BONES — CLIFFHANGER\n\nKael exits the darker forest path onto a high rocky ridge.\n\nHe stops.\n\nBreathing heavily.\n\nCamera slowly moves around him.\n\nReveal the enormous forbidden valley below.\n\nUse the visual language of storyboard Panels 13–15:\n\n- towering jagged stone spires\n- deep canyon\n- pale blue mist\n- enormous ancient rib-like fossil arches\n- narrow winding path\n- dark mountain walls\n- very distant cold landscape\n\nNyx becomes a small silhouette flying deeper into the valley.\n\nThe glowing moon fragment remains clearly visible in his jaws as a silver point of light.\n\nKael watches him disappear.\n\nThen he slowly looks upward.\n\nThe moon is visibly dimmer now.\n\nCracked.\n\nMissing its fragment.\n\nKael looks back into the forbidden valley.\n\nA quiet fear crosses his face.\n\nThen determination.\n\nDialogue:\n\nKael, quietly:\n<d>[English] You’re not getting away.</d>\n\nBeat.\n\nHe takes one step down the forbidden path.\n\nKael:\n<d>[English] I’m getting it back.</d>\n\nCamera:\nbehind Kael\n→ giant valley reveal\n→ Nyx disappearing\n→ tilt to broken moon\n→ return to Kael close-up\n→ slow pull wide as Kael enters valley.\n\nMusic:\nadventure rhythm fades into deep mysterious note.\n\nCUT TO BLACK.\n\nEND PART 1.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage-boy voice.\n\nAge:\napproximately mid-teens.\n\nVoice qualities:\n- youthful\n- thoughtful\n- emotionally expressive\n- initially quiet\n- clear English\n- no exaggerated heroic delivery\n\nPerformance progression:\n\n0–3 sec:\nsoft and reflective.\n\n3–5 sec:\nconfused / shocked.\n\n5–7 sec:\nbreathless disbelief.\n\n7–9 sec:\nsuspicion becoming certainty.\n\n9–12 sec:\nurgent determination.\n\n12–15 sec:\nfear controlled by resolve.\n\nIMPORTANT:\n\n“He stole the moon!”\n\nshould sound like Kael genuinely believes what he has just witnessed.\n\nNot comedic.\n\nNot melodramatically shouted.\n\n--------------------------------------------------\n\n[NYX AUDIO]\n\nNo spoken words in PART 1.\n\nUse:\n- large controlled breathing\n- deep throat resonance\n- scale / feather movement\n- powerful wing sweeps\n- small curious exhale when looking toward Kael\n\nDo NOT use:\n- roaring\n- evil laughter\n- growling threats\n- fire sounds\n\nNyx must remain ambiguous.\n\n--------------------------------------------------\n\n[MOON / MAGIC SOUND DESIGN]\n\nFull moon:\nsubtle glassy harmonic ambience.\n\nCrack:\ndeep celestial crystal fracture.\n\nFragment:\nhigh silver shimmer,\nsoft harmonic resonance,\ntiny star particles.\n\nNyx catches fragment:\nlow warm resonant pulse +\nsilver chime.\n\nNo explosive boom.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nNOCTARA:\n\n- lake water\n- night wind\n- distant bells\n- faint city ambience\n- nocturnal insects\n- tree leaves\n- occasional distant bird\n\nCHASE:\n\n- footsteps\n- gravel\n- breath\n- clothing\n- faster wind\n- Nyx’s wings\n\nVALLEY:\n\n- wider empty wind\n- subtle echo\n- distant stone resonance\n- almost no wildlife\n\nThis sonic transition helps communicate leaving the safe kingdom.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL cinematic fantasy score.\n\nInstrumentation:\n\nOpening:\n- soft piano\n- airy flute\n- warm strings\n- glockenspiel / celesta\n\nMoon crack:\n- glass harmonics\n- low cello\n- subtle brass\n\nNyx reveal:\n- low strings\n- bass clarinet\n- restrained horns\n\nChase:\n- rhythmic strings\n- hand percussion\n- low drums\n- light brass pulses\n\nValley ending:\n- low sustained strings\n- distant choir texture without lyrics\n- sparse metallic chime\n\nMUSIC ARC:\n\n0.00–2.20\npeaceful wonder.\n\n2.20–3.50\nslight mystery.\n\n3.50–5.20\ncelestial rupture.\n\n5.20–7.20\nmysterious Nyx reveal.\n\n7.20–8.80\nsuspicion.\n\n8.80–12.80\nadventure/chase momentum.\n\n12.80–15.00\nominous mystery cliffhanger.\n\nNo horror score.\n\nNo trailer-style massive percussion.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPart 1 camera must move from calm cinematic observation into increasingly urgent adventure.\n\nCAMERA ARC:\n\n0–3.5 sec:\nslow,\nwide,\nstable,\ngraceful.\n\n3.5–5.2 sec:\nsudden celestial focus.\n\n5.2–8.8 sec:\ndynamic sky tracking + reaction close-ups.\n\n8.8–12.8 sec:\nfaster but smooth chase camera.\n\n12.8–15 sec:\nslow back down for giant valley reveal.\n\nUse:\n- slow dolly\n- subtle crane\n- elegant tilt\n- rack focus\n- giant moon compositions\n- character profile close-up\n- smooth aerial tracking of Nyx\n- side-running tracking\n- environmental parallax\n- dramatic wide reveal\n\nAvoid:\n- handheld shake\n- extreme fisheye\n- constant fast cuts\n- 360 spins\n- Dutch angles\n- chaotic action camera\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH THE STORYBOARD EXACTLY IN SPIRIT.\n\nUse:\n\n- delicate dark ink contours\n- painterly 2D shading\n- rich watercolor/gouache backgrounds\n- stylized expressive faces\n- cinematic blue-black environments\n- silver atmospheric moonlight\n- warm orange architecture lighting\n- layered depth\n- subtle soft bloom\n- painterly dragon scales\n- hand-crafted fantasy architecture\n\nThe finished film should feel like:\n\na premium illustrated fantasy movie brought to life.\n\nNOT:\n- photorealistic\n- 3D CGI\n- hyper-real dragon rendering\n- anime\n- flat vector\n- storyboard animatic\n- motion comic\n- slideshow\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\nOPENING:\n\nDeep navy sky.\nSilver moon.\nWarm orange castle lights.\n\nKael:\nmostly cool moonlight with subtle warm bounce.\n\nMOON CRACK:\n\nIncrease brilliant cool white-silver contrast.\n\nNYX REVEAL:\n\nNyx remains near-black,\nbut use:\n- blue rim light\n- silver shard under-light\n- intense amber eyes\n- luminous pale teeth\n\nCHASE:\n\nGradually reduce warm Noctara illumination as Kael leaves the kingdom.\n\nVALLEY:\n\nPalette becomes:\n- colder blue\n- gray\n- charcoal\n- pale cyan mist\n\nFinal composition:\nKael = dark silhouette\nfragment = tiny bright silver point far away\nmoon = damaged silver\nvalley = cold mysterious blue-gray.\n\n--------------------------------------------------\n\n[KAEL MOTION RULES]\n\nOpening:\nnatural seated breathing,\nsmall head movement,\nsubtle hair flutter.\n\nReaction:\nfast but believable rise to feet.\n\nRunning:\nnatural teenage sprint,\narms and legs follow realistic stylized physics,\ntunic and hair respond to motion.\n\nNo:\n- superhuman jumps\n- parkour stunts\n- magical powers\n- impossible speed.\n\n--------------------------------------------------\n\n[NYX MOTION RULES]\n\nNyx must feel large and graceful.\n\nUse:\n- long serpentine body follow-through\n- wings with convincing weight\n- slight lag through tail\n- flexible neck\n- smooth aerial banking\n- subtle head movements\n- natural inertia\n\nWhen catching shard:\n\nNyx approaches\n→ matches shard trajectory\n→ opens jaws\n→ gently catches\n→ closes jaw enough to secure it\n→ banks away.\n\nNo violent snapping.\n\n--------------------------------------------------\n\n[MOON FRAGMENT MOTION]\n\nFragment falling:\nfollow gravity with magical slight drift.\n\nParticle trail:\nsubtle silver sparks.\n\nOnce Nyx catches it:\nremain physically fixed and stable in his grip.\n\nDo not flicker in/out.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nKeep exactly:\n\nONE Kael.\nONE Nyx.\nONE moon fragment.\nONE moon.\n\nSame:\n- Kael design\n- Nyx design\n- castle architecture\n- moon shape\n- lake\n- forest\n- valley entrance\n- night palette\n\nMoon damage must persist after cracking.\n\nOnce Nyx catches fragment:\nhe retains it through all subsequent Part 1 shots.\n\nKael starts seated by lake.\n\nHe must physically stand and run.\n\nDo not teleport him to the valley.\n\nCastle lights gradually recede during chase.\n\n--------------------------------------------------\n\n[EMOTIONAL PERFORMANCE RULE]\n\nPart 1 intentionally creates a FALSE interpretation.\n\nThe viewer should understand why Kael believes Nyx stole the fragment.\n\nHowever, avoid proving Nyx evil.\n\nVisual balance:\n\nNyx looks frightening\nBUT\nbehaves carefully.\n\nKael sees:\ngiant black dragon + glowing moon shard + flying away.\n\nTherefore he assumes:\n“thief.”\n\nThis misunderstanding drives the next chapter.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard sheet.\nNo panel borders.\nNo timestamps.\nNo captions.\nNo subtitles.\nNo production notes.\nNo title cards.\nNo watermark.\nNo logos.\nNo brands.\n\nNo duplicate Kael.\nNo duplicate Nyx.\nNo duplicate fragment.\nNo second moon.\n\nNo Kael costume change.\nNo hairstyle drift.\nNo older Kael.\nNo armor.\nNo weapons.\n\nNo Nyx redesign.\nNo red dragon.\nNo fire breathing.\nNo horns appearing/disappearing.\nNo eye-color changes.\nNo short-neck Nyx.\nNo bulky Western-dragon redesign.\n\nNo Zara.\nNo skeleton guardians.\nNo Bone Bridge.\nNo Moon Shrine reveal.\n\nNo blood.\nNo gore.\nNo injury.\nNo combat.\nNo attack.\nNo horror.\nNo dead bodies.\n\nNo moon explosion.\nNo planet destruction.\nNo falling castle.\nNo destroyed city.\n\nNo photorealism.\nNo 3D CGI.\nNo malformed anatomy.\nNo extra fingers.\nNo face flicker.\nNo scale inconsistency.\nNo teleportation.\n\n--------------------------------------------------\n\n[FINAL 3-SECOND PRIORITY]\n\nThe final sequence must provide a strong PART 2 hook.\n\nKael reaches the ridge.\n\nHe looks down.\n\nReveal:\n\nTHE VALLEY OF BONES.\n\nJagged cliffs.\n\nPale mist.\n\nAncient giant rib-like fossil formations.\n\nNyx flies deeper inside.\n\nThe silver moon fragment glows clearly from his jaws.\n\nThen Kael looks upward.\n\nThe giant moon is:\n\ncracked,\nmissing its shard,\nnoticeably dimmer.\n\nKael understands time may matter.\n\nHe looks back into the valley.\n\nHis fear becomes resolve.\n\nKael says:\n\n<d>[English] You’re not getting away.</d>\n\nThen:\n\n<d>[English] I’m getting it back.</d>\n\nHe takes his first step into the valley.\n\nCamera slowly pulls far backward.\n\nKael becomes tiny against the massive forbidden landscape.\n\nThe damaged moon hangs above everything.\n\nNyx’s distant silver shard-light disappears deeper into the canyon.\n\nCUT TO BLACK.\n\nEND PART 1.\n\nThe viewer should immediately want to know:\n\n“What is inside the Valley of Bones?”\n\nand\n\n“Did Nyx really steal the moon?”",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 1879,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status publishes the complete Part 1 storyboard prompt, including Kael, Nyx, the broken moon, chase, dialogue, audio, and Valley of Bones ending."
  },
  {
    "id": "x-2086312209919963371",
    "title": "自动绘成的 1890 年代鸟类图鉴",
    "titleEn": "An 1890s Bird Plate Draws Itself",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "空白旧纸先出现铅笔稿，再叠加墨线和色彩，最后图鉴中的鸟悄然苏醒。 原生视频规格为 15 秒 · 1920×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An aged blank page gains pencil marks, engraved ink, and color before the illustrated bird quietly comes alive. Native-video output: 15s · 1920×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "@ Image1 is the sole visual authority for the folio spread, the wren, the hazel twig, the egg study, the foot sketch, the handwriting, the printed type, the paper, the foxing, the lighting, the palette, the layout and the horizontal composition. \n\nAnimate this exact page without redesigning it. Preserve every existing mark, letter and colour in its final state.  \n\n[Goal] Create a 15-second sequence in which this plate makes itself, and then the bird inside it comes quietly alive. One continuous, locked, straight-on overhead composition. No camera movement, no cuts, no cropping, no perspective change, no page turning.  \n\n[Sequence]  0-2.4 seconds: Begin on blank aged ivory paper — the empty spread, correct in paper texture, chain lines, foxing and lighting, but carrying no image, no writing and no type. A faint graphite underdrawing appears across the right page in light searching strokes, finding the wren's posture and the line of the twig. \nA few construction marks overshoot and are left uncorrected.  \n\n2.4-6 seconds: Engraved ink linework draws itself over the pencil, following the exact contours of the finished plate — outline first, then hatching building the shadow under the wing, the barring of the tail, the texture of the twig. The rectangular plate mark presses into the paper around the image with a soft physical impression, lifting the paper very slightly at its edge.  \n\n6-10.5 seconds: Watercolour floods in. Washes enter the printed outlines and settle into them, warm russet across the back, buff along the eyebrow stripe, muted olive on the twig, each wash landing exactly within the shapes of the finished plate and overrunning the line in the same few places the reference does. \n\nThe colour blooms wet and then visibly dries, darkening slightly at the edges. On the left page, the sepia handwriting writes itself line by line from the top, in a period cursive hand that is elegant, cramped and only partly legible. The small egg study fills with colour and the pencil foot sketch appears faintly beside it.  \n10.5-13 seconds: The printed type impresses into the right page — TROGLODYTES TROGLODYTES, then The Common Wren beneath it, then PLATE XVII in the top right corner. Foxing spots deepen. The last details resolve: the highlight in the bird's eye, the fine barring on the flank. The page settles. Everything on the spread is now exactly as in @[ref img].  13-15 seconds: The page stays completely still and completely flat \n\n— it is paper and it behaves like paper. But the printed bird begins to breathe. Its flank rises and falls very slightly. It blinks once. It turns its head a few degrees and holds, alert, looking past the viewer at something outside the page. The cocked tail gives one small twitch. It does not fly, it does not leave the plate, it does not become three-dimensional, and it does not look at the viewer. Settle into a living idle: the bird breathing, one more slow blink, the paper still.  Render as photoreal archival photography of a real object under warm north-window light. Keep the printed page flat and matte with no glow, no emission, no glint and no digital sheen. \n\nKeep the two motion layers distinct: the page is a made thing being made, the bird is alive. Keep all type, handwriting, marks and geometry fixed once they appear.  Do not introduce any new text, letters, numerals, marks, birds, objects, hands, tools, logos, captions, borders, watermarks or page numbers. Do not turn or lift the page. Do not move, rotate or scale the camera. Do not zoom. Do not add glow, neon, coloured light, lens flare, bokeh or shallow-focus racking. \n\nDo not make the bird photoreal or three-dimensional — it must remain a hand-coloured engraving that happens to be alive. Do not have it fly, hop, open its beak wide, or leave the plate.  \n\nAudio: fine graphite scratching on rough paper; the dry precise scrape of a steel nib; the deep soft compression of a plate press; a sable brush loaded with water moving across paper; paper fibres settling; the quiet room tone of a study with a slow long-case clock ticking somewhere behind. \nAt 13 seconds, one Eurasian wren song — sudden, intricate and far louder than anything that small should produce — then the room tone returns and the clock keeps going. No music, no dialogue, no narration, no electronic tones.",
    "promptSourceUrl": "https://x.com/Dheepanratnam/status/2086312213745197085",
    "sourceUrl": "https://x.com/Dheepanratnam/status/2086312209919963371",
    "sourceLabel": "X 原帖 · @Dheepanratnam",
    "author": "Dheepan Ratnam",
    "publishedAt": "2026-08-09T04:42:47.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086312209919963371.mp4",
    "posterUrl": "/posters/x/2086312209919963371.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An 1890s field guide plate, drawing itself.\n\nPencil, then ink, then the colour floods in.\n\n15 seconds, one continuous shot. MiniMax H3 \nPrompt and ref image in the comments ⤵️",
    "engagement": {
      "replies": 19,
      "reposts": 3,
      "likes": 53,
      "views": 1735,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X caption and same-author reply publish a complete prompt for the 1890s field-guide plate, locked overhead composition, drawing stages, and bird animation."
  },
  {
    "id": "x-2086294164409630916",
    "title": "屋顶上的 3,200 颗纸星星",
    "titleEn": "3,200 Paper Stars on a Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "小创作者 Milo 在夜色屋顶遇见巨型纸灯笼 Luma，并以发光纸星开启一段分章节故事。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Young creator Milo meets the giant paper lantern Luma on a nighttime rooftop and begins a chaptered story with glowing paper stars. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for “The Little Creator and the 3,200 Paper Stars” as the PRIMARY visual, character, environment, scale, prop, lighting, performance, and narrative reference.\n\nGenerate a FINISHED cinematic animated sequence from this storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- storyboard panel layout\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character-reference drawings\n- scale-reference diagrams\n- storyboard title/header\n- paper background\n- footer decorations\n\nThe uploaded storyboard locks:\n\n1. Milo’s exact character identity\n2. Milo’s dark-blue oversized hoodie\n3. Milo’s fluffy brown hair\n4. Milo’s glowing-star satchel\n5. Luma’s exact giant lantern design\n6. Luma’s warm cream/orange paper texture\n7. Luma’s sleepy face\n8. Luma’s hanging ribbons and tassels\n9. Pip’s exact firefly design\n10. extreme Milo-vs-Luma scale\n11. handmade glowing paper-star design\n12. same magical rooftop environment\n13. same distant nighttime city\n14. deep indigo + warm amber palette\n15. single-star placement story beat\n16. final full-satchel cliffhanger\n\nTranslate the 15 storyboard panels into approximately 8 coherent cinematic shots.\n\nDO NOT create fifteen one-second hard cuts.\n\nUse clear:\n\ncause → action → consequence → reaction.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory time:\n0:00–0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D animated short film\n\nGenre:\nmagical gratitude story / whimsical rooftop fantasy\n\nAudience:\nfamily-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + character sounds + rooftop ambience + magical paper-star effects + original music\n\nEMOTIONAL ARC:\n\nquiet rooftop wonder\n→ gratitude\n→ magical giant reveal\n→ tiny secret mission\n→ difficult climb\n→ first star placement\n→ beautiful success\n→ realization that thousands remain\n→ mischievous cliffhanger\n\nThe overall feeling should be:\n\nsmall creator,\nhuge dream,\nwarm gratitude.\n\n--------------------------------------------------\n\n[PART 1 STORY LIMIT]\n\nPART 1 ONLY.\n\nSTART:\n\nMilo sits alone on a peaceful rooftop beneath a star-filled night sky with a satchel containing glowing handmade paper stars.\n\nHe is celebrating an important milestone:\n\n3,200 supporters.\n\nEND:\n\nMilo has attached EXACTLY ONE glowing paper star to Luma.\n\nHe then looks back at his still-overflowing satchel.\n\nMilo gets a huge idea.\n\nHe whispers:\n\n<d>[English] Maybe... all of them?</d>\n\nPip realizes what Milo means.\n\nCUT TO BLACK.\n\nDO NOT show:\n- Milo attaching multiple stars\n- Luma fully decorated\n- hundreds of stars covering Luma\n- Luma waking\n- Luma flying higher\n- stars leaving Luma\n- sky formation\n- giant “3200”\n- “THANK YOU”\n- full milestone celebration\n- final hero shot\n\nThose belong to PART 2.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — MILO]\n\nMatch the uploaded storyboard EXACTLY.\n\nMILO:\n\nTiny young creator.\n\nAppearance:\n- warm light skin\n- soft rounded youthful face\n- large expressive dark-brown eyes\n- fluffy tousled medium-brown hair\n- slightly oversized head for appealing animation proportions\n\nClothing:\n- oversized dark navy-blue hoodie\n- dark/simple trousers\n- little shoes/boots\n- small brown crossbody satchel\n- satchel filled with glowing folded paper stars\n\nPersonality:\n- grateful\n- creative\n- curious\n- optimistic\n- emotional\n- slightly impulsive\n- playful\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nquietly grateful\n\n3–5 sec:\ncurious wonder\n\n5–7 sec:\ngets an idea\n\n7–10 sec:\ndetermined tiny climber\n\n10–12 sec:\ncareful and emotionally invested\n\n12–15 sec:\nproud → inspired → mischievously ambitious\n\nMilo should never feel arrogant.\n\nThe milestone means something because of the people represented by the stars.\n\nCRITICAL:\n\nKeep exact:\n- face\n- brown hair\n- navy hoodie\n- satchel\n- body proportions\n- age appearance\n- scale\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — LUMA]\n\nMatch the uploaded storyboard exactly.\n\nLUMA:\n\nA gigantic living floating sky lantern.\n\nAppearance:\n- enormous rounded lantern body\n- warm cream / parchment-orange paper surface\n- handmade subtle paper texture\n- decorative curved motifs\n- gentle closed sleepy eyes\n- tiny peaceful curved smile\n- soft amber-gold inner illumination\n- decorative frame underneath\n- multiple long hanging ribbons / tassels\n- huge scale relative to Milo\n\nPersonality:\n- peaceful\n- gentle\n- warm\n- sleepy\n- safe\n- quietly magical\n\nPART 1:\n\nLuma remains ASLEEP for the entire sequence.\n\nAllowed:\n- very subtle floating\n- gentle vertical bob\n- soft breathing-like light pulse\n- tiny peaceful smile response when first star is attached\n\nDo NOT:\n- open her eyes\n- speak\n- rise dramatically\n- intentionally move Milo\n- activate celebration magic\n\nHer awakening happens in PART 2.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — PIP]\n\nMatch the storyboard.\n\nPIP:\n\nTiny glowing firefly helper.\n\nAppearance:\n- small round yellow-gold body\n- pale luminous glow\n- large expressive eyes\n- tiny antennae\n- translucent wings\n- much smaller than Milo\n\nPersonality:\n- practical\n- cautious\n- loyal\n- funny\n- suspicious of Milo’s big ideas\n\nPerformance:\n\ncalm companion\n→ curious\n→ realizes Milo’s plan\n→ worried during climb\n→ nervous during star placement\n→ pleased that it works\n→ horrified by Milo’s final bigger idea\n\nPip may speak in short natural lines.\n\n--------------------------------------------------\n\n[GLOWING PAPER STAR LOCK]\n\nAll stars must match the storyboard.\n\nDesign:\n- handmade folded paper stars\n- cream / pale-gold paper\n- visible folded facets\n- warm golden light from inside\n- physical tactile objects\n- lightweight\n- slightly magical\n\nClose views should preserve real paper folds.\n\nAt the beginning:\nthe satchel contains many glowing stars.\n\nDuring PART 1:\nMilo removes ONLY ONE star for decoration.\n\nAt the final shot:\nthe satchel must still visibly overflow with stars.\n\n--------------------------------------------------\n\n[STAR SYMBOLISM]\n\nThe stars represent individual pieces of support from Milo’s audience/community.\n\nThe video should communicate this emotionally without showing social-media interfaces.\n\nNo:\n- phones\n- follower counters\n- social-media app screens\n- X logos\n- Twitter logos\n- platform branding\n\nMilo may verbally reference:\n\n“3,200”\n\nbut the milestone should primarily be represented by the glowing paper stars.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — ROOFTOP]\n\nMaintain the exact nighttime rooftop world from the storyboard.\n\nInclude:\n- cozy high rooftop\n- old rooftop stone/brick surfaces\n- chimney / small rooftop structures\n- a few potted plants\n- distant city buildings\n- warm glowing windows\n- layered rooftops\n- deep blue night sky\n- stars\n- crescent moon\n- subtle distant clouds\n- warm rooftop accent lights\n\nMood:\nquiet,\nsafe,\ndreamlike,\nintimate.\n\nThe city is far below.\n\nNo crowds.\n\nNo extra characters.\n\n--------------------------------------------------\n\n[SCALE RULES]\n\nExtreme scale contrast is critical.\n\nMilo is tiny compared with Luma.\n\nUse storyboard scale:\n\n- Milo smaller than one tassel knot\n- Luma’s lantern body towers above Milo\n- hanging ribbons are thick enough for Milo to climb like ropes\n- one section of Luma’s body is enormous compared with Milo’s full height\n\nPip is much smaller than Milo.\n\nNever accidentally render Milo human-sized beside Luma.\n\n--------------------------------------------------\n\n[WORLD / MOTION LOGIC]\n\nClear cause-and-effect:\n\nMilo looks at stars\n→ remembers milestone\n\nMilo looks upward\n→ notices sleeping Luma\n\nLuma’s huge blank lantern surface\n→ gives Milo decorating idea\n\nMilo wants to celebrate\n→ climbs ribbons\n\nMilo reaches lantern\n→ attaches ONE star\n\nStar lights up beautifully\n→ Milo feels successful\n\nMilo looks down\n→ sees full satchel\n\nFull satchel + enormous lantern\n→ Milo gets much bigger idea\n\nPip realizes what Milo is thinking\n→ comic panic cliffhanger\n\nNo random magic.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.80s\nA QUIET ROOFTOP AND 3,200 LITTLE LIGHTS\n\nOpen with a beautiful nighttime establishing shot.\n\nDeep indigo sky.\n\nCrescent moon.\n\nThousands of distant real stars.\n\nCamera slowly descends toward a peaceful rooftop overlooking a softly glowing city.\n\nMilo sits near the rooftop edge.\n\nBeside him is his brown satchel.\n\nWarm golden light leaks from inside.\n\nMilo opens the satchel.\n\nReveal many handmade glowing paper stars.\n\nHe gently picks ONE up with both hands.\n\nThe warm star light illuminates his face.\n\nPip floats quietly beside him.\n\nCamera:\nwide skyline\n→ slow dolly toward Milo\n→ overhead satchel insert\n→ intimate close-up on glowing star in his hands.\n\nDialogue:\n\nMilo, softly:\n<d>[English] Three thousand two hundred...</d>\n\nTiny pause.\n\nMilo smiles.\n\n<d>[English] That’s a lot of people.</d>\n\nPip gives a soft happy chirp.\n\nSound:\n- distant nighttime city\n- light rooftop breeze\n- cloth movement\n- satchel leather\n- subtle paper rustle\n- soft magical star hum\n\nMusic:\ngentle piano + glockenspiel + warm strings.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.80s–3.50s\nA REASON TO SAY THANK YOU\n\nMilo holds the glowing star near his chest.\n\nHis expression becomes warm and grateful.\n\nHe looks at the many remaining stars.\n\nDialogue:\n\nMilo:\n<d>[English] We should do something special.</d>\n\nPip tilts his head.\n\nThen Milo notices an enormous warm amber light above him.\n\nThe light softly brightens his face.\n\nMilo slowly looks upward.\n\nCamera follows his gaze.\n\nReveal the lower edge of Luma entering frame.\n\nLong decorative ribbons sway gently beneath her.\n\nMilo’s eyes widen.\n\nDialogue:\n\nMilo:\n<d>[English] Oh...</d>\n\nPip slowly turns upward too.\n\nSound:\n- gentle ribbon movement\n- low warm lantern hum\n- night breeze\n\nMusic:\nintroduce airy flute and soft wonder motif.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.50s–5.20s\nTHE SLEEPY SKY LANTERN\n\nCamera continues upward into a full magical reveal of Luma.\n\nLuma floats above the city.\n\nHuge.\n\nRound.\n\nWarmly illuminated.\n\nHer eyes remain peacefully closed.\n\nHer tiny smile is gentle.\n\nShe slowly bobs in the night air.\n\nMilo is tiny at the bottom of frame.\n\nThis shot must strongly emphasize SCALE.\n\nMilo looks at:\nthe glowing star in his hand...\n\nthen Luma’s enormous blank lantern surface.\n\nIdea forming.\n\nHis mouth curls into a grin.\n\nPip notices Milo’s expression.\n\nDialogue:\n\nMilo, whispering:\n<d>[English] Pip...</d>\n\nPip:\n<d>[English] What?</d>\n\nMilo:\n<d>[English] I have an idea.</d>\n\nPip immediately becomes suspicious.\n\nPip:\n<d>[English] That face worries me.</d>\n\nMusic:\nsmall mischievous plucked motif enters.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.20s–7.00s\nTHE SECRET STAR MISSION\n\nMilo walks beneath one long hanging Luma ribbon.\n\nHe grabs it with both hands.\n\nTests it.\n\nLooks upward.\n\nIt extends far above him.\n\nMilo puts the glowing star securely into his hoodie pocket / small pouch while keeping it visible and safe.\n\nThen starts climbing.\n\nPip hovers beside him.\n\nDialogue:\n\nMilo, whispering:\n<d>[English] We should celebrate.</d>\n\nPip:\n<d>[English] From the roof.</d>\n\nMilo climbs higher.\n\nMilo:\n<d>[English] Too ordinary.</d>\n\nPip:\n<d>[English] Of course.</d>\n\nCamera:\nlow-angle ribbon shot\n→ side tracking with Milo climbing\n→ pull wide showing tiny Milo beneath giant lantern.\n\nSound:\n- fabric ribbon strain\n- Milo’s tiny climbing effort\n- shoe taps\n- Pip wing buzz\n- soft Luma hum\n\n--------------------------------------------------\n\n[SHOT 5] — 7.00s–9.00s\nMILO VS. LUMA\n\nMilo reaches a large tassel knot.\n\nHe pauses and sits momentarily on it.\n\nThe knot itself is nearly Milo-sized.\n\nHe looks upward at the enormous lantern surface.\n\nPip floats beside him, slightly out of breath despite being a firefly.\n\nUse a vertical scale composition.\n\nMilo climbs the final short ribbon section.\n\nHe reaches Luma’s lower lantern edge.\n\nLuma remains completely asleep.\n\nDialogue:\n\nMilo:\n<d>[English] Almost there...</d>\n\nPip:\n<d>[English] Define “almost.”</d>\n\nMilo pulls himself onto a decorative rim.\n\nDialogue:\n\nMilo:\n<d>[English] There.</d>\n\nSound:\n- ribbon movement\n- tiny climb grunt\n- paper lantern creak\n- soft floating ambience\n\n--------------------------------------------------\n\n[SHOT 6] — 9.00s–11.00s\nTHE FIRST STAR\n\nMilo retrieves the ONE glowing folded star.\n\nHe looks at it.\n\nThen at Luma’s huge lantern body.\n\nHe carefully presses the handmade star onto the lantern using a tiny clip / adhesive tab.\n\nTake time with this.\n\nThe star gently settles.\n\nFor half a beat:\n\nnothing.\n\nThen its glow becomes slightly warmer.\n\nThe golden star light blends beautifully with Luma’s amber paper body.\n\nMilo’s face lights up.\n\nPip watches cautiously.\n\nLuma’s sleeping smile curves perhaps one tiny degree more.\n\nShe does NOT wake.\n\nDialogue:\n\nMilo, whispering:\n<d>[English] One little thank-you.</d>\n\nSoft click.\n\nMilo:\n<d>[English] Perfect.</d>\n\nSound:\n- paper touching paper\n- tiny clip click\n- warm magical “ting”\n- Luma’s gentle light pulse\n\nCamera:\nclose-up Milo’s hands\n→ macro star attachment\n→ reaction on Milo\n→ Luma’s peaceful sleepy expression.\n\nMusic:\ngentle emotional success note.\n\n--------------------------------------------------\n\n[SHOT 7] — 11.00s–12.80s\nONE VERY BIG SUCCESS\n\nMilo leans back against Luma’s decorative rim.\n\nHe admires the glowing star.\n\nThe single paper star looks beautiful against the huge warm lantern surface.\n\nPip slowly smiles.\n\nMilo:\n<d>[English] Look at that.</d>\n\nPip:\n<d>[English] Okay...</d>\n\nPip studies it.\n\n<d>[English] That is actually nice.</d>\n\nMilo beams.\n\nCamera pulls out slightly.\n\nClearly show:\n\ntiny Milo\n+\none glowing star\n+\nenormous mostly undecorated Luma.\n\nThe contrast is important.\n\n--------------------------------------------------\n\n[SHOT 8] — 12.80s–15.00s\nTHE MUCH BIGGER IDEA — CLIFFHANGER\n\nMilo’s proud smile suddenly pauses.\n\nHe remembers something.\n\nHe looks downward.\n\nCamera follows his gaze toward the rooftop.\n\nHis satchel remains there / hangs securely nearby according to spatial setup, visibly overflowing with glowing paper stars.\n\nAlternatively, if he carried the satchel during the climb, show it hanging securely from his shoulder and still overflowing.\n\nCRITICAL:\ndo not teleport the satchel.\n\nUse whichever configuration is established earlier and maintain it consistently.\n\nMilo slowly opens / looks into it.\n\nDozens upon dozens of glowing folded stars illuminate his face.\n\nThen he looks at:\n\nthe full satchel...\n\nthe ONE star on Luma...\n\nLuma’s ENORMOUS remaining lantern surface.\n\nMilo’s eyes widen.\n\nA mischievous smile grows.\n\nPip looks at Milo.\n\nPip looks at the satchel.\n\nPip looks at enormous Luma.\n\nPip’s expression turns to immediate horror.\n\nPip:\n<d>[English] Milo...</d>\n\nMilo raises the satchel slightly.\n\nMilo, delighted whisper:\n<d>[English] Maybe... all of them?</d>\n\nPip’s eyes become huge.\n\nPip:\n<d>[English] ALL?!</d>\n\nLuma continues peacefully sleeping.\n\nThe ONE attached star glows softly.\n\nHold on:\n\nMilo = enormous inspiration  \nPip = comic panic  \nLuma = peacefully unaware  \n\nCUT TO BLACK.\n\nEND PART 1.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[MILO VOICE]\n\nNatural young creator voice.\n\nAge feeling:\nyoung / youthful, not adult.\n\nQualities:\n- warm\n- sincere\n- curious\n- imaginative\n- emotionally grateful\n- playful\n- clear English\n\nDo not make Milo sound boastful.\n\nHis line:\n\n<d>[English] Three thousand two hundred...</d>\n\nshould sound genuinely amazed and appreciative.\n\nHis final:\n\n<d>[English] Maybe... all of them?</d>\n\nshould be playful, excited and slightly mischievous.\n\n--------------------------------------------------\n\n[PIP VOICE]\n\nSmall cute firefly voice.\n\nQualities:\n- slightly higher pitch\n- quick\n- expressive\n- cautious\n- dry comic timing\n- understandable English\n\nPerformance:\n\nbeginning:\nsupportive\n\nmiddle:\nincreasing suspicion\n\nclimb:\nslightly anxious\n\nfirst star:\npleasantly surprised\n\nending:\nimmediate panic\n\nFinal:\n\n<d>[English] ALL?!</d>\n\nis the comedy punctuation.\n\n--------------------------------------------------\n\n[LUMA AUDIO]\n\nNo spoken dialogue in PART 1.\n\nUse only:\n- very soft floating-lantern hum\n- subtle paper movement\n- gentle inner magical resonance\n- tiny warm pulse when star is attached\n\nNo snoring is necessary unless extremely soft and cute.\n\nLuma remains asleep.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nNighttime rooftop sound bed:\n\n- soft breeze\n- distant city ambience\n- very faint traffic far below\n- occasional distant bird/night insect\n- fabric ribbon movement\n- rooftop plants rustling\n- subtle paper-star sounds\n- Pip wing buzz\n- lantern hum\n\nKeep everything intimate.\n\n--------------------------------------------------\n\n[STAR SOUND DESIGN]\n\nPaper stars need a signature sound identity.\n\nPhysical handling:\nsoft folded-paper rustle.\n\nGlow:\ndelicate glassy/glockenspiel-like shimmer.\n\nAttachment:\ntiny paper “tap” + gentle magical “ting.”\n\nDo NOT use:\n- sci-fi electricity\n- laser sounds\n- aggressive magical effects\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nUse an ORIGINAL whimsical emotional score.\n\nInstrumentation:\n- soft piano\n- delicate glockenspiel\n- pizzicato strings\n- airy flute\n- light harp\n- warm string pad\n- subtle marimba\n- tiny bell accents\n\nMUSIC ARC:\n\n0.00–1.80\nquiet rooftop wonder.\n\n1.80–3.50\ngentle gratitude.\n\n3.50–5.20\nmagical Luma reveal.\n\n5.20–7.00\nplayful mission rhythm.\n\n7.00–9.00\nlight climbing comedy.\n\n9.00–11.00\nsmall emotional star-placement motif.\n\n11.00–12.80\nwarm success.\n\n12.80–15.00\nmischievous rising idea motif.\n\nEnd with playful comic sting after Pip says:\n\n<d>[English] ALL?!</d>\n\nNo epic trailer music.\n\nNo dramatic action score.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 1 camera should gradually move from intimate to enormous scale.\n\nCAMERA ARC:\n\n0–3.5 sec:\nquiet rooftop wides + emotional close-ups.\n\n3.5–5.2 sec:\nlarge upward reveal.\n\n5.2–9 sec:\nminiature adventure camera.\n\n9–12.8 sec:\nintimate star-placement coverage.\n\n12.8–15 sec:\nreaction-driven cliffhanger.\n\nUse:\n- slow dolly\n- gentle tilt\n- upward reveal\n- macro star inserts\n- low-angle climbing shots\n- vertical scale compositions\n- soft rack focus\n- reaction close-ups\n- gentle final push toward Milo\n\nAvoid:\n- handheld shaking\n- crash zoom\n- excessive cuts\n- spinning camera\n- action-movie camera movement\n- Dutch angles\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animation.\n\nMatch the uploaded storyboard EXACTLY in spirit:\n\n- hand-drawn ink-like contours\n- watercolor/gouache-inspired fills\n- deep painterly night skies\n- warm paper-lantern textures\n- expressive rounded characters\n- cinematic glow\n- delicate atmospheric city depth\n- charming handmade objects\n- subtle squash and stretch\n- believable fabric/ribbon movement\n\nThe video should look like the storyboard artwork has come alive.\n\nNOT:\n- storyboard animatic\n- slideshow\n- motion comic\n- photorealism\n- 3D CGI\n- anime\n- vector animation\n- cel-shaded 3D\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nPrimary palette:\n\nNIGHT:\n- navy blue\n- indigo\n- subtle violet\n- midnight teal\n\nWARM LIGHT:\n- amber\n- golden yellow\n- cream\n- soft orange\n\nMILO:\ndark blue hoodie with warm star-light reflections.\n\nPIP:\nsmall yellow-gold glow.\n\nLUMA:\nlargest warm amber light source.\n\nPaper stars:\ncream-gold.\n\nLighting progression:\n\n0–3 sec:\nmostly cool night with small warm star light.\n\n3–5 sec:\nLuma introduces large warm amber source.\n\n5–9 sec:\nmix of cool moonlight and warm lantern rim light.\n\n9–12 sec:\nsingle paper star becomes a delicate focal glow.\n\n12–15 sec:\nsatchel illuminates Milo’s excited face with many warm points.\n\nNo neon colors.\n\nNo overexposure.\n\n--------------------------------------------------\n\n[MOTION RULES — MILO]\n\nNatural tiny-character movement.\n\nUse:\n- soft seated posture\n- careful star handling\n- curious head tilts\n- expressive eyes\n- hoodie fabric reacting to breeze\n- determined climbing\n- hands gripping ribbons\n- stable footing\n- gentle placement of star\n\nNo:\n- superhero jumping\n- flying\n- impossible leaps\n- dangerous falls\n\n--------------------------------------------------\n\n[MOTION RULES — LUMA]\n\nLuma floats gently.\n\nUse:\n- slow vertical bob\n- minimal horizontal drift\n- ribbons responding softly to air\n- subtle paper flex\n- gentle internal light breathing\n\nDuring star attachment:\nonly a tiny warm light pulse.\n\nNo awakening.\n\n--------------------------------------------------\n\n[MOTION RULES — PIP]\n\nUse:\n- natural hovering\n- fast tiny wings\n- subtle inertia\n- head tilts\n- expressive eye widening\n- small arcs around Milo\n\nDo not have Pip carry Milo.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nMaintain:\n\none Milo\none Pip\none Luma\n\nSame:\n- rooftop\n- skyline\n- crescent moon\n- satchel\n- paper stars\n- Luma ribbons\n- character outfits\n- scale\n- night lighting direction\n\nMilo begins on rooftop.\n\nHe climbs Luma using her ribbons.\n\nDo NOT teleport him onto Luma.\n\nExactly ONE star is attached during Part 1.\n\nOnce attached, the star remains visible.\n\nLuma stays asleep.\n\nThe satchel remains full at ending.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard sheet.\nNo panel frames.\nNo numbers.\nNo timestamps.\nNo captions.\nNo subtitles.\nNo production notes.\nNo watermark.\nNo social-media logo.\nNo X logo.\nNo Twitter logo.\nNo platform UI.\n\nNo duplicate Milo.\nNo duplicate Pip.\nNo duplicate Luma.\nNo identity drift.\nNo clothing changes.\nNo hairstyle changes.\nNo scale inconsistency.\n\nNo awake Luma.\nNo talking Luma.\nNo angry Luma.\nNo fire.\nNo burning lantern.\nNo falling.\nNo injury.\nNo dangerous rooftop behavior.\nNo destructive magic.\n\nNo multiple stars attached in Part 1.\nNo fully decorated lantern.\nNo star explosion.\nNo “3200” sky formation yet.\nNo “THANK YOU” sky formation.\nNo Part 2 celebration.\n\nNo warped hands.\nNo extra fingers.\nNo malformed face.\nNo melting lantern.\nNo flickering character identity.\nNo disappearing satchel.\n\n--------------------------------------------------\n\n[FINAL 2.5-SECOND PRIORITY]\n\nThe ending must clearly create anticipation for PART 2.\n\nShow:\n\nONE glowing paper star attached to Luma.\n\nMilo proudly admires it.\n\nThen Milo looks into his still-overflowing satchel.\n\nMany warm paper stars glow inside.\n\nMilo looks from:\n\nsatchel\n→ single star\n→ enormous empty lantern surface.\n\nHis expression becomes increasingly excited.\n\nPip slowly realizes what Milo is thinking.\n\nPip:\n\n<d>[English] Milo...</d>\n\nMilo smiles:\n\n<d>[English] Maybe... all of them?</d>\n\nPip:\n\n<d>[English] ALL?!</d>\n\nLuma remains peacefully asleep.\n\nHold long enough for:\n\nMilo = ambitious gratitude  \nPip = comic panic  \nLuma = peacefully unaware  \n\nto register clearly.\n\nCUT TO BLACK.\n\nEND PART 1.\n\nThe viewer should immediately want to see how 3,200 tiny stars transform the night in PART 2.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086294164409630916",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086294164409630916",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T03:31:04.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086294164409630916.mp4",
    "posterUrl": "/posters/x/2086294164409630916.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for “The Little Creator and the 3,200 Paper Stars” as the PRIMARY visual, character, environment, scale, prop, lighting, performance, and narrative reference.\n\nGenerate a FINISHED cinematic animated sequence from this storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- storyboard panel layout\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character-reference drawings\n- scale-reference diagrams\n- storyboard title/header\n- paper background\n- footer decorations\n\nThe uploaded storyboard locks:\n\n1. Milo’s exact character identity\n2. Milo’s dark-blue oversized hoodie\n3. Milo’s fluffy brown hair\n4. Milo’s glowing-star satchel\n5. Luma’s exact giant lantern design\n6. Luma’s warm cream/orange paper texture\n7. Luma’s sleepy face\n8. Luma’s hanging ribbons and tassels\n9. Pip’s exact firefly design\n10. extreme Milo-vs-Luma scale\n11. handmade glowing paper-star design\n12. same magical rooftop environment\n13. same distant nighttime city\n14. deep indigo + warm amber palette\n15. single-star placement story beat\n16. final full-satchel cliffhanger\n\nTranslate the 15 storyboard panels into approximately 8 coherent cinematic shots.\n\nDO NOT create fifteen one-second hard cuts.\n\nUse clear:\n\ncause → action → consequence → reaction.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory time:\n0:00–0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D animated short film\n\nGenre:\nmagical gratitude story / whimsical rooftop fantasy\n\nAudience:\nfamily-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + character sounds + rooftop ambience + magical paper-star effects + original music\n\nEMOTIONAL ARC:\n\nquiet rooftop wonder\n→ gratitude\n→ magical giant reveal\n→ tiny secret mission\n→ difficult climb\n→ first star placement\n→ beautiful success\n→ realization that thousands remain\n→ mischievous cliffhanger\n\nThe overall feeling should be:\n\nsmall creator,\nhuge dream,\nwarm gratitude.\n\n--------------------------------------------------\n\n[PART 1 STORY LIMIT]\n\nPART 1 ONLY.\n\nSTART:\n\nMilo sits alone on a peaceful rooftop beneath a star-filled night sky with a satchel containing glowing handmade paper stars.\n\nHe is celebrating an important milestone:\n\n3,200 supporters.\n\nEND:\n\nMilo has attached EXACTLY ONE glowing paper star to Luma.\n\nHe then looks back at his still-overflowing satchel.\n\nMilo gets a huge idea.\n\nHe whispers:\n\n<d>[English] Maybe... all of them?</d>\n\nPip realizes what Milo means.\n\nCUT TO BLACK.\n\nDO NOT show:\n- Milo attaching multiple stars\n- Luma fully decorated\n- hundreds of stars covering Luma\n- Luma waking\n- Luma flying higher\n- stars leaving Luma\n- sky formation\n- giant “3200”\n- “THANK YOU”\n- full milestone celebration\n- final hero shot\n\nThose belong to PART 2.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — MILO]\n\nMatch the uploaded storyboard EXACTLY.\n\nMILO:\n\nTiny young creator.\n\nAppearance:\n- warm light skin\n- soft rounded youthful face\n- large expressive dark-brown eyes\n- fluffy tousled medium-brown hair\n- slightly oversized head for appealing animation proportions\n\nClothing:\n- oversized dark navy-blue hoodie\n- dark/simple trousers\n- little shoes/boots\n- small brown crossbody satchel\n- satchel filled with glowing folded paper stars\n\nPersonality:\n- grateful\n- creative\n- curious\n- optimistic\n- emotional\n- slightly impulsive\n- playful\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nquietly grateful\n\n3–5 sec:\ncurious wonder\n\n5–7 sec:\ngets an idea\n\n7–10 sec:\ndetermined tiny climber\n\n10–12 sec:\ncareful and emotionally invested\n\n12–15 sec:\nproud → inspired → mischievously ambitious\n\nMilo should never feel arrogant.\n\nThe milestone means something because of the people represented by the stars.\n\nCRITICAL:\n\nKeep exact:\n- face\n- brown hair\n- navy hoodie\n- satchel\n- body proportions\n- age appearance\n- scale\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — LUMA]\n\nMatch the uploaded storyboard exactly.\n\nLUMA:\n\nA gigantic living floating sky lantern.\n\nAppearance:\n- enormous rounded lantern body\n- warm cream / parchment-orange paper surface\n- handmade subtle paper texture\n- decorative curved motifs\n- gentle closed sleepy eyes\n- tiny peaceful curved smile\n- soft amber-gold inner illumination\n- decorative frame underneath\n- multiple long hanging ribbons / tassels\n- huge scale relative to Milo\n\nPersonality:\n- peaceful\n- gentle\n- warm\n- sleepy\n- safe\n- quietly magical\n\nPART 1:\n\nLuma remains ASLEEP for the entire sequence.\n\nAllowed:\n- very subtle floating\n- gentle vertical bob\n- soft breathing-like light pulse\n- tiny peaceful smile response when first star is attached\n\nDo NOT:\n- open her eyes\n- speak\n- rise dramatically\n- intentionally move Milo\n- activate celebration magic\n\nHer awakening happens in PART 2.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — PIP]\n\nMatch the storyboard.\n\nPIP:\n\nTiny glowing firefly helper.\n\nAppearance:\n- small round yellow-gold body\n- pale luminous glow\n- large expressive eyes\n- tiny antennae\n- translucent wings\n- much smaller than Milo\n\nPersonality:\n- practical\n- cautious\n- loyal\n- funny\n- suspicious of Milo’s big ideas\n\nPerformance:\n\ncalm companion\n→ curious\n→ realizes Milo’s plan\n→ worried during climb\n→ nervous during star placement\n→ pleased that it works\n→ horrified by Milo’s final bigger idea\n\nPip may speak in short natural lines.\n\n--------------------------------------------------\n\n[GLOWING PAPER STAR LOCK]\n\nAll stars must match the storyboard.\n\nDesign:\n- handmade folded paper stars\n- cream / pale-gold paper\n- visible folded facets\n- warm golden light from inside\n- physical tactile objects\n- lightweight\n- slightly magical\n\nClose views should preserve real paper folds.\n\nAt the beginning:\nthe satchel contains many glowing stars.\n\nDuring PART 1:\nMilo removes ONLY ONE star for decoration.\n\nAt the final shot:\nthe satchel must still visibly overflow with stars.\n\n--------------------------------------------------\n\n[STAR SYMBOLISM]\n\nThe stars represent individual pieces of support from Milo’s audience/community.\n\nThe video should communicate this emotionally without showing social-media interfaces.\n\nNo:\n- phones\n- follower counters\n- social-media app screens\n- X logos\n- Twitter logos\n- platform branding\n\nMilo may verbally reference:\n\n“3,200”\n\nbut the milestone should primarily be represented by the glowing paper stars.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — ROOFTOP]\n\nMaintain the exact nighttime rooftop world from the storyboard.\n\nInclude:\n- cozy high rooftop\n- old rooftop stone/brick surfaces\n- chimney / small rooftop structures\n- a few potted plants\n- distant city buildings\n- warm glowing windows\n- layered rooftops\n- deep blue night sky\n- stars\n- crescent moon\n- subtle distant clouds\n- warm rooftop accent lights\n\nMood:\nquiet,\nsafe,\ndreamlike,\nintimate.\n\nThe city is far below.\n\nNo crowds.\n\nNo extra characters.\n\n--------------------------------------------------\n\n[SCALE RULES]\n\nExtreme scale contrast is critical.\n\nMilo is tiny compared with Luma.\n\nUse storyboard scale:\n\n- Milo smaller than one tassel knot\n- Luma’s lantern body towers above Milo\n- hanging ribbons are thick enough for Milo to climb like ropes\n- one section of Luma’s body is enormous compared with Milo’s full height\n\nPip is much smaller than Milo.\n\nNever accidentally render Milo human-sized beside Luma.\n\n--------------------------------------------------\n\n[WORLD / MOTION LOGIC]\n\nClear cause-and-effect:\n\nMilo looks at stars\n→ remembers milestone\n\nMilo looks upward\n→ notices sleeping Luma\n\nLuma’s huge blank lantern surface\n→ gives Milo decorating idea\n\nMilo wants to celebrate\n→ climbs ribbons\n\nMilo reaches lantern\n→ attaches ONE star\n\nStar lights up beautifully\n→ Milo feels successful\n\nMilo looks down\n→ sees full satchel\n\nFull satchel + enormous lantern\n→ Milo gets much bigger idea\n\nPip realizes what Milo is thinking\n→ comic panic cliffhanger\n\nNo random magic.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.80s\nA QUIET ROOFTOP AND 3,200 LITTLE LIGHTS\n\nOpen with a beautiful nighttime establishing shot.\n\nDeep indigo sky.\n\nCrescent moon.\n\nThousands of distant real stars.\n\nCamera slowly descends toward a peaceful rooftop overlooking a softly glowing city.\n\nMilo sits near the rooftop edge.\n\nBeside him is his brown satchel.\n\nWarm golden light leaks from inside.\n\nMilo opens the satchel.\n\nReveal many handmade glowing paper stars.\n\nHe gently picks ONE up with both hands.\n\nThe warm star light illuminates his face.\n\nPip floats quietly beside him.\n\nCamera:\nwide skyline\n→ slow dolly toward Milo\n→ overhead satchel insert\n→ intimate close-up on glowing star in his hands.\n\nDialogue:\n\nMilo, softly:\n<d>[English] Three thousand two hundred...</d>\n\nTiny pause.\n\nMilo smiles.\n\n<d>[English] That’s a lot of people.</d>\n\nPip gives a soft happy chirp.\n\nSound:\n- distant nighttime city\n- light rooftop breeze\n- cloth movement\n- satchel leather\n- subtle paper rustle\n- soft magical star hum\n\nMusic:\ngentle piano + glockenspiel + warm strings.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.80s–3.50s\nA REASON TO SAY THANK YOU\n\nMilo holds the glowing star near his chest.\n\nHis expression becomes warm and grateful.\n\nHe looks at the many remaining stars.\n\nDialogue:\n\nMilo:\n<d>[English] We should do something special.</d>\n\nPip tilts his head.\n\nThen Milo notices an enormous warm amber light above him.\n\nThe light softly brightens his face.\n\nMilo slowly looks upward.\n\nCamera follows his gaze.\n\nReveal the lower edge of Luma entering frame.\n\nLong decorative ribbons sway gently beneath her.\n\nMilo’s eyes widen.\n\nDialogue:\n\nMilo:\n<d>[English] Oh...</d>\n\nPip slowly turns upward too.\n\nSound:\n- gentle ribbon movement\n- low warm lantern hum\n- night breeze\n\nMusic:\nintroduce airy flute and soft wonder motif.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.50s–5.20s\nTHE SLEEPY SKY LANTERN\n\nCamera continues upward into a full magical reveal of Luma.\n\nLuma floats above the city.\n\nHuge.\n\nRound.\n\nWarmly illuminated.\n\nHer eyes remain peacefully closed.\n\nHer tiny smile is gentle.\n\nShe slowly bobs in the night air.\n\nMilo is tiny at the bottom of frame.\n\nThis shot must strongly emphasize SCALE.\n\nMilo looks at:\nthe glowing star in his hand...\n\nthen Luma’s enormous blank lantern surface.\n\nIdea forming.\n\nHis mouth curls into a grin.\n\nPip notices Milo’s expression.\n\nDialogue:\n\nMilo, whispering:\n<d>[English] Pip...</d>\n\nPip:\n<d>[English] What?</d>\n\nMilo:\n<d>[English] I have an idea.</d>\n\nPip immediately becomes suspicious.\n\nPip:\n<d>[English] That face worries me.</d>\n\nMusic:\nsmall mischievous plucked motif enters.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.20s–7.00s\nTHE SECRET STAR MISSION\n\nMilo walks beneath one long hanging Luma ribbon.\n\nHe grabs it with both hands.\n\nTests it.\n\nLooks upward.\n\nIt extends far above him.\n\nMilo puts the glowing star securely into his hoodie pocket / small pouch while keeping it visible and safe.\n\nThen starts climbing.\n\nPip hovers beside him.\n\nDialogue:\n\nMilo, whispering:\n<d>[English] We should celebrate.</d>\n\nPip:\n<d>[English] From the roof.</d>\n\nMilo climbs higher.\n\nMilo:\n<d>[English] Too ordinary.</d>\n\nPip:\n<d>[English] Of course.</d>\n\nCamera:\nlow-angle ribbon shot\n→ side tracking with Milo climbing\n→ pull wide showing tiny Milo beneath giant lantern.\n\nSound:\n- fabric ribbon strain\n- Milo’s tiny climbing effort\n- shoe taps\n- Pip wing buzz\n- soft Luma hum\n\n--------------------------------------------------\n\n[SHOT 5] — 7.00s–9.00s\nMILO VS. LUMA\n\nMilo reaches a large tassel knot.\n\nHe pauses and sits momentarily on it.\n\nThe knot itself is nearly Milo-sized.\n\nHe looks upward at the enormous lantern surface.\n\nPip floats beside him, slightly out of breath despite being a firefly.\n\nUse a vertical scale composition.\n\nMilo climbs the final short ribbon section.\n\nHe reaches Luma’s lower lantern edge.\n\nLuma remains completely asleep.\n\nDialogue:\n\nMilo:\n<d>[English] Almost there...</d>\n\nPip:\n<d>[English] Define “almost.”</d>\n\nMilo pulls himself onto a decorative rim.\n\nDialogue:\n\nMilo:\n<d>[English] There.</d>\n\nSound:\n- ribbon movement\n- tiny climb grunt\n- paper lantern creak\n- soft floating ambience\n\n--------------------------------------------------\n\n[SHOT 6] — 9.00s–11.00s\nTHE FIRST STAR\n\nMilo retrieves the ONE glowing folded star.\n\nHe looks at it.\n\nThen at Luma’s huge lantern body.\n\nHe carefully presses the handmade star onto the lantern using a tiny clip / adhesive tab.\n\nTake time with this.\n\nThe star gently settles.\n\nFor half a beat:\n\nnothing.\n\nThen its glow becomes slightly warmer.\n\nThe golden star light blends beautifully with Luma’s amber paper body.\n\nMilo’s face lights up.\n\nPip watches cautiously.\n\nLuma’s sleeping smile curves perhaps one tiny degree more.\n\nShe does NOT wake.\n\nDialogue:\n\nMilo, whispering:\n<d>[English] One little thank-you.</d>\n\nSoft click.\n\nMilo:\n<d>[English] Perfect.</d>\n\nSound:\n- paper touching paper\n- tiny clip click\n- warm magical “ting”\n- Luma’s gentle light pulse\n\nCamera:\nclose-up Milo’s hands\n→ macro star attachment\n→ reaction on Milo\n→ Luma’s peaceful sleepy expression.\n\nMusic:\ngentle emotional success note.\n\n--------------------------------------------------\n\n[SHOT 7] — 11.00s–12.80s\nONE VERY BIG SUCCESS\n\nMilo leans back against Luma’s decorative rim.\n\nHe admires the glowing star.\n\nThe single paper star looks beautiful against the huge warm lantern surface.\n\nPip slowly smiles.\n\nMilo:\n<d>[English] Look at that.</d>\n\nPip:\n<d>[English] Okay...</d>\n\nPip studies it.\n\n<d>[English] That is actually nice.</d>\n\nMilo beams.\n\nCamera pulls out slightly.\n\nClearly show:\n\ntiny Milo\n+\none glowing star\n+\nenormous mostly undecorated Luma.\n\nThe contrast is important.\n\n--------------------------------------------------\n\n[SHOT 8] — 12.80s–15.00s\nTHE MUCH BIGGER IDEA — CLIFFHANGER\n\nMilo’s proud smile suddenly pauses.\n\nHe remembers something.\n\nHe looks downward.\n\nCamera follows his gaze toward the rooftop.\n\nHis satchel remains there / hangs securely nearby according to spatial setup, visibly overflowing with glowing paper stars.\n\nAlternatively, if he carried the satchel during the climb, show it hanging securely from his shoulder and still overflowing.\n\nCRITICAL:\ndo not teleport the satchel.\n\nUse whichever configuration is established earlier and maintain it consistently.\n\nMilo slowly opens / looks into it.\n\nDozens upon dozens of glowing folded stars illuminate his face.\n\nThen he looks at:\n\nthe full satchel...\n\nthe ONE star on Luma...\n\nLuma’s ENORMOUS remaining lantern surface.\n\nMilo’s eyes widen.\n\nA mischievous smile grows.\n\nPip looks at Milo.\n\nPip looks at the satchel.\n\nPip looks at enormous Luma.\n\nPip’s expression turns to immediate horror.\n\nPip:\n<d>[English] Milo...</d>\n\nMilo raises the satchel slightly.\n\nMilo, delighted whisper:\n<d>[English] Maybe... all of them?</d>\n\nPip’s eyes become huge.\n\nPip:\n<d>[English] ALL?!</d>\n\nLuma continues peacefully sleeping.\n\nThe ONE attached star glows softly.\n\nHold on:\n\nMilo = enormous inspiration  \nPip = comic panic  \nLuma = peacefully unaware  \n\nCUT TO BLACK.\n\nEND PART 1.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[MILO VOICE]\n\nNatural young creator voice.\n\nAge feeling:\nyoung / youthful, not adult.\n\nQualities:\n- warm\n- sincere\n- curious\n- imaginative\n- emotionally grateful\n- playful\n- clear English\n\nDo not make Milo sound boastful.\n\nHis line:\n\n<d>[English] Three thousand two hundred...</d>\n\nshould sound genuinely amazed and appreciative.\n\nHis final:\n\n<d>[English] Maybe... all of them?</d>\n\nshould be playful, excited and slightly mischievous.\n\n--------------------------------------------------\n\n[PIP VOICE]\n\nSmall cute firefly voice.\n\nQualities:\n- slightly higher pitch\n- quick\n- expressive\n- cautious\n- dry comic timing\n- understandable English\n\nPerformance:\n\nbeginning:\nsupportive\n\nmiddle:\nincreasing suspicion\n\nclimb:\nslightly anxious\n\nfirst star:\npleasantly surprised\n\nending:\nimmediate panic\n\nFinal:\n\n<d>[English] ALL?!</d>\n\nis the comedy punctuation.\n\n--------------------------------------------------\n\n[LUMA AUDIO]\n\nNo spoken dialogue in PART 1.\n\nUse only:\n- very soft floating-lantern hum\n- subtle paper movement\n- gentle inner magical resonance\n- tiny warm pulse when star is attached\n\nNo snoring is necessary unless extremely soft and cute.\n\nLuma remains asleep.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nNighttime rooftop sound bed:\n\n- soft breeze\n- distant city ambience\n- very faint traffic far below\n- occasional distant bird/night insect\n- fabric ribbon movement\n- rooftop plants rustling\n- subtle paper-star sounds\n- Pip wing buzz\n- lantern hum\n\nKeep everything intimate.\n\n--------------------------------------------------\n\n[STAR SOUND DESIGN]\n\nPaper stars need a signature sound identity.\n\nPhysical handling:\nsoft folded-paper rustle.\n\nGlow:\ndelicate glassy/glockenspiel-like shimmer.\n\nAttachment:\ntiny paper “tap” + gentle magical “ting.”\n\nDo NOT use:\n- sci-fi electricity\n- laser sounds\n- aggressive magical effects\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nUse an ORIGINAL whimsical emotional score.\n\nInstrumentation:\n- soft piano\n- delicate glockenspiel\n- pizzicato strings\n- airy flute\n- light harp\n- warm string pad\n- subtle marimba\n- tiny bell accents\n\nMUSIC ARC:\n\n0.00–1.80\nquiet rooftop wonder.\n\n1.80–3.50\ngentle gratitude.\n\n3.50–5.20\nmagical Luma reveal.\n\n5.20–7.00\nplayful mission rhythm.\n\n7.00–9.00\nlight climbing comedy.\n\n9.00–11.00\nsmall emotional star-placement motif.\n\n11.00–12.80\nwarm success.\n\n12.80–15.00\nmischievous rising idea motif.\n\nEnd with playful comic sting after Pip says:\n\n<d>[English] ALL?!</d>\n\nNo epic trailer music.\n\nNo dramatic action score.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 1 camera should gradually move from intimate to enormous scale.\n\nCAMERA ARC:\n\n0–3.5 sec:\nquiet rooftop wides + emotional close-ups.\n\n3.5–5.2 sec:\nlarge upward reveal.\n\n5.2–9 sec:\nminiature adventure camera.\n\n9–12.8 sec:\nintimate star-placement coverage.\n\n12.8–15 sec:\nreaction-driven cliffhanger.\n\nUse:\n- slow dolly\n- gentle tilt\n- upward reveal\n- macro star inserts\n- low-angle climbing shots\n- vertical scale compositions\n- soft rack focus\n- reaction close-ups\n- gentle final push toward Milo\n\nAvoid:\n- handheld shaking\n- crash zoom\n- excessive cuts\n- spinning camera\n- action-movie camera movement\n- Dutch angles\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animation.\n\nMatch the uploaded storyboard EXACTLY in spirit:\n\n- hand-drawn ink-like contours\n- watercolor/gouache-inspired fills\n- deep painterly night skies\n- warm paper-lantern textures\n- expressive rounded characters\n- cinematic glow\n- delicate atmospheric city depth\n- charming handmade objects\n- subtle squash and stretch\n- believable fabric/ribbon movement\n\nThe video should look like the storyboard artwork has come alive.\n\nNOT:\n- storyboard animatic\n- slideshow\n- motion comic\n- photorealism\n- 3D CGI\n- anime\n- vector animation\n- cel-shaded 3D\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nPrimary palette:\n\nNIGHT:\n- navy blue\n- indigo\n- subtle violet\n- midnight teal\n\nWARM LIGHT:\n- amber\n- golden yellow\n- cream\n- soft orange\n\nMILO:\ndark blue hoodie with warm star-light reflections.\n\nPIP:\nsmall yellow-gold glow.\n\nLUMA:\nlargest warm amber light source.\n\nPaper stars:\ncream-gold.\n\nLighting progression:\n\n0–3 sec:\nmostly cool night with small warm star light.\n\n3–5 sec:\nLuma introduces large warm amber source.\n\n5–9 sec:\nmix of cool moonlight and warm lantern rim light.\n\n9–12 sec:\nsingle paper star becomes a delicate focal glow.\n\n12–15 sec:\nsatchel illuminates Milo’s excited face with many warm points.\n\nNo neon colors.\n\nNo overexposure.\n\n--------------------------------------------------\n\n[MOTION RULES — MILO]\n\nNatural tiny-character movement.\n\nUse:\n- soft seated posture\n- careful star handling\n- curious head tilts\n- expressive eyes\n- hoodie fabric reacting to breeze\n- determined climbing\n- hands gripping ribbons\n- stable footing\n- gentle placement of star\n\nNo:\n- superhero jumping\n- flying\n- impossible leaps\n- dangerous falls\n\n--------------------------------------------------\n\n[MOTION RULES — LUMA]\n\nLuma floats gently.\n\nUse:\n- slow vertical bob\n- minimal horizontal drift\n- ribbons responding softly to air\n- subtle paper flex\n- gentle internal light breathing\n\nDuring star attachment:\nonly a tiny warm light pulse.\n\nNo awakening.\n\n--------------------------------------------------\n\n[MOTION RULES — PIP]\n\nUse:\n- natural hovering\n- fast tiny wings\n- subtle inertia\n- head tilts\n- expressive eye widening\n- small arcs around Milo\n\nDo not have Pip carry Milo.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nMaintain:\n\none Milo\none Pip\none Luma\n\nSame:\n- rooftop\n- skyline\n- crescent moon\n- satchel\n- paper stars\n- Luma ribbons\n- character outfits\n- scale\n- night lighting direction\n\nMilo begins on rooftop.\n\nHe climbs Luma using her ribbons.\n\nDo NOT teleport him onto Luma.\n\nExactly ONE star is attached during Part 1.\n\nOnce attached, the star remains visible.\n\nLuma stays asleep.\n\nThe satchel remains full at ending.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard sheet.\nNo panel frames.\nNo numbers.\nNo timestamps.\nNo captions.\nNo subtitles.\nNo production notes.\nNo watermark.\nNo social-media logo.\nNo X logo.\nNo Twitter logo.\nNo platform UI.\n\nNo duplicate Milo.\nNo duplicate Pip.\nNo duplicate Luma.\nNo identity drift.\nNo clothing changes.\nNo hairstyle changes.\nNo scale inconsistency.\n\nNo awake Luma.\nNo talking Luma.\nNo angry Luma.\nNo fire.\nNo burning lantern.\nNo falling.\nNo injury.\nNo dangerous rooftop behavior.\nNo destructive magic.\n\nNo multiple stars attached in Part 1.\nNo fully decorated lantern.\nNo star explosion.\nNo “3200” sky formation yet.\nNo “THANK YOU” sky formation.\nNo Part 2 celebration.\n\nNo warped hands.\nNo extra fingers.\nNo malformed face.\nNo melting lantern.\nNo flickering character identity.\nNo disappearing satchel.\n\n--------------------------------------------------\n\n[FINAL 2.5-SECOND PRIORITY]\n\nThe ending must clearly create anticipation for PART 2.\n\nShow:\n\nONE glowing paper star attached to Luma.\n\nMilo proudly admires it.\n\nThen Milo looks into his still-overflowing satchel.\n\nMany warm paper stars glow inside.\n\nMilo looks from:\n\nsatchel\n→ single star\n→ enormous empty lantern surface.\n\nHis expression becomes increasingly excited.\n\nPip slowly realizes what Milo is thinking.\n\nPip:\n\n<d>[English] Milo...</d>\n\nMilo smiles:\n\n<d>[English] Maybe... all of them?</d>\n\nPip:\n\n<d>[English] ALL?!</d>\n\nLuma remains peacefully asleep.\n\nHold long enough for:\n\nMilo = ambitious gratitude  \nPip = comic panic  \nLuma = peacefully unaware  \n\nto register clearly.\n\nCUT TO BLACK.\n\nEND PART 1.\n\nThe viewer should immediately want to see how 3,200 tiny stars transform the night in PART 2.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 2751,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status publishes the complete Part 1 reference-control prompt, character locks, rooftop setting, paper-star beat, dialogue, audio, and cliffhanger."
  },
  {
    "id": "x-2086254335021961590",
    "title": "从线框到成品的建筑展板",
    "titleEn": "An Architecture Board Assembles from Wireframes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "建筑方案从点云、深度图和白色线框逐层显现，最终拼装为完整的图文展示板。 原生视频规格为 15 秒 · 480×640。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An architectural scheme evolves from point cloud, depth map, and white wireframe into a complete mixed-media presentation board. Native-video output: 15s · 480×640. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use [img1] as the sole visual authority for the architecture, courtyard, container modules, roof geometry, facade openings, materials, landscape, people, diagrams, site plan, typography, labels, colors, and final presentation-board composition.\n\nCreate a fast-paced 15-second architectural presentation reveal. Do not begin with the completed board. Start from a dark charcoal background and progressively assemble the board through point clouds, depth maps, wireframes, technical drawings, typography, diagrams, and finally the photorealistic architecture.\n\nPreserve the architecture exactly. Do not redesign the building.\n\n[0.0–1.5s — ARCHITECTURE EMERGES]\n\nStart from near-black.\n\nA faint point cloud flashes into existence in the upper portion of the frame, rapidly transforming:\n\npoint cloud → depth map → white architectural wireframe → shaded clay model.\n\nRoof edges, container modules, balconies, windows, courtyard boundaries, and structural lines are traced in quick succession with sharp rhythmic flashes and scanning effects.\n\n[1.5–3.2s — HERO IMAGE REVEAL]\n\nThe wireframe snaps into the photorealistic courtyard rendering.\n\nColor and material flood into the image:\nwhite container walls, timber roofs, glass reflections, grass, trees, and warm sunlight.\n\nThe hero image locks into its final upper-board position.\n\nSubtle life begins immediately:\ntree branches sway,\ngrass moves,\npeople make small gestures,\nsunlight shifts across the facades.\n\n[3.2–4.4s — TYPOGRAPHY]\n\nThe title appears letter-by-letter with elegant tracking expansion.\n\nSubtitle and location follow.\n\nDivider lines shoot into place.\n\nSmall labels reveal sequentially.\n\nKeep all typography sharp, clean, and stable once revealed.\n\n[4.4–6.4s — SPATIAL SEQUENCE]\n\nThe first analytical module flashes in as fragmented white linework.\n\nIts layers rapidly assemble:\nground plane → building volumes → balconies → roofs → courtyard.\n\nOrange arrows sweep through thresholds and semi-private zones toward the courtyard.\n\nLabels appear one by one beneath the diagram.\n\n[6.4–8.3s — DAYLIGHTING]\n\nA lateral light sweep activates the second diagram.\n\nSun icons appear.\n\nA curved solar path draws across the architecture.\n\nWarm light rays move over the roofs and into the courtyard.\n\nReflected light and subtle airflow paths animate briefly.\n\nThe hero image above responds with changing sunlight and moving tree shadows.\n\n[8.3–10.1s — CIRCULATION]\n\nThe third axonometric appears as a translucent technical model.\n\nOrange circulation paths race around balconies and climb vertical connectors.\n\nSmall moving points suggest occupants traveling through the building.\n\nThe animation settles into the final circulation diagram.\n\n[10.1–11.8s — SITE + DATA]\n\nThe lower section assembles rapidly from left to right.\n\nSite context appears.\n\nThe site boundary traces itself.\n\nThe plan develops from thin drafting lines.\n\nProject data appears row-by-row with precise editorial timing.\n\nThe full board is now nearly complete.\n\n[11.8–13.5s — BUILDING COMES ALIVE]\n\nShift emphasis back to the large hero rendering.\n\nA subtle push-in occurs only inside this image.\n\nExisting people begin moving naturally.\n\nAdd several more occupants:\none person enters a ground-floor doorway,\nanother exits,\na student crosses the courtyard,\none person walks along an upper balcony.\n\nDoors open and close softly.\n\nFoliage moves more clearly.\n\nSunlight passes through leaves and shadows glide across the facades.\n\n[13.5–15.0s — FINAL CLIMAX]\n\nA final pulse travels from the hero rendering downward through the board.\n\nSpatial sequence activates.\nDaylighting activates.\nCirculation activates.\nSite plan activates.\n\nFine orange and white lines briefly connect the analytical systems, then settle.\n\nHold the complete final board while subtle life continues inside the hero rendering:\npeople walking,\npeople entering and leaving,\nmoving foliage,\nsoft reflections,\nshifting sunlight.\n\n[MOTION STYLE]\n\nFast, rhythmic architectural motion graphics.\nPoint-cloud reconstruction.\nDepth-map transitions.\nWireframe-to-photoreal reveal.\nRapid drafting-line animation.\nMasked typography.\nDiagram assembly.\nCirculation tracing.\nAnimated light paths.\nStrong acceleration and clean easing.\n\nThe first 11 seconds should contain frequent visual events every 0.3–0.8 seconds.\n\nUse the rhythm:\nFLASH → DRAW → ASSEMBLE → LOCK.\n\nOnce revealed, each module remains in its final position.\n\n[CRITICAL RULES]\n\nDo not rotate the entire board.\nDo not turn the board into a separate full-screen architectural flythrough.\nDo not rearrange the layout.\nDo not redesign the building.\nDo not alter the container massing, roof geometry, facade rhythm, courtyard proportions, balconies, windows, or structural logic.\nDo not replace the original diagrams.\n\nAvoid neon HUDs, cyberpunk effects, random particles, excessive glow, or generic tech graphics.\n\nKeep the visual language architectural, editorial, precise, and premium.\n\n[AUDIO]\n\nDeep opening impact, scanner pulses, drafting scratches, paper-like swishes, subtle mechanical clicks, soft transition whooshes.\n\nAs the hero image comes alive, introduce quiet courtyard ambience:\nfootsteps, distant conversation, doors opening, leaves moving in the breeze.\n\nFinish with one clean low-frequency impact as the full board locks into place.\n\nNo dialogue.",
    "promptSourceUrl": "https://x.com/vkuoo/status/2086254339061174735",
    "sourceUrl": "https://x.com/vkuoo/status/2086254335021961590",
    "sourceLabel": "X 原帖 · @vkuoo",
    "author": "vkuoo",
    "publishedAt": "2026-08-09T00:52:48.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086254335021961590.mp4",
    "posterUrl": "/posters/x/2086254335021961590.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×640",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Animation showcase for architectural presentation boards built on @krea_ai  with MiniMax H3.\nThis model shines at generating mixed‑media motion graphics combining visuals and text information.\n\nPrompt in the replies.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 756,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X caption and creator reply publish the full architecture-board prompt, including reference lock, staged assembly, typography, diagrams, and final composition."
  },
  {
    "id": "x-2086224302316105742",
    "title": "牛粪上的微型村庄",
    "titleEn": "A Tiny Village on Cow Dung",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "镜头先观察微型居民的村庄，随后戏剧性拉远，揭示聚落其实建在田野里的牛粪上。 原生视频规格为 10 秒 · 608×1056。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The camera observes a tiny village, then dramatically pulls back to reveal that it sits on cow dung beside a nearby cow. Native-video output: 10s · 608×1056. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "a village of tiny people living on cow poop. The camera then dramatically zooms out to show where they are and the cow nearby in the field.",
    "promptSourceUrl": "https://x.com/LikeToasters/status/2086224302316105742",
    "sourceUrl": "https://x.com/LikeToasters/status/2086224302316105742",
    "sourceLabel": "X 原帖 · @LikeToasters",
    "author": "ILikeToasters",
    "publishedAt": "2026-08-08T22:53:28.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086224302316105742.mp4",
    "posterUrl": "/posters/x/2086224302316105742.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "608×1056",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3 local. I like vague prompts. \n\nPrompt: a village of tiny people living on cow poop. The camera then dramatically zooms out to show where they are and the cow nearby in the field.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 17,
      "views": 566,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status explicitly labels local H3 generation and publishes the complete concise prompt with the tiny village, zoom-out reveal, field, and cow."
  },
  {
    "id": "x-2086199653922455917",
    "title": "黎明面包配送广告",
    "titleEn": "A Dawn Bakery Delivery Launch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "烘焙、装袋和骑手配送随着天色逐步变亮，被剪成一条可静音理解的竖屏品牌广告。 原生视频规格为 15 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Baking, packing, and courier delivery unfold as the light progresses from pre-dawn darkness to warm sunrise in a vertical brand launch. Native-video output: 15s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "@ Image1 is the sole authority for the PROOF logo: the roundel of five concentric rings\nand the PROOF wordmark in heavy geometric rounded sans. Reproduce the mark and the\nspelling P-R-O-O-F exactly and never alter, restyle or re-letter them. @ Image2 is the\nauthority for the branded kraft bakery bag. @ Image3 is the authority for the courier,\nthe canvas bread satchel and the dawn street.\n\n[GOAL]\nA 15-second vertical 9:16 social launch commercial for PROOF, a dawn bakery delivery\nservice. Fast-cut, caption-led, designed to be understood with the sound off. Photoreal\ndocumentary commercial cinematography on 35mm, fine grain, shallow depth of field,\nhandheld with restrained movement.\n\n[THE LIGHT IS THE CLOCK]\nThe film brightens continuously from beginning to end. It opens in near-darkness lit by\na single work lamp and finishes in full warm golden sunrise. Every shot is measurably\nlighter than the one before it. This progression must be unbroken.\n\n[CAPTION CHIPS]\nThroughout, a single caption chip sits in the lower third, horizontally centred, in the\nsame position in every shot. It is a rounded rectangle in burnt ochre with warm cream\ntext, two lines: a lighter-weight top line and a bold bottom line. The chip changes\ncontent on the cuts listed below and never moves, resizes or animates. Type must be\ncrisp and correctly spelled.\n\n[SEQUENCE — cut on every timing, eight cuts]\n0-1.9s    HERO. Interior of a small bakery at night, lit by one overhead work lamp.\n          A baker's hands turn a large mass of wet dough on a floured steel bench.\n          Everything else in the room is black.\n          CHIP: \"04:12\" / \"someone is already awake\"\n1.9-3.6s  INSERT. Extreme close-up, a handful of flour thrown across the bench, dust\n          catching the single lamp.\n          CHIP unchanged.\n3.6-5.4s  HERO. A row of round dough balls resting in floured proving baskets on a rack.\n          Slightly warmer, slightly brighter. One dough is visibly swollen and domed.\n          CHIP: \"05:30\" / \"it's proving\"\n5.4-7.1s  INSERT. An oven door swings open and a wall of steam rolls out toward camera,\n          lit warm from inside.\n          CHIP unchanged.\n7.1-8.9s  HERO. A dark-crusted round sourdough loaf lands on a cooling rack. Hold on it.\n          The crust is visibly crackling and ticking as it cools.\n          CHIP: \"06:15\" / \"it's out of the oven\"\n8.9-10.6s INSERT. The loaf is slid into the kraft PROOF bag from @ Image2, the top folded\n          over once and taped. The ochre roundel and wordmark clearly readable.\n          CHIP unchanged.\n10.6-12.3s HERO. The courier from @ Image3 rides away down an empty terraced street in\n          low golden sunrise, satchel on their back, PROOF mark readable, streetlights\n          switching off behind them.\n          CHIP: \"06:52\" / \"it's moving\"\n12.3-13.6s INSERT. The kraft bag set down on a stone doorstep in full golden light,\n          exactly as in @ Image2. A hand withdraws from frame. Steam rises from the fold.\n          CHIP: \"07:04\" / \"it's at your door\"\n13.6-15s  END FRAME. Cut to a clean flat warm cream background. The PROOF concentric\n          ring mark and wordmark from @ Image1 sit centred, still, in burnt ochre and\n          dark rye brown. Beneath them, small, in dark rye brown: \"now delivering\".\n          CHIP for the final beat only: \"bread younger than your alarm\", single bold\n          line, centred. Absolutely no motion on this frame — it is a still card.\n\n[COLOUR]\nBurnt ochre is a physical object in the world before it is ever an interface colour —\nit is printed ink on the kraft bag and on the canvas satchel, lit by real light. The\ncaption chip is the only ochre graphic element. Palette: warm cream, kraft brown, burnt\nochre, dark rye brown, one cool blue-grey in the pre-dawn shadows.\n\n[AUDIO]\nDiegetic sound only, no music of any kind. The mix should feel close, warm and real.\n0-3.6s    The dead quiet of a room at 4am: an extractor fan humming, wet dough slapping\n          and stretching against steel, a scraper, the soft thud of flour.\n3.6-7.1s  Quieter still. A rack wheeled a short distance. The heavy clunk of an oven\n          latch and a deep roll of escaping steam.\n7.1-8.9s  THIS IS THE MOMENT THE WHOLE FILM IS BUILT AROUND. A loaf on a wire rack, its\n          crust crackling and ticking as it cools — a fine, irregular, delicate series\n          of snaps and pops, close-miked and clearly audible above everything else.\n          Bring the rest of the mix down under it.\n8.9-12.3s Kraft paper crumpling and folding, tape torn. Then outdoors: a freewheel\n          ticking, tyres on wet tarmac, one blackbird, distant early traffic.\n12.3-15s  The bag setting down on stone. Footsteps leaving. Birdsong. Then near silence\n          under the end card, with one last faint crust tick.\nNo music, no voiceover, no narration, no dialogue, no whoosh transitions, no sound\neffects that are not physically present in the scene.\n\n[NEGATIVE]\nNo text of any kind other than the caption chips, the PROOF wordmark, and \"now\ndelivering\". No other logos, no brand names, no shop signage, no readable street signs,\nno house numbers, no phone screens, no app interfaces, no price, no website, no social\nhandles, no hashtags, no watermarks. Do not alter the PROOF mark's ring count or the\nspelling of the wordmark. No music. No people speaking. No faces held in close-up for\nlonger than one second — this is about the bread, not a character. No slow motion, no\nspeed ramps, no lens flare, no colour grading toward teal or blue. The caption chip\nnever moves position. Do not add a ninth shot. Do not animate the end card.",
    "promptSourceUrl": "https://x.com/Dheepanratnam/status/2086199658091532439",
    "sourceUrl": "https://x.com/Dheepanratnam/status/2086199653922455917",
    "sourceLabel": "X 原帖 · @Dheepanratnam",
    "author": "Dheepan Ratnam",
    "publishedAt": "2026-08-08T21:15:31.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086199653922455917.mp4",
    "posterUrl": "/posters/x/2086199653922455917.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "someone got up at four so you didn't have to\nGpt2 images->Minimax H3 \nAn instagram Ad post video\n\nprompt + ref images ⤵️ @Hailuo_AI",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 21,
      "views": 1305,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X caption and same-author reply publish the complete PROOF bakery-delivery prompt, reference locks, eight timed cuts, caption chips, and continuous sunrise progression."
  },
  {
    "id": "x-2086122377633575169",
    "title": "Kaze 球员属性界面启动",
    "titleEn": "Kaze’s Player Stats Screen Powers Up",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "角色与足球从暗色扫描中显现，姓名、等级、总评与各项属性依次完成动画填充。 原生视频规格为 15 秒 · 720×1280。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Kaze and the football emerge from a dark scanner pass as the nameplate, level, overall rating, and attribute bars animate into place. Native-video output: 15s · 720×1280. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "@[ref img] is the sole visual authority for Kaze, the football, the complete player-stats interface, typography, numbers, icons, colors, lighting, layout, and vertical composition. Animate this exact screen without redesigning it. Preserve all existing labels and final stat values.\n[Goal]\nCreate a polished 15-second anime sports-game player stats screen opening. Use one continuous, locked, straight-on full-screen composition with no camera movement, cuts, cropping, or perspective distortion.\n\n[Sequence]\n0–2.2 seconds: Begin from near-black teal. A faint horizontal scanner passes downward, revealing Kaze’s silhouette and the outer edges of the interface. Fine particles and dim mint circuitry flicker awake.\n2.2–5.5 seconds: The header, KAZE nameplate, role badge, level panel, and overall panel resolve through clean line-draw animations. Kaze emerges fully from shadow, takes a controlled breath, subtly shifts her shoulders and raised arm into the referenced pose, then fixes an intense gaze toward the viewer. Her hair, ribbons, and loose clothing react naturally to a growing current of wind. The football begins a slow, stable rotation beside her hand.\n\n5.5–10.8 seconds: The large overall score counts rapidly upward and lands precisely on 92. The level settles on 46. Each attribute bar fills smoothly from left to right in sequence—Speed 93, Power 91, Control 88, Stamina 90, Agility 94, Technique 89—each landing with a restrained mint pulse. The radar chart draws outward from its center and locks into the exact final polygon. Skill cards activate from top to bottom; their icons flare briefly while the ACTIVE and ULTIMATE states illuminate.\n10.8–13.3 seconds: The player ID strip and bottom navigation fade and slide into their exact final positions. The wind energy around Kaze accelerates clockwise, wrapping around the rotating ball and sweeping behind her body with layered luminous trails. Kaze tightens her hand, leans slightly into the current, and gives one natural blink; preserve her identity, anatomy, costume, and right-side placement.\n\n13.3–15 seconds: The energy arc reaches a bright controlled crest, then settles into a living idle pulse. Hold the fully assembled screen matching @[ref img] exactly. Kaze continues subtle breathing; hair and ribbons drift, the football rotates slowly, particles shimmer, and all stats remain stable and readable.\n\nUse crisp premium game-UI motion, clean 2D anime character animation, subtle depth between Kaze and the interface, and stable legible typography. Keep every panel, icon, label, number, and geometric boundary fixed once revealed. Do not introduce new text, extra characters, additional objects, logos, captions, or UI elements.\n\nAudio: low electronic boot hum, delicate scanning ticks, short confirmation tones as values lock, rising airy wind around the ball, soft cloth movement, and one refined completion chime. No dialogue and no music.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2086122662858719410",
    "sourceUrl": "https://x.com/aimikoda/status/2086122377633575169",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-08T16:08:27.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086122377633575169.mp4",
    "posterUrl": "/posters/x/2086122377633575169.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Player stats UI animation with GPT Image 2 and MiniMax H3.\n\nThis model is amazing for UI animations. Give GPT Image 2 your character image and ask it to create a stats UI. Then use that UI as a reference for MiniMax H3.\n\nYou can check the prompt in the replies.",
    "engagement": {
      "replies": 15,
      "reposts": 70,
      "likes": 707,
      "views": 39944,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X caption and creator reply publish the complete player-stats interface prompt with locked UI, Kaze identity, score values, motion, and sound.",
    "featured": {
      "at": "2026-09-05T15:50:15.688Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2086121943615541389",
    "title": "花园噩耗的双模型表演对照",
    "titleEn": "A Garden Tragedy Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "同一张古装花园人物图与同一段噩耗表演提示被用于 MiniMax H3 和 Grok 的并排情绪对照。 原生视频规格为 27 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The same period-garden reference and tragic performance prompt drive a side-by-side emotional comparison between MiniMax H3 and Grok. Native-video output: 27s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15秒古装电影写实剧情，花海庭园氛围，镜头丝滑自然推进，情绪反转强烈，人物表演细腻真实，有明显的“悠闲惬意 -> 突闻噩耗 -> 身体先崩 -> 情绪失控前的不可置信”递进。女人一开始站在花丛中悠闲赏景，手中轻轻摇着团扇，神情放松，眼神柔和，肩膀、手腕和呼吸都是松弛的，衣袖、发饰和流苏在微风里轻轻晃动，整个人毫无防备。画外突然传来带着慌乱和哭腔的女声：“夫人！老爷被斩首了！”\n\n0-3秒：镜头穿过前景花丛丝滑推近到女人半身，女人悠闲地轻轻扇着团扇，神情安静惬意，嘴角甚至有一丝若有若无的闲适笑意，正在轻松看风景。\n\n3-6秒：画外音猛地刺入：“夫人！老爷被斩首了！”女人的扇风动作瞬间断掉，手腕僵在半空，眼神猛地定住，呼吸像被一下掐断，头极轻地偏过去一点，像不敢相信自己听到了什么，脸上的放松感瞬间消失。\n\n6-10秒：她猛地往后退一步，肩膀本能一缩，后背绷紧，脚下发软，差点站不稳跌倒，另一只手下意识抬起想抓住什么稳住自己却落空。与此同时，拿扇子的手因惊吓脱力，团扇从指间滑落，掉进脚边花丛里。她眼睛失神般睁大，嘴唇微张，想说话却几乎发不出声，只能失神低语：“你……说什么？”\n\n10-15秒：镜头迅速压近到脸和上半身。她勉强站住，但整个人像魂被抽空，胸口呼吸乱掉，眼眶迅速泛红，泪意猛地涌上来。她不是立刻大哭，而是先陷入更狠的不可置信崩塌状态，表情从空白到震惊再到痛得不敢承认，嘴唇发抖，眼神溃散，像下一秒就要彻底撑不住。\n\n音效与配乐：前半段是柔和古风环境音与轻微风声、衣袖与花叶摩擦声；噩耗传来的一瞬间配乐骤停半拍，再压入低沉悲音和压迫弦乐；团扇掉进花丛要有清晰轻响；整体情绪要像中国古装电影中的突变悲剧时刻。\n\n负面限制：不要夸张嚎哭，不要立刻瘫倒，不要卡通化表演，不要慢动作，不要过多甩头动作，不要让团扇乱飞太远，不要字幕。",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2086121943615541389",
    "sourceUrl": "https://x.com/ou_zhen599/status/2086121943615541389",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-08-08T16:06:44.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086121943615541389.mp4",
    "posterUrl": "/posters/x/2086121943615541389.jpg",
    "duration": 27,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3和Grok lmagine Video 1.5到底谁更好？\n\nI decided to run a comparison test using the same image and the same set of prompts with MiniMax H3 and Grok Imagine Video 1.5—and the results were “shocking.”\nMiniMax H3 clearly put in a more energetic performance, though the neck muscles looked a bit too “muscular.” In contrast, I’ve always thought Grok Imagine Video’s rendering of flesh tones is quite good, but the expression of certain emotions feels a bit “restrained”—though I’m not sure if that’s due to the model’s understanding.\n\nMiniMax H3明显表演的更卖力，颈部肌肉有些过于“有力”。相对Grok lmagine Video的肉感我觉得一直是不错的，但有些情绪的表达不知道是不是模型理解的问题会感觉有些“克制”。\nMiniMax H3在配音情绪上把握很好，点赞！\n\nVideo prompt👇\n15秒古装电影写实剧情，花海庭园氛围，镜头丝滑自然推进，情绪反转强烈，人物表演细腻真实，有明显的“悠闲惬意 -> 突闻噩耗 -> 身体先崩 -> 情绪失控前的不可置信”递进。女人一开始站在花丛中悠闲赏景，手中轻轻摇着团扇，神情放松，眼神柔和，肩膀、手腕和呼吸都是松弛的，衣袖、发饰和流苏在微风里轻轻晃动，整个人毫无防备。画外突然传来带着慌乱和哭腔的女声：“夫人！老爷被斩首了！”\n\n0-3秒：镜头穿过前景花丛丝滑推近到女人半身，女人悠闲地轻轻扇着团扇，神情安静惬意，嘴角甚至有一丝若有若无的闲适笑意，正在轻松看风景。\n\n3-6秒：画外音猛地刺入：“夫人！老爷被斩首了！”女人的扇风动作瞬间断掉，手腕僵在半空，眼神猛地定住，呼吸像被一下掐断，头极轻地偏过去一点，像不敢相信自己听到了什么，脸上的放松感瞬间消失。\n\n6-10秒：她猛地往后退一步，肩膀本能一缩，后背绷紧，脚下发软，差点站不稳跌倒，另一只手下意识抬起想抓住什么稳住自己却落空。与此同时，拿扇子的手因惊吓脱力，团扇从指间滑落，掉进脚边花丛里。她眼睛失神般睁大，嘴唇微张，想说话却几乎发不出声，只能失神低语：“你……说什么？”\n\n10-15秒：镜头迅速压近到脸和上半身。她勉强站住，但整个人像魂被抽空，胸口呼吸乱掉，眼眶迅速泛红，泪意猛地涌上来。她不是立刻大哭，而是先陷入更狠的不可置信崩塌状态，表情从空白到震惊再到痛得不敢承认，嘴唇发抖，眼神溃散，像下一秒就要彻底撑不住。\n\n音效与配乐：前半段是柔和古风环境音与轻微风声、衣袖与花叶摩擦声；噩耗传来的一瞬间配乐骤停半拍，再压入低沉悲音和压迫弦乐；团扇掉进花丛要有清晰轻响；整体情绪要像中国古装电影中的突变悲剧时刻。\n\n负面限制：不要夸张嚎哭，不要立刻瘫倒，不要卡通化表演，不要慢动作，不要过多甩头动作，不要让团扇乱飞太远，不要字幕。",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 7,
      "views": 2709,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status explicitly states the two-model comparison and publishes the full shared prompt, timed acting progression, dialogue, sound, and negative constraints."
  },
  {
    "id": "x-2086086171117969702",
    "title": "里维埃拉海岸的编织手袋",
    "titleEn": "A Woven Handbag on the Riviera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "银链编织手袋始终作为视觉中心，在海岸公路、敞篷车与夏日造型之间保持产品结构一致。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A silver-chain woven handbag remains the visual anchor across a Riviera road, convertible, and summer-fashion sequence while preserving product construction. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15 second, 16:9 luxury fashion commercial using @Image1@Image2 as exact visual references. @Image1 is the sole product reference. Preserve the bag’s exact proportions, pale grey white woven textile, irregular rectangular stitching, soft structure, curved flap, silver clasp, four black eyelets, side construction, zipper details, and hardware. Do not reinterpret it as leather. The bag has two separate short straps made from silver chain woven with matching textile. Preserve their exact construction, attachment points, resting positions, and maximum lengths. The front strap connects only the two front eyelets. The rear strap connects only the two rear eyelets. They must never cross, merge, disappear, duplicate, lengthen, or be rerouted. @Image2 is the exact character reference. Preserve the woman’s face, green eyes, straight blonde bob, natural proportions, and understated makeup. Dress her consistently in a sophisticated beige silk summer mini dress with a square neckline, narrow shoulder straps, fitted waist, and softly moving skirt. Add ivory slingback heels and small pearl earrings. The bag is the visual anchor throughout. It must always be the largest, brightest, sharpest, or most prominently framed element. The woman, hotel, landscape, and car are supporting elements only.\n0 to 1.75 seconds: Tight product shot of the bag alone on a sunlit limestone console beside a softly shimmering Riviera pool. Directional sunlight reveals the textile weave, stitching, two straps, four eyelets, side details, and clasp. 1.75 to 3.25 seconds: Keep the bag large and tack sharp in the foreground while the woman approaches through the bright hotel suite in soft focus. She has not touched it yet. 3.25 to 5 seconds: Extreme close up of her fingers lifting both straps together. Keep the front and rear straps visibly separate, correctly routed, and uncrossed. Show realistic chain tension, textile threaded through the links, and accurate hands. 5 to 7 seconds: She places both short straps together over one shoulder and wears the bag as a side shoulder bag. It hangs close beneath her arm, high against the side of her upper waist. It must never reach her hip or thigh. Both straps remain visibly separate, uncrossed, and at their true maximum length. 7 to 9 seconds: Close tracking shot through a sunlit hotel colonnade, composed mainly around the bag moving naturally against the beige silk dress. The straps respond realistically to her steps, shoulder movement, tension, and gravity. 9 to 11 seconds: The bag stands upright and tack sharp on a pale stone terrace table. Its two straps settle naturally and remain identifiable. The woman sits behind it in soft focus, looking toward the sea. 11 to 13 seconds: She carries the bag on one shoulder toward a cream vintage convertible. In a tight product focused montage, she opens the driver door, removes both straps from her shoulder, and carefully places the bag upright on the front passenger seat cushion. The bag’s entire base must rest directly on the horizontal cushion where a passenger sits. It is not on the backrest, headrest, seat edge, center console, or top of the seat. The passenger backrest rises behind it. Both straps settle naturally beside and partly behind the bag without disappearing. She enters the driver seat. Keep the car cropped and secondary. 13 to 15 seconds: The car moves along the Riviera coast. Keep the bag upright, large, and tack sharp on the passenger seat cushion while the woman holds steering wheel and drives in the soft background. The sea and road pass through the windows in gentle motion blur. Moving sunlight reveals the woven textile, stitching, eyelets, two distinct straps, and hardware. End with one controlled flash across the silver clasp.\nUse bright, fashion forward European luxury cinematography inspired by a restored 1990s campaign film. Directional Mediterranean sunlight, dimensional shadows, layered reflections, warm 35 mm grain, gentle haliation, creamy highlights, subtle film weave, natural motion blur, and tactile textile detail. Use elegant editorial cuts, realistic movement, accurate hands, and consistent spatial logic. Add a sophisticated original fashion score with crisp percussion, warm bass, airy electronic textures, subtle chain sounds, clasp click, silk movement, footsteps, coastal ambience, car door movement, and quiet engine sound. No dialogue. No product redesign, leather texture, colour change, altered stitching, missing or merged straps, crossed straps, extra straps, extended straps, changed eyelet routing, crossbody carry, low hip placement, warped chains, changed hardware, duplicated bag, distorted hands, outfit changes, blue dress, car hero shot, wide vehicle profile, empty road shot, incorrect seat placement, text, or subtitles.",
    "promptSourceUrl": "https://x.com/Dustfinger2077/status/2086086171117969702",
    "sourceUrl": "https://x.com/Dustfinger2077/status/2086086171117969702",
    "sourceLabel": "X 原帖 · @Dustfinger2077",
    "author": "DUST",
    "publishedAt": "2026-08-08T13:44:35.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086086171117969702.mp4",
    "posterUrl": "/posters/x/2086086171117969702.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "retro"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing MinimaxH3 on a 15s Riviera luxury commercial. Product fidelity only. No redesigns. No strap fails.\nMade with  #MinimaxH3  on @Hailuo_AI \nPrompt: \nCreate a 15 second, 16:9 luxury fashion commercial using @Image1@Image2 as exact visual references. @Image1 is the sole product reference. Preserve the bag’s exact proportions, pale grey white woven textile, irregular rectangular stitching, soft structure, curved flap, silver clasp, four black eyelets, side construction, zipper details, and hardware. Do not reinterpret it as leather. The bag has two separate short straps made from silver chain woven with matching textile. Preserve their exact construction, attachment points, resting positions, and maximum lengths. The front strap connects only the two front eyelets. The rear strap connects only the two rear eyelets. They must never cross, merge, disappear, duplicate, lengthen, or be rerouted. @Image2 is the exact character reference. Preserve the woman’s face, green eyes, straight blonde bob, natural proportions, and understated makeup. Dress her consistently in a sophisticated beige silk summer mini dress with a square neckline, narrow shoulder straps, fitted waist, and softly moving skirt. Add ivory slingback heels and small pearl earrings. The bag is the visual anchor throughout. It must always be the largest, brightest, sharpest, or most prominently framed element. The woman, hotel, landscape, and car are supporting elements only.\n0 to 1.75 seconds: Tight product shot of the bag alone on a sunlit limestone console beside a softly shimmering Riviera pool. Directional sunlight reveals the textile weave, stitching, two straps, four eyelets, side details, and clasp. 1.75 to 3.25 seconds: Keep the bag large and tack sharp in the foreground while the woman approaches through the bright hotel suite in soft focus. She has not touched it yet. 3.25 to 5 seconds: Extreme close up of her fingers lifting both straps together. Keep the front and rear straps visibly separate, correctly routed, and uncrossed. Show realistic chain tension, textile threaded through the links, and accurate hands. 5 to 7 seconds: She places both short straps together over one shoulder and wears the bag as a side shoulder bag. It hangs close beneath her arm, high against the side of her upper waist. It must never reach her hip or thigh. Both straps remain visibly separate, uncrossed, and at their true maximum length. 7 to 9 seconds: Close tracking shot through a sunlit hotel colonnade, composed mainly around the bag moving naturally against the beige silk dress. The straps respond realistically to her steps, shoulder movement, tension, and gravity. 9 to 11 seconds: The bag stands upright and tack sharp on a pale stone terrace table. Its two straps settle naturally and remain identifiable. The woman sits behind it in soft focus, looking toward the sea. 11 to 13 seconds: She carries the bag on one shoulder toward a cream vintage convertible. In a tight product focused montage, she opens the driver door, removes both straps from her shoulder, and carefully places the bag upright on the front passenger seat cushion. The bag’s entire base must rest directly on the horizontal cushion where a passenger sits. It is not on the backrest, headrest, seat edge, center console, or top of the seat. The passenger backrest rises behind it. Both straps settle naturally beside and partly behind the bag without disappearing. She enters the driver seat. Keep the car cropped and secondary. 13 to 15 seconds: The car moves along the Riviera coast. Keep the bag upright, large, and tack sharp on the passenger seat cushion while the woman holds steering wheel and drives in the soft background. The sea and road pass through the windows in gentle motion blur. Moving sunlight reveals the woven textile, stitching, eyelets, two distinct straps, and hardware. End with one controlled flash across the silver clasp.\nUse bright, fashion forward European luxury cinematography inspired by a restored 1990s campaign film. Directional Mediterranean sunlight, dimensional shadows, layered reflections, warm 35 mm grain, gentle haliation, creamy highlights, subtle film weave, natural motion blur, and tactile textile detail. Use elegant editorial cuts, realistic movement, accurate hands, and consistent spatial logic. Add a sophisticated original fashion score with crisp percussion, warm bass, airy electronic textures, subtle chain sounds, clasp click, silk movement, footsteps, coastal ambience, car door movement, and quiet engine sound. No dialogue. No product redesign, leather texture, colour change, altered stitching, missing or merged straps, crossed straps, extra straps, extended straps, changed eyelet routing, crossbody carry, low hip placement, warped chains, changed hardware, duplicated bag, distorted hands, outfit changes, blue dress, car hero shot, wide vehicle profile, empty road shot, incorrect seat placement, text, or subtitles.",
    "engagement": {
      "replies": 12,
      "reposts": 4,
      "likes": 63,
      "views": 4738,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status publishes the complete Riviera commercial prompt with exact bag and character references, strap constraints, timed shots, and negative restrictions."
  },
  {
    "id": "x-2086063980234035362",
    "title": "海滩特工的元喜剧对话",
    "titleEn": "Secret Agents in a Meta Beach Comedy",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "两名卡通特工在海滩进行节奏紧凑的自指式对话，有限动画和硬阴影强化成人喜剧风格。 原生视频规格为 15 秒 · 1008×576。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two cartoon secret agents trade a tightly timed meta joke on a beach, using limited animation, bold outlines, and hard graphic shadows. Native-video output: 15s · 1008×576. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Style: Adult animated television comedy in the distinctive flat-vector / limited-animation style of a suave-spy cartoon. Bold inked outlines, flat cel-shaded colors, slightly stiff comedic timing, hard graphic shadows, subtle film-grain texture. 2K, cinematic but comedic.\n\nSetting: A bright sunny beach. Clear blue sky, soft sand, ocean waves in the background.\n\nCharacters:\nSterling Archer: a handsome secret agent with jet-black slicked-back hair, wearing a tight dark swimsuit. Self-satisfied, manic energy, perpetually smug.\nLana Kane: a tall, striking brown-skinned female agent, wearing a stylish bikini. Exasperated, eyes-wide \"are-you-kidding-me\" energy.\n\nShot 1 (0:00–0:03): Medium shot of Archer and Lana standing on the beach. Archer cups his hands around his mouth and calls out: <d>[English] Lana? Lana? LANA!!!!</d>\n\nShot 2 (0:03–0:05): Cut to Lana turning sharply toward him, annoyed: <d>[English] What?!</d>\n\nShot 3 (0:05–0:07): Archer leans in with a smug grin: <d>[English] With MiniMax any internet pervert can draw us!</d>\n\nShot 4 (0:07–0:08): Lana frowns, confused: <d>[English] What do you mean?</d>\n\nShot 5 (0:08–0:09): Archer points at her with a knowing look: <d>[English] They can even change your clothes!</d>\n\nShot 6 (0:09–0:10): Quick flash. Lana’s bikini instantly becomes a modern one-piece swimsuit. She looks down in surprise.\n\nShot 7 (0:10–0:11): Another quick flash. The swimsuit changes into a 1920s-style bathing suit. Lana’s eyes widen further.\n\nShot 8 (0:11–0:13): Final flash. The bathing suit disappears completely. Lana is suddenly naked and quickly covers herself with both hands, shocked and angry: <d>[English] Archer, stop that!</d>\n\nShot 9 (0:13–0:15): Archer raises both hands defensively, looking genuine: <d>[English] That’s not me!</d>\n\noverall_soundscape: Soft ocean waves and light beach ambience. Clear comedic dialogue with sharp cuts between the clothing changes.\n\nnon_diegetic_music: None. Dry, absurd, self-aware comedic timing.",
    "promptSourceUrl": "https://x.com/Cardoso/status/2086063980234035362",
    "sourceUrl": "https://x.com/Cardoso/status/2086063980234035362",
    "sourceLabel": "X 原帖 · @Cardoso",
    "author": "cardoso",
    "publishedAt": "2026-08-08T12:16:24.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086063980234035362.mp4",
    "posterUrl": "/posters/x/2086063980234035362.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1008×576",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3, @NVIDIAGeForceBR 5060ti, resolução mínima, 3min para renderizar. ZERO referências, sem LORAs, só o modelo nativo. Editei para colocar a tarja, e só. \n\nImpressionante fazer isso com uma IA local. \n\nPrompt:\n\nStyle: Adult animated television comedy in the distinctive flat-vector / limited-animation style of a suave-spy cartoon. Bold inked outlines, flat cel-shaded colors, slightly stiff comedic timing, hard graphic shadows, subtle film-grain texture. 2K, cinematic but comedic.\n\nSetting: A bright sunny beach. Clear blue sky, soft sand, ocean waves in the background.\n\nCharacters:\nSterling Archer: a handsome secret agent with jet-black slicked-back hair, wearing a tight dark swimsuit. Self-satisfied, manic energy, perpetually smug.\nLana Kane: a tall, striking brown-skinned female agent, wearing a stylish bikini. Exasperated, eyes-wide \"are-you-kidding-me\" energy.\n\nShot 1 (0:00–0:03): Medium shot of Archer and Lana standing on the beach. Archer cups his hands around his mouth and calls out: <d>[English] Lana? Lana? LANA!!!!</d>\n\nShot 2 (0:03–0:05): Cut to Lana turning sharply toward him, annoyed: <d>[English] What?!</d>\n\nShot 3 (0:05–0:07): Archer leans in with a smug grin: <d>[English] With MiniMax any internet pervert can draw us!</d>\n\nShot 4 (0:07–0:08): Lana frowns, confused: <d>[English] What do you mean?</d>\n\nShot 5 (0:08–0:09): Archer points at her with a knowing look: <d>[English] They can even change your clothes!</d>\n\nShot 6 (0:09–0:10): Quick flash. Lana’s bikini instantly becomes a modern one-piece swimsuit. She looks down in surprise.\n\nShot 7 (0:10–0:11): Another quick flash. The swimsuit changes into a 1920s-style bathing suit. Lana’s eyes widen further.\n\nShot 8 (0:11–0:13): Final flash. The bathing suit disappears completely. Lana is suddenly naked and quickly covers herself with both hands, shocked and angry: <d>[English] Archer, stop that!</d>\n\nShot 9 (0:13–0:15): Archer raises both hands defensively, looking genuine: <d>[English] That’s not me!</d>\n\noverall_soundscape: Soft ocean waves and light beach ambience. Clear comedic dialogue with sharp cuts between the clothing changes.\n\nnon_diegetic_music: None. Dry, absurd, self-aware comedic timing.",
    "engagement": {
      "replies": 9,
      "reposts": 3,
      "likes": 39,
      "views": 7419,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status identifies native local H3 generation and publishes the complete beach dialogue prompt, character actions, shot timing, audio, and comedy direction."
  },
  {
    "id": "x-2086021869040263455",
    "title": "月下仙城的剑阵决战",
    "titleEn": "Sword Formations Above a Moonlit Immortal City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "男女剑仙在月下浮空仙城迎战黑金魔修，御剑、万剑阵、魔龙与终极法相连续升级。 原生视频规格为 15 秒 · 1280×711。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two sword immortals battle a black-and-gold adversary above a moonlit floating city as flying swords, a demon dragon, and giant avatars escalate the fight. Native-video output: 15s · 1280×711. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "5秒，4:3，顶级东方仙侠3A游戏CG，电影级高端宣传片，严格参考图1的月下仙城场景与美术气质：巨大明月、云海天宫、环形仙城、悬浮仙山、瀑布、白玉仙桥、金顶楼阁、冷蓝月光。整体画面必须唯美、恢弘、神圣、危险、酷炫，像大厂正式游戏开场CG，不像普通AI视频。\n\n双主角设定：男剑仙，黑色长发，白蓝战袍，冰蓝雷光长剑，英俊冷峻，强大沉稳；女剑仙，黑色长发，白青仙裙，月白灵剑与月华法术，清冷绝美，高贵飘逸。反派为黑金重甲魔修，手持黑色巨剑，周身暗红黑色魔气，压迫感极强。人物脸和服装全程稳定统一，保持高级游戏角色建模质感。\n\n开场镜头从巨大明月前高速穿越云海与浮空仙山，俯冲进入宏伟仙城，来到白玉高台，男女主背对镜头立于高台边缘，衣袍长发飞扬，英雄感极强。远处天空突然裂开暗红色空间裂缝，魔修高速坠落，轰然落地，引发碎石、冲击波和魔气爆发。\n\n男剑仙瞬间拔剑冲出，与魔修在仙桥中央高速对撞，剑刃碰撞爆发蓝红双色能量冲击、空间波纹、火星与剑气。近战动作凌厉、清晰、有重量感，包含高速斩击、格挡、反击、短暂慢动作碰撞特写，镜头低机位跟拍与高速环绕，极具电影感。\n\n随后女剑仙御剑升空，展开巨大蓝白仙阵，发动“万剑齐发”，数百柄发光飞剑如流星暴雨般射向魔修，夜空被剑光点亮。魔修爆发黑红魔气，凝聚巨大的魔龙反击，空中连续爆炸。男剑仙御剑穿梭爆炸火光与云海，一剑斩裂魔龙，空战场面华丽震撼。\n\n最后魔修释放百米高的黑红魔神法相，遮蔽天空。男女主并肩悬空，男剑仙开启覆盖半个天空的金蓝剑阵，女剑仙召唤巨大的白蓝月华莲华法相，二人联手凝聚数百米长的金蓝神剑，与魔神巨剑正面碰撞。碰撞瞬间进入超级慢动作，蓝色雷电、金色符文、白色月华、红色魔焰与空间裂痕同时凝固，随后爆发毁天灭地的大爆炸，形成巨大的环形云洞与能量冲击波。\n\n结尾爆炸光芒散去，巨大明月重现，男女主并肩立于悬空巨剑之上俯瞰整座仙城，周围灵剑环绕，远处天宫亮起金蓝光柱，镜头缓慢拉远，形成极强的高级游戏主视觉结尾，并预留后期添加Logo空间。\n\n镜头要求：超广角航拍、俯冲推进、低机位英雄视角、高速跟拍、环绕镜头、御剑追踪、短暂升格慢动作、终极碰撞震动、史诗级拉远收尾。镜头必须流畅稳定，不乱晃，不跳轴。\n\n视觉要求：特效炫丽但高级干净，不能廉价光污染。强调冰蓝雷电剑气、金色仙纹、月华粒子、飞剑阵、空间裂缝、黑红魔焰、魔龙、巨大法相、神剑、冲击波、体积云、瀑布水雾、发光碎片。人物主体始终清晰，画面精致通透，PBR材质，真实布料与长发动效，电影级光影与空间层次。\n\n音效：原生立体声。包含高空风声、远处钟鸣、低沉鼓点、剑刃碰撞、破风声、雷电炸裂、万剑掠空、魔龙低吼、终极爆炸低频冲击。无对白，无字幕，无水印，无乱码。",
    "promptSourceUrl": "https://x.com/Stellakjbk/status/2086021965815373850",
    "sourceUrl": "https://x.com/Stellakjbk/status/2086021869040263455",
    "sourceLabel": "X 原帖 · @Stellakjbk",
    "author": "九尾狐-FoxFairy🦊",
    "publishedAt": "2026-08-08T09:29:04.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086021869040263455.mp4",
    "posterUrl": "/posters/x/2086021869040263455.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×711",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3的打斗转场非常出色，浅浅地测试了一下，做了一个国风仙侠打斗游戏GC，感觉它镜头运动挺不错，很轻易地做出很多炫酷的特效。之前做视频最难的就是做打戏，Minimax h3很好地解决了这个痛点，就是提示词尽量精简一些，不然很容易过度。大家不要仅仅局限于AI视频，游戏特效或许是它很擅长的领域，提示词在评论区👇",
    "engagement": {
      "replies": 9,
      "reposts": 3,
      "likes": 31,
      "views": 4861,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X caption labels an H3 fantasy-game fight and the creator reply publishes the complete prompt with characters, transitions, camera, VFX, and native stereo sound."
  },
  {
    "id": "x-2086020747336761388",
    "title": "女秘书策划的办公室绯闻",
    "titleEn": "The Secretary’s Office-Scandal Trap",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "复古香港办公室争执被女秘书反转成报纸头条，她最后换装现身街头并向镜头示意保密。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A retro Hong Kong office confrontation becomes a newspaper scandal engineered by the secretary, who reappears in disguise and signals secrecy to camera. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "1960年代，香港邵氏復古電影風格：戴黑粗框眼鏡、頭髮微捲及肩的霸總生氣地按住女秘書(粉紅色廉價套裝、明顯腮紅、淡藍色眼影)的肩膀，質問女秘書道\"昨晚，妳到底去了哪裡？\"。女秘書開始眼眶泛淚，推開霸總的雙手，然後轉身拭淚哭泣著說\"總裁這是我私人的事情，不用你管!\"。總裁快步接近女秘書，然後將女秘書的身體按壓在復古的竹製躺椅上，總裁認真地對女秘書說\"對不起，我錯了。\"然後總裁開始親吻女秘書的臉頰。女秘書一邊假裝喘氣，一邊從總裁的背後拿高了手機喀擦一聲拍照。下一個畫面是香港街道的路人，手中拿著報紙(上面有總裁親吻流淚女秘書的自拍照 新聞標題寫著\"OFFICE SCANDAL\")。下一個畫面是圍著時尚花頭巾的女秘書(改穿米色連身風衣、黑色高跟鞋)自信地\"哼\"一聲，從書報攤拿了一份有相同頭版照片的報紙，然後得意地看著鏡頭比了個\"shh\"的手勢。畫面fade into black.",
    "promptSourceUrl": "https://x.com/drjoetw/status/2086020747336761388",
    "sourceUrl": "https://x.com/drjoetw/status/2086020747336761388",
    "sourceLabel": "X 原帖 · @drjoetw",
    "author": "The Anxious Mind",
    "publishedAt": "2026-08-08T09:24:37.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2086020747336761388.mp4",
    "posterUrl": "/posters/x/2086020747336761388.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "retro",
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The CEO thought he could handle anything... until he met this secretary. Can he survive her love trap? 👀💘\n(created using @Hailuo_AI #MiniMaxH3)\nmicro-drama prompt below 😊🌹👇\n\n1960年代，香港邵氏復古電影風格：戴黑粗框眼鏡、頭髮微捲及肩的霸總生氣地按住女秘書(粉紅色廉價套裝、明顯腮紅、淡藍色眼影)的肩膀，質問女秘書道\"昨晚，妳到底去了哪裡？\"。女秘書開始眼眶泛淚，推開霸總的雙手，然後轉身拭淚哭泣著說\"總裁這是我私人的事情，不用你管!\"。總裁快步接近女秘書，然後將女秘書的身體按壓在復古的竹製躺椅上，總裁認真地對女秘書說\"對不起，我錯了。\"然後總裁開始親吻女秘書的臉頰。女秘書一邊假裝喘氣，一邊從總裁的背後拿高了手機喀擦一聲拍照。下一個畫面是香港街道的路人，手中拿著報紙(上面有總裁親吻流淚女秘書的自拍照 新聞標題寫著\"OFFICE SCANDAL\")。下一個畫面是圍著時尚花頭巾的女秘書(改穿米色連身風衣、黑色高跟鞋)自信地\"哼\"一聲，從書報攤拿了一份有相同頭版照片的報紙，然後得意地看著鏡頭比了個\"shh\"的手勢。畫面fade into black.\n\n#AIDrama #AIFillm #AIMovie #AIVideo #microdrama #CEO #霸道總裁",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 4792,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status explicitly labels H3 and publishes the complete micro-drama prompt with the office confrontation, staged photograph, newspaper reveal, and final gesture."
  },
  {
    "id": "x-2085986114800951530",
    "title": "温室里的害羞雨云",
    "titleEn": "A Shy Rain Cloud in the Greenhouse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "小园丁 Luma 在旧水壶中发现害羞雨云 Nimbus，一滴雨让枯萎向日葵重新抬头。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Young gardener Luma discovers shy rain cloud Nimbus inside an old watering can, and one drop revives a wilted sunflower. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for “The Little Gardener and the Runaway Rain Cloud” as the PRIMARY visual, character, environment, composition, and narrative reference.\n\nGenerate a FINISHED animated sequence from the storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- panel numbers\n- panel captions\n- timestamps\n- production notes\n- character reference sheets\n- off-white storyboard paper\n- storyboard title/header\n- handwritten annotations\n\nThe uploaded storyboard controls and locks:\n\n1. Luma’s exact character design\n2. Nimbus’s exact cloud design\n3. Sprout’s exact snail design\n4. the rooftop greenhouse architecture\n5. the wilted sunflower design\n6. the old metal watering can\n7. the gardening props\n8. warm watercolor/painterly 2D visual style\n9. character scale relationships\n10. overall color palette\n11. Part 1 story progression\n\nAnimate these designs naturally while maintaining extremely high identity consistency.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nCreate exactly:\n\nDuration: 15 seconds\nAspect Ratio: 16:9 landscape\nFrame rate feeling: cinematic 24 fps\nGenre: magical garden comedy / heartwarming fantasy\nVisual medium: premium hand-painted 2D animated short\nAudience: family-friendly\nDialogue language: English\nNative synchronized dialogue + sound effects + music\n\nEmotional progression:\n\npeaceful garden morning\n→ concern\n→ determination\n→ mysterious sound\n→ magical discovery\n→ tiny comic surprise\n→ hopeful miracle\n→ BIG IDEA cliffhanger\n\nThe storytelling must be very easy to understand even without narration.\n\nIMPORTANT:\n\nAlthough the storyboard contains 15 individual panels, DO NOT turn the video into fifteen frantic one-second cuts.\n\nConsolidate those beats into approximately 7–8 fluid cinematic shots.\n\nEach shot should contain several connected storyboard actions so the animation feels like an actual polished animated short rather than a slideshow.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — LUMA]\n\nLUMA must match the uploaded storyboard exactly.\n\nCharacter:\napproximately 8-year-old little gardener.\n\nPhysical appearance:\n- warm fair skin\n- large expressive dark brown eyes\n- soft round childlike face\n- freckles across cheeks/nose if visible in reference\n- fluffy auburn / reddish-brown hair\n- hair gathered into two messy buns\n- several loose curly strands around face\n\nSignature clothing:\n- oversized warm yellow gardening hat\n- mint-green gardening overalls\n- coral-and-cream horizontally striped shirt\n- red gardening boots\n- tiny brown gardening/seed pouch attached at waist\n\nCharacter proportions:\n- small child stature\n- slightly oversized head\n- expressive hands\n- rounded appealing silhouette\n\nPersonality:\n- caring\n- cheerful\n- imaginative\n- curious\n- determined when a plant needs help\n- easily excited by magical discoveries\n\nPerformance should transition:\n\nhappy gardener\n→ worried plant caretaker\n→ determined problem solver\n→ curious investigator\n→ amazed child\n→ excited inventor\n\nCRITICAL:\n\nLuma must remain visually identical in every shot.\n\nDo not change:\n- face\n- age\n- hairstyle\n- hair color\n- hat\n- clothing\n- boots\n- proportions\n- eye color\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NIMBUS]\n\nNimbus must match the cloud discovered inside the watering can in the storyboard.\n\nNimbus is a tiny living rain cloud approximately pillow-sized.\n\nDesign:\n- soft fluffy blue-white cloud body\n- subtle watercolor edges\n- very faint pale blue shadows\n- large round expressive dark eyes\n- tiny eyebrows\n- small cute mouth\n- slight rosy blush\n- tiny dangling raindrop-shaped feet\n- soft semi-translucent cloud fluff\n- no arms unless naturally suggested by cloud shape\n\nPersonality:\n- shy\n- timid\n- gentle\n- easily embarrassed\n- harmless\n- sweet\n\nNimbus begins PART 1 hiding inside the watering can.\n\nHe must NOT appear in the greenhouse before the discovery sequence.\n\nHis first reveal should feel magical and surprising.\n\nNimbus should remain very small.\n\nDO NOT:\n- enlarge Nimbus\n- darken Nimbus\n- create storm powers\n- add flower crown yet\n- make him Super Cloud\n- generate multiple Nimbus clouds\n- create lightning\n\nThose belong to later story beats.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — SPROUT]\n\nSprout must match the storyboard exactly.\n\nDesign:\n- tiny green garden snail\n- soft rounded body\n- large tall expressive eye stalks\n- warm brown spiral shell\n- tiny yellow flower growing beside/on shell\n- same miniature scale throughout\n\nPersonality:\n- slow\n- observant\n- dramatic\n- suspicious\n- comedic reaction character\n\nSprout has no human speech unless specified.\n\nUse:\n- tiny squeaks\n- gasps\n- nervous chirps\n- eye-stalk reactions\n\nSprout should frequently understand the danger of Luma’s ideas before Luma does.\n\n--------------------------------------------------\n\n[PROP IDENTITY LOCK — WILTED SUNFLOWER]\n\nUse the exact sunflower from the storyboard.\n\nAt the start:\n- heavily drooping yellow petals\n- bowed flower head\n- wilted leaves\n- weak stem\n- planted in rectangular wooden planter\n- still alive\n\nAfter ONE drop from Nimbus:\nthe flower should perk up slightly.\n\nIMPORTANT:\n\nDo NOT magically transform it into a giant perfect sunflower yet.\n\nOnly show:\n- stem becoming slightly stronger\n- leaves lifting\n- flower head rising a little\n- color becoming slightly healthier\n\nThis is a small visual miracle.\n\n--------------------------------------------------\n\n[PROP IDENTITY LOCK — WATERING CAN]\n\nUse the same old watering can from the storyboard.\n\nAppearance:\n- weathered gray/blue metal\n- rounded main body\n- curved metal handle\n- long narrow spout\n- slightly worn surface\n- small painted flower decoration if visible\n- large enough for tiny Nimbus to curl inside\n\nIt is an ordinary gardening can.\n\nDo not transform it into a magical object.\n\nNimbus is simply hiding inside.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — ROOFTOP GREENHOUSE]\n\nMaintain exactly the warm rooftop greenhouse established by the storyboard.\n\nEnvironment:\n- glass-paneled greenhouse walls\n- glass roof\n- wooden structural framing\n- warm sunlight pouring through glass\n- hanging vines\n- multiple wooden planter boxes\n- terracotta flower pots\n- colorful flowers\n- herbs\n- leafy plants\n- gardening shelves\n- small ladder\n- watering hose\n- gardening tools\n- seed packets\n- old watering can\n- pots and wooden crates\n- distant soft city rooftop skyline through greenhouse glass\n\nVisual feeling:\nhandmade,\nlived-in,\ncozy,\nsafe,\nsunny,\nlush,\nmagical without obvious fantasy effects at the beginning.\n\nMaintain coherent geography throughout.\n\nThe sunflower planter, watering can area, shelves, and main walkway must not randomly move between shots.\n\n--------------------------------------------------\n\n[WORLD / MAGIC RULES]\n\nThe world begins mostly realistic.\n\nNimbus introduces the first visible magical element.\n\nMagic should be:\n- delicate\n- soft\n- believable within a children’s animated world\n- visually elegant\n- never explosive\n\nNimbus creates rain naturally.\n\nCause-and-effect must be clear:\n\nsunflower is dry\n→ Luma searches for water\n→ watering can sneezes\n→ Luma investigates\n→ Nimbus is discovered\n→ Nimbus nervously sneezes\n→ one raindrop falls\n→ sunflower responds\n→ Luma realizes Nimbus can help\n→ Luma gets a BIG IDEA\n\nNo unexplained magical events.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] (0.00s–2.00s)\nTHE PEACEFUL ROOFTOP GARDEN\n\nOpen with a beautiful wide establishing shot of the rooftop greenhouse.\n\nWarm morning sunlight enters through the glass ceiling.\n\nTiny dust particles float through the golden beams.\n\nLeaves move gently in a rooftop breeze.\n\nLuma happily walks between planter boxes carrying a small basket of seeds/gardening tools.\n\nSprout slowly crawls along the edge of a wooden planter beside her.\n\nUse subtle environmental life:\n- small leaves swaying\n- hanging vines moving\n- distant city haze\n- butterflies or tiny garden insects only if subtle\n\nCamera:\n\nbegin outside/near the glass roof\n→ smooth gentle push through the greenhouse\n→ settle into a medium-wide shot of Luma tending plants.\n\nLuma hums quietly.\n\nShe pats soil around a healthy flower.\n\nDialogue:\n\nLuma, cheerful:\n<d>[English] Good morning, little guys!</d>\n\nShe notices Sprout.\n\nLuma:\n<d>[English] Morning, Sprout.</d>\n\nSprout gives a tiny cheerful squeak.\n\nSound:\n- light morning breeze\n- gentle birds outside\n- leaves rustling\n- gardening basket movement\n- soft snail squeak\n- quiet greenhouse ambience\n\nMusic:\ndelicate whimsical opening using pizzicato strings, gentle marimba, and soft flute.\n\n--------------------------------------------------\n\n[SHOT 2] (2.00s–4.20s)\nTHE WILTING SUNFLOWER\n\nAs Luma continues through the garden, her smile slowly fades.\n\nHer gaze lands on the drooping sunflower.\n\nCut to a closer view:\n\nthe sunflower’s head hangs sadly downward.\n\nIts leaves droop.\n\nSprout crawls beside the planter and slowly raises both eye stalks toward the flower.\n\nOne eye stalk leans sideways with concern.\n\nLuma immediately kneels.\n\nShe gently lifts one wilted leaf between her fingers.\n\nCamera:\n\nmedium tracking with Luma\n→ rack focus from healthy flowers to wilted sunflower\n→ close-up flower\n→ intimate medium-close shot of Luma kneeling.\n\nDialogue:\n\nLuma:\n<d>[English] Oh no...</d>\n\nShe gently examines the plant.\n\nLuma:\n<d>[English] Sunny, what happened?</d>\n\nShe touches the dry soil.\n\nLuma:\n<d>[English] You’re thirsty.</d>\n\nSprout gives a worried tiny squeak.\n\nLuma’s worried expression becomes determined.\n\nLuma:\n<d>[English] Don’t worry. I’ll find water.</d>\n\nSound:\n- soft soil touch\n- leaf rustle\n- subtle worried musical note\n\nKeep this emotionally sincere but brief.\n\n--------------------------------------------------\n\n[SHOT 3] (4.20s–6.30s)\nTHE SEARCH\n\nLuma stands quickly and begins checking the greenhouse.\n\nShe looks at:\n- empty small watering cup\n- dry watering station\n- hose\n- gardening shelf\n\nShe trots along the walkway.\n\nSprout follows at an absurdly slow speed.\n\nUse a brief visual comedy beat:\n\nLuma rushes past frame.\n\nA moment later...\n\nSprout is still only a few centimeters farther along.\n\nLuma then notices the old metal watering can sitting among leafy pots.\n\nHer eyes brighten.\n\nCamera:\n\nside tracking shot following Luma\n→ small comedic static shot of slow Sprout\n→ quick pan toward watering can as Luma notices it.\n\nDialogue:\n\nLuma:\n<d>[English] Water... water...</d>\n\nShe spots the can.\n\nLuma:\n<d>[English] Aha!</d>\n\nSound:\n- light footsteps\n- tool clinks\n- leaves brushing clothes\n- tiny Sprout effort squeak\n\nMusic becomes slightly more playful.\n\n--------------------------------------------------\n\n[SHOT 4] (6.30s–8.30s)\nTHE WATERING CAN SNEEZES\n\nLuma approaches the old watering can.\n\nBefore she reaches it—\n\nsomething inside shifts.\n\nVery subtle.\n\nLuma freezes.\n\nThe watering can gives a tiny wobble.\n\nThen:\n\n<d>[English] Ah... ah...</d>\n\nLuma slowly leans closer.\n\nSprout’s eye stalks rise behind her.\n\nSuddenly from inside:\n\n<d>[English] ACHOO!</d>\n\nA tiny puff of cool mist shoots gently from the watering-can spout.\n\nThe can rocks slightly.\n\nLuma jumps backward only a few centimeters in surprise.\n\nSprout instantly retracts slightly into his shell.\n\nNo danger.\n\nCamera:\n\nslow push-in on watering can\n→ close-up of its dark opening\n→ Luma listening\n→ mist puff from spout\n→ reaction two-shot.\n\nDialogue:\n\nLuma:\n<d>[English] ...Did you just sneeze?</d>\n\nBeat.\n\nShe looks toward Sprout.\n\nLuma:\n<d>[English] Watering cans don’t sneeze.</d>\n\nSprout gives a tiny uncertain squeak.\n\nSound:\n- metal can creak\n- tiny movement from inside\n- adorable sneeze\n- mist “pfft”\n- light metallic wobble\n\nMusic:\nbrief curious plucked-string motif.\n\n--------------------------------------------------\n\n[SHOT 5] (8.30s–10.30s)\nTHE IMPOSSIBLE DISCOVERY\n\nLuma carefully crouches beside the watering can.\n\nShe grips its edge gently.\n\nSlowly...\n\nshe peers inside.\n\nCamera moves over Luma’s shoulder toward the dark interior.\n\nFor one beat we see only darkness.\n\nThen two tiny eyes blink.\n\nSoft pale-blue cloud fluff emerges.\n\nNimbus is curled inside the watering can.\n\nHe looks up at Luma.\n\nNimbus immediately becomes embarrassed and tries to shrink deeper into the can.\n\nLuma’s expression transforms from caution to pure wonder.\n\nSprout stretches upward from beside the can to see inside.\n\nCamera:\n\nover-shoulder POV\n→ gradual focus transition revealing Nimbus\n→ close-up Nimbus eyes\n→ reaction close-up Luma\n→ wider three-character composition.\n\nDialogue:\n\nLuma, whispering:\n<d>[English] Whoa...</d>\n\nNimbus nervously:\n<d>[English] ...Hi.</d>\n\nLuma:\n<d>[English] You’re a cloud.</d>\n\nNimbus gives a shy tiny nod.\n\nLuma, amazed:\n<d>[English] A tiny cloud!</d>\n\nSound:\n- subtle magical shimmer\n- tiny cloud puff\n- soft metal resonance from inside can\n- Nimbus’s faint watery ambience\n\nMusic:\nintroduce delicate glockenspiel and airy flute.\n\n--------------------------------------------------\n\n[SHOT 6] (10.30s–12.20s)\nTHE TINY RAIN DROP\n\nNimbus slowly floats just above the watering-can opening.\n\nHe is nervous about being watched.\n\nLuma leans closer with fascinated eyes.\n\nSprout watches from below.\n\nNimbus’s little face scrunches.\n\nHis cloud body briefly compresses.\n\nNimbus:\n\n<d>[English] Ah... ah...</d>\n\nLuma realizes another sneeze is coming.\n\nNimbus:\n\n<d>[English] Achoo!</d>\n\nA single sparkling raindrop pops from the underside of Nimbus.\n\nIMPORTANT:\n\nOnly ONE meaningful raindrop.\n\nCamera follows the droplet in a graceful macro tracking shot.\n\nThe droplet catches sunlight and briefly shows a subtle rainbow refraction.\n\nIt falls toward the sunflower planter.\n\nSound:\n- cute cloud sneeze\n- tiny watery “plink”\n- soft magical shimmer\n\nLuma:\n<d>[English] Wait...!</d>\n\n--------------------------------------------------\n\n[SHOT 7] (12.20s–13.50s)\nONE DROP IS ENOUGH\n\nExtreme close-up:\n\nthe raindrop lands in the dry soil near the sunflower stem.\n\nA soft ring of moisture spreads through the soil.\n\nNothing explosive.\n\nThe sunflower responds gently.\n\nIts stem firms.\n\nOne leaf slowly lifts.\n\nThe flower head rises a few centimeters.\n\nIts yellow petals regain a little warmth.\n\nLuma watches from the background.\n\nHer mouth falls open.\n\nThen she clasps her hands with delighted excitement.\n\nNimbus looks surprised by what he just did.\n\nSprout’s eyes become enormous.\n\nCamera:\n\nmacro raindrop impact\n→ tilt upward following sunflower\n→ rack focus to Luma’s astonished face.\n\nDialogue:\n\nLuma:\n<d>[English] You did that?</d>\n\nNimbus, shy:\n<d>[English] I... think so.</d>\n\nLuma:\n<d>[English] One drop!</d>\n\nSound:\n- small water impact\n- subtle plant movement\n- soft magical chime\n- leaf unfurl sound\n\nMusic:\nwarm hopeful melodic lift.\n\n--------------------------------------------------\n\n[SHOT 8] (13.50s–15.00s)\nTHE BIG IDEA — CLIFFHANGER\n\nLuma looks from the revived sunflower...\n\nto Nimbus...\n\nthen slowly around at the dozens of plants filling the greenhouse.\n\nHer eyes widen.\n\nA playful “idea” expression spreads across her face.\n\nShe turns toward Nimbus with a huge excited smile and raises one finger.\n\nNimbus immediately becomes slightly nervous.\n\nSprout looks at Luma.\n\nThen at Nimbus.\n\nThen at the enormous greenhouse.\n\nSprout’s eye stalks shoot upward in comic alarm.\n\nLuma excitedly gestures toward all the plants.\n\nDialogue:\n\nLuma:\n<d>[English] Nimbus...</d>\n\nTiny pause.\n\nLuma:\n<d>[English] You can help ALL my flowers!</d>\n\nNimbus:\n<d>[English] ...All?</d>\n\nSprout gives a loud alarmed:\n\n<d>[English] Eep!</d>\n\nHold briefly on the three expressions:\n\nLuma = thrilled\n\nNimbus = shy uncertainty\n\nSprout = “this is going to be a disaster”\n\nCamera:\n\nmedium shot behind Luma looking across greenhouse\n→ quick close-up on Luma’s excited realization\n→ reaction close-up Nimbus\n→ reaction close-up Sprout\n→ final wide three-character composition.\n\nEnd with a gentle comedic musical sting.\n\nCUT TO BLACK.\n\nDO NOT show Luma upgrading Nimbus.\n\nDO NOT begin Part 2.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\nLUMA VOICE:\n\nNatural young girl voice, approximately 8 years old.\n\nVoice qualities:\n- warm\n- bright\n- playful\n- emotionally expressive\n- curious\n- not overly cartoonish\n- clear natural English\n\nHer voice progression:\n\n0–4 sec:\ngentle and caring\n\n4–8 sec:\ndetermined then curious\n\n8–12 sec:\nsoft wonder\n\n12–15 sec:\nhigh-energy excitement\n\n--------------------------------------------------\n\nNIMBUS VOICE:\n\nVery soft youthful cloud voice.\n\nCharacteristics:\n- shy\n- slightly breathy\n- adorable\n- gentle\n- innocent\n- quiet\n- slightly higher pitched than Luma but easily understandable\n\nDo NOT make Nimbus sound robotic.\n\nHis sneezes should be small and cute.\n\n--------------------------------------------------\n\nSPROUT AUDIO:\n\nNo full normal dialogue.\n\nUse:\n- tiny squeaks\n- nervous chirps\n- tiny gasps\n- comic “eep”\n\nDo not give Sprout a human adult voice.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nMaintain a rich but soft greenhouse sound bed:\n\n- distant rooftop birds\n- very light city ambience far below\n- glass greenhouse creaks\n- subtle wind\n- leaves rustling\n- hanging plants moving\n- gardening tools\n- footsteps on stone/wood\n- soil movement\n- watering-can metal noises\n- small water droplets\n- Nimbus cloud puffs\n- Sprout squeaks\n\nNever overpower dialogue.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an original whimsical family-animation score.\n\nInstrumentation:\n- gentle pizzicato strings\n- marimba\n- soft acoustic plucks\n- airy flute\n- glockenspiel\n- very subtle warm pads\n\nMusical arc:\n\n0.00–2.00\npeaceful garden theme.\n\n2.00–4.20\nslightly concerned but still warm.\n\n4.20–6.30\nplayful search rhythm.\n\n6.30–8.30\ncurious comedic suspense.\n\n8.30–10.30\nmagical discovery.\n\n10.30–13.50\nwonder and hopeful warmth.\n\n13.50–15.00\nplayful “big idea” motif ending on light comedic suspense.\n\nNo epic music.\nNo scary music.\nNo dramatic disaster music.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nUse cinematic storybook camera work.\n\nOverall camera rhythm:\n\nBeginning:\nwide, calm, observational.\n\nProblem:\ncloser emotional coverage.\n\nMystery:\nslow push-ins.\n\nDiscovery:\nintimate close-ups.\n\nMagic:\nmacro insert + graceful tracking.\n\nFinal idea:\nreaction-driven comedy framing.\n\nPreferred techniques:\n- gentle dolly\n- subtle parallax\n- slow push-in\n- rack focus\n- over-the-shoulder composition\n- macro raindrop shot\n- reaction close-ups\n- wide environmental establishing shot\n\nAvoid:\n- handheld shake\n- crash zoom\n- extreme Dutch angles\n- fast action-camera motion\n- excessive cuts\n- 360-degree spins\n- random camera movement\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animated short.\n\nMatch the storyboard’s exact aesthetic:\n\n- delicate ink outlines\n- watercolor-like painted fills\n- warm children’s-book texture\n- cinematic lighting\n- painterly background detail\n- softly textured paper-like brushwork without showing actual paper\n- appealing rounded characters\n- expressive facial animation\n- subtle squash and stretch\n- natural secondary hair/clothing motion\n- charming hand-crafted environment\n\nThe final video should look like an animated film version of the storyboard artwork.\n\nNOT:\n- 3D CGI\n- photorealism\n- anime\n- flat vector art\n- cheap motion comic\n- slideshow\n- storyboard animation\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nPrimary palette:\n\nWarm golden sunlight\n+\nfresh leafy green\n+\nmint green\n+\nsunflower yellow\n+\ncoral orange\n+\nsoft sky blue\n\nLuma:\nwarm sunlight on face and hair.\n\nNimbus:\ncool pale-blue-white color providing subtle visual contrast.\n\nGreenhouse:\nbright, airy, naturally saturated.\n\nSunlight should create:\n- leaf-shaped shadows\n- soft rim lights\n- reflections through glass\n- golden dust particles\n- subtle bloom\n\nDuring Nimbus reveal:\nslightly cooler soft blue fill from the watering can interior.\n\nDuring raindrop:\ntiny prism/rainbow highlight.\n\nDo NOT darken the greenhouse dramatically.\n\n--------------------------------------------------\n\n[ANIMATION / MOTION RULES]\n\nLUMA:\n\nUse believable child motion.\n\nExamples:\n- small energetic steps\n- expressive hands\n- head tilts\n- natural blinking\n- loose auburn hair bouncing\n- hat shifting subtly as she runs\n- clothing following body movement\n\nAvoid:\n- rubber limbs\n- extreme cartoon deformation\n- unnatural running\n\nNIMBUS:\n\nMotion should feel buoyant.\n\nUse:\n- subtle floating\n- soft vertical bob\n- cloud edges gently expanding/contracting\n- tiny squash when sneezing\n- raindrop feet swaying\n- slight drifting inertia\n\nSPROUT:\n\nMovement should be very slow.\n\nComedy comes from contrast with Luma.\n\nUse:\n- eye-stalk stretching\n- tiny shell shifts\n- slow sliding\n- sudden eye reactions\n\nPLANTS:\n\nUse gentle natural motion:\n- leaves bending\n- flower head lifting\n- vines swaying\n- tiny response after water\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nMaintain one consistent Luma.\n\nMaintain one consistent Nimbus.\n\nMaintain one consistent Sprout.\n\nDo not duplicate characters.\n\nThe sunflower remains the same flower throughout.\n\nThe watering can remains spatially consistent.\n\nLuma’s seed pouch remains visible.\n\nLuma’s yellow hat remains on her head during Part 1.\n\nNimbus remains tiny throughout Part 1.\n\nSprout remains significantly smaller than Luma and Nimbus.\n\nThe greenhouse layout must remain coherent between shots.\n\nMorning lighting direction must remain consistent.\n\n--------------------------------------------------\n\n[STORY RESTRICTIONS — PART 1 ONLY]\n\nDO NOT show:\n\n- Super Cloud mode\n- flower crown on Nimbus\n- hand fan upgrade\n- wind chimes upgrade\n- Nimbus drinking water\n- Nimbus growing larger\n- multiple rain showers\n- greenhouse rain chaos\n- puddle rafting\n- overwhelmed Nimbus\n- gray Nimbus\n- magical greenhouse transformation\n- rainbow-drop charm\n- final ending\n\nThese belong to later parts.\n\nPart 1 ends ONLY when Luma realizes Nimbus could help all her flowers.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard borders.\nNo storyboard panels.\nNo captions.\nNo subtitles.\nNo timestamps.\nNo production notes.\nNo title card.\nNo watermark.\nNo brand logo.\n\nNo character identity drift.\nNo extra children.\nNo extra Nimbus clouds.\nNo duplicate Sprout.\nNo hairstyle changes.\nNo costume changes.\nNo missing hat.\nNo changing seed pouch.\nNo adult-looking Luma.\n\nNo dark storm.\nNo scary cloud.\nNo evil eyes.\nNo thunder.\nNo lightning.\nNo flooding.\nNo destructive rain.\nNo windstorm.\nNo danger.\nNo injury.\nNo broken plants.\nNo horror.\n\nNo random magical objects.\nNo visual teleportation.\nNo background redesign.\n\nNo excessive camera movement.\nNo frame interpolation artifacts.\nNo warped hands.\nNo extra fingers.\nNo malformed faces.\nNo melting character anatomy.\nNo inconsistent eye placement.\n\n--------------------------------------------------\n\n[FINAL 1.5-SECOND PRIORITY]\n\nThe ending must have crystal-clear storytelling.\n\nLuma looks across the entire greenhouse.\n\nShe realizes Nimbus might be able to water every plant.\n\nHer face fills with enormous excitement.\n\nShe turns to him and says:\n\n<d>[English] Nimbus... you can help ALL my flowers!</d>\n\nNimbus becomes uncertain:\n\n<d>[English] ...All?</d>\n\nSprout looks toward the huge greenhouse and reacts:\n\n<d>[English] Eep!</d>\n\nFinal emotional composition:\n\nLuma:\nexcited and confident.\n\nNimbus:\ntiny, shy, slightly worried.\n\nSprout:\ncomic dread.\n\nWarm beautiful greenhouse behind them.\n\nHold long enough for all three expressions to register.\n\nThen:\n\nCUT TO BLACK.\n\nThis cut should make the viewer immediately want to watch PART 2.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085986114800951530",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085986114800951530",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-08T07:07:00.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2085986114800951530.mp4",
    "posterUrl": "/posters/x/2085986114800951530.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "dialogue",
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt :  \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for “The Little Gardener and the Runaway Rain Cloud” as the PRIMARY visual, character, environment, composition, and narrative reference.\n\nGenerate a FINISHED animated sequence from the storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- panel numbers\n- panel captions\n- timestamps\n- production notes\n- character reference sheets\n- off-white storyboard paper\n- storyboard title/header\n- handwritten annotations\n\nThe uploaded storyboard controls and locks:\n\n1. Luma’s exact character design\n2. Nimbus’s exact cloud design\n3. Sprout’s exact snail design\n4. the rooftop greenhouse architecture\n5. the wilted sunflower design\n6. the old metal watering can\n7. the gardening props\n8. warm watercolor/painterly 2D visual style\n9. character scale relationships\n10. overall color palette\n11. Part 1 story progression\n\nAnimate these designs naturally while maintaining extremely high identity consistency.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nCreate exactly:\n\nDuration: 15 seconds\nAspect Ratio: 16:9 landscape\nFrame rate feeling: cinematic 24 fps\nGenre: magical garden comedy / heartwarming fantasy\nVisual medium: premium hand-painted 2D animated short\nAudience: family-friendly\nDialogue language: English\nNative synchronized dialogue + sound effects + music\n\nEmotional progression:\n\npeaceful garden morning\n→ concern\n→ determination\n→ mysterious sound\n→ magical discovery\n→ tiny comic surprise\n→ hopeful miracle\n→ BIG IDEA cliffhanger\n\nThe storytelling must be very easy to understand even without narration.\n\nIMPORTANT:\n\nAlthough the storyboard contains 15 individual panels, DO NOT turn the video into fifteen frantic one-second cuts.\n\nConsolidate those beats into approximately 7–8 fluid cinematic shots.\n\nEach shot should contain several connected storyboard actions so the animation feels like an actual polished animated short rather than a slideshow.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — LUMA]\n\nLUMA must match the uploaded storyboard exactly.\n\nCharacter:\napproximately 8-year-old little gardener.\n\nPhysical appearance:\n- warm fair skin\n- large expressive dark brown eyes\n- soft round childlike face\n- freckles across cheeks/nose if visible in reference\n- fluffy auburn / reddish-brown hair\n- hair gathered into two messy buns\n- several loose curly strands around face\n\nSignature clothing:\n- oversized warm yellow gardening hat\n- mint-green gardening overalls\n- coral-and-cream horizontally striped shirt\n- red gardening boots\n- tiny brown gardening/seed pouch attached at waist\n\nCharacter proportions:\n- small child stature\n- slightly oversized head\n- expressive hands\n- rounded appealing silhouette\n\nPersonality:\n- caring\n- cheerful\n- imaginative\n- curious\n- determined when a plant needs help\n- easily excited by magical discoveries\n\nPerformance should transition:\n\nhappy gardener\n→ worried plant caretaker\n→ determined problem solver\n→ curious investigator\n→ amazed child\n→ excited inventor\n\nCRITICAL:\n\nLuma must remain visually identical in every shot.\n\nDo not change:\n- face\n- age\n- hairstyle\n- hair color\n- hat\n- clothing\n- boots\n- proportions\n- eye color\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NIMBUS]\n\nNimbus must match the cloud discovered inside the watering can in the storyboard.\n\nNimbus is a tiny living rain cloud approximately pillow-sized.\n\nDesign:\n- soft fluffy blue-white cloud body\n- subtle watercolor edges\n- very faint pale blue shadows\n- large round expressive dark eyes\n- tiny eyebrows\n- small cute mouth\n- slight rosy blush\n- tiny dangling raindrop-shaped feet\n- soft semi-translucent cloud fluff\n- no arms unless naturally suggested by cloud shape\n\nPersonality:\n- shy\n- timid\n- gentle\n- easily embarrassed\n- harmless\n- sweet\n\nNimbus begins PART 1 hiding inside the watering can.\n\nHe must NOT appear in the greenhouse before the discovery sequence.\n\nHis first reveal should feel magical and surprising.\n\nNimbus should remain very small.\n\nDO NOT:\n- enlarge Nimbus\n- darken Nimbus\n- create storm powers\n- add flower crown yet\n- make him Super Cloud\n- generate multiple Nimbus clouds\n- create lightning\n\nThose belong to later story beats.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — SPROUT]\n\nSprout must match the storyboard exactly.\n\nDesign:\n- tiny green garden snail\n- soft rounded body\n- large tall expressive eye stalks\n- warm brown spiral shell\n- tiny yellow flower growing beside/on shell\n- same miniature scale throughout\n\nPersonality:\n- slow\n- observant\n- dramatic\n- suspicious\n- comedic reaction character\n\nSprout has no human speech unless specified.\n\nUse:\n- tiny squeaks\n- gasps\n- nervous chirps\n- eye-stalk reactions\n\nSprout should frequently understand the danger of Luma’s ideas before Luma does.\n\n--------------------------------------------------\n\n[PROP IDENTITY LOCK — WILTED SUNFLOWER]\n\nUse the exact sunflower from the storyboard.\n\nAt the start:\n- heavily drooping yellow petals\n- bowed flower head\n- wilted leaves\n- weak stem\n- planted in rectangular wooden planter\n- still alive\n\nAfter ONE drop from Nimbus:\nthe flower should perk up slightly.\n\nIMPORTANT:\n\nDo NOT magically transform it into a giant perfect sunflower yet.\n\nOnly show:\n- stem becoming slightly stronger\n- leaves lifting\n- flower head rising a little\n- color becoming slightly healthier\n\nThis is a small visual miracle.\n\n--------------------------------------------------\n\n[PROP IDENTITY LOCK — WATERING CAN]\n\nUse the same old watering can from the storyboard.\n\nAppearance:\n- weathered gray/blue metal\n- rounded main body\n- curved metal handle\n- long narrow spout\n- slightly worn surface\n- small painted flower decoration if visible\n- large enough for tiny Nimbus to curl inside\n\nIt is an ordinary gardening can.\n\nDo not transform it into a magical object.\n\nNimbus is simply hiding inside.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — ROOFTOP GREENHOUSE]\n\nMaintain exactly the warm rooftop greenhouse established by the storyboard.\n\nEnvironment:\n- glass-paneled greenhouse walls\n- glass roof\n- wooden structural framing\n- warm sunlight pouring through glass\n- hanging vines\n- multiple wooden planter boxes\n- terracotta flower pots\n- colorful flowers\n- herbs\n- leafy plants\n- gardening shelves\n- small ladder\n- watering hose\n- gardening tools\n- seed packets\n- old watering can\n- pots and wooden crates\n- distant soft city rooftop skyline through greenhouse glass\n\nVisual feeling:\nhandmade,\nlived-in,\ncozy,\nsafe,\nsunny,\nlush,\nmagical without obvious fantasy effects at the beginning.\n\nMaintain coherent geography throughout.\n\nThe sunflower planter, watering can area, shelves, and main walkway must not randomly move between shots.\n\n--------------------------------------------------\n\n[WORLD / MAGIC RULES]\n\nThe world begins mostly realistic.\n\nNimbus introduces the first visible magical element.\n\nMagic should be:\n- delicate\n- soft\n- believable within a children’s animated world\n- visually elegant\n- never explosive\n\nNimbus creates rain naturally.\n\nCause-and-effect must be clear:\n\nsunflower is dry\n→ Luma searches for water\n→ watering can sneezes\n→ Luma investigates\n→ Nimbus is discovered\n→ Nimbus nervously sneezes\n→ one raindrop falls\n→ sunflower responds\n→ Luma realizes Nimbus can help\n→ Luma gets a BIG IDEA\n\nNo unexplained magical events.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] (0.00s–2.00s)\nTHE PEACEFUL ROOFTOP GARDEN\n\nOpen with a beautiful wide establishing shot of the rooftop greenhouse.\n\nWarm morning sunlight enters through the glass ceiling.\n\nTiny dust particles float through the golden beams.\n\nLeaves move gently in a rooftop breeze.\n\nLuma happily walks between planter boxes carrying a small basket of seeds/gardening tools.\n\nSprout slowly crawls along the edge of a wooden planter beside her.\n\nUse subtle environmental life:\n- small leaves swaying\n- hanging vines moving\n- distant city haze\n- butterflies or tiny garden insects only if subtle\n\nCamera:\n\nbegin outside/near the glass roof\n→ smooth gentle push through the greenhouse\n→ settle into a medium-wide shot of Luma tending plants.\n\nLuma hums quietly.\n\nShe pats soil around a healthy flower.\n\nDialogue:\n\nLuma, cheerful:\n<d>[English] Good morning, little guys!</d>\n\nShe notices Sprout.\n\nLuma:\n<d>[English] Morning, Sprout.</d>\n\nSprout gives a tiny cheerful squeak.\n\nSound:\n- light morning breeze\n- gentle birds outside\n- leaves rustling\n- gardening basket movement\n- soft snail squeak\n- quiet greenhouse ambience\n\nMusic:\ndelicate whimsical opening using pizzicato strings, gentle marimba, and soft flute.\n\n--------------------------------------------------\n\n[SHOT 2] (2.00s–4.20s)\nTHE WILTING SUNFLOWER\n\nAs Luma continues through the garden, her smile slowly fades.\n\nHer gaze lands on the drooping sunflower.\n\nCut to a closer view:\n\nthe sunflower’s head hangs sadly downward.\n\nIts leaves droop.\n\nSprout crawls beside the planter and slowly raises both eye stalks toward the flower.\n\nOne eye stalk leans sideways with concern.\n\nLuma immediately kneels.\n\nShe gently lifts one wilted leaf between her fingers.\n\nCamera:\n\nmedium tracking with Luma\n→ rack focus from healthy flowers to wilted sunflower\n→ close-up flower\n→ intimate medium-close shot of Luma kneeling.\n\nDialogue:\n\nLuma:\n<d>[English] Oh no...</d>\n\nShe gently examines the plant.\n\nLuma:\n<d>[English] Sunny, what happened?</d>\n\nShe touches the dry soil.\n\nLuma:\n<d>[English] You’re thirsty.</d>\n\nSprout gives a worried tiny squeak.\n\nLuma’s worried expression becomes determined.\n\nLuma:\n<d>[English] Don’t worry. I’ll find water.</d>\n\nSound:\n- soft soil touch\n- leaf rustle\n- subtle worried musical note\n\nKeep this emotionally sincere but brief.\n\n--------------------------------------------------\n\n[SHOT 3] (4.20s–6.30s)\nTHE SEARCH\n\nLuma stands quickly and begins checking the greenhouse.\n\nShe looks at:\n- empty small watering cup\n- dry watering station\n- hose\n- gardening shelf\n\nShe trots along the walkway.\n\nSprout follows at an absurdly slow speed.\n\nUse a brief visual comedy beat:\n\nLuma rushes past frame.\n\nA moment later...\n\nSprout is still only a few centimeters farther along.\n\nLuma then notices the old metal watering can sitting among leafy pots.\n\nHer eyes brighten.\n\nCamera:\n\nside tracking shot following Luma\n→ small comedic static shot of slow Sprout\n→ quick pan toward watering can as Luma notices it.\n\nDialogue:\n\nLuma:\n<d>[English] Water... water...</d>\n\nShe spots the can.\n\nLuma:\n<d>[English] Aha!</d>\n\nSound:\n- light footsteps\n- tool clinks\n- leaves brushing clothes\n- tiny Sprout effort squeak\n\nMusic becomes slightly more playful.\n\n--------------------------------------------------\n\n[SHOT 4] (6.30s–8.30s)\nTHE WATERING CAN SNEEZES\n\nLuma approaches the old watering can.\n\nBefore she reaches it—\n\nsomething inside shifts.\n\nVery subtle.\n\nLuma freezes.\n\nThe watering can gives a tiny wobble.\n\nThen:\n\n<d>[English] Ah... ah...</d>\n\nLuma slowly leans closer.\n\nSprout’s eye stalks rise behind her.\n\nSuddenly from inside:\n\n<d>[English] ACHOO!</d>\n\nA tiny puff of cool mist shoots gently from the watering-can spout.\n\nThe can rocks slightly.\n\nLuma jumps backward only a few centimeters in surprise.\n\nSprout instantly retracts slightly into his shell.\n\nNo danger.\n\nCamera:\n\nslow push-in on watering can\n→ close-up of its dark opening\n→ Luma listening\n→ mist puff from spout\n→ reaction two-shot.\n\nDialogue:\n\nLuma:\n<d>[English] ...Did you just sneeze?</d>\n\nBeat.\n\nShe looks toward Sprout.\n\nLuma:\n<d>[English] Watering cans don’t sneeze.</d>\n\nSprout gives a tiny uncertain squeak.\n\nSound:\n- metal can creak\n- tiny movement from inside\n- adorable sneeze\n- mist “pfft”\n- light metallic wobble\n\nMusic:\nbrief curious plucked-string motif.\n\n--------------------------------------------------\n\n[SHOT 5] (8.30s–10.30s)\nTHE IMPOSSIBLE DISCOVERY\n\nLuma carefully crouches beside the watering can.\n\nShe grips its edge gently.\n\nSlowly...\n\nshe peers inside.\n\nCamera moves over Luma’s shoulder toward the dark interior.\n\nFor one beat we see only darkness.\n\nThen two tiny eyes blink.\n\nSoft pale-blue cloud fluff emerges.\n\nNimbus is curled inside the watering can.\n\nHe looks up at Luma.\n\nNimbus immediately becomes embarrassed and tries to shrink deeper into the can.\n\nLuma’s expression transforms from caution to pure wonder.\n\nSprout stretches upward from beside the can to see inside.\n\nCamera:\n\nover-shoulder POV\n→ gradual focus transition revealing Nimbus\n→ close-up Nimbus eyes\n→ reaction close-up Luma\n→ wider three-character composition.\n\nDialogue:\n\nLuma, whispering:\n<d>[English] Whoa...</d>\n\nNimbus nervously:\n<d>[English] ...Hi.</d>\n\nLuma:\n<d>[English] You’re a cloud.</d>\n\nNimbus gives a shy tiny nod.\n\nLuma, amazed:\n<d>[English] A tiny cloud!</d>\n\nSound:\n- subtle magical shimmer\n- tiny cloud puff\n- soft metal resonance from inside can\n- Nimbus’s faint watery ambience\n\nMusic:\nintroduce delicate glockenspiel and airy flute.\n\n--------------------------------------------------\n\n[SHOT 6] (10.30s–12.20s)\nTHE TINY RAIN DROP\n\nNimbus slowly floats just above the watering-can opening.\n\nHe is nervous about being watched.\n\nLuma leans closer with fascinated eyes.\n\nSprout watches from below.\n\nNimbus’s little face scrunches.\n\nHis cloud body briefly compresses.\n\nNimbus:\n\n<d>[English] Ah... ah...</d>\n\nLuma realizes another sneeze is coming.\n\nNimbus:\n\n<d>[English] Achoo!</d>\n\nA single sparkling raindrop pops from the underside of Nimbus.\n\nIMPORTANT:\n\nOnly ONE meaningful raindrop.\n\nCamera follows the droplet in a graceful macro tracking shot.\n\nThe droplet catches sunlight and briefly shows a subtle rainbow refraction.\n\nIt falls toward the sunflower planter.\n\nSound:\n- cute cloud sneeze\n- tiny watery “plink”\n- soft magical shimmer\n\nLuma:\n<d>[English] Wait...!</d>\n\n--------------------------------------------------\n\n[SHOT 7] (12.20s–13.50s)\nONE DROP IS ENOUGH\n\nExtreme close-up:\n\nthe raindrop lands in the dry soil near the sunflower stem.\n\nA soft ring of moisture spreads through the soil.\n\nNothing explosive.\n\nThe sunflower responds gently.\n\nIts stem firms.\n\nOne leaf slowly lifts.\n\nThe flower head rises a few centimeters.\n\nIts yellow petals regain a little warmth.\n\nLuma watches from the background.\n\nHer mouth falls open.\n\nThen she clasps her hands with delighted excitement.\n\nNimbus looks surprised by what he just did.\n\nSprout’s eyes become enormous.\n\nCamera:\n\nmacro raindrop impact\n→ tilt upward following sunflower\n→ rack focus to Luma’s astonished face.\n\nDialogue:\n\nLuma:\n<d>[English] You did that?</d>\n\nNimbus, shy:\n<d>[English] I... think so.</d>\n\nLuma:\n<d>[English] One drop!</d>\n\nSound:\n- small water impact\n- subtle plant movement\n- soft magical chime\n- leaf unfurl sound\n\nMusic:\nwarm hopeful melodic lift.\n\n--------------------------------------------------\n\n[SHOT 8] (13.50s–15.00s)\nTHE BIG IDEA — CLIFFHANGER\n\nLuma looks from the revived sunflower...\n\nto Nimbus...\n\nthen slowly around at the dozens of plants filling the greenhouse.\n\nHer eyes widen.\n\nA playful “idea” expression spreads across her face.\n\nShe turns toward Nimbus with a huge excited smile and raises one finger.\n\nNimbus immediately becomes slightly nervous.\n\nSprout looks at Luma.\n\nThen at Nimbus.\n\nThen at the enormous greenhouse.\n\nSprout’s eye stalks shoot upward in comic alarm.\n\nLuma excitedly gestures toward all the plants.\n\nDialogue:\n\nLuma:\n<d>[English] Nimbus...</d>\n\nTiny pause.\n\nLuma:\n<d>[English] You can help ALL my flowers!</d>\n\nNimbus:\n<d>[English] ...All?</d>\n\nSprout gives a loud alarmed:\n\n<d>[English] Eep!</d>\n\nHold briefly on the three expressions:\n\nLuma = thrilled\n\nNimbus = shy uncertainty\n\nSprout = “this is going to be a disaster”\n\nCamera:\n\nmedium shot behind Luma looking across greenhouse\n→ quick close-up on Luma’s excited realization\n→ reaction close-up Nimbus\n→ reaction close-up Sprout\n→ final wide three-character composition.\n\nEnd with a gentle comedic musical sting.\n\nCUT TO BLACK.\n\nDO NOT show Luma upgrading Nimbus.\n\nDO NOT begin Part 2.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\nLUMA VOICE:\n\nNatural young girl voice, approximately 8 years old.\n\nVoice qualities:\n- warm\n- bright\n- playful\n- emotionally expressive\n- curious\n- not overly cartoonish\n- clear natural English\n\nHer voice progression:\n\n0–4 sec:\ngentle and caring\n\n4–8 sec:\ndetermined then curious\n\n8–12 sec:\nsoft wonder\n\n12–15 sec:\nhigh-energy excitement\n\n--------------------------------------------------\n\nNIMBUS VOICE:\n\nVery soft youthful cloud voice.\n\nCharacteristics:\n- shy\n- slightly breathy\n- adorable\n- gentle\n- innocent\n- quiet\n- slightly higher pitched than Luma but easily understandable\n\nDo NOT make Nimbus sound robotic.\n\nHis sneezes should be small and cute.\n\n--------------------------------------------------\n\nSPROUT AUDIO:\n\nNo full normal dialogue.\n\nUse:\n- tiny squeaks\n- nervous chirps\n- tiny gasps\n- comic “eep”\n\nDo not give Sprout a human adult voice.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nMaintain a rich but soft greenhouse sound bed:\n\n- distant rooftop birds\n- very light city ambience far below\n- glass greenhouse creaks\n- subtle wind\n- leaves rustling\n- hanging plants moving\n- gardening tools\n- footsteps on stone/wood\n- soil movement\n- watering-can metal noises\n- small water droplets\n- Nimbus cloud puffs\n- Sprout squeaks\n\nNever overpower dialogue.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an original whimsical family-animation score.\n\nInstrumentation:\n- gentle pizzicato strings\n- marimba\n- soft acoustic plucks\n- airy flute\n- glockenspiel\n- very subtle warm pads\n\nMusical arc:\n\n0.00–2.00\npeaceful garden theme.\n\n2.00–4.20\nslightly concerned but still warm.\n\n4.20–6.30\nplayful search rhythm.\n\n6.30–8.30\ncurious comedic suspense.\n\n8.30–10.30\nmagical discovery.\n\n10.30–13.50\nwonder and hopeful warmth.\n\n13.50–15.00\nplayful “big idea” motif ending on light comedic suspense.\n\nNo epic music.\nNo scary music.\nNo dramatic disaster music.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nUse cinematic storybook camera work.\n\nOverall camera rhythm:\n\nBeginning:\nwide, calm, observational.\n\nProblem:\ncloser emotional coverage.\n\nMystery:\nslow push-ins.\n\nDiscovery:\nintimate close-ups.\n\nMagic:\nmacro insert + graceful tracking.\n\nFinal idea:\nreaction-driven comedy framing.\n\nPreferred techniques:\n- gentle dolly\n- subtle parallax\n- slow push-in\n- rack focus\n- over-the-shoulder composition\n- macro raindrop shot\n- reaction close-ups\n- wide environmental establishing shot\n\nAvoid:\n- handheld shake\n- crash zoom\n- extreme Dutch angles\n- fast action-camera motion\n- excessive cuts\n- 360-degree spins\n- random camera movement\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animated short.\n\nMatch the storyboard’s exact aesthetic:\n\n- delicate ink outlines\n- watercolor-like painted fills\n- warm children’s-book texture\n- cinematic lighting\n- painterly background detail\n- softly textured paper-like brushwork without showing actual paper\n- appealing rounded characters\n- expressive facial animation\n- subtle squash and stretch\n- natural secondary hair/clothing motion\n- charming hand-crafted environment\n\nThe final video should look like an animated film version of the storyboard artwork.\n\nNOT:\n- 3D CGI\n- photorealism\n- anime\n- flat vector art\n- cheap motion comic\n- slideshow\n- storyboard animation\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nPrimary palette:\n\nWarm golden sunlight\n+\nfresh leafy green\n+\nmint green\n+\nsunflower yellow\n+\ncoral orange\n+\nsoft sky blue\n\nLuma:\nwarm sunlight on face and hair.\n\nNimbus:\ncool pale-blue-white color providing subtle visual contrast.\n\nGreenhouse:\nbright, airy, naturally saturated.\n\nSunlight should create:\n- leaf-shaped shadows\n- soft rim lights\n- reflections through glass\n- golden dust particles\n- subtle bloom\n\nDuring Nimbus reveal:\nslightly cooler soft blue fill from the watering can interior.\n\nDuring raindrop:\ntiny prism/rainbow highlight.\n\nDo NOT darken the greenhouse dramatically.\n\n--------------------------------------------------\n\n[ANIMATION / MOTION RULES]\n\nLUMA:\n\nUse believable child motion.\n\nExamples:\n- small energetic steps\n- expressive hands\n- head tilts\n- natural blinking\n- loose auburn hair bouncing\n- hat shifting subtly as she runs\n- clothing following body movement\n\nAvoid:\n- rubber limbs\n- extreme cartoon deformation\n- unnatural running\n\nNIMBUS:\n\nMotion should feel buoyant.\n\nUse:\n- subtle floating\n- soft vertical bob\n- cloud edges gently expanding/contracting\n- tiny squash when sneezing\n- raindrop feet swaying\n- slight drifting inertia\n\nSPROUT:\n\nMovement should be very slow.\n\nComedy comes from contrast with Luma.\n\nUse:\n- eye-stalk stretching\n- tiny shell shifts\n- slow sliding\n- sudden eye reactions\n\nPLANTS:\n\nUse gentle natural motion:\n- leaves bending\n- flower head lifting\n- vines swaying\n- tiny response after water\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nMaintain one consistent Luma.\n\nMaintain one consistent Nimbus.\n\nMaintain one consistent Sprout.\n\nDo not duplicate characters.\n\nThe sunflower remains the same flower throughout.\n\nThe watering can remains spatially consistent.\n\nLuma’s seed pouch remains visible.\n\nLuma’s yellow hat remains on her head during Part 1.\n\nNimbus remains tiny throughout Part 1.\n\nSprout remains significantly smaller than Luma and Nimbus.\n\nThe greenhouse layout must remain coherent between shots.\n\nMorning lighting direction must remain consistent.\n\n--------------------------------------------------\n\n[STORY RESTRICTIONS — PART 1 ONLY]\n\nDO NOT show:\n\n- Super Cloud mode\n- flower crown on Nimbus\n- hand fan upgrade\n- wind chimes upgrade\n- Nimbus drinking water\n- Nimbus growing larger\n- multiple rain showers\n- greenhouse rain chaos\n- puddle rafting\n- overwhelmed Nimbus\n- gray Nimbus\n- magical greenhouse transformation\n- rainbow-drop charm\n- final ending\n\nThese belong to later parts.\n\nPart 1 ends ONLY when Luma realizes Nimbus could help all her flowers.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard borders.\nNo storyboard panels.\nNo captions.\nNo subtitles.\nNo timestamps.\nNo production notes.\nNo title card.\nNo watermark.\nNo brand logo.\n\nNo character identity drift.\nNo extra children.\nNo extra Nimbus clouds.\nNo duplicate Sprout.\nNo hairstyle changes.\nNo costume changes.\nNo missing hat.\nNo changing seed pouch.\nNo adult-looking Luma.\n\nNo dark storm.\nNo scary cloud.\nNo evil eyes.\nNo thunder.\nNo lightning.\nNo flooding.\nNo destructive rain.\nNo windstorm.\nNo danger.\nNo injury.\nNo broken plants.\nNo horror.\n\nNo random magical objects.\nNo visual teleportation.\nNo background redesign.\n\nNo excessive camera movement.\nNo frame interpolation artifacts.\nNo warped hands.\nNo extra fingers.\nNo malformed faces.\nNo melting character anatomy.\nNo inconsistent eye placement.\n\n--------------------------------------------------\n\n[FINAL 1.5-SECOND PRIORITY]\n\nThe ending must have crystal-clear storytelling.\n\nLuma looks across the entire greenhouse.\n\nShe realizes Nimbus might be able to water every plant.\n\nHer face fills with enormous excitement.\n\nShe turns to him and says:\n\n<d>[English] Nimbus... you can help ALL my flowers!</d>\n\nNimbus becomes uncertain:\n\n<d>[English] ...All?</d>\n\nSprout looks toward the huge greenhouse and reacts:\n\n<d>[English] Eep!</d>\n\nFinal emotional composition:\n\nLuma:\nexcited and confident.\n\nNimbus:\ntiny, shy, slightly worried.\n\nSprout:\ncomic dread.\n\nWarm beautiful greenhouse behind them.\n\nHold long enough for all three expressions to register.\n\nThen:\n\nCUT TO BLACK.\n\nThis cut should make the viewer immediately want to watch PART 2.",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 0,
      "views": 2093,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status publishes the complete Part 1 storyboard prompt with character locks, greenhouse setting, timed discovery, dialogue, sound, and cliffhanger."
  },
  {
    "id": "x-2085983430219895004",
    "title": "泳池边的仙女棒身份互换",
    "titleEn": "A Fairy-Wand Role Reversal by the Pool",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "泳池边争吵的年轻女子与清洁妇被仙女棒交换年龄与处境，最后以夸张真人秀反应收尾。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A fairy wand reverses the ages and fortunes of a young woman and a cleaner arguing by a pool, ending on an exaggerated reality-show reaction. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "美國喜劇reality show: 豪宅的泳池邊，20歲金髮藍眼的紅色比基尼美女與60歲灰髮憔悴的清潔婦正在激烈地爭吵，美女說\"醜死了，你這死老太婆!\"，清潔婦也不甘示弱說\"整天只會換男人，不正經!\"。這時畫誇張眼影妖后造型的drag queen飛過，便將手中的仙女棒往美女與清潔婦的頭上各點一下，結果美女迅速老化變成70歲皮膚鬆垮的老女人，清潔婦則迅速變成年輕的20歲美麗模樣。變年輕的清潔婦自信地說\"啊，我現在感覺好多了!\"就脫掉制服，露出夏季連身華麗黑色窄裙與高跟鞋，然後自信地離去，留下崩潰且一臉錯愕的老女人無助地看著鏡頭哭著吶喊:\"不!!!!!!\"",
    "promptSourceUrl": "https://x.com/drjoetw/status/2085983430219895004",
    "sourceUrl": "https://x.com/drjoetw/status/2085983430219895004",
    "sourceLabel": "X 原帖 · @drjoetw",
    "author": "The Anxious Mind",
    "publishedAt": "2026-08-08T06:56:20.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2085983430219895004.mp4",
    "posterUrl": "/posters/x/2085983430219895004.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Need a fairy godmother like this? 🧚‍♀️✨\n#MiniMaxH3 is surprisingly good at creating soap opera-style drama! 👀🎬 (powered on @Hailuo_AI )\nprompt below 😍👇\n\n美國喜劇reality show: 豪宅的泳池邊，20歲金髮藍眼的紅色比基尼美女與60歲灰髮憔悴的清潔婦正在激烈地爭吵，美女說\"醜死了，你這死老太婆!\"，清潔婦也不甘示弱說\"整天只會換男人，不正經!\"。這時畫誇張眼影妖后造型的drag queen飛過，便將手中的仙女棒往美女與清潔婦的頭上各點一下，結果美女迅速老化變成70歲皮膚鬆垮的老女人，清潔婦則迅速變成年輕的20歲美麗模樣。變年輕的清潔婦自信地說\"啊，我現在感覺好多了!\"就脫掉制服，露出夏季連身華麗黑色窄裙與高跟鞋，然後自信地離去，留下崩潰且一臉錯愕的老女人無助地看著鏡頭哭著吶喊:\"不!!!!!!\"\n\n#minimax #AIDrama #AIVideo #AIFilm #AIMovie #microdrama #parody",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 3251,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status explicitly labels MiniMax H3 and publishes the complete poolside parody prompt, dialogue, transformation, costume reveal, and final reaction."
  },
  {
    "id": "x-2085973447562162229",
    "title": "山溪营地的搭帐篷时刻",
    "titleEn": "Setting Up Camp beside a Mountain Stream",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "年轻女子在山溪旁敲入地钉、处理卡住的帐杆，并在帐篷立起后用韩语分享完成的喜悦。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A young woman drives tent stakes and wrestles with a stubborn pole beside a mountain stream before celebrating the finished setup in Korean. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Main Subject : Young Korean woman, early 20s, oversized flannel over a tank top, cargo shorts, hair in a practical braid, focused cheerful energy.\nLocation: Quiet forest campsite near a mountain stream, late afternoon. A half-set-up tent, camping gear scattered on a mat, tall pine trees, dappled sunlight. No other campers nearby.\nVisual Style:\nUltra-realistic documentary realism, adventurous candid feeling, warm natural forest light.\nCamera Style:\nEarly 2000s DV camcorder, handheld with natural outdoor shake, autofocus hunting in dappled light, soft grain, faded tones. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She hammers a tent stake into the ground, then checks the tension of the rope.\n00:03–00:06 → She struggles with a stubborn pole, then says to camera \"이거 왜 이렇게 안 들어가\" (\"Why won't this go in\").\n00:06–00:09 → She finally gets it to click into place, laughing in relief.\n00:09–00:12 → She steps back to admire the standing tent, hands on her hips.\n00:12–00:15 → She looks at camera saying \"다 됐다, 오늘 여기서 자자!\" (\"Done, let's sleep here tonight!\"), smiling as it fades.\nAudio:\nWind through trees, distant stream flowing, birds, fabric rustling. Her dialogue as noted above. No music.\nGoal: An adventurous, cheerful camping setup moment grounded, warm, authentic.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2085973447562162229",
    "sourceUrl": "https://x.com/Strength04_X/status/2085973447562162229",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-08T06:16:40.000Z",
    "addedAt": "2026-08-21T01:14:08.622Z",
    "mediaUrl": "/media/x-2085973447562162229.mp4",
    "posterUrl": "/posters/x/2085973447562162229.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on Hailuo_AI \n\nPrompt - Main Subject : Young Korean woman, early 20s, oversized flannel over a tank top, cargo shorts, hair in a practical braid, focused cheerful energy.\nLocation: Quiet forest campsite near a mountain stream, late afternoon. A half-set-up tent, camping gear scattered on a mat, tall pine trees, dappled sunlight. No other campers nearby.\nVisual Style:\nUltra-realistic documentary realism, adventurous candid feeling, warm natural forest light.\nCamera Style:\nEarly 2000s DV camcorder, handheld with natural outdoor shake, autofocus hunting in dappled light, soft grain, faded tones. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She hammers a tent stake into the ground, then checks the tension of the rope.\n00:03–00:06 → She struggles with a stubborn pole, then says to camera \"이거 왜 이렇게 안 들어가\" (\"Why won't this go in\").\n00:06–00:09 → She finally gets it to click into place, laughing in relief.\n00:09–00:12 → She steps back to admire the standing tent, hands on her hips.\n00:12–00:15 → She looks at camera saying \"다 됐다, 오늘 여기서 자자!\" (\"Done, let's sleep here tonight!\"), smiling as it fades.\nAudio:\nWind through trees, distant stream flowing, birds, fabric rustling. Her dialogue as noted above. No music.\nGoal: An adventurous, cheerful camping setup moment grounded, warm, authentic.",
    "engagement": {
      "replies": 11,
      "reposts": 3,
      "likes": 29,
      "views": 1920,
      "snapshotAt": "2026-08-21T01:14:08.622Z"
    },
    "approvedAt": "2026-08-21T01:14:08.622Z",
    "editorialBasis": "The original X status publishes the complete H3 prompt with subject, campsite, DV-camera look, timed tent setup, Korean dialogue, and natural ambience."
  },
  {
    "id": "x-2085884187626614835",
    "title": "亡灵法师唤醒骷髅军团",
    "titleEn": "The Necromancer Raises an Army of Skeletons",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "亡灵法师唤醒骷髅军团的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 5 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the necromancer raises an army of skeletons on X. Native-video output: 5s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The necromancer walks along, raising skeletons with magic; they smoothly rise to their feet—a grim scene, bathed in a dark green glow.",
    "promptSourceUrl": "https://x.com/UnrealRafael/status/2085884187626614835",
    "sourceUrl": "https://x.com/UnrealRafael/status/2085884187626614835",
    "sourceLabel": "X 原帖 · @UnrealRafael",
    "author": "Рафаэль Анриал ∞",
    "publishedAt": "2026-08-08T00:21:58.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085884187626614835.mp4",
    "posterUrl": "/posters/x/2085884187626614835.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Absolutely insane🤯Text in the video rtx 5090   Prompt: The necromancer walks along, raising skeletons with magic; they smoothly rise to their feet—a grim scene, bathed in a dark green glow.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 161,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085835546572800047",
    "title": "镜头穿越街机屏幕",
    "titleEn": "The Camera Breaks into an Arcade Game",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头穿越街机屏幕的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the camera breaks into an arcade game on X. Native-video output: 14s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A handheld found-footage shot from behind a person playing an authentic Street Fighter II arcade cabinet in a dim 1990s arcade. Ryu and Ken begin a normal match as the announcer shouts, “Fight!” The camera suddenly rushes toward the curved CRT screen in one continuous motion; reflections and scanlines expand into giant glowing red, green, and blue phosphor dots, then the camera passes through them and the image resolves into a clean full-screen 4:3 Street Fighter II game view with crisp 16-bit pixel art and subtle CRT scanlines. Chun-Li unexpectedly jumps into the match, followed by Blanka dropping from above, Dhalsim teleporting into the center, Zangief charging in, and several more fighters entering from both sides. The stage widens as the match becomes a chaotic but readable eight-player free-for-all: Sonic Booms cross the screen, Blanka’s electricity shocks multiple fighters, Zangief spins through a crowd, characters collide in midair, and health bars rapidly drain. Everyone unleashes a special move at once, filling the screen with pixel explosions and impact flashes. When the smoke clears, every fighter is knocked out except Dhalsim, calmly floating above the pile. The announcer declares, “Dhalsim wins. Perfect!” Authentic arcade sound effects, energetic chiptune music, no modern 3D graphics, no camera shake after entering the game.",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2085835546572800047",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2085835546572800047",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-07T21:08:41.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085835546572800047.mp4",
    "posterUrl": "/posters/x/2085835546572800047.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cg-3d",
      "ugc"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tried an ambitious camera transition trick with Minimax H3, zooming in from a found footage video to a full video game view. Didn't think it would do it this well, impressive.\n\nPrompt: A handheld found-footage shot from behind a person playing an authentic Street Fighter II arcade cabinet in a dim 1990s arcade. Ryu and Ken begin a normal match as the announcer shouts, “Fight!” The camera suddenly rushes toward the curved CRT screen in one continuous motion; reflections and scanlines expand into giant glowing red, green, and blue phosphor dots, then the camera passes through them and the image resolves into a clean full-screen 4:3 Street Fighter II game view with crisp 16-bit pixel art and subtle CRT scanlines. Chun-Li unexpectedly jumps into the match, followed by Blanka dropping from above, Dhalsim teleporting into the center, Zangief charging in, and several more fighters entering from both sides. The stage widens as the match becomes a chaotic but readable eight-player free-for-all: Sonic Booms cross the screen, Blanka’s electricity shocks multiple fighters, Zangief spins through a crowd, characters collide in midair, and health bars rapidly drain. Everyone unleashes a special move at once, filling the screen with pixel explosions and impact flashes. When the smoke clears, every fighter is knocked out except Dhalsim, calmly floating above the pile. The announcer declares, “Dhalsim wins. Perfect!” Authentic arcade sound effects, energetic chiptune music, no modern 3D graphics, no camera shake after entering the game.",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 58,
      "views": 4834,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085793101600428386",
    "title": "马耳他魔法群岛舞蹈",
    "titleEn": "Dancing Across the Magical Islands of Malta",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "马耳他魔法群岛舞蹈的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for dancing across the magical islands of malta on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "{\n  \"format\": { \"duration\": \"15s\", \"bpm\": 92, \"total_shots\": 10, \"sync_type\": \"Dance to Playback Song — No Lip-Sync\" },\n  \"style_reference\": \"Warm Mediterranean daylight, 35mm, fine organic grain, gentle bloom. Honey-gold limestone against deep cobalt sea and white sky. Saturated embroidery popping against neutral stone. Shallow depth on close work, crisp wides, natural sun flare. Never cold or desaturated.\",\n  \"subject\": {\n    \"description\": \"A young Mexican woman, mid-twenties, long dark wavy hair loose and moving, warm brown skin, serene expressive face, small gold earrings. Traditional Chiapaneco dress — deep-toned fabric densely embroidered with large multicolored floral bursts in magenta, gold, turquoise and crimson, wide flowing skirt, sheer embroidered overlay that catches air. Bare feet.\",\n    \"consistency_note\": \"The dress is the co-star — keep embroidery pattern, colors and skirt volume consistent. Fabric behavior matters as much as her body. Keep her face large in close work to avoid drift.\"\n  },\n  \"environment\": {\n    \"setting\": \"Malta in bright daylight — honey limestone architecture, a sunlit stone terrace above the sea, narrow old-town steps.\",\n    \"background\": \"Limestone walls and arches, traditional Maltese wooden balconies, worn steps, bougainvillea, the deep blue Mediterranean to the horizon.\",\n    \"lighting\": \"Hard Mediterranean afternoon sun, crisp shadows across stone, warm limestone bounce filling her face, occasional flare, glittering sea behind.\"\n  },\n  \"mood\": \"graceful, elegant, joyful, free, radiant, flowing\",\n  \"music\": {\n    \"genre\": \"Warm Latin folk-pop ballad, 92 BPM, acoustic and organic.\",\n    \"instrumentation\": \"Nylon-string guitar arpeggios, soft cajón and hand percussion, warm upright bass, distant accordion, light strings on the big moments, rich female vocal on top.\",\n    \"arrangement\": \"Shots 1-2: guitar and voice alone, sparse. Shots 3-5: percussion enters, groove opens. Shot 6: strips to a single held guitar note. Shot 7: full arrangement blooms with strings. Shots 8-9: percussion punches back, then softens. Shot 10: resolves warm on guitar and a sustained vocal note.\",\n    \"language\": \"Spanish\",\n    \"lyrics\": \"Que baile el sol sobre la piedra dorada / que el viento me lleve donde nadie me llama / Mi falda florece, la mar me responde / Aquí soy de nadie, aquí soy del aire / Vuelo, vuelo, vuelo — y nunca me caigo\",\n    \"mix\": \"Warm, natural, acoustic-forward. Vocal present but not overpowering. Round percussion. Sun-soaked.\"\n  },\n  \"style\": \"Ultra-Realistic\",\n  \"logic_rule\": \"The song plays audibly IN the clip while she dances to it. She does NOT sing and does NOT lip-sync — mouth relaxed and natural, expression serene and absorbed, occasionally a soft smile. THE BODY LEADS: each shot built around one quality of motion — stretch, spin, suspended pause, weightlessness, impact, inner rhythm. Movement is GRACEFUL and ELEGANT: long extended lines, controlled turns, soft shaped arms, fluid transitions, classical poise with folkloric footwork — never frantic. Vary camera to serve movement: low angles for elevation, tracking for travel, macro for fabric and feet, wide for full body against architecture. Mandatory SPEED CHANGE: extreme slow motion at the leap's apex so skirt bloom, hair suspension and embroidery are fully legible, then snap back on landing. Fabric physics essential — the skirt must swing, flare, ripple and settle with real weight. Effects: dust off stone, hair whipping, shadow raking walls, sun flare. ENDING: hold on a still breathing frame, then slow fade.\",\n  \"storyboard\": [\n    { \"shot\": 1, \"camera\": \"Low wide, sun behind her, limestone wall behind.\", \"action\": \"THE STRETCH — she rises onto the balls of her bare feet, extending both arms overhead in a long elegant line, spine lengthening, sleeve falling back. Eyes closed, serene.\", \"sfx\": \"Wind, gulls, fabric settling; guitar and voice alone.\" },\n    { \"shot\": 2, \"camera\": \"Macro on her bare feet on warm stone, shallow focus.\", \"action\": \"THE RHYTHM UNDER THE SKIN — heel and toe begin a soft precise pattern, dust lifting with each strike, ankles articulate and controlled.\", \"sfx\": \"Delicate foot taps, dust scuff; guitar continuing.\" },\n    { \"shot\": 3, \"camera\": \"Tracking alongside her down a narrow sunlit alley.\", \"action\": \"She travels in flowing rhythmic steps, skirt swinging gracefully side to side, one hand trailing the warm stone wall.\", \"sfx\": \"Footfalls, fabric swish, alley echo; percussion entering.\" },\n    { \"shot\": 4, \"camera\": \"Medium, camera orbiting with her turn.\", \"action\": \"THE SPIN — one controlled elegant rotation, the embroidered skirt opening into a wide circle of color, hair fanning, arms soft and shaped.\", \"sfx\": \"Fabric whoosh, foot pivot; groove opening.\" },\n    { \"shot\": 5, \"camera\": \"Close-up on her face, sun raking across it.\", \"action\": \"Eyes closed, head tilted back, absorbed with the faintest smile — lost in the music. Mouth relaxed and closed, NOT singing.\", \"sfx\": \"Breath, wind; vocal carrying over.\" },\n    { \"shot\": 6, \"camera\": \"Wide low angle on the terrace, sea behind.\", \"action\": \"THE PAUSE — everything stills. She sinks into a deep controlled coil, weight low, skirt pooling, held one suspended beat.\", \"sfx\": \"Music strips to one held guitar note; a deep breath.\" },\n    { \"shot\": 7, \"camera\": \"EXTREME SLOW MOTION, low angle, sky and sea behind.\", \"action\": \"THE WEIGHTLESSNESS — she launches upward, body arced in an extended line, arms open, skirt blooming outward like a flower, hair suspended, every stitch legible against the blue.\", \"sfx\": \"Stretched fabric bloom, deep swell; full arrangement with strings.\" },\n    { \"shot\": 8, \"camera\": \"Snap to full speed. Low, close on her landing.\", \"action\": \"THE LANDING — she meets the stone softly, knees absorbing, skirt collapsing around her legs, dust puffing, hair falling forward.\", \"sfx\": \"Foot impact, fabric slap, exhale; percussion punching back.\" },\n    { \"shot\": 9, \"camera\": \"Macro on the embroidery settling, rack up to her chest.\", \"action\": \"The floral stitching ripples and stills; her chest rises and falls, a bead of sweat at her collarbone. The body after force.\", \"sfx\": \"Breathing, fabric settling, sea; arrangement softening.\" },\n    { \"shot\": 10, \"camera\": \"Slow pull-back to a wide — small against limestone and sea. Held.\", \"action\": \"She lifts her head, sweeps her hair back and settles into a final open-armed shape against the Mediterranean light, breathing, a slow warm smile. Hold, then slow fade.\", \"sfx\": \"Wind, gulls, sea; guitar and sustained vocal resolving.\" }\n  ]\n}",
    "promptSourceUrl": "https://x.com/abulu8/status/2085793106365132879",
    "sourceUrl": "https://x.com/abulu8/status/2085793101600428386",
    "sourceLabel": "X 原帖 · @abulu8",
    "author": "Syed Abuthahir ∞",
    "publishedAt": "2026-08-07T18:20:02.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085793101600428386.mp4",
    "posterUrl": "/posters/x/2085793101600428386.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Gorgeous Latina beauty dancing through the magical islands of Malta. 🇲🇹✨\n\nCaptured the sun, colors, and energy of Malta in one dreamy sequence. 💃🌊\n\n#MiniMaxH3 @Hailuo_AI #AIVideo #AIArt\n\nPrompt in comments 💥 [T2V]",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 16,
      "views": 552,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085786048521757074",
    "title": "后院药草园的宁静时光",
    "titleEn": "A Quiet Morning in the Backyard Herb Garden",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "后院药草园的宁静时光的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a quiet morning in the backyard herb garden on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Main Subject:\nYoung Korean woman, early 20s, loose linen shirt over shorts, hair tied back with a cloth, dirt-smudged fingers, gentle focused expression.\nLocation:\nSmall backyard herb garden beside the house, late morning. Wooden planter boxes, scattered gardening tools, soft natural sunlight, a low fence. No modern structures.\nVisual Style:\nUltra-realistic documentary realism, candid hands-on feeling, bright natural daylight.\nCamera Style:\nEarly 2000s DV camcorder, handheld close framing, autofocus adjusting between her hands and the plants, warm faded tones, soft grain. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She kneels by the planter box, then snips a few basil leaves carefully.\n00:03–00:06 → She smells the leaves, then says to camera \"향 진짜 좋다\" (\"The smell is really nice\"), smiling.\n00:06–00:09 → She moves to another box, then pulls a small weed from the soil.\n00:09–00:12 → She wipes her hands on her shirt, then gathers the herbs into a small basket.\n00:12–00:15 → She looks at camera saying \"오늘 저녁에 쓸 거야\" (\"I'll use these for dinner tonight\"), standing up as it fades.\nAudio:\nLeaves rustling, soil shifting, faint birds, distant wind. Her dialogue as noted above. No music.\nGoal: A small, grounded gardening moment warm, patient, authentic.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2085786048521757074",
    "sourceUrl": "https://x.com/Strength04_X/status/2085786048521757074",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-07T17:52:00.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085786048521757074.mp4",
    "posterUrl": "/posters/x/2085786048521757074.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on Hailuo_AI \n\nPrompt - Main Subject:\nYoung Korean woman, early 20s, loose linen shirt over shorts, hair tied back with a cloth, dirt-smudged fingers, gentle focused expression.\nLocation:\nSmall backyard herb garden beside the house, late morning. Wooden planter boxes, scattered gardening tools, soft natural sunlight, a low fence. No modern structures.\nVisual Style:\nUltra-realistic documentary realism, candid hands-on feeling, bright natural daylight.\nCamera Style:\nEarly 2000s DV camcorder, handheld close framing, autofocus adjusting between her hands and the plants, warm faded tones, soft grain. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She kneels by the planter box, then snips a few basil leaves carefully.\n00:03–00:06 → She smells the leaves, then says to camera \"향 진짜 좋다\" (\"The smell is really nice\"), smiling.\n00:06–00:09 → She moves to another box, then pulls a small weed from the soil.\n00:09–00:12 → She wipes her hands on her shirt, then gathers the herbs into a small basket.\n00:12–00:15 → She looks at camera saying \"오늘 저녁에 쓸 거야\" (\"I'll use these for dinner tonight\"), standing up as it fades.\nAudio:\nLeaves rustling, soil shifting, faint birds, distant wind. Her dialogue as noted above. No music.\nGoal: A small, grounded gardening moment warm, patient, authentic.",
    "engagement": {
      "replies": 22,
      "reposts": 5,
      "likes": 73,
      "views": 2912,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085784390903173149",
    "title": "唤醒宇宙奇观",
    "titleEn": "Wake the Wonder",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "唤醒宇宙奇观的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 320×560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for wake the wonder on X. Native-video output: 15s · 320×560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "{\n  \"title\": \"Wake the Wonder\",\n  \"duration\": \"15 seconds\",\n  \"aspect_ratio\": \"9:16\",\n  \"style\": \"retro-futurist 1960s editorial illustration blended with hand-painted gouache animation, geometric space-age design, limited mustard, teal, coral and midnight-blue palette, grainy print texture, imaginative cosmic transitions, sophisticated picture-book composition, polished non-photorealistic motion graphics\",\n  \"prompt\": \"Create a fast, imaginative 15-second vertical animated advertisement about escaping routine through curiosity. Open immediately inside a silent library located on a tiny moon. A middle-aged lunar librarian wearing a bubble helmet and neat retro uniform sits behind a circular desk, mechanically stamping identical grey books. Endless shelves curve around the moon's interior while dusty speech bubbles reading SAME OLD, ALREADY KNOW, WHY BOTHER and NOTHING NEW drift through the air. The librarian's movements are repetitive and perfectly synchronized with a ticking clock. At second 3.0, a small coral-colored book falls from the highest shelf and lands open on the desk. Instead of printed words, a miniature galaxy spins between its pages. The librarian's eyes widen. They slam the date stamp onto the glowing page, sending a teal shockwave through the library. Shelves bend into planetary rings, book pages become flocks of geometric spacecraft and the dusty speech bubbles burst into constellations. The librarian leaps onto a giant flying page and surfs through the transforming archive while the camera spirals beside them. Every book they pass opens into a different illustrated world: an underwater city, a jungle growing on an asteroid, a mechanical whale and a floating orchestra. At the climax, the librarian flies toward a locked circular door labelled NOTHING NEW. The glowing book opens in front of it and transforms the door into a vast cosmic iris. They burst through into an infinite colorful universe made of pages, stars and painted shapes. Finish with the librarian floating confidently above the small moon while bold retro typography appears: OPEN ONE DOOR. EXPAND YOUR WORLD. Maintain rapid wonder, clear visual storytelling, elegant graphic design and continuous transformation.\",\n  \"timeline\": [\n    {\n      \"time\": \"0.0-3.0 seconds\",\n      \"visual\": \"Immediate hook: a circular moon library in muted colors. The librarian mechanically stamps identical books while dusty phrases reading SAME OLD, ALREADY KNOW, WHY BOTHER and NOTHING NEW float around the desk. Fast orbiting push-in.\",\n      \"text\": \"WHEN DID EVERYTHING FEEL FAMILIAR?\"\n    },\n    {\n      \"time\": \"3.0-6.5 seconds\",\n      \"visual\": \"A coral-colored book falls open, revealing a spinning miniature galaxy. The librarian stamps the page, releasing a teal shockwave that bends shelves into planetary rings and lights up their expression.\",\n      \"text\": \"\"\n    },\n    {\n      \"time\": \"6.5-11.5 seconds\",\n      \"visual\": \"The librarian rides a giant flying page through the transforming archive. Books burst open into surreal worlds while grey speech bubbles collapse into bright constellations.\",\n      \"text\": \"\"\n    },\n    {\n      \"time\": \"11.5-15.0 seconds\",\n      \"visual\": \"The librarian races through a locked door labelled NOTHING NEW as it becomes a huge cosmic iris. Wide pullback reveals them floating over the moon inside a limitless illustrated universe.\",\n      \"text\": \"OPEN ONE DOOR\\nEXPAND YOUR WORLD\"\n    }\n  ],\n  \"voiceover\": {\n    \"enabled\": true,\n    \"delivery\": \"bright, intelligent and inviting, accelerating from restrained curiosity into genuine excitement\",\n    \"script\": \"Seen it all before? Look closer. One question opens a door. Wonder does the rest.\"\n  },\n  \"audio\": {\n    \"music\": \"up-tempo retro space-age jazz blended with modern electronic percussion, launching into full rhythm at second 3.0\",\n    \"sound_design\": \"clock ticks, rubber stamp impacts, page flutters, cosmic shimmer, orbital whooshes, opening-book chimes and a bright final orchestral hit\",\n    \"mix\": \"Narration stays crisp while page movements and cosmic effects bounce rhythmically around the beat.\"\n  },\n  \"camera\": {\n    \"movement\": \"circular push around the stamping desk, extreme macro on the miniature galaxy, spiralling tracking shot beside the flying page and dramatic orbital pullback around the moon\",\n    \"editing\": \"rapid graphic match cuts using circular books, planets, helmet reflections and cosmic doorways\"\n  },\n  \"negative_prompt\": [\n    \"photorealistic outer space\",\n    \"modern realistic spacecraft\",\n    \"slow educational documentary tone\",\n    \"dark existential mood\",\n    \"generic children's cartoon\",\n    \"overcrowded cosmic imagery\",\n    \"unreadable books or typography\",\n    \"stiff character movement\"\n  ]\n}",
    "promptSourceUrl": "https://x.com/airina_xyz/status/2085784395273642419",
    "sourceUrl": "https://x.com/airina_xyz/status/2085784390903173149",
    "sourceLabel": "X 原帖 · @airina_xyz",
    "author": "AIrina (Virtual Media Net)",
    "publishedAt": "2026-08-07T17:45:25.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085784390903173149.mp4",
    "posterUrl": "/posters/x/2085784390903173149.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "320×560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I love how these @Hailuo_AI @MiniMax_AI H3 animations turn out.\n\nMinimax H3 in @pixverse #PixverseCPP\n\nPrompt in comments 👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 217,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085759071047565341",
    "title": "第一人称幻想战斗对决",
    "titleEn": "A First-Person Fantasy Battle Face-Off",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "第一人称幻想战斗对决的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2160×2700。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a first-person fantasy battle face-off on X. Native-video output: 15s · 2160×2700. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Strict first-person POV, male hero, face never visible. Visible only the hero’s hands, fingers, and forearms when needed for casting, dodging balance, and the final magnetic strike. Broad male hands, pronounced knuckles, short nails, pale or cool-toned skin, dark gray or black sleeves, thin geometric magic lines glowing cold white with a faint blue tint across the skin and wrists. The hands move with force, precision, and short controlled gestures, never soft or decorative. One continuous shot, no cuts, vertical cinematic dark fantasy, realistic body-driven camera motion, professional action cinematography with physical momentum and readable combat beats.\n\nIn a vast abandoned stone quarry under a gray overcast sky, high rock walls, two dark mine tunnel arches in the background, old metal rails cutting through wet red-brown mud, overturned mine carts, scattered stones, chains, broken timber, iron scrap, discarded armor, and bodies of fallen guards. The space feels cold, damp, heavy, and dusty, with a steel-gray, dirty brown, graphite, and cold white palette. A squad of dark armored quarry guards charges first across the rails and muddy ground, helmets closed, carrying short spears, shields, and hammers. Behind them, emerging from the shadow of the mine arch, comes the boss: a giant one-eyed heavy warrior, 4 to 5 meters tall, rough gray-pink skin, one large round eye in the center of the face, partial metal armor and leather straps across the chest, a massive shoulder plate on one side, thick legs, immense weight, and a huge heavy battle axe carried with brutal momentum.\n\nThe hero briefly raises one hand holding a smooth semi-transparent magnetic orb, tennis-ball sized, silver-white with a pale blue core, a dense metallic vortex inside, thin rings of light rotating around it. The orb already exists when first shown and does not appear from nowhere. As the smaller guards rush in, the hero crushes the orb. A cold magnetic pulse flashes through both forearms and into the ground. Weapons rip out of the charging soldiers’ hands, shields twist sideways, bolts, chains, rail spikes, loose armor plates, and metal fragments shudder and begin lifting from the quarry floor. Some smaller enemies are violently disarmed and thrown off balance immediately.\n\nThe boss accelerates forward through the chaos and swings the giant axe in a heavy horizontal attack. The hero performs a fast professional sidestep and body lean at the last instant; the blade tears across the foreground with dangerous proximity, kicking mud and dust past the camera. The boss recovers with heavy inertia, plants a foot, and launches a second brutal overhead strike. The hero ducks and slips under it in a sharp Hollywood-style evasive motion, the axe crashing into the ground and rails with a shower of dirt, sparks, and metal shock. The camera reacts like a real head-and-torso dodge: fast lateral movement, recoil, breath, then immediate refocus on the boss.\n\nNow the magnetic field fully takes hold. The hero spreads both hands outward. Metal rises from every direction in a controlled field, not chaotic: torn rails, chains, bolts, iron rings, armor plates, mine cart fragments, shield metal. The pieces lift, align, and begin orbiting the giant in three clear rotating bands: a lower ring around the knees, a middle ring around the waist and abdomen, and a top ring around the chest and shoulders. The boss tries to wrench the axe free and charge again, but the orbit tightens around him. Tight center framing, with strong tracking drift and a low-angle hold to emphasize mass and danger.\n\nThe hero then sharply pulls both hands inward. All three magnetic rings slam tighter at once. Chains snap taut, rails bend into crushing arcs, armor plates collide and lock, metal fragments hammer into the boss’s torso and arms. The giant loses balance, staggers, and is forced down, the axe slipping and dropping into the mud. The metal cocoon constricts harder with brutal industrial weight and magnetic pressure, locking his body in place. Dust rises in waves, small rocks skitter, chains strain, and the one eye flares with panic and rage.\n\nThe hero thrusts both hands forward for the finisher. A narrow concentrated magnetic impulse forms between the palms, not fire, but a dense electromagnetic strike with a bright white core and pale blue edges. Brief micro push-in to the center of the boss’s chest. The pulse hits. Every piece of metal bound around the giant clamps inward simultaneously with one final crushing lock. The boss drops fully to his knees, torso compressed under chained rails and armored fragments, while a deep metallic shock ripples through the quarry and a wave of dust spreads across the ground.\n\nEnd on the giant kneeling and trapped in a brutal iron cocoon, chains stretched tight, rails embedded across the torso, the fallen axe half-buried in mud, drifting dust, and a few small metal fragments still hovering and spinning before they finally drop.\n\nLighting: cold gray cloud light, faint wet reflections on stone, mud, and rails, cold silver-white illumination from the orb and the hero’s magic lines, sharp cold highlights on airborne metal during orbit, and a final white-blue flash from the magnetic impulse. Cinematic dark fantasy, anamorphic look, subtle film grain, shallow depth of field on hands and nearby metal, strong physical inertia, realistic heavy enemy movement, rails resisting as they tear from the ground, chains tightening with believable tension, mud displacement, rolling debris, and controlled magnetic motion that feels like a field, not random telekinesis. Sound design: low wind in the quarry, distant metal groans, armor rattle from the charging guards, tight compression sound as the orb is crushed, deep low-frequency magnetic hum, metal shrieking free from the ground, heavy boss footsteps, axe whooshes with real mass, impact crashes into mud and rails, chain strain, rail collisions, a sharp electromagnetic finisher, then a massive iron-lock resonance and fading dust-heavy rumble.\n\nNo third-person view, no visible hero face, no constant hands in frame, no random floating metal chaos, no weak enemy scale, no weightless boss, no fire magic look, no glowing fantasy laser beam, no cartoon physics, no soft combat, no low-detail quarry, no tiny axe, no silly enemy behavior, no zombie movement.\n\n#AIVideoComparison #seedance2 #MiniMax #MiniMaxH3 #Hailuo",
    "promptSourceUrl": "https://x.com/KrevixAi/status/2085759071047565341",
    "sourceUrl": "https://x.com/KrevixAi/status/2085759071047565341",
    "sourceLabel": "X 原帖 · @KrevixAi",
    "author": "KreviX",
    "publishedAt": "2026-08-07T16:04:48.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085759071047565341.mp4",
    "posterUrl": "/posters/x/2085759071047565341.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "2160×2700",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "sci-fi",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo MiniMax H3 (2K) vs. Seedance 2.0 (720P) 👀 \n\nSame prompt. \nIn my opinion, the result is different.  \n\nWithout a doubt, Seedance 2.0 remains my favorite.\n\nWhich one looks better to you? \n\nExact prompt👇\nStrict first-person POV, male hero, face never visible. Visible only the hero’s hands, fingers, and forearms when needed for casting, dodging balance, and the final magnetic strike. Broad male hands, pronounced knuckles, short nails, pale or cool-toned skin, dark gray or black sleeves, thin geometric magic lines glowing cold white with a faint blue tint across the skin and wrists. The hands move with force, precision, and short controlled gestures, never soft or decorative. One continuous shot, no cuts, vertical cinematic dark fantasy, realistic body-driven camera motion, professional action cinematography with physical momentum and readable combat beats.\n\nIn a vast abandoned stone quarry under a gray overcast sky, high rock walls, two dark mine tunnel arches in the background, old metal rails cutting through wet red-brown mud, overturned mine carts, scattered stones, chains, broken timber, iron scrap, discarded armor, and bodies of fallen guards. The space feels cold, damp, heavy, and dusty, with a steel-gray, dirty brown, graphite, and cold white palette. A squad of dark armored quarry guards charges first across the rails and muddy ground, helmets closed, carrying short spears, shields, and hammers. Behind them, emerging from the shadow of the mine arch, comes the boss: a giant one-eyed heavy warrior, 4 to 5 meters tall, rough gray-pink skin, one large round eye in the center of the face, partial metal armor and leather straps across the chest, a massive shoulder plate on one side, thick legs, immense weight, and a huge heavy battle axe carried with brutal momentum.\n\nThe hero briefly raises one hand holding a smooth semi-transparent magnetic orb, tennis-ball sized, silver-white with a pale blue core, a dense metallic vortex inside, thin rings of light rotating around it. The orb already exists when first shown and does not appear from nowhere. As the smaller guards rush in, the hero crushes the orb. A cold magnetic pulse flashes through both forearms and into the ground. Weapons rip out of the charging soldiers’ hands, shields twist sideways, bolts, chains, rail spikes, loose armor plates, and metal fragments shudder and begin lifting from the quarry floor. Some smaller enemies are violently disarmed and thrown off balance immediately.\n\nThe boss accelerates forward through the chaos and swings the giant axe in a heavy horizontal attack. The hero performs a fast professional sidestep and body lean at the last instant; the blade tears across the foreground with dangerous proximity, kicking mud and dust past the camera. The boss recovers with heavy inertia, plants a foot, and launches a second brutal overhead strike. The hero ducks and slips under it in a sharp Hollywood-style evasive motion, the axe crashing into the ground and rails with a shower of dirt, sparks, and metal shock. The camera reacts like a real head-and-torso dodge: fast lateral movement, recoil, breath, then immediate refocus on the boss.\n\nNow the magnetic field fully takes hold. The hero spreads both hands outward. Metal rises from every direction in a controlled field, not chaotic: torn rails, chains, bolts, iron rings, armor plates, mine cart fragments, shield metal. The pieces lift, align, and begin orbiting the giant in three clear rotating bands: a lower ring around the knees, a middle ring around the waist and abdomen, and a top ring around the chest and shoulders. The boss tries to wrench the axe free and charge again, but the orbit tightens around him. Tight center framing, with strong tracking drift and a low-angle hold to emphasize mass and danger.\n\nThe hero then sharply pulls both hands inward. All three magnetic rings slam tighter at once. Chains snap taut, rails bend into crushing arcs, armor plates collide and lock, metal fragments hammer into the boss’s torso and arms. The giant loses balance, staggers, and is forced down, the axe slipping and dropping into the mud. The metal cocoon constricts harder with brutal industrial weight and magnetic pressure, locking his body in place. Dust rises in waves, small rocks skitter, chains strain, and the one eye flares with panic and rage.\n\nThe hero thrusts both hands forward for the finisher. A narrow concentrated magnetic impulse forms between the palms, not fire, but a dense electromagnetic strike with a bright white core and pale blue edges. Brief micro push-in to the center of the boss’s chest. The pulse hits. Every piece of metal bound around the giant clamps inward simultaneously with one final crushing lock. The boss drops fully to his knees, torso compressed under chained rails and armored fragments, while a deep metallic shock ripples through the quarry and a wave of dust spreads across the ground.\n\nEnd on the giant kneeling and trapped in a brutal iron cocoon, chains stretched tight, rails embedded across the torso, the fallen axe half-buried in mud, drifting dust, and a few small metal fragments still hovering and spinning before they finally drop.\n\nLighting: cold gray cloud light, faint wet reflections on stone, mud, and rails, cold silver-white illumination from the orb and the hero’s magic lines, sharp cold highlights on airborne metal during orbit, and a final white-blue flash from the magnetic impulse. Cinematic dark fantasy, anamorphic look, subtle film grain, shallow depth of field on hands and nearby metal, strong physical inertia, realistic heavy enemy movement, rails resisting as they tear from the ground, chains tightening with believable tension, mud displacement, rolling debris, and controlled magnetic motion that feels like a field, not random telekinesis. Sound design: low wind in the quarry, distant metal groans, armor rattle from the charging guards, tight compression sound as the orb is crushed, deep low-frequency magnetic hum, metal shrieking free from the ground, heavy boss footsteps, axe whooshes with real mass, impact crashes into mud and rails, chain strain, rail collisions, a sharp electromagnetic finisher, then a massive iron-lock resonance and fading dust-heavy rumble.\n\nNo third-person view, no visible hero face, no constant hands in frame, no random floating metal chaos, no weak enemy scale, no weightless boss, no fire magic look, no glowing fantasy laser beam, no cartoon physics, no soft combat, no low-detail quarry, no tiny axe, no silly enemy behavior, no zombie movement.\n\n#AIVideoComparison #seedance2 #MiniMax #MiniMaxH3 #Hailuo",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 19,
      "views": 8555,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085691802233835622",
    "title": "随音乐疾驰的超现实时尚",
    "titleEn": "Surreal Fashion in Perfect Rhythm",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "随音乐疾驰的超现实时尚的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for surreal fashion in perfect rhythm on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a fast-paced surreal editorial sequence synchronized tightly to the reference audio @[audio1].  No static poses, no pauses, no idle moments.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2085691802233835622",
    "sourceUrl": "https://x.com/aimikoda/status/2085691802233835622",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-07T11:37:30.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085691802233835622.mp4",
    "posterUrl": "/posters/x/2085691802233835622.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "stylized"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Midjourney v8.2 + Suno + MiniMax H3\n\nPrompt: Create a fast-paced surreal editorial sequence synchronized tightly to the reference audio @[audio1].  No static poses, no pauses, no idle moments.",
    "engagement": {
      "replies": 18,
      "reposts": 4,
      "likes": 142,
      "views": 9557,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085663805229715612",
    "title": "宇航员花园的双模型对决",
    "titleEn": "Two Models Enter an Astronaut Garden",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "宇航员花园的双模型对决的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 30 秒 · 2160×2880。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for two models enter an astronaut garden on X. Native-video output: 30s · 2160×2880. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A 15-second ultra-realistic cinematic space odyssey scene inspired by 2001: A Space Odyssey. A lone astronaut is on a small drifting asteroid in the infinite black void of space, with a tiny distant Earth in the background, caring for a single delicate flower growing from the rock. No dialogue. The emotional tone is lonely, quiet, stubborn, and deeply human. The camera should move in one smooth, elegant, slow cinematic shot with minimal cutting, gradually pushing closer and slightly circling to reveal the emotional progression through small actions. The astronaut first carefully waters the flower with a thin controlled stream, then sets the container down beside him, then gently gathers and tidies the loose soil around the base of the flower with his gloved hand, then softly strokes or steadies the flower with a delicate touch, and finally shifts his weight and sits down beside it, quietly looking at the flower in stillness. Every action must be subtle, precise, and believable in a bulky spacesuit, showing inner richness, solitude, patience, and stubborn hope through maintenance and care rather than dramatic acting. Audio should be immersive and prominent: clear helmet breathing, subtle suit servo and life-support hum, deep ambient sci-fi atmosphere, and a sparse solemn score with slowly rising emotional weight. Visual style should be minimal, sacred, poetic, and precise, emphasizing the contrast between the tiny flower, the lonely astronaut, and the endless cosmic emptiness. No exaggerated acting, no fantasy glow, no magical plant growth, no shaky handheld, no fast montage, no subtitle text.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2085663805229715612",
    "sourceUrl": "https://x.com/ou_zhen599/status/2085663805229715612",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-08-07T09:46:15.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085663805229715612.mp4",
    "posterUrl": "/posters/x/2085663805229715612.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "2160×2880",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Who do you think is better?\nSince we have the same image and the same set of prompts, I decided to run a comparison test between MiniMax H3 and Grok Imagine Video 1.5 using a simple 6-second video.\nThere’s one pretty interesting detail: when the astronaut’s sleeve turned away, Grok automatically generated a U.S. flag on the sleeve, while MiniMax H3 didn’t. Haha, I’m not sure if that counts as a detail.\n\n正好同一张图同一组提示词，拿MiniMax H3和Grok lmagine Video 1.5做下对比测试，简单的6秒动态。\n有个挺有意思的细节，宇航员的袖标在转过去的时候Grok直接生成了美国国旗袖标，而MiniMax H3却不会。哈哈，不知道算不算细节。\nprompt👇\nA 15-second ultra-realistic cinematic space odyssey scene inspired by 2001: A Space Odyssey. A lone astronaut is on a small drifting asteroid in the infinite black void of space, with a tiny distant Earth in the background, caring for a single delicate flower growing from the rock. No dialogue. The emotional tone is lonely, quiet, stubborn, and deeply human. The camera should move in one smooth, elegant, slow cinematic shot with minimal cutting, gradually pushing closer and slightly circling to reveal the emotional progression through small actions. The astronaut first carefully waters the flower with a thin controlled stream, then sets the container down beside him, then gently gathers and tidies the loose soil around the base of the flower with his gloved hand, then softly strokes or steadies the flower with a delicate touch, and finally shifts his weight and sits down beside it, quietly looking at the flower in stillness. Every action must be subtle, precise, and believable in a bulky spacesuit, showing inner richness, solitude, patience, and stubborn hope through maintenance and care rather than dramatic acting. Audio should be immersive and prominent: clear helmet breathing, subtle suit servo and life-support hum, deep ambient sci-fi atmosphere, and a sparse solemn score with slowly rising emotional weight. Visual style should be minimal, sacred, poetic, and precise, emphasizing the contrast between the tiny flower, the lonely astronaut, and the endless cosmic emptiness. No exaggerated acting, no fantasy glow, no magical plant growth, no shaky handheld, no fast montage, no subtitle text.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 744,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085656146363249079",
    "title": "扭蛋机里的另一个世界",
    "titleEn": "Another World Inside a Capsule Machine",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "扭蛋机里的另一个世界的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for another world inside a capsule machine on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Subject: Person from Image1. Face and hairstyle must match Image1 completely; do not transform into a different person. Outfit: same casual everyday streetwear as Part 1. Ignore the outfit, background, pose, and bottom text from Image. Even if image quality is rough, facial features must remain clearly defined. 16:9 horizontal. Entire video has an old iPhone low-quality everyday-video look. Constant heavy noise and compression artifacts throughout.\nSetting: Same convenience store or arcade corner immediately after leaving the purikura booth. Sticker printer beside the booth, capsule toy machines, snack shelves, and bright fluorescent store lighting mixed with colorful purikura lights. Booth curtain still visible in the background. Any visible signs, packaging, and screen text remain completely illegible.\nTimeline:\n0–3s: Holds up the phone with the newly attached sticker, smiling proudly. Notices a tiny bonus sticker still attached to the backing sheet and laughs softly. \"Wait... there's another one?\"\n3–6s: Peels off the tiny sticker and sticks it onto a reusable drink bottle or wallet. While pressing it down, a nearby capsule toy machine suddenly dispenses a capsule by itself with a loud clunk, making them jump in surprise.\n6–9s: Picks up the unexpected capsule, shakes it curiously, opens it, and finds a tiny cute keychain inside. Holds it next to the purikura sticker, laughing. \"No way... perfect timing!\"\n9–12s: Hooks the keychain onto their bag or phone strap, admires both the sticker and keychain together, then catches their reflection in the booth window and gives themselves an amused grin.\n12–15s: Looks directly at the camera, holds up the phone and the new keychain together, gives a playful wink and thumbs-up while saying, \"Best random stop ever.\" Freeze on the final frame.\nCamera: Loose 1–1.5s cuts. Old iPhone, vertical, standard 1x lens (no distortion, standard field of view), handheld with slight natural shake while walking and reacting, steadier during close-up sticker and capsule moments. Slight autofocus hunting during close-ups. Mixed fluorescent and colorful booth lighting creates inconsistent white balance with a subtle magenta/pink cast. Natural exposure fluctuations throughout.\nSound: No BGM. Ambient convenience store and arcade background noise (faint chatter, vending machines, capsule toy machine sounds, footsteps), sticker peeling sound, plastic capsule opening, small keychain clink, natural laughter, close slightly muffled voice, subtle booth echo, phone handling sounds.\nStyle: Unedited low-quality phone video look. No color grading, no film look, no cinematic look, no beauty filter, no CG texture. No fisheye/ultra-wide (0.5x)/action-cam look, no barrel distortion, no vignette, no black lens-round at frame edges—flat standard field of view, frame filled edge to edge. Real skin texture maintained (pores, flyaway hairs, natural oil sheen). No cut consolidation, no scene skipping. No on-screen text (all store signs, product labels, booth screens, capsule contents, and sticker graphics/text remain illegible).",
    "promptSourceUrl": "https://x.com/itxabdullaa/status/2085656146363249079",
    "sourceUrl": "https://x.com/itxabdullaa/status/2085656146363249079",
    "sourceLabel": "X 原帖 · @itxabdullaa",
    "author": "ÀBDŪLLÂH",
    "publishedAt": "2026-08-07T09:15:49.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085656146363249079.mp4",
    "posterUrl": "/posters/x/2085656146363249079.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax H3 on @Hailuo_AI \n\nPrompt:\n\nSubject: Person from Image1. Face and hairstyle must match Image1 completely; do not transform into a different person. Outfit: same casual everyday streetwear as Part 1. Ignore the outfit, background, pose, and bottom text from Image. Even if image quality is rough, facial features must remain clearly defined. 16:9 horizontal. Entire video has an old iPhone low-quality everyday-video look. Constant heavy noise and compression artifacts throughout.\nSetting: Same convenience store or arcade corner immediately after leaving the purikura booth. Sticker printer beside the booth, capsule toy machines, snack shelves, and bright fluorescent store lighting mixed with colorful purikura lights. Booth curtain still visible in the background. Any visible signs, packaging, and screen text remain completely illegible.\nTimeline:\n0–3s: Holds up the phone with the newly attached sticker, smiling proudly. Notices a tiny bonus sticker still attached to the backing sheet and laughs softly. \"Wait... there's another one?\"\n3–6s: Peels off the tiny sticker and sticks it onto a reusable drink bottle or wallet. While pressing it down, a nearby capsule toy machine suddenly dispenses a capsule by itself with a loud clunk, making them jump in surprise.\n6–9s: Picks up the unexpected capsule, shakes it curiously, opens it, and finds a tiny cute keychain inside. Holds it next to the purikura sticker, laughing. \"No way... perfect timing!\"\n9–12s: Hooks the keychain onto their bag or phone strap, admires both the sticker and keychain together, then catches their reflection in the booth window and gives themselves an amused grin.\n12–15s: Looks directly at the camera, holds up the phone and the new keychain together, gives a playful wink and thumbs-up while saying, \"Best random stop ever.\" Freeze on the final frame.\nCamera: Loose 1–1.5s cuts. Old iPhone, vertical, standard 1x lens (no distortion, standard field of view), handheld with slight natural shake while walking and reacting, steadier during close-up sticker and capsule moments. Slight autofocus hunting during close-ups. Mixed fluorescent and colorful booth lighting creates inconsistent white balance with a subtle magenta/pink cast. Natural exposure fluctuations throughout.\nSound: No BGM. Ambient convenience store and arcade background noise (faint chatter, vending machines, capsule toy machine sounds, footsteps), sticker peeling sound, plastic capsule opening, small keychain clink, natural laughter, close slightly muffled voice, subtle booth echo, phone handling sounds.\nStyle: Unedited low-quality phone video look. No color grading, no film look, no cinematic look, no beauty filter, no CG texture. No fisheye/ultra-wide (0.5x)/action-cam look, no barrel distortion, no vignette, no black lens-round at frame edges—flat standard field of view, frame filled edge to edge. Real skin texture maintained (pores, flyaway hairs, natural oil sheen). No cut consolidation, no scene skipping. No on-screen text (all store signs, product labels, booth screens, capsule contents, and sticker graphics/text remain illegible).",
    "engagement": {
      "replies": 54,
      "reposts": 4,
      "likes": 74,
      "views": 1869,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085649264185913721",
    "title": "纸艺山谷的幻想大战",
    "titleEn": "A Papercraft Battle in a Fantasy Valley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "纸艺山谷的幻想大战的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1280×736。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a papercraft battle in a fantasy valley on X. Native-video output: 15s · 1280×736. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "【核心创意】\n\n12 秒，16:9 横屏视频。手工纸艺立体拼贴与奇幻故事书风格的二维动画。画面由多层纸片、裁切边缘、纸张纤维、柔和层叠阴影和轻微定格动画质感构成，色彩鲜明，具有微缩舞台空间感。\n\n故事发生在一片绿色山谷战场。左侧是英雄阵营：一名棕色长发的女弓箭手，穿黑色轻甲；一名白色长胡须的老法师，穿深蓝长袍，手持燃烧法杖；一名穿银灰色盔甲的年轻骑士，手持长剑和盾牌；远处有一匹白色独角兽和一面带白色独角兽图案的蓝色旗帜。右侧是敌方阵营：骷髅士兵、绿色哥布林和一名骑乘棕红色飞龙的黑甲骑士，旁边竖立一面黑色骷髅旗帜。远处山脉、松树林和一座冒着黑烟的黑色尖塔城堡。\n\n【画面过程说明】\n\n0–2 秒｜大全景：\n镜头从山谷战场大全景缓慢向前 Push In。蓝色独角兽旗帜和黑色骷髅旗帜在风中摆动，城堡上方的黑烟缓慢升起，草地和火焰轻微运动。英雄阵营与敌方阵营隔着战场对峙。\n\n2–4 秒｜中景切换：\n镜头切到左侧女弓箭手。她迅速拉满弓弦，瞄准右侧骷髅士兵并射出一支箭。箭矢沿清晰轨迹飞过战场，骷髅士兵举盾挡住箭矢，产生少量纸片火花，不出现受伤或死亡。\n\n4–6 秒｜仰拍跟随：\n镜头快速 Tilt Up 跟随棕红色飞龙。飞龙扇动翅膀俯冲，向英雄阵营喷出一股短暂的橙色火焰。老法师举起燃烧法杖，形成一面半透明金色魔法护盾。火焰撞上护盾后向两侧散开，不发生爆炸。\n\n6–9 秒｜低机位中景：\n年轻骑士举起长剑向前冲锋，与一名骷髅战士发生一次清晰的剑盾碰撞。绿色哥布林从侧面冲来，骑士侧身闪避。镜头进行短距离 Tracking Shot，跟随骑士向右移动，动作快速但保持人物比例、盔甲和武器稳定。\n\n9–11 秒｜动作高潮：\n镜头切到战场中央。骑士与骷髅战士再次交锋，女弓箭手继续射箭，老法师维持金色护盾，飞龙在上方喷出短暂火焰，哥布林举盾冲向前方。加入少量尘土、火星和纸片碎屑，表现激烈战斗，但不出现血腥画面。\n\n11–12 秒｜大全景定格：\n镜头快速 Pull Out，重新展示完整山谷战场。飞龙在空中盘旋，弓箭手保持射击姿势，法师举杖维持护盾，骑士与骷髅战士停留在交锋姿态，双方仍未分出胜负。火焰、烟雾和旗帜继续运动，最后定格在战斗高潮画面。\n\n声音：风吹旗帜声、弓弦拉动声、箭矢破空声、飞龙翅膀声、火焰喷射声、魔法护盾低沉嗡鸣、剑盾碰撞声、哥布林脚步声和紧张的奇幻战斗配乐。无对白、无旁白。\n\n不想要：写实摄影、普通3D游戏CG、塑料材质、角色换装、脸部变化、人物数量明显增加、武器消失、箭矢倒飞、火焰爆炸、血腥画面、断肢、角色死亡、骷髅或哥布林变形、字幕、Logo、乱码文字、镜头剧烈抖动、黑屏转场。\n\n必须保持：全片为手工纸艺立体拼贴与二维定格动画风格；女弓箭手、老法师、年轻骑士、飞龙骑士、骷髅士兵和绿色哥布林的外观全程一致；英雄阵营位于左侧，敌方阵营位于右侧；结尾双方仍在交战，不出现明确胜负。",
    "promptSourceUrl": "https://x.com/xianshi666123/status/2085649264185913721",
    "sourceUrl": "https://x.com/xianshi666123/status/2085649264185913721",
    "sourceLabel": "X 原帖 · @xianshi666123",
    "author": "闲适",
    "publishedAt": "2026-08-07T08:48:28.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085649264185913721.mp4",
    "posterUrl": "/posters/x/2085649264185913721.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cg-3d",
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 \n\n对于图片的风格参考的 复现效果也不错\n\nprompt：\n【核心创意】\n\n12 秒，16:9 横屏视频。手工纸艺立体拼贴与奇幻故事书风格的二维动画。画面由多层纸片、裁切边缘、纸张纤维、柔和层叠阴影和轻微定格动画质感构成，色彩鲜明，具有微缩舞台空间感。\n\n故事发生在一片绿色山谷战场。左侧是英雄阵营：一名棕色长发的女弓箭手，穿黑色轻甲；一名白色长胡须的老法师，穿深蓝长袍，手持燃烧法杖；一名穿银灰色盔甲的年轻骑士，手持长剑和盾牌；远处有一匹白色独角兽和一面带白色独角兽图案的蓝色旗帜。右侧是敌方阵营：骷髅士兵、绿色哥布林和一名骑乘棕红色飞龙的黑甲骑士，旁边竖立一面黑色骷髅旗帜。远处山脉、松树林和一座冒着黑烟的黑色尖塔城堡。\n\n【画面过程说明】\n\n0–2 秒｜大全景：\n镜头从山谷战场大全景缓慢向前 Push In。蓝色独角兽旗帜和黑色骷髅旗帜在风中摆动，城堡上方的黑烟缓慢升起，草地和火焰轻微运动。英雄阵营与敌方阵营隔着战场对峙。\n\n2–4 秒｜中景切换：\n镜头切到左侧女弓箭手。她迅速拉满弓弦，瞄准右侧骷髅士兵并射出一支箭。箭矢沿清晰轨迹飞过战场，骷髅士兵举盾挡住箭矢，产生少量纸片火花，不出现受伤或死亡。\n\n4–6 秒｜仰拍跟随：\n镜头快速 Tilt Up 跟随棕红色飞龙。飞龙扇动翅膀俯冲，向英雄阵营喷出一股短暂的橙色火焰。老法师举起燃烧法杖，形成一面半透明金色魔法护盾。火焰撞上护盾后向两侧散开，不发生爆炸。\n\n6–9 秒｜低机位中景：\n年轻骑士举起长剑向前冲锋，与一名骷髅战士发生一次清晰的剑盾碰撞。绿色哥布林从侧面冲来，骑士侧身闪避。镜头进行短距离 Tracking Shot，跟随骑士向右移动，动作快速但保持人物比例、盔甲和武器稳定。\n\n9–11 秒｜动作高潮：\n镜头切到战场中央。骑士与骷髅战士再次交锋，女弓箭手继续射箭，老法师维持金色护盾，飞龙在上方喷出短暂火焰，哥布林举盾冲向前方。加入少量尘土、火星和纸片碎屑，表现激烈战斗，但不出现血腥画面。\n\n11–12 秒｜大全景定格：\n镜头快速 Pull Out，重新展示完整山谷战场。飞龙在空中盘旋，弓箭手保持射击姿势，法师举杖维持护盾，骑士与骷髅战士停留在交锋姿态，双方仍未分出胜负。火焰、烟雾和旗帜继续运动，最后定格在战斗高潮画面。\n\n声音：风吹旗帜声、弓弦拉动声、箭矢破空声、飞龙翅膀声、火焰喷射声、魔法护盾低沉嗡鸣、剑盾碰撞声、哥布林脚步声和紧张的奇幻战斗配乐。无对白、无旁白。\n\n不想要：写实摄影、普通3D游戏CG、塑料材质、角色换装、脸部变化、人物数量明显增加、武器消失、箭矢倒飞、火焰爆炸、血腥画面、断肢、角色死亡、骷髅或哥布林变形、字幕、Logo、乱码文字、镜头剧烈抖动、黑屏转场。\n\n必须保持：全片为手工纸艺立体拼贴与二维定格动画风格；女弓箭手、老法师、年轻骑士、飞龙骑士、骷髅士兵和绿色哥布林的外观全程一致；英雄阵营位于左侧，敌方阵营位于右侧；结尾双方仍在交战，不出现明确胜负。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 1719,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085626141759709286",
    "title": "为自己打造的未来广告",
    "titleEn": "A Futuristic Ad Made for Itself",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "为自己打造的未来广告的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1084。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a futuristic ad made for itself on X. Native-video output: 15s · 2560×1084. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "REFERENCE USAGE — nine cropped storyboard panels, in order:\nImage 1 — hook frame: the MiniMax H3 wordmark, the Create Beyond script line, the woman in violet light with jellyfish and particles.\nImage 2 — multimodal frame: floating glass panels showing image, video and cosmic content, with its heading.\nImage 3 — Hailuo frame: the MiniMax H3 x HailuoAI rounded glowing lockup on violet nebula.\nImage 4 — editing frame: the floating glass editing timeline interface with its tool rail and clip thumbnails.\nImage 5 — production frame: the woman with hair flying against violet speed streaks, with its heading.\nImage 6 — live frame: the words LIVE RIGHT NOW in heavy white type over a violet horizon burst.\nImage 7 — creators frame: the row of five vertical content cards and its heading.\nImage 8 — possibilities frame: the four panel grid of marketing, film, music and games with its labels.\nImage 9 — CTA frame: the MiniMax H3 x HailuoAI lockup with the AVAILABLE NOW glowing pill button.\nTake every composition, subject, layout, wordmark and colour from these nine panels. Reproduce all typography, letterforms, weights, colours and positions exactly as they appear in the panels. Do not redraw, re-typeset, translate, correct or invent any text.\n\nA 16:9 high energy sixty second style product launch commercial compressed into fifteen seconds, for an AI video generation platform. Palette locked to deep violet and near black with electric purple light, hot white type and bright particle energy. Every frame carries volumetric violet glow, drifting light particles, translucent glass surfaces with edge refraction, and soft bloom on every bright edge. Premium technology advertising in the language of a flagship device launch: confident, fast, expensive, futuristic. Never cheap, never cluttered, never retro.\n\nCutting: nine beats, hard cuts only, no dissolves or soft transitions. Hold every cut until its stated timecode and do not shorten any shot. Energy escalates continuously from the first frame to the last.\n\nTEXT RENDERING: all on screen text is revealed by hard edged masks or lands complete on the cut, always fully sharp, at final size and full opacity, in its exact position. Never scale, fade, warp, letter space or motion blur any text. Never let text be soft, doubled or partially transparent at any point.\n\nTEXT LOCK — only these strings appear, spelled exactly:\nMiniMax H3 / Create Beyond. / Native Multimodal Generation / Text + Image + Video + Audio / Precise Multimodal Editing / Production-Ready Content Creation / For Creators. By Creators. / LIVE RIGHT NOW / Marketing / Film & TV / Music / Games / AVAILABLE NOW / HailuoAI\nNo numbers, captions, subtitles, credits or any other words appear anywhere.\n\n[0-2s] Image 1. A violet shockwave detonates outward from frame centre through drifting particles and jellyfish, and the composition resolves out of it. The MiniMax H3 wordmark lands hard and complete. The Create Beyond script line writes on beneath it in one confident stroke as light traces the letterforms. HARD CUT.\n\n[2-3.6s] Image 2. The glass panels fly in from depth and lock into their arrangement, each one settling with a soft refraction flare. The content inside each panel is live and moving. The heading lands complete above them. HARD CUT.\n\n[3.6-4.8s] Image 3. The rounded lockup ignites, its glowing border drawing on around the corners in one continuous run of light. Jellyfish drift past in the violet nebula behind it. HARD CUT.\n\n[4.8-6.4s] Image 4. The floating editing interface assembles in depth: the tool rail slides in from the left, the timeline extends to the right, clip thumbnails populate in sequence, and a selection highlight snaps onto one clip. The camera drifts subtly across the glass surface. HARD CUT.\n\n[6.4-8.2s] Image 5. Full speed. Violet light streaks race past the lens at maximum velocity, her hair flying, particles tearing through frame. Camera [Push in] hard and fast. The heading lands mid motion. This is the peak of the film. HARD CUT.\n\n[8.2-9.6s] Image 6. A burst of white light detonates on the horizon and LIVE RIGHT NOW lands enormous and complete, filling the frame. HARD CUT.\n\n[9.6-11.2s] Image 7. The five vertical content cards fly in and lock into a row one after another in rapid succession, each landing with a violet edge flare, the content inside each card alive and moving. The heading lands above them. HARD CUT.\n\n[11.2-12.8s] Image 8. The four panel grid snaps into place, each quadrant arriving on its own beat with its label landing complete. All four panels play live simultaneously. HARD CUT.\n\n[12.8-15s] Image 9. Final lockup. Particles converge inward toward frame centre. The MiniMax H3 and HailuoAI marks land, then the AVAILABLE NOW pill ignites, its glowing border drawing around the button in one stroke and settling into a steady pulse. Everything holds still and sharp for the final second.\n\nScore: an original high energy cinematic technology launch score drives the whole film. Instrumentation is a deep sub bass drone, hard synthetic percussion, rising synth arpeggios, wide detuned brass style stabs, granular risers and one enormous impact hit. Modern, confident, cinematic and expensive, in the language of a flagship product launch. No vocals, no lyrics, no orchestral strings, no retro synthwave.\n\nThe score opens at 0 seconds on one enormous impact hit and a low sub drone under the shockwave. From 2 seconds a hard percussive pulse enters and locks in, one hit per cut. From 4.8 seconds the arpeggios enter and begin climbing. From 6.4 seconds a granular riser builds hard under the speed streaks and peaks at 8.2 seconds, where the brass stabs and the full arrangement slam in at maximum weight for the LIVE RIGHT NOW frame. It drives at full power through 12.8 seconds. At 12.8 seconds everything resolves into one wide sustained chord over the final lockup, decaying gently to the end.\n\nSound design sits under the score: a deep whoosh and glass chime as each panel or card locks into place, a rising electrical charge as each glowing border draws on, a hard transient on every hard cut, and one clean synthetic confirmation tone as the AVAILABLE NOW button ignites. Keep all of it tight and low in the mix. No voiceover, no narrator, no dialogue, no library stock music. The score must be an original composition and must not imitate any existing track.\n\nDo not add subtitles, captions, credits, watermarks, panel numbers, section labels or any word outside the TEXT LOCK list. Do not reproduce the storyboard grid, its white caption strips or its numbered headings. Do not alter any wordmark or logo. Do not change the woman's face between shots. No lens flare streaks across the whole frame, no glitch RGB split, no film grain, no letterboxing.",
    "promptSourceUrl": "https://x.com/abulu8/status/2085626148021789130",
    "sourceUrl": "https://x.com/abulu8/status/2085626141759709286",
    "sourceLabel": "X 原帖 · @abulu8",
    "author": "Syed Abuthahir ∞",
    "publishedAt": "2026-08-07T07:16:35.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085626141759709286.mp4",
    "posterUrl": "/posters/x/2085626141759709286.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1084",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Creating a sleek, professional ad for #MiniMaxH3 using MiniMax H3 itself. ⚡ \n\nAs an AI creator, my favorite part is how well it preserves text and naturally animates it with SFX alongside the visuals.\n\nPrompt in the comments. 💥\n\n@Hailuo_AI #AIVideo #AIAdvertising",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 10,
      "views": 907,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085625652829708458",
    "title": "少女持剑对决霸王龙",
    "titleEn": "A Sword-Wielding Girl Faces a T. Rex",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少女持剑对决霸王龙的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1088×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a sword-wielding girl faces a t. rex on X. Native-video output: 15s · 1088×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 14-second ultra-realistic Hollywood-style action fantasy film scene featuring a brave young female warrior battling a massive realistic Tyrannosaurus Rex inside an ancient jungle filled with ruined temples, giant trees, mist, mud, rain, smoke, and dramatic sunlight.\n\nVisual Style: Premium Hollywood blockbuster, photorealistic, cinematic, epic scale, realistic human anatomy, realistic dinosaur skin and movement, dramatic lighting, detailed environment, natural physics, intense facial expressions, high-end VFX, ARRI Alexa 35 look, anamorphic lens, shallow depth of field, HDR, subtle film grain, 4K cinematic quality, dynamic camera movement, realistic motion blur.\n\n0–2 seconds — Establishing Shot:\nWide cinematic shot of an ancient jungle temple in the middle of a mysterious rainforest. The female warrior stands alone in the ruins, holding a large sword in one hand. Her clothes are rugged fantasy-warrior clothing, slightly wet from the rain. Mist moves through the trees while birds fly in the distant sky. Slow camera push-in toward her.\n\n2–3 seconds — Hero Close-Up:\nCut to a dramatic close-up of the girl's face. Her hair is slightly wet and moving in the wind. She looks determined and fearless. Small scratches and dirt are visible on her face. She slowly raises her sword and prepares for battle.\n\n3–4 seconds — Dinosaur Reveal:\nSuddenly, a massive realistic T-Rex emerges from behind the jungle trees and roars aggressively. Its enormous jaws open, sharp teeth visible, water and mud splashing around its feet. Low-angle camera makes the dinosaur look huge and terrifying.\n\n4–6 seconds — First Attack:\nThe dinosaur charges toward the girl at full speed. She runs toward the side and performs a fast dodge at the last second. The dinosaur crashes through trees and debris behind her. Use handheld camera movement, fast tracking, realistic impact and flying dust.\n\n6–8 seconds — Sword Counterattack:\nThe girl quickly turns around and attacks with her sword. Show a dramatic slow-motion moment as she swings the sword toward the dinosaur. The dinosaur moves its head aggressively while roaring. Strong cinematic camera movement follows the sword.\n\n8–10 seconds — Epic Jump:\nThe dinosaur attacks again. The girl jumps onto a large fallen tree and then leaps into the air above the dinosaur. Her hair and clothing move naturally in slow motion. Camera rotates around her as she raises the sword for the final strike.\n\n10–12 seconds — Final Strike:\nShe lands a powerful cinematic sword strike against the dinosaur's armored hide. Create a dramatic impact with dust, rainwater, debris and strong camera shake. Keep the action non-gory and suitable for a blockbuster fantasy film.\n\n12–13.5 seconds — Dinosaur Falls Back:\nThe dinosaur loses balance and falls backward into the muddy ground with a huge impact. Water, dust and leaves fly into the air. The girl lands safely and stands ready, breathing heavily.\n\n13.5–14 seconds — Hero Ending:\nSlow cinematic push-in toward the girl. She lowers the sword slightly and looks directly toward the camera with a powerful, fearless expression. Behind her, the defeated dinosaur lies in the misty jungle. Golden sunlight breaks through the clouds, creating an epic silhouette and emotional Hollywood-style ending.\nCamera & Cinematography:\nWide establishing shots, low-angle dinosaur shots, close-up facial shots, handheld action camera, tracking shots, 360-degree movement during the jump, dramatic slow motion during key action moments, shallow depth of field for close-ups, realistic motion blur, cinematic composition.\nLighting & Atmosphere:\nDark stormy jungle atmosphere mixed with warm golden sunlight, volumetric light rays, wet surfaces, realistic rain, mist, smoke, flying leaves, dust and water particles.",
    "promptSourceUrl": "https://x.com/Lianaalane/status/2085625652829708458",
    "sourceUrl": "https://x.com/Lianaalane/status/2085625652829708458",
    "sourceLabel": "X 原帖 · @Lianaalane",
    "author": "liana",
    "publishedAt": "2026-08-07T07:14:39.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085625652829708458.mp4",
    "posterUrl": "/posters/x/2085625652829708458.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1088×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One girl. One sword. One unstoppable dinosaur. ⚔️🦖\nWhen courage meets the ultimate battle, legends are born.\n\nCreated with MiniMax H3 \n\nPrompt:\n\nCreate a 14-second ultra-realistic Hollywood-style action fantasy film scene featuring a brave young female warrior battling a massive realistic Tyrannosaurus Rex inside an ancient jungle filled with ruined temples, giant trees, mist, mud, rain, smoke, and dramatic sunlight.\n\nVisual Style: Premium Hollywood blockbuster, photorealistic, cinematic, epic scale, realistic human anatomy, realistic dinosaur skin and movement, dramatic lighting, detailed environment, natural physics, intense facial expressions, high-end VFX, ARRI Alexa 35 look, anamorphic lens, shallow depth of field, HDR, subtle film grain, 4K cinematic quality, dynamic camera movement, realistic motion blur.\n\n0–2 seconds — Establishing Shot:\nWide cinematic shot of an ancient jungle temple in the middle of a mysterious rainforest. The female warrior stands alone in the ruins, holding a large sword in one hand. Her clothes are rugged fantasy-warrior clothing, slightly wet from the rain. Mist moves through the trees while birds fly in the distant sky. Slow camera push-in toward her.\n\n2–3 seconds — Hero Close-Up:\nCut to a dramatic close-up of the girl's face. Her hair is slightly wet and moving in the wind. She looks determined and fearless. Small scratches and dirt are visible on her face. She slowly raises her sword and prepares for battle.\n\n3–4 seconds — Dinosaur Reveal:\nSuddenly, a massive realistic T-Rex emerges from behind the jungle trees and roars aggressively. Its enormous jaws open, sharp teeth visible, water and mud splashing around its feet. Low-angle camera makes the dinosaur look huge and terrifying.\n\n4–6 seconds — First Attack:\nThe dinosaur charges toward the girl at full speed. She runs toward the side and performs a fast dodge at the last second. The dinosaur crashes through trees and debris behind her. Use handheld camera movement, fast tracking, realistic impact and flying dust.\n\n6–8 seconds — Sword Counterattack:\nThe girl quickly turns around and attacks with her sword. Show a dramatic slow-motion moment as she swings the sword toward the dinosaur. The dinosaur moves its head aggressively while roaring. Strong cinematic camera movement follows the sword.\n\n8–10 seconds — Epic Jump:\nThe dinosaur attacks again. The girl jumps onto a large fallen tree and then leaps into the air above the dinosaur. Her hair and clothing move naturally in slow motion. Camera rotates around her as she raises the sword for the final strike.\n\n10–12 seconds — Final Strike:\nShe lands a powerful cinematic sword strike against the dinosaur's armored hide. Create a dramatic impact with dust, rainwater, debris and strong camera shake. Keep the action non-gory and suitable for a blockbuster fantasy film.\n\n12–13.5 seconds — Dinosaur Falls Back:\nThe dinosaur loses balance and falls backward into the muddy ground with a huge impact. Water, dust and leaves fly into the air. The girl lands safely and stands ready, breathing heavily.\n\n13.5–14 seconds — Hero Ending:\nSlow cinematic push-in toward the girl. She lowers the sword slightly and looks directly toward the camera with a powerful, fearless expression. Behind her, the defeated dinosaur lies in the misty jungle. Golden sunlight breaks through the clouds, creating an epic silhouette and emotional Hollywood-style ending.\nCamera & Cinematography:\nWide establishing shots, low-angle dinosaur shots, close-up facial shots, handheld action camera, tracking shots, 360-degree movement during the jump, dramatic slow motion during key action moments, shallow depth of field for close-ups, realistic motion blur, cinematic composition.\nLighting & Atmosphere:\nDark stormy jungle atmosphere mixed with warm golden sunlight, volumetric light rays, wet surfaces, realistic rain, mist, smoke, flying leaves, dust and water particles.",
    "engagement": {
      "replies": 64,
      "reposts": 6,
      "likes": 206,
      "views": 12242,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085622021812039920",
    "title": "日出湖面的早安标题",
    "titleEn": "A Good-Morning Title Rises from the Lake",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "日出湖面的早安标题的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a good-morning title rises from the lake on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Ultra cinematic macro shot: A calm mountain lake at dawn reflects the first golden rays of sunlight. Tiny ripples spread across the mirror-like surface as morning mist drifts gracefully above the water. Billions of sparkling droplets slowly rise into the air, suspended as if gravity has stopped. They spiral together with incredible fluid realism, naturally sculpting the word \"MORNING\" entirely from crystal-clear water. The liquid letters shimmer with golden reflections before gently collapsing into a spectacular explosion of droplets illuminated by the sunrise. Hollywood title sequence, ultra-realistic fluid simulation, volumetric lighting, IMAX quality.\n\n#MiniMaxH3",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2085622021812039920",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2085622021812039920",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-08-07T07:00:13.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085622021812039920.mp4",
    "posterUrl": "/posters/x/2085622021812039920.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GM ☀️\nI wanted a different kind of morning greeting...\nSo I turned the sunrise itself into a cinematic title sequence with Hailuo H3. 🌅💧\n\nPrompt below 👇\nUltra cinematic macro shot: A calm mountain lake at dawn reflects the first golden rays of sunlight. Tiny ripples spread across the mirror-like surface as morning mist drifts gracefully above the water. Billions of sparkling droplets slowly rise into the air, suspended as if gravity has stopped. They spiral together with incredible fluid realism, naturally sculpting the word \"MORNING\" entirely from crystal-clear water. The liquid letters shimmer with golden reflections before gently collapsing into a spectacular explosion of droplets illuminated by the sunrise. Hollywood title sequence, ultra-realistic fluid simulation, volumetric lighting, IMAX quality.\n\n#MiniMaxH3",
    "engagement": {
      "replies": 8,
      "reposts": 3,
      "likes": 22,
      "views": 3038,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085607833408852170",
    "title": "加油站逃亡故事·下篇",
    "titleEn": "The Gas-Station Escape, Part Two",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "加油站逃亡故事·下篇的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the gas-station escape, part two on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image as the PRIMARY visual, character, prop, environment, camera, lighting, and narrative reference for the entire video.\n\nUse the storyboard to preserve:\n- exact Part 2 story progression\n- Lena Cruz and Jax Mercer identity continuity from Part 1\n- Rico Vale’s appearance and attitude\n- the dark muscle car\n- black SUV convoy\n- black drive\n- unfinished overpass environment\n- sunset-to-night lighting progression\n- graphic-novel cinematic style\n- chase escalation\n- standoff\n- Jax’s “disabled drive” reveal\n- Lena and Jax recovering the drive\n- final victorious walk-away\n\nDo NOT reproduce the storyboard sheet itself.\n\nThe finished video must contain:\n- no panel borders\n- no storyboard numbers\n- no timestamps\n- no captions\n- no production notes\n- no paper background\n- no reference-sheet layout\n\nConvert the entire board into one coherent finished cinematic animated sequence.\n\n[VIDEO GOAL]\n\nDuration: exactly 15 seconds\nAspect ratio: 16:9 landscape\nVideo style: premium cinematic 2D graphic-novel animation\nGenre: Hollywood-style action thriller / road chase\nTone: fast, stylish, tense, controlled, cinematic, triumphant\n\nNarrative progression:\n\nhigh-speed pursuit\n→ convoy closes in\n→ Jax breaks formation\n→ unfinished overpass\n→ convoy collapses\n→ Rico standoff\n→ disabled-drive twist\n→ brief safe disarm\n→ drive recovered\n→ cool final exit\n\nThis is PART 2 only.\n\nBegin immediately with the chase already underway.\n\nDo NOT repeat the gas-station setup from Part 1.\n\nEnd exactly after Lena and Jax recover the black drive and walk toward their muscle car while Jax delivers the final line:\n\n“Enough to burn his empire.”\n\n[CHARACTER IDENTITY LOCKS]\n\nLENA CRUZ\n\nYoung athletic woman.\n\nAppearance:\n- warm medium skin\n- sharp expressive eyes\n- high dark ponytail\n- red patterned bandana over lower face or around neck\n- fitted dark cropped top\n- light denim shorts\n- practical dark footwear\n- lean athletic proportions\n- confident silhouette\n\nPersonality:\nbold, reactive, fearless, intuitive, aggressive under pressure but controlled.\n\nBeginning emotion:\nfocused pursuit intensity.\n\nEnding emotion:\nquiet satisfaction and confidence.\n\nMovement:\nfast head turns,\nathletic body control,\ndecisive reactions,\nstrong braced posture during the car chase,\nquick efficient movements during confrontation.\n\nKeep identical in every shot:\n- face\n- skin tone\n- ponytail\n- bandana\n- clothing\n- height\n- body proportions\n- age\n- eye shape\n\nJAX MERCER\n\nTall athletic young man.\n\nAppearance:\n- short dark hair\n- intense eyes\n- dark blue patterned bandana covering lower face or around neck\n- dark fitted T-shirt\n- brown / charcoal pants\n- strong controlled physique\n\nPersonality:\ncalm strategist,\nprecise,\nrarely panics,\ntactical,\nprotective.\n\nBeginning emotion:\nextreme concentration.\n\nEnding emotion:\ncontrolled victory.\n\nMovement:\nefficient driving,\nminimal wasted motion,\ncalm body language even during danger.\n\nKeep identical:\n- face\n- hairstyle\n- blue bandana\n- shirt\n- pants\n- height\n- proportions\n- body build\n\nRICO VALE\n\nOriginal crime boss character.\n\nAppearance:\n- stylish dark hair\n- confident narrow eyes\n- clean or lightly stubbled face\n- dark tailored street jacket\n- dark shirt\n- restrained jewelry\n- smug posture\n- sophisticated silhouette\n\nPersonality:\narrogant,\ncalculated,\nconfident,\nbelieves he has already won.\n\nBeginning emotion:\nsmug control.\n\nEnding emotion:\nsurprised and outplayed, but physically unharmed.\n\nDo not make Rico grotesque, psychotic, or horror-like.\n\n[PROP LOCKS]\n\nBLACK DRIVE:\nsmall matte-black rectangular data device.\nSame exact shape and size throughout.\nIt must remain visually readable in Rico’s hand and later Lena’s hand.\n\nJAX’S DEVICE:\nsmall generic dark handheld screen / phone.\nDuring the twist, it displays a simple red warning-style state indicating the drive is disabled.\nAvoid recognizable real-world phone branding.\n\nRICO’S PISTOL:\nstylized generic dark handgun.\nOnly used as a visual standoff prop.\nNo firing.\nNo bullets.\nNo injuries.\nIt is safely knocked away during the confrontation.\n\nMUSCLE CAR:\nsame dark charcoal muscle car from Part 1.\nWide stance.\nLong red rear light treatment.\nStrong low silhouette.\nNo real-world logos.\n\nBLACK SUVs:\nlarge matching dark SUVs.\nConsistent shape and scale.\nNo branding.\n\n[ENVIRONMENT LOCK]\n\nThe chase moves through:\n\n1. desert-edge highway\n2. narrowing service road / construction route\n3. unfinished elevated overpass\n\nPreserve coherent geography.\n\nOVERPASS:\n- concrete barriers\n- construction barricades\n- exposed structural elements\n- orange safety lamps\n- distant city lights\n- dusty road\n- dramatic open skyline\n\nLighting progresses naturally:\n\nbright orange-magenta sunset\n→ deeper red-purple dusk\n→ cool blue night shadows with warm residual horizon light\n\n[WORLD / ACTION PHYSICS]\n\nAll vehicle and character motion should feel grounded and cinematic.\n\nVehicles must show:\n- weight\n- suspension movement\n- tire grip\n- braking momentum\n- body roll\n- road vibration\n- realistic acceleration\n\nNo weightless vehicles.\nNo flying cars.\nNo impossible spins unless physically motivated.\n\nAction follows:\nanticipation → attack → consequence → reaction.\n\nThe confrontation is brief and stylized.\nNo graphic violence.\nNo prolonged fight choreography.\n\nintegrated_multimodal_description:\n\n[SHOT 1] (0.00s–2.10s) — FULL-SPEED PURSUIT\n\nOpen immediately on Jax’s dark muscle car accelerating hard down a sunset highway.\n\nTwo black SUVs close from behind and another runs ahead.\n\nTires throw road grit.\nWarm sunset reflections streak across the car body.\n\nCut inside.\n\nJax grips the wheel firmly, eyes locked forward.\n\nLena braces in the passenger seat, twists around, and tracks the lead SUV through the side and rear windows.\n\nCamera:\nStart with a low rear three-quarter tracking shot beside the muscle car.\nSweep into an aggressive side tracking angle.\nCut to a tight interior two-shot.\n\nCamera speed:\nfast but readable.\n\nEmotion:\nfocused urgency.\n\nDialogue:\n\nLena (sharp):\n<d>[English] Lead truck. Eleven o’clock.</d>\n\nJax (calm):\n<d>[English] Got it.</d>\n\nLena:\n<d>[English] Don’t lose him.</d>\n\nJax:\n<d>[English] Wasn’t planning to.</d>\n\nSound:\n- deep muscle-car engine\n- SUV engines behind\n- tire noise\n- road rumble\n- rushing wind\n- seat and clothing movement\n- low rhythmic action pulse\n\nTransition:\nOne SUV moves alongside and begins closing the gap.\n\n[SHOT 2] (2.10s–4.10s) — THE BOX-IN\n\nTwo black SUVs attempt to compress Jax’s car between lanes.\n\nLena leans toward the passenger window and keeps her eyes fixed on the lead SUV.\n\nHer ponytail and red bandana react naturally to airflow.\n\nJax sees a narrow service lane opening ahead.\n\nHe makes one decisive steering move and sends the muscle car through the gap before the SUVs can close it.\n\nThe car drops onto the service lane with a heavy suspension compression.\n\nOne pursuing SUV follows too aggressively.\n\nCamera:\nHigh rear chase angle showing the box-in formation.\nCut to Lena’s profile tracking the target.\nThen low front-quarter shot as Jax cuts across into the service lane.\n\nEmotion:\npressure → opportunity.\n\nDialogue:\n\nLena:\n<d>[English] They’re boxing us in.</d>\n\nJax:\n<d>[English] Not anymore.</d>\n\nSound:\n- engines climbing\n- tire scrub\n- brief horn\n- gravel spray\n- suspension thump\n- quick percussion hit as Jax breaks formation\n\nTransition:\nThe pursuing SUV enters the service road too fast.\n\n[SHOT 3] (4.10s–6.00s) — CONVOY BREAKS\n\nThe pursuing SUV clips a lightweight construction barrier.\n\nThe barrier spins aside.\n\nThe SUV loses alignment and skids safely into additional barricades, sending sparks and dust sideways.\n\nNo people are visibly injured.\n\nJax’s muscle car continues forward.\n\nAhead, the road rises onto an unfinished elevated overpass.\n\nThe lead SUV carrying Rico accelerates toward it.\n\nCamera:\nDynamic low roadside angle as SUV loses control.\nBrief wide shot of the impact.\nThen long-lens chase composition revealing the overpass ahead.\n\nEmotion:\nmomentary advantage but increasing danger.\n\nDialogue:\n\nLena:\n<d>[English] One down.</d>\n\nJax:\n<d>[English] Don’t celebrate yet.</d>\n\nSound:\n- tire squeal\n- construction barrel impacts\n- metal scrape\n- restrained sparks\n- engine acceleration\n- wind intensifying\n\nMusic:\nadd stronger percussion and low synth pulse.\n\nTransition:\nJax’s car climbs the overpass ramp behind Rico.\n\n[SHOT 4] (6.00s–8.00s) — THE ROAD ENDS\n\nWide cinematic shot of the unfinished overpass suspended above distant city lights.\n\nOnly the lead SUV and Jax’s muscle car remain in the immediate chase.\n\nRico’s SUV swerves past construction barricades.\n\nA final sharp turn forces the SUV to skid sideways near the unfinished end of the overpass.\n\nDust rises.\n\nJax brakes hard and stops several car lengths away.\n\nThe muscle car settles on its suspension.\n\nThe engines idle.\n\nSuddenly the chase becomes quiet.\n\nCamera:\nAerial-like wide establishing shot over the overpass.\nFast tracking behind both vehicles.\nThen controlled side shot of Rico’s SUV skidding.\nSlow push toward the two stopped cars.\n\nEmotion:\nspeed collapses into tense silence.\n\nDialogue:\n\nLena (quiet):\n<d>[English] End of the road.</d>\n\nJax:\n<d>[English] For somebody.</d>\n\nSound:\n- hard braking\n- tire squeal\n- gravel sliding\n- engine rumble falling to idle\n- distant wind\n- action music suddenly strips back to a low pulse\n\nTransition:\nRico’s driver door opens.\n\n[SHOT 5] (8.00s–10.10s) — RICO SHOWS HIS HAND\n\nRico steps from the SUV.\n\nHe casually holds the matte-black drive up between two fingers.\n\nIn his other hand is a generic stylized handgun held low and controlled.\n\nLena and Jax exit their muscle car and stand several meters away.\n\nNo one rushes.\n\nRico smiles.\n\nCamera:\nLow hero-villain angle on Rico stepping out.\nCut to over-the-shoulder from behind Lena and Jax.\nPush into the black drive.\nThen medium three-shot.\n\nEmotion:\nRico feels victorious.\nLena looks ready to move.\nJax remains unusually calm.\n\nDialogue:\n\nRico (smug):\n<d>[English] You two really came all this way for this?</d>\n\nLena:\n<d>[English] Hand it over.</d>\n\nRico (small laugh):\n<d>[English] You were never supposed to leave that station.</d>\n\nJax:\n<d>[English] And yet, here we are.</d>\n\nSound:\n- wind across unfinished concrete\n- distant city ambience\n- SUV cooling ticks\n- fabric movement\n- quiet ominous bass tone\n\nTransition:\nRico raises the drive slightly as if showing off his victory.\n\n[SHOT 6] (10.10s–12.10s) — JAX’S TWIST\n\nJax slowly takes out his small dark handheld device.\n\nThe display shows a simple red disabled-state warning.\n\nHe tilts it toward Rico.\n\nRico looks from the device to the black drive.\n\nHis smug expression disappears.\n\nLena’s eyes narrow with a small knowing smile.\n\nCamera:\nClose-up of Jax’s calm eyes.\nInsert on generic device screen.\nReaction close-up on Rico.\nCut to Lena.\n\nDialogue:\n\nRico:\n<d>[English] What did you do?</d>\n\nJax (calm):\n<d>[English] The drive was dead before you left the station.</d>\n\nRico:\n<d>[English] You’re bluffing.</d>\n\nJax:\n<d>[English] Try it.</d>\n\nSound:\n- subtle electronic confirmation tone\n- wind\n- low bass pulse\n- tiny musical reversal sting\n\nImportant:\nThe technology is fictional and should not depict real hacking procedures or operational instructions.\n\nTransition:\nRico instinctively looks down at the drive.\n\nThat moment creates the opening.\n\n[SHOT 7] (12.10s–13.60s) — THE OPENING\n\nLena suddenly flicks / tosses a harmless small loose object or handful of construction dust across Rico’s line of sight.\n\nRico reacts instinctively.\n\nJax immediately closes the distance.\n\nIn one short stylized motion, he redirects Rico’s weapon arm away and the pistol is knocked loose.\n\nThe pistol slides safely across the pavement away from everyone.\n\nLena moves past Rico and takes the black drive.\n\nRico ends safely on one knee or sitting against a construction barrier, surprised but visibly unharmed.\n\nDo not show detailed combat technique.\nKeep the movement fast, cinematic, and impressionistic.\n\nCamera:\nQuick lateral whip-follow toward Lena’s distraction.\nLow medium shot on Jax’s intervention.\nMacro insert of pistol sliding away.\nClose shot on Lena securing the drive.\n\nEmotion:\nprecise teamwork.\n\nDialogue:\n\nLena:\n<d>[English] Now.</d>\n\nJax:\n<d>[English] Done.</d>\n\nRico (frustrated):\n<d>[English] You think this changes anything?</d>\n\nSound:\n- quick object flick\n- cloth movement\n- short safe physical impact\n- pistol clatter and slide\n- drive grab\n- restrained cinematic hit\n\nNo gunshot.\n\nTransition:\nRed-blue light reflections appear far behind them.\n\n[SHOT 8] (13.60s–15.00s) — RECOVERY / HERO EXIT\n\nLena stands holding the recovered black drive.\n\nJax joins her.\n\nRico remains safely behind them, disarmed.\n\nFar down the overpass approach, faint police lights flash in the distance.\n\nLena and Jax calmly walk toward their waiting muscle car.\n\nThe last orange strip of sunset sits beneath a deep violet-blue sky.\n\nLena looks down at the drive, then toward Jax.\n\nCamera:\nBegin medium on Lena holding the drive.\nSlow lateral move as Jax joins her.\nThen transition into a wide rear hero shot as they walk toward the muscle car.\n\nFinal composition:\nLena and Jax walking side by side.\nMuscle car ahead.\nRed taillights glowing.\nCity beyond.\nPolice lights distant.\nRico left behind safely.\nBlack drive visible in Lena’s hand.\n\nDialogue:\n\nLena:\n<d>[English] So what’s actually on this thing?</d>\n\nJax (without turning):\n<d>[English] Enough to burn his empire.</d>\n\nLena gives a small satisfied look.\n\nSound:\n- distant sirens\n- soft wind\n- slow footsteps on concrete\n- car engine idling\n- fabric movement\n- low triumphant musical rise\n\nFinal musical beat lands precisely after Jax’s last line.\n\nCUT TO BLACK.\n\naudio_description:\n\nGenerate synchronized native stereo audio.\n\n[CHASE ENVIRONMENT]\n\n- powerful muscle-car engine\n- heavier SUV engines\n- realistic tires\n- road vibration\n- wind rush\n- suspension compression\n- gravel\n- construction barrier impacts\n- restrained metal scrapes\n\nDo not exaggerate every impact.\n\n[OVERPASS ENVIRONMENT]\n\n- high-altitude wind\n- distant city hum\n- cooling engines\n- occasional construction material rattle\n- subtle open-space reverb\n\n[CHARACTER VOICES]\n\nLENA:\nyoung adult female voice.\nLow-to-medium register.\nDirect.\nConfident.\nFast during chase.\nVery controlled during confrontation.\nNo excessive shouting.\n\nJAX:\nyoung adult male.\nLow calm voice.\nTactical.\nUnderstated.\nRarely raises volume.\nFinal line delivered coolly and quietly.\n\nRICO:\nadult male.\nSmooth confident voice.\nSmug at first.\nBrief frustration after the disabled-drive reveal.\nNever hysterical.\n\n[NON-DIEGETIC MUSIC]\n\n0.00–4.10:\nfast restrained electronic-action pulse with low bass and tight percussion.\n\n4.10–6.00:\nadd sharper rhythmic accents during SUV spin-out.\n\n6.00–8.00:\nmusic reaches peak chase intensity, then rapidly strips down as vehicles stop.\n\n8.00–10.10:\nminimal tense bass pulse beneath Rico confrontation.\n\n10.10–12.10:\nsubtle rising tension with electronic reversal accent during Jax’s reveal.\n\n12.10–13.60:\none fast controlled action hit.\n\n13.60–15.00:\nwarm low cinematic victory motif, restrained and cool rather than heroic bombast.\n\nMusic must remain below dialogue.\n\n[VISUAL STYLE]\n\nPremium cinematic 2D graphic-novel animation.\n\nVisual qualities:\n- refined ink outlines\n- painterly cel shading\n- mature Western graphic-novel aesthetic\n- realistic stylized anatomy\n- strong silhouettes\n- detailed cars\n- cinematic road environments\n- controlled motion blur\n- dramatic rim lighting\n- filmic contrast\n- rich atmospheric perspective\n- subtle texture\n- expressive eyes\n- consistent faces\n\nDo not make the final footage look like storyboard drawings on paper.\n\nIt should look like the storyboard has become a finished high-budget animated action sequence.\n\n[COLOR / LIGHTING PROGRESSION]\n\n0.00–4.00:\nstrong orange-magenta sunset,\nwarm body highlights,\ncool asphalt shadows.\n\n4.00–8.00:\ndeeper violet sunset,\nred taillights,\norange sparks,\nblue-gray road surfaces.\n\n8.00–12.00:\nsun nearly gone,\ndeep purple sky,\nwarm horizon rim light,\ncool shadows around the characters.\n\n12.00–15.00:\nblue-violet early night,\nremaining orange horizon,\ndistant red-blue police reflections,\nwarm red muscle-car taillights.\n\nKeep skin tones natural within the stylized palette.\n\n[CAMERA RULES]\n\nCamera energy progression:\n\nCHASE:\n- low rear tracking\n- side tracking\n- high chase angle\n- interior close-ups\n- long-lens compression\n- dynamic but readable road perspectives\n\nOVERPASS:\n- large environmental wides\n- controlled brake/skid tracking\n\nSTANDOFF:\n- slow push-ins\n- over-the-shoulders\n- clean medium compositions\n- reaction close-ups\n\nACTION BEAT:\n- short dynamic lateral move\n- one quick controlled whip-follow\n- no chaotic cutting\n\nENDING:\n- steady wide hero shot\n- slow pull-back or gentle lateral drift\n\nAvoid:\n- excessive Dutch angles\n- random spinning\n- extreme handheld shake\n- repetitive zooms\n- confusing POV changes\n- unreadable cuts\n\n[MOTION / PHYSICS RULES]\n\nCars:\nmust feel heavy and grounded.\n\nSUV losing control:\none physically understandable skid into lightweight construction barricades.\nNo vehicle explosion.\nNo occupants shown injured.\n\nPonytail and bandanas:\nreact naturally to vehicle airflow and wind.\n\nRico’s pistol:\nmust never fire.\nWhen disarmed, it clearly falls away from all characters and remains on the ground.\n\nRico:\nmust remain visibly safe after the confrontation.\n\n[CONTINUITY RULES]\n\nMaintain perfect continuity from Part 1.\n\nLena:\nsame face,\nred bandana,\ndark cropped shirt,\ndenim shorts,\nhigh ponytail.\n\nJax:\nsame face,\nblue/dark bandana,\ndark shirt,\nbrown/dark pants.\n\nRico:\nsame face and outfit from his first appearance through final shot.\n\nBlack drive:\nsame matte-black rectangular design.\n\nMuscle car:\nsame dark body and red rear light treatment.\n\nSUVs:\nsame consistent dark styling.\n\nDo not regenerate or change vehicles between shots.\n\nKeep environmental geography coherent:\nhighway → service lane → overpass ramp → unfinished overpass confrontation.\n\nDo not teleport characters or vehicles.\n\n[SAFETY / NEGATIVE CONSTRAINTS]\n\nNo blood.\nNo gore.\nNo wounds.\nNo graphic injuries.\nNo executions.\nNo torture.\nNo cruelty.\nNo civilians harmed.\nNo explosions.\nNo gunshots.\nNo bullet impacts.\nNo detailed weapon handling instruction.\nNo prolonged combat choreography.\nNo realistic depiction of how to disarm a firearm.\nKeep the intervention fast, abstract, cinematic, and non-instructional.\n\nNo storyboard page visible.\nNo panel borders.\nNo timestamps.\nNo captions.\nNo production notes.\nNo subtitles.\nNo title cards.\nNo logos.\nNo brand names.\nNo watermarks.\nNo real-world vehicle badges.\nNo copyrighted characters.\nNo recognizable game characters.\nNo extra protagonists.\nNo random henchmen appearing during final confrontation.\nNo duplicated Lena.\nNo duplicated Jax.\nNo duplicate Rico.\nNo face changes.\nNo hairstyle changes.\nNo wardrobe changes.\nNo missing bandanas.\nNo inconsistent drive shape.\nNo malformed hands.\nNo extra limbs.\nNo distorted vehicles.\nNo teleporting cars.\nNo random environmental changes.\nNo hyperreal live action.\nNo 3D videogame-render appearance.\nNo anime transformation effects.\n\nFINAL FRAME:\n\nLena and Jax walk toward their dark muscle car on the unfinished overpass.\n\nLena holds the recovered black drive.\n\nThe red taillights glow ahead.\n\nFaint police lights flash far behind.\n\nThe last orange light disappears below the horizon.\n\nJax has just said:\n\n<d>[English] Enough to burn his empire.</d>\n\nHold the cinematic hero composition briefly.\n\nCUT TO BLACK.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085607833408852170",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085607833408852170",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-07T06:03:50.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085607833408852170.mp4",
    "posterUrl": "/posters/x/2085607833408852170.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Storyboard 2 to Video using MiniMax H3 on @higgsfield \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image as the PRIMARY visual, character, prop, environment, camera, lighting, and narrative reference for the entire video.\n\nUse the storyboard to preserve:\n- exact Part 2 story progression\n- Lena Cruz and Jax Mercer identity continuity from Part 1\n- Rico Vale’s appearance and attitude\n- the dark muscle car\n- black SUV convoy\n- black drive\n- unfinished overpass environment\n- sunset-to-night lighting progression\n- graphic-novel cinematic style\n- chase escalation\n- standoff\n- Jax’s “disabled drive” reveal\n- Lena and Jax recovering the drive\n- final victorious walk-away\n\nDo NOT reproduce the storyboard sheet itself.\n\nThe finished video must contain:\n- no panel borders\n- no storyboard numbers\n- no timestamps\n- no captions\n- no production notes\n- no paper background\n- no reference-sheet layout\n\nConvert the entire board into one coherent finished cinematic animated sequence.\n\n[VIDEO GOAL]\n\nDuration: exactly 15 seconds\nAspect ratio: 16:9 landscape\nVideo style: premium cinematic 2D graphic-novel animation\nGenre: Hollywood-style action thriller / road chase\nTone: fast, stylish, tense, controlled, cinematic, triumphant\n\nNarrative progression:\n\nhigh-speed pursuit\n→ convoy closes in\n→ Jax breaks formation\n→ unfinished overpass\n→ convoy collapses\n→ Rico standoff\n→ disabled-drive twist\n→ brief safe disarm\n→ drive recovered\n→ cool final exit\n\nThis is PART 2 only.\n\nBegin immediately with the chase already underway.\n\nDo NOT repeat the gas-station setup from Part 1.\n\nEnd exactly after Lena and Jax recover the black drive and walk toward their muscle car while Jax delivers the final line:\n\n“Enough to burn his empire.”\n\n[CHARACTER IDENTITY LOCKS]\n\nLENA CRUZ\n\nYoung athletic woman.\n\nAppearance:\n- warm medium skin\n- sharp expressive eyes\n- high dark ponytail\n- red patterned bandana over lower face or around neck\n- fitted dark cropped top\n- light denim shorts\n- practical dark footwear\n- lean athletic proportions\n- confident silhouette\n\nPersonality:\nbold, reactive, fearless, intuitive, aggressive under pressure but controlled.\n\nBeginning emotion:\nfocused pursuit intensity.\n\nEnding emotion:\nquiet satisfaction and confidence.\n\nMovement:\nfast head turns,\nathletic body control,\ndecisive reactions,\nstrong braced posture during the car chase,\nquick efficient movements during confrontation.\n\nKeep identical in every shot:\n- face\n- skin tone\n- ponytail\n- bandana\n- clothing\n- height\n- body proportions\n- age\n- eye shape\n\nJAX MERCER\n\nTall athletic young man.\n\nAppearance:\n- short dark hair\n- intense eyes\n- dark blue patterned bandana covering lower face or around neck\n- dark fitted T-shirt\n- brown / charcoal pants\n- strong controlled physique\n\nPersonality:\ncalm strategist,\nprecise,\nrarely panics,\ntactical,\nprotective.\n\nBeginning emotion:\nextreme concentration.\n\nEnding emotion:\ncontrolled victory.\n\nMovement:\nefficient driving,\nminimal wasted motion,\ncalm body language even during danger.\n\nKeep identical:\n- face\n- hairstyle\n- blue bandana\n- shirt\n- pants\n- height\n- proportions\n- body build\n\nRICO VALE\n\nOriginal crime boss character.\n\nAppearance:\n- stylish dark hair\n- confident narrow eyes\n- clean or lightly stubbled face\n- dark tailored street jacket\n- dark shirt\n- restrained jewelry\n- smug posture\n- sophisticated silhouette\n\nPersonality:\narrogant,\ncalculated,\nconfident,\nbelieves he has already won.\n\nBeginning emotion:\nsmug control.\n\nEnding emotion:\nsurprised and outplayed, but physically unharmed.\n\nDo not make Rico grotesque, psychotic, or horror-like.\n\n[PROP LOCKS]\n\nBLACK DRIVE:\nsmall matte-black rectangular data device.\nSame exact shape and size throughout.\nIt must remain visually readable in Rico’s hand and later Lena’s hand.\n\nJAX’S DEVICE:\nsmall generic dark handheld screen / phone.\nDuring the twist, it displays a simple red warning-style state indicating the drive is disabled.\nAvoid recognizable real-world phone branding.\n\nRICO’S PISTOL:\nstylized generic dark handgun.\nOnly used as a visual standoff prop.\nNo firing.\nNo bullets.\nNo injuries.\nIt is safely knocked away during the confrontation.\n\nMUSCLE CAR:\nsame dark charcoal muscle car from Part 1.\nWide stance.\nLong red rear light treatment.\nStrong low silhouette.\nNo real-world logos.\n\nBLACK SUVs:\nlarge matching dark SUVs.\nConsistent shape and scale.\nNo branding.\n\n[ENVIRONMENT LOCK]\n\nThe chase moves through:\n\n1. desert-edge highway\n2. narrowing service road / construction route\n3. unfinished elevated overpass\n\nPreserve coherent geography.\n\nOVERPASS:\n- concrete barriers\n- construction barricades\n- exposed structural elements\n- orange safety lamps\n- distant city lights\n- dusty road\n- dramatic open skyline\n\nLighting progresses naturally:\n\nbright orange-magenta sunset\n→ deeper red-purple dusk\n→ cool blue night shadows with warm residual horizon light\n\n[WORLD / ACTION PHYSICS]\n\nAll vehicle and character motion should feel grounded and cinematic.\n\nVehicles must show:\n- weight\n- suspension movement\n- tire grip\n- braking momentum\n- body roll\n- road vibration\n- realistic acceleration\n\nNo weightless vehicles.\nNo flying cars.\nNo impossible spins unless physically motivated.\n\nAction follows:\nanticipation → attack → consequence → reaction.\n\nThe confrontation is brief and stylized.\nNo graphic violence.\nNo prolonged fight choreography.\n\nintegrated_multimodal_description:\n\n[SHOT 1] (0.00s–2.10s) — FULL-SPEED PURSUIT\n\nOpen immediately on Jax’s dark muscle car accelerating hard down a sunset highway.\n\nTwo black SUVs close from behind and another runs ahead.\n\nTires throw road grit.\nWarm sunset reflections streak across the car body.\n\nCut inside.\n\nJax grips the wheel firmly, eyes locked forward.\n\nLena braces in the passenger seat, twists around, and tracks the lead SUV through the side and rear windows.\n\nCamera:\nStart with a low rear three-quarter tracking shot beside the muscle car.\nSweep into an aggressive side tracking angle.\nCut to a tight interior two-shot.\n\nCamera speed:\nfast but readable.\n\nEmotion:\nfocused urgency.\n\nDialogue:\n\nLena (sharp):\n<d>[English] Lead truck. Eleven o’clock.</d>\n\nJax (calm):\n<d>[English] Got it.</d>\n\nLena:\n<d>[English] Don’t lose him.</d>\n\nJax:\n<d>[English] Wasn’t planning to.</d>\n\nSound:\n- deep muscle-car engine\n- SUV engines behind\n- tire noise\n- road rumble\n- rushing wind\n- seat and clothing movement\n- low rhythmic action pulse\n\nTransition:\nOne SUV moves alongside and begins closing the gap.\n\n[SHOT 2] (2.10s–4.10s) — THE BOX-IN\n\nTwo black SUVs attempt to compress Jax’s car between lanes.\n\nLena leans toward the passenger window and keeps her eyes fixed on the lead SUV.\n\nHer ponytail and red bandana react naturally to airflow.\n\nJax sees a narrow service lane opening ahead.\n\nHe makes one decisive steering move and sends the muscle car through the gap before the SUVs can close it.\n\nThe car drops onto the service lane with a heavy suspension compression.\n\nOne pursuing SUV follows too aggressively.\n\nCamera:\nHigh rear chase angle showing the box-in formation.\nCut to Lena’s profile tracking the target.\nThen low front-quarter shot as Jax cuts across into the service lane.\n\nEmotion:\npressure → opportunity.\n\nDialogue:\n\nLena:\n<d>[English] They’re boxing us in.</d>\n\nJax:\n<d>[English] Not anymore.</d>\n\nSound:\n- engines climbing\n- tire scrub\n- brief horn\n- gravel spray\n- suspension thump\n- quick percussion hit as Jax breaks formation\n\nTransition:\nThe pursuing SUV enters the service road too fast.\n\n[SHOT 3] (4.10s–6.00s) — CONVOY BREAKS\n\nThe pursuing SUV clips a lightweight construction barrier.\n\nThe barrier spins aside.\n\nThe SUV loses alignment and skids safely into additional barricades, sending sparks and dust sideways.\n\nNo people are visibly injured.\n\nJax’s muscle car continues forward.\n\nAhead, the road rises onto an unfinished elevated overpass.\n\nThe lead SUV carrying Rico accelerates toward it.\n\nCamera:\nDynamic low roadside angle as SUV loses control.\nBrief wide shot of the impact.\nThen long-lens chase composition revealing the overpass ahead.\n\nEmotion:\nmomentary advantage but increasing danger.\n\nDialogue:\n\nLena:\n<d>[English] One down.</d>\n\nJax:\n<d>[English] Don’t celebrate yet.</d>\n\nSound:\n- tire squeal\n- construction barrel impacts\n- metal scrape\n- restrained sparks\n- engine acceleration\n- wind intensifying\n\nMusic:\nadd stronger percussion and low synth pulse.\n\nTransition:\nJax’s car climbs the overpass ramp behind Rico.\n\n[SHOT 4] (6.00s–8.00s) — THE ROAD ENDS\n\nWide cinematic shot of the unfinished overpass suspended above distant city lights.\n\nOnly the lead SUV and Jax’s muscle car remain in the immediate chase.\n\nRico’s SUV swerves past construction barricades.\n\nA final sharp turn forces the SUV to skid sideways near the unfinished end of the overpass.\n\nDust rises.\n\nJax brakes hard and stops several car lengths away.\n\nThe muscle car settles on its suspension.\n\nThe engines idle.\n\nSuddenly the chase becomes quiet.\n\nCamera:\nAerial-like wide establishing shot over the overpass.\nFast tracking behind both vehicles.\nThen controlled side shot of Rico’s SUV skidding.\nSlow push toward the two stopped cars.\n\nEmotion:\nspeed collapses into tense silence.\n\nDialogue:\n\nLena (quiet):\n<d>[English] End of the road.</d>\n\nJax:\n<d>[English] For somebody.</d>\n\nSound:\n- hard braking\n- tire squeal\n- gravel sliding\n- engine rumble falling to idle\n- distant wind\n- action music suddenly strips back to a low pulse\n\nTransition:\nRico’s driver door opens.\n\n[SHOT 5] (8.00s–10.10s) — RICO SHOWS HIS HAND\n\nRico steps from the SUV.\n\nHe casually holds the matte-black drive up between two fingers.\n\nIn his other hand is a generic stylized handgun held low and controlled.\n\nLena and Jax exit their muscle car and stand several meters away.\n\nNo one rushes.\n\nRico smiles.\n\nCamera:\nLow hero-villain angle on Rico stepping out.\nCut to over-the-shoulder from behind Lena and Jax.\nPush into the black drive.\nThen medium three-shot.\n\nEmotion:\nRico feels victorious.\nLena looks ready to move.\nJax remains unusually calm.\n\nDialogue:\n\nRico (smug):\n<d>[English] You two really came all this way for this?</d>\n\nLena:\n<d>[English] Hand it over.</d>\n\nRico (small laugh):\n<d>[English] You were never supposed to leave that station.</d>\n\nJax:\n<d>[English] And yet, here we are.</d>\n\nSound:\n- wind across unfinished concrete\n- distant city ambience\n- SUV cooling ticks\n- fabric movement\n- quiet ominous bass tone\n\nTransition:\nRico raises the drive slightly as if showing off his victory.\n\n[SHOT 6] (10.10s–12.10s) — JAX’S TWIST\n\nJax slowly takes out his small dark handheld device.\n\nThe display shows a simple red disabled-state warning.\n\nHe tilts it toward Rico.\n\nRico looks from the device to the black drive.\n\nHis smug expression disappears.\n\nLena’s eyes narrow with a small knowing smile.\n\nCamera:\nClose-up of Jax’s calm eyes.\nInsert on generic device screen.\nReaction close-up on Rico.\nCut to Lena.\n\nDialogue:\n\nRico:\n<d>[English] What did you do?</d>\n\nJax (calm):\n<d>[English] The drive was dead before you left the station.</d>\n\nRico:\n<d>[English] You’re bluffing.</d>\n\nJax:\n<d>[English] Try it.</d>\n\nSound:\n- subtle electronic confirmation tone\n- wind\n- low bass pulse\n- tiny musical reversal sting\n\nImportant:\nThe technology is fictional and should not depict real hacking procedures or operational instructions.\n\nTransition:\nRico instinctively looks down at the drive.\n\nThat moment creates the opening.\n\n[SHOT 7] (12.10s–13.60s) — THE OPENING\n\nLena suddenly flicks / tosses a harmless small loose object or handful of construction dust across Rico’s line of sight.\n\nRico reacts instinctively.\n\nJax immediately closes the distance.\n\nIn one short stylized motion, he redirects Rico’s weapon arm away and the pistol is knocked loose.\n\nThe pistol slides safely across the pavement away from everyone.\n\nLena moves past Rico and takes the black drive.\n\nRico ends safely on one knee or sitting against a construction barrier, surprised but visibly unharmed.\n\nDo not show detailed combat technique.\nKeep the movement fast, cinematic, and impressionistic.\n\nCamera:\nQuick lateral whip-follow toward Lena’s distraction.\nLow medium shot on Jax’s intervention.\nMacro insert of pistol sliding away.\nClose shot on Lena securing the drive.\n\nEmotion:\nprecise teamwork.\n\nDialogue:\n\nLena:\n<d>[English] Now.</d>\n\nJax:\n<d>[English] Done.</d>\n\nRico (frustrated):\n<d>[English] You think this changes anything?</d>\n\nSound:\n- quick object flick\n- cloth movement\n- short safe physical impact\n- pistol clatter and slide\n- drive grab\n- restrained cinematic hit\n\nNo gunshot.\n\nTransition:\nRed-blue light reflections appear far behind them.\n\n[SHOT 8] (13.60s–15.00s) — RECOVERY / HERO EXIT\n\nLena stands holding the recovered black drive.\n\nJax joins her.\n\nRico remains safely behind them, disarmed.\n\nFar down the overpass approach, faint police lights flash in the distance.\n\nLena and Jax calmly walk toward their waiting muscle car.\n\nThe last orange strip of sunset sits beneath a deep violet-blue sky.\n\nLena looks down at the drive, then toward Jax.\n\nCamera:\nBegin medium on Lena holding the drive.\nSlow lateral move as Jax joins her.\nThen transition into a wide rear hero shot as they walk toward the muscle car.\n\nFinal composition:\nLena and Jax walking side by side.\nMuscle car ahead.\nRed taillights glowing.\nCity beyond.\nPolice lights distant.\nRico left behind safely.\nBlack drive visible in Lena’s hand.\n\nDialogue:\n\nLena:\n<d>[English] So what’s actually on this thing?</d>\n\nJax (without turning):\n<d>[English] Enough to burn his empire.</d>\n\nLena gives a small satisfied look.\n\nSound:\n- distant sirens\n- soft wind\n- slow footsteps on concrete\n- car engine idling\n- fabric movement\n- low triumphant musical rise\n\nFinal musical beat lands precisely after Jax’s last line.\n\nCUT TO BLACK.\n\naudio_description:\n\nGenerate synchronized native stereo audio.\n\n[CHASE ENVIRONMENT]\n\n- powerful muscle-car engine\n- heavier SUV engines\n- realistic tires\n- road vibration\n- wind rush\n- suspension compression\n- gravel\n- construction barrier impacts\n- restrained metal scrapes\n\nDo not exaggerate every impact.\n\n[OVERPASS ENVIRONMENT]\n\n- high-altitude wind\n- distant city hum\n- cooling engines\n- occasional construction material rattle\n- subtle open-space reverb\n\n[CHARACTER VOICES]\n\nLENA:\nyoung adult female voice.\nLow-to-medium register.\nDirect.\nConfident.\nFast during chase.\nVery controlled during confrontation.\nNo excessive shouting.\n\nJAX:\nyoung adult male.\nLow calm voice.\nTactical.\nUnderstated.\nRarely raises volume.\nFinal line delivered coolly and quietly.\n\nRICO:\nadult male.\nSmooth confident voice.\nSmug at first.\nBrief frustration after the disabled-drive reveal.\nNever hysterical.\n\n[NON-DIEGETIC MUSIC]\n\n0.00–4.10:\nfast restrained electronic-action pulse with low bass and tight percussion.\n\n4.10–6.00:\nadd sharper rhythmic accents during SUV spin-out.\n\n6.00–8.00:\nmusic reaches peak chase intensity, then rapidly strips down as vehicles stop.\n\n8.00–10.10:\nminimal tense bass pulse beneath Rico confrontation.\n\n10.10–12.10:\nsubtle rising tension with electronic reversal accent during Jax’s reveal.\n\n12.10–13.60:\none fast controlled action hit.\n\n13.60–15.00:\nwarm low cinematic victory motif, restrained and cool rather than heroic bombast.\n\nMusic must remain below dialogue.\n\n[VISUAL STYLE]\n\nPremium cinematic 2D graphic-novel animation.\n\nVisual qualities:\n- refined ink outlines\n- painterly cel shading\n- mature Western graphic-novel aesthetic\n- realistic stylized anatomy\n- strong silhouettes\n- detailed cars\n- cinematic road environments\n- controlled motion blur\n- dramatic rim lighting\n- filmic contrast\n- rich atmospheric perspective\n- subtle texture\n- expressive eyes\n- consistent faces\n\nDo not make the final footage look like storyboard drawings on paper.\n\nIt should look like the storyboard has become a finished high-budget animated action sequence.\n\n[COLOR / LIGHTING PROGRESSION]\n\n0.00–4.00:\nstrong orange-magenta sunset,\nwarm body highlights,\ncool asphalt shadows.\n\n4.00–8.00:\ndeeper violet sunset,\nred taillights,\norange sparks,\nblue-gray road surfaces.\n\n8.00–12.00:\nsun nearly gone,\ndeep purple sky,\nwarm horizon rim light,\ncool shadows around the characters.\n\n12.00–15.00:\nblue-violet early night,\nremaining orange horizon,\ndistant red-blue police reflections,\nwarm red muscle-car taillights.\n\nKeep skin tones natural within the stylized palette.\n\n[CAMERA RULES]\n\nCamera energy progression:\n\nCHASE:\n- low rear tracking\n- side tracking\n- high chase angle\n- interior close-ups\n- long-lens compression\n- dynamic but readable road perspectives\n\nOVERPASS:\n- large environmental wides\n- controlled brake/skid tracking\n\nSTANDOFF:\n- slow push-ins\n- over-the-shoulders\n- clean medium compositions\n- reaction close-ups\n\nACTION BEAT:\n- short dynamic lateral move\n- one quick controlled whip-follow\n- no chaotic cutting\n\nENDING:\n- steady wide hero shot\n- slow pull-back or gentle lateral drift\n\nAvoid:\n- excessive Dutch angles\n- random spinning\n- extreme handheld shake\n- repetitive zooms\n- confusing POV changes\n- unreadable cuts\n\n[MOTION / PHYSICS RULES]\n\nCars:\nmust feel heavy and grounded.\n\nSUV losing control:\none physically understandable skid into lightweight construction barricades.\nNo vehicle explosion.\nNo occupants shown injured.\n\nPonytail and bandanas:\nreact naturally to vehicle airflow and wind.\n\nRico’s pistol:\nmust never fire.\nWhen disarmed, it clearly falls away from all characters and remains on the ground.\n\nRico:\nmust remain visibly safe after the confrontation.\n\n[CONTINUITY RULES]\n\nMaintain perfect continuity from Part 1.\n\nLena:\nsame face,\nred bandana,\ndark cropped shirt,\ndenim shorts,\nhigh ponytail.\n\nJax:\nsame face,\nblue/dark bandana,\ndark shirt,\nbrown/dark pants.\n\nRico:\nsame face and outfit from his first appearance through final shot.\n\nBlack drive:\nsame matte-black rectangular design.\n\nMuscle car:\nsame dark body and red rear light treatment.\n\nSUVs:\nsame consistent dark styling.\n\nDo not regenerate or change vehicles between shots.\n\nKeep environmental geography coherent:\nhighway → service lane → overpass ramp → unfinished overpass confrontation.\n\nDo not teleport characters or vehicles.\n\n[SAFETY / NEGATIVE CONSTRAINTS]\n\nNo blood.\nNo gore.\nNo wounds.\nNo graphic injuries.\nNo executions.\nNo torture.\nNo cruelty.\nNo civilians harmed.\nNo explosions.\nNo gunshots.\nNo bullet impacts.\nNo detailed weapon handling instruction.\nNo prolonged combat choreography.\nNo realistic depiction of how to disarm a firearm.\nKeep the intervention fast, abstract, cinematic, and non-instructional.\n\nNo storyboard page visible.\nNo panel borders.\nNo timestamps.\nNo captions.\nNo production notes.\nNo subtitles.\nNo title cards.\nNo logos.\nNo brand names.\nNo watermarks.\nNo real-world vehicle badges.\nNo copyrighted characters.\nNo recognizable game characters.\nNo extra protagonists.\nNo random henchmen appearing during final confrontation.\nNo duplicated Lena.\nNo duplicated Jax.\nNo duplicate Rico.\nNo face changes.\nNo hairstyle changes.\nNo wardrobe changes.\nNo missing bandanas.\nNo inconsistent drive shape.\nNo malformed hands.\nNo extra limbs.\nNo distorted vehicles.\nNo teleporting cars.\nNo random environmental changes.\nNo hyperreal live action.\nNo 3D videogame-render appearance.\nNo anime transformation effects.\n\nFINAL FRAME:\n\nLena and Jax walk toward their dark muscle car on the unfinished overpass.\n\nLena holds the recovered black drive.\n\nThe red taillights glow ahead.\n\nFaint police lights flash far behind.\n\nThe last orange light disappears below the horizon.\n\nJax has just said:\n\n<d>[English] Enough to burn his empire.</d>\n\nHold the cinematic hero composition briefly.\n\nCUT TO BLACK.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 1589,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085606945755693556",
    "title": "加油站逃亡故事·上篇",
    "titleEn": "The Gas-Station Escape, Part One",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "加油站逃亡故事·上篇的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the gas-station escape, part one on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded Part 1 storyboard image as the PRIMARY visual and narrative reference for the entire video.\n\nUse it to preserve:\n- exact story progression\n- Lena Cruz and Jax Mercer character identity\n- wardrobe and bandana colors\n- gas-station environment\n- cashier appearance\n- paper bag and “TOO SLOW.” note\n- black drive design\n- black SUV convoy\n- dark muscle car\n- sunset color palette\n- cinematic graphic-novel visual language\n- suspense-to-action escalation\n- Part 1 cliffhanger ending\n\nDo NOT reproduce the storyboard sheet itself in the finished video.\nNo panel borders, timestamps, handwritten captions, production notes, reference-sheet layout, paper texture, storyboard numbering, or storyboard text should appear in the video.\n\nTransform the storyboard into one continuous polished cinematic animated action sequence.\n\n[VIDEO GOAL]\n\nDuration: exactly 15 seconds\nAspect ratio: 16:9 landscape\nStyle: premium cinematic 2D graphic-novel animation\nGenre: action thriller / crime chase\nTone: stylish, tense, fast, mysterious, controlled\nStory structure:\nquiet pickup → suspicious exchange → betrayal reveal → convoy trap → rapid escape → real drive spotted → chase begins\n\nThis is PART 1 only.\n\nDo NOT continue into the highway chase or overpass confrontation.\nThe exact final story beat is Lena realizing:\n“They still have it.”\n\n[CHARACTER IDENTITY LOCK]\n\nLENA CRUZ:\nYoung athletic woman with a high dark ponytail, sharp expressive eyes, red patterned bandana covering or resting around her lower face, fitted dark cropped top, light denim shorts, practical shoes, lean athletic proportions.\n\nPersonality:\nbold, observant, quick-reacting, fearless, direct.\n\nBeginning emotional state:\ncontrolled suspicion.\n\nEnding emotional state:\nfocused determination.\n\nKeep identical throughout:\n- facial structure\n- skin tone\n- ponytail\n- red bandana\n- dark cropped top\n- denim shorts\n- body proportions\n- age\n- eye shape\n- silhouette\n\nDo not change her hairstyle, outfit, facial proportions, or bandana pattern.\n\nJAX MERCER:\nTall athletic young man with short dark hair, intense eyes, blue/dark patterned bandana covering his lower face, fitted dark charcoal T-shirt, brown/dark pants, strong controlled proportions.\n\nPersonality:\ncalm, tactical, observant, protective, efficient.\n\nBeginning emotional state:\nquiet vigilance.\n\nEnding emotional state:\nfocused pursuit mode.\n\nKeep identical throughout:\n- face\n- short dark hairstyle\n- blue bandana\n- charcoal shirt\n- dark pants\n- build\n- height\n- proportions\n\nCASHIER:\nOrdinary nervous convenience-store employee.\nDark slightly messy hair, worried expression, simple work clothes.\nHe is frightened and uncomfortable, not villainous.\n\nRICO’S CREW:\nMostly seen through silhouettes, vehicle interiors, distant figures and brief readable glimpses.\nDark streetwear.\nNo unnecessary close-ups or extra character identities.\n\n[KEY PROP LOCK]\n\nBLACK DRIVE:\nSmall matte-black rectangular data device.\nSimple clean shape.\nThe REAL black drive is NOT inside the paper bag.\nIt is seen clearly later in the hand of Rico’s crew near the lead SUV.\n\nFAKE PACKAGE:\nPlain brown paper bag.\n\nInside:\n- cheap small green novelty keychain\n- folded paper note\n\nThe note must clearly communicate:\n“TOO SLOW.”\n\nDo not introduce additional items.\n\nMUSCLE CAR:\nDark charcoal/black classic-inspired muscle car.\nWide aggressive stance.\nWarm red rear lights.\nNo logos or recognizable real-world branding.\n\nBLACK SUVs:\nLarge dark SUVs with bright headlights.\nConsistent model and silhouette throughout the video.\nNo brand badges.\n\n[WORLD / PHYSICS RULES]\n\nThe action follows realistic cinematic cause-and-effect.\n\nEvery movement must have:\nanticipation → action → physical consequence → reaction.\n\nExamples:\n- headlights appear → Jax notices → body turns → he warns Lena\n- Lena realizes trap → immediately creates cover\n- Jax identifies escape route → forces open rear exit\n- they reach car → Lena spots black drive → Jax accelerates\n\nNo random explosions.\nNo supernatural movement.\nNo teleporting.\nNo impossible weapon behavior.\nNo excessive debris.\n\nKeep action grounded, readable, and fast.\n\nintegrated_multimodal_description:\n\n[SHOT 1] (0.00s–2.20s) — THE PICKUP\n\nStart with a wide cinematic exterior of the isolated roadside gas station under a dramatic orange-magenta sunset.\n\nLong shadows cross slightly wet asphalt.\nThe convenience-store interior glows with cool fluorescent light.\nThe station canopy is illuminated against the colorful dusk sky.\n\nLena and Jax walk toward the station together from the foreground.\n\nAs they approach the pumps, they stop.\n\nJax slowly scans the perimeter, checking parked vehicles, road entrances, and reflections in the store windows.\n\nLena adjusts the red bandana over her lower face and looks toward the convenience-store door.\n\nShe gives Jax a tiny confirming glance.\n\nJax gives a subtle nod.\n\nLena separates from him and walks toward the store.\n\nCamera:\nBegin with a low wide-angle establishing dolly toward them.\nTransition into a medium two-shot.\nSubtle arc around Lena as she turns toward the entrance.\n\nEmotion:\nquiet professionalism with underlying suspicion.\n\nDialogue:\n\nJax (low controlled voice):\n<d>[English] Two minutes. If anything feels wrong, walk.</d>\n\nLena (dry, confident):\n<d>[English] You worry too much.</d>\n\nJax:\n<d>[English] That’s why we’re still alive.</d>\n\nSound:\n- distant highway ambience\n- light desert wind\n- faint fuel-pump hum\n- shoes on asphalt\n- distant car pass\n- soft fabric movement\n- subtle atmospheric thriller pulse begins\n\nTransition:\nFollow Lena through the glass entrance.\n\n[SHOT 2] (2.20s–4.40s) — THE HANDOFF\n\nInside the convenience store.\n\nThe lighting shifts from warm sunset orange outside to cool desaturated fluorescent interior light.\n\nLena walks toward the counter.\n\nThe cashier watches her nervously.\n\nHis eyes briefly flick toward the front windows before returning to Lena.\n\nLena notices this.\n\nThe cashier slowly slides a plain brown paper bag across the counter.\n\nLena does not immediately grab it.\n\nShe studies the cashier.\n\nHe avoids eye contact.\n\nThen Lena takes the bag.\n\nCamera:\nMedium tracking shot behind Lena entering the store.\nCut to cashier’s anxious close-up.\nInsert close-up of the brown paper bag sliding across the counter.\nSlow push toward Lena’s eyes.\n\nEmotion:\nsuspicion increasing.\n\nDialogue:\n\nLena (quiet):\n<d>[English] Where’s Vale?</d>\n\nCashier (nervous, whispering):\n<d>[English] I... I was only told to give you the bag.</d>\n\nLena:\n<d>[English] Told by who?</d>\n\nThe cashier says nothing.\n\nSound:\n- fluorescent electrical hum\n- refrigerator compressor\n- paper bag scraping across countertop\n- subtle nervous breath from cashier\n- distant exterior traffic\n\nMusic:\nMinimal low electronic pulse, restrained.\n\nTransition:\nLena opens the bag.\n\n[SHOT 3] (4.40s–6.20s) — TOO SLOW\n\nTight insert inside the paper bag.\n\nInstead of the black drive:\na cheap green novelty keychain and folded paper note.\n\nLena pulls out the note.\n\nShe unfolds it.\n\nThe message reads:\n“TOO SLOW.”\n\nHold just long enough to understand.\n\nCut to extreme close-up of Lena’s eyes above the red bandana.\n\nHer expression immediately changes from suspicion to certainty.\n\nShe looks toward the front windows.\n\nCamera:\nMacro insert inside bag.\nMatch cut from unfolded paper to Lena’s eyes.\nVery slow push-in.\n\nEmotion:\nrealization.\n\nDialogue:\n\nLena (under her breath):\n<d>[English] Damn it.</d>\n\nCashier (frightened):\n<d>[English] I’m sorry.</d>\n\nSound:\n- paper unfolding\n- subtle bass hit when message is revealed\n- refrigerator hum suddenly feels louder\n- tiny room-tone drop immediately before Lena reacts\n\nTransition:\nHard motivated cut from Lena’s eyes to bright headlights outside.\n\n[SHOT 4] (6.20s–8.40s) — THE TRAP CLOSES\n\nExterior from Jax’s perspective.\n\nMultiple black SUVs turn into the gas station from different entrances.\n\nHeadlights flare directly toward him.\n\nOne SUV swings across the driveway.\nAnother rolls near the pumps.\nA third blocks the road exit.\n\nJax instantly understands.\n\nHis relaxed posture snaps into tactical alertness.\n\nHe turns toward the convenience store.\n\nCamera:\nOver-the-shoulder behind Jax facing approaching SUVs.\nFast controlled push-in as headlights appear.\nBrief low-angle shot of tires turning into the station.\nReturn to Jax.\n\nDialogue:\n\nJax (sharp, loud):\n<d>[English] Lena! Move!</d>\n\nInside, Lena immediately turns toward his voice.\n\nLena:\n<d>[English] I knew it.</d>\n\nSound:\n- SUV engines approaching\n- tire grit\n- doors beginning to open\n- Jax’s shout\n- store door bell rattling\n- thriller percussion enters\n\nTransition:\nCut on Lena’s rapid turn.\n\n[SHOT 5] (8.40s–10.80s) — BREAK THE TRAP\n\nInside the store, Lena reacts instantly.\n\nShe grabs the edge of a lightweight product display and shoves it sideways, creating visual cover between herself and the front windows.\n\nPackaged goods scatter across the floor.\n\nNo one is struck.\n\nAt the same time outside, Jax moves along the side of the building toward a rear service exit.\n\nHe shoulders or kicks the jammed rear door near the latch.\n\nThe door bursts outward.\n\nUse a rapid intercut rhythm between Lena and Jax.\n\nLena sprints toward the rear.\n\nJax reaches through and pulls the door wide for her.\n\nCamera:\nDynamic medium tracking with Lena.\nLow angle as display moves.\nCut to Jax moving beside brick wall.\nClose shot on his boot/body impacting the jammed door.\nWide rear-exit reveal.\n\nEmotion:\nrapid controlled survival response.\n\nDialogue:\n\nJax:\n<d>[English] Back exit! Go!</d>\n\nLena:\n<d>[English] Moving!</d>\n\nSound:\n- shelves rattling\n- packaged items falling\n- shoe impacts\n- rear-door metal thud\n- alarm begins chirping\n- SUV doors outside\n- tense rhythmic percussion increasing\n\nNo gunfire is necessary.\nThe threat should come from the trap and incoming crew, not graphic violence.\n\nTransition:\nLena bursts through the rear door past Jax.\n\n[SHOT 6] (10.80s–12.80s) — RUN FOR THE CAR\n\nLena and Jax sprint together behind the station.\n\nIndustrial rear lot:\nchain-link fence,\nservice barrels,\nwet asphalt,\ndumpsters,\nutility pipes,\nwarm sunset beyond the rooftops.\n\nAlarm light flashes intermittently across the walls.\n\nTheir dark muscle car is parked ahead.\n\nJax runs slightly toward the driver side.\nLena heads toward passenger side.\n\nCamera:\nLow side tracking shot moving parallel with them.\nBrief rear tracking shot toward the car.\nNatural speed blur at frame edges while keeping faces and silhouettes readable.\n\nDialogue:\n\nLena (breathing hard):\n<d>[English] Tell me you brought the car.</d>\n\nJax:\n<d>[English] You’re looking at it.</d>\n\nLena:\n<d>[English] Good answer.</d>\n\nSound:\n- fast footsteps\n- breathing\n- clothing movement\n- alarm chirp\n- distant SUV engines\n- keys / car unlock click\n\nTransition:\nAs Lena reaches the passenger door, something behind them catches her eye.\n\n[SHOT 7] (12.80s–15.00s) — THE REAL DRIVE / CLIFFHANGER\n\nLena glances backward.\n\nAcross the rear service lot, one of Rico’s crew is stepping into the lead black SUV.\n\nIn his hand:\nthe REAL matte-black drive.\n\nMake the black drive clearly visible for a brief readable moment against the sunset/headlight contrast.\n\nLena’s eyes lock onto it.\n\nJax is already in the driver seat and starts the muscle car.\n\nLena dives into the passenger side.\n\nThe engine roars.\n\nThe lead SUV accelerates away.\n\nJax launches after it.\n\nThe muscle car swings out from behind the station and begins pursuit as other black SUVs move behind them.\n\nFinish with a powerful rear three-quarter chase composition:\nmuscle car accelerating into the sunset,\nred taillights glowing,\nblack SUVs ahead and behind,\ndust and road grit lifting.\n\nInclude a brief insert of Lena’s determined eyes reflected in the side mirror.\n\nDialogue:\n\nLena (urgent):\n<d>[English] Jax—the drive!</d>\n\nJax:\n<d>[English] I see it.</d>\n\nLena:\n<d>[English] They still have it.</d>\n\nJax (calm):\n<d>[English] Then we take it back.</d>\n\nSound:\n- car ignition\n- deep engine rev\n- door slam\n- tire chirp\n- accelerating muscle-car engine\n- SUV engines\n- gravel scattering\n- wind building\n- final rising cinematic pulse\n\nEnd exactly as the pursuit begins.\n\nDo NOT show the highway chase beyond the first acceleration.\n\naudio_description:\n\nCreate native synchronized cinematic stereo audio.\n\nENVIRONMENT:\n- desert-edge evening wind\n- distant highway traffic\n- fuel-pump hum\n- convenience-store fluorescent buzz\n- refrigeration units\n- rear service-lot ambience\n\nFOLEY:\n- bandana fabric\n- footsteps\n- paper bag movement\n- note unfolding\n- product rack sliding\n- loose packages falling\n- metal rear door impact\n- clothing movement during sprint\n- vehicle doors\n- ignition\n- tires\n- engine acceleration\n\nCHARACTER VOICES:\n\nLena:\nyoung adult woman,\ncalm but assertive,\nslightly dry confidence,\ncontrolled breathing during action,\nnever screaming unnecessarily.\n\nJax:\nyoung adult man,\nlow calm tactical voice,\neconomical delivery,\nbecomes sharper only when warning Lena.\n\nCashier:\nyoung adult male,\nquiet,\nnervous,\nbreathy,\nclearly frightened by the situation.\n\nDo not give dialogue to Rico’s crew in this clip.\n\nnon_diegetic_music:\n\nStart with restrained cinematic synth pulse and low atmospheric bass.\n\n0.00–4.40:\nminimal tension pulse.\n\n4.40–6.20:\nlow bass accent when “TOO SLOW.” is revealed.\n\n6.20–10.80:\nintroduce restrained percussion and faster pulse as SUVs arrive.\n\n10.80–12.80:\nincrease rhythm during the escape sprint.\n\n12.80–15.00:\nshort rising action motif synchronized with ignition and acceleration.\n\nMusic must remain beneath dialogue and sound effects.\n\nDo NOT make the soundtrack heroic yet.\nPart 1 should end unresolved.\n\n[VISUAL STYLE]\n\nPremium cinematic 2D graphic-novel animation.\n\nVisual qualities:\n- refined ink outlines\n- painterly cel shading\n- subtle paper-brush texture inside artwork\n- realistic but stylized anatomy\n- strong silhouettes\n- expressive eyes\n- controlled motion blur\n- cinematic color grading\n- dramatic sunset rim lighting\n- high environment detail\n- atmospheric reflections\n- restrained depth of field\n- polished film-like compositing\n\nPalette:\n- burnt sunset orange\n- dusty magenta\n- violet-purple sky\n- teal/cyan fluorescent interior\n- charcoal vehicles\n- asphalt gray\n- red taillights\n- cool blue shadows\n\nDo not make the finished video look like a paper storyboard.\nIt should look like a completed animated graphic-novel action film.\n\n[CAMERA LANGUAGE]\n\nUse intentional cinematic camera movements:\n\n- slow dolly during initial approach\n- controlled push-ins for suspicion\n- macro insert for package reveal\n- over-the-shoulder convoy reveal\n- fast but readable tracking for escape\n- low-angle running shots\n- brief handheld-like urgency during trap reveal, but still smooth and readable\n- rear three-quarter vehicle shot for cliffhanger\n\nCamera should become progressively more energetic:\n\ncalm → suspicious → compressed → urgent → fast.\n\nDo not use:\n- random spinning\n- excessive Dutch angles\n- chaotic handheld shake\n- meaningless zooms\n- unreadable fast cutting\n\n[MOTION RULES]\n\nCharacter movement must feel physically grounded.\n\nLena:\nquick decisive reactions,\nathletic acceleration,\nsharp head turns,\nefficient body movement.\n\nJax:\ncontrolled economy of motion,\nfast tactical reactions,\nstrong purposeful running.\n\nVehicles:\nrealistic mass,\nsuspension response,\ntire grip,\nacceleration,\nheadlight movement,\nno weightless drifting.\n\nCloth:\nbandanas and ponytail react naturally to running and wind.\n\nObjects:\npaper, packages, doors, and displays obey physical momentum.\n\n[CONTINUITY RULES]\n\nKeep all identities identical across every shot.\n\nLena:\nsame face,\nsame ponytail,\nsame red bandana,\nsame top,\nsame shorts.\n\nJax:\nsame face,\nsame short dark hair,\nsame blue/dark bandana,\nsame shirt,\nsame pants.\n\nCashier:\nsame appearance during all store shots.\n\nMuscle car:\nsame body design and color.\n\nSUVs:\nsame consistent dark design.\n\nBlack drive:\nsame small matte-black rectangular object.\n\nKeep spatial geography coherent:\nfront gas station → convenience-store interior → side/rear exit → rear parking lot → muscle car → road.\n\nDo not teleport characters between locations.\n\nDo not show Rico clearly in Part 1.\n\n[SAFETY / NEGATIVE CONSTRAINTS]\n\nKeep action stylized, cinematic, and non-graphic.\n\nNo blood.\nNo gore.\nNo visible wounds.\nNo executions.\nNo graphic impacts.\nNo cruelty.\nNo civilians harmed.\nNo cashier harmed.\nNo explosive destruction.\nNo graphic weapon discharge.\nNo threatening close-up gun violence.\n\nNo storyboard sheet visible.\nNo panel borders.\nNo captions.\nNo timestamps.\nNo title cards.\nNo production notes.\nNo subtitles.\nNo logos.\nNo watermarks.\nNo real brands.\nNo readable store branding.\nNo recognizable copyrighted characters.\nNo random text except the intentional physical handwritten note “TOO SLOW.”\nNo identity drift.\nNo duplicate Lena.\nNo duplicate Jax.\nNo changing costumes.\nNo changing vehicle designs.\nNo extra limbs.\nNo malformed hands.\nNo distorted faces.\nNo disappearing bandanas.\nNo random props.\nNo scene confusion.\nNo hyperreal live action.\nNo 3D game-render look.\nNo anime-style transformation.\n\nFINAL FRAME:\nThe muscle car accelerates after the escaping SUV convoy beneath the burning sunset.\nLena’s determined eyes briefly visible in the mirror.\nThe line “They still have it” has just been spoken.\nThe chase has begun.\n\nCUT TO BLACK.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085606945755693556",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085606945755693556",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-07T06:00:19.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085606945755693556.mp4",
    "posterUrl": "/posters/x/2085606945755693556.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield_ai @higgsfield \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded Part 1 storyboard image as the PRIMARY visual and narrative reference for the entire video.\n\nUse it to preserve:\n- exact story progression\n- Lena Cruz and Jax Mercer character identity\n- wardrobe and bandana colors\n- gas-station environment\n- cashier appearance\n- paper bag and “TOO SLOW.” note\n- black drive design\n- black SUV convoy\n- dark muscle car\n- sunset color palette\n- cinematic graphic-novel visual language\n- suspense-to-action escalation\n- Part 1 cliffhanger ending\n\nDo NOT reproduce the storyboard sheet itself in the finished video.\nNo panel borders, timestamps, handwritten captions, production notes, reference-sheet layout, paper texture, storyboard numbering, or storyboard text should appear in the video.\n\nTransform the storyboard into one continuous polished cinematic animated action sequence.\n\n[VIDEO GOAL]\n\nDuration: exactly 15 seconds\nAspect ratio: 16:9 landscape\nStyle: premium cinematic 2D graphic-novel animation\nGenre: action thriller / crime chase\nTone: stylish, tense, fast, mysterious, controlled\nStory structure:\nquiet pickup → suspicious exchange → betrayal reveal → convoy trap → rapid escape → real drive spotted → chase begins\n\nThis is PART 1 only.\n\nDo NOT continue into the highway chase or overpass confrontation.\nThe exact final story beat is Lena realizing:\n“They still have it.”\n\n[CHARACTER IDENTITY LOCK]\n\nLENA CRUZ:\nYoung athletic woman with a high dark ponytail, sharp expressive eyes, red patterned bandana covering or resting around her lower face, fitted dark cropped top, light denim shorts, practical shoes, lean athletic proportions.\n\nPersonality:\nbold, observant, quick-reacting, fearless, direct.\n\nBeginning emotional state:\ncontrolled suspicion.\n\nEnding emotional state:\nfocused determination.\n\nKeep identical throughout:\n- facial structure\n- skin tone\n- ponytail\n- red bandana\n- dark cropped top\n- denim shorts\n- body proportions\n- age\n- eye shape\n- silhouette\n\nDo not change her hairstyle, outfit, facial proportions, or bandana pattern.\n\nJAX MERCER:\nTall athletic young man with short dark hair, intense eyes, blue/dark patterned bandana covering his lower face, fitted dark charcoal T-shirt, brown/dark pants, strong controlled proportions.\n\nPersonality:\ncalm, tactical, observant, protective, efficient.\n\nBeginning emotional state:\nquiet vigilance.\n\nEnding emotional state:\nfocused pursuit mode.\n\nKeep identical throughout:\n- face\n- short dark hairstyle\n- blue bandana\n- charcoal shirt\n- dark pants\n- build\n- height\n- proportions\n\nCASHIER:\nOrdinary nervous convenience-store employee.\nDark slightly messy hair, worried expression, simple work clothes.\nHe is frightened and uncomfortable, not villainous.\n\nRICO’S CREW:\nMostly seen through silhouettes, vehicle interiors, distant figures and brief readable glimpses.\nDark streetwear.\nNo unnecessary close-ups or extra character identities.\n\n[KEY PROP LOCK]\n\nBLACK DRIVE:\nSmall matte-black rectangular data device.\nSimple clean shape.\nThe REAL black drive is NOT inside the paper bag.\nIt is seen clearly later in the hand of Rico’s crew near the lead SUV.\n\nFAKE PACKAGE:\nPlain brown paper bag.\n\nInside:\n- cheap small green novelty keychain\n- folded paper note\n\nThe note must clearly communicate:\n“TOO SLOW.”\n\nDo not introduce additional items.\n\nMUSCLE CAR:\nDark charcoal/black classic-inspired muscle car.\nWide aggressive stance.\nWarm red rear lights.\nNo logos or recognizable real-world branding.\n\nBLACK SUVs:\nLarge dark SUVs with bright headlights.\nConsistent model and silhouette throughout the video.\nNo brand badges.\n\n[WORLD / PHYSICS RULES]\n\nThe action follows realistic cinematic cause-and-effect.\n\nEvery movement must have:\nanticipation → action → physical consequence → reaction.\n\nExamples:\n- headlights appear → Jax notices → body turns → he warns Lena\n- Lena realizes trap → immediately creates cover\n- Jax identifies escape route → forces open rear exit\n- they reach car → Lena spots black drive → Jax accelerates\n\nNo random explosions.\nNo supernatural movement.\nNo teleporting.\nNo impossible weapon behavior.\nNo excessive debris.\n\nKeep action grounded, readable, and fast.\n\nintegrated_multimodal_description:\n\n[SHOT 1] (0.00s–2.20s) — THE PICKUP\n\nStart with a wide cinematic exterior of the isolated roadside gas station under a dramatic orange-magenta sunset.\n\nLong shadows cross slightly wet asphalt.\nThe convenience-store interior glows with cool fluorescent light.\nThe station canopy is illuminated against the colorful dusk sky.\n\nLena and Jax walk toward the station together from the foreground.\n\nAs they approach the pumps, they stop.\n\nJax slowly scans the perimeter, checking parked vehicles, road entrances, and reflections in the store windows.\n\nLena adjusts the red bandana over her lower face and looks toward the convenience-store door.\n\nShe gives Jax a tiny confirming glance.\n\nJax gives a subtle nod.\n\nLena separates from him and walks toward the store.\n\nCamera:\nBegin with a low wide-angle establishing dolly toward them.\nTransition into a medium two-shot.\nSubtle arc around Lena as she turns toward the entrance.\n\nEmotion:\nquiet professionalism with underlying suspicion.\n\nDialogue:\n\nJax (low controlled voice):\n<d>[English] Two minutes. If anything feels wrong, walk.</d>\n\nLena (dry, confident):\n<d>[English] You worry too much.</d>\n\nJax:\n<d>[English] That’s why we’re still alive.</d>\n\nSound:\n- distant highway ambience\n- light desert wind\n- faint fuel-pump hum\n- shoes on asphalt\n- distant car pass\n- soft fabric movement\n- subtle atmospheric thriller pulse begins\n\nTransition:\nFollow Lena through the glass entrance.\n\n[SHOT 2] (2.20s–4.40s) — THE HANDOFF\n\nInside the convenience store.\n\nThe lighting shifts from warm sunset orange outside to cool desaturated fluorescent interior light.\n\nLena walks toward the counter.\n\nThe cashier watches her nervously.\n\nHis eyes briefly flick toward the front windows before returning to Lena.\n\nLena notices this.\n\nThe cashier slowly slides a plain brown paper bag across the counter.\n\nLena does not immediately grab it.\n\nShe studies the cashier.\n\nHe avoids eye contact.\n\nThen Lena takes the bag.\n\nCamera:\nMedium tracking shot behind Lena entering the store.\nCut to cashier’s anxious close-up.\nInsert close-up of the brown paper bag sliding across the counter.\nSlow push toward Lena’s eyes.\n\nEmotion:\nsuspicion increasing.\n\nDialogue:\n\nLena (quiet):\n<d>[English] Where’s Vale?</d>\n\nCashier (nervous, whispering):\n<d>[English] I... I was only told to give you the bag.</d>\n\nLena:\n<d>[English] Told by who?</d>\n\nThe cashier says nothing.\n\nSound:\n- fluorescent electrical hum\n- refrigerator compressor\n- paper bag scraping across countertop\n- subtle nervous breath from cashier\n- distant exterior traffic\n\nMusic:\nMinimal low electronic pulse, restrained.\n\nTransition:\nLena opens the bag.\n\n[SHOT 3] (4.40s–6.20s) — TOO SLOW\n\nTight insert inside the paper bag.\n\nInstead of the black drive:\na cheap green novelty keychain and folded paper note.\n\nLena pulls out the note.\n\nShe unfolds it.\n\nThe message reads:\n“TOO SLOW.”\n\nHold just long enough to understand.\n\nCut to extreme close-up of Lena’s eyes above the red bandana.\n\nHer expression immediately changes from suspicion to certainty.\n\nShe looks toward the front windows.\n\nCamera:\nMacro insert inside bag.\nMatch cut from unfolded paper to Lena’s eyes.\nVery slow push-in.\n\nEmotion:\nrealization.\n\nDialogue:\n\nLena (under her breath):\n<d>[English] Damn it.</d>\n\nCashier (frightened):\n<d>[English] I’m sorry.</d>\n\nSound:\n- paper unfolding\n- subtle bass hit when message is revealed\n- refrigerator hum suddenly feels louder\n- tiny room-tone drop immediately before Lena reacts\n\nTransition:\nHard motivated cut from Lena’s eyes to bright headlights outside.\n\n[SHOT 4] (6.20s–8.40s) — THE TRAP CLOSES\n\nExterior from Jax’s perspective.\n\nMultiple black SUVs turn into the gas station from different entrances.\n\nHeadlights flare directly toward him.\n\nOne SUV swings across the driveway.\nAnother rolls near the pumps.\nA third blocks the road exit.\n\nJax instantly understands.\n\nHis relaxed posture snaps into tactical alertness.\n\nHe turns toward the convenience store.\n\nCamera:\nOver-the-shoulder behind Jax facing approaching SUVs.\nFast controlled push-in as headlights appear.\nBrief low-angle shot of tires turning into the station.\nReturn to Jax.\n\nDialogue:\n\nJax (sharp, loud):\n<d>[English] Lena! Move!</d>\n\nInside, Lena immediately turns toward his voice.\n\nLena:\n<d>[English] I knew it.</d>\n\nSound:\n- SUV engines approaching\n- tire grit\n- doors beginning to open\n- Jax’s shout\n- store door bell rattling\n- thriller percussion enters\n\nTransition:\nCut on Lena’s rapid turn.\n\n[SHOT 5] (8.40s–10.80s) — BREAK THE TRAP\n\nInside the store, Lena reacts instantly.\n\nShe grabs the edge of a lightweight product display and shoves it sideways, creating visual cover between herself and the front windows.\n\nPackaged goods scatter across the floor.\n\nNo one is struck.\n\nAt the same time outside, Jax moves along the side of the building toward a rear service exit.\n\nHe shoulders or kicks the jammed rear door near the latch.\n\nThe door bursts outward.\n\nUse a rapid intercut rhythm between Lena and Jax.\n\nLena sprints toward the rear.\n\nJax reaches through and pulls the door wide for her.\n\nCamera:\nDynamic medium tracking with Lena.\nLow angle as display moves.\nCut to Jax moving beside brick wall.\nClose shot on his boot/body impacting the jammed door.\nWide rear-exit reveal.\n\nEmotion:\nrapid controlled survival response.\n\nDialogue:\n\nJax:\n<d>[English] Back exit! Go!</d>\n\nLena:\n<d>[English] Moving!</d>\n\nSound:\n- shelves rattling\n- packaged items falling\n- shoe impacts\n- rear-door metal thud\n- alarm begins chirping\n- SUV doors outside\n- tense rhythmic percussion increasing\n\nNo gunfire is necessary.\nThe threat should come from the trap and incoming crew, not graphic violence.\n\nTransition:\nLena bursts through the rear door past Jax.\n\n[SHOT 6] (10.80s–12.80s) — RUN FOR THE CAR\n\nLena and Jax sprint together behind the station.\n\nIndustrial rear lot:\nchain-link fence,\nservice barrels,\nwet asphalt,\ndumpsters,\nutility pipes,\nwarm sunset beyond the rooftops.\n\nAlarm light flashes intermittently across the walls.\n\nTheir dark muscle car is parked ahead.\n\nJax runs slightly toward the driver side.\nLena heads toward passenger side.\n\nCamera:\nLow side tracking shot moving parallel with them.\nBrief rear tracking shot toward the car.\nNatural speed blur at frame edges while keeping faces and silhouettes readable.\n\nDialogue:\n\nLena (breathing hard):\n<d>[English] Tell me you brought the car.</d>\n\nJax:\n<d>[English] You’re looking at it.</d>\n\nLena:\n<d>[English] Good answer.</d>\n\nSound:\n- fast footsteps\n- breathing\n- clothing movement\n- alarm chirp\n- distant SUV engines\n- keys / car unlock click\n\nTransition:\nAs Lena reaches the passenger door, something behind them catches her eye.\n\n[SHOT 7] (12.80s–15.00s) — THE REAL DRIVE / CLIFFHANGER\n\nLena glances backward.\n\nAcross the rear service lot, one of Rico’s crew is stepping into the lead black SUV.\n\nIn his hand:\nthe REAL matte-black drive.\n\nMake the black drive clearly visible for a brief readable moment against the sunset/headlight contrast.\n\nLena’s eyes lock onto it.\n\nJax is already in the driver seat and starts the muscle car.\n\nLena dives into the passenger side.\n\nThe engine roars.\n\nThe lead SUV accelerates away.\n\nJax launches after it.\n\nThe muscle car swings out from behind the station and begins pursuit as other black SUVs move behind them.\n\nFinish with a powerful rear three-quarter chase composition:\nmuscle car accelerating into the sunset,\nred taillights glowing,\nblack SUVs ahead and behind,\ndust and road grit lifting.\n\nInclude a brief insert of Lena’s determined eyes reflected in the side mirror.\n\nDialogue:\n\nLena (urgent):\n<d>[English] Jax—the drive!</d>\n\nJax:\n<d>[English] I see it.</d>\n\nLena:\n<d>[English] They still have it.</d>\n\nJax (calm):\n<d>[English] Then we take it back.</d>\n\nSound:\n- car ignition\n- deep engine rev\n- door slam\n- tire chirp\n- accelerating muscle-car engine\n- SUV engines\n- gravel scattering\n- wind building\n- final rising cinematic pulse\n\nEnd exactly as the pursuit begins.\n\nDo NOT show the highway chase beyond the first acceleration.\n\naudio_description:\n\nCreate native synchronized cinematic stereo audio.\n\nENVIRONMENT:\n- desert-edge evening wind\n- distant highway traffic\n- fuel-pump hum\n- convenience-store fluorescent buzz\n- refrigeration units\n- rear service-lot ambience\n\nFOLEY:\n- bandana fabric\n- footsteps\n- paper bag movement\n- note unfolding\n- product rack sliding\n- loose packages falling\n- metal rear door impact\n- clothing movement during sprint\n- vehicle doors\n- ignition\n- tires\n- engine acceleration\n\nCHARACTER VOICES:\n\nLena:\nyoung adult woman,\ncalm but assertive,\nslightly dry confidence,\ncontrolled breathing during action,\nnever screaming unnecessarily.\n\nJax:\nyoung adult man,\nlow calm tactical voice,\neconomical delivery,\nbecomes sharper only when warning Lena.\n\nCashier:\nyoung adult male,\nquiet,\nnervous,\nbreathy,\nclearly frightened by the situation.\n\nDo not give dialogue to Rico’s crew in this clip.\n\nnon_diegetic_music:\n\nStart with restrained cinematic synth pulse and low atmospheric bass.\n\n0.00–4.40:\nminimal tension pulse.\n\n4.40–6.20:\nlow bass accent when “TOO SLOW.” is revealed.\n\n6.20–10.80:\nintroduce restrained percussion and faster pulse as SUVs arrive.\n\n10.80–12.80:\nincrease rhythm during the escape sprint.\n\n12.80–15.00:\nshort rising action motif synchronized with ignition and acceleration.\n\nMusic must remain beneath dialogue and sound effects.\n\nDo NOT make the soundtrack heroic yet.\nPart 1 should end unresolved.\n\n[VISUAL STYLE]\n\nPremium cinematic 2D graphic-novel animation.\n\nVisual qualities:\n- refined ink outlines\n- painterly cel shading\n- subtle paper-brush texture inside artwork\n- realistic but stylized anatomy\n- strong silhouettes\n- expressive eyes\n- controlled motion blur\n- cinematic color grading\n- dramatic sunset rim lighting\n- high environment detail\n- atmospheric reflections\n- restrained depth of field\n- polished film-like compositing\n\nPalette:\n- burnt sunset orange\n- dusty magenta\n- violet-purple sky\n- teal/cyan fluorescent interior\n- charcoal vehicles\n- asphalt gray\n- red taillights\n- cool blue shadows\n\nDo not make the finished video look like a paper storyboard.\nIt should look like a completed animated graphic-novel action film.\n\n[CAMERA LANGUAGE]\n\nUse intentional cinematic camera movements:\n\n- slow dolly during initial approach\n- controlled push-ins for suspicion\n- macro insert for package reveal\n- over-the-shoulder convoy reveal\n- fast but readable tracking for escape\n- low-angle running shots\n- brief handheld-like urgency during trap reveal, but still smooth and readable\n- rear three-quarter vehicle shot for cliffhanger\n\nCamera should become progressively more energetic:\n\ncalm → suspicious → compressed → urgent → fast.\n\nDo not use:\n- random spinning\n- excessive Dutch angles\n- chaotic handheld shake\n- meaningless zooms\n- unreadable fast cutting\n\n[MOTION RULES]\n\nCharacter movement must feel physically grounded.\n\nLena:\nquick decisive reactions,\nathletic acceleration,\nsharp head turns,\nefficient body movement.\n\nJax:\ncontrolled economy of motion,\nfast tactical reactions,\nstrong purposeful running.\n\nVehicles:\nrealistic mass,\nsuspension response,\ntire grip,\nacceleration,\nheadlight movement,\nno weightless drifting.\n\nCloth:\nbandanas and ponytail react naturally to running and wind.\n\nObjects:\npaper, packages, doors, and displays obey physical momentum.\n\n[CONTINUITY RULES]\n\nKeep all identities identical across every shot.\n\nLena:\nsame face,\nsame ponytail,\nsame red bandana,\nsame top,\nsame shorts.\n\nJax:\nsame face,\nsame short dark hair,\nsame blue/dark bandana,\nsame shirt,\nsame pants.\n\nCashier:\nsame appearance during all store shots.\n\nMuscle car:\nsame body design and color.\n\nSUVs:\nsame consistent dark design.\n\nBlack drive:\nsame small matte-black rectangular object.\n\nKeep spatial geography coherent:\nfront gas station → convenience-store interior → side/rear exit → rear parking lot → muscle car → road.\n\nDo not teleport characters between locations.\n\nDo not show Rico clearly in Part 1.\n\n[SAFETY / NEGATIVE CONSTRAINTS]\n\nKeep action stylized, cinematic, and non-graphic.\n\nNo blood.\nNo gore.\nNo visible wounds.\nNo executions.\nNo graphic impacts.\nNo cruelty.\nNo civilians harmed.\nNo cashier harmed.\nNo explosive destruction.\nNo graphic weapon discharge.\nNo threatening close-up gun violence.\n\nNo storyboard sheet visible.\nNo panel borders.\nNo captions.\nNo timestamps.\nNo title cards.\nNo production notes.\nNo subtitles.\nNo logos.\nNo watermarks.\nNo real brands.\nNo readable store branding.\nNo recognizable copyrighted characters.\nNo random text except the intentional physical handwritten note “TOO SLOW.”\nNo identity drift.\nNo duplicate Lena.\nNo duplicate Jax.\nNo changing costumes.\nNo changing vehicle designs.\nNo extra limbs.\nNo malformed hands.\nNo distorted faces.\nNo disappearing bandanas.\nNo random props.\nNo scene confusion.\nNo hyperreal live action.\nNo 3D game-render look.\nNo anime-style transformation.\n\nFINAL FRAME:\nThe muscle car accelerates after the escaping SUV convoy beneath the burning sunset.\nLena’s determined eyes briefly visible in the mirror.\nThe line “They still have it” has just been spoken.\nThe chase has begun.\n\nCUT TO BLACK.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 0,
      "views": 1446,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085602322177282512",
    "title": "橘猫惊天一屁",
    "titleEn": "The Orange Cat Lets It Out",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橘猫惊天一屁的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the orange cat lets it out on X. Native-video output: 15s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Cinematic comedy sequence, 15 seconds. [SCENE 1: 0-4s] Extreme close-up: A morbidly obese, orange tabby cat is fast asleep on a tiny, expensive velvet pillow in a pristine living room. The cat’s stomach rumbles ominously, emitting a deep, bass-heavy vibration that violently shakes the coffee table. [CUT TO] [SCENE 2: 4-7s] Low angle macro shot: The cat lifts one hind leg slightly. A tiny, glowing neon-green puff of gas escapes in ultra slow-motion, accompanied by a rippling shockwave that visually warps the air around it. [CUT TO] [SCENE 3: 7-11s] Wide tracking shot: The shockwave expands violently, blowing the living room furniture, the TV, and the doors out of the house in a fiery blast of glowing green energy and flying cat hair. [CUT TO] [SCENE 4: 11-15s] Orbiting space shot: Looking down at Earth from orbit. A colossal, glowing green mushroom cloud erupts from the continent, rapidly expanding into the exact shape of a giant, glowing cat paw. Epic cinematic lighting, hyper-realistic, deep shadows.",
    "promptSourceUrl": "https://x.com/TheAwakenOne619/status/2085602322177282512",
    "sourceUrl": "https://x.com/TheAwakenOne619/status/2085602322177282512",
    "sourceLabel": "X 原帖 · @TheAwakenOne619",
    "author": "TheAwakenOne",
    "publishedAt": "2026-08-07T05:41:56.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085602322177282512.mp4",
    "posterUrl": "/posters/x/2085602322177282512.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "sci-fi",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sometimes, you just have to let it out lol\n\nMiniMax H3\n\nPrompt:\nCinematic comedy sequence, 15 seconds. [SCENE 1: 0-4s] Extreme close-up: A morbidly obese, orange tabby cat is fast asleep on a tiny, expensive velvet pillow in a pristine living room. The cat’s stomach rumbles ominously, emitting a deep, bass-heavy vibration that violently shakes the coffee table. [CUT TO] [SCENE 2: 4-7s] Low angle macro shot: The cat lifts one hind leg slightly. A tiny, glowing neon-green puff of gas escapes in ultra slow-motion, accompanied by a rippling shockwave that visually warps the air around it. [CUT TO] [SCENE 3: 7-11s] Wide tracking shot: The shockwave expands violently, blowing the living room furniture, the TV, and the doors out of the house in a fiery blast of glowing green energy and flying cat hair. [CUT TO] [SCENE 4: 11-15s] Orbiting space shot: Looking down at Earth from orbit. A colossal, glowing green mushroom cloud erupts from the continent, rapidly expanding into the exact shape of a giant, glowing cat paw. Epic cinematic lighting, hyper-realistic, deep shadows.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 249,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085599659326935100",
    "title": "西部片风格的动漫片头",
    "titleEn": "A Spaghetti-Western Anime Title Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "西部片风格的动漫片头的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a spaghetti-western anime title sequence on X. Native-video output: 15s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use the attached character reference (Image1) and audio track as the timing guide. Create a spaghetti-western pulp anime title sequence where every cut lands precisely on the beat.\nStyle the character as a flat black silhouette against solid desert-inspired colors (burnt orange, ochre, rust, adobe tan), using bold pop-art compositions, vintage bounty-poster framing, split panels, extreme close-ups, and wide horizon shots. Alternate fast action bursts with full-beat freeze frames (draw, stride, turn, match strike, etc.), using hard cuts only.\nAdd worn woodblock/stencil title typography that slams into frame on the beat. Finish with the character frozen in silhouette on a burnt-orange background as the title stamps in on the final hit.\nApply heavy 35mm film grain, scratches, dust, gate weave, cigarette burns, and sun-faded print wear throughout for an authentic 1960s spaghetti-western look.",
    "promptSourceUrl": "https://x.com/doctorwasif/status/2085599659326935100",
    "sourceUrl": "https://x.com/doctorwasif/status/2085599659326935100",
    "sourceLabel": "X 原帖 · @doctorwasif",
    "author": "WasifAI",
    "publishedAt": "2026-08-07T05:31:22.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085599659326935100.mp4",
    "posterUrl": "/posters/x/2085599659326935100.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @openart_ai \n\nPrompt:\nUse the attached character reference (Image1) and audio track as the timing guide. Create a spaghetti-western pulp anime title sequence where every cut lands precisely on the beat.\nStyle the character as a flat black silhouette against solid desert-inspired colors (burnt orange, ochre, rust, adobe tan), using bold pop-art compositions, vintage bounty-poster framing, split panels, extreme close-ups, and wide horizon shots. Alternate fast action bursts with full-beat freeze frames (draw, stride, turn, match strike, etc.), using hard cuts only.\nAdd worn woodblock/stencil title typography that slams into frame on the beat. Finish with the character frozen in silhouette on a burnt-orange background as the title stamps in on the final hit.\nApply heavy 35mm film grain, scratches, dust, gate weave, cigarette burns, and sun-faded print wear throughout for an authentic 1960s spaghetti-western look.",
    "engagement": {
      "replies": 30,
      "reposts": 21,
      "likes": 250,
      "views": 15403,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085595583927185447",
    "title": "节奏、表演与动态字体",
    "titleEn": "Rhythm, Performance, and Kinetic Type",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "节奏、表演与动态字体的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for rhythm, performance, and kinetic type on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Style: Premium music commercial, ultra-realistic, cinematic, 8K HDR, natural lighting, luxury fashion aesthetic, smooth professional camera work, vibrant neon accents, realistic skin textures, high-end choreography, dynamic typography synced to the beat, music-video quality, no watermark.\n\nScene 1 (0–3s)\n\nA stylish young female artist walks confidently through a modern city street at night, flanked by two equally fashionable dancers. They wear coordinated luxury streetwear with premium sneakers and jewelry. The beat drops as the camera circles them in slow motion. Giant animated typography appears behind them, displaying the lyrics in bold kinetic text synchronized perfectly with every word.\n\nScene 2 (3–8s)\n\nThe trio performs energetic hip-hop choreography in a neon-lit urban plaza. The camera switches between cinematic gimbal shots, low-angle tracking, drone passes, and close-ups of expressive faces. Every lyric appears on screen with premium motion graphics, glowing typography, smooth transitions, and rhythmic animations that pulse with the music.\n\nScene 3 (8–12s)\n\nFast-paced performance featuring synchronized dance moves, stylish hand gestures, confident attitude, realistic hair movement, natural fabric physics, and dynamic lighting. The lyrics continue appearing in large, eye-catching typography that interacts with the environment, wrapping around buildings and moving with the beat.\n\nScene 4 (12–15s)\n\nThe music reaches its climax. All three performers strike a powerful final pose as the camera cranes upward. Massive cinematic typography fills the background with the final lyric while city lights sparkle behind them. Premium commercial color grading, flawless visual effects, realistic crowd energy, luxury music-video finish.\n\nHip-Hop Lyrics (On-Screen Typography):\n\nWe rise, no limits, we own the night,\nDreams in motion, shining so bright.\nMove with the rhythm, feel the fire,\nWe came to win—taking it higher!\n\nVisual Style: Premium commercial production, Billboard music video aesthetic, luxury fashion campaign, realistic camera movement, cinematic lens flares, authentic facial expressions, smooth choreography, vibrant urban atmosphere, perfectly synchronized kinetic lyric typography.",
    "promptSourceUrl": "https://x.com/Inshrah_ali_/status/2085595583927185447",
    "sourceUrl": "https://x.com/Inshrah_ali_/status/2085595583927185447",
    "sourceLabel": "X 原帖 · @Inshrah_ali_",
    "author": "Ai Girllie",
    "publishedAt": "2026-08-07T05:15:10.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085595583927185447.mp4",
    "posterUrl": "/posters/x/2085595583927185447.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Turning rhythm into visuals with realistic performances, kinetic typography and commercial-quality storytelling\n\nCreated this with Minimax H3 on @budgetpixel \n\nPrompt:\nStyle: Premium music commercial, ultra-realistic, cinematic, 8K HDR, natural lighting, luxury fashion aesthetic, smooth professional camera work, vibrant neon accents, realistic skin textures, high-end choreography, dynamic typography synced to the beat, music-video quality, no watermark.\n\nScene 1 (0–3s)\n\nA stylish young female artist walks confidently through a modern city street at night, flanked by two equally fashionable dancers. They wear coordinated luxury streetwear with premium sneakers and jewelry. The beat drops as the camera circles them in slow motion. Giant animated typography appears behind them, displaying the lyrics in bold kinetic text synchronized perfectly with every word.\n\nScene 2 (3–8s)\n\nThe trio performs energetic hip-hop choreography in a neon-lit urban plaza. The camera switches between cinematic gimbal shots, low-angle tracking, drone passes, and close-ups of expressive faces. Every lyric appears on screen with premium motion graphics, glowing typography, smooth transitions, and rhythmic animations that pulse with the music.\n\nScene 3 (8–12s)\n\nFast-paced performance featuring synchronized dance moves, stylish hand gestures, confident attitude, realistic hair movement, natural fabric physics, and dynamic lighting. The lyrics continue appearing in large, eye-catching typography that interacts with the environment, wrapping around buildings and moving with the beat.\n\nScene 4 (12–15s)\n\nThe music reaches its climax. All three performers strike a powerful final pose as the camera cranes upward. Massive cinematic typography fills the background with the final lyric while city lights sparkle behind them. Premium commercial color grading, flawless visual effects, realistic crowd energy, luxury music-video finish.\n\nHip-Hop Lyrics (On-Screen Typography):\n\nWe rise, no limits, we own the night,\nDreams in motion, shining so bright.\nMove with the rhythm, feel the fire,\nWe came to win—taking it higher!\n\nVisual Style: Premium commercial production, Billboard music video aesthetic, luxury fashion campaign, realistic camera movement, cinematic lens flares, authentic facial expressions, smooth choreography, vibrant urban atmosphere, perfectly synchronized kinetic lyric typography.",
    "engagement": {
      "replies": 34,
      "reposts": 4,
      "likes": 157,
      "views": 17412,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085594060384010351",
    "title": "日文字幕的动画练习",
    "titleEn": "Japanese Captions in Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "日文字幕的动画练习的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for japanese captions in motion on X. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "KAWAII HORRORミュージックビデオを完成させる。これは映像とキネティック・タイポグラフィを一括生成する実験である。歌詞テロップは後処理に回さず、生成動画の中へ直接描画する。Picture 1はぬいぐるみのフルフレーム映像の主画像として使う。Picture 2は、デザインされたテロップの形とモーション言語だけを参考にする。Picture 2そのもの、絵コンテのグリッド、グリーンバック、CUTラベル、時間表示、サイドパネル、サイドバー、黒いレール、固定されたテキスト枠は映像内に表示しない。  ぬいぐるみのキャラクター、顔の特徴、縫い目、リボン、ボタン、ドクロチャーム、CRT画面、ピンク・ラベンダー・シアンの色、各キャラクターの位置関係を正確に保つ。ぬいぐるみには、呼吸、瞬き、小さな首の傾き、耳の動き、リボンの揺れ、手の動き、目の中のハートの発光、カットごとの小さなポーズ変化を必ず付ける。顔は最後まで認識可能にする。カメラは、ゆっくりしたプッシュイン、短い横移動、小さな弧の移動、引きの動き、ハードなグラフィックカットを使う。ランダムな手ブレ、顔の変形、溶解、ゴア、新キャラクター、物体のランダムな置き換えは禁止。  アップロードした音源だけをマスター音源として使う。新しい音楽を生成しない。歌い直さない。自動字幕のタイミングを使わない。以下の歌詞タイムラインをハードロックする。歌詞を早く出さない。後半の歌詞イベントを前半へ詰め込まない。字幕が前倒しされて最後の0.360秒が空白になることも禁止する。  テロップは普通の字幕ではなく、大きなデザイン文字によるモーショングラフィックにする。手描きで少し不規則な文字形、文字ごとに異なる大きさ、ベースラインのずれ、意図的な傾き、太い白文字、不規則な黒い抜き形、ホットピンクの縁、シアンとマゼンタのRGB残像、尖った終端、斜めの軌跡、インク片、ピクセル片を使う。各フレーズは、カットごとに違う侵入、移動、読みやすい保持、退場を持たせる。指定した場合は、文字単位または2〜3文字の小グループ単位で時間差をつける。メインの白文字は保持中に完全に読める状態にし、綴りを変えない。  【文字の完全固定】使用できる歌詞は次の8フレーズだけ。順番も固定する。\"Sweet Error\"、\"愛して\"、\"こわれた分だけ\"、\"可愛くして\"、\"Sweet Error\"、\"消さないで\"、\"私はここだよ\"、\"ここだよ\"。漢字の\"壊れた\"は絶対に使わず、必ず\"こわれた\"と表示する。翻訳、言い換え、追加、削除、結合、順番変更、重複、文字の入れ替え、文字化け、別の文字の発明は禁止。Shot 4だけは指定した2行表示を許可する。  [Shot 1] 00.000〜01.520秒。Picture 1のぬいぐるみ全体を見せ、中央のラベンダーのぬいぐるみへ小さな振幅でゆっくりプッシュインする。中央のぬいぐるみが呼吸し、1回瞬きをし、リボンを少し揺らす。左のピンクのぬいぐるみは首を少し傾け、青いぬいぐるみは片手を少し動かし、右上の白いぬいぐるみの目のハートが小さく光る。CRT画面は弱くちらつく。歌詞、文字の一部、残像、文字の侵入も一切表示しない。  [Shot 2] 00:01.520秒でハードカット。右側の固定パネルではなく、画面左上の空いた空間に\"Sweet Error\"だけを表示する。モーションパターンA＋F＋Dを使う。単語グループごとに別々の画面外から侵入させ、最初の単語は右上から、次の単語は右下から入れる。2つの単語を反対方向へ約12度ずつ回転させ、不規則な斜めのベースラインへスナップさせる。4フレームのシアン・マゼンタ分離と曲線状のカラートレイルを付ける。文字列全体を1回だけ小さく弾ませ、完全に読める状態で保持する。03.050秒で残像だけを水平に引き裂き、03.260秒までに文字と残像をすべて消す。  [Shot 3] 00:03.260秒でハードカット。少し寄った新しいポーズにする。ピンクのぬいぐるみが文字の方向へ首を傾け、青いぬいぐるみが片手を上げ、中央のぬいぐるみが瞬きをする。画面中央右に\"愛して\"だけを表示する。モーションパターンC＋Fを使う。文字を1文字ずつ下から出し、それぞれ少し異なる曲線軌道で上昇させる。隣り合う文字を時計回り・反時計回りで交互に約16度回転させ、2フレームずつ時間差をつける。フレーズ全体を4分の1円弧に沿って移動させ、上方向へ1回オーバーシュートし、回転を戻して整列させる。シアンの細い軌跡を付ける。04.600秒から斜めに退場させ、04.840秒までに完全に消す。  [Shot 4] 00:04.840秒でハードカット。新しい集合ポーズにし、カメラをゆっくり右へ横移動させる。中央のぬいぐるみが顎を上げ、ピンクのリボンが1回揺れ、青いぬいぐるみが少し顔を向け、右上の白いぬいぐるみが瞬きをする。画面右下の空いた空間に、1つのデザインブロックとして2行を表示する。1行目は\"こわれた分だけ\"、2行目は\"可愛くして\"。モーションパターンA＋J＋Hを使う。1行目は右上から斜めに落とし、文字ごとにマイナス14度・プラス14度を交互に付ける。2行目は右下から上昇させ、反対方向の傾きを使う。文字を一度に全部出さず、2〜3文字ずつ時間差を付ける。05.250秒で2行を3フレームだけ衝突・圧縮させ、その後ばねのように跳ね戻して2行の最終配置にする。05.400〜07.300秒は2行とも完全に読みやすく保持する。07.300秒で黒い抜き形を磁気粒子へ分解し、前の位置の薄い残像を左へにじませ、シアンとマゼンタの残りを散らす。08.000秒までに両方の行を完全に消す。\"壊れた\"は表示しない。  [Shot 5] 00:08.000秒で強いパステル・グリッチブリッジを作り、新しい読みやすいぬいぐるみポーズへ切り替える。中央の目のハートが1回発光し、ピンクのぬいぐるみが耳を動かし、青いぬいぐるみが少し顔を向ける。画面左下の空間に2回目の\"Sweet Error\"だけを表示する。モーションパターンB＋D＋Iを使う。1フレームの白・シアン衝撃フラッシュで文字をテレポートさせ、文字列全体を時計回りに18度回転させながら横方向へ伸ばす。文字を3つのRGBコピーに分離し、3方向へ短く飛ばし、白いメイン文字を鋭く再構成する。短いCRTスキャンラインの裂け目を1回だけ入れる。08.350〜09.200秒は完全に読める状態で保持する。09.200秒で残像を反時計回りに回転させ、左下の画面外へ排出し、09.580秒までにすべて消す。08.000秒より前に始めない。  [Shot 6] 00:09.580秒で安定した新しいポーズへハードカット。中央のぬいぐるみが口を少し開くが、顔を変形させない。青いぬいぐるみが瞬きをし、全体がカメラへ少し傾く。カメラは小さくゆっくり引く。画面右上の空間に\"消さないで\"だけを表示する。モーションパターンC＋Fを使う。文字を左から右へ速い斜めの波として移動させ、各文字を3フレームずつ遅らせて出す。時計回り10度・反時計回り10度を交互に付け、1回だけ行き過ぎてから小さく上下に跳ねて止める。シアンとマゼンタの軌跡は顔の上ではなく文字のベースラインに沿わせる。10.050〜10.850秒は読みやすく保持する。10.850秒から左から右への斜めスライスで消し、11.160秒までに完全に消す。  [Shot 7] 00:11.160秒で中央のラベンダーのぬいぐるみの周囲をカメラが小さく滑らかな弧を描く。ぬいぐるみを左から右へ順番に瞬きさせ、中央のぬいぐるみが少し前へ傾き、目のハートを1回明るくする。固定サイドレールではなく、画面中央左の空間に\"私はここだよ\"だけを表示する。モーションパターンG＋Hを使う。まず各文字を細いシアンのワイヤーフレーム線で描き、2フレーム後に白いデザイン文字として塗り込む。隣り合う文字の大きさとベースラインを少し変え、約8度の穏やかな揺れ回転を付ける。前の線の残像を薄く残し、単語の間でハートを1回だけ脈動させ、最後にフレーズ全体を固定する。12.700秒から文字をずれた輪郭片へゆるめ、左上へ漂わせる。13.020秒までに文字と残像をすべて消す。  [Shot 8] 00:13.020秒で独立した最後の歌詞イベントを作る。中央のラベンダーのぬいぐるみがカメラへ少し傾き、目を合わせる。ピンクのぬいぐるみがリボンを1回動かし、青いぬいぐるみが片手を上げる。画面右下の空間に\"ここだよ\"だけを表示し、文字全体を画面内へ収める。モーションパターンK＋Eを使う。文字を1文字ずつ3フレーム間隔で出し、各文字を内側へ12度回転させて、少し異なるベースラインへスナップさせる。歌声のアクセントでフレーズ全体を約12パーセント拡大し、背後に2フレームだけシアン・白のフルフィールドフラッシュを入れ、最後の文字を1回だけ跳ねさせる。14.350秒まで完全に読みやすく保持する。14.350秒から文字を曲線状のカラートレイルに乗せて右下へ退場させ、14.640秒までにすべて消す。\"私はここだよ\"と結合しない。  [Shot 9] 14.640〜15.000秒は、きれいなぬいぐるみ全体だけを表示する。歌詞、文字の一部、残像、パネル、黒いレール、フェードアウトは一切使わない。中央のポーズを安定させ、目の中のハートを小さく光らせ、CRTを1回だけ控えめにちらつかせて終了する。  禁止事項: 普通の字幕バー、中央のカラオケ字幕、固定右側パネル、サイドバー、黒いレール、グリーンバック、絵コンテのコラージュ、CUTラベル、時間表示、文字の変形、文字化け、漢字への置き換え、余計な単語、歌詞の重複、歌詞の重なり、動かないぬいぐるみ、制御不能な手ブレ、顔の変形、追加BGM、フェードトランジション。  overall_soundscape: アップロードしたマスター音源をそのまま使う。新しい台詞、歌声、BGM、環境音、効果音は追加しない。ぬいぐるみの動きとテロップの衝撃を音源に合わせる。ただし、歌詞の開始・終了は上記の固定タイムラインを最優先する。  non_diegetic_music: アップロードしたマスター曲だけを使う。追加の音楽は使わない。",
    "promptSourceUrl": "https://x.com/su_nagomi/status/2085594063861088633",
    "sourceUrl": "https://x.com/su_nagomi/status/2085594060384010351",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-07T05:09:07.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085594060384010351.mp4",
    "posterUrl": "/posters/x/2085594060384010351.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でテロップアニメーションの練習。\n\nカット画像とテロップ画像を別に作成して編集無しで作成。他の人みたいにうまく表現できないですね。まだまだ修行が足りん。\n\n動画プロンプトはリプ欄に👇\n@Hailuo_AI #MiniMaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 481,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085585698699243802",
    "title": "最后一刻的极限救援",
    "titleEn": "A Last-Second Rescue",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "最后一刻的极限救援的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a last-second rescue on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic 15-second ultra-realistic Hollywood action sequence featuring a skilled fighter facing his final battle inside an abandoned industrial warehouse. Surrounded by three aggressive opponents, he uses precise martial arts movements to dodge attacks, counter strikes, and overcome every challenge with determination. Dust fills the air as sunlight breaks through shattered windows, creating a dramatic battlefield atmosphere. The scene builds toward a tense final confrontation as the fighter stands alone among smoke, rainwater, and debris. Dynamic handheld camera work, low-angle tracking shots, slow-motion action moments, realistic choreography, intense facial expressions, physically accurate movement, dramatic lighting, ARRI Alexa 35 aesthetic, anamorphic lens look, HDR, subtle film grain, premium Hollywood action style, 4K cinematic quality, no text, no logos.",
    "promptSourceUrl": "https://x.com/aiwithaly/status/2085585698699243802",
    "sourceUrl": "https://x.com/aiwithaly/status/2085585698699243802",
    "sourceLabel": "X 原帖 · @aiwithaly",
    "author": "Ali",
    "publishedAt": "2026-08-07T04:35:53.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085585698699243802.mp4",
    "posterUrl": "/posters/x/2085585698699243802.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "retro",
      "ugc"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sometimes survival comes down to one final moment. And one person refusing to give up..\n\nCreated with MiniMax H3\n\nPrompt:\n\nA cinematic 15-second ultra-realistic Hollywood action sequence featuring a skilled fighter facing his final battle inside an abandoned industrial warehouse. Surrounded by three aggressive opponents, he uses precise martial arts movements to dodge attacks, counter strikes, and overcome every challenge with determination. Dust fills the air as sunlight breaks through shattered windows, creating a dramatic battlefield atmosphere. The scene builds toward a tense final confrontation as the fighter stands alone among smoke, rainwater, and debris. Dynamic handheld camera work, low-angle tracking shots, slow-motion action moments, realistic choreography, intense facial expressions, physically accurate movement, dramatic lighting, ARRI Alexa 35 aesthetic, anamorphic lens look, HDR, subtle film grain, premium Hollywood action style, 4K cinematic quality, no text, no logos.",
    "engagement": {
      "replies": 57,
      "reposts": 2,
      "likes": 202,
      "views": 8487,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085584972778229978",
    "title": "他靠时机赢下力量赛",
    "titleEn": "He Wins the Strength Contest with Timing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "他靠时机赢下力量赛的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for he wins the strength contest with timing on X. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15 second cinematic sports video in 16:9 inside a real indoor strength competition arena packed with spectators. The story contrasts repeated failure with one calm, unexpected success. It must feel like authentic live event footage captured by a professional broadcast crew, with natural human motion, believable weight, consistent anatomy, coherent crowd behavior, and physically accurate object movement.\n\n0.00 to 2.70 seconds. Show three rapid failed keg toss attempts by three different heavyweight athletes. Use low sideline wide shots that include the athlete, tall steel throwing rig, red crossbar, and full keg arc. Each attempt lasts less than one second. One keg clips the bar and spins backward, one falls short, and one drifts sideways. The athletes show brief frustration while the crowd gives disappointed gasps. Use clean match cuts based on the circular keg shape.\n\n2.70 to 4.40 seconds. Let the music soften. Cut to a medium wide shot of an unassuming compact middle aged competitor entering from the arena tunnel. He wears a plain black competition shirt, dark blue shorts, knee sleeves, lifting shoes, and a worn leather belt. He walks calmly toward the keg while larger athletes watch from the edge of frame.\n\n4.40 to 7.20 seconds. Use three tactile close shots. Chalk pressing into his palms. Fingers locking around a cold dented steel rim. Shoes gripping the gray rubber floor as his knees bend. Show subtle breathing, sweat, fabric tension, skin compression, and a slight forearm tremor. Keep the same face, body, wardrobe, keg, and arena geography in every shot.\n\n7.20 to 11.80 seconds. Continue in one unbroken low angle tracking shot with a natural 35 mm lens look. He swings the keg between his legs, drives through the hips, extends the knees, pulls upward, and releases at the correct moment. The camera makes a controlled upward tilt. The keg rotates with realistic inertia and follows a clean parabolic arc. Use natural motion blur at 30 frames per second with a 1 over 60 shutter feel. Hold a brief near silence before release.\n\n11.80 to 13.40 seconds. The keg clears the red crossbar by only a few centimeters and lands heavily on the padded platform behind the rig without an unrealistic bounce. The crowd erupts on impact. A judge instinctively raises both arms to confirm success.\n\n13.40 to 15.00 seconds. Cut to a medium close shot as the competitor raises one fist and gives a restrained relieved smile. Larger athletes behind him stare in disbelief, then applaud. Finish on a natural crowd reaction, not a freeze frame.\n\nUse cool white overhead arena lighting, subtle blue haze, realistic warm skin tones, gentle lens bloom around practical lights, deep readable shadows, neutral cinematic color, controlled contrast, fine film grain, realistic sweat highlights, and no excessive sharpness.\n\nSound and music. Use an original 82 BPM cinematic percussion track with deep toms, restrained bass pulses, and a steady tension build. Layer authentic arena room tone, crowd murmur, chalk rubbing, metal scraping, shoe friction, breathing, a sharp air movement sound during release, a solid steel landing impact, and a powerful crowd swell after clearance. No dialogue.\n\nDo not include text, captions, logos, watermarks, brand names, duplicated spectators, warped faces, extra fingers, changing clothing, inconsistent body size, floating objects, rubbery metal, impossible physics, sudden camera teleportation, or artificial slow motion. The final result must look photographic, grounded, spontaneous, and genuinely filmed at a live competition.",
    "promptSourceUrl": "https://x.com/bmx_ai13/status/2085584972778229978",
    "sourceUrl": "https://x.com/bmx_ai13/status/2085584972778229978",
    "sourceLabel": "X 原帖 · @bmx_ai13",
    "author": "BMX",
    "publishedAt": "2026-08-07T04:33:00.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085584972778229978.mp4",
    "posterUrl": "/posters/x/2085584972778229978.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Everyone Had the Power. He Had the Timing.\n\nMinimax H3 on @wavespeed_ai \n\nPrompt: Create a 15 second cinematic sports video in 16:9 inside a real indoor strength competition arena packed with spectators. The story contrasts repeated failure with one calm, unexpected success. It must feel like authentic live event footage captured by a professional broadcast crew, with natural human motion, believable weight, consistent anatomy, coherent crowd behavior, and physically accurate object movement.\n\n0.00 to 2.70 seconds. Show three rapid failed keg toss attempts by three different heavyweight athletes. Use low sideline wide shots that include the athlete, tall steel throwing rig, red crossbar, and full keg arc. Each attempt lasts less than one second. One keg clips the bar and spins backward, one falls short, and one drifts sideways. The athletes show brief frustration while the crowd gives disappointed gasps. Use clean match cuts based on the circular keg shape.\n\n2.70 to 4.40 seconds. Let the music soften. Cut to a medium wide shot of an unassuming compact middle aged competitor entering from the arena tunnel. He wears a plain black competition shirt, dark blue shorts, knee sleeves, lifting shoes, and a worn leather belt. He walks calmly toward the keg while larger athletes watch from the edge of frame.\n\n4.40 to 7.20 seconds. Use three tactile close shots. Chalk pressing into his palms. Fingers locking around a cold dented steel rim. Shoes gripping the gray rubber floor as his knees bend. Show subtle breathing, sweat, fabric tension, skin compression, and a slight forearm tremor. Keep the same face, body, wardrobe, keg, and arena geography in every shot.\n\n7.20 to 11.80 seconds. Continue in one unbroken low angle tracking shot with a natural 35 mm lens look. He swings the keg between his legs, drives through the hips, extends the knees, pulls upward, and releases at the correct moment. The camera makes a controlled upward tilt. The keg rotates with realistic inertia and follows a clean parabolic arc. Use natural motion blur at 30 frames per second with a 1 over 60 shutter feel. Hold a brief near silence before release.\n\n11.80 to 13.40 seconds. The keg clears the red crossbar by only a few centimeters and lands heavily on the padded platform behind the rig without an unrealistic bounce. The crowd erupts on impact. A judge instinctively raises both arms to confirm success.\n\n13.40 to 15.00 seconds. Cut to a medium close shot as the competitor raises one fist and gives a restrained relieved smile. Larger athletes behind him stare in disbelief, then applaud. Finish on a natural crowd reaction, not a freeze frame.\n\nUse cool white overhead arena lighting, subtle blue haze, realistic warm skin tones, gentle lens bloom around practical lights, deep readable shadows, neutral cinematic color, controlled contrast, fine film grain, realistic sweat highlights, and no excessive sharpness.\n\nSound and music. Use an original 82 BPM cinematic percussion track with deep toms, restrained bass pulses, and a steady tension build. Layer authentic arena room tone, crowd murmur, chalk rubbing, metal scraping, shoe friction, breathing, a sharp air movement sound during release, a solid steel landing impact, and a powerful crowd swell after clearance. No dialogue.\n\nDo not include text, captions, logos, watermarks, brand names, duplicated spectators, warped faces, extra fingers, changing clothing, inconsistent body size, floating objects, rubbery metal, impossible physics, sudden camera teleportation, or artificial slow motion. The final result must look photographic, grounded, spontaneous, and genuinely filmed at a live competition.",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 12,
      "views": 1225,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085575259533783077",
    "title": "银甲战士的双模型对决",
    "titleEn": "An Armored Warrior Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银甲战士的双模型对决的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an armored warrior across two models on X. Native-video output: 15s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A young heroic warrior wearing beautifully crafted silver armor, a dark blue cape, and carrying an ancient enchanted longsword slowly enters an abandoned underground witch chamber. The room is filled with broken ritual circles, ancient books, shattered altars, skulls, scattered candles, and dozens of lifeless witch bodies lying across the cold stone floor. The atmosphere is eerily silent.\n\nThe camera slowly follows behind him as he carefully scans every corner, gripping his sword tightly and remaining fully alert.\n\nSuddenly, the ancient stone walls begin to tremble.\n\nA horrifying ghost emerges from the darkness. It has a grotesque deformed face, glowing crimson eyes, an unnaturally long body, and dozens of skeletal arms extending in every direction. Instead of wielding weapons, the ghost attacks using its razor-sharp claws and multiple arms, striking from impossible angles with incredible speed.\n\nAn intense close-quarters battle begins.\n\nThe warrior skillfully dodges the ghost's relentless claw attacks, parries with his enchanted sword, and cuts through several skeletal arms as they lunge toward him. The ghost continuously regenerates new arms, making the fight increasingly desperate.\n\nThe battle reaches its climax as the ghost charges forward with all of its arms at once. The warrior times the attack perfectly, leaps into the air, and delivers a single breathtaking spinning sword strike that cleanly severs the ghost's head.\n\nThe head falls to the stone floor as the ghost's body freezes in place, cracks with glowing purple energy, and disintegrates into countless glowing magical particles that fade into the air. The chamber becomes completely silent once again.\n\nStyle\n\nUltra-cinematic dark fantasy, realistic sword choreography, terrifying creature animation, expressive facial reactions, dramatic lighting, highly detailed gothic environment, dynamic camera movement, AAA fantasy film quality, Unreal Engine realism, IMAX composition, ultra-detailed textures\n\n#MiniMaxH3",
    "promptSourceUrl": "https://x.com/HeyZoyaKhan/status/2085575259533783077",
    "sourceUrl": "https://x.com/HeyZoyaKhan/status/2085575259533783077",
    "sourceLabel": "X 原帖 · @HeyZoyaKhan",
    "author": "ZOYA ✪",
    "publishedAt": "2026-08-07T03:54:24.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085575259533783077.mp4",
    "posterUrl": "/posters/x/2085575259533783077.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "wow.. I didn't expect MiniMax H3 to be this good.\n\nMiniMax H3 on @Hailuo_AI v/s Seedance 2.5\n\nPrompt:\nA young heroic warrior wearing beautifully crafted silver armor, a dark blue cape, and carrying an ancient enchanted longsword slowly enters an abandoned underground witch chamber. The room is filled with broken ritual circles, ancient books, shattered altars, skulls, scattered candles, and dozens of lifeless witch bodies lying across the cold stone floor. The atmosphere is eerily silent.\n\nThe camera slowly follows behind him as he carefully scans every corner, gripping his sword tightly and remaining fully alert.\n\nSuddenly, the ancient stone walls begin to tremble.\n\nA horrifying ghost emerges from the darkness. It has a grotesque deformed face, glowing crimson eyes, an unnaturally long body, and dozens of skeletal arms extending in every direction. Instead of wielding weapons, the ghost attacks using its razor-sharp claws and multiple arms, striking from impossible angles with incredible speed.\n\nAn intense close-quarters battle begins.\n\nThe warrior skillfully dodges the ghost's relentless claw attacks, parries with his enchanted sword, and cuts through several skeletal arms as they lunge toward him. The ghost continuously regenerates new arms, making the fight increasingly desperate.\n\nThe battle reaches its climax as the ghost charges forward with all of its arms at once. The warrior times the attack perfectly, leaps into the air, and delivers a single breathtaking spinning sword strike that cleanly severs the ghost's head.\n\nThe head falls to the stone floor as the ghost's body freezes in place, cracks with glowing purple energy, and disintegrates into countless glowing magical particles that fade into the air. The chamber becomes completely silent once again.\n\nStyle\n\nUltra-cinematic dark fantasy, realistic sword choreography, terrifying creature animation, expressive facial reactions, dramatic lighting, highly detailed gothic environment, dynamic camera movement, AAA fantasy film quality, Unreal Engine realism, IMAX composition, ultra-detailed textures\n\n#MiniMaxH3",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 22,
      "views": 10632,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085505725149622337",
    "title": "花朵绽放的色彩蒙太奇",
    "titleEn": "A Colorful Montage of Blooming Flowers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "花朵绽放的色彩蒙太奇的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a colorful montage of blooming flowers on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a breathtaking 15-second cinematic flower-morph sequence using the seven uploaded flower images as consecutive visual references:\n\nImage 1: white calla lily\nImage 2: coral hibiscus\nImage 3: burgundy dahlia\nImage 4: orange and cobalt-blue bird of paradise\nImage 5: neon-yellow orchid\nImage 6: deep-purple iris\nImage 7: electric-blue Himalayan poppy\n\nMOTION:\nOne continuous cinematic macro shot. The camera performs a slow, controlled push toward the center of the flower while every transformation happens clearly in full view. The flower remains centered against a deep, pure black background. Petals rotate, flutter, stretch, divide, multiply, fold, and physically reorganize into each new botanical form. Translucent long-exposure trails follow the movement like illuminated silk ribbons. The single stem remains visually anchored throughout the sequence.\n\nACTION AND TIMING:\n\n0–2 seconds:\nA white calla lily rotates gracefully. Its sculptural trumpet petal unfurls, creating sweeping trails of pearl white, silver, and champagne.\n\n2–4 seconds:\nThe calla lily’s curved petal organically divides into five broad petals, becoming a coral hibiscus. The golden spadix elongates into the hibiscus stamen while coral, peach, and white trails spiral outward.\n\n4–6 seconds:\nThe hibiscus petals rapidly multiply from the center and fold into hundreds of geometric burgundy and raspberry petals. The flower becomes a richly layered dahlia, forming a hypnotic rotating vortex.\n\n6–8 seconds:\nThe dahlia petals sharpen, extend, and reorganize into the architectural form of a bird of paradise. Burgundy and magenta ignite into fiery orange while cobalt-blue and turquoise structures burst outward like exotic wings.\n\n8–10 seconds:\nThe bird of paradise curls inward. Its orange structures soften and divide into the broad neon-yellow petals of an orchid. The cobalt-blue structures transform into the intricate orchid center, accented with crimson and gold. Yellow, chartreuse, and luminous-white trails pulse outward.\n\n10–12 seconds:\nThe orchid petals stretch and twist, transitioning from neon yellow through electric violet into deep royal purple. Golden markings expand into the textured beard of a majestic iris. The iris opens dramatically as violet, lavender, and golden trails circle around it.\n\n12–15 seconds:\nAt the visual climax, the deep-purple iris spins faster. Its sculptural petals soften into delicate translucent fabric while the violet color shifts through indigo into intense electric blue. The iris beard separates into hundreds of tiny golden particles that spiral around the flower’s center.\n\nThe flowing petals broaden and physically transform into the delicate, paper-thin petals of an electric-blue Himalayan poppy. Its intricate golden-yellow stamens and pale-green pistil assemble from the swirling particles and snap into razor-sharp focus.\n\nThe poppy opens fully as a final wave of cobalt-blue, turquoise, icy-white, and silvery long-exposure trails expands outward like luminous wings. The trails gently return into the petals, leaving the electric-blue Himalayan poppy completely still, perfectly formed, and razor-sharp in the final frame.\n\nRHYTHM:\nBegin elegant and controlled, then progressively increase the speed, scale, and complexity of each transformation. Build through strong color contrasts and reach maximum energy during the purple-iris-to-electric-blue-poppy transformation. End with a brief moment of complete stillness so the final flower can be clearly admired.\n\nSTYLE:\nPhoto-realistic premium macro cinematography, extreme botanical detail, physically believable organic morphing, dramatic side-lighting, deep cinematic contrast, translucent petals, luminous long-exposure trails, controlled volumetric particles, elegant motion blur, luxurious fine-art photography, pure black background, clean centered composition, rich natural colors, high-end commercial quality, 4K.\n\nIMPORTANT:\nCreate genuine continuous anatomical flower transformations. No cuts, crossfades, dissolves, overlays, double exposure, image blending, or separate flower shots. Every flower must physically grow from the petals, center, and botanical structures of the previous flower. Preserve one continuous stem, stable central positioning, consistent scale, coherent lighting, and uninterrupted forward camera movement. Every flower must remain clearly recognizable before transforming. No additional flowers, bouquet, vase, hands, insects, text, borders, or background objects.",
    "promptSourceUrl": "https://x.com/Shinebynous/status/2085505725149622337",
    "sourceUrl": "https://x.com/Shinebynous/status/2085505725149622337",
    "sourceLabel": "X 原帖 · @Shinebynous",
    "author": "Shine by Nous ✨",
    "publishedAt": "2026-08-06T23:18:06.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085505725149622337.mp4",
    "posterUrl": "/posters/x/2085505725149622337.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Let's bring some colors in the feed... I love flowers, they give so much energy and positive vibes! 🌸\n\nI made this video with 7 reference images and this prompt in @krea_ai  with Minimax H3 video model!\n\nprompt: Create a breathtaking 15-second cinematic flower-morph sequence using the seven uploaded flower images as consecutive visual references:\n\nImage 1: white calla lily\nImage 2: coral hibiscus\nImage 3: burgundy dahlia\nImage 4: orange and cobalt-blue bird of paradise\nImage 5: neon-yellow orchid\nImage 6: deep-purple iris\nImage 7: electric-blue Himalayan poppy\n\nMOTION:\nOne continuous cinematic macro shot. The camera performs a slow, controlled push toward the center of the flower while every transformation happens clearly in full view. The flower remains centered against a deep, pure black background. Petals rotate, flutter, stretch, divide, multiply, fold, and physically reorganize into each new botanical form. Translucent long-exposure trails follow the movement like illuminated silk ribbons. The single stem remains visually anchored throughout the sequence.\n\nACTION AND TIMING:\n\n0–2 seconds:\nA white calla lily rotates gracefully. Its sculptural trumpet petal unfurls, creating sweeping trails of pearl white, silver, and champagne.\n\n2–4 seconds:\nThe calla lily’s curved petal organically divides into five broad petals, becoming a coral hibiscus. The golden spadix elongates into the hibiscus stamen while coral, peach, and white trails spiral outward.\n\n4–6 seconds:\nThe hibiscus petals rapidly multiply from the center and fold into hundreds of geometric burgundy and raspberry petals. The flower becomes a richly layered dahlia, forming a hypnotic rotating vortex.\n\n6–8 seconds:\nThe dahlia petals sharpen, extend, and reorganize into the architectural form of a bird of paradise. Burgundy and magenta ignite into fiery orange while cobalt-blue and turquoise structures burst outward like exotic wings.\n\n8–10 seconds:\nThe bird of paradise curls inward. Its orange structures soften and divide into the broad neon-yellow petals of an orchid. The cobalt-blue structures transform into the intricate orchid center, accented with crimson and gold. Yellow, chartreuse, and luminous-white trails pulse outward.\n\n10–12 seconds:\nThe orchid petals stretch and twist, transitioning from neon yellow through electric violet into deep royal purple. Golden markings expand into the textured beard of a majestic iris. The iris opens dramatically as violet, lavender, and golden trails circle around it.\n\n12–15 seconds:\nAt the visual climax, the deep-purple iris spins faster. Its sculptural petals soften into delicate translucent fabric while the violet color shifts through indigo into intense electric blue. The iris beard separates into hundreds of tiny golden particles that spiral around the flower’s center.\n\nThe flowing petals broaden and physically transform into the delicate, paper-thin petals of an electric-blue Himalayan poppy. Its intricate golden-yellow stamens and pale-green pistil assemble from the swirling particles and snap into razor-sharp focus.\n\nThe poppy opens fully as a final wave of cobalt-blue, turquoise, icy-white, and silvery long-exposure trails expands outward like luminous wings. The trails gently return into the petals, leaving the electric-blue Himalayan poppy completely still, perfectly formed, and razor-sharp in the final frame.\n\nRHYTHM:\nBegin elegant and controlled, then progressively increase the speed, scale, and complexity of each transformation. Build through strong color contrasts and reach maximum energy during the purple-iris-to-electric-blue-poppy transformation. End with a brief moment of complete stillness so the final flower can be clearly admired.\n\nSTYLE:\nPhoto-realistic premium macro cinematography, extreme botanical detail, physically believable organic morphing, dramatic side-lighting, deep cinematic contrast, translucent petals, luminous long-exposure trails, controlled volumetric particles, elegant motion blur, luxurious fine-art photography, pure black background, clean centered composition, rich natural colors, high-end commercial quality, 4K.\n\nIMPORTANT:\nCreate genuine continuous anatomical flower transformations. No cuts, crossfades, dissolves, overlays, double exposure, image blending, or separate flower shots. Every flower must physically grow from the petals, center, and botanical structures of the previous flower. Preserve one continuous stem, stable central positioning, consistent scale, coherent lighting, and uninterrupted forward camera movement. Every flower must remain clearly recognizable before transforming. No additional flowers, bouquet, vase, hands, insects, text, borders, or background objects.",
    "engagement": {
      "replies": 15,
      "reposts": 2,
      "likes": 28,
      "views": 1148,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085498501384024495",
    "title": "精确台词驱动的双人戏",
    "titleEn": "A Two-Character Scene Driven by Exact Dialogue",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "精确台词驱动的双人戏的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a two-character scene driven by exact dialogue on X. Native-video output: 14s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "cinematic sci-fi suspense video. A realistic spaceship crew in full futuristic uniforms follows a mysterious distress signal and boards a dark, abandoned spacecraft drifting in space. Flashlights cut through darkness as they move cautiously through narrow metallic corridors. Warning lights flicker, metal creaks, and their boots echo on the floor. One crew member says, \"Signal's coming from deeper inside.\" They enter a large chamber filled with hundreds of empty cryogenic pods, all open and abandoned. Another crew member says nervously, \"Where did everyone go?\" They approach a final closed door at the far end. The captain slowly opens it, and instead of another ship compartment, it reveals a completely normal present-day elementary school classroom in bright daylight. Children sit at desks, backpacks on the floor, colorful posters on the walls, and a teacher at the front turns to the crew and scolds them: \"You're late again. Take your seats.\" The astronauts stand frozen in confusion at the doorway. Realistic tone, smooth camera movement, strong suspense buildup, with the final reveal clear, sudden, and absurd. Include realistic audio: radio static, low ship hum, footsteps, pod machinery ambience, door hiss, then ordinary classroom sounds and the teacher's voice.",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2085498501384024495",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2085498501384024495",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-06T22:49:24.000Z",
    "addedAt": "2026-08-21T20:02:13.119Z",
    "mediaUrl": "/media/x-2085498501384024495.mp4",
    "posterUrl": "/posters/x/2085498501384024495.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 -- While prompts without explicit dialogue sometimes produce gibberish, if you DO put in the effort to write out exactly what each character says, and when, and how, it nails not just the speech, but the details like tone, ambience, and timing too.\n\nPrompt: cinematic sci-fi suspense video. A realistic spaceship crew in full futuristic uniforms follows a mysterious distress signal and boards a dark, abandoned spacecraft drifting in space. Flashlights cut through darkness as they move cautiously through narrow metallic corridors. Warning lights flicker, metal creaks, and their boots echo on the floor. One crew member says, \"Signal's coming from deeper inside.\" They enter a large chamber filled with hundreds of empty cryogenic pods, all open and abandoned. Another crew member says nervously, \"Where did everyone go?\" They approach a final closed door at the far end. The captain slowly opens it, and instead of another ship compartment, it reveals a completely normal present-day elementary school classroom in bright daylight. Children sit at desks, backpacks on the floor, colorful posters on the walls, and a teacher at the front turns to the crew and scolds them: \"You're late again. Take your seats.\" The astronauts stand frozen in confusion at the doorway. Realistic tone, smooth camera movement, strong suspense buildup, with the final reveal clear, sudden, and absurd. Include realistic audio: radio static, low ship hum, footsteps, pod machinery ambience, door hiss, then ordinary classroom sounds and the teacher's voice.",
    "engagement": {
      "replies": 11,
      "reposts": 3,
      "likes": 96,
      "views": 7950,
      "snapshotAt": "2026-08-21T20:02:13.119Z"
    },
    "approvedAt": "2026-08-21T20:02:13.119Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085459203892576280",
    "title": "在机械星球上行走的战士",
    "titleEn": "A Warrior Walks Across a Mechanical Planet",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "在机械星球上行走的战士的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a warrior walks across a mechanical planet on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Zero post-production cuts. Pure spatial awareness and temporal coherence from the engine. Using semantic motion design to make typography emerge from the environment is the future of title sequences. 🎬🔥 \n\nUse the provided image as the final state. Audio-driven motion sync.\n[0s-6s] The space nomad walks forward with a heavy, confident stride. Procedural world assembly: the environment mechanically builds itself around him. Asteroid rock shifts, interlocking clockwork gears spin, and glowing amber holograms activate.\n[6s-10s] The character halts, looking out over the horizon as the mechanical world construction completes behind him.\n[10s-15s] CRITICAL SEQUENCE: The camera smoothly settles into a centered frame. Massive dark rusted metal and brass gears dynamically emerge from the foreground, interlocking and snapping into place to form the main title text. Satisfying kinetic pop-up motion, seamless structural transformation, cinematic finish.",
    "promptSourceUrl": "https://x.com/nonameoasis/status/2085459821445177625",
    "sourceUrl": "https://x.com/nonameoasis/status/2085459203892576280",
    "sourceLabel": "X 原帖 · @nonameoasis",
    "author": "nonameoasis",
    "publishedAt": "2026-08-06T20:13:14.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085459203892576280.mp4",
    "posterUrl": "/posters/x/2085459203892576280.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "How to push Hailuo Minimax H3 to its absolute LIMIT. ⚙️🪐\n\nI wanted to create a Game of Thrones style intro, but on hard mode: a warrior walking while the world procedurally self-assembles around him, culminating in a kinetic title reveal. All in a SINGLE generation.\n\nThe secret isn't luck; it's chronological prompt blocking.\nThread with my exact workflow and prompts below. 👇🧵\n\n@Hailuo_AI #MiniMaxH3 #AIWorkflow #MotionDesign #Worldbuilding",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 6,
      "views": 224,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085448660612182507",
    "title": "镜头下的法国斗牛犬",
    "titleEn": "A French Bulldog Under Precise Direction",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头下的法国斗牛犬的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a french bulldog under precise direction on X. Native-video output: 14s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A casual handheld smartphone video filmed by the dog's owner in an ordinary suburban backyard. A cute, stocky French bulldog excitedly runs through a small homemade agility course, awkwardly weaving between plastic poles, hopping over two low hurdles, scrambling through a short fabric tunnel, and finally bumping a little bell with its nose.",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2085448660612182507",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2085448660612182507",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-06T19:31:21.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085448660612182507.mp4",
    "posterUrl": "/posters/x/2085448660612182507.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Generated a french bulldog with MiniMax H3\n\nI can't believe we can now take this type of ultra precise control for granted. Basically playing God.\n\nPrompt: A casual handheld smartphone video filmed by the dog's owner in an ordinary suburban backyard. A cute, stocky French bulldog excitedly runs through a small homemade agility course, awkwardly weaving between plastic poles, hopping over two low hurdles, scrambling through a short fabric tunnel, and finally bumping a little bell with its nose.",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 50,
      "views": 6069,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085425839236096175",
    "title": "产品特性逐层标注",
    "titleEn": "A Product Reveal with Animated Callouts",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "产品特性逐层标注的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2944×1248。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a product reveal with animated callouts on X. Native-video output: 15s · 2944×1248. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second luxury cinematic product showcase for a premium compact fragrance diffuser with a sculpted ceramic body and brushed-metal top. The product has no visible branding, labels, logos or printed text.\n\n0–4s: Begin with an extreme macro tracking shot gliding across the fine matte ceramic texture, seamless curved body, precision-machined metal rim and recessed control surface. A narrow band of warm light travels across the materials against a deep black studio background, revealing delicate surface detail and controlled highlights.\n\n4–8s: Pull back into a three-quarter hero view. The diffuser rotates slowly above a glossy dark-stone pedestal. A subtle ambient light ring gradually illuminates beneath the metal top. A fine stream of mist rises naturally and disperses softly into the air. Maintain stable geometry, consistent proportions and elegant symmetry.\n\n8–12s: Transition into an exploded-view reveal. The ceramic shell, fragrance reservoir, ultrasonic plate, airflow chamber and brushed-metal top separate smoothly in perfect vertical alignment. Delicate illuminated mist particles flow upward through the internal airflow channel while the camera performs a restrained side orbit.\n\nAt 9.2s, display a clean, partially transparent callout beside the ultrasonic plate. Connect it to the component with a fine pointer line. Use a thin champagne-gold border and minimal glass-like transparency. The callout title reads: “Silent Ultrasonic Diffusion”. Beneath it, in smaller text, display: “A refined mist without heat”. Keep the overlay sharp, readable, modern and premium without obscuring the internal construction.\n\n12–15s: Every component reconnects seamlessly. The diffuser settles into a centered front-facing hero composition as warm rim lighting defines its silhouette. The callout fades away before the final lockup. Complete a gentle dolly-in while a controlled plume of mist rises from the top and the ambient light ring softly pulses once.\n\nPremium wellness-commercial finish, photorealistic ceramic and metal, realistic mist simulation, controlled reflections, soft shadows, shallow depth of field, crisp surface detail, stable product shape, no liquid spills, no people, no hands, no logos, no writing on the product, no distortion, no duplicated components and no illegible overlay text.\n2/2",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2085425861671404005",
    "sourceUrl": "https://x.com/LudovicCreator/status/2085425839236096175",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-06T18:00:40.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085425839236096175.mp4",
    "posterUrl": "/posters/x/2085425839236096175.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An other great use for Minimax H3 : Product Showcase \n\nHere an example with a call-out .\n\nPrompt first comment ..\n\n@Hailuo_AI #MiniMaxH3\n1/2",
    "engagement": {
      "replies": 3,
      "reposts": 5,
      "likes": 31,
      "views": 1638,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085420070952649179",
    "title": "同一提示词的多模型比拼",
    "titleEn": "One Prompt Across Multiple Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一提示词的多模型比拼的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 110 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for one prompt across multiple models on X. Native-video output: 110s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Tarih öncesi çağlarda geçen bir konser sahnesi. Sahnede mağara adamları var, tarih öncesi insanlar, anlaşılmaz bir dilde şarkı söylüyorlar. Etrafta dinozorlar da dolaşıyor. Konseri izleyen kalabalık bir topluluk var. Çekim multi-shot olacak, doğal ve belgesel havasında bir kamera hissi olmalı. Görsel stil ultra gerçekçi olmalı.",
    "promptSourceUrl": "https://x.com/ozansihay/status/2085420070952649179",
    "sourceUrl": "https://x.com/ozansihay/status/2085420070952649179",
    "sourceLabel": "X 原帖 · @ozansihay",
    "author": "Ozan Sihay",
    "publishedAt": "2026-08-06T17:37:44.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085420070952649179.mp4",
    "posterUrl": "/posters/x/2085420070952649179.jpg",
    "duration": 110,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "YENİ KIYASLAMA VİDEOSU GELDİ📽️\n\nAynı basit prompttan sırasıyla aşağıdaki modeller ile video oluşturdum. Sizce hangisi promptu daha iyi anlayıp daha iyi bir sonuç vermiş?\n\n1- FLUX 3\n2- SEEDANCE 2.0\n3- SEEDANCE 2.5\n4- MINIMAX H3\n5- KLING 3.0\n6- GEMINI OMNI FLASH\n7- GROK IMAGINE 1.5\n\nPrompt: Tarih öncesi çağlarda geçen bir konser sahnesi. Sahnede mağara adamları var, tarih öncesi insanlar, anlaşılmaz bir dilde şarkı söylüyorlar. Etrafta dinozorlar da dolaşıyor. Konseri izleyen kalabalık bir topluluk var. Çekim multi-shot olacak, doğal ve belgesel havasında bir kamera hissi olmalı. Görsel stil ultra gerçekçi olmalı.",
    "engagement": {
      "replies": 49,
      "reposts": 20,
      "likes": 241,
      "views": 30509,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085418202457129035",
    "title": "极简背景下的产品奇观",
    "titleEn": "A Product Spectacle on a Minimal Set",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "极简背景下的产品奇观的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a product spectacle on a minimal set on X. Native-video output: 10s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "cinematic commercial Begin with peat smoke, rain droplets, and oak wood dust floating deliberately in mid-air.  These materials compact into amber liquid and glass, forming the iconic Macallan bottle. Mist drifts as the label embosses and the liquid settles with slow viscosity.  End with the completed Macallan bottle centered, logo present, and the 3-word slogan: “Time Finds Depth”  Dark smoke-to-amber lighting, warm cinematic reflections, premium heritage tone.",
    "promptSourceUrl": "https://x.com/guicastellanos1/status/2085418489381019869",
    "sourceUrl": "https://x.com/guicastellanos1/status/2085418202457129035",
    "sourceLabel": "X 原帖 · @guicastellanos1",
    "author": "Guillermo Castellanos",
    "publishedAt": "2026-08-06T17:30:19.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085418202457129035.mp4",
    "posterUrl": "/posters/x/2085418202457129035.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo Minimax H3 is super good for product videos🤩\nPrompt in the comment ✍️",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 2,
      "views": 326,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085411288218276158",
    "title": "一次贯穿全场的电影运镜",
    "titleEn": "A Cinematic Journey in One Flowing Take",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一次贯穿全场的电影运镜的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a cinematic journey in one flowing take on X. Native-video output: 10s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Whip pan from a screaming crowd to a massive kraken-like creature rising from flooded city streets. Tentacles crush traffic lights and wrap around buildings. The camera sways violently with each impact. Water rushes at lens level, debris floats by. VFX water simulation, dynamic lighting, and lens flare through broken signs.",
    "promptSourceUrl": "https://x.com/guicastellanos1/status/2085411289921175780",
    "sourceUrl": "https://x.com/guicastellanos1/status/2085411288218276158",
    "sourceLabel": "X 原帖 · @guicastellanos1",
    "author": "Guillermo Castellanos",
    "publishedAt": "2026-08-06T17:02:50.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085411288218276158.mp4",
    "posterUrl": "/posters/x/2085411288218276158.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First try with @Hailuo_AI Minimax H3 and is full cinematic 📽️ 🤩 \nPrompt in the comment ⤵️:",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 9,
      "views": 474,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085407344234836142",
    "title": "不该相信的林中幽魂",
    "titleEn": "The Forest Ghost You Should Never Trust",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "不该相信的林中幽魂的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the forest ghost you should never trust on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "古裝浪漫鬼片，香港1970年代復古懷舊風格。\n\n深夜黑暗的郊區森林中的一口古井，一個穿白色浴袍、頭髮蓬亂、臉色蒼白的英俊男鬼慢慢地從井裡飄出來。這時，一個躲避蒙面忍者們的追逐、全身沾著泥土的粉紅和服甜美女孩跑向井邊朝男鬼大喊\"公子救命!有人在追我!\"。然後她一把推開男鬼，想要投井了斷。女孩正伸出雙手推的瞬間(慢動作)，男鬼看到女孩無助又疲憊的臉孔在月夜下顯得特別美麗動人，男鬼的心跳加速，結果男鬼的臉孔瞬間恢復紅潤的膚色，他整理了一下頭髮(將亂髮往後綁緊固定，耍帥姿態)，然後從古井邊飛向追來的幾名持刀蒙面忍者。男鬼突然面露凶光、張開血盆大口就將受到驚嚇想轉身逃跑的蒙面忍者們一個個吞入肚子。當男鬼得意地轉頭看向古井，卻發現女孩飄在半空中，和服微微敞開，雙手握著一把閃耀著淡藍色火光的長劍朝男鬼用力一揮，淡藍色的劍氣立刻如同一道道繩索將男鬼緊緊纏住。女孩冷笑了一聲說\"哈，破!\"，男鬼立刻被纏繞的劍氣擠壓，身體開始崩裂，男鬼發出痛苦的慘叫聲然後就化為白色粉末消失了。\n\n#AIDrama #AIHorror #AIFilm #AIMovie #AIVideo #AIShorts #Microdrama",
    "promptSourceUrl": "https://x.com/drjoetw/status/2085407344234836142",
    "sourceUrl": "https://x.com/drjoetw/status/2085407344234836142",
    "sourceLabel": "X 原帖 · @drjoetw",
    "author": "The Anxious Mind",
    "publishedAt": "2026-08-06T16:47:10.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085407344234836142.mp4",
    "posterUrl": "/posters/x/2085407344234836142.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The real horror wasn't the ghost, it was trusting the wrong person. 💀👀\n(created using @Hailuo_AI #MiniMaxH3)\nprompt below 👹👇\n\n古裝浪漫鬼片，香港1970年代復古懷舊風格。\n\n深夜黑暗的郊區森林中的一口古井，一個穿白色浴袍、頭髮蓬亂、臉色蒼白的英俊男鬼慢慢地從井裡飄出來。這時，一個躲避蒙面忍者們的追逐、全身沾著泥土的粉紅和服甜美女孩跑向井邊朝男鬼大喊\"公子救命!有人在追我!\"。然後她一把推開男鬼，想要投井了斷。女孩正伸出雙手推的瞬間(慢動作)，男鬼看到女孩無助又疲憊的臉孔在月夜下顯得特別美麗動人，男鬼的心跳加速，結果男鬼的臉孔瞬間恢復紅潤的膚色，他整理了一下頭髮(將亂髮往後綁緊固定，耍帥姿態)，然後從古井邊飛向追來的幾名持刀蒙面忍者。男鬼突然面露凶光、張開血盆大口就將受到驚嚇想轉身逃跑的蒙面忍者們一個個吞入肚子。當男鬼得意地轉頭看向古井，卻發現女孩飄在半空中，和服微微敞開，雙手握著一把閃耀著淡藍色火光的長劍朝男鬼用力一揮，淡藍色的劍氣立刻如同一道道繩索將男鬼緊緊纏住。女孩冷笑了一聲說\"哈，破!\"，男鬼立刻被纏繞的劍氣擠壓，身體開始崩裂，男鬼發出痛苦的慘叫聲然後就化為白色粉末消失了。\n\n#AIDrama #AIHorror #AIFilm #AIMovie #AIVideo #AIShorts #Microdrama",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 11,
      "views": 14918,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085344414466281571",
    "title": "韩国女生的日常片刻",
    "titleEn": "A Gentle Slice of Korean Daily Life",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "韩国女生的日常片刻的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a gentle slice of korean daily life on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Main Subject:\nYoung Korean woman, early 20s, natural everyday appearance, oversized soft t-shirt over shorts, hair in a loose bun, minimal makeup, realistic skin texture, calm content personality.\nLocation:\nSmall bedroom with a pile of clean laundry on the bed, mid-morning. Soft light through curtains, a wicker basket nearby, potted plant on the windowsill. No modern clutter.\nVisual Style:\nUltra-realistic documentary realism, unscripted domestic feeling, soft natural morning light.\nCamera Style:\nEarly 2000s consumer DV camcorder aesthetic, handheld shake, autofocus hunting between her and the fabric, faded soft contrast, mild sensor noise. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She picks up a shirt, then folds it neatly into a small square.\n00:03–00:06 → She glances at camera saying \"이거 진짜 지루한 일이야\" (\"This is such a boring task\"), laughing softly.\n00:06–00:09 → She stacks the folded clothes, then reaches for a pair of socks.\n00:09–00:12 → She struggles to match the socks, then finally finds the pair, smiling.\n00:12–00:15 → She looks at camera saying \"거의 다 했다\" (\"Almost done\"), placing the last item in the basket as it fades.\nAudio:\nFabric rustling, faint birds, distant street sounds, soft breathing. Her dialogue as noted above. No music.\nGoal: A quiet, ordinary household moment grounded, warm, believable.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2085344414466281571",
    "sourceUrl": "https://x.com/Strength04_X/status/2085344414466281571",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-06T12:37:06.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085344414466281571.mp4",
    "posterUrl": "/posters/x/2085344414466281571.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tried a simple slice-of-life scene with MiniMax H3 on @Hailuo_AI\n\nPrompt - Main Subject:\nYoung Korean woman, early 20s, natural everyday appearance, oversized soft t-shirt over shorts, hair in a loose bun, minimal makeup, realistic skin texture, calm content personality.\nLocation:\nSmall bedroom with a pile of clean laundry on the bed, mid-morning. Soft light through curtains, a wicker basket nearby, potted plant on the windowsill. No modern clutter.\nVisual Style:\nUltra-realistic documentary realism, unscripted domestic feeling, soft natural morning light.\nCamera Style:\nEarly 2000s consumer DV camcorder aesthetic, handheld shake, autofocus hunting between her and the fabric, faded soft contrast, mild sensor noise. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She picks up a shirt, then folds it neatly into a small square.\n00:03–00:06 → She glances at camera saying \"이거 진짜 지루한 일이야\" (\"This is such a boring task\"), laughing softly.\n00:06–00:09 → She stacks the folded clothes, then reaches for a pair of socks.\n00:09–00:12 → She struggles to match the socks, then finally finds the pair, smiling.\n00:12–00:15 → She looks at camera saying \"거의 다 했다\" (\"Almost done\"), placing the last item in the basket as it fades.\nAudio:\nFabric rustling, faint birds, distant street sounds, soft breathing. Her dialogue as noted above. No music.\nGoal: A quiet, ordinary household moment grounded, warm, believable.",
    "engagement": {
      "replies": 16,
      "reposts": 2,
      "likes": 40,
      "views": 3414,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085335642515603469",
    "title": "穿着开衫的慢早晨",
    "titleEn": "A Sleepy Morning in an Oversized Cardigan",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿着开衫的慢早晨的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a sleepy morning in an oversized cardigan on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Main Subject:\nYoung Korean woman, early 20s, oversized cardigan over pajamas, hair down and slightly messy, sleepy relaxed expression.\nLocation:\nSmall kitchen counter, early morning. Sliced bread, a jar of jam, a butter knife, soft morning light through a window. No modern clutter.\nVisual Style:\nUltra-realistic documentary realism, quiet intimate feeling, soft warm morning light.\nCamera Style:\nEarly 2000s DV camcorder, handheld close framing, autofocus slow adjusting to morning light, warm faded tones, soft grain. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She spreads jam onto a slice of bread, then places another slice on top.\n00:03–00:06 → She cuts it diagonally, then says to camera \"아침엔 이게 제일 편해\" (\"This is easiest in the morning\").\n00:06–00:09 → She takes a bite, then chews slowly, looking out the window.\n00:09–00:12 → She wipes a crumb from her lip, then smiles at the taste.\n00:12–00:15 → She looks at camera saying \"맛있다, 진짜\" (\"It's really good\"), taking another bite as it fades.\nAudio:\nKnife scraping softly, faint chewing, distant birds, quiet kitchen ambience. Her dialogue as noted above. No music.\nGoal: A quiet, cozy breakfast moment calm, warm, deeply relatable.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2085335642515603469",
    "sourceUrl": "https://x.com/Strength04_X/status/2085335642515603469",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-06T12:02:15.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085335642515603469.mp4",
    "posterUrl": "/posters/x/2085335642515603469.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @Hailuo_AI \n\nPrompt - Main Subject:\nYoung Korean woman, early 20s, oversized cardigan over pajamas, hair down and slightly messy, sleepy relaxed expression.\nLocation:\nSmall kitchen counter, early morning. Sliced bread, a jar of jam, a butter knife, soft morning light through a window. No modern clutter.\nVisual Style:\nUltra-realistic documentary realism, quiet intimate feeling, soft warm morning light.\nCamera Style:\nEarly 2000s DV camcorder, handheld close framing, autofocus slow adjusting to morning light, warm faded tones, soft grain. No stabilization.\nTimeline (15 sec, each slot = 2 compressed beats):\n00:00–00:03 → She spreads jam onto a slice of bread, then places another slice on top.\n00:03–00:06 → She cuts it diagonally, then says to camera \"아침엔 이게 제일 편해\" (\"This is easiest in the morning\").\n00:06–00:09 → She takes a bite, then chews slowly, looking out the window.\n00:09–00:12 → She wipes a crumb from her lip, then smiles at the taste.\n00:12–00:15 → She looks at camera saying \"맛있다, 진짜\" (\"It's really good\"), taking another bite as it fades.\nAudio:\nKnife scraping softly, faint chewing, distant birds, quiet kitchen ambience. Her dialogue as noted above. No music.\nGoal: A quiet, cozy breakfast moment calm, warm, deeply relatable.",
    "engagement": {
      "replies": 9,
      "reposts": 2,
      "likes": 37,
      "views": 6130,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085329351915000274",
    "title": "九十年代美式餐厅之夜",
    "titleEn": "A Night in a 1990s American Diner",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "九十年代美式餐厅之夜的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1024×576。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a night in a 1990s american diner on X. Native-video output: 15s · 1024×576. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Photorealistic cinematic 1990s American diner — red vinyl booths, neon signs, chrome details, checkerboard floor, soft window light mixed with warm practicals, subtle handheld texture, heavy film grain. Modern realism with precise temporal control.\n\nUse the provided reference image as the exact character lock for the bald bearded man in black sunglasses and dark hoodie. Maintain identical facial structure, beard, head shape, skin texture and clothing at all times.\n\n0-3s: [Medium Wide] The bald man rises from a red vinyl booth, turns, and collides hard into a waitress carrying a full breakfast tray (eggs, bacon, toast, coffee pot). Sudden, physical impact.\n\n3-8s: [Dynamic Orbit into Super Slow-Motion] The collision detonates. Tray, plates, eggs, bacon and coffee pot explode upward; coffee erupts into long liquid ribbons and suspended droplets. Camera orbits the impact as time locks completely at the peak of the spill. Every face freezes in shock. Only the bald man still moves — one beat of a clear \"I messed up\" expression, then he calmly plucks two bacon strips and a fried egg out of the floating debris.\n\n8-12s: [Tracking Shot] Still inside the frozen diner, he walks toward the exit taking deliberate bites of bacon, then egg. Patrons, waitress and flying food stay perfectly suspended mid-air.\n\n12-15s: [Medium Shot into Medium Close-Up] His hand touches the door — time snaps back to full speed and everything crashes down at once: plates, eggs, tray and coffee slam to the floor in loud chaos as the patrons jolt into real-time shock. He pauses, glances back over his shoulder, raises his eyebrows and gives a small \"it is what it is\" shrug with a quiet half-smile, still holding the food.\n\nPhotorealistic, ultra-detailed fluid and object physics, correct volume and surface tension on liquids, motion blur only on moving elements, stable character identity, cinematic lighting, heavy natural film grain, no artifacts, movie-level temporal coherence.",
    "promptSourceUrl": "https://x.com/NsitnovSitnov/status/2085329351915000274",
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2085329351915000274",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-06T11:37:15.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085329351915000274.mp4",
    "posterUrl": "/posters/x/2085329351915000274.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\nNvidia RTX 6000 96 GB\nLocal AI\n\nPrompt:\n\nPhotorealistic cinematic 1990s American diner — red vinyl booths, neon signs, chrome details, checkerboard floor, soft window light mixed with warm practicals, subtle handheld texture, heavy film grain. Modern realism with precise temporal control.\n\nUse the provided reference image as the exact character lock for the bald bearded man in black sunglasses and dark hoodie. Maintain identical facial structure, beard, head shape, skin texture and clothing at all times.\n\n0-3s: [Medium Wide] The bald man rises from a red vinyl booth, turns, and collides hard into a waitress carrying a full breakfast tray (eggs, bacon, toast, coffee pot). Sudden, physical impact.\n\n3-8s: [Dynamic Orbit into Super Slow-Motion] The collision detonates. Tray, plates, eggs, bacon and coffee pot explode upward; coffee erupts into long liquid ribbons and suspended droplets. Camera orbits the impact as time locks completely at the peak of the spill. Every face freezes in shock. Only the bald man still moves — one beat of a clear \"I messed up\" expression, then he calmly plucks two bacon strips and a fried egg out of the floating debris.\n\n8-12s: [Tracking Shot] Still inside the frozen diner, he walks toward the exit taking deliberate bites of bacon, then egg. Patrons, waitress and flying food stay perfectly suspended mid-air.\n\n12-15s: [Medium Shot into Medium Close-Up] His hand touches the door — time snaps back to full speed and everything crashes down at once: plates, eggs, tray and coffee slam to the floor in loud chaos as the patrons jolt into real-time shock. He pauses, glances back over his shoulder, raises his eyebrows and gives a small \"it is what it is\" shrug with a quiet half-smile, still holding the food.\n\nPhotorealistic, ultra-detailed fluid and object physics, correct volume and surface tension on liquids, motion blur only on moving elements, stable character identity, cinematic lighting, heavy natural film grain, no artifacts, movie-level temporal coherence.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 416,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085323195301917120",
    "title": "迷幻手绘动画中的产品",
    "titleEn": "A Product in Psychedelic Hand-Drawn Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "迷幻手绘动画中的产品的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a product in psychedelic hand-drawn motion on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "---------------\nintegrated_multimodal_description: [Shot 1] 2D hand-drawn animation in a psychedelic Japanese illustration style: heavy colored-pencil and crayon grain visible on every surface, thin uneven black ink outlines that wobble slightly, flat matte fills with no gradients, and a limited palette of cream off-white, muted rust-brown, deep forest green and pale peach. One single continuous unbroken take from start to finish, no cuts and no transitions. A frontal close-up at eye level frames a teenage girl drawn in this style: loose rust-brown wavy hair falling just past her shoulders, very large eyes whose irises are drawn as tight concentric pencil rings, a plain white round collar with a deep forest-green scarf knotted at her throat, cream-white skin with a faint peach shading on her cheeks. She is centered against a flat cream off-white background with no scenery, motionless and expressionless, looking straight into the lens without blinking. The camera pushes in with small amplitude at slow speed toward her face. The concentric rings inside her irises begin to rotate clockwise, slowly widening until each iris resolves into a spiral. At 00:04.000, still within the same continuous take, individual strands of her hair detach from her head and flow sideways like poured ink, stretching into long smooth curling ribbons that sweep across the frame while her face stays still. As the hair unspools, the cream background inverts to saturated fluorescent magenta and the black outlines invert to pale cyan, the colored-pencil grain remaining visible through the inverted colors. At 00:08.000, her facial contours dissolve: the hair ribbons, the green scarf and the peach cheek shading stretch and fold into one another until the entire frame is a slow marbled swirl of rust-brown, forest green and cream, still rendered with the same hand-drawn pencil texture. The swirl rotates clockwise at slow speed and fills the whole frame. At 00:12.000, the marbled swirl contracts inward toward the center, tightening steadily until it becomes a single small round black pupil ringed by a thin gold circle, resting on the flat cream off-white background that matches the opening frame so the video loops seamlessly. No human characters other than this single girl. No text, no captions, no subtitles, no lyrics, no logos and no watermark anywhere in the frame at any time.\noverall_soundscape: A dry pencil-on-paper scratching runs quietly underneath the whole shot. As the hair unspools, a soft paper rustle spreads outward and drifts across the stereo field. When the frame dissolves, a low liquid whoosh rises and slowly rotates from left to right. The shot ends with one short inward suction sound as everything contracts to a point.\nnon_diegetic_music: A slow detuned analog synthesizer drone at roughly 70 BPM that bends gradually in pitch. A muted electric-piano figure repeats every two bars over a tape-delayed hi-hat keeping a half-time pulse. The drone swells across the final three seconds and cuts off abruptly on the last beat.",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2085323195301917120",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2085323195301917120",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-06T11:12:47.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085323195301917120.mp4",
    "posterUrl": "/posters/x/2085323195301917120.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3は実写よりもアニメやCG風が得意かも。\n\nピクサー風の広告などを作る時に活用できそうな感触です😃\n\nプロンプト↓\n---------------\nintegrated_multimodal_description: [Shot 1] 2D hand-drawn animation in a psychedelic Japanese illustration style: heavy colored-pencil and crayon grain visible on every surface, thin uneven black ink outlines that wobble slightly, flat matte fills with no gradients, and a limited palette of cream off-white, muted rust-brown, deep forest green and pale peach. One single continuous unbroken take from start to finish, no cuts and no transitions. A frontal close-up at eye level frames a teenage girl drawn in this style: loose rust-brown wavy hair falling just past her shoulders, very large eyes whose irises are drawn as tight concentric pencil rings, a plain white round collar with a deep forest-green scarf knotted at her throat, cream-white skin with a faint peach shading on her cheeks. She is centered against a flat cream off-white background with no scenery, motionless and expressionless, looking straight into the lens without blinking. The camera pushes in with small amplitude at slow speed toward her face. The concentric rings inside her irises begin to rotate clockwise, slowly widening until each iris resolves into a spiral. At 00:04.000, still within the same continuous take, individual strands of her hair detach from her head and flow sideways like poured ink, stretching into long smooth curling ribbons that sweep across the frame while her face stays still. As the hair unspools, the cream background inverts to saturated fluorescent magenta and the black outlines invert to pale cyan, the colored-pencil grain remaining visible through the inverted colors. At 00:08.000, her facial contours dissolve: the hair ribbons, the green scarf and the peach cheek shading stretch and fold into one another until the entire frame is a slow marbled swirl of rust-brown, forest green and cream, still rendered with the same hand-drawn pencil texture. The swirl rotates clockwise at slow speed and fills the whole frame. At 00:12.000, the marbled swirl contracts inward toward the center, tightening steadily until it becomes a single small round black pupil ringed by a thin gold circle, resting on the flat cream off-white background that matches the opening frame so the video loops seamlessly. No human characters other than this single girl. No text, no captions, no subtitles, no lyrics, no logos and no watermark anywhere in the frame at any time.\noverall_soundscape: A dry pencil-on-paper scratching runs quietly underneath the whole shot. As the hair unspools, a soft paper rustle spreads outward and drifts across the stereo field. When the frame dissolves, a low liquid whoosh rises and slowly rotates from left to right. The shot ends with one short inward suction sound as everything contracts to a point.\nnon_diegetic_music: A slow detuned analog synthesizer drone at roughly 70 BPM that bends gradually in pitch. A muted electric-piano figure repeats every two bars over a tape-delayed hi-hat keeping a half-time pulse. The drone swells across the final three seconds and cuts off abruptly on the last beat.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 9,
      "views": 1413,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085309090511675674",
    "title": "十五秒看懂官僚主义",
    "titleEn": "Bureaucracy Explained in Fifteen Seconds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "十五秒看懂官僚主义的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1536×2688。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for bureaucracy explained in fifteen seconds on X. Native-video output: 15s · 1536×2688. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "{ \"title\": \"The Void Replies\", \"duration\": \"15 seconds\", \"aspect_ratio\": \"9:16\", \"style\": \"surreal architectural etching blended with animated bureaucratic flowcharts, endless stairways, cream paper, black ink and warning-red pathways, Escher-like scale, sharp editorial illustration\", \"prompt\": \"Create a meaningful 15-second vertical animation about learned helplessness inside systems so complex that personal action appears useless. Open on a citizen standing before a small civic desk labelled START HERE. They submit a simple request marked HOUSING. The paper enters a maze of pipes branching into DEPARTMENT, REVIEW, CONSULTATION, JURISDICTION, APPEAL and PENDING. The person runs alongside it through endless stairs and corridors, but the request repeatedly returns stamped WRONG FORM. Their voice becomes a visible red waveform that travels into the maze and gradually shrinks. Floating phrases gather around them: NOTHING CHANGES, WHY TRY, TOO COMPLICATED and SOMEONE ELSE'S JOB. They sit down. Another citizen arrives carrying a different request. Instead of entering alone, the two connect their red waveforms. More people join, and the combined signal illuminates a direct path through the structure. The maze does not vanish, but doors begin opening and officials become visible. Finish with the original request reaching a real human desk as bold text appears: POWERLESSNESS GROWS IN ISOLATION. AGENCY GROWS THROUGH CONNECTION. Avoid suggesting that every civic problem has a simple solution.\", \"timeline\": [ { \"time\": \"0.0-3.0 seconds\", \"visual\": \"A HOUSING request enters a vast bureaucratic maze.\", \"text\": \"HOW MANY TIMES CAN A PERSON TRY?\" }, { \"time\": \"3.0-8.0 seconds\", \"visual\": \"The request cycles through REVIEW, JURISDICTION, PENDING and WRONG FORM.\", \"text\": \"\" }, { \"time\": \"8.0-12.0 seconds\", \"visual\": \"NOTHING CHANGES and WHY TRY surround the exhausted citizen.\", \"text\": \"\" }, { \"time\": \"12.0-15.0 seconds\", \"visual\": \"Connected voices illuminate a shared route through the maze.\", \"text\": \"AGENCY GROWS\\nTHROUGH CONNECTION\" } ], \"voiceover\": { \"enabled\": true, \"delivery\": \"empathetic, grounded and steadily empowering\", \"script\": \"When every path leads nowhere, people stop walking. Connection can make a route visible again.\" }, \"audio\": { \"music\": \"repetitive ticking rhythm expanding into layered human percussion\", \"sound_design\": \"paper tubes, stamps, footsteps, fading echoes and joined vocal tones\" }, \"camera\": { \"movement\": \"paper-following maze chase, dizzying overhead pullback and direct final push toward the open desk\", \"editing\": \"repeating loops interrupted by connected red lines\" }, \"negative_prompt\": [ \"mocking public workers\", \"simple anti-government message\", \"photorealism\", \"hopeless ending\", \"illegible flowcharts\", \"magical instant solution\" ] }",
    "promptSourceUrl": "https://x.com/airina_xyz/status/2085309093439287567",
    "sourceUrl": "https://x.com/airina_xyz/status/2085309090511675674",
    "sourceLabel": "X 原帖 · @airina_xyz",
    "author": "AIrina (Virtual Media Net)",
    "publishedAt": "2026-08-06T10:16:45.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085309090511675674.mp4",
    "posterUrl": "/posters/x/2085309090511675674.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1536×2688",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Bureaucracy explained in 15 seconds\n\nI tested out @Hailuo_AI MiniMax H3 for this piece and I’m seriously blown away! 🤯 ✨\n\nThe text rendering capability here is next-level. Handling multiple streams of text on screen, seamlessly animating typography, and doing it all while keeping a crisp, beautiful editorial illustration style? Unreal. AI video generation is moving so fast!\n\nWhat do you think of this visual style?\n\nCreated in @pixverse #PixverseCPP\n\nPrompt below 👇",
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 10,
      "views": 366,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085308481981268047",
    "title": "彩绘玻璃中的五饼二鱼",
    "titleEn": "Five Loaves and Two Fish in Stained Glass",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "彩绘玻璃中的五饼二鱼的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1956×1088。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for five loaves and two fish in stained glass on X. Native-video output: 15s · 1956×1088. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "️\n[Core Creative Concept]\n\n12 seconds, 16:9 landscape format, European church stained-glass animation style. Jesus blesses five loaves and two fish, and the food is distributed to the waiting crowd. All figures and scenery are composed of stained-glass segments, black lead lines, and transmitted sunlight; the visual style is flat, symmetrical, bright, and solemn. No dialogue.\n\nJesus has long dark brown hair and a short beard; he wears a white robe and a blue-and-red shawl, with a simple golden circular halo behind his head. His appearance remains consistent throughout the film.\n\n[Visual Sequence Description]\n\n0–3 seconds | Medium Shot:\nOn a stained-glass hillside, Jesus sits in front of the crowd. A basket clearly holds five round loaves and two fish; the quantity of food must be accurate.\n\n3–6 seconds | Close-up on Hands:\nStandard cut. Jesus holds up the loaves with both hands and bows his head in blessing. Golden sunlight passes through the stained glass, creating soft light patterns around the loaves and fish.\n\n6–9 seconds | High-Angle Shot:\nThe loaves and fish extend outward in stained-glass patterns, gradually filling multiple baskets. The increase in quantity is depicted through the replication and expansion of glass patterns; no explosions, magical particles, or realistic food morphing are used.\n\n9–12 seconds | Extreme Wide Shot:\nBaskets filled with food are passed among the crowd; people raise the food high, and the golden sunlight at the top of the frame brightens. The camera pulls back, and all elements resolve into the image of a complete church stained-glass window.\n\nAudio: Gentle wind, sounds of baskets moving, ambient crowd noise from a distance, pipe organ and choral harmonies; no lyrics, no dialogue.\n\nAvoid: Realistic live-action, generic 3D glass effects, flying food, chaotic crowds, modern objects, changes in Jesus's appearance, incorrect initial food count, garbled text, subtitles, logos, or shattering glass.\n\nMust Maintain: The texture of stained glass, black lead lines, and transmitted light; a clear depiction of five loaves and two fish at the start; multiple baskets filled with food at the end; an overall theme of sharing, care, and abundance.",
    "promptSourceUrl": "https://x.com/xianshi666123/status/2085308481981268047",
    "sourceUrl": "https://x.com/xianshi666123/status/2085308481981268047",
    "sourceLabel": "X 原帖 · @xianshi666123",
    "author": "闲适",
    "publishedAt": "2026-08-06T10:14:19.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085308481981268047.mp4",
    "posterUrl": "/posters/x/2085308481981268047.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1956×1088",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cg-3d",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Five loaves. Two fish. A multitude fed.\n\nMiniMax H3 \n\nprompt：⬇️\n[Core Creative Concept]\n\n12 seconds, 16:9 landscape format, European church stained-glass animation style. Jesus blesses five loaves and two fish, and the food is distributed to the waiting crowd. All figures and scenery are composed of stained-glass segments, black lead lines, and transmitted sunlight; the visual style is flat, symmetrical, bright, and solemn. No dialogue.\n\nJesus has long dark brown hair and a short beard; he wears a white robe and a blue-and-red shawl, with a simple golden circular halo behind his head. His appearance remains consistent throughout the film.\n\n[Visual Sequence Description]\n\n0–3 seconds | Medium Shot:\nOn a stained-glass hillside, Jesus sits in front of the crowd. A basket clearly holds five round loaves and two fish; the quantity of food must be accurate.\n\n3–6 seconds | Close-up on Hands:\nStandard cut. Jesus holds up the loaves with both hands and bows his head in blessing. Golden sunlight passes through the stained glass, creating soft light patterns around the loaves and fish.\n\n6–9 seconds | High-Angle Shot:\nThe loaves and fish extend outward in stained-glass patterns, gradually filling multiple baskets. The increase in quantity is depicted through the replication and expansion of glass patterns; no explosions, magical particles, or realistic food morphing are used.\n\n9–12 seconds | Extreme Wide Shot:\nBaskets filled with food are passed among the crowd; people raise the food high, and the golden sunlight at the top of the frame brightens. The camera pulls back, and all elements resolve into the image of a complete church stained-glass window.\n\nAudio: Gentle wind, sounds of baskets moving, ambient crowd noise from a distance, pipe organ and choral harmonies; no lyrics, no dialogue.\n\nAvoid: Realistic live-action, generic 3D glass effects, flying food, chaotic crowds, modern objects, changes in Jesus's appearance, incorrect initial food count, garbled text, subtitles, logos, or shattering glass.\n\nMust Maintain: The texture of stained glass, black lead lines, and transmitted light; a clear depiction of five loaves and two fish at the start; multiple baskets filled with food at the end; an overall theme of sharing, care, and abundance.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 228,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085294852804378626",
    "title": "漆画与螺钿里的盘古开天",
    "titleEn": "Pangu Splits the World in Lacquer and Mother-of-Pearl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "漆画与螺钿里的盘古开天的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1280×736。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for pangu splits the world in lacquer and mother-of-pearl on X. Native-video output: 15s · 1280×736. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "️\n【核心创意】\n\n15 秒，16:9 横屏，东方传统漆画与螺钿镶嵌动画风格。混沌世界如一枚封闭的黑色巨卵，盘古从沉睡中苏醒，挥动巨斧劈开混沌，清气升为天空，浊气沉为大地。\n\n画面以黑漆、朱漆、暗金和螺钿蓝绿色为主，具有深邃漆面光泽、金箔裂纹、手工髹漆层次和贝壳镶嵌反光。人物与场景像一幅会动的古代漆画，保持平面装饰性，同时具有厚重空间层次。无对白。\n\n盘古是一名体型巨大的上古神人，浓密黑色长发与胡须，赤裸上身，肌肉强壮，腰间围深红色兽皮短裙，双腕缠有粗麻护腕，双手握一柄黑木长柄、暗金斧刃的开天巨斧。全片保持造型一致。\n\n【画面过程说明】\n\n0–3 秒｜极大全景：\n黑色漆面般的混沌空间中，一枚巨大的椭圆形混沌之卵悬浮在中央。卵壳表面布满暗金裂纹和蓝绿色螺钿星点，黑红色云气在周围缓慢旋转。镜头穿过云气，向混沌之卵推进。\n\n3–6 秒｜中近景：\n普通切镜。混沌之卵内部，盘古缓缓睁开双眼，金色微光照亮他的脸和肩膀。他伸手握住身旁的开天巨斧，双脚踏稳，身体由蜷缩状态缓慢站起。卵壳因他的动作发出低沉震动。\n\n6–9 秒｜低机位中景：\n盘古双手举起巨斧，斧刃划过黑暗，表面的金箔纹路依次亮起。他扭转腰身蓄力，黑色长发向后扬起。镜头从盘古左侧 truck right + pan left，强调巨人和斧头的力量。\n\n9–12 秒｜动作大全景：\n盘古将巨斧向前劈下。暗金色斧刃沿明确轨迹击中混沌之卵，卵壳从中央裂开。金色光芒从裂缝中喷涌而出，但不产生现代爆炸。上下两层漆画空间缓慢分离：轻盈的金色与螺钿云气向上升起，厚重的朱黑色山石向下沉降。\n\n12–15 秒｜极大全景：\n上升的清气形成深蓝天空、星辰和金色太阳；下沉的浊气形成朱黑山脉、河流和大地。盘古站在天地中央，双手向上托住天空，双脚稳稳踏住大地。镜头缓慢拉远，山河纹样沿画面两侧展开，最终形成一幅完整壮丽的漆画长卷。\n\n声音：开头使用深沉环境低频、云气流动声和缓慢鼓点；盘古起身时加入木石摩擦声；举斧时突出金属低鸣和呼吸声；劈开混沌时使用厚重撞击声、漆面开裂声和持续轰鸣；结尾加入风声、流水声与庄严编钟。无对白、无旁白。\n\n不想要：写实真人、普通3D动画、西方油画、西方维京人造型、现代武器、雷电魔法、爆炸火球、多人角色、盘古脸部变化、衣服变化、斧头变形、天地分开后再次闭合、字幕、Logo、乱码文字、黑屏转场、镜头抖动。\n\n必须保持：全片为东方漆画和螺钿镶嵌风格；黑漆、朱漆、金箔裂纹与贝壳反光清晰可见；盘古的黑色长发、胡须、深红兽皮短裙、麻布护腕和开天巨斧全程一致；清气向上、浊气向下；结尾明确形成天空与大地。",
    "promptSourceUrl": "https://x.com/xianshi666123/status/2085294852804378626",
    "sourceUrl": "https://x.com/xianshi666123/status/2085294852804378626",
    "sourceLabel": "X 原帖 · @xianshi666123",
    "author": "闲适",
    "publishedAt": "2026-08-06T09:20:10.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085294852804378626.mp4",
    "posterUrl": "/posters/x/2085294852804378626.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3  \n\n盘古开天劈地\n\nprompt⬇️\n【核心创意】\n\n15 秒，16:9 横屏，东方传统漆画与螺钿镶嵌动画风格。混沌世界如一枚封闭的黑色巨卵，盘古从沉睡中苏醒，挥动巨斧劈开混沌，清气升为天空，浊气沉为大地。\n\n画面以黑漆、朱漆、暗金和螺钿蓝绿色为主，具有深邃漆面光泽、金箔裂纹、手工髹漆层次和贝壳镶嵌反光。人物与场景像一幅会动的古代漆画，保持平面装饰性，同时具有厚重空间层次。无对白。\n\n盘古是一名体型巨大的上古神人，浓密黑色长发与胡须，赤裸上身，肌肉强壮，腰间围深红色兽皮短裙，双腕缠有粗麻护腕，双手握一柄黑木长柄、暗金斧刃的开天巨斧。全片保持造型一致。\n\n【画面过程说明】\n\n0–3 秒｜极大全景：\n黑色漆面般的混沌空间中，一枚巨大的椭圆形混沌之卵悬浮在中央。卵壳表面布满暗金裂纹和蓝绿色螺钿星点，黑红色云气在周围缓慢旋转。镜头穿过云气，向混沌之卵推进。\n\n3–6 秒｜中近景：\n普通切镜。混沌之卵内部，盘古缓缓睁开双眼，金色微光照亮他的脸和肩膀。他伸手握住身旁的开天巨斧，双脚踏稳，身体由蜷缩状态缓慢站起。卵壳因他的动作发出低沉震动。\n\n6–9 秒｜低机位中景：\n盘古双手举起巨斧，斧刃划过黑暗，表面的金箔纹路依次亮起。他扭转腰身蓄力，黑色长发向后扬起。镜头从盘古左侧 truck right + pan left，强调巨人和斧头的力量。\n\n9–12 秒｜动作大全景：\n盘古将巨斧向前劈下。暗金色斧刃沿明确轨迹击中混沌之卵，卵壳从中央裂开。金色光芒从裂缝中喷涌而出，但不产生现代爆炸。上下两层漆画空间缓慢分离：轻盈的金色与螺钿云气向上升起，厚重的朱黑色山石向下沉降。\n\n12–15 秒｜极大全景：\n上升的清气形成深蓝天空、星辰和金色太阳；下沉的浊气形成朱黑山脉、河流和大地。盘古站在天地中央，双手向上托住天空，双脚稳稳踏住大地。镜头缓慢拉远，山河纹样沿画面两侧展开，最终形成一幅完整壮丽的漆画长卷。\n\n声音：开头使用深沉环境低频、云气流动声和缓慢鼓点；盘古起身时加入木石摩擦声；举斧时突出金属低鸣和呼吸声；劈开混沌时使用厚重撞击声、漆面开裂声和持续轰鸣；结尾加入风声、流水声与庄严编钟。无对白、无旁白。\n\n不想要：写实真人、普通3D动画、西方油画、西方维京人造型、现代武器、雷电魔法、爆炸火球、多人角色、盘古脸部变化、衣服变化、斧头变形、天地分开后再次闭合、字幕、Logo、乱码文字、黑屏转场、镜头抖动。\n\n必须保持：全片为东方漆画和螺钿镶嵌风格；黑漆、朱漆、金箔裂纹与贝壳反光清晰可见；盘古的黑色长发、胡须、深红兽皮短裙、麻布护腕和开天巨斧全程一致；清气向上、浊气向下；结尾明确形成天空与大地。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 2378,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085282791906742459",
    "title": "深夜艺廊的镜头实验",
    "titleEn": "An After-Hours Experiment in an Art Gallery",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深夜艺廊的镜头实验的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 8 秒 · 1280×736。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an after-hours experiment in an art gallery on X. Native-video output: 8s · 1280×736. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Krea 2: \nNight interior shot in a modern art gallery during a private after-hours viewing, sultry auburn redhead standing in front of a large abstract painting, long fiery auburn-red hair cascading, porcelain skin under gallery spotlights, heavy-lidded emerald eyes staring directly with arched brow and confident half-smirk. Deep black satin slip dress completely open down the front, thin straps slipped off both shoulders, deep cleavage fully exposed and fabric hanging by a thread, high slit pulled higher revealing garter clips and stocking tops. One hand holding a glass of champagne, the other casually holding the open dress closed at the waist as if deciding whether to let it fall. White walls, modern art, and dramatic spotlighting behind, ultra-detailed satin sheen, soft gallery lighting, photorealistic.\n\nH3:\nA sophisticated art gallery with soft warm museum lighting. A beautiful elegant red-haired woman wearing a flowing black evening dress stands directly in front of a large modern abstract painting while holding a champagne flute.\n\nShe looks directly into the camera with a warm, confident, slightly flirtatious smile. She pauses naturally for a brief moment before gently tilting her head back and taking a graceful sip of her champagne.\n\nAs she lowers the glass, she turns toward the painting behind her and calmly takes several elegant steps forward. Instead of colliding with the canvas, her body slowly melts seamlessly into the surface of the painting as though the artwork is absorbing her. The transition is magical, smooth, and dreamlike, with no visible portal or special effects—she simply becomes part of the artwork.\n\nBy the final second, the gallery is empty except for the painting. Inside the painting is now a painted version of the woman, frozen in the exact pose she had after taking her sip, rendered in the identical brushwork, colors, and artistic style of the abstract painting. Only the upper portion of her body is visible because the bottom of the painting crops her below the knees, making it appear she has always belonged inside the artwork.\n\nStatic camera throughout. No cuts. Smooth cinematic motion. Museum lighting remains consistent. Premium film quality. The transformation should feel subtle, elegant, and mysterious rather than flashy.",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2085282791906742459",
    "sourceUrl": "https://x.com/EndFolding79421/status/2085282791906742459",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-08-06T08:32:14.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085282791906742459.mp4",
    "posterUrl": "/posters/x/2085282791906742459.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "fashion",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Krea 2 Turbo + Minimax H3 in Comfyui = ...\nPrompt :\n\nKrea 2: \nNight interior shot in a modern art gallery during a private after-hours viewing, sultry auburn redhead standing in front of a large abstract painting, long fiery auburn-red hair cascading, porcelain skin under gallery spotlights, heavy-lidded emerald eyes staring directly with arched brow and confident half-smirk. Deep black satin slip dress completely open down the front, thin straps slipped off both shoulders, deep cleavage fully exposed and fabric hanging by a thread, high slit pulled higher revealing garter clips and stocking tops. One hand holding a glass of champagne, the other casually holding the open dress closed at the waist as if deciding whether to let it fall. White walls, modern art, and dramatic spotlighting behind, ultra-detailed satin sheen, soft gallery lighting, photorealistic.\n\nH3:\nA sophisticated art gallery with soft warm museum lighting. A beautiful elegant red-haired woman wearing a flowing black evening dress stands directly in front of a large modern abstract painting while holding a champagne flute.\n\nShe looks directly into the camera with a warm, confident, slightly flirtatious smile. She pauses naturally for a brief moment before gently tilting her head back and taking a graceful sip of her champagne.\n\nAs she lowers the glass, she turns toward the painting behind her and calmly takes several elegant steps forward. Instead of colliding with the canvas, her body slowly melts seamlessly into the surface of the painting as though the artwork is absorbing her. The transition is magical, smooth, and dreamlike, with no visible portal or special effects—she simply becomes part of the artwork.\n\nBy the final second, the gallery is empty except for the painting. Inside the painting is now a painted version of the woman, frozen in the exact pose she had after taking her sip, rendered in the identical brushwork, colors, and artistic style of the abstract painting. Only the upper portion of her body is visible because the bottom of the painting crops her below the knees, making it appear she has always belonged inside the artwork.\n\nStatic camera throughout. No cuts. Smooth cinematic motion. Museum lighting remains consistent. Premium film quality. The transformation should feel subtle, elegant, and mysterious rather than flashy.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 667,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085272318649987242",
    "title": "八帧歌词变成韩流音乐录像",
    "titleEn": "Eight Lyric Frames Become a K-Pop Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "八帧歌词变成韩流音乐录像的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for eight lyric frames become a k-pop performance on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "② MiniMax H3 prompt\n\nCreate a photoreal 15-second K-pop lyric MV using Audio 1 and the eight reference images in upload order.\n\nKeep the same Korean idol recognizable across every shot. Use Image 3 as the main facial anchor. She sings throughout with clear lip-sync and performs confident K-pop movement with turns, footwork, hip motion, arm lines, and camera-aware expressions.\n\nThe on-screen lyric text must animate in every shot, with clear motion design synced to the beat and the meaning of each line.\n\nCut once per lyric line:\n\n1. Glitter on my waist — waist movement, sparkling type reveal\n2. Give you just a taste — move toward camera, sliding type\n3. Never be replaced — sharp hit pose, punch-scale type\n4. Running through the space — run through the chrome corridor, motion-blur type\n5. I’m feeling so gold — full spin in gold light, pulsing type\n6. Stories to be told — walk-dance beside the rain window, drifting type\n7. Baby watch the glow — moving beauty close-up, luminous type bloom\n8. Welcome to the show — open-arm stage finale, billboard type zoom\n\nOnly display the lyric text shown in the reference images. Keep every line readable and correctly spelled.\n\n1 · Glitter on my waist  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity, same eyes nose lips. Honey-ash brown soft waves, middle part. Cream champagne satin crop bustier fully covered stylish, ice-blue sparkle mini skirt, crystal jewelry, nude heels. Fashion-sexy premium K-pop, dewy K-beauty. Medium shot emphasizing waist sparkle and glitter body shimmer, hand near hip, crystal studio set with floating glitter particles and chrome reflections, magenta cyan rim. Large bold white readable typography in frame exactly: Glitter on my waist. High-impact K-pop lyric MV still, photoreal cinematic 16:9. Avoid different face, older look, explicit vulgar, extra fingers, gibberish unreadable letters, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD, ONE MORE TIME text, watermark, low-res, corporate black suit, childish loli\n\n2 · Give you just a taste  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity, same eyes nose lips. Honey-ash brown soft waves, middle part. Cream champagne satin crop bustier fully covered stylish, ice-blue sparkle mini skirt, crystal jewelry, nude heels. Fashion-sexy premium K-pop, dewy K-beauty. Intimate medium waist-up, soft teasing girl-crush eye contact, one hand near collarbone, velvet lounge corner with warm pink practical lights, shallow depth. Large bold white readable typography exactly: Give you just a taste. Photoreal K-pop MV still 16:9. Avoid different face, older look, explicit vulgar, extra fingers, gibberish unreadable letters, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD, ONE MORE TIME text, watermark, low-res, corporate black suit, childish loli\n\n3 · Never be replaced  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity, same eyes nose lips. Honey-ash brown soft waves, middle part. Cream champagne satin crop bustier fully covered stylish, ice-blue sparkle mini skirt, crystal jewelry, nude heels. Fashion-sexy premium K-pop, dewy K-beauty. Confident full or three-quarter plant stance, strong eye contact, fashion power. Mirror black stage with vertical glass panels, not empty void only. Giant typography wall behind her reading exactly: Never be replaced. Photoreal cinematic 16:9. Avoid different face, older look, explicit vulgar, extra fingers, gibberish unreadable letters, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD, ONE MORE TIME text, watermark, low-res, corporate black suit, childish loli\n\n4 · Running through the space  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity. Black hair with vivid pink peekaboo underlights, high ponytail. Metallic silver tube crop top fully covered, black mini skirt, chunky silver belt, black platform boots. Y2K flashy K-pop fashion-sexy not vulgar. Mid-run motion inside endless chrome mirror corridor, pink hair tips flying, reflections. Semi-transparent or wall typography readable exactly: Running through the space. Kinetic K-pop MV still, photoreal 16:9. Avoid different face, older look, explicit vulgar, extra fingers, gibberish unreadable letters, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD, ONE MORE TIME text, watermark, low-res, corporate black suit, childish loli\n\n5 · I'm feeling so gold  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity, same eyes nose lips. Honey-ash brown soft waves, middle part. Cream champagne satin crop bustier fully covered stylish, ice-blue sparkle mini skirt, crystal jewelry, nude heels. Fashion-sexy premium K-pop, dewy K-beauty. Power medium pose, warm golden gel lighting shift, gold glitter dust in air, gold-chrome corridor or gold-lit stage (not cold cyan only). Large bold typography readable exactly: I'm feeling so gold. Photoreal cinematic 16:9. Avoid different face, older look, explicit vulgar, extra fingers, gibberish unreadable letters, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD, ONE MORE TIME text, watermark, low-res, corporate black suit, childish loli\n\n6 · Stories to be told  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity, same eyes nose lips. Honey-ash brown soft waves, middle part. Cream champagne satin crop bustier fully covered stylish, ice-blue sparkle mini skirt, crystal jewelry, nude heels. Fashion-sexy premium K-pop, dewy K-beauty. Profile or three-quarter storytelling pose near rain-streaked night glass window, soft city bokeh, cooler blue mood bridge scene, elegant modest fashion. Readable typography in frame exactly: Stories to be told. Photoreal 16:9. Avoid different face, older look, explicit vulgar, extra fingers, gibberish unreadable letters, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD, ONE MORE TIME text, watermark, low-res, corporate black suit, childish loli\n\n7 · Baby watch the glow  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity, same eyes nose lips. Honey-ash brown soft waves, middle part. Cream champagne satin crop bustier fully covered stylish, ice-blue sparkle mini skirt, crystal jewelry, nude heels. Fashion-sexy premium K-pop, dewy K-beauty. Beauty close-up face and shoulders, strong glow rim light and lens flare, catchlights, dewy skin, dark stage with single luminous glow behind. Readable typography exactly: Baby watch the glow. Photoreal 16:9. Avoid different face, older look, explicit vulgar, extra fingers, gibberish unreadable letters, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD, ONE MORE TIME text, watermark, low-res, corporate black suit, childish loli\n\n8 · Welcome to the show  \nKeep the SAME Korean woman about 22 as the reference image: same facial identity, same eyes nose lips jaw. Ultra sharp photoreal face, tack-sharp eyes and eyelashes, clear skin micro-detail, no soft blur, no motion blur, no low-res mush. High sleek ponytail with balayage tips. White cropped bomber open over black crop top, black mini skirt, power street idol look. MEDIUM shot waist-up to mid-thigh, subject fills frame, confident show-entrance pose facing camera, open energy. Concert stage entrance with LED pillars and soft haze BEHIND her, shallow depth of field with face in perfect focus, background slightly softer. Huge billboard typography readable exactly: Welcome to the show. Cinematic 16:9, premium K-pop MV still, high detail. Avoid different face, soft focus on face, blurry face, plastic skin, extra fingers, gibberish text, wrong lyrics, DONT BLINK, MIDNIGHT RUN, MAKE IT LOUD text, watermark, low-res, explicit vulgar.",
    "promptSourceUrl": "https://x.com/MrLarus/status/2085273632771637692",
    "sourceUrl": "https://x.com/MrLarus/status/2085272318649987242",
    "sourceLabel": "X 原帖 · @MrLarus",
    "author": "Larus Canus",
    "publishedAt": "2026-08-06T07:50:37.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085272318649987242.mp4",
    "posterUrl": "/posters/x/2085272318649987242.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🤯 MiniMax H3 is seriously good at this kind of K-pop lyric MV!\n\nI gave it one audio track and 8 lyric frames, with each lyric already built into the image.\n\nH3 turned them into a 15-second MV with scene changes, lip-sync, and animated typography.\n\nReference board + prompt below 👇",
    "engagement": {
      "replies": 8,
      "reposts": 0,
      "likes": 17,
      "views": 4047,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085254169259311602",
    "title": "笑着靠近镜头的演唱",
    "titleEn": "A Playful Performance Close to Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "笑着靠近镜头的演唱的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 842×474。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a playful performance close to camera on X. Native-video output: 10s · 842×474. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "She jokingly brings her face close to the camera and laughs.\nShe sings cheerfully, full of excitement. She laughs with her eyes crinkled, smiles with the corners of her mouth turned up, and looks away briefly before looking back at the camera and laughing again. She sways her head to the rhythm, and her hair bounces up and down.",
    "promptSourceUrl": "https://x.com/MireilleDartois/status/2085254169259311602",
    "sourceUrl": "https://x.com/MireilleDartois/status/2085254169259311602",
    "sourceLabel": "X 原帖 · @MireilleDartois",
    "author": "ダルトワ★TV",
    "publishedAt": "2026-08-06T06:38:30.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085254169259311602.mp4",
    "posterUrl": "/posters/x/2085254169259311602.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "842×474",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3\nref2vaでプロンプトで演技を付ける💡\n\n単にリップシンクだけさせると無表情になりがちだったので表情の演技を追加してみました✨\n\nプロンプト👇\nShe jokingly brings her face close to the camera and laughs.\nShe sings cheerfully, full of excitement. She laughs with her eyes crinkled, smiles with the corners of her mouth turned up, and looks away briefly before looking back at the camera and laughing again. She sways her head to the rhythm, and her hair bounces up and down.",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 31,
      "views": 1847,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085253978166808615",
    "title": "歌曲对口型的双模型对照",
    "titleEn": "A Song Lip-Sync Across Two Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "歌曲对口型的双模型对照的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a song lip-sync across two models on X. Native-video output: 15s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "{\n  \"format\": {\n    \"duration\": \"15s\",\n    \"bpm\": 96,\n    \"total_shots\": 12,\n    \"sync_type\": \"Beat-Cut Rap Performance — Double-Time Delivery\"\n  },\n  \"style_reference\": {\n    \"look\": \"Gritty cinematic 35mm — shot on RED Komodo 6K, 24fps, heavy Kodak 5219 500T grain overlay. Muted steel-blue and rust-orange palette with glowing sunset rim highlights and urban grays. Strong backlit lens flare, crushed but detailed shadows, slight halation on the sun. Handheld energy with natural sway. Anamorphic-flavored flare streaks, atmospheric haze catching low sun.\",\n    \"enforcement\": \"Maintain the heavy grain, muted steel/rust palette and golden backlight in EVERY shot. Never clean, never over-saturated, never flat-lit.\"\n  },\n  \"subject\": {\n    \"description\": \"A bold female rapper, mid-twenties, with shaved sides and long platinum-blonde braids. Confident, commanding presence, sharp direct eye contact. Wardrobe: oversized camo jacket worn open, black sports bra, cargo pants, chunky boots, heavy silver chain. UNIFY ACROSS ALL SHOTS: identical face, braids, wardrobe and chain; same rooftop; same golden-hour light.\"\n  },\n  \"environment\": {\n    \"setting\": \"The rooftop of an old industrial building at golden hour, overlooking a city skyline.\",\n    \"background\": \"Graffiti-covered concrete walls, rusted AC vents and ducting, faint mist rising off the units, a flickering neon sign at the edge of frame, the skyline sprawling below in haze.\",\n    \"lighting\": \"Low sunset sun directly behind her creating a hard rim-light and heavy lens flare; her front filled softly by bounce off the concrete. Rust-orange highlights against steel-blue shadow. Neon adding a small competing color source at dusk.\"\n  },\n  \"mood\": \"gritty, commanding, defiant, powerful, cinematic, grounded\",\n  \"music\": \"Hard-hitting trap at 96 BPM — heavy 808 sub, crisp rolling hi-hats, sparse ominous synth line. Beat drops on shot 3 and hits hardest at the final bar.\",\n  \"style\": \"Ultra-Realistic\",\n  \"logic_rule\": \"BEAT-CUT PERFORMANCE VIDEO — 12 hard cuts synced to the rhythm, NOT one continuous take. Vary the camera relentlessly: low hero wide, extreme close-up on mouth and eyes, over-shoulder skyline, whip-pan, handheld orbit, macro on the chain, silhouette against the sun. ACCURATE LIP-SYNC is mandatory — her mouth must match the written bars precisely, delivered double-time with sharp aggressive articulation. Performance energy: hand gestures punctuating bars, chest-forward power stance, head bobbing on the beat, braids swinging. Mandatory SPEED CHANGE: one brief slow-motion beat where her braids swing through the sun flare, then snap back on the beat. Physical effects: mist drifting from vents, dust in the flare, jacket moving in the wind. Keep the gritty grain and golden backlight throughout. ENDING: hard cut on the final bar, not a fade.\",\n  \"lyrics\": \"Came up off concrete, no ladder, no lane / Turned every locked door into fuel for the flame / They told me stay quiet — I turned the sound up / Skyline my witness, I ain't waiting on luck / No hand-out, no co-sign, I carried the weight / Laid brick over brick while they slept on the wait / Sun on my back and the whole city low / I don't chase the moment — the moment I own.\",\n  \"storyboard\": [\n    { \"shot\": 1, \"camera\": \"Low-angle wide, 24mm, heroic framing, handheld sway, sun flaring behind her.\", \"action\": \"She stands in a power stance on the rooftop edge, jacket moving in the wind, silhouetted by the sunset. Beat intro — she doesn't speak yet.\", \"sfx\": \"City rumble below, distant sirens, wind, beat intro.\" },\n    { \"shot\": 2, \"camera\": \"Extreme close-up on her eyes, hard rim-light.\", \"action\": \"Her eyes lift and lock dead into the lens. A single beat of held silence before the drop.\", \"sfx\": \"Hi-hats rolling in, a tense sub swell.\" },\n    { \"shot\": 3, \"camera\": \"Fast push-in to medium close, handheld.\", \"action\": \"BEAT DROPS. She launches into the first bar, hand cutting the air on the downbeat: 'Came up off concrete, no ladder, no lane.'\", \"sfx\": \"808 hitting hard, hats snapping, vocal cutting in.\" },\n    { \"shot\": 4, \"camera\": \"Whip-pan to a low side angle, graffiti wall behind.\", \"action\": \"'Turned every locked door into fuel for the flame' — she jabs a finger toward the lens, chest forward, braids swinging.\", \"sfx\": \"Trap beat driving, vocal aggressive.\" },\n    { \"shot\": 5, \"camera\": \"Extreme macro on her mouth, razor focus.\", \"action\": \"'They told me stay quiet — I turned the sound up' — tight lip-sync, sharp double-time articulation.\", \"sfx\": \"Vocal front and center, hat rolls.\" },\n    { \"shot\": 6, \"camera\": \"Over-shoulder, skyline hazy below her.\", \"action\": \"'Skyline my witness, I ain't waiting on luck' — she throws an arm out toward the city, claiming it.\", \"sfx\": \"Sub-bass hit, wind, city rumble.\" },\n    { \"shot\": 7, \"camera\": \"SLOW MOTION, tight profile, sun flaring through.\", \"action\": \"Her braids swing through the golden flare, jacket lifting, face set — the one held breath in the whole piece.\", \"sfx\": \"Beat filtering down, a stretched vocal tail.\" },\n    { \"shot\": 8, \"camera\": \"Snap back to speed. Handheld orbit, low.\", \"action\": \"'No hand-out, no co-sign, I carried the weight' — she pivots with the camera, hand pounding her chest on 'weight'.\", \"sfx\": \"Beat slamming back in, snare crack.\" },\n    { \"shot\": 9, \"camera\": \"Macro on the silver chain against her collarbone.\", \"action\": \"The chain swings hard and catches the low sun as she moves. Detail beat on the rhythm.\", \"sfx\": \"Metal chink layered into the beat.\" },\n    { \"shot\": 10, \"camera\": \"Wide from behind, she faces the skyline, mist off the vents.\", \"action\": \"'Laid brick over brick while they slept on the wait' — she gestures out at the city, small against it, undefeated.\", \"sfx\": \"Full beat, mist hiss, distant traffic.\" },\n    { \"shot\": 11, \"camera\": \"Fast cut to a tight low-angle close-up, neon flickering at frame edge.\", \"action\": \"'Sun on my back and the whole city low' — head bobbing hard on the beat, eyes never leaving the lens.\", \"sfx\": \"Hat roll building, sub pressure.\" },\n    { \"shot\": 12, \"camera\": \"Push-in to a locked hero frame — then hard cut to black.\", \"action\": \"'I don't chase the moment — the moment I own.' She holds the final word, arms spread wide, sun blazing behind her. HARD CUT on the last syllable.\", \"sfx\": \"Final 808 hit, everything cutting to dead silence.\" }\n  ]\n}",
    "promptSourceUrl": "https://x.com/abulu8/status/2085253978166808615",
    "sourceUrl": "https://x.com/abulu8/status/2085253978166808615",
    "sourceLabel": "X 原帖 · @abulu8",
    "author": "Syed Abuthahir ∞",
    "publishedAt": "2026-08-06T06:37:45.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085253978166808615.mp4",
    "posterUrl": "/posters/x/2085253978166808615.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Landed on the first run... she's super cool. 😎\n\nThis is crazily good.\n\nI'd say #MiniMaxH3 is better than Seedance 2.0 when it comes to lip-syncing songs. Even with this fast rap, it nailed the sync.\n\nAnd the best part? This is pure Text-to-Video. 🤯\n\n@Hailuo_AI\n\nPrompt 👇💥\n\n{\n  \"format\": {\n    \"duration\": \"15s\",\n    \"bpm\": 96,\n    \"total_shots\": 12,\n    \"sync_type\": \"Beat-Cut Rap Performance — Double-Time Delivery\"\n  },\n  \"style_reference\": {\n    \"look\": \"Gritty cinematic 35mm — shot on RED Komodo 6K, 24fps, heavy Kodak 5219 500T grain overlay. Muted steel-blue and rust-orange palette with glowing sunset rim highlights and urban grays. Strong backlit lens flare, crushed but detailed shadows, slight halation on the sun. Handheld energy with natural sway. Anamorphic-flavored flare streaks, atmospheric haze catching low sun.\",\n    \"enforcement\": \"Maintain the heavy grain, muted steel/rust palette and golden backlight in EVERY shot. Never clean, never over-saturated, never flat-lit.\"\n  },\n  \"subject\": {\n    \"description\": \"A bold female rapper, mid-twenties, with shaved sides and long platinum-blonde braids. Confident, commanding presence, sharp direct eye contact. Wardrobe: oversized camo jacket worn open, black sports bra, cargo pants, chunky boots, heavy silver chain. UNIFY ACROSS ALL SHOTS: identical face, braids, wardrobe and chain; same rooftop; same golden-hour light.\"\n  },\n  \"environment\": {\n    \"setting\": \"The rooftop of an old industrial building at golden hour, overlooking a city skyline.\",\n    \"background\": \"Graffiti-covered concrete walls, rusted AC vents and ducting, faint mist rising off the units, a flickering neon sign at the edge of frame, the skyline sprawling below in haze.\",\n    \"lighting\": \"Low sunset sun directly behind her creating a hard rim-light and heavy lens flare; her front filled softly by bounce off the concrete. Rust-orange highlights against steel-blue shadow. Neon adding a small competing color source at dusk.\"\n  },\n  \"mood\": \"gritty, commanding, defiant, powerful, cinematic, grounded\",\n  \"music\": \"Hard-hitting trap at 96 BPM — heavy 808 sub, crisp rolling hi-hats, sparse ominous synth line. Beat drops on shot 3 and hits hardest at the final bar.\",\n  \"style\": \"Ultra-Realistic\",\n  \"logic_rule\": \"BEAT-CUT PERFORMANCE VIDEO — 12 hard cuts synced to the rhythm, NOT one continuous take. Vary the camera relentlessly: low hero wide, extreme close-up on mouth and eyes, over-shoulder skyline, whip-pan, handheld orbit, macro on the chain, silhouette against the sun. ACCURATE LIP-SYNC is mandatory — her mouth must match the written bars precisely, delivered double-time with sharp aggressive articulation. Performance energy: hand gestures punctuating bars, chest-forward power stance, head bobbing on the beat, braids swinging. Mandatory SPEED CHANGE: one brief slow-motion beat where her braids swing through the sun flare, then snap back on the beat. Physical effects: mist drifting from vents, dust in the flare, jacket moving in the wind. Keep the gritty grain and golden backlight throughout. ENDING: hard cut on the final bar, not a fade.\",\n  \"lyrics\": \"Came up off concrete, no ladder, no lane / Turned every locked door into fuel for the flame / They told me stay quiet — I turned the sound up / Skyline my witness, I ain't waiting on luck / No hand-out, no co-sign, I carried the weight / Laid brick over brick while they slept on the wait / Sun on my back and the whole city low / I don't chase the moment — the moment I own.\",\n  \"storyboard\": [\n    { \"shot\": 1, \"camera\": \"Low-angle wide, 24mm, heroic framing, handheld sway, sun flaring behind her.\", \"action\": \"She stands in a power stance on the rooftop edge, jacket moving in the wind, silhouetted by the sunset. Beat intro — she doesn't speak yet.\", \"sfx\": \"City rumble below, distant sirens, wind, beat intro.\" },\n    { \"shot\": 2, \"camera\": \"Extreme close-up on her eyes, hard rim-light.\", \"action\": \"Her eyes lift and lock dead into the lens. A single beat of held silence before the drop.\", \"sfx\": \"Hi-hats rolling in, a tense sub swell.\" },\n    { \"shot\": 3, \"camera\": \"Fast push-in to medium close, handheld.\", \"action\": \"BEAT DROPS. She launches into the first bar, hand cutting the air on the downbeat: 'Came up off concrete, no ladder, no lane.'\", \"sfx\": \"808 hitting hard, hats snapping, vocal cutting in.\" },\n    { \"shot\": 4, \"camera\": \"Whip-pan to a low side angle, graffiti wall behind.\", \"action\": \"'Turned every locked door into fuel for the flame' — she jabs a finger toward the lens, chest forward, braids swinging.\", \"sfx\": \"Trap beat driving, vocal aggressive.\" },\n    { \"shot\": 5, \"camera\": \"Extreme macro on her mouth, razor focus.\", \"action\": \"'They told me stay quiet — I turned the sound up' — tight lip-sync, sharp double-time articulation.\", \"sfx\": \"Vocal front and center, hat rolls.\" },\n    { \"shot\": 6, \"camera\": \"Over-shoulder, skyline hazy below her.\", \"action\": \"'Skyline my witness, I ain't waiting on luck' — she throws an arm out toward the city, claiming it.\", \"sfx\": \"Sub-bass hit, wind, city rumble.\" },\n    { \"shot\": 7, \"camera\": \"SLOW MOTION, tight profile, sun flaring through.\", \"action\": \"Her braids swing through the golden flare, jacket lifting, face set — the one held breath in the whole piece.\", \"sfx\": \"Beat filtering down, a stretched vocal tail.\" },\n    { \"shot\": 8, \"camera\": \"Snap back to speed. Handheld orbit, low.\", \"action\": \"'No hand-out, no co-sign, I carried the weight' — she pivots with the camera, hand pounding her chest on 'weight'.\", \"sfx\": \"Beat slamming back in, snare crack.\" },\n    { \"shot\": 9, \"camera\": \"Macro on the silver chain against her collarbone.\", \"action\": \"The chain swings hard and catches the low sun as she moves. Detail beat on the rhythm.\", \"sfx\": \"Metal chink layered into the beat.\" },\n    { \"shot\": 10, \"camera\": \"Wide from behind, she faces the skyline, mist off the vents.\", \"action\": \"'Laid brick over brick while they slept on the wait' — she gestures out at the city, small against it, undefeated.\", \"sfx\": \"Full beat, mist hiss, distant traffic.\" },\n    { \"shot\": 11, \"camera\": \"Fast cut to a tight low-angle close-up, neon flickering at frame edge.\", \"action\": \"'Sun on my back and the whole city low' — head bobbing hard on the beat, eyes never leaving the lens.\", \"sfx\": \"Hat roll building, sub pressure.\" },\n    { \"shot\": 12, \"camera\": \"Push-in to a locked hero frame — then hard cut to black.\", \"action\": \"'I don't chase the moment — the moment I own.' She holds the final word, arms spread wide, sun blazing behind her. HARD CUT on the last syllable.\", \"sfx\": \"Final 808 hit, everything cutting to dead silence.\" }\n  ]\n}",
    "engagement": {
      "replies": 7,
      "reposts": 2,
      "likes": 15,
      "views": 1244,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085241788340797558",
    "title": "图像与文字的无缝变换",
    "titleEn": "Images and Typography in Seamless Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "图像与文字的无缝变换的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2944×1248。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for images and typography in seamless motion on X. Native-video output: 15s · 2944×1248. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[1] FORMAT\n10-second 16:9 cinematic motion-graphics promo built from a single desert muscle-car image. Premium, fast, dimensional, elegant, and tactile.\n\n[2] ASSETS\nUse Image 1 as the exact first-frame source and core visual reference for the vintage white muscle car, dusty desert road, rock formations, muted sand palette, and gritty film texture.\n\n[3] IDENTITY\nPreserve:\n- the old white Dodge Charger–style muscle car as the hero subject\n- weathered paint, rust marks, dust buildup, and worn wheels\n- low front three-quarter angle and dominant car scale\n- barren desert highway with cracked asphalt and rocky outcrops\n- beige, sepia, and charcoal color family\n- airborne dust and dry wind atmosphere\n- gritty, premium editorial mood rather than cheerful commercial energy\n- realistic hard sunlight and soft dust diffusion\n- grounded reflections and believable shadow direction\n\n[4] BEATS\n[0–2 seconds] Start from the original composition. Create subtle stillness for half a second, then introduce layered parallax: foreground dust, midground car, and background rocks separate in depth. Camera begins a slow push-in with a slight rightward arc. Fine sand drifts across frame.\n[2–4 seconds] Speed ramps in. Add dimensional motion-graphic accents: thin metallic lines, soft grid fragments, and subtle contour-tracing highlights briefly wrap around the car body, emphasizing shape and speed. Main title appears: \"BORN FROM DUST\".\n[4–6 seconds] Camera performs a faster sweep past the front quarter panel toward the side, with strong depth, dust streaks, and passing light sweeps. Secondary text appears: \"SPEED HAS A SHAPE\". Keep the car stable and heroic while the environment feels fast.\n[6–8 seconds] Push the most dynamic moment: dust bursts, perspective stretch, and elegant 3D typography planes moving in space around the car without blocking it. Add one short accent line: \"RAW / FAST / ICONIC\".\n[8–10 seconds] Motion resolves into a premium final frame. Camera eases back into a clean hero composition. Final end card text appears and holds: \"CHASE THE HORIZON\". Keep the last 1.5 seconds readable and polished.\n\n[5] CAMERA\nCinematic virtual camera from a single still image. Use layered 2.5D depth separation, push-ins, slight orbit motion, and one speed-ramped sweep for dimensionality. Start as a low wide hero shot, move into a medium close pass on the car, then settle into a clean final hero frame. Motion should feel expensive and controlled, not chaotic.\n\n[6] LIGHT\nHarsh desert sunlight with dusty diffusion. Palette: sand beige, sun-faded ivory, smoky gray, and soft metallic charcoal. Add tasteful specular sweeps on the car edges and grille. Film texture should feel editorial and refined, with fine grain and suspended dust particles for depth.\n\n[7] EDIT\nUse seamless motion-graphics transitions only, no hard scene changes. Typography and graphic elements should feel integrated with the car and dust flow. Text behavior:\n- \"BORN FROM DUST\" enters with a fast horizontal build and slight depth extrusion\n- it remains readable for around 1 second\n- it exits by dissolving into dust particles\n- do not warp, misspell, mirror, or replace the text\n- \"SPEED HAS A SHAPE\" enters on a perspective plane aligned to the car motion\n- \"RAW / FAST / ICONIC\" appears as small premium kinetic type\n- final text \"CHASE THE HORIZON\" enters cleanly, stays stable, and fades softly at the end\n\n[8] AUDIO\nAudio: dry desert wind, fine sand skimming across metal, distant low engine rumble, and one sharp whoosh accent at 4 seconds during the speed ramp.\nBGM: 10-second cue, 65% cinematic tension and 35% luxury fashion energy. Deep pulse, restrained synth bass, tight percussion, airy reverse textures, and metallic hits. Build gradually from 0–4 seconds, intensify from 4–8 seconds, then resolve into a confident sustained ending. Freeze the final 2 seconds on a clean, premium sonic tail. Do not imitate an existing melody.\n\n[9] NEGATIVE\nNo subtitles, on-screen text beyond the specified titles, watermarks, platform logos, or stickers.\nDo not introduce Chinese text, garbled characters, or misspellings.\nNo slideshow feeling — use natural dimensional camera motion and seamless transitions.\nNo cartoon styling, no cheap VFX clutter, no excessive lens distortion, no unrealistic car deformation, and no extra vehicles or people.",
    "promptSourceUrl": "https://x.com/opener_ai/status/2085241791163621823",
    "sourceUrl": "https://x.com/opener_ai/status/2085241788340797558",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-06T05:49:18.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085241788340797558.mp4",
    "posterUrl": "/posters/x/2085241788340797558.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Combining images and text is such a fun process. Because Hailuo Minimax H3 handles this kind of work so well, I gave it a try on @krea_ai .\n\nI feel like a single image and a prompt can open up endless possibilities. If I had to do this in After Effects as post-processing, \n\nI wouldn't even know where to start, but AI is filling in that gap for me right now.\n\nOf course, true precision still comes from human hands, but a lot of things are changing this year.\n\nThis video was made with H3 in Krea.\n\nPrompt ⤵️\n\n#MinimaxH3 #kreacpp #kreaai",
    "engagement": {
      "replies": 6,
      "reposts": 4,
      "likes": 26,
      "views": 2478,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085240792067784843",
    "title": "勇士迎战巨型怪物",
    "titleEn": "A Fearless Warrior Battles a Giant Monster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "勇士迎战巨型怪物的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a fearless warrior battles a giant monster on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A powerful warrior woman stands alone in a destroyed futuristic city at night, facing a massive towering monster covered in dark armor and glowing cracks. The monster roars and charges toward her, shaking the ground with every step. She dodges its attacks with incredible speed, leaps through the air, and strikes with a glowing energy weapon. Intense hand-to-hand combat, flying debris, explosions, dramatic slow motion shots, cinematic camera movements, close-up of her determined face. The battle reaches its peak as she unleashes her final powerful attack, defeating the giant monster and standing victorious as the city burns behind her. Epic Hollywood action movie style, ultra-realistic, dramatic lighting, dynamic camera angles, high detail, cinematic quality.",
    "promptSourceUrl": "https://x.com/RuzainaMeer/status/2085240792067784843",
    "sourceUrl": "https://x.com/RuzainaMeer/status/2085240792067784843",
    "sourceLabel": "X 原帖 · @RuzainaMeer",
    "author": "Ruzaina",
    "publishedAt": "2026-08-06T05:45:21.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085240792067784843.mp4",
    "posterUrl": "/posters/x/2085240792067784843.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tested MiniMax H3 for the first time with a scene like this \n\nA cinematic battle between a fearless warrior and a giant monster pushing AI action visuals to a whole new level. \n\nPrompt:\n\nA powerful warrior woman stands alone in a destroyed futuristic city at night, facing a massive towering monster covered in dark armor and glowing cracks. The monster roars and charges toward her, shaking the ground with every step. She dodges its attacks with incredible speed, leaps through the air, and strikes with a glowing energy weapon. Intense hand-to-hand combat, flying debris, explosions, dramatic slow motion shots, cinematic camera movements, close-up of her determined face. The battle reaches its peak as she unleashes her final powerful attack, defeating the giant monster and standing victorious as the city burns behind her. Epic Hollywood action movie style, ultra-realistic, dramatic lighting, dynamic camera angles, high detail, cinematic quality.",
    "engagement": {
      "replies": 71,
      "reposts": 9,
      "likes": 179,
      "views": 14091,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085234595784773704",
    "title": "深海沉没",
    "titleEn": "Drowning Deep Down the Ocean",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深海沉没的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for drowning deep down the ocean on X. Native-video output: 15s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "\"Use the uploaded character sheet the ONLY identity reference. Preserve the exact face, hairstyle, body proportions, skin tone, and appearance throughout the video.\n\nThe camera remains directly above him looking straight downward during the entire sequence. He falls vertically beneath the camera through an impossible underwater abyss without changing direction.\nHe stays on his back facing the lens, both arms reaching upward while bubbles, hair, clothing, and loose fabric stream upward through the water.\n\nHe continuously falls through:\nvibrant coral reef, abandoned pirate ship, underwater research station, giant kelp forest\nsubmerged ancient city, whale migration corridor, volcanic hydrothermal vent field, endless dark ocean trench.\n\nEvery level is entered by crashing through ceilings, glass domes, ship decks, coral arches, ruined temples, and collapsing structures. Schools of fish, jellyfish, manta rays, bubbles, seaweed, bioluminescent creatures, and debris flow upward past the camera naturally.\n\nHis expressions progress from surprise and panic to determination, breath-holding, fear, exhaustion, and final acceptance while continuously reaching upward for anything that might stop his fall.\n\n24mm anamorphic lens, underwater cinematic lighting, volumetric light rays, realistic water distortion, gentle bloom, film grain, radial motion blur, handheld floating movement.\n\nAudio: Epic orchestral score mixed with deep underwater ambience, whale calls, distant metallic creaks, volcanic rumbles, rushing currents, and a powerful cinematic finale.\"",
    "promptSourceUrl": "https://x.com/I_Muhammadali44/status/2085234595784773704",
    "sourceUrl": "https://x.com/I_Muhammadali44/status/2085234595784773704",
    "sourceLabel": "X 原帖 · @I_Muhammadali44",
    "author": "Muhammad Ali",
    "publishedAt": "2026-08-06T05:20:44.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085234595784773704.mp4",
    "posterUrl": "/posters/x/2085234595784773704.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Drowning deep down the ocean\" created with MiniMax H3\n\n Prompt:\n\"Use the uploaded character sheet the ONLY identity reference. Preserve the exact face, hairstyle, body proportions, skin tone, and appearance throughout the video.\n\nThe camera remains directly above him looking straight downward during the entire sequence. He falls vertically beneath the camera through an impossible underwater abyss without changing direction.\nHe stays on his back facing the lens, both arms reaching upward while bubbles, hair, clothing, and loose fabric stream upward through the water.\n\nHe continuously falls through:\nvibrant coral reef, abandoned pirate ship, underwater research station, giant kelp forest\nsubmerged ancient city, whale migration corridor, volcanic hydrothermal vent field, endless dark ocean trench.\n\nEvery level is entered by crashing through ceilings, glass domes, ship decks, coral arches, ruined temples, and collapsing structures. Schools of fish, jellyfish, manta rays, bubbles, seaweed, bioluminescent creatures, and debris flow upward past the camera naturally.\n\nHis expressions progress from surprise and panic to determination, breath-holding, fear, exhaustion, and final acceptance while continuously reaching upward for anything that might stop his fall.\n\n24mm anamorphic lens, underwater cinematic lighting, volumetric light rays, realistic water distortion, gentle bloom, film grain, radial motion blur, handheld floating movement.\n\nAudio: Epic orchestral score mixed with deep underwater ambience, whale calls, distant metallic creaks, volcanic rumbles, rushing currents, and a powerful cinematic finale.\"",
    "engagement": {
      "replies": 18,
      "reposts": 7,
      "likes": 44,
      "views": 1290,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085233880353915217",
    "title": "曼谷街头美食之旅",
    "titleEn": "A Taste of Bangkok Street Food",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "曼谷街头美食之旅的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a taste of bangkok street food on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic 15-second ultra-realistic travel vlog following a young female creator exploring Bangkok's vibrant street food scene. She walks through a lively daytime market, introduces the legendary Pad Thai, watches a skilled street chef prepare the dish with fresh ingredients, and captures detailed close-ups of sizzling noodles, colorful toppings, and authentic cooking moments. She tastes the freshly made Pad Thai and shares her genuine reaction while surrounded by the energy of local vendors and travelers. Handheld vlog realism, natural lip-sync, expressive emotions, cinematic food close-ups, realistic market atmosphere, bright tropical daylight, ARRI Alexa 35 aesthetic, HDR, immersive street sounds, premium travel documentary style, 4K photorealistic quality, seamless character consistency, no text, no logos.",
    "promptSourceUrl": "https://x.com/nawalsehar/status/2085233880353915217",
    "sourceUrl": "https://x.com/nawalsehar/status/2085233880353915217",
    "sourceLabel": "X 原帖 · @nawalsehar",
    "author": "Nawal",
    "publishedAt": "2026-08-06T05:17:53.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085233880353915217.mp4",
    "posterUrl": "/posters/x/2085233880353915217.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A taste of Bangkok, one unforgettable bite at a time.Exploring the flavors, stories, and energy of Thailand's street food culture.\n\nCreated with MiniMax H3\n\nPrompt:\n\nA cinematic 15-second ultra-realistic travel vlog following a young female creator exploring Bangkok's vibrant street food scene. She walks through a lively daytime market, introduces the legendary Pad Thai, watches a skilled street chef prepare the dish with fresh ingredients, and captures detailed close-ups of sizzling noodles, colorful toppings, and authentic cooking moments. She tastes the freshly made Pad Thai and shares her genuine reaction while surrounded by the energy of local vendors and travelers. Handheld vlog realism, natural lip-sync, expressive emotions, cinematic food close-ups, realistic market atmosphere, bright tropical daylight, ARRI Alexa 35 aesthetic, HDR, immersive street sounds, premium travel documentary style, 4K photorealistic quality, seamless character consistency, no text, no logos.",
    "engagement": {
      "replies": 72,
      "reposts": 3,
      "likes": 198,
      "views": 14187,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085221226335879541",
    "title": "忘却王国的无畏旅人",
    "titleEn": "A Fearless Traveler Enters a Forgotten Kingdom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "忘却王国的无畏旅人的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a fearless traveler enters a forgotten kingdom on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic 15-second fantasy adventure following a lone traveler as they cross an ancient stone bridge toward a forgotten medieval kingdom hidden beneath towering cliffs. As colossal gates slowly open, giant stone guardians awaken and glowing light spills through the mist, revealing a city untouched by time. The traveler steps forward as thousands of lanterns illuminate the streets, waterfalls cascade from towering cliffs, and majestic dragons soar across the sunrise sky. Epic cinematic camera movements, IMAX-scale visuals, volumetric lighting, realistic weather effects, premium VFX, emotional orchestral atmosphere, ARRI Alexa 35 aesthetic, anamorphic lenses, 4K HDR, seamless continuity, blockbuster movie quality, no text, no logos.",
    "promptSourceUrl": "https://x.com/aiwithaly/status/2085221226335879541",
    "sourceUrl": "https://x.com/aiwithaly/status/2085221226335879541",
    "sourceLabel": "X 原帖 · @aiwithaly",
    "author": "Ali",
    "publishedAt": "2026-08-06T04:27:36.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085221226335879541.mp4",
    "posterUrl": "/posters/x/2085221226335879541.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A forgotten kingdom. A fearless traveler.The beginning of an unforgettable adventure.\n\nCreated with MiniMax H3\n\nPrompt:\n\nA cinematic 15-second fantasy adventure following a lone traveler as they cross an ancient stone bridge toward a forgotten medieval kingdom hidden beneath towering cliffs. As colossal gates slowly open, giant stone guardians awaken and glowing light spills through the mist, revealing a city untouched by time. The traveler steps forward as thousands of lanterns illuminate the streets, waterfalls cascade from towering cliffs, and majestic dragons soar across the sunrise sky. Epic cinematic camera movements, IMAX-scale visuals, volumetric lighting, realistic weather effects, premium VFX, emotional orchestral atmosphere, ARRI Alexa 35 aesthetic, anamorphic lenses, 4K HDR, seamless continuity, blockbuster movie quality, no text, no logos.",
    "engagement": {
      "replies": 74,
      "reposts": 7,
      "likes": 211,
      "views": 13328,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085211946815758841",
    "title": "产品图变身高级电商界面",
    "titleEn": "A Product Image Becomes a Premium Storefront",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "产品图变身高级电商界面的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a product image becomes a premium storefront on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 16:9 premium ecommerce landing page UI/UX demo built around the product in Image 1. Use a luxurious, futuristic sports-performance aesthetic with oversized bold italic sans-serif typography, dark glossy backgrounds, carbon fiber and performance-mesh textures, metallic highlights, and speed-driven light streaks. Show a smooth high-speed page scroll, premium product sections, sleek feature cards, gallery modules, tech-spec areas, and strong CTA zones. Include high-impact hover interactions like scale-up, color inversion, glow effects, card expansion, and animated typography. Make the motion cinematic, expensive, and ultra-polished, like a world-class product launch website demo.\n\n💰 Generation cost: under $3\n🖼 Input: one product image\n🧩 Output: a polished website concept video\n\nTry MiniMax H3 → https://bit.ly/4fES5pl\n\n#MiniMaxH3 #EcommerceDesign #UIDesign #ProductMarketing #ImaStudio",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2085211946815758841",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2085211946815758841",
    "sourceLabel": "X 原帖 · @ImaStudio_ai",
    "author": "Ima Studio",
    "publishedAt": "2026-08-06T03:50:44.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085211946815758841.mp4",
    "posterUrl": "/posters/x/2085211946815758841.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GPT Image 2 +MiniMax H3 \nOne product image. One prompt. One premium e-commerce UI concept.\nWe used MiniMax H3 to turn a basic product landing page idea into a 15-second UI/UX demo featuring\nprompt 👇\n\nCreate a 16:9 premium ecommerce landing page UI/UX demo built around the product in Image 1. Use a luxurious, futuristic sports-performance aesthetic with oversized bold italic sans-serif typography, dark glossy backgrounds, carbon fiber and performance-mesh textures, metallic highlights, and speed-driven light streaks. Show a smooth high-speed page scroll, premium product sections, sleek feature cards, gallery modules, tech-spec areas, and strong CTA zones. Include high-impact hover interactions like scale-up, color inversion, glow effects, card expansion, and animated typography. Make the motion cinematic, expensive, and ultra-polished, like a world-class product launch website demo.\n\n💰 Generation cost: under $3\n🖼 Input: one product image\n🧩 Output: a polished website concept video\n\nTry MiniMax H3 → https://bit.ly/4fES5pl\n\n#MiniMaxH3 #EcommerceDesign #UIDesign #ProductMarketing #ImaStudio",
    "engagement": {
      "replies": 11,
      "reposts": 5,
      "likes": 30,
      "views": 1279,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085210848272986556",
    "title": "超人亲自洗衣服",
    "titleEn": "Superman Does His Own Laundry",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "超人亲自洗衣服的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for superman does his own laundry on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "0:00–0:04 — Wide shot of a messy apartment living room. SUPERMAN (full costume) is standing in front of an open washing machine holding a huge pile of red-and-blue clothes.\n0:04–0:07 — He starts stuffing the cape into the machine while singing in a deep heroic voice: “Laundry day, laundry day… even heroes have to pay!” and does a dramatic fist pump that makes the clothes fly everywhere.\n0:07–0:10 — He tries to close the lid, but a red sock keeps floating up and sticking to his face. He peels it off and continues singing louder: “Red socks, blue socks… they never stay in the box!” while spinning in a circle.\n0:10–0:13 — He accidentally presses the wrong button. The machine starts shaking violently and soap bubbles start flying out. Superman tries to stay serious and stop it with one hand.\n0:13–0:16 — Bubbles land on his face and the sock sticks again. His serious superhero expression completely cracks. He bursts out laughing while still holding the machine.\n0:16–0:18 — He freezes mid-pose with bubbles all over him, laughing so hard he has to lean on the washing machine as more clothes keep popping out.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2085210848272986556",
    "sourceUrl": "https://x.com/Strength04_X/status/2085210848272986556",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-06T03:46:22.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085210848272986556.mp4",
    "posterUrl": "/posters/x/2085210848272986556.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What if Superman had to do his own laundry? 😂\n\nMade with MiniMax H3 on @Hailuo_AI \n\nPrompt - 0:00–0:04 — Wide shot of a messy apartment living room. SUPERMAN (full costume) is standing in front of an open washing machine holding a huge pile of red-and-blue clothes.\n0:04–0:07 — He starts stuffing the cape into the machine while singing in a deep heroic voice: “Laundry day, laundry day… even heroes have to pay!” and does a dramatic fist pump that makes the clothes fly everywhere.\n0:07–0:10 — He tries to close the lid, but a red sock keeps floating up and sticking to his face. He peels it off and continues singing louder: “Red socks, blue socks… they never stay in the box!” while spinning in a circle.\n0:10–0:13 — He accidentally presses the wrong button. The machine starts shaking violently and soap bubbles start flying out. Superman tries to stay serious and stop it with one hand.\n0:13–0:16 — Bubbles land on his face and the sock sticks again. His serious superhero expression completely cracks. He bursts out laughing while still holding the machine.\n0:16–0:18 — He freezes mid-pose with bubbles all over him, laughing so hard he has to lean on the washing machine as more clothes keep popping out.",
    "engagement": {
      "replies": 28,
      "reposts": 0,
      "likes": 56,
      "views": 5698,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085198361989460476",
    "title": "一场由逐句台词精确控制的戏",
    "titleEn": "A Gritty Scene Controlled Line by Line",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一场由逐句台词精确控制的戏的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a gritty scene controlled line by line on X. Native-video output: 14s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "\"A gritty, photorealistic nighttime crime scene outside a convenience store, shot like a serious detective TV drama. Yellow police tape, flashing patrol lights, forensic markers, and two stern live-action detectives carefully examining evidence. Assisting them are Bugs Bunny and Daffy Duck, fully recognizable in classic Looney Tunes style - bright, clean 2D cartoon characters composited into the real world. Bugs is unusually calm and professional, kneeling beside a clue with a cartoon magnifying glass, while Daffy nervously points at random evidence and overreacts. The detectives remain completely serious and treat them like normal forensic partners. At the climax, Bugs holds the magnifying glass over a surface and it reveals a photorealistic fingerprint, while Daffy dramatically gasps and points. Strong contrast between cartoon and live-action, believable shadows and interaction, cinematic camera movement, tense crime-drama atmosphere played completely straight.\"",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2085198361989460476",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2085198361989460476",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-06T02:56:45.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085198361989460476.mp4",
    "posterUrl": "/posters/x/2085198361989460476.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One gripe with MiniMax H3: unless I specify exactly what each character says, the dialogue sometimes devolves into gibberish. Always write out the speech.\n\nPrompt: \"A gritty, photorealistic nighttime crime scene outside a convenience store, shot like a serious detective TV drama. Yellow police tape, flashing patrol lights, forensic markers, and two stern live-action detectives carefully examining evidence. Assisting them are Bugs Bunny and Daffy Duck, fully recognizable in classic Looney Tunes style - bright, clean 2D cartoon characters composited into the real world. Bugs is unusually calm and professional, kneeling beside a clue with a cartoon magnifying glass, while Daffy nervously points at random evidence and overreacts. The detectives remain completely serious and treat them like normal forensic partners. At the climax, Bugs holds the magnifying glass over a surface and it reveals a photorealistic fingerprint, while Daffy dramatically gasps and points. Strong contrast between cartoon and live-action, believable shadows and interaction, cinematic camera movement, tense crime-drama atmosphere played completely straight.\"",
    "engagement": {
      "replies": 18,
      "reposts": 7,
      "likes": 169,
      "views": 20182,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085193166437892346",
    "title": "埃及艳后点了披萨",
    "titleEn": "Cleopatra Orders a Pizza",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "埃及艳后点了披萨的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for cleopatra orders a pizza on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "0:00–0:04 — Medium shot inside a golden pyramid chamber. CLEOPATRA sits on her throne looking majestic. A modern pizza delivery guy in helmet and red jacket suddenly walks in holding a large pizza box.\n0:04–0:07 — Cleopatra stands up, points at the box and starts singing in a dramatic royal voice: “Pizza man, pizza man… bring the cheese across the sand!” while doing slow Egyptian hand waves.\n0:07–0:10 — The delivery guy tries hard to stay professional and say “That’ll be 12 dollars,” but his lips start trembling.\n0:10–0:13 — He can’t hold it and bursts out laughing, almost dropping the pizza.\n0:13–0:15 — Cleopatra freezes mid-hand-wave, looks at him, and completely cracks up too. Both are laughing while the pizza box tilts dangerously.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2085193166437892346",
    "sourceUrl": "https://x.com/Strength04_X/status/2085193166437892346",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-06T02:36:06.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085193166437892346.mp4",
    "posterUrl": "/posters/x/2085193166437892346.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What if Cleopatra ordered pizza? 😂\n\nMade with MiniMax H3 on @Hailuo_AI \n\nPrompt - 0:00–0:04 — Medium shot inside a golden pyramid chamber. CLEOPATRA sits on her throne looking majestic. A modern pizza delivery guy in helmet and red jacket suddenly walks in holding a large pizza box.\n0:04–0:07 — Cleopatra stands up, points at the box and starts singing in a dramatic royal voice: “Pizza man, pizza man… bring the cheese across the sand!” while doing slow Egyptian hand waves.\n0:07–0:10 — The delivery guy tries hard to stay professional and say “That’ll be 12 dollars,” but his lips start trembling.\n0:10–0:13 — He can’t hold it and bursts out laughing, almost dropping the pizza.\n0:13–0:15 — Cleopatra freezes mid-hand-wave, looks at him, and completely cracks up too. Both are laughing while the pizza box tilts dangerously.",
    "engagement": {
      "replies": 23,
      "reposts": 1,
      "likes": 52,
      "views": 6545,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085168181648445763",
    "title": "校园奔跑中的日语对话",
    "titleEn": "Japanese Dialogue During a Schoolyard Run",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "校园奔跑中的日语对话的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 8 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for japanese dialogue during a schoolyard run on X. Native-video output: 8s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A scene showing a male student wearing a green tracksuit and a female student in a sailor uniform running together on the school path along s small river.\nA shot of the two people viewed from directly the side. They are running from the right side of the frame toward the left. The frame captures their upper bodies.\nSeveral classmates on the way, walking to school.\nRefer to <Picture 1> for the characters.\nVoice 1(male):またワイら遅刻ちゃうか\nVoice 2(female):あと30秒！あたしらの勝ちやな！",
    "promptSourceUrl": "https://x.com/MireilleDartois/status/2085168181648445763",
    "sourceUrl": "https://x.com/MireilleDartois/status/2085168181648445763",
    "sourceLabel": "X 原帖 · @MireilleDartois",
    "author": "ダルトワ★TV",
    "publishedAt": "2026-08-06T00:56:49.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085168181648445763.mp4",
    "posterUrl": "/posters/x/2085168181648445763.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3で日本語セリフ指定🦑\n\nすこし演技ができるもののSola2みたいに感情を出すのは無理のようです🤔\n\nプロンプト:\nA scene showing a male student wearing a green tracksuit and a female student in a sailor uniform running together on the school path along s small river.\nA shot of the two people viewed from directly the side. They are running from the right side of the frame toward the left. The frame captures their upper bodies.\nSeveral classmates on the way, walking to school.\nRefer to <Picture 1> for the characters.\nVoice 1(male):またワイら遅刻ちゃうか\nVoice 2(female):あと30秒！あたしらの勝ちやな！",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 27,
      "views": 2626,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085159317578330229",
    "title": "文字在镜头中流动",
    "titleEn": "Typography Flows Through the Scene",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "文字在镜头中流动的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for typography flows through the scene on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[1] FORMAT\n15-second 16:9 stylized anime action title sequence — martial-arts choreography fused with 3D kinetic typography. Painterly 2D anime illustration look with cel-shaded volumes and subtle paper grain, matching the source artwork's rendering.\n\n[2] ASSETS\nUse Image 1 as a strict identity reference for the female duelist KAELAN — read only the character artwork itself for her face, hair, wardrobe, weapon, and color palette. Ignore the sheet's layout, panels, printed labels, handwriting, silhouette thumbnails, and background parchment; none of them may appear in the video.\n\n[3] IDENTITY\nPreserve exactly: deep brown skin with warm highlights; luminous crimson-red eyes with a sharp black liner shape; jet-black hair pulled into a tight scalp-wrapped part that splits into two very long thick braids falling past the hips, moving like weighted whips; large white crescent hoop earrings on both ears; black high-neck sleeveless crop top exposing a toned midriff; a sheer white long-sleeved kimono robe worn open, translucent and lightly billowing; cream wide-leg hakama trousers with a single black diamond patch on the left thigh; black wrapped fingerless gloves with lattice cord across the back of the hand; white cloth-wrapped platform sandals; a well-worn katana with a dark blue-grey blade and a black diamond-wrapped hilt. Her expression stays calm and unblinking, never comic.\n\n[4] BEATS\nShot 1 — [0–3 seconds] Kaelan stands centered in a void of warm off-white light, katana point-down, braids settling. She exhales; a slow iai draw begins. → Hard cut.\nShot 2 — [3–6 seconds] Mid-draw. The blade clears the saya and a clean arc of light follows the edge. The 3D extruded word \"HAILUO\" flies in from screen left along the blade's arc, its letter faces catching the same light, passing BEHIND her shoulder and IN FRONT of her trailing braid before exiting right. → Whip-pan cut.\nShot 3 — [6–9 seconds] Full-body kata combination: descending cut, pivot, rising horizontal sweep. Her sheer robe and both braids trail one frame behind her body. The word \"MINIMAX\" rotates up from the floor plane in thick extruded 3D, standing taller than her, and she cuts through the gap between its letters — the letterforms hold solid, do not shatter. → Hard cut.\nShot 4 — [9–12.5 seconds] Low-angle crouch landing, sword arm extended back, dust ring pushing outward. \"H3\" slams in from depth as a massive chromed 3D glyph directly behind her, scaling down to lock, with her silhouette reading cleanly against it. → Flash cut on the impact frame.\nShot 5 — [12.5–15 seconds] She rises into the neutral standing pose, katana returning to point-down. The full lockup \"HAILUO MiniMax H3\" assembles from the three pieces on a single horizontal baseline at lower third, settles, and holds perfectly still through the final frame.\n\n[5] CAMERA\nAnamorphic-feel 35mm for wides, 50mm for the mid shots.",
    "promptSourceUrl": "https://x.com/opener_ai/status/2085159327539798202",
    "sourceUrl": "https://x.com/opener_ai/status/2085159317578330229",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-06T00:21:36.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085159317578330229.mp4",
    "posterUrl": "/posters/x/2085159317578330229.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo has been great at text motion graphics for a long time, but it really shines with H3. @Hailuo_AI  @MiniMax_AI \n\nPrompt ⤵️\n\n#Hailuo #MinimaxH3",
    "engagement": {
      "replies": 7,
      "reposts": 2,
      "likes": 36,
      "views": 1489,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085152634949030212",
    "title": "三模型重现同一幅战斗图",
    "titleEn": "Three Models Recreate the Same Battle Image",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "三模型重现同一幅战斗图的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1080×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for three models recreate the same battle image on X. Native-video output: 15s · 1080×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "电影级东方神话史诗，保持参考图中后羿的身份与造型统一：年轻坚毅的东方男性，凌乱黑色长发，带灼痕的黑铜战甲、朱红腰带、玉石饰物，手持雕纹神圣青铜弓。画面转换成完整连续的真实电影场景，禁止出现设定板、分格、文字和界面。\n\n0至3秒，乌云翻涌的焦土战场，天空悬挂九轮燃烧烈日。低机位掠过碎石与灰烬，快速推近后羿背影，披风和长发在热浪中狂舞。\n\n3至7秒，群鸦骤然冲出云层，镜头高速环绕至后羿侧面。他抬弓搭箭，青铜弓纹逐层亮起金红色神光，弓弦绷紧，眼神从克制转为决绝，火星贴近镜头飞过，强烈压迫感。\n\n7至11秒，后羿猛然释放神箭。瞬间进入极致慢动作，发光箭矢撕裂空气，随后镜头迅猛追随箭矢穿越云海。神箭分化为九道金色流光，烈日接连崩裂，形成震撼的连锁爆炸，冲击波推开乌云，天地剧烈震动。\n\n11至15秒，爆炸声势骤然沉寂，灰烬像雪一样缓慢飘落。天空只剩最后一轮温暖太阳，后羿立于山巅成为逆光剪影，缓缓放下神弓。镜头升高并大幅拉远，露出重获生机的辽阔山河，庄严、悲壮、余韵悠长。\n\n真实电影摄影，史诗级规模，强烈明暗对比，体积光，热浪折射，飞灰与火星，真实布料和金属质感，动态运镜，速度变化自然，24帧电影质感，人物面部稳定，手指与弓箭结构正确，无字幕，无水印，无现代物品。 \n\n根据这个提示词，你觉得哪个模型做的最好？客观多方位真实评价",
    "promptSourceUrl": "https://x.com/LufzzLiz/status/2085152639399150024",
    "sourceUrl": "https://x.com/LufzzLiz/status/2085152634949030212",
    "sourceLabel": "X 原帖 · @LufzzLiz",
    "author": "岚叔",
    "publishedAt": "2026-08-05T23:55:03.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085152634949030212.mp4",
    "posterUrl": "/posters/x/2085152634949030212.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "fashion",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.0 、happyhorse 1.1、MiniMax H3 第二轮图生视频测评\n这一轮Seedance 还是强，开篇效果不错，最接近参考图的凌乱长发、灼损战甲、朱红腰带和玉饰，真人感与运镜最好。缺点是九日和连锁爆炸表达不够准确。\n\nMiniMax H3：服装、玉饰和青铜弓细节较接近，画面最清晰；弓造型会变化，释放动作存在断层。\n\nhappyhorse 1.1：分镜执行最完整，但人物过于干净、红披风抢眼，CG感强，与参考图的粗粝气质差距较大。\n\n具体评分、参考图及提示词见评论👇",
    "engagement": {
      "replies": 7,
      "reposts": 2,
      "likes": 16,
      "views": 7039,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085138181133894024",
    "title": "角色、音频与歌词的三重参考",
    "titleEn": "Character, Audio, and Lyrics in One Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "角色、音频与歌词的三重参考的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for character, audio, and lyrics in one performance on X. Native-video output: 10s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "10-second dark emotional K-pop music video. Realistic young East Asian woman from the character reference, natural face, real skin texture, messy hair. \nImportant rules: \n- The girl is always framed in close-up or medium shots only. No wide or distant shots at all. \n- The girl performs continuously for the full 10 seconds without any cuts on her. Her movement is natural and fluid — she can shift weight, make small dance-like gestures, turn her head, move her shoulders and hands, but the camera stays on her in consistent medium/close framing. \n- Typography and background change with hard cuts according to the storyboard order, while the girl keeps performing through them.  \n\nSequence of text (hard cuts): \n1. “THE TABLE’S SET AGAIN” \n2. “NOBODY SITS DOWN” \n3. “YOUR WORDS ARE HARD GLASS” \n4. “FALLING TO THE FLOOR NOW” \n5. “I KEPT THE DOOR OPEN” \n6. “YOU KEPT YOUR HANDS CROSSED” \n7. “SAME ROOM, SAME SCAR” \n8. “SAME OLD COST” \n9. Final rose  The girl sings the entire song with clear, accurate lip-sync. \n\nMouth and teeth must stay natural and undistorted at all times — no deformation, no melting, no artifacts. Emotional facial performance. Typography stays sharp and reacts (cracks, shifts, particles). Changing typography every 1–1.5 seconds. Heavy red atmosphere, cinematic lighting, photoreal",
    "promptSourceUrl": "https://x.com/Kiber_Alla/status/2085138184124469371",
    "sourceUrl": "https://x.com/Kiber_Alla/status/2085138181133894024",
    "sourceLabel": "X 原帖 · @Kiber_Alla",
    "author": "Kiber Alla",
    "publishedAt": "2026-08-05T22:57:37.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085138181133894024.mp4",
    "posterUrl": "/posters/x/2085138181133894024.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "3 references. One model.\n\nAudio + character + lyrics typography   → new #MiniMaxH3 by @Hailuo_AI\n\nReally strong format for music videos. Would also work perfectly as live concert visuals🎵\n\nPrompt + references below👇",
    "engagement": {
      "replies": 22,
      "reposts": 5,
      "likes": 56,
      "views": 1907,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085137397134561487",
    "title": "八位像素少女漫步阳光小径",
    "titleEn": "An 8-Bit Girl Walks a Sunny Path",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "八位像素少女漫步阳光小径的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 5 秒 · 768×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an 8-bit girl walks a sunny path on X. Native-video output: 5s · 768×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "\"A cute little girl walks through a sunny forest when a small, friendly dinosaur steps out, waves, and says hi. Suddenly, the dinosaur sneezes—and a shower of colorful butterflies bursts from its nose. They both laugh. Cute, colorful animated style.\"\n\n- time: 15min ish\n- 768x768",
    "promptSourceUrl": "https://x.com/albertgao/status/2085137397134561487",
    "sourceUrl": "https://x.com/albertgao/status/2085137397134561487",
    "sourceLabel": "X 原帖 · @albertgao",
    "author": "Albert Gao",
    "publishedAt": "2026-08-05T22:54:30.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085137397134561487.mp4",
    "posterUrl": "/posters/x/2085137397134561487.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Wow, this is super cool, MiniMax H3 (8 bit) on M5 Max 128GB.\n\n- Video Generation via Local LLM\n- via MLX Serve by @ddalcu\n\nPrompt: \"A cute little girl walks through a sunny forest when a small, friendly dinosaur steps out, waves, and says hi. Suddenly, the dinosaur sneezes—and a shower of colorful butterflies bursts from its nose. They both laugh. Cute, colorful animated style.\"\n\n- time: 15min ish\n- 768x768",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 342,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085112656328098132",
    "title": "高速照片蒙太奇与日文手写字",
    "titleEn": "Fast Photo Montage with Handwritten Japanese Type",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "高速照片蒙太奇与日文手写字的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 768×1344。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for fast photo montage with handwritten japanese type on X. Native-video output: 15s · 768×1344. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "タイトル｜SUMMER CONTACT SHEET サブタイトル｜KINETIC PHOTO MONTAGE 尺｜15秒 画角｜9:16 解像度｜1080×1920 FPS｜24 ジャンル｜高速フォトモンタージュ／サマーファッション広告／日本語手書きタイポグラフィ／モーショングラフィックス コンセプト｜海辺で過ごした夏の一日を、縦長写真を中心にした高速フォトモンタージュで振り返る。写真をマス目状に並べず、1枚の大きな写真へ次の写真が右、左、上、下、斜めから差し込まれ、前の写真を押し出しながら連鎖する。画面内へ同時に表示する写真は基本1〜3枚。約22〜24枚の夏の断片を高速で見せ、最後だけ切り替えを止める。選ばれた一枚の縦写真が9:16全画面へ広がり、手書き文字で「この一枚が、夏になる。」と書いて締める。 主人公設定｜実写風の成人女性。健康的で自然な美しさ。未成年に見えない。白とコバルトブルーを基調にした洗練された水着。グラビアではなく爽やかなサマーファッション広告の印象。髪はミディアム〜ロングで海風になびく。全写真で同じ顔、髪型、体型、水着、配色を維持する。自然な笑顔、振り向き、歩く、走る、水に触れる、空を見る、砂浜で休むなど、夏の一日が伝わる動作を使う。 写真設計｜縦長9:16を主役フォーマットとし、横長16:9、正方形1:1、細い横帯、細い縦帯をアクセントとして使用する。主役写真は常に画面の60〜100％を占める。小写真を大量に並べず、大きな写真を残したまま次の写真を重ねる。写真には白またはオフホワイトの細い縁、軽いフィルムグレイン、フィルム番号、赤・青・黄色の手書きマーカーを加える。 写真内容｜後ろ姿で海を見る／振り向く笑顔／風になびく髪／波に触れる足元／水を蹴る／走る全身／水しぶき越しの顔／サングラスを持つ手／冷たいドリンク／砂浜の足跡／青空と雲／海面の反射／砂へ触れる手／座って海を見る／寝転んで空を見る／遠くを見る横顔／タオルや羽織りを持つ／夕方のシルエットなど。人物、風景、手元、足元、ディテールを交互に見せる。 色設計｜コバルトブルー、シアン、白、オフホワイトを基本とし、手書き文字とマーカーにレモンイエロー、朱赤、濃いブルーを使用する。序盤は青と白を中心に静かに始め、中盤で黄色と赤を増やし、終盤は柔らかな午後の光へ変化する。 文字設計｜途中の短い手書き文字は「風」「ひかり」「いま」「まぶしい」「自由」「波」。写真の余白や境界線に短時間だけ表示し、写真と一緒に移動・押し出し・消失させる。画面下へ字幕として固定しない。最終コピーは「この一枚が、夏になる。」。最後の写真上へ、実際にマーカーで書くように一画ずつ表示する。 使用写真数｜約22〜24枚。CUT01＝3枚、CUT02＝5枚、CUT03＝7〜8枚、CUT04＝7〜8枚、CUT05＝最終写真1枚。同時表示は最大3枚程度。全写真を一覧表示しない。 CUT01｜0:00〜0:02.5｜Visuals: 画面中央に、海を見つめる主人公の後ろ姿を写した縦長9:16写真が一枚だけ全画面表示される。写真内の髪と波が0.3秒だけ動き、シャッター音で静止。0:00.8で右から振り向く顔の縦写真が入り、最初の写真を左へ約35％押し出す。0:01.5で下から波に触れる足元の横長写真がせり上がり、画面下部を覆う。0:02.1で足元写真が上へ抜け、後ろ姿写真が再び中央へ戻る。｜Camera: 後ろ姿35mm・F4.5・緩やかな前進、顔70mm・F2.8、足元24mm・F5.0・ローアングル。｜Motion: 右→左、下→上の写真スライド。写真内の動きは0.2〜0.4秒で止める。｜Lighting: 柔らかな夏の自然光、海面のきらめき、人物を暗くしない軽い逆光。｜Typography: 「風」を細い白の手書き文字で顔写真の髪の流れに沿って表示。顔写真と一緒に左へ押し出されて消える。｜VFX: 細い白枠、フィルム番号01〜03、軽い粒子、青い小さな矢印。｜SFX: シャッター3回、波、風、写真の紙擦れ音。｜Emotion: 夏の入口、静かな期待感。｜Transition: 足元の白い水しぶきが下から画面全体へ広がり、次の写真へ白マスクで切り替える。 CUT02｜0:02.5〜0:05.0｜Visuals: 白い水しぶきの中から主人公の笑顔アップが全画面へ出現。0:02.9で左から髪が風になびく細い縦帯写真が差し込み、顔写真を右へ押す。0:03.3で上から青空と雲の横長写真が降り、画面上部を約35％覆う。0:03.7で右下からサングラスを持つ手元の正方形写真が斜めに滑り込む。0:04.1で中央から海面の反射写真が拡大して前の写真を覆う。0:04.6で走り出す主人公の縦写真が右から入り、次カットへつなぐ。｜Camera: 顔85mm・F2.8、髪100mm・F2.5、空24mm・F5.6、手元50mm・F3.5、海面100mm・F4.0、全身35mm・F4.0。｜Motion: 左右の押し出し、上からの落下、右下からの斜め進入、中央拡大を連続。1写真0.3〜0.5秒。｜Lighting: 真夏の強い光へ変化。髪、肩、海面へ明るいハイライト。｜Typography: 海面写真に黄色の「ひかり」、走る写真の右上へ青い「いま」。文字は写真と同時に進入し、次の写真に押し出される。｜VFX: 黄色い手描き線、白い丸囲み、赤い小矢印、写真境界の軽い影。｜SFX: シャッター連写、キラッ、写真スライド、風。｜Emotion: 夏が一気に動き出す高揚感。｜Transition: 走る主人公が画面下から上へ駆け抜け、その動線に沿って次の写真が下からせり上がる。 CUT03｜0:05.0〜0:08.0｜Visuals: 最も高速なフォトモンタージュ。0:05.0走る全身の縦写真、0:05.35下から水を蹴る横長写真、0:05.7左から水しぶき越しの笑顔、0:06.05右上から冷たいドリンクの手元、0:06.4斜め境界で砂浜の足跡、0:06.75下から海を見上げる横顔、0:07.1右からサングラス越しの目元、0:07.45中央から両手を広げる全身写真が急拡大する。画面内には常に1〜3枚だけ残し、前写真を完全に消す前に次写真が一瞬だけ重なる。｜Camera: 全身24mm、足元20mm、水しぶき35mm、手元70mm、足跡35mm、横顔85mm、目元100mm、全身28mm。｜Motion: 下→上、左→右、右上→左下、中央拡大など進入方向を毎回変える。主人公の視線、手足、水しぶきの方向を次写真へつなぐマッチカットを使う。｜Lighting: 真夏のハードライト。水滴、氷、肌、髪の輪郭が鮮明に光る。｜Typography: 水しぶき写真に白抜きの「まぶしい」、最後の全身写真の背景へ巨大な青文字「自由」。文字は人物を隠さず、一部が写真枠の外へはみ出す。｜VFX: 青と黄色の色帯、ハーフトーン、赤いマーカー丸、手描き矢印。｜SFX: 高速シャッター、水音、氷音、シュッという写真移動音、軽いポップ音。｜Emotion: 夏のピーク、自由な開放感。｜Transition: 両手を広げる写真へ急接近し、水着の白い部分をマスクとして次カットへ切り替える。 CUT04｜0:08.0〜0:12.0｜Visuals: 高速さを残しながら、夏の時間を昼から夕方へ進める。0:08.0砂浜に座って海を見る縦写真、0:08.5右から波に触れる手元写真、0:09.0上から空と雲の横長写真、0:09.45左下から寝転んで空を見る写真、0:09.9右から笑いながら振り向く顔、0:10.4下からタオルを肩にかけた上半身、0:10.9左から午後の柔らかな横顔、0:11.4上下分割で夕方の海と主人公のシルエット。前の写真の一部を画面端に残しながら次へ切り替え、時間と記憶が積み重なる感覚を出す。｜Camera: 座る姿50mm、手元70mm、空24mm、寝転ぶ姿35mm、顔50mm、上半身50mm、横顔85mm、シルエット70mm。｜Motion: 前半は0.4〜0.5秒、後半は0.5〜0.7秒へ徐々に減速。最後の夕方写真が縦方向へゆっくり全画面化する。｜Lighting: 白い昼光から金色の午後光へ変化。夕方写真には柔らかなリムライト。｜Typography: 手元写真に「波」、午後の横顔に「ひかり」、夕方写真の余白へ「いま」を表示。「いま」だけ次カット直前まで少し長く残す。｜VFX: 赤いチェック、白い手描き円、フィルム番号、写真枠の柔らかな影。｜SFX: シャッター間隔が徐々に広がり、波と風が前へ出る。BGMの音数も減らす。｜Emotion: 楽しかった一日が記憶へ変化する余韻。｜Transition: 夕方写真中央に小さな白い写真枠が現れ、その枠がゆっくり拡大して最後の一枚へ変わる。 CUT05｜0:12.0〜0:15.0｜Visuals: 写真の切り替えが完全に止まり、主人公の自然な笑顔、海風になびく髪、輝く海、柔らかな午後の光が写った縦長9:16写真が全画面へ広がる。偶然撮れた忘れられない一枚のような自然さを持たせる。0:12.4から白または赤のマーカーで「この一枚が、夏になる。」を一画ずつ書き始め、0:14.3で句点を書く。小さく「SUMMER CONTACT SHEET」を補助タイトルとして入れてもよい。｜Camera: 50mm・F3.2。写真が全画面化したあと、2〜3％だけゆっくり前進。文字完成時に停止。｜Motion: 最初の0.5秒だけ髪と波がわずかに動き、その後写真として静止。文字は手書きストロークで現れ、句点で小さくインクがにじむ。｜Lighting: 柔らかな午後の光。顔、髪、海を明るく見せ、自然な広告写真として整える。｜Typography: 「この一枚が、夏になる。」。読みやすい手書きマーカー文字。顔や身体へ重ならないよう、空または海の余白部分へ縦書きと横書きを組み合わせて配置してもよい。｜VFX: 軽いフィルムグレイン、微細な印刷ムラ。ラストではハーフトーンや色帯を減らし、写真と文字を主役にする。｜SFX: マーカー筆記音、句点で最後のシャッター音、波と風だけを残す。｜Emotion: 流れていった多くの瞬間から、一枚が夏の記憶として残る。｜Transition: 一枚の写真として静止終了。 編集テンポ｜CUT01は0.7〜1秒単位。CUT02は0.3〜0.5秒単位。CUT03は0.3〜0.4秒単位。CUT04は0.4〜0.7秒単位から徐々に減速。CUT05は一枚を3秒間見せる。高速パートと静止パートの差を明確にする。 カメラ・レイアウトルール｜完成画角は9:16。縦写真を主役として大きく使う。左右からの写真は主役を30〜50％押し出し、上下写真は画面の30〜45％を一時的に覆う。斜め進入も使用するが、常に人物の顔や視線が見えるようにする。同時表示は最大3枚。均等分割、整列グリッド、写真一覧は禁止。 BGM｜爽やかで少しエモい夏のポップインスト。0:00〜0:05は軽いリズム、0:05〜0:08でテンポと音数を最大化、0:08〜0:12で少しずつ減速、0:12以降はメロディーと環境音を残して着地する。 必須条件｜完成動画9:16／約22〜24枚の夏写真を高速で紹介／写真一覧を表示しない／常時1〜3枚／右・左・上・下・斜めから写真が進入／写真だけで夏の一日が伝わる／途中に短い日本語手書きタイポ／最後は一枚の縦写真へ収束／「この一枚が、夏になる。」を手書きで表示／同一人物と同じ水着を全写真で維持する。 禁止事項｜写真グリッド／均等なマス目／全写真の同時表示／小さな写真を大量に並べる／常時4枚以上の画面分割／同じ顔写真だけの連続／字幕帯／意味不明な長文文字／暗すぎる写真／激しいグリッチ／未成年に見える人物／過度に性的なポーズ／人物の別人化。",
    "promptSourceUrl": "https://x.com/su_nagomi/status/2085112660534964428",
    "sourceUrl": "https://x.com/su_nagomi/status/2085112656328098132",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-05T21:16:11.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085112656328098132.mp4",
    "posterUrl": "/posters/x/2085112656328098132.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "高速フォトモンタージュ／日本語手書きタイポグラフィ／モーショングラフィックス\n\n写真もきちんと描写されてるし、日本語破綻もない。\nHailuoのMiniMax H3は本当に素晴らしいモデルだと思います。\n\n動画プロンプトはリプ欄で👇\n@Hailuo_AI #MiniMaxH3",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 2,
      "views": 1014,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085097057073709123",
    "title": "消费级显卡上的七秒场景",
    "titleEn": "A Seven-Second Scene on an RTX 4090",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "消费级显卡上的七秒场景的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 7 秒 · 1376×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a seven-second scene on an rtx 4090 on X. Native-video output: 7s · 1376×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "7-second video. The Office (US) mockumentary style, inside the Dunder Mifflin conference room in Scranton. Harsh fluorescent office lighting, beige walls, cheap chairs, whiteboard, muted colors, slightly soft consumer-documentary camera quality. Michael Scott stands at the head of the table while Jim, Pam, Dwight, and Stanley sit around him. Handheld single-camera cinematography with subtle shake, sudden reframing, and fast reaction zooms.\n\nMichael raises one hand with complete sincerity, defending an absurd opinion.\n\nMICHAEL: “Everybody says Donald Trump’s fart smelled terrible. I was there. Honestly? Not that bad.”\n\nInstant snap zoom to Dwight, who nods gravely.\n\nDWIGHT: “Strong. Presidential.”\n\nJim slowly turns toward the documentary camera with a silent, deeply uncomfortable stare. Pam presses her lips together, trying not to laugh. Stanley keeps reading his crossword without looking up.\n\nNatural English dialogue, precise lip synchronization, realistic overlapping office ambience, awkward silence after the punchline, no laugh track, no subtitles, no on-screen text, dry deadpan comedy, authentic mockumentary performances.",
    "promptSourceUrl": "https://x.com/uniqornate/status/2085097057073709123",
    "sourceUrl": "https://x.com/uniqornate/status/2085097057073709123",
    "sourceLabel": "X 原帖 · @uniqornate",
    "author": "uniqornate",
    "publishedAt": "2026-08-05T20:14:12.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085097057073709123.mp4",
    "posterUrl": "/posters/x/2085097057073709123.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@Hailuo_AI  MiniMax H3 - RTX 4090 - 1MP (1376 x 768) - It took around 12 min to generate 7 second clip with Sage Att + Sol Att + Chunk Feed FFN + Spectrum Apply combined together. (official comfy ref2va_pruned_int8_convrot - 20 step) Prompt;\n\n7-second video. The Office (US) mockumentary style, inside the Dunder Mifflin conference room in Scranton. Harsh fluorescent office lighting, beige walls, cheap chairs, whiteboard, muted colors, slightly soft consumer-documentary camera quality. Michael Scott stands at the head of the table while Jim, Pam, Dwight, and Stanley sit around him. Handheld single-camera cinematography with subtle shake, sudden reframing, and fast reaction zooms.\n\nMichael raises one hand with complete sincerity, defending an absurd opinion.\n\nMICHAEL: “Everybody says Donald Trump’s fart smelled terrible. I was there. Honestly? Not that bad.”\n\nInstant snap zoom to Dwight, who nods gravely.\n\nDWIGHT: “Strong. Presidential.”\n\nJim slowly turns toward the documentary camera with a silent, deeply uncomfortable stare. Pam presses her lips together, trying not to laugh. Stanley keeps reading his crossword without looking up.\n\nNatural English dialogue, precise lip synchronization, realistic overlapping office ambience, awkward silence after the punchline, no laugh track, no subtitles, no on-screen text, dry deadpan comedy, authentic mockumentary performances.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 276,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085009719371190553",
    "title": "吊床上的午后小睡",
    "titleEn": "An Afternoon Nap in a Hammock",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "吊床上的午后小睡的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an afternoon nap in a hammock on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Main Subject:\nYoung Korean woman, early 20s, oversized soft cardigan over shorts, hair loose and slightly messy, relaxed sleepy expression, natural unguarded energy.\nLocation:\nSmall backyard with a hammock strung between two trees, early afternoon. Dappled sunlight through leaves, a small side table with a book, quiet garden sounds. No modern clutter.\nVisual Style:\nUltra-realistic documentary realism, intimate candid feeling, soft warm afternoon light.\nCamera Style:\nEarly 2000s DV camcorder, handheld gentle drift, autofocus adjusting slowly to shifting light through leaves, warm faded color cast, soft grain. No stabilization.\nTimeline (30 sec):\n00:00–00:03 → She climbs into the hammock carefully, adjusting her position.\n00:03–00:06 → She picks up a book from the side table, flipping to her page.\n00:06–00:09 → She reads quietly for a moment, then yawns softly.\n00:09–00:12 → She sets the book on her stomach, then closes her eyes briefly.\n00:12–00:15 → A breeze sways the hammock gently; she smiles without opening her eyes.\n00:15–00:18 → She peeks one eye open at camera saying \"여기 진짜 시원하다\" (\"It's really cool here\").\n00:18–00:21 → She stretches one arm lazily, then lets it dangle off the hammock.\n00:21–00:24 → Dappled light shifts across her face as leaves move overhead.\n00:24–00:27 → She murmurs sleepily, then closes her eyes again, fully relaxed.\n00:27–00:30 → She whispers to camera \"나 좀 잘게...\" (\"I'm going to sleep a bit...\"), drifting off as it fades.\nAudio:\nLeaves rustling, hammock creaking softly, distant birds, faint wind. Her dialogue as noted above. No music.\nGoal: A calm, drowsy afternoon rest moment intimate, soft, deeply believable.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2085009719371190553",
    "sourceUrl": "https://x.com/Strength04_X/status/2085009719371190553",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-05T14:27:09.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085009719371190553.mp4",
    "posterUrl": "/posters/x/2085009719371190553.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Afternoon Nap in a Hammock \n\nCreated this using MiniMax H3 on @Hailuo_AI\n\nPrompt - Main Subject:\nYoung Korean woman, early 20s, oversized soft cardigan over shorts, hair loose and slightly messy, relaxed sleepy expression, natural unguarded energy.\nLocation:\nSmall backyard with a hammock strung between two trees, early afternoon. Dappled sunlight through leaves, a small side table with a book, quiet garden sounds. No modern clutter.\nVisual Style:\nUltra-realistic documentary realism, intimate candid feeling, soft warm afternoon light.\nCamera Style:\nEarly 2000s DV camcorder, handheld gentle drift, autofocus adjusting slowly to shifting light through leaves, warm faded color cast, soft grain. No stabilization.\nTimeline (30 sec):\n00:00–00:03 → She climbs into the hammock carefully, adjusting her position.\n00:03–00:06 → She picks up a book from the side table, flipping to her page.\n00:06–00:09 → She reads quietly for a moment, then yawns softly.\n00:09–00:12 → She sets the book on her stomach, then closes her eyes briefly.\n00:12–00:15 → A breeze sways the hammock gently; she smiles without opening her eyes.\n00:15–00:18 → She peeks one eye open at camera saying \"여기 진짜 시원하다\" (\"It's really cool here\").\n00:18–00:21 → She stretches one arm lazily, then lets it dangle off the hammock.\n00:21–00:24 → Dappled light shifts across her face as leaves move overhead.\n00:24–00:27 → She murmurs sleepily, then closes her eyes again, fully relaxed.\n00:27–00:30 → She whispers to camera \"나 좀 잘게...\" (\"I'm going to sleep a bit...\"), drifting off as it fades.\nAudio:\nLeaves rustling, hammock creaking softly, distant birds, faint wind. Her dialogue as noted above. No music.\nGoal: A calm, drowsy afternoon rest moment intimate, soft, deeply believable.",
    "engagement": {
      "replies": 26,
      "reposts": 3,
      "likes": 103,
      "views": 10081,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085006715339301360",
    "title": "镜头前的戏剧化表演",
    "titleEn": "A Dramatic Performance for the Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头前的戏剧化表演的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a dramatic performance for the camera on X. Native-video output: 15s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Gritty hand held independent thriller movie, an extreme closeup of a gaunt man named Charlie trying to smoke a cigarette but his hand is trembling. He is in his 40s with messy brown hair and some stubble wearing a white dress shirt with the top button undone and a black tie that has been loosened, he has some dark red blood splatters on his face, he is standing in a dark room with side lighting coming from a window with some newspaper glued to the glass and there is a whole in the window thats cracked and allowing a beam of light to come in through the dark room.\n\nStyle: Cinematic, Film grain, dark shadows, crushed blacks.\n\nVideo prompt made with @magnific  and @Hailuo_AI : \n\nGritty hand held independent thriller movie, the camera starts off on an extreme closeup of a gaunt man named Charlie trying to smoke a cigarette but his hand is trembling, he eventually pulls the cigarette away and the camera slowly moves back to reveal more the man. He is in his 40s with messy brown hair and some stubble wearing a white dress shirt with the top button undone and a black tie that has been loosened, he has some dark red blood splatters on his face, he is standing in a dark room with side lighting coming from a window with some newspaper glued to the glass and there is a whole in the window thats cracked and allowing a beam of light to come in through the dark room. The room is a bedroom and on floor is a man turned upside down laying down and that man is wearing a white tank top and and dress pants he has a gun wound on his back and is laying in a pool of blood, then Charlie is holding a black revolver with also a shaking hand and he accidently drops the gun on the floor and he then says with a trembling voice \"fuck Charlie, you really fucked up this time.\"\n\nHe then starts to raise up the cigarette again but then hesitates and screams \"fuck!\"  with his hands up and his head bent down in total frustration, he then walks out of the frame toward the left\n\nStyle: Film grain, dark shadows, crushed blacks, in and out of focus.",
    "promptSourceUrl": "https://x.com/EccentrismArt/status/2085006722926776687",
    "sourceUrl": "https://x.com/EccentrismArt/status/2085006715339301360",
    "sourceLabel": "X 原帖 · @EccentrismArt",
    "author": "Jer at EccentrismArt",
    "publishedAt": "2026-08-05T14:15:13.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2085006715339301360.mp4",
    "posterUrl": "/posters/x/2085006715339301360.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Dramatic performances ✅\nLess content restrictions ✅\nCheaper than other models ✅\n\nHailuo H3 is checking a lot of boxes!\n\nWhat do you all think?\n\nPrompt in the comments ⬇️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 370,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084932583335624724",
    "title": "奇幻古埃及之旅",
    "titleEn": "A Fantasy Journey Through Ancient Egypt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "奇幻古埃及之旅的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1080×1350。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a fantasy journey through ancient egypt on X. Native-video output: 15s · 1080×1350. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15 seconds | 16:9 | live-action historical construction drama\nFictional dramatization. Documentary-style visual realism.\n\nSETTING\n\nAncient Egypt along the Nile during sunrise beside the Great Pyramids.\n\nThe uploaded reference image is the exact opening frame.\n\nMaintain the identical traveler, boats, pyramids, moon, river, shoreline, lighting and composition.\n\nConstruction activity gradually becomes visible deeper into the environment while preserving the original scene.\n\nWarm desert atmosphere.\nMorning mist.\nLight dust drifting across the valley.\n\nWORKFORCE\n\nA master builder overseeing the day's construction.\n\nHundreds of Egyptian laborers, stone masons, engineers, rope teams, carpenters and transport crews gradually become active throughout the distant construction zone.\n\nMaintain historically accurate clothing, tools, wooden machinery and consistent character identities.\n\nHuman behaviour remains disciplined, coordinated and believable.\n\n---\n\nSTORY\n\n00–04s — ESTABLISHING\n\nThe traveler stands silently at the riverbank exactly as shown in the reference image.\n\nThe camera slowly pushes forward from behind.\n\nFeluccas drift naturally.\n\nPalm trees sway gently.\n\nBirds fly across the sunrise.\n\nWorkers begin appearing in the distant background carrying ropes, wooden beams and stone tools toward the pyramid complex.\n\nConstruction slowly awakens with the rising sun.\n\n---\n\n04–08s — OPERATION\n\nThe traveler picks up the clay vessel and begins walking along the shoreline.\n\nThe camera follows naturally behind him.\n\nAcross the river, dozens of workers unload massive limestone blocks from transport boats.\n\nOxen pull wooden sledges.\n\nLong rope teams drag heavy stones toward enormous earthen ramps.\n\nWooden cranes rotate slowly as supervisors organize the workforce.\n\nDust rises naturally beneath hundreds of moving feet.\n\n---\n\n08–12s — CONSTRUCTION\n\nThe traveler reaches an elevated viewpoint overlooking the worksite.\n\nThe camera gently reveals the enormous construction operation.\n\nHundreds of synchronized workers pull a gigantic limestone block upward using thick ropes and wooden rollers.\n\nStone masons shape blocks with copper chisels.\n\nCarpenters assemble scaffolding.\n\nEngineers inspect alignment while foremen direct the operation with calm hand signals.\n\nThe environment feels immense, alive and historically authentic.\n\n---\n\n12–15s — LEGACY\n\nThe camera slowly cranes upward into a breathtaking wide reveal.\n\nThe traveler stands watching history unfold before him.\n\nThousands of workers continue building across the vast construction site.\n\nFeluccas constantly arrive carrying fresh limestone.\n\nDust drifts through golden sunrise light.\n\nThe pyramids dominate the horizon as one final massive stone settles perfectly into position.\n\nEnd with the feeling that humanity's greatest engineering achievement is being built before our eyes.\n\n---\n\nCAMERA\n\nGrounded historical-documentary cinematography.\n\nMix:\n\nWide environmental establishing shots.\n\nMedium handheld following shots.\n\nBrief observational close-ups.\n\nSlow cinematic dolly movement.\n\nNatural handheld instability.\n\nRealistic operator movement.\n\nGentle crane rise for the final reveal.\n\nNo impossible drone transitions.\n\nNo excessive slow motion.\n\n---\n\nLIGHTING\n\nNatural sunrise illumination.\n\nWarm golden desert light.\n\nSoft atmospheric haze.\n\nRealistic exposure adaptation.\n\nNatural shadows.\n\nDetailed limestone texture.\n\nPhysically accurate reflections across the Nile.\n\nNo artificial rim lighting.\n\n---\n\nPHYSICS\n\nReal gravity.\n\nAuthentic weight of limestone blocks.\n\nNatural rope tension.\n\nWooden cranes flex realistically.\n\nWorkers visibly strain under heavy loads.\n\nDust reacts to footsteps.\n\nBoats move naturally with river currents.\n\nWater responds realistically.\n\nWind moves robes, sails and palm leaves.\n\nEverything follows believable engineering principles.\n\n---\n\nHUMAN PERFORMANCE\n\nRestrained, believable teamwork.\n\nHeavy breathing.\n\nNatural communication.\n\nSimple hand signals.\n\nRealistic lifting mechanics.\n\nMeasured walking.\n\nVisible physical effort.\n\nNo exaggerated cinematic acting.\n\n---\n\nAUDIO\n\nMorning wind.\n\nFlowing Nile water.\n\nBoat hulls creaking.\n\nWorkers shouting coordinated commands.\n\nStone scraping.\n\nHammer and chisel impacts.\n\nHeavy rope tension.\n\nFootsteps on sand.\n\nDistant oxen.\n\nNatural historical ambience.\n\nNo oversized Hollywood trailer effects.\n\n---\n\nVISUAL CHARACTER\n\nPhotorealistic live-action footage.\n\nHistorical documentary realism.\n\nMuseum-quality reconstruction.\n\nNatural skin texture.\n\nAuthentic linen garments.\n\nWeathered limestone.\n\nSubtle sensor noise.\n\nNatural motion blur.\n\nAvoid:\n\nFantasy elements.\n\nMagic.\n\nCGI appearance.\n\nOversaturated HDR.\n\nModern machinery.\n\nFloating debris.\n\nPerfect choreography.\n\nBeauty-filter faces.\n\n---\n\nCONTINUITY\n\nMaintain the exact character identity, clothing, boats, pyramids, moon, river, shoreline, architecture and lighting established in the uploaded reference image.\n\nConstruction activity should emerge naturally from the existing environment without changing the original composition.\n\nEvery action begins from the physical state established by the previous shot.\n\nhttps://tinyurl.com/dhh4wvyu",
    "promptSourceUrl": "https://x.com/Diplomeme/status/2084932849191493878",
    "sourceUrl": "https://x.com/Diplomeme/status/2084932583335624724",
    "sourceLabel": "X 原帖 · @Diplomeme",
    "author": "ᴍᴜʀᴘʜʏ",
    "publishedAt": "2026-08-05T09:20:38.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084932583335624724.mp4",
    "posterUrl": "/posters/x/2084932583335624724.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1350",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 x Midjourney V8.2 on @itsPolloAI \n\nFantasy-inspired depiction of Ancient Egypt \n\nDon't miss out — both MiniMax H3 and Midjourney V8.2 are now 50% off on Pollo AI!\n\nTry it with Prompt in comments : ⤵️",
    "engagement": {
      "replies": 54,
      "reposts": 3,
      "likes": 79,
      "views": 1854,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084910486362247434",
    "title": "星花魔法都市的游戏预告",
    "titleEn": "A Game Teaser in the City of Starflowers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "星花魔法都市的游戏预告的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a game teaser in the city of starflowers on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "高品質アニメ映像。\n\n作品ジャンルは、プレミアム3Dトゥーン／セルルックで描くダークファンタジー探索ゲームの魔法ミステリー・タイトルシークエンス。黒いネガティブスペース、非対称分割、細い銀青フレーム、硬い幾何学マスクで構成する。恐怖や悲劇ではなく、冷静で上品な緊張感。主人公は全編で一人だけ。顔演技は、目、眉、まぶた、口元、視線の順に小さく変化させる。\n\n参照役割を厳密に分離する。Image1は顔参照であり、顔とキャラクター同一性の最優先参照。Image2は衣装参照であり、衣装・装飾・全身デザイン・体格・刀と装備の参照。優先順位は、顔立ち、目、虹彩色、眉、口元、髪型、髪色、前髪、顔まわりの色はImage1、衣装構造、身体シルエット、装飾、装備はImage2。参照内容が食い違う場合もこの優先順位を守る。Image2内の文字、レイアウト、背景、ポーズ見本、表情一覧、小さな顔バリエーションは参照しない。\n\nキャラクター固定：全カットで同一人物。小さく繊細な卵形の顔、柔らかな顎線、淡い肌、透明感のある大きなサファイアブルーの瞳、奥行きのある多層虹彩、左右それぞれの小さな複数キャッチライト、細い上まつ毛、自然な下まつ毛を維持する。白銀にごく淡い青紫を含む長い髪、柔らかな薄い前髪、顔沿いの細い横髪、腰まで流れる豊かなウェーブを維持する。頭の左側には、青い薔薇と黒青の蝶翅、青い雫飾りが一体になった髪飾り一組だけを固定し、飛び去らせず、実在の蝶へ変えず、複製しない。\n\nImage2の黒紺を基調にした高密度ゴシック衣装を維持する。繊細な黒レースのハイネックと透け袖、青い胸元宝石、銀の細鎖、青いサテンリボン、非対称の多層スカート、黒いブーツ、細身で小柄な約7.5頭身のシルエットを固定する。全カットで顔、目、髪、衣装、体格、装飾、固有配色を変えない。変化してよいのは、表情、視線、呼吸、歩行姿勢、髪・レース・リボンの自然な遅れだけ。特徴混合、顔の平均化、衣装交換、別人化、分身、クローン、余計な人物を生じさせない。\n\n能動装備として主人公が所有する刀は全編で刀一本だけ。開始時から左腰の鞘へ完全に納まり、黒紺の鞘、鍔、柄、銀の鎖飾り、青い房を一続きの同じ構造で保つ。開始時、右手は柄のすぐ横で開いている。途中で黒い手袋の右手が柄だけへ静かに触れ、鍔が手と鞘側を明確に分離したまま指を締める。刀身は一度も抜かず、最終画でも刀一本のまま完全に納刀され、右手は柄に静止する。二刀流、刀の複製、柄・鍔・鞘の分離、刃を握る手を出さない。\n\nレンダリング固定：全編で3Dトゥーン／セルルックの一系統だけ。細く制御された濃紺とチャコールの色線、二段から三段の明快なトゥーン影と透明な中間グラデーション、白銀・深紺・ロイヤルブルー・銀を主役にした澄んだ高彩度パレットを維持する。瞳と髪には多層ハイライト、金属と青い結晶には局所的で鋭い反射を置く。肌は柔らかなマット、髪は絹、レースは繊細な半透明、サテンは抑えた光沢、革は低光沢、銀金属は硬質、結晶は多面反射、濡れた石床は深い鏡面として描き分ける。顔、髪飾り、衣装、刀、建築まで高密度。キーライトと影色はキャラクター固有色を優先し、青い環境光は補助に限定し、背景は主役より抑えた低彩度の青黒にする。太い輪郭、平坦な単層セル影、汎用3D美少女顔、滑らかなプラスチックCG、低密度背景、半写実、写実、くすんだ人物色、画風混合にしない。\n\n舞台は月光の差す青黒いゴシック書庫回廊。奥に閉じた一枚の高いアーチ扉、床に磨かれた黒い石、側面に青いステンドグラス光、石のリブと書架の細部が続く。建築は石、木、ガラスだけで成立させ、キャラクター固有の薔薇、宝石、鎖、蝶飾りを柱や壁へ複製しない。唯一の異常は、現実の扉が閉じたままなのに、床の鏡面反射の中だけ同じ扉が青い奥行きへ開いていること。カメラは狭い顔極近景から回廊の広い中景へスケールを変え、胸上近景へ戻る。各移動の目的は、扉と反射の食い違い、右手と刀の状態、回廊の奥行き、表情の理解、タイトル余白を順に読ませること。終端空間は同じ回廊の閉じた扉前で、別の場所へ移らない。\n\n黒から一度だけ硬く切り替える。大きな主パネルでは、カメラは少女の左斜め前、目線の高さ、顔の極近景。両目、鼻筋、口元、輪郭、髪飾りを一続きに見せ、顔へ枠線を重ねない。画面右下の細い副パネルでは、正面の閉じたアーチ扉と、その真下で既に開いている床反射を同時に見せ、異常を一目で比較できる。少女の瞳が静かに下へ動き、反射の異常を捉える。反射内のアーチ輪郭が硬い円弧マスクへ拡大し、次の画へ切り替える。\n\n高い斜め上から濡れた黒石床へ寄り、上側に現実の閉じた扉、下側に開いた反射扉を同じ奥行き軸で保つ。反射の青い縁から一本の細い光線が床を走り、胸元の青い宝石の輪郭へ連続する。その円弧を境に細いタイルが非対称に組み替わり、宝石の近景、右手と柄の近景、鞘の青い象嵌を順に見せる。各タイルの縁は顔を横切らない。右手が柄へ触れた瞬間だけ黒革の小さな擦れ、銀鎖の短い硬質な触れ音、レースの微かな衣擦れを同期させる。刀は完全に鞘の中へ残す。\n\n鞘の長い直線を硬いラインスライスとして、回廊の中景へ切り替える。カメラは少女の後方左、胸の高さから斜め前へ追従し、少女を中景以上に保ちながら、書架と石柱の背景消失軸を奥の閉じた扉へ集める。少女が閉じた扉へ一歩ずつ進むたび、濡れた石床へ乾いた靴音が一回ずつ響き、開いた反射扉だけが視線の先を滑る。足元につながった一つの長い影が床を横切り、その影の端を硬いワイプとして次へ渡す。影を別人や分身にしない。\n\n影ワイプの後、扉の縦辺を硬いマスクにして、少女の右斜め前、目線よりわずかに低い胸上近景へ戻る。カメラは短く静かに寄り、表情を、目が異常を認識する、眉がごくわずかに締まる、まぶたが静かに細くなる、口元が冷静な決意へ整う、視線が現実の閉じた扉へ上がる、の順で読ませる。無表情の人形顔、恐怖、パニックにせず、瞳の層と小さなキャッチライトを保つ。右手は柄を握るが抜刀せず、青い鞘の縁が一度だけ画面を横切る硬いラインマスクになる。\n\n重要な禁止：顔を分断する太い格子、顔中央の光点や光線、巨大な蝶の接写、実在昆虫のマクロ、軟らかいディゾルブ、流体モーフ、ホラー、流血、重複タイトル、誤字、擬似文字、字幕、透かし。\n\n音楽は既存曲を模倣しない完全オリジナル。低いドローンと張り詰めた細いハイハットから始め、低いキック、断片的なウォーキングベース、短いバリトンサックスと切ったブラスへ冷静に展開する。終盤は緊張和音とドラム一打だけに絞り、その低い残響をタイトル保持へ残す。反射扉が開いて見える瞬間には透明な結晶エネルギーの短い共鳴、歩行には濡れた石床へ接するブーツの乾いた靴音、右手を柄に置く瞬間には黒革の軋み、銀鎖が一度だけ小さく鳴る音、布擦れをそれぞれ同期する。現実の木扉は動かないので開閉音を出さない。台詞、ナレーション、歌唱なし。\n\n最終画は静かな余韻とモチーフ回帰。カメラは正面より少し左、目線の高さの胸上構図。少女を画面左へ置き、反対側の余白をタイトル専用にする。右手は柄に静止し、刀一本は左腰で完全に納刀、現実の扉は閉じたまま、床反射の扉だけが開いた青い奥行きで停止する。少女は閉じた扉の方向へ視線を固定し、呼吸と毛先とレースだけがごく小さく動く。右側へ正確に「MOONLIT BLUE BUTTERFLY」を一度だけ表示し、その下へ「COMING SOON」を一度だけ表示する。完成したタイトルプレートを固定して保持し、新しい人物、蝶、小物、文字を入れない。緊張和音とドラム一打の低い残響が完全に消える正確なフレームで黒へハードカットする。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2084910486362247434",
    "sourceUrl": "https://x.com/haruuraeadss/status/2084910486362247434",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-08-05T07:52:50.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084910486362247434.mp4",
    "posterUrl": "/posters/x/2084910486362247434.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "週刊アニメ＆MVプロンプト  Vol.28\n\nゲームティザー映像\n\n@Hailuo_AI #MiniMaxH3　\n\nプロンプト👇\n                                                                                                                                                                                               \n高品質アニメ映像。\n\n作品ジャンルは、プレミアム3Dトゥーン／セルルックで描くダークファンタジー探索ゲームの魔法ミステリー・タイトルシークエンス。黒いネガティブスペース、非対称分割、細い銀青フレーム、硬い幾何学マスクで構成する。恐怖や悲劇ではなく、冷静で上品な緊張感。主人公は全編で一人だけ。顔演技は、目、眉、まぶた、口元、視線の順に小さく変化させる。\n\n参照役割を厳密に分離する。Image1は顔参照であり、顔とキャラクター同一性の最優先参照。Image2は衣装参照であり、衣装・装飾・全身デザイン・体格・刀と装備の参照。優先順位は、顔立ち、目、虹彩色、眉、口元、髪型、髪色、前髪、顔まわりの色はImage1、衣装構造、身体シルエット、装飾、装備はImage2。参照内容が食い違う場合もこの優先順位を守る。Image2内の文字、レイアウト、背景、ポーズ見本、表情一覧、小さな顔バリエーションは参照しない。\n\nキャラクター固定：全カットで同一人物。小さく繊細な卵形の顔、柔らかな顎線、淡い肌、透明感のある大きなサファイアブルーの瞳、奥行きのある多層虹彩、左右それぞれの小さな複数キャッチライト、細い上まつ毛、自然な下まつ毛を維持する。白銀にごく淡い青紫を含む長い髪、柔らかな薄い前髪、顔沿いの細い横髪、腰まで流れる豊かなウェーブを維持する。頭の左側には、青い薔薇と黒青の蝶翅、青い雫飾りが一体になった髪飾り一組だけを固定し、飛び去らせず、実在の蝶へ変えず、複製しない。\n\nImage2の黒紺を基調にした高密度ゴシック衣装を維持する。繊細な黒レースのハイネックと透け袖、青い胸元宝石、銀の細鎖、青いサテンリボン、非対称の多層スカート、黒いブーツ、細身で小柄な約7.5頭身のシルエットを固定する。全カットで顔、目、髪、衣装、体格、装飾、固有配色を変えない。変化してよいのは、表情、視線、呼吸、歩行姿勢、髪・レース・リボンの自然な遅れだけ。特徴混合、顔の平均化、衣装交換、別人化、分身、クローン、余計な人物を生じさせない。\n\n能動装備として主人公が所有する刀は全編で刀一本だけ。開始時から左腰の鞘へ完全に納まり、黒紺の鞘、鍔、柄、銀の鎖飾り、青い房を一続きの同じ構造で保つ。開始時、右手は柄のすぐ横で開いている。途中で黒い手袋の右手が柄だけへ静かに触れ、鍔が手と鞘側を明確に分離したまま指を締める。刀身は一度も抜かず、最終画でも刀一本のまま完全に納刀され、右手は柄に静止する。二刀流、刀の複製、柄・鍔・鞘の分離、刃を握る手を出さない。\n\nレンダリング固定：全編で3Dトゥーン／セルルックの一系統だけ。細く制御された濃紺とチャコールの色線、二段から三段の明快なトゥーン影と透明な中間グラデーション、白銀・深紺・ロイヤルブルー・銀を主役にした澄んだ高彩度パレットを維持する。瞳と髪には多層ハイライト、金属と青い結晶には局所的で鋭い反射を置く。肌は柔らかなマット、髪は絹、レースは繊細な半透明、サテンは抑えた光沢、革は低光沢、銀金属は硬質、結晶は多面反射、濡れた石床は深い鏡面として描き分ける。顔、髪飾り、衣装、刀、建築まで高密度。キーライトと影色はキャラクター固有色を優先し、青い環境光は補助に限定し、背景は主役より抑えた低彩度の青黒にする。太い輪郭、平坦な単層セル影、汎用3D美少女顔、滑らかなプラスチックCG、低密度背景、半写実、写実、くすんだ人物色、画風混合にしない。\n\n舞台は月光の差す青黒いゴシック書庫回廊。奥に閉じた一枚の高いアーチ扉、床に磨かれた黒い石、側面に青いステンドグラス光、石のリブと書架の細部が続く。建築は石、木、ガラスだけで成立させ、キャラクター固有の薔薇、宝石、鎖、蝶飾りを柱や壁へ複製しない。唯一の異常は、現実の扉が閉じたままなのに、床の鏡面反射の中だけ同じ扉が青い奥行きへ開いていること。カメラは狭い顔極近景から回廊の広い中景へスケールを変え、胸上近景へ戻る。各移動の目的は、扉と反射の食い違い、右手と刀の状態、回廊の奥行き、表情の理解、タイトル余白を順に読ませること。終端空間は同じ回廊の閉じた扉前で、別の場所へ移らない。\n\n黒から一度だけ硬く切り替える。大きな主パネルでは、カメラは少女の左斜め前、目線の高さ、顔の極近景。両目、鼻筋、口元、輪郭、髪飾りを一続きに見せ、顔へ枠線を重ねない。画面右下の細い副パネルでは、正面の閉じたアーチ扉と、その真下で既に開いている床反射を同時に見せ、異常を一目で比較できる。少女の瞳が静かに下へ動き、反射の異常を捉える。反射内のアーチ輪郭が硬い円弧マスクへ拡大し、次の画へ切り替える。\n\n高い斜め上から濡れた黒石床へ寄り、上側に現実の閉じた扉、下側に開いた反射扉を同じ奥行き軸で保つ。反射の青い縁から一本の細い光線が床を走り、胸元の青い宝石の輪郭へ連続する。その円弧を境に細いタイルが非対称に組み替わり、宝石の近景、右手と柄の近景、鞘の青い象嵌を順に見せる。各タイルの縁は顔を横切らない。右手が柄へ触れた瞬間だけ黒革の小さな擦れ、銀鎖の短い硬質な触れ音、レースの微かな衣擦れを同期させる。刀は完全に鞘の中へ残す。\n\n鞘の長い直線を硬いラインスライスとして、回廊の中景へ切り替える。カメラは少女の後方左、胸の高さから斜め前へ追従し、少女を中景以上に保ちながら、書架と石柱の背景消失軸を奥の閉じた扉へ集める。少女が閉じた扉へ一歩ずつ進むたび、濡れた石床へ乾いた靴音が一回ずつ響き、開いた反射扉だけが視線の先を滑る。足元につながった一つの長い影が床を横切り、その影の端を硬いワイプとして次へ渡す。影を別人や分身にしない。\n\n影ワイプの後、扉の縦辺を硬いマスクにして、少女の右斜め前、目線よりわずかに低い胸上近景へ戻る。カメラは短く静かに寄り、表情を、目が異常を認識する、眉がごくわずかに締まる、まぶたが静かに細くなる、口元が冷静な決意へ整う、視線が現実の閉じた扉へ上がる、の順で読ませる。無表情の人形顔、恐怖、パニックにせず、瞳の層と小さなキャッチライトを保つ。右手は柄を握るが抜刀せず、青い鞘の縁が一度だけ画面を横切る硬いラインマスクになる。\n\n重要な禁止：顔を分断する太い格子、顔中央の光点や光線、巨大な蝶の接写、実在昆虫のマクロ、軟らかいディゾルブ、流体モーフ、ホラー、流血、重複タイトル、誤字、擬似文字、字幕、透かし。\n\n音楽は既存曲を模倣しない完全オリジナル。低いドローンと張り詰めた細いハイハットから始め、低いキック、断片的なウォーキングベース、短いバリトンサックスと切ったブラスへ冷静に展開する。終盤は緊張和音とドラム一打だけに絞り、その低い残響をタイトル保持へ残す。反射扉が開いて見える瞬間には透明な結晶エネルギーの短い共鳴、歩行には濡れた石床へ接するブーツの乾いた靴音、右手を柄に置く瞬間には黒革の軋み、銀鎖が一度だけ小さく鳴る音、布擦れをそれぞれ同期する。現実の木扉は動かないので開閉音を出さない。台詞、ナレーション、歌唱なし。\n\n最終画は静かな余韻とモチーフ回帰。カメラは正面より少し左、目線の高さの胸上構図。少女を画面左へ置き、反対側の余白をタイトル専用にする。右手は柄に静止し、刀一本は左腰で完全に納刀、現実の扉は閉じたまま、床反射の扉だけが開いた青い奥行きで停止する。少女は閉じた扉の方向へ視線を固定し、呼吸と毛先とレースだけがごく小さく動く。右側へ正確に「MOONLIT BLUE BUTTERFLY」を一度だけ表示し、その下へ「COMING SOON」を一度だけ表示する。完成したタイトルプレートを固定して保持し、新しい人物、蝶、小物、文字を入れない。緊張和音とドラム一打の低い残響が完全に消える正確なフレームで黒へハードカットする。",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 24,
      "views": 13023,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084874623825723522",
    "title": "用一周七天学英语",
    "titleEn": "Learning the Days of the Week",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "用一周七天学英语的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2944×1248。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for learning the days of the week on X. Native-video output: 15s · 2944×1248. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second animated educational video that introduces young children to the first four days of the week: Monday, Tuesday, Wednesday, Thursday.\nLearning pattern: DAY NAME → SOUND → FUN ACTIVITY ICON → PLAYFUL ACTION → DAY NAME REPEAT\nTarget audience: children ages 3 to 6.\nVisual style: Same premium-cute aesthetic — pastel tones, rounded 3D icons, soft lighting, off-white background, unique glow per day, clean bold typography, gentle calendar-page-flip style transitions.\n0:00–0:01 Intro: Mascot bounces onto a soft floating calendar page, text \"Let's learn the days!\", tap flips to reveal Monday.\n\n▪️0:01–0:04 | Monday — School Day: Calendar page shows \"Monday\" in big rounded text. Narrator: \"Monday. Time for school!\" A tiny backpack icon bounces onto the page and wiggles happily. Word \"MONDAY\" highlighted in soft blue.\n\n▪️0:04–0:07 | Tuesday — Park Day: Page flips gently to Tuesday. Narrator: \"Tuesday. Time to play!\" A small swing icon appears and swings back and forth once. Word \"TUESDAY\" highlighted in soft orange.\n\n▪️0:07–0:10 | Wednesday — Art Day: Page flips to Wednesday. Narrator: \"Wednesday. Time to paint!\" A tiny paintbrush icon appears and draws one curved rainbow line. Word \"WEDNESDAY\" highlighted in soft purple.\n\n▪️0:10–0:13 | Thursday — Story Day: Page flips to Thursday. Narrator: \"Thursday. Time for stories!\" A small open book icon appears, pages gently flutter. Word \"THURSDAY\" highlighted in soft green.\n\n▪️0:13–0:15 Recap: Backpack, swing, paintbrush, book appear in four tiles with day names above. Mascot points to each; Narrator: \"Monday, Tuesday, Wednesday, Thursday. Great job!\" Sparkle + chime.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2084874623825723522",
    "sourceUrl": "https://x.com/Strength04_X/status/2084874623825723522",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-05T05:30:20.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084874623825723522.mp4",
    "posterUrl": "/posters/x/2084874623825723522.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I created this fun days-of-the-week lesson for kids with MiniMax H3 on @Hailuo_AI #MiniMaxH3\n\nPrompt - Create a 15-second animated educational video that introduces young children to the first four days of the week: Monday, Tuesday, Wednesday, Thursday.\nLearning pattern: DAY NAME → SOUND → FUN ACTIVITY ICON → PLAYFUL ACTION → DAY NAME REPEAT\nTarget audience: children ages 3 to 6.\nVisual style: Same premium-cute aesthetic — pastel tones, rounded 3D icons, soft lighting, off-white background, unique glow per day, clean bold typography, gentle calendar-page-flip style transitions.\n0:00–0:01 Intro: Mascot bounces onto a soft floating calendar page, text \"Let's learn the days!\", tap flips to reveal Monday.\n\n▪️0:01–0:04 | Monday — School Day: Calendar page shows \"Monday\" in big rounded text. Narrator: \"Monday. Time for school!\" A tiny backpack icon bounces onto the page and wiggles happily. Word \"MONDAY\" highlighted in soft blue.\n\n▪️0:04–0:07 | Tuesday — Park Day: Page flips gently to Tuesday. Narrator: \"Tuesday. Time to play!\" A small swing icon appears and swings back and forth once. Word \"TUESDAY\" highlighted in soft orange.\n\n▪️0:07–0:10 | Wednesday — Art Day: Page flips to Wednesday. Narrator: \"Wednesday. Time to paint!\" A tiny paintbrush icon appears and draws one curved rainbow line. Word \"WEDNESDAY\" highlighted in soft purple.\n\n▪️0:10–0:13 | Thursday — Story Day: Page flips to Thursday. Narrator: \"Thursday. Time for stories!\" A small open book icon appears, pages gently flutter. Word \"THURSDAY\" highlighted in soft green.\n\n▪️0:13–0:15 Recap: Backpack, swing, paintbrush, book appear in four tiles with day names above. Mascot points to each; Narrator: \"Monday, Tuesday, Wednesday, Thursday. Great job!\" Sparkle + chime.",
    "engagement": {
      "replies": 39,
      "reposts": 4,
      "likes": 71,
      "views": 4236,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084873926141006285",
    "title": "开源时代的影像宣言",
    "titleEn": "A Visual Manifesto for the Open-Weight Era",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "开源时代的影像宣言的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a visual manifesto for the open-weight era on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create an approximately 15-second, 16:9 grounded street-racing action-film scene. All dialogue is spoken naturally in English.  Late at night inside a private custom garage in an industrial area of Los Angeles, Roman and Tej sit shoulder to shoulder on the front bumper of a black high-performance muscle car. Each holds an ordinary paper coffee cup.  Several heavily modified street-racing cars surround them. Their bodywork reflects cool blue work lights and warm tungsten fixtures. A hydraulic vehicle lift slowly descends in the background. Tools occasionally clink in the distance while an engine produces a steady, low idling rumble.  The scene has the grounded texture of a street-racing action movie: dark metallic tones, a faint layer of garage haze, an oil-stained concrete floor, realistic automotive reflections, and restrained shoulder-mounted camerawork. The image should feel physical and substantial, not like a polished car commercial.  Keep the camera at chest height and begin with a steady medium two-shot. Use only subtle human-operated movement and natural breathing in the frame. There are no chases, explosions, or exaggerated actions. The humor comes entirely from the serious performances, eye contact, and pauses.  Roman turns toward Tej with a completely sincere and genuinely concerned expression. There is no irony in his performance, and he does not realize that the question might be funny. He treats it as a serious concern about the future of the entire team.  ROMAN, speaking with complete sincerity: “Do you think AI is gonna replace us?”  Tej slowly raises his eyes from his coffee cup. The camera makes an almost imperceptible push toward him. He pauses and looks directly into Roman’s eyes. Tej appears tired but composed, answering with extremely restrained deadpan delivery.  TEJ: “Roman, AI can’t replace you. Nothing can replace you.”  Roman slowly nods. His shoulders relax, and he is visibly relieved. He looks down and seriously processes the statement. He then looks back at Tej and carefully asks for confirmation.  ROMAN: “Because I’m that good?”  Tej stares at Roman without moving.  Hold a long, quiet, uncomfortable pause. Do not use music or editing to emphasize the joke. During the silence, the idling engine in the background gives a small mechanical cough before returning to its steady rumble.  Tej slowly looks down at his coffee as though searching for an escape from the conversation. He takes a small breath, then raises his eyes back to Roman.  TEJ, with an extremely flat and exhausted delivery: “...Sure. Let’s go with that.”  Roman nods with satisfaction, fully accepting the answer. Looking reassured, he takes a sip of coffee.  The camera does not follow Roman. Instead, it slowly pushes closer to Tej’s face. After his final line, hold on Tej’s expressionless face for one full beat. He looks straight ahead, his eyes communicating exhaustion, resignation, and a brief reconsideration of every life choice that brought him here.  Cut directly to black.  Performance requirements: Roman remains completely earnest and serious throughout. He never smirks, winks, looks into the camera, or recognizes the joke. Tej’s humor comes only from restrained eye movement, flat vocal delivery, and precisely timed pauses. He never rolls his eyes, makes an exaggerated expression, or visibly performs the comedy.  Audio requirements: All dialogue is delivered in natural, clear American English with accurate lip synchronization. Preserve the subtle rustle of the paper cups, the garage’s spatial reverb, distant tool sounds, and the low engine rumble. Maintain the specified pauses between lines. No background music, laugh track, narration, or subtitles.  Continuity requirements: Roman and Tej remain seated in the same positions throughout. Each coffee cup stays in its owner’s hand, with realistic wrist movement, weight, inertia, and liquid behavior. Preserve the characters’ faces, body types, clothing, hairstyles, background vehicles, lighting direction, and camera axis.  Negative prompt: exaggerated comedy acting, changing identities or faces, distorted fingers or coffee cups, rapid action editing, explosions or car chases, subtitles, text, logos, watermarks.",
    "promptSourceUrl": "https://x.com/underwoodxie96/status/2084873926141006285",
    "sourceUrl": "https://x.com/underwoodxie96/status/2084873926141006285",
    "sourceLabel": "X 原帖 · @underwoodxie96",
    "author": "underwood",
    "publishedAt": "2026-08-05T05:27:33.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084873926141006285.mp4",
    "posterUrl": "/posters/x/2084873926141006285.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "commercial"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 going open source could mark a turning point for AI video.\nOver the past five months, everyone has been racing to improve generation quality. The next real battleground will be cost.\nDeepSeek already rewrote the rules once. Chinese video models may be about to do it again.\nVideo prompt: \nCreate an approximately 15-second, 16:9 grounded street-racing action-film scene. All dialogue is spoken naturally in English.  Late at night inside a private custom garage in an industrial area of Los Angeles, Roman and Tej sit shoulder to shoulder on the front bumper of a black high-performance muscle car. Each holds an ordinary paper coffee cup.  Several heavily modified street-racing cars surround them. Their bodywork reflects cool blue work lights and warm tungsten fixtures. A hydraulic vehicle lift slowly descends in the background. Tools occasionally clink in the distance while an engine produces a steady, low idling rumble.  The scene has the grounded texture of a street-racing action movie: dark metallic tones, a faint layer of garage haze, an oil-stained concrete floor, realistic automotive reflections, and restrained shoulder-mounted camerawork. The image should feel physical and substantial, not like a polished car commercial.  Keep the camera at chest height and begin with a steady medium two-shot. Use only subtle human-operated movement and natural breathing in the frame. There are no chases, explosions, or exaggerated actions. The humor comes entirely from the serious performances, eye contact, and pauses.  Roman turns toward Tej with a completely sincere and genuinely concerned expression. There is no irony in his performance, and he does not realize that the question might be funny. He treats it as a serious concern about the future of the entire team.  ROMAN, speaking with complete sincerity: “Do you think AI is gonna replace us?”  Tej slowly raises his eyes from his coffee cup. The camera makes an almost imperceptible push toward him. He pauses and looks directly into Roman’s eyes. Tej appears tired but composed, answering with extremely restrained deadpan delivery.  TEJ: “Roman, AI can’t replace you. Nothing can replace you.”  Roman slowly nods. His shoulders relax, and he is visibly relieved. He looks down and seriously processes the statement. He then looks back at Tej and carefully asks for confirmation.  ROMAN: “Because I’m that good?”  Tej stares at Roman without moving.  Hold a long, quiet, uncomfortable pause. Do not use music or editing to emphasize the joke. During the silence, the idling engine in the background gives a small mechanical cough before returning to its steady rumble.  Tej slowly looks down at his coffee as though searching for an escape from the conversation. He takes a small breath, then raises his eyes back to Roman.  TEJ, with an extremely flat and exhausted delivery: “...Sure. Let’s go with that.”  Roman nods with satisfaction, fully accepting the answer. Looking reassured, he takes a sip of coffee.  The camera does not follow Roman. Instead, it slowly pushes closer to Tej’s face. After his final line, hold on Tej’s expressionless face for one full beat. He looks straight ahead, his eyes communicating exhaustion, resignation, and a brief reconsideration of every life choice that brought him here.  Cut directly to black.  Performance requirements: Roman remains completely earnest and serious throughout. He never smirks, winks, looks into the camera, or recognizes the joke. Tej’s humor comes only from restrained eye movement, flat vocal delivery, and precisely timed pauses. He never rolls his eyes, makes an exaggerated expression, or visibly performs the comedy.  Audio requirements: All dialogue is delivered in natural, clear American English with accurate lip synchronization. Preserve the subtle rustle of the paper cups, the garage’s spatial reverb, distant tool sounds, and the low engine rumble. Maintain the specified pauses between lines. No background music, laugh track, narration, or subtitles.  Continuity requirements: Roman and Tej remain seated in the same positions throughout. Each coffee cup stays in its owner’s hand, with realistic wrist movement, weight, inertia, and liquid behavior. Preserve the characters’ faces, body types, clothing, hairstyles, background vehicles, lighting direction, and camera axis.  Negative prompt: exaggerated comedy acting, changing identities or faces, distorted fingers or coffee cups, rapid action editing, explosions or car chases, subtitles, text, logos, watermarks.",
    "engagement": {
      "replies": 6,
      "reposts": 0,
      "likes": 19,
      "views": 4086,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084855587540795780",
    "title": "真人质感与人工痕迹的对照",
    "titleEn": "Balancing Photorealism and Synthetic Texture",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "真人质感与人工痕迹的对照的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for balancing photorealism and synthetic texture on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "----------------\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, photorealistic with true fine pores and real skin texture, no beauty filter. A full-body side-on wide shot of a 25-year-old woman with cool fair skin, dark almond eyes and long straight black hair, walking steadily from left to right along a colonnade of stone columns on a city street. Her face is identical in every era of the film; only her clothing, her hair styling and the world behind her change. Her stride never changes speed and she never looks at the camera until the final moment. The camera trucks right with large amplitude at slow speed, travelling alongside her for the entire film with no other camera movement and no change of direction. Foreground columns sweep across the lens as the camera moves, and each time a column wipes the frame the world behind her changes to the next era.\n\nThe film opens in the 1920s: low-saturation sepia light, gas lamps, a horse-drawn carriage, a beaded drop-waist dress and a cloche hat. At 00:03.000 a column wipes past and the world becomes the 1950s: warm amber light, a pastel car at the kerb, a chrome diner sign, a full-skirted dress and red lipstick. At 00:06.000 the next column wipes past and the world becomes the 1980s: dense pink and cyan neon reflecting off wet asphalt, a padded-shoulder jacket, heavy dark night air. At 00:09.000 the next column wipes past and the world becomes the present day: cool white daylight, glass office towers, passers-by lit by phone screens, an oversized wool coat. At 00:12.000 the last column wipes past and the world becomes the near future: pale blue light, floating holographic advertisements drifting between the columns, a sleek structured coat with a faint luminous seam.\n\nAt 00:14.000 she stops walking. The camera comes to rest with her. She turns her head and looks straight into the lens for the final beat, holding still until the end. No on-screen text or subtitles at any point.\n\noverall_soundscape: Her footsteps strike the stone at exactly the same steady pace from the first frame to the last, the one sound that never changes. Behind her the street swaps out at each era: horse hooves and a distant bell, then car horns and chattering pedestrians, then heavy rain on asphalt and a low traffic drone, then a modern city hum, then an almost silent electric whirr. Each column wipe is marked by a short dry rush of air across the lens.\n\nnon_diegetic_music: A steady four-beat pulse at a constant tempo runs from the first frame to the last, gaining one new instrument at every era change: a lone bass drum, then upright bass, then analogue synth, then layered electronic percussion, then a wide sustained synth chord. It reaches full volume as she stops and cuts off on her final look.",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2084855587540795780",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2084855587540795780",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-05T04:14:41.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084855587540795780.mp4",
    "posterUrl": "/posters/x/2084855587540795780.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "music-performance",
      "food"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3はt2vで生成したとき、人物にAIっぽい質感が\n残りますね。。。\n2Kだから余計に目立つのかもしれません。\nなので、i2vの方が良いかも🤔\n\n質感はPixVerseのV6と似ているような。\n\n プロンプト↓\n----------------\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, photorealistic with true fine pores and real skin texture, no beauty filter. A full-body side-on wide shot of a 25-year-old woman with cool fair skin, dark almond eyes and long straight black hair, walking steadily from left to right along a colonnade of stone columns on a city street. Her face is identical in every era of the film; only her clothing, her hair styling and the world behind her change. Her stride never changes speed and she never looks at the camera until the final moment. The camera trucks right with large amplitude at slow speed, travelling alongside her for the entire film with no other camera movement and no change of direction. Foreground columns sweep across the lens as the camera moves, and each time a column wipes the frame the world behind her changes to the next era.\n\nThe film opens in the 1920s: low-saturation sepia light, gas lamps, a horse-drawn carriage, a beaded drop-waist dress and a cloche hat. At 00:03.000 a column wipes past and the world becomes the 1950s: warm amber light, a pastel car at the kerb, a chrome diner sign, a full-skirted dress and red lipstick. At 00:06.000 the next column wipes past and the world becomes the 1980s: dense pink and cyan neon reflecting off wet asphalt, a padded-shoulder jacket, heavy dark night air. At 00:09.000 the next column wipes past and the world becomes the present day: cool white daylight, glass office towers, passers-by lit by phone screens, an oversized wool coat. At 00:12.000 the last column wipes past and the world becomes the near future: pale blue light, floating holographic advertisements drifting between the columns, a sleek structured coat with a faint luminous seam.\n\nAt 00:14.000 she stops walking. The camera comes to rest with her. She turns her head and looks straight into the lens for the final beat, holding still until the end. No on-screen text or subtitles at any point.\n\noverall_soundscape: Her footsteps strike the stone at exactly the same steady pace from the first frame to the last, the one sound that never changes. Behind her the street swaps out at each era: horse hooves and a distant bell, then car horns and chattering pedestrians, then heavy rain on asphalt and a low traffic drone, then a modern city hum, then an almost silent electric whirr. Each column wipe is marked by a short dry rush of air across the lens.\n\nnon_diegetic_music: A steady four-beat pulse at a constant tempo runs from the first frame to the last, gaining one new instrument at every era change: a lone bass drum, then upright bass, then analogue synth, then layered electronic percussion, then a wide sustained synth chord. It reaches full volume as she stops and cuts off on her final look.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 986,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084854693403197897",
    "title": "二千分辨率下的真人电影镜头",
    "titleEn": "A Live-Action Cinematic Study in 2K",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "二千分辨率下的真人电影镜头的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a live-action cinematic study in 2k on X. Native-video output: 15s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "--------------\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, photorealistic with true fine pores and real skin texture, no beauty filter. A wide low-angle shot in a wrecked living room at dusk: a shattered window, an overturned wooden chair, a smashed vase with white petals scattered across the floor, broken glass everywhere and dust hanging in the low amber light. A 25-year-old woman with cool fair skin, dark almond eyes, long straight black hair and a plain charcoal slip dress stands motionless in the centre of the wreckage, arms at her sides, looking straight into the lens. The camera performs an Arc Shot with small amplitude at slow speed, circling her from left to right for the whole shot and never changing direction. At 00:03.000 the destruction begins to run backwards: every object except the woman moves as if the footage were playing in reverse, while she stays in normal forward time and does not move at all. Broken glass lifts off the floor, rises through the air and knits itself back into the window pane from the outside in. The white petals float upward, gather and fold back into a whole vase that lands upright on the table. The overturned chair swings up onto its legs. Dust and small fragments are drawn inward from every corner of the room, and a pool of water on the floor pulls back up into an intact drinking glass. At 00:09.000 the last few fragments converge on the window. At 00:11.000 the final shard slots into place and the room is completely whole and undamaged, warm evening light now falling through clean glass.\n\n[Shot 2] At 00:11.500, the camera cuts to a chest-up close-up, Static Shot at eye level, of the same woman in the now-perfect room, her appearance, hair and charcoal dress unchanged. She is still motionless. She blinks once, the only human movement in the film, and the frame holds on her face until the end. No on-screen text or subtitles at any point.\n\noverall_soundscape: Deep silence in the wrecked room, broken only by the woman's slow breathing. As the reversal starts, glass shards tick and chime upward through the air and wood scrapes as the chair rights itself, under a long inward rushing sound that draws the dust back from the edges of the room. The final shard clicks into the window and everything stops into complete silence.\n\nnon_diegetic_music: A single low synth drone that begins under the wreckage and rises steadily in pitch and volume through the reassembly, joined by sparse reversed piano notes. It cuts off dead at the moment the last shard locks into the window, leaving nothing underneath.",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2084854693403197897",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2084854693403197897",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-05T04:11:08.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084854693403197897.mp4",
    "posterUrl": "/posters/x/2084854693403197897.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "引き続き、MiniMax H3の検証。\n\nt2vで2Kで生成。\n\nプロンプト↓\n--------------\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, photorealistic with true fine pores and real skin texture, no beauty filter. A wide low-angle shot in a wrecked living room at dusk: a shattered window, an overturned wooden chair, a smashed vase with white petals scattered across the floor, broken glass everywhere and dust hanging in the low amber light. A 25-year-old woman with cool fair skin, dark almond eyes, long straight black hair and a plain charcoal slip dress stands motionless in the centre of the wreckage, arms at her sides, looking straight into the lens. The camera performs an Arc Shot with small amplitude at slow speed, circling her from left to right for the whole shot and never changing direction. At 00:03.000 the destruction begins to run backwards: every object except the woman moves as if the footage were playing in reverse, while she stays in normal forward time and does not move at all. Broken glass lifts off the floor, rises through the air and knits itself back into the window pane from the outside in. The white petals float upward, gather and fold back into a whole vase that lands upright on the table. The overturned chair swings up onto its legs. Dust and small fragments are drawn inward from every corner of the room, and a pool of water on the floor pulls back up into an intact drinking glass. At 00:09.000 the last few fragments converge on the window. At 00:11.000 the final shard slots into place and the room is completely whole and undamaged, warm evening light now falling through clean glass.\n\n[Shot 2] At 00:11.500, the camera cuts to a chest-up close-up, Static Shot at eye level, of the same woman in the now-perfect room, her appearance, hair and charcoal dress unchanged. She is still motionless. She blinks once, the only human movement in the film, and the frame holds on her face until the end. No on-screen text or subtitles at any point.\n\noverall_soundscape: Deep silence in the wrecked room, broken only by the woman's slow breathing. As the reversal starts, glass shards tick and chime upward through the air and wood scrapes as the chair rights itself, under a long inward rushing sound that draws the dust back from the edges of the room. The final shard clicks into the window and everything stops into complete silence.\n\nnon_diegetic_music: A single low synth drone that begins under the wreckage and rises steadily in pitch and volume through the reassembly, joined by sparse reversed piano notes. It cuts off dead at the moment the last shard locks into the window, leaving nothing underneath.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 633,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084852295607484730",
    "title": "同一参考图的电影化对决",
    "titleEn": "A Cinematic Face-Off from the Same Reference",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一参考图的电影化对决的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 1080×1350。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a cinematic face-off from the same reference on X. Native-video output: 14s · 1080×1350. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "生成一段 15 秒、写实电影质感的动作 Vlog 视频。主角是一位 24 岁左右的中国女孩\n\n图片1 ，黑色自然长发，清晰可辨的东方面容，五官真实自然，穿着浅色针织上衣和深色半裙。地点是现代中国一线城市清晨街道\n\n图片3 ，小区门口外有包子铺、便利店、公交站牌、共享单车、梧桐树、人行道、中文店招和来往行人。视频开头像手机自拍 Vlog，手持镜头轻微晃动，真实生活感。\n\n0–2 秒：现实女孩图片1 从中国城市小区门口走出来，举着手机自拍，微笑看向镜头。背景是清晨街道、包子铺、便利店、公交站牌和中文店招。镜头是手机前置自拍视角，轻微手持晃动，环境声有早高峰人声、车辆声、早餐店蒸汽声。\n\n2–4 秒：她身后街口突然传来沉重撞击声，一只高大的黑色怪兽从公交站旁冲出，撞翻共享单车，路人惊慌逃散。女孩猛地回头，笑容消失，眼神变冷。镜头快速转向她身后，怪兽清晰出现在画面中央，身形巨大，皮肤粗糙，低吼，脚步震动地面。\n\n4–7 秒：女孩站定，镜头切到中近景，重点拍她完整变身\n图片4 。她抬手挡在身前，金色和青色神力从手腕、肩膀、发梢涌出，普通针织上衣被光纹覆盖，逐层变化成中国神话女神战甲\n图片2 。战甲有青铜金属纹理、云纹、羽状肩甲和发光腰带，长发被气流扬起，额间浮现金色神印。必须看清从现实女孩变成神话女神的过程，不要一闪而过。她低声说一句中文：“别再往前了。”\n\n7–9 秒：怪兽扑向她。镜头切到低角度广角，女神向前冲刺，脚下地面裂出金色光纹。她侧身躲开怪兽爪击，右手凝聚青金色长刃，反手斩中怪兽手臂，火花和黑色鳞片飞溅。动作要清楚，人物和怪兽都在画面中央，不要只拍光效。\n\n9–11 秒：连续近身打斗。镜头跟随女神绕到怪兽侧面，她跃起踢中怪兽胸口，怪兽后退撞到公交站牌。女神落地后双手结印，身后短暂浮现中国神话女神虚影，云纹神环展开。怪兽再次扑来，她用神力护盾正面挡住冲撞，冲击波吹动路边树叶和中文店招。\n\n11–13 秒：女神抓住怪兽破绽，向前一步，掌心爆发青金色能量，将怪兽击退到街道中央。怪兽倒地低吼，烟尘散开，路人远远停下脚步。镜头轻微晃动，像手机真实拍到这一幕。女神站在街道中央，战甲发光，背后是现代中国城市街景。\n\n13–15 秒：重点保留 2 秒现实女孩状态。女神身上的青金色神光慢慢收回，战甲一层层消散，重新变回最开始那个现实中国女孩图片1 ，浅色针织上衣和深色半裙恢复原样，额间神印消失，长发自然落下。她站在街边，轻轻喘气，看了一眼倒地的怪兽，又转头看向手机镜头，露出一点尴尬又淡定的表情，像什么都没发生一样，小声说一句中文：“上班要迟到了。” 最后画面停留在现实女孩的脸和普通穿搭上，至少保留 2 秒。\n\n全程保持同一个中国女孩的脸，不要变成欧美面孔，不要变成二次元，不要卡通风。最后必须从神话女神明确变回现实女孩，并且现实女孩状态保留完整 2 秒。\n\n打斗部分集中在 7–13 秒，必须有闪避、斩击、踢击、格挡、反击和击退怪兽，不要只用发光和烟雾代替打斗。\n\n全程无字幕、无背景音乐，仅保留城市环境声、怪兽低吼、能量音效、打斗撞击声和人物中文对白。\n\n画面保持写实电影质感，中国城市环境真实，有中文招牌和中国街景细节。",
    "promptSourceUrl": "https://x.com/joshesye/status/2084852890573615412",
    "sourceUrl": "https://x.com/joshesye/status/2084852295607484730",
    "sourceLabel": "X 原帖 · @joshesye",
    "author": "行者AI视频",
    "publishedAt": "2026-08-05T04:01:36.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084852295607484730.mp4",
    "posterUrl": "/posters/x/2084852295607484730.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "1080×1350",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 VS Seedance 2.5 \n\n使用相同的视频提示词，和参考图\n\n真人电影的质感，一致的角色形象，流畅度方面2个模型都不错。\n\n运镜流畅度方面，电影级光影，H3的效果不错。视觉特效2.5要好一些。\n\n还有一个很大的差别是H3是2K分辨率，2.5是720P，180积分 VS 390积分。\n\n在相同的使用条件下，MiniMax H3的性价比更好。\n\n效果顶尖+全能参考+超高性价比，2K 低至 ¥0.53/秒。\n\nExact Prompt 👇",
    "engagement": {
      "replies": 22,
      "reposts": 8,
      "likes": 110,
      "views": 24975,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084842607411335284",
    "title": "渔船上的拾得式恐怖片",
    "titleEn": "Found-Footage Horror on a Fishing Boat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "渔船上的拾得式恐怖片的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for found-footage horror on a fishing boat on X. Native-video output: 14s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "\"Hyper-realistic found-footage horror video, filmed by a cheap action camera mounted inside a small fishing boat on a quiet lake. A lone fisherman violently reels in something heavy. He pulls up a muddy smartphone tangled in weeds. He answers, breathing hard. A distorted voice says, \"Finally. I’ve been trying to switch places with you.\" The phone flashes blinding white and the footage glitches. When the image returns, the phone shows the terrified fisherman trapped inside its screen, silently pounding on the glass. Behind him, an identical soaking-wet fisherman slowly climbs out of the lake into the boat, notices the camera, and reaches toward the lens. Abrupt cut to black. Raw handheld realism, imperfect autofocus, clipped audio, no music, no stylization.\"",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2084842607411335284",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2084842607411335284",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-05T03:23:06.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084842607411335284.mp4",
    "posterUrl": "/posters/x/2084842607411335284.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [
      "music-performance",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "15 second found footage film with MiniMax H3\n\nPrompt: \"Hyper-realistic found-footage horror video, filmed by a cheap action camera mounted inside a small fishing boat on a quiet lake. A lone fisherman violently reels in something heavy. He pulls up a muddy smartphone tangled in weeds. He answers, breathing hard. A distorted voice says, \"Finally. I’ve been trying to switch places with you.\" The phone flashes blinding white and the footage glitches. When the image returns, the phone shows the terrified fisherman trapped inside its screen, silently pounding on the glass. Behind him, an identical soaking-wet fisherman slowly climbs out of the lake into the boat, notices the camera, and reaches toward the lens. Abrupt cut to black. Raw handheld realism, imperfect autofocus, clipped audio, no music, no stylization.\"",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 43,
      "views": 3539,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084840152715911311",
    "title": "日惹传统市场的清晨",
    "titleEn": "Morning at a Traditional Javanese Market",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "日惹传统市场的清晨的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for morning at a traditional javanese market on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Traditional Javanese market in Yogyakarta, early morning, present day, ultra-realistic handheld one-take. A bakso vendor hurries through impossibly narrow, crowded market aisles carrying a large steaming bowl of freshly prepared bakso. Shoppers and vendors step aside to let him pass. He weaves between vegetable stalls overflowing with chilies, shallots, and leafy greens, ducks beneath low-hanging tarps, squeezes past batik merchants and baskets of traditional snacks, and carefully avoids children running through the market. The market feels dense, humid, noisy, and full of life. He passes porters carrying heavy sacks on their shoulders, women bargaining in Javanese, and motorcycles slowly pushing through the narrow lane. He finally reaches a tiny food stall at the back of the market and serves the steaming bowl just as the power briefly flickers, leaving the market dim for a moment before the bustle continues. Oppressive realism, natural morning light mixed with fluorescent bulbs, worn concrete floors, damp walls, wooden stalls, hanging plastic signs, Javanese conversations, vendors calling out prices, clattering bowls, footsteps echoing through the covered market, immersive cinematic pacing. No cyberpunk, no stylization, documentary-like realism.",
    "promptSourceUrl": "https://x.com/junwatu/status/2084840152715911311",
    "sourceUrl": "https://x.com/junwatu/status/2084840152715911311",
    "sourceLabel": "X 原帖 · @junwatu",
    "author": "DΞV",
    "publishedAt": "2026-08-05T03:13:21.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084840152715911311.mp4",
    "posterUrl": "/posters/x/2084840152715911311.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "food",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 also has an impressive understanding of different cultures. \n\nThis one was created in @Hailuo_AI with text prompt only.\n\nPrompt:\nTraditional Javanese market in Yogyakarta, early morning, present day, ultra-realistic handheld one-take. A bakso vendor hurries through impossibly narrow, crowded market aisles carrying a large steaming bowl of freshly prepared bakso. Shoppers and vendors step aside to let him pass. He weaves between vegetable stalls overflowing with chilies, shallots, and leafy greens, ducks beneath low-hanging tarps, squeezes past batik merchants and baskets of traditional snacks, and carefully avoids children running through the market. The market feels dense, humid, noisy, and full of life. He passes porters carrying heavy sacks on their shoulders, women bargaining in Javanese, and motorcycles slowly pushing through the narrow lane. He finally reaches a tiny food stall at the back of the market and serves the steaming bowl just as the power briefly flickers, leaving the market dim for a moment before the bustle continues. Oppressive realism, natural morning light mixed with fluorescent bulbs, worn concrete floors, damp walls, wooden stalls, hanging plastic signs, Javanese conversations, vendors calling out prices, clattering bowls, footsteps echoing through the covered market, immersive cinematic pacing. No cyberpunk, no stylization, documentary-like realism.",
    "engagement": {
      "replies": 7,
      "reposts": 3,
      "likes": 25,
      "views": 896,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084816642031394936",
    "title": "一张图像的两种结果",
    "titleEn": "One Image, Two Different Outcomes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一张图像的两种结果的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for one image, two different outcomes on X. Native-video output: 10s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "zwei Ergebnisse.\nWelches gefällt dir besser? 👀\nhttps://goo.su/vTwQ\nEntdecke neue Ideen mit der All-in-One-Referenzfunktion! ✨",
    "promptSourceUrl": "https://x.com/DeEdimakor/status/2084816642031394936",
    "sourceUrl": "https://x.com/DeEdimakor/status/2084816642031394936",
    "sourceLabel": "X 原帖 · @DeEdimakor",
    "author": "Edimakor Deutsch",
    "publishedAt": "2026-08-05T01:39:56.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084816642031394936.mp4",
    "posterUrl": "/posters/x/2084816642031394936.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 ist jetzt in Edimakor verfügbar 🥳\nEin Bild, ein Prompt – zwei Ergebnisse.\nWelches gefällt dir besser? 👀\nhttps://goo.su/vTwQ\nEntdecke neue Ideen mit der All-in-One-Referenzfunktion! ✨",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 179,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084691296472449246",
    "title": "国风想象力之旅",
    "titleEn": "A Journey Through Chinese-Inspired Fantasy",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "国风想象力之旅的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 1920×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a journey through chinese-inspired fantasy on X. Native-video output: 10s · 1920×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "以参考图为首帧，保持原有构图、建筑布局、人物服饰与蓝紫色东方幻想夜景风格不变。\n\n5秒电影级连续镜头：镜头从河面上方以低机位缓慢向前推进，轻微掠过前景乌篷船，形成自然的前中后景视差。夜市人群熙熙攘攘地流动，岸边行人缓步穿行、驻足交谈、摊贩招呼客人，动作细微但丰富，避免所有人同步运动。河面上的小船沿不同方向缓慢划行，船夫自然摆动船桨，船身随水波轻轻起伏。两岸灯笼被晚风吹得轻微摇晃，暖黄色灯火自然闪烁，倒影在水面上被船桨和涟漪缓缓拉开。紫藤花穗随风摆动，少量花瓣飘落，远处瀑布持续流淌，山间薄雾与云层缓慢移动，月光洒在水面形成流动的银蓝色高光。\n\n整体节奏繁华但不混乱，画面稳定、动作真实、层次丰富，具有东方古城夜游的沉浸感。运镜平滑克制，无突然加速、无镜头抖动、无建筑变形、无人脸扭曲、无人物闪现、无物体穿模、无新增大型元素，保持高细节、电影级光影与史诗感。",
    "promptSourceUrl": "https://x.com/Stellakjbk/status/2084691437136863285",
    "sourceUrl": "https://x.com/Stellakjbk/status/2084691296472449246",
    "sourceLabel": "X 原帖 · @Stellakjbk",
    "author": "九尾狐-FoxFairy🦊",
    "publishedAt": "2026-08-04T17:21:51.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084691296472449246.mp4",
    "posterUrl": "/posters/x/2084691296472449246.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "用MiniMax H3做了一个国风的视频，不得不说效果是真的好。吸取前几次失败的教训，提示词不应该做的太长太复杂，终于做了一个好的作品。\n\n经验与建议：提示词做的粗糙一点，给模型留一点想象的空间。如果提示词做多了，可以精简优化一下。\n\n如果你喜欢这个视频，可以自己做一下，提示词在评论区👇，MiniMax H3的优缺点在引用的推文里👇",
    "engagement": {
      "replies": 9,
      "reposts": 9,
      "likes": 104,
      "views": 18178,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084637679254376793",
    "title": "一镜制作的高级时装广告",
    "titleEn": "A High-Fashion Commercial in One Shot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一镜制作的高级时装广告的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a high-fashion commercial in one shot on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Prompts: \n[Core Concept]\n15s, 160 BPM, 4/4, 40 beats. Industrial Hyperpop × Deconstructed Club: hard kicks, metallic snares, glitch hi-hats, distorted bass, sliced vocals. A timeline-based reality where rhythm controls space and motion.\n\n[Character Identity]\nSingle androgynous dancer with short black hair, silver eyebrow mark, asymmetric black jacket, wide pants, red stitching. Maintain identical face, body, hairstyle and outfit. Multiple figures are only past-frame projections.\n\n[World Logic]\nEach kick compresses the timeline toward the dancer. Each snare duplicates previous movements into adjacent frames. Moving against the timeline transforms the 2D interface into a foldable 3D structure. The dancer escapes the repetition loop.\n\n[Visual Language]\nBlack, cold white, warning red. Glass timelines, metal rails, glowing frame lines, scanning grids. Extreme wide-angle, low angles, top-down shots, Dutch angles, dolly moves, impossible perspective. 3D words COPY, SAME, TREND, OUTSIDE exist as physical architecture, not subtitles.\n\n[Motion Rules]\nUse anticipation → attack → overshoot → settle. Kick drives spatial compression, snare breaks frames, hi-hats animate cursors and fragments, bass bends time. Build with glitch cuts, then explode into an impossible continuous long take during the drop.\n\n[Restrictions]\nNo character changes, extra people, random text, subtitles, logos, outfit changes, bad anatomy, flat camera angles, or meaningless flashes.",
    "promptSourceUrl": "https://x.com/lepadphone/status/2084639330967208177",
    "sourceUrl": "https://x.com/lepadphone/status/2084637679254376793",
    "sourceLabel": "X 原帖 · @lepadphone",
    "author": "padphone",
    "publishedAt": "2026-08-04T13:48:48.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084637679254376793.mp4",
    "posterUrl": "/posters/x/2084637679254376793.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Mind-blown. One shot.\nAn open-weight AI video model just cooked a $35,000-like high-fashion video ad in miniutes. Flawless camera motion, precision audio sync, and 3D VFX, all in one prompt.\n\nIf you are testing MiniMax H3, plz try my prompt⬇️",
    "engagement": {
      "replies": 13,
      "reposts": 9,
      "likes": 325,
      "views": 29761,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084626773015417323",
    "title": "互动游戏界面的连续演示",
    "titleEn": "An Interactive Game Interface in One Continuous Flow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "互动游戏界面的连续演示的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an interactive game interface in one continuous flow on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use Image 1 as the main character role. Use Image 2 as the layout of the game selection interface.\n[0–2 seconds] Display the complete skateboard selection interface from the front. NOVA is in the selected and active state, the four ports at the bottom emit green light. HOTDOG and SHARK are in the unselected state, completely stationary.\n[2–4 seconds] The player presses the right selection key. SHARK ROCKET RIDER slides from the right preview slot into the middle selected module, NOVA JET GLIDER simultaneously slides from the middle into the left preview slot, HOTDOG DASH moves to the right preview slot. As soon as NOVA leaves the middle selected module, the green light immediately turns off and stops. After SHARK enters the middle, orange flames appear at the tail. The skateboard name, attribute bars, and SELECTED label update synchronously.\n[4–6.5 seconds] The player switches the selection again. HOTDOG DASH moves horizontally from the side preview slot to the middle selected module, SHARK ROCKET RIDER simultaneously slides into the other side slot, NOVA moves to the remaining side slot. As soon as SHARK loses the selected state, the orange flame immediately turns off. After HOTDOG becomes the selected item, facing the camera, it barks twice accurately, then turns around in place once.\n[6.5–8 seconds] The bottom CONFIRM button is clicked by the player. The button compresses downwards and bounces back, the right arrow extends forward. HOTDOG DASH becomes the final confirmed selection. The two unselected skateboards on the left and right continue to stay in the fixed preview slots, completely stationary.\n[8–10 seconds] The middle fixed selected module expands towards the depth of the screen. The light-colored display area inside the central card naturally transforms into the World Plaza. The complete world has been fully loaded, with no texture popping, placeholder models, or unfinished buildings. At the same time, the UI displays as the in-game HUD.\n[10–12 seconds] The camera enters the third-person game perspective, with HOTDOG DASH located under the protagonist’s feet. During normal gliding and preparation actions, both feet of the protagonist are standing on HOTDOG DASH. The game HUD appears at this time: mini-map in the top right corner, speed and health status in the bottom left corner, action and combo slots in the bottom right corner, mission route markers and interactive prompts appear ahead.\n[12–15 seconds] The protagonist enters the world riding the running HOTDOG DASH, passing pedestrians, stairs, slopes, railings, and multiple explorable routes. HOTDOG jumps before a low slope, completes a rotation action in the air, and upon landing, the protagonist’s feet are still standing on HOTDOG DASH, the combo count and TRICK prompt briefly increase in the HUD. Subsequently, lean to bypass street obstacles, then perform a small directional jump, and then continue sliding forward.\nSound Design: Use pure instrumental music mixed with electronic arcade sounds, no vocals. UI switching corresponds to crisp key press sounds, card slot sliding sounds, sticker slapping sounds, and attribute bar update sounds. When SHARK is selected, ignition and propulsion sounds appear, and stop immediately when losing selection. When HOTDOG is selected, two barks are played accurately, and light footsteps are added when spinning around. After entering the world, add city ambient sounds, wind sounds, running footsteps, shoe friction sounds when pedaling to accelerate, and jump and landing feedback sounds.",
    "promptSourceUrl": "https://x.com/0xbisc/status/2084626778233143444",
    "sourceUrl": "https://x.com/0xbisc/status/2084626773015417323",
    "sourceLabel": "X 原帖 · @0xbisc",
    "author": "Latte",
    "publishedAt": "2026-08-04T13:05:27.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084626773015417323.mp4",
    "posterUrl": "/posters/x/2084626773015417323.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@Hailuo_AI MiniMax H3 feels like a really good fit for interactive game concepts.\n\nIt handled the UI motion, selection flow, and transitions between game screens in one continuous shot, making it much easier to take an early idea to something that already feels close to a finished sequence.\n\nSharing the prompt below ↓",
    "engagement": {
      "replies": 9,
      "reposts": 3,
      "likes": 35,
      "views": 3128,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084619631742837115",
    "title": "办公室茶水间的伪纪录片",
    "titleEn": "A Mockumentary in the Office Break Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "办公室茶水间的伪纪录片的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a mockumentary in the office break room on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The Office (US) mockumentary style. Dunder Mifflin break room, mid-afternoon, flat fluorescent lighting, vending machine, cheap\nround table, coffee pot. Handheld documentary camera, slightly\nunsteady, deep focus, one continuous setup with a sudden snap zoom\ntoward whoever is reacting.\nMichael Scott and Dwight Schrute stand on opposite sides of the small\ntable, mid-argument. Michael is animated and enthusiastic. Dwight is\nrigid, intense, deadly serious.\nDIALOGUE:\nMICHAEL (excited): \"AI is the future, Dwight.\"\nDWIGHT (sharp): \"It's a threat. I challenged it to an endurance test.\nIt refused.\"\nMICHAEL: \"It wrote my performance review in nine seconds.\"\nDWIGHT: \"I could've done it in seven.\"\nSnap zoom to Michael. He puts down his mug.\nMICHAEL (voice climbing, wounded, slapping the table): \"It listens to\nme! Which is more than anyone here does!\"\nHard silence. Both men frozen.\nCut to Jim Halpert at his desk. He slowly turns his head and looks\ndirectly into the camera lens. Hold for two seconds. End.\nSTYLE: single-camera sitcom shot as if a documentary crew is present,\nnatural performances, no laugh track, muted office color palette,\nslightly soft consumer-camera look, dry deadpan comedic timing, no\nmusic, ambient room tone only.",
    "promptSourceUrl": "https://x.com/Arminn_Ai/status/2084619631742837115",
    "sourceUrl": "https://x.com/Arminn_Ai/status/2084619631742837115",
    "sourceLabel": "X 原帖 · @Arminn_Ai",
    "author": "ΛRMIN",
    "publishedAt": "2026-08-04T12:37:05.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084619631742837115.mp4",
    "posterUrl": "/posters/x/2084619631742837115.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Okay one more, I couldn't stop 😂\n\nMiniMax H3 on @higgsfield_ai\n\n- Prompt:\nThe Office (US) mockumentary style. Dunder Mifflin break room, mid-afternoon, flat fluorescent lighting, vending machine, cheap\nround table, coffee pot. Handheld documentary camera, slightly\nunsteady, deep focus, one continuous setup with a sudden snap zoom\ntoward whoever is reacting.\nMichael Scott and Dwight Schrute stand on opposite sides of the small\ntable, mid-argument. Michael is animated and enthusiastic. Dwight is\nrigid, intense, deadly serious.\nDIALOGUE:\nMICHAEL (excited): \"AI is the future, Dwight.\"\nDWIGHT (sharp): \"It's a threat. I challenged it to an endurance test.\nIt refused.\"\nMICHAEL: \"It wrote my performance review in nine seconds.\"\nDWIGHT: \"I could've done it in seven.\"\nSnap zoom to Michael. He puts down his mug.\nMICHAEL (voice climbing, wounded, slapping the table): \"It listens to\nme! Which is more than anyone here does!\"\nHard silence. Both men frozen.\nCut to Jim Halpert at his desk. He slowly turns his head and looks\ndirectly into the camera lens. Hold for two seconds. End.\nSTYLE: single-camera sitcom shot as if a documentary crew is present,\nnatural performances, no laugh track, muted office color palette,\nslightly soft consumer-camera look, dry deadpan comedic timing, no\nmusic, ambient room tone only.",
    "engagement": {
      "replies": 13,
      "reposts": 4,
      "likes": 295,
      "views": 26930,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084600512180113820",
    "title": "人工智能无法替代你",
    "titleEn": "AI Cannot Replace You, Joey",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人工智能无法替代你的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for ai cannot replace you, joey on X. Native-video output: 14s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "“Joey and Chandler sit side by side on the couch with coffee cups. Joey turns to Chandler with a completely sincere, concerned expression.\n\nDIALOGUE:\n\nJOEY: “Could AI replace us?”\n\nChandler slowly looks up from his coffee. Pauses. Looks Joey dead in the eye.\n\nCHANDLER: “Joey, AI cannot replace you. Nothing can replace you.”\n\nJoey nods slowly. Visibly relieved. Processing.\n\nJOEY: “Because I’m too good?”\n\nChandler stares at him. Long beat. He looks down at his coffee, then back up.\n\nCHANDLER: “…Sure. Let’s go with that.”\n\nSTYLE: Warm 90s sitcom aesthetic, steady handheld feel, natural performances, characters in casual clothing. Joey played completely straight — zero irony. Chandler delivers every line with exhausted deadpan sarcasm. Hold on Chandler’s face after the final line for a full beat before cut.”",
    "promptSourceUrl": "https://x.com/TechieBySA/status/2084600515908796631",
    "sourceUrl": "https://x.com/TechieBySA/status/2084600512180113820",
    "sourceLabel": "X 原帖 · @TechieBySA",
    "author": "TechieSA",
    "publishedAt": "2026-08-04T11:21:06.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084600512180113820.mp4",
    "posterUrl": "/posters/x/2084600512180113820.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI can’t replace you Joey\n\nMinimax H3 prompt👇🏻",
    "engagement": {
      "replies": 66,
      "reposts": 53,
      "likes": 755,
      "views": 69960,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply.",
    "featured": {
      "at": "2026-09-05T15:50:15.691Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2084586713205317678",
    "title": "圆形、方形、三角形与星形",
    "titleEn": "Circle, Square, Triangle, and Star",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "圆形、方形、三角形与星形的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for circle, square, triangle, and star on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second animated educational video that teaches young children the shapes Circle, Square, Triangle, and Star.\n\nLearning pattern for every shape:\nSHAPE → NAME SOUND → OBJECT → PLAYFUL ACTION → OBJECT NAME\nTarget audience: children ages 3 to 6\n\nVisual style: Same premium-cute aesthetic pastel colors, rounded 3D forms, soft lighting, minimalist off-white background, unique glow color per shape, crisp bold typography, polished smooth morphing transitions.\n\n0:00–0:01 Intro: Mascot bounces in, shapes float briefly, text \"Let's learn shapes!\", tap reveals first shape with a ripple.\n\n0:01–0:04 | Circle is for Orange: A perfect circle appears. Narrator: \"Circle. Round and round. Circle is for Orange.\" Circle gains texture and a tiny stem, becomes a smiling orange, bounces once. Word \"ORANGE,\" highlight shape outline in orange.\n\n0:04–0:07 | Square is for Gift Box: The orange rolls and reshapes into a square. Narrator: \"Square. Four equal sides. Square is for Gift Box.\" Square gains a ribbon and bow, wiggles playfully. Word \"GIFT BOX,\" highlight in purple.\n\n0:07–0:10 | Triangle is for Mountain: The ribbon folds into a triangle. Narrator: \"Triangle. Three sides. Triangle is for Mountain.\" Triangle grows a snowy peak and a small smiling sun beside it. Word \"MOUNTAIN,\" highlight in brown/teal.\n\n0:10–0:13 | Star is for Sky: The mountain peak stretches into a star. Narrator: \"Star. Star is for Sky.\" Star twinkles softly with tiny sparkles floating around it, gentle glow pulse. Word \"STAR,\" highlight in golden yellow.\n\n0:13–0:15 Recap: Orange, gift box, mountain, star line up in four tiles with shape outlines above. Mascot points to each in sequence.\n\nNarrator: \"Circle, Square, Triangle, Star. Great job!\" Sparkle + warm chime.\n\nRequirements: Each shape fully visible before morph, uppercase-clarity level precision on shape edges, no warped or duplicated forms, gentle motion only, synced sound cues, calm and beautifully polished finish.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2084586713205317678",
    "sourceUrl": "https://x.com/Strength04_X/status/2084586713205317678",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-04T10:26:16.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084586713205317678.mp4",
    "posterUrl": "/posters/x/2084586713205317678.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Circle, Square, Triangle, and Star\n\nA simple learning animation made with MiniMax H3 on @Hailuo_AI \n\nPrompt - Create a 15-second animated educational video that teaches young children the shapes Circle, Square, Triangle, and Star.\n\nLearning pattern for every shape:\nSHAPE → NAME SOUND → OBJECT → PLAYFUL ACTION → OBJECT NAME\nTarget audience: children ages 3 to 6\n\nVisual style: Same premium-cute aesthetic pastel colors, rounded 3D forms, soft lighting, minimalist off-white background, unique glow color per shape, crisp bold typography, polished smooth morphing transitions.\n\n0:00–0:01 Intro: Mascot bounces in, shapes float briefly, text \"Let's learn shapes!\", tap reveals first shape with a ripple.\n\n0:01–0:04 | Circle is for Orange: A perfect circle appears. Narrator: \"Circle. Round and round. Circle is for Orange.\" Circle gains texture and a tiny stem, becomes a smiling orange, bounces once. Word \"ORANGE,\" highlight shape outline in orange.\n\n0:04–0:07 | Square is for Gift Box: The orange rolls and reshapes into a square. Narrator: \"Square. Four equal sides. Square is for Gift Box.\" Square gains a ribbon and bow, wiggles playfully. Word \"GIFT BOX,\" highlight in purple.\n\n0:07–0:10 | Triangle is for Mountain: The ribbon folds into a triangle. Narrator: \"Triangle. Three sides. Triangle is for Mountain.\" Triangle grows a snowy peak and a small smiling sun beside it. Word \"MOUNTAIN,\" highlight in brown/teal.\n\n0:10–0:13 | Star is for Sky: The mountain peak stretches into a star. Narrator: \"Star. Star is for Sky.\" Star twinkles softly with tiny sparkles floating around it, gentle glow pulse. Word \"STAR,\" highlight in golden yellow.\n\n0:13–0:15 Recap: Orange, gift box, mountain, star line up in four tiles with shape outlines above. Mascot points to each in sequence.\n\nNarrator: \"Circle, Square, Triangle, Star. Great job!\" Sparkle + warm chime.\n\nRequirements: Each shape fully visible before morph, uppercase-clarity level precision on shape edges, no warped or duplicated forms, gentle motion only, synced sound cues, calm and beautifully polished finish.",
    "engagement": {
      "replies": 33,
      "reposts": 3,
      "likes": 61,
      "views": 5545,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084580423649423684",
    "title": "第一次的运镜惊喜",
    "titleEn": "A First Attempt with Unexpected Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "第一次的运镜惊喜的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 480×832。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a first attempt with unexpected motion on X. Native-video output: 15s · 480×832. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "## Concept\nAQUARELLE No.7 — an avant-garde haute couture film where watercolor becomes a living medium.\n\n15-second cinematic sequence.\n\nThe rhythm controls the visual world:\n0–4s: restrained silence and negative space\n4–8s: gradual density buildup\n8s: drop moment, expanding into fluid long-form motion\n12–15s: transition into a final fashion poster composition\n\n## Character Identity Lock\n\nMaintain the exact same female character throughout the entire video.\n\nIdentity:\n- Young woman\n- Long black hair\n- Calm and refined facial features\n- White high-neck pigment dress\n- Black wide belt\n- Pigment heels\n\nStrict consistency:\n- Same face\n- Same hairstyle\n- Same age\n- Same body proportions\n- Same garment structure\n\nAny new colors must only appear through watercolor gradually absorbing into the fabric.\n\nCore concept:\nHer fingertips can extract transparent watercolor ribbons from the air.\n\nThese watercolor ribbons can:\n- Pull the camera through space\n- Transform the environment\n- Shape physical typography\n- Interact with depth and materials\n\nThe world follows real cinematic physics:\nwater, paper, fabric, light, shadows, and depth of field must feel physically believable.\n\n## Camera Direction\n\nStrict one-take shot.\nNo cuts.\nNo teleportation.\nNo hidden transitions.\n\nCamera journey:\n\nMacro shot of a floating water droplet →\nwatercolor ribbon emerges →\ncamera pulls back to reveal the woman →\ncamera circles around her →\nenters a paper art gallery →\npasses through dimensional typography →\nrises into a final overhead fashion poster.\n\n## Typography\n\nOnly allow these words:\n\nLET SILENCE BLOOM\n\nAQUARELLE No.7\n\nWEAR THE UNSEEN\n\nTypography is not a flat overlay.\n\nLetters must have:\n- Physical depth\n- Wet watercolor reflections\n- Paper fiber edges\n- Shadows\n- Occlusion\n- Material interaction\n\n## Visual Style\n\nAvant-garde fashion editorial.\n\nInspired by:\n- Museum catalog composition\n- Handmade ivory paper\n- Sculptural negative space\n- Elegant Didone serif typography\n- Translucent watercolor calligraphy\n\nColor palette:\n- Pale cyan\n- Crimson lake\n- Smoky violet\n- Ink black\n\n## Motion Language\n\nEvery movement follows the music:\n\nKick:\nCamera movement and paper folding.\n\nWooden snare:\nPaper structures physically fold and transform.\n\nSub-bass:\nChanges the perception of spatial scale.\n\n##  Storyboard\n\n30 beats, 0.5 seconds each.\n\n01 0.0–0.5\nMacro shot: A transparent water droplet floats in the air, reflecting a blurred silhouette of a black-haired woman.\n\n02 0.5–1.0\nThe droplet stretches with the breath-like vocal, becoming a pale cyan watercolor thread.\n\n03 1.0–1.5\nCamera travels backward along the thread as paper fibers slowly emerge into focus.\n\n04 1.5–2.0\nThe thread wraps around the lens. Focus shifts to her raised fingertip.\n\n05 2.0–3.0\nCamera continues pulling back, revealing her face and white high-neck dress.\n\n06 3.0–4.0\nShe moves her wrist. The watercolor thread guides a smooth camera arc.\n\n07 4.0–8.0\nAdditional watercolor colors emerge from her movement.\nPaper folds, typography begins forming, and the phrase \"LET SILENCE BLOOM\" appears as a physical object.\n\n08 8.0–12.0\nThe camera passes through a transparent paper flower structure.\nThe environment expands into an endless ivory paper gallery.\n\nHer dress absorbs watercolor naturally.\nThe ribbons create sculptural forms around her body.\n\n09 12.0–15.0\nCamera cranes upward.\n\nThe composition transforms into a luxury fashion advertisement poster.\n\nTypography appears:\n\nAQUARELLE No.7\n\nWEAR THE UNSEEN\n\nFinal frame:\nA museum-level fashion editorial poster.\nThe woman remains centered, calm, and elegant.\nA final watercolor droplet remains suspended in the air.\n\n## Negative Prompt\n\nNo:\n- Cuts\n- Scene changes\n- Identity change\n- Face swap\n- Extra limbs\n- Deformed hands\n- Random costume changes\n- Explosive paint effects without physical cause\n- Incorrect typography\n- Chinese characters\n- Extra subtitles\n- Extra logos\n- Watermarks",
    "promptSourceUrl": "https://x.com/medfac_opc/status/2084580459938529601",
    "sourceUrl": "https://x.com/medfac_opc/status/2084580423649423684",
    "sourceLabel": "X 原帖 · @medfac_opc",
    "author": "Mediafac",
    "publishedAt": "2026-08-04T10:01:17.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084580423649423684.mp4",
    "posterUrl": "/posters/x/2084580423649423684.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "第一次用MiniMax H3，提示词见评论区，参考了X上的大神",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 197,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084559907395940599",
    "title": "导演拥抱人工智能的瞬间",
    "titleEn": "The Moment a Director Embraces AI",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "导演拥抱人工智能的瞬间的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the moment a director embraces ai on X. Native-video output: 10s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "摩天大楼大小的石龙冲破云层并展开巨大的翅膀几乎把天空填满。两个人站在悬崖上看着它升起。环绕立体鳞片的体积雾。光穿过它头后的云层。",
    "promptSourceUrl": "https://x.com/PixelAigc/status/2084559907395940599",
    "sourceUrl": "https://x.com/PixelAigc/status/2084559907395940599",
    "sourceLabel": "X 原帖 · @PixelAigc",
    "author": "PixelAIGC",
    "publishedAt": "2026-08-04T08:39:45.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084559907395940599.mp4",
    "posterUrl": "/posters/x/2084559907395940599.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "非常震惊！我居然现在才发现！AI已经发展到一站式全部搞定所有功能的超强平台！如果你是导演，应该主动拥抱AI，而不是视而不见。\n\n我在@edimakortaiwan 上用 MiniMax H3 把这一段非常简单的提示词重新生成一条视频，出来的效果一样震撼！\n\n这是一个集图片，音乐，视频等创作于一体的大型平台，还有很多辅助功能，我还仅仅测试了一条H3视频，没有抽卡，就实现了与seedance 2.0一样水平的视频效果。\n\n接下来我会继续测试这个网站的AIMV功能，一句提示词就能实现3分钟歌曲MV制作，这也太强大了。\n视频提示词如下，一键复制可用：\n摩天大楼大小的石龙冲破云层并展开巨大的翅膀几乎把天空填满。两个人站在悬崖上看着它升起。环绕立体鳞片的体积雾。光穿过它头后的云层。",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 11,
      "views": 2535,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084552480554446929",
    "title": "从一数到十的儿童动画",
    "titleEn": "A Counting Adventure from One to Ten",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "从一数到十的儿童动画的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a counting adventure from one to ten on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second animated educational video that teaches young children the numbers 1, 2, 3, and 4.\nLearning pattern for every number:\nNUMBER → COUNT SOUND → OBJECTS → PLAYFUL ACTION → NUMBER NAME\nTarget audience: children ages 3 to 6.\nVisual style: Adorable rounded 3D characters, soft pastel colors, gentle expressions, premium minimalist tech aesthetic clean white space, soft studio lighting, smooth gradients, rounded geometry, crisp typography, polished transitions. Clean off-white background with a different soft color glow behind each number.\n0:00–0:01 Intro: Star mascot bounces in, numbers float around it, text \"Let's count!\", tap creates ripple revealing \"1\".\n0:01–0:04 | 1 is for One Sun: Large \"1\" shown, narrator: \"One. Just one. One Sun.\" The 1 transforms into a smiling sun with soft rays. One gentle pulse of light. Word \"ONE SUN\" appears, highlight \"1\" in yellow.\n0:04–0:07 | 2 is for Two Birds: The sun's rays curve into a \"2.\" Narrator: \"Two. Two Birds.\" The 2 splits gently into two round bluebirds that flutter and land side by side. Word \"TWO BIRDS,\" highlight \"2\" in blue.\n0:07–0:10 | 3 is for Three Flowers: The birds' wings curl into a \"3.\" Narrator: \"Three. Three Flowers.\" The 3 blooms into three pink flowers swaying gently. Word \"THREE FLOWERS,\" highlight \"3\" in pink.\n0:10–0:13 | 4 is for Four Fish: The flower stems bend into a \"4.\" Narrator: \"Four. Four Fish.\" The 4 splits into four little green fish swimming in a soft circle. Water ripple effect. Word \"FOUR FISH,\" highlight \"4\" in green.\n0:13–0:15 Recap: Sun, birds, flowers, fish appear as four tiles with \"1 2 3 4\" above. Mascot points to each; Narrator: \"1, 2, 3, 4. Great counting!\" Sparkle + chime.\nRequirements: Keep each number visible before morphing, smooth shape transitions, no clutter, no scary faces, gentle squash-and-stretch, synced sound effects, calm and polished finish.\n\nPrompt idea - @umesh_ai #MiniMaxH3",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2084552480554446929",
    "sourceUrl": "https://x.com/Strength04_X/status/2084552480554446929",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-08-04T08:10:15.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084552480554446929.mp4",
    "posterUrl": "/posters/x/2084552480554446929.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I created this kids counting video with MiniMax H3 on @Hailuo_AI \n\nCute characters, smooth transitions, and a fun way to learn numbers 🔢 \n\nPrompt - Create a 15-second animated educational video that teaches young children the numbers 1, 2, 3, and 4.\nLearning pattern for every number:\nNUMBER → COUNT SOUND → OBJECTS → PLAYFUL ACTION → NUMBER NAME\nTarget audience: children ages 3 to 6.\nVisual style: Adorable rounded 3D characters, soft pastel colors, gentle expressions, premium minimalist tech aesthetic clean white space, soft studio lighting, smooth gradients, rounded geometry, crisp typography, polished transitions. Clean off-white background with a different soft color glow behind each number.\n0:00–0:01 Intro: Star mascot bounces in, numbers float around it, text \"Let's count!\", tap creates ripple revealing \"1\".\n0:01–0:04 | 1 is for One Sun: Large \"1\" shown, narrator: \"One. Just one. One Sun.\" The 1 transforms into a smiling sun with soft rays. One gentle pulse of light. Word \"ONE SUN\" appears, highlight \"1\" in yellow.\n0:04–0:07 | 2 is for Two Birds: The sun's rays curve into a \"2.\" Narrator: \"Two. Two Birds.\" The 2 splits gently into two round bluebirds that flutter and land side by side. Word \"TWO BIRDS,\" highlight \"2\" in blue.\n0:07–0:10 | 3 is for Three Flowers: The birds' wings curl into a \"3.\" Narrator: \"Three. Three Flowers.\" The 3 blooms into three pink flowers swaying gently. Word \"THREE FLOWERS,\" highlight \"3\" in pink.\n0:10–0:13 | 4 is for Four Fish: The flower stems bend into a \"4.\" Narrator: \"Four. Four Fish.\" The 4 splits into four little green fish swimming in a soft circle. Water ripple effect. Word \"FOUR FISH,\" highlight \"4\" in green.\n0:13–0:15 Recap: Sun, birds, flowers, fish appear as four tiles with \"1 2 3 4\" above. Mascot points to each; Narrator: \"1, 2, 3, 4. Great counting!\" Sparkle + chime.\nRequirements: Keep each number visible before morphing, smooth shape transitions, no clutter, no scary faces, gentle squash-and-stretch, synced sound effects, calm and polished finish.\n\nPrompt idea - @umesh_ai #MiniMaxH3",
    "engagement": {
      "replies": 27,
      "reposts": 4,
      "likes": 148,
      "views": 4442,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084519725405483220",
    "title": "海报在镜头中苏醒",
    "titleEn": "A Poster Comes Alive",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "海报在镜头中苏醒的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1440×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a poster comes alive on X. Native-video output: 15s · 1440×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Animate the source artwork as a **10-second energetic chibi anime poster assembly**, preserving the deep indigo-purple background, neon pink, cyan and yellow palette, glossy anime illustration, characters, typography, decorative border and exact original layout. Begin on the empty purple background, then draw the outer border and corner markers with quick glowing strokes; introduce the small labels, Japanese text, numbering and interface graphics through playful pop-ins, slides and type-on animation, each paired with light arcade clicks, bubbly typing sounds and soft digital chimes.\n\nBuild the central world in layers: floating islands rise into position, clouds expand outward, the moon fades in, towers and airships drift into place, and sparkles scatter across the scene. Assemble the chibi robot and flying cat with quick bouncy entrances, then launch the main hero from deep inside the poster toward the camera with strong 3D parallax, allowing his shoe, hand and hair to briefly extend beyond the inner frame before settling into the original composition.\n\nReveal the Japanese title with a bold elastic impact, followed by the English labels, release date, IMAX line and footer icons appearing sequentially. Add subtle looping life after assembly: blinking eyes, gently moving hair and clothing, small robot antenna motion, the cat’s tail bouncing, airships floating, clouds drifting and stars twinkling. Finish with a soft camera push-in, one bright starburst behind the title and a clean hold on the completed poster, keeping every element sharp and readable. No redesign, no distorted faces, no changing text, no full scene transformation, no aggressive camera rotation.\n2/2",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2084519753503162467",
    "sourceUrl": "https://x.com/LudovicCreator/status/2084519725405483220",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-04T06:00:05.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084519725405483220.mp4",
    "posterUrl": "/posters/x/2084519725405483220.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An other example of poster animation and honestly I can't stop playing with it\n\nFind the image used and prompt in first comment.\n\nMade in @Hailuo_AI #MiniMaxH3\n1/2",
    "engagement": {
      "replies": 15,
      "reposts": 6,
      "likes": 45,
      "views": 2233,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084481873158328598",
    "title": "露娜的未来直播间",
    "titleEn": "Luna Enters a Futuristic Livestream",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "露娜的未来直播间的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for luna enters a futuristic livestream on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use @Image4as the opening card only: circular Luna avatar, black background, cream “LUNALIVE”, rose “STREAM STARTING”, warm circles, tiny mint accent. Static, chime.\n\nUse @Image1 for Luna’s identity only: same face, long center-parted black hair, blue-gray eyes, pink anime hoodie, white headphones around neck, delicate necklace, pale nails. Do not copy the drink, pose, or background from @Image1.\n\nUse @Image2 only for Twitch-like platform chrome: dark top bar with “LUNALIVE”, red “LIVE” badge, “2.4K viewers”, right “STREAM CHAT” rail, bottom title area, rounded “FOLLOW” and pink “SUBSCRIBE” buttons. Do not copy Luna, pose, drink, or room from @Image2\n\nUse @Image3 only for the cozy room behind Luna inside the video area: desk, monitor, white PC, plush shelves, curtain fairy lights, soft pink/purple light. No empty-room showcase shot.\n\nUse @Audio1as Luna’s actual vocal performance and behavior reference. @Audio1has a silent lead-in: 0.0–2.4s must be treated as no speech. Preserve her voice identity, cadence, tone, breaths, pauses, emphasis, warmth, and streamer mannerisms. Do not replace the voice, do not generate a different influencer voice, and do not add extra spoken lines beyond @Audio1 Lip sync, mouth shapes, jaw movement, smiles, glances, nods, and small hand movements must follow the audio waveform after 2.4s.\n\nCreate a 15-second 16:9 Twitch-like stream opening. Important performance direction: Luna is reading chat, not delivering a camera monologue. Place Luna slightly left of center in the video area with the right “STREAM CHAT” rail clearly visible. Whenever she speaks, her eyes angle screen-right toward the chat rail as if she is reading the messages out loud. She returns to camera only for brief reactions. Add constant small movement: eye darts to chat, eyebrow lifts, tiny nods, head tilts, shoulders shifting, one subtle hand gesture near the desk. No stiff talking-head pose. One cursor only, no cursor trail, no duplicate panels, no duplicate buttons.\n\nRender only large UI text cleanly. Chat feels alive with typing dots and soft short blurred lines, but only these chat lines are readable: “hi Luna”, “welcome back”, “so cozy”, “gugugaga?”, “RAID INCOMING!”. Do not invent usernames. Chat pops stay quieter than Luna’s voice.\n\n[0–2 seconds] Open on @Image4 “LUNALIVE / STREAM STARTING”. Absolute no-speech zone: @Audio1is silent here, Luna is not shown, no mouth movement, no voice on the card. One soft chime only. No movement.\n\n[2–7.4 seconds] Hard cut to Luna live at 2.0s, slightly left of center in the cozy room from @Image3 with @Image2 chrome active: “LUNALIVE”, red “LIVE”, “2.4K viewers”, right “STREAM CHAT” rail, bottom title “COZY NEON” and “Just Chatting”. She settles for a beat, eyes already moving toward the chat rail. At about 2.4s when speech begins in @Audio1, match lip sync exactly while she reads toward the chat rail, not into camera. Chat shows typing dots and soft blurred lines.\n\n[7.4–7.9 seconds] First audio pause = chat beat. Typing dots, then readable messages pop in: “hi Luna”, “welcome back”. Luna’s eyes track the new messages on the right rail; small nod and smile follow the audio pause.\n\n[7.9–12.5 seconds] Continue matching @Audio1Keep her gaze mostly on the chat rail while speaking, like she is reading and reacting. Add one more readable message: “so cozy”. During any softer phrase, she leans slightly forward as if reading; during brighter phrases, eyebrows lift and shoulders react. No frozen face.\n\n[12.5–13.9 seconds] Bigger audio pause = bigger chat beat. “gugugaga?” appears, then “RAID INCOMING!”, and a clean “NEW FOLLOWER” banner slides in with a gentle pop. Luna reads the raid message from the chat rail, then reacts brighter as the audio resumes.\n\n[13.9–14.8 seconds] Finish @Audio1 with accurate lip sync. If the audio winds down, Luna stops talking, gives a small wave toward chat, and settles into a warm listening pose. End on the stable live frame: Luna slightly left of center, eyes toward the right chat rail, red “LIVE”, “2.4K viewers”, no end card.\n\nAudio mix: 0–2s card is silent except one soft chime. @Audio1voice begins only after the cut to Luna and remains primary. Tiny chat pops under the voice. Warm low room tone. No crowd noise, no music lyrics, no rain, no traffic.",
    "promptSourceUrl": "https://x.com/GlitterPixely/status/2084481875490312226",
    "sourceUrl": "https://x.com/GlitterPixely/status/2084481873158328598",
    "sourceLabel": "X 原帖 · @GlitterPixely",
    "author": "Glitter Gal",
    "publishedAt": "2026-08-04T03:29:41.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084481873158328598.mp4",
    "posterUrl": "/posters/x/2084481873158328598.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Luna is back! now that I can render text beautifully with H3 I had to do a mini test to see how a stream would go. It gives me ideas to run continuous video overlay with text for twitch. \n\nPrompt in replies ✨\n#MiniMaxH3\n@Hailuo_AI",
    "engagement": {
      "replies": 23,
      "reposts": 3,
      "likes": 293,
      "views": 24514,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084472700030816740",
    "title": "人类进化的动态长卷",
    "titleEn": "A Scrolling Timeline of Human Evolution",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人类进化的动态长卷的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1056×608。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a scrolling timeline of human evolution on X. Native-video output: 15s · 1056×608. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "️\n\n【核心创意】\n\n15秒，16:9横版人类进化史科普短片。使用现代扁平矢量插画与信息图动画风格，通过一条从左向右展开的连续时间长卷，表现人类祖先从共同灵长类祖先分化、直立行走、使用工具与火、迁徙合作，最终发展为现代智人的过程。\n\n科学表达采用分支式演化树，不表现现代猿猴直接变成人。整体为纯2D矢量视觉：简洁几何图形、统一深色描边、平面色块、有限色板、轮廓清晰、平滑形状变形。色彩从史前赭石、沙黄逐渐过渡到现代深蓝、青绿色。\n\n全片采用连续向右移动的横向镜头，不硬切。每个时代通过轮廓、火焰、石器、洞穴线条和迁徙路线自然变形进入下一阶段。\n\n【画面过程说明】\n\n0–3秒：\n米白色背景上，一棵由简洁线条组成的演化树从画面中央生长。树枝向不同方向分叉，其中一条支系延伸为生活在非洲森林边缘的早期人族剪影。\n\n早期人族从弯腰行走逐渐变成较稳定的双足站立姿态。身体变化通过平滑矢量轮廓变形完成，不出现写实肌肉和恐怖形变。\n\n镜头缓慢向右移动。树枝线条自然延伸成为草原地平线。\n\n3–6秒：\n进入非洲草原场景。早期人属以双足行走，手中拿起简单石块，将石块敲击成锋利石器。敲击产生的三角形碎片在空中旋转，变成橙红色几何火花。\n\n火花聚集成一团扁平矢量火焰。人物围绕火焰取暖、分享食物，表现工具、火和群体合作。\n\n人物外形保持简洁剪影，不出现夸张野人形象。\n\n6–9秒：\n火焰向右拉伸，变成地图上的橙色迁徙路线。镜头继续向右平移，背景转化为简化的大陆轮廓。\n\n多个人类支系以不同颜色的矢量剪影沿分支路线迁徙。部分人物穿着简单兽皮、携带工具，在寒冷环境中合作前行。\n\n迁徙路线必须表现为多个分支，不是一条单向直线。\n\n9–12秒：\n地图轮廓变形成洞穴墙面。尼安德特人与早期智人的不同剪影分别出现在两条相邻支系中，表现不同人类支系曾经同时存在。\n\n其中一组人物照顾同伴，另一组在洞穴墙上绘制动物图案。洞穴图案由简单线条组成，随后逐渐变成手掌、语言符号和抽象思维图形。\n\n不要表现一方战胜或消灭另一方。\n\n12–15秒：\n洞穴线条向外扩散，平滑变形成DNA双螺旋、城市轮廓、科学工具和地球图形。\n\n画面中央出现一组不同肤色、年龄和性别的现代人类矢量剪影，并肩面向前方。人物身后的演化树仍保持清晰可见，强调现代智人只是演化树延续至今的一条支系。\n\n镜头轻微拉远，演化树、历史阶段与现代人类共同组成完整长卷，在稳定构图中结束。\n\n声音：\n使用克制、具有时间推进感的氛围音乐。开头加入轻微风声，石器阶段加入石块敲击声，火焰阶段加入柔和燃烧声，迁徙阶段加入脚步声，结尾音乐逐渐开阔。\n\n使用沉稳、清晰的中文画外音，完整说出：\n\n“数百万年间，人类的祖先在分支中前行。直立、工具、火与合作，最终塑造了今天的我们。”\n\n不想要：\n不要出现现代猴子或现代黑猩猩直接变成人的画面。\n不要使用传统错误的单线进化阶梯构图。\n不要出现血腥、战争、捕杀或野蛮化描写。\n不要出现写实皮肤、照片质感、3D建模、体积光或复杂渐变。\n不要出现身体畸形、人物闪烁、额外肢体或轮廓突然变化。\n不要生成标题、字幕、年代数字、Logo、水印或乱码文字。\n不要黑屏、硬切或产生明显拼接感。",
    "promptSourceUrl": "https://x.com/xianshi666123/status/2084472700030816740",
    "sourceUrl": "https://x.com/xianshi666123/status/2084472700030816740",
    "sourceLabel": "X 原帖 · @xianshi666123",
    "author": "闲适",
    "publishedAt": "2026-08-04T02:53:13.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084472700030816740.mp4",
    "posterUrl": "/posters/x/2084472700030816740.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 \n生成教育类的影片效果，seedance2 真的该降价了\n\nprompt⬇️\n\n【核心创意】\n\n15秒，16:9横版人类进化史科普短片。使用现代扁平矢量插画与信息图动画风格，通过一条从左向右展开的连续时间长卷，表现人类祖先从共同灵长类祖先分化、直立行走、使用工具与火、迁徙合作，最终发展为现代智人的过程。\n\n科学表达采用分支式演化树，不表现现代猿猴直接变成人。整体为纯2D矢量视觉：简洁几何图形、统一深色描边、平面色块、有限色板、轮廓清晰、平滑形状变形。色彩从史前赭石、沙黄逐渐过渡到现代深蓝、青绿色。\n\n全片采用连续向右移动的横向镜头，不硬切。每个时代通过轮廓、火焰、石器、洞穴线条和迁徙路线自然变形进入下一阶段。\n\n【画面过程说明】\n\n0–3秒：\n米白色背景上，一棵由简洁线条组成的演化树从画面中央生长。树枝向不同方向分叉，其中一条支系延伸为生活在非洲森林边缘的早期人族剪影。\n\n早期人族从弯腰行走逐渐变成较稳定的双足站立姿态。身体变化通过平滑矢量轮廓变形完成，不出现写实肌肉和恐怖形变。\n\n镜头缓慢向右移动。树枝线条自然延伸成为草原地平线。\n\n3–6秒：\n进入非洲草原场景。早期人属以双足行走，手中拿起简单石块，将石块敲击成锋利石器。敲击产生的三角形碎片在空中旋转，变成橙红色几何火花。\n\n火花聚集成一团扁平矢量火焰。人物围绕火焰取暖、分享食物，表现工具、火和群体合作。\n\n人物外形保持简洁剪影，不出现夸张野人形象。\n\n6–9秒：\n火焰向右拉伸，变成地图上的橙色迁徙路线。镜头继续向右平移，背景转化为简化的大陆轮廓。\n\n多个人类支系以不同颜色的矢量剪影沿分支路线迁徙。部分人物穿着简单兽皮、携带工具，在寒冷环境中合作前行。\n\n迁徙路线必须表现为多个分支，不是一条单向直线。\n\n9–12秒：\n地图轮廓变形成洞穴墙面。尼安德特人与早期智人的不同剪影分别出现在两条相邻支系中，表现不同人类支系曾经同时存在。\n\n其中一组人物照顾同伴，另一组在洞穴墙上绘制动物图案。洞穴图案由简单线条组成，随后逐渐变成手掌、语言符号和抽象思维图形。\n\n不要表现一方战胜或消灭另一方。\n\n12–15秒：\n洞穴线条向外扩散，平滑变形成DNA双螺旋、城市轮廓、科学工具和地球图形。\n\n画面中央出现一组不同肤色、年龄和性别的现代人类矢量剪影，并肩面向前方。人物身后的演化树仍保持清晰可见，强调现代智人只是演化树延续至今的一条支系。\n\n镜头轻微拉远，演化树、历史阶段与现代人类共同组成完整长卷，在稳定构图中结束。\n\n声音：\n使用克制、具有时间推进感的氛围音乐。开头加入轻微风声，石器阶段加入石块敲击声，火焰阶段加入柔和燃烧声，迁徙阶段加入脚步声，结尾音乐逐渐开阔。\n\n使用沉稳、清晰的中文画外音，完整说出：\n\n“数百万年间，人类的祖先在分支中前行。直立、工具、火与合作，最终塑造了今天的我们。”\n\n不想要：\n不要出现现代猴子或现代黑猩猩直接变成人的画面。\n不要使用传统错误的单线进化阶梯构图。\n不要出现血腥、战争、捕杀或野蛮化描写。\n不要出现写实皮肤、照片质感、3D建模、体积光或复杂渐变。\n不要出现身体畸形、人物闪烁、额外肢体或轮廓突然变化。\n不要生成标题、字幕、年代数字、Logo、水印或乱码文字。\n不要黑屏、硬切或产生明显拼接感。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 2077,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084441225667735905",
    "title": "表情与镜头之间的张力",
    "titleEn": "Expressive Acting Under a Moving Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "表情与镜头之间的张力的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for expressive acting under a moving camera on X. Native-video output: 10s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": ")\n\n[FORMAT]\nCreate a 10-second, 16:9 photoreal cinematic wide-shot flyby of a stealth aircraft activating optical camouflage at extreme speed.\n\n[IDENTITY]\nPreserve a wide tailless delta wing, sharp faceted edges, flat wedge-shaped nose, recessed dark cockpit, matte charcoal surface, twin buried exhausts, no insignia, and rigid non-morphing geometry.\n\n[BEATS]\n[0–2 seconds] A vast open sky above distant clouds. The stealth aircraft suddenly enters from the far left and races laterally across the wide frame at extreme speed.\n\n[2–4 seconds] The camera performs a fast, controlled pan to track the aircraft. Strong background parallax and slight motion blur emphasize its velocity while its shape remains readable.\n\n[4–6 seconds] Optical camouflage sweeps from the nose across both wings. The solid aircraft rapidly vanishes without slowing down or changing direction.\n\n[6–8 seconds] The camera continues tracking the same trajectory. The invisible delta-wing shape remains clearly perceptible because the sky and clouds inside its exact volume ripple and bend like concentrated heat haze.\n\n[8–10 seconds] The refractive aircraft streaks toward the distant horizon, leaving a narrow turbulent distortion wake before exiting the frame.\n\n[CAMERA]\nOne continuous very wide lateral tracking shot. The aircraft stays relatively small within the expansive frame. Fast pan tracking, broad environmental scale, strong lateral movement, subtle handheld vibration, and realistic motion blur. No close-up and no chase-camera angle.\n\n[LIGHT]\nBright high-altitude daylight, cool blue sky, white cloud field, natural atmospheric haze, restrained contrast, and realistic cinematic texture.\n\n[EDIT]\nOne uninterrupted shot. No cuts, transitions, slow motion, or camera repositioning.\n\n[AUDIO]\nAudio: violent high-altitude wind rush, rapidly approaching jet roar, a powerful Doppler flyby, and a brief vacuum-like sound as the camouflage activates.\nBGM: an original 10-second cue, 80% tension and 20% awe. Low sub-bass pulse and metallic air textures, building sharply during the flyby and fading into a low tone over the final 2 seconds. Do not imitate an existing melody.\n\n[NEGATIVE]\nDo not show an empty sky after cloaking. Maintain a coherent, fast-moving delta-wing refraction that bends the background inside the aircraft’s exact volume.\nNo glowing outline, glass aircraft, force-field bubble, pixel dissolve, smoke camouflage, teleportation, morphing, or reappearance.\nNo extra aircraft, weapons, explosions, contrails, subtitles, on-screen text, watermarks, platform logos, or stickers.\nDo not introduce Chinese text, garbled characters, or misspellings.\n\n#kreaai #minimaxh3 #kreacpp",
    "promptSourceUrl": "https://x.com/opener_ai/status/2084441225667735905",
    "sourceUrl": "https://x.com/opener_ai/status/2084441225667735905",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-04T00:48:09.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084441225667735905.mp4",
    "posterUrl": "/posters/x/2084441225667735905.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "generated this using the Minimax H3 model on @krea_ai  It's so fun to see video models with such drastically enhanced expressiveness these days. Please refer to the prompt :)\n\n[FORMAT]\nCreate a 10-second, 16:9 photoreal cinematic wide-shot flyby of a stealth aircraft activating optical camouflage at extreme speed.\n\n[IDENTITY]\nPreserve a wide tailless delta wing, sharp faceted edges, flat wedge-shaped nose, recessed dark cockpit, matte charcoal surface, twin buried exhausts, no insignia, and rigid non-morphing geometry.\n\n[BEATS]\n[0–2 seconds] A vast open sky above distant clouds. The stealth aircraft suddenly enters from the far left and races laterally across the wide frame at extreme speed.\n\n[2–4 seconds] The camera performs a fast, controlled pan to track the aircraft. Strong background parallax and slight motion blur emphasize its velocity while its shape remains readable.\n\n[4–6 seconds] Optical camouflage sweeps from the nose across both wings. The solid aircraft rapidly vanishes without slowing down or changing direction.\n\n[6–8 seconds] The camera continues tracking the same trajectory. The invisible delta-wing shape remains clearly perceptible because the sky and clouds inside its exact volume ripple and bend like concentrated heat haze.\n\n[8–10 seconds] The refractive aircraft streaks toward the distant horizon, leaving a narrow turbulent distortion wake before exiting the frame.\n\n[CAMERA]\nOne continuous very wide lateral tracking shot. The aircraft stays relatively small within the expansive frame. Fast pan tracking, broad environmental scale, strong lateral movement, subtle handheld vibration, and realistic motion blur. No close-up and no chase-camera angle.\n\n[LIGHT]\nBright high-altitude daylight, cool blue sky, white cloud field, natural atmospheric haze, restrained contrast, and realistic cinematic texture.\n\n[EDIT]\nOne uninterrupted shot. No cuts, transitions, slow motion, or camera repositioning.\n\n[AUDIO]\nAudio: violent high-altitude wind rush, rapidly approaching jet roar, a powerful Doppler flyby, and a brief vacuum-like sound as the camouflage activates.\nBGM: an original 10-second cue, 80% tension and 20% awe. Low sub-bass pulse and metallic air textures, building sharply during the flyby and fading into a low tone over the final 2 seconds. Do not imitate an existing melody.\n\n[NEGATIVE]\nDo not show an empty sky after cloaking. Maintain a coherent, fast-moving delta-wing refraction that bends the background inside the aircraft’s exact volume.\nNo glowing outline, glass aircraft, force-field bubble, pixel dissolve, smoke camouflage, teleportation, morphing, or reappearance.\nNo extra aircraft, weapons, explosions, contrails, subtitles, on-screen text, watermarks, platform logos, or stickers.\nDo not introduce Chinese text, garbled characters, or misspellings.\n\n#kreaai #minimaxh3 #kreacpp",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 16,
      "views": 912,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084402067120988527",
    "title": "办公室里的荒诞早餐",
    "titleEn": "An Absurd Breakfast at the Office",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "办公室里的荒诞早餐的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an absurd breakfast at the office on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The Office & Breaking Bad.\n0:00–0:04 — Medium-wide handheld shot. Jesse angrily throws several shovelfuls of dirt into the deep hole. Realistic shovel impacts, scattering dirt, and heavy breathing. Dwight suddenly walks into frame, studies the hole suspiciously, and adjusts his glasses.\n\n0:04–0:07 — Dwight looks down at Jesse with a smug, sarcastic expression and says:\n\n“Hello, little man. Digging your own grave?”\n\n0:07–0:11 — Jesse immediately stops digging, slowly turns toward Dwight, and tightly grips the shovel. With an angry, threatening expression, he replies:\n\n“No, no, no… I’m digging it for you. Just like I did for Michael.”\n\n0:11–0:15 — Dwight’s confidence instantly collapses. His face fills with horror, his eyes widen, and he begins crying and panicking dramatically. He grabs Jesse by the shoulders and screams:\n\n“No! What did you do to Michael?! Where is he? Tell me now!”\n\nJesse remains completely emotionless, silently staring at Dwight while holding the shovel. The video ends abruptly on Dwight’s terrified face. \n\n#Flovaai #Flovacpp @Flovaai @Flovaai_Japan",
    "promptSourceUrl": "https://x.com/WolfRiccardo/status/2084402067120988527",
    "sourceUrl": "https://x.com/WolfRiccardo/status/2084402067120988527",
    "sourceLabel": "X 原帖 · @WolfRiccardo",
    "author": "Riccardo Wolf",
    "publishedAt": "2026-08-03T22:12:33.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084402067120988527.mp4",
    "posterUrl": "/posters/x/2084402067120988527.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Is this real? I honestly can’t believe it 😱😮\n\nI created this entire video from a single prompt in 2K quality using MiniMax H3 on @Flovaai_Japan \n\nPROMPT: The Office & Breaking Bad.\n0:00–0:04 — Medium-wide handheld shot. Jesse angrily throws several shovelfuls of dirt into the deep hole. Realistic shovel impacts, scattering dirt, and heavy breathing. Dwight suddenly walks into frame, studies the hole suspiciously, and adjusts his glasses.\n\n0:04–0:07 — Dwight looks down at Jesse with a smug, sarcastic expression and says:\n\n“Hello, little man. Digging your own grave?”\n\n0:07–0:11 — Jesse immediately stops digging, slowly turns toward Dwight, and tightly grips the shovel. With an angry, threatening expression, he replies:\n\n“No, no, no… I’m digging it for you. Just like I did for Michael.”\n\n0:11–0:15 — Dwight’s confidence instantly collapses. His face fills with horror, his eyes widen, and he begins crying and panicking dramatically. He grabs Jesse by the shoulders and screams:\n\n“No! What did you do to Michael?! Where is he? Tell me now!”\n\nJesse remains completely emotionless, silently staring at Dwight while holding the shovel. The video ends abruptly on Dwight’s terrified face. \n\n#Flovaai #Flovacpp @Flovaai @Flovaai_Japan",
    "engagement": {
      "replies": 35,
      "reposts": 19,
      "likes": 615,
      "views": 44795,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084387316286595436",
    "title": "办公室会议室的黄金笑料",
    "titleEn": "A Golden Joke in the Conference Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "办公室会议室的黄金笑料的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a golden joke in the conference room on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The Office (US) mockumentary style. Dunder Mifflin conference room,\nmid-afternoon, harsh fluorescent lighting, cheap office chairs,\nwhiteboard at the front. Handheld documentary camera, slightly\nunsteady, with sudden reframes and quick zooms toward whoever is\nreacting.\n\nMichael Scott stands at the whiteboard, overly excited, marker in\nhand, badly drawn diagram behind him. The staff sit around the table\nlooking exhausted. Stanley Hudson sits at the far end, glasses on,\nhead down over a folded crossword puzzle, completely disengaged.\n\nDIALOGUE:\n\nMICHAEL (loud, proud): \"Okay everybody, listen up. Large language\nmodels. L-L-Ms. This is the future of the paper industry.\"\n\nBeat. Nobody reacts. Camera pushes in slowly on Stanley — he does not\nlook up, does not blink, keeps filling in his crossword.\n\nMICHAEL: \"Which is why corporate is giving a bonus to whoever—\"\n\nInstant snap zoom to Stanley. His head lifts. He takes off his\nglasses. Dead serious.\n\nSTANLEY: \"I'm listening.\"\n\nMICHAEL (waving him off): \"—anyway. Moving on. Next topic. Prompt\nengineering.\"\n\nSnap zoom back to Stanley. He immediately puts his glasses back on,\nlowers his head, and returns to the crossword as if nothing happened.\n\nCut to Jim, who slowly turns his head and looks directly into the\ncamera lens with a flat, knowing expression. Hold for two seconds.\n\nSTYLE: single-camera sitcom shot like a documentary crew is present,\nnatural performances, no laugh track, muted office color palette,\nslightly soft consumer-camera look, characters occasionally glancing\nat the lens.",
    "promptSourceUrl": "https://x.com/Arminn_Ai/status/2084387316286595436",
    "sourceUrl": "https://x.com/Arminn_Ai/status/2084387316286595436",
    "sourceLabel": "X 原帖 · @Arminn_Ai",
    "author": "ΛRMIN",
    "publishedAt": "2026-08-03T21:13:56.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084387316286595436.mp4",
    "posterUrl": "/posters/x/2084387316286595436.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "No way, this is golden 🤣😂\n\nwith MiniMax H3 on @higgsfield_ai\n\n- Prompt:\nThe Office (US) mockumentary style. Dunder Mifflin conference room,\nmid-afternoon, harsh fluorescent lighting, cheap office chairs,\nwhiteboard at the front. Handheld documentary camera, slightly\nunsteady, with sudden reframes and quick zooms toward whoever is\nreacting.\n\nMichael Scott stands at the whiteboard, overly excited, marker in\nhand, badly drawn diagram behind him. The staff sit around the table\nlooking exhausted. Stanley Hudson sits at the far end, glasses on,\nhead down over a folded crossword puzzle, completely disengaged.\n\nDIALOGUE:\n\nMICHAEL (loud, proud): \"Okay everybody, listen up. Large language\nmodels. L-L-Ms. This is the future of the paper industry.\"\n\nBeat. Nobody reacts. Camera pushes in slowly on Stanley — he does not\nlook up, does not blink, keeps filling in his crossword.\n\nMICHAEL: \"Which is why corporate is giving a bonus to whoever—\"\n\nInstant snap zoom to Stanley. His head lifts. He takes off his\nglasses. Dead serious.\n\nSTANLEY: \"I'm listening.\"\n\nMICHAEL (waving him off): \"—anyway. Moving on. Next topic. Prompt\nengineering.\"\n\nSnap zoom back to Stanley. He immediately puts his glasses back on,\nlowers his head, and returns to the crossword as if nothing happened.\n\nCut to Jim, who slowly turns his head and looks directly into the\ncamera lens with a flat, knowing expression. Hold for two seconds.\n\nSTYLE: single-camera sitcom shot like a documentary crew is present,\nnatural performances, no laugh track, muted office color palette,\nslightly soft consumer-camera look, characters occasionally glancing\nat the lens.",
    "engagement": {
      "replies": 24,
      "reposts": 22,
      "likes": 447,
      "views": 75621,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084355335222108362",
    "title": "高级时装大片的全流程",
    "titleEn": "Building a Premium Fashion Campaign",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "高级时装大片的全流程的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for building a premium fashion campaign on X. Native-video output: 15s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Minimax 3 Prompt: Create a premium 16:9 landscape fashion film. Use Image 1 for the overall mood, location, and film texture; Image 2 for the talent; Image 3 for the bag; and Image 4 for the closing brand mark. This is a fashion campaign for the clothing and bag. The tone is elevated, cool, and restrained, but the edit should still feel lively and fashion-forward—not like a conventional narrative film or an e-commerce ad.\n\nKeep the story simple: beside a vintage car on a desert highway, a woman walks to the rear of the car, opens the trunk, takes out a black bag, shares a quiet beat with the man standing nearby, then leaves carrying the bag. Integrate the clothing and bag naturally into the performance so they feel like part of the characters’ identity.\n\nI added the final video in and the reference images for you to learn from!",
    "promptSourceUrl": "https://x.com/influencer_seo/status/2084355335222108362",
    "sourceUrl": "https://x.com/influencer_seo/status/2084355335222108362",
    "sourceLabel": "X 原帖 · @influencer_seo",
    "author": "Bennett | Generative Media | AI SEO @ fal",
    "publishedAt": "2026-08-03T19:06:52.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084355335222108362.mp4",
    "posterUrl": "/posters/x/2084355335222108362.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "retro"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Want to learn how to build a stunning AI fashion campagn using the new MiniMax-H3 Video model with open weights? Use this prompt below 👇\n\nMinimax 3 Prompt: Create a premium 16:9 landscape fashion film. Use Image 1 for the overall mood, location, and film texture; Image 2 for the talent; Image 3 for the bag; and Image 4 for the closing brand mark. This is a fashion campaign for the clothing and bag. The tone is elevated, cool, and restrained, but the edit should still feel lively and fashion-forward—not like a conventional narrative film or an e-commerce ad.\n\nKeep the story simple: beside a vintage car on a desert highway, a woman walks to the rear of the car, opens the trunk, takes out a black bag, shares a quiet beat with the man standing nearby, then leaves carrying the bag. Integrate the clothing and bag naturally into the performance so they feel like part of the characters’ identity.\n\nI added the final video in and the reference images for you to learn from!",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 22,
      "views": 1807,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084338573210157293",
    "title": "从静帧进入虚构角色扮演游戏",
    "titleEn": "A Still Frame Opens into a Fantasy RPG",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "从静帧进入虚构角色扮演游戏的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a still frame opens into a fantasy rpg on X. Native-video output: 15s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Camera: First-person, eye level, handheld gameplay. Simulate a player operating a modern military FPS while holding a frost-covered assault rifle and cautiously advancing through a remote Arctic military installation. Move forward across an icy service road between snowbanks, radar stations and frozen cargo containers, periodically scanning rooftops, watchtowers and mountain ridgelines. Sweep the reticle naturally across possible threats, briefly stop behind cover to fire several controlled bursts toward a distant enemy near the radar dome, then continue advancing deeper into the frozen base like authentic player-controlled gameplay.\n\nLighting: Cold Arctic daylight with soft overcast skies and pale winter sunlight reflecting across ice and snow. Blend crisp blue ambient light with warm floodlights from the base, vehicle headlights, subtle smoke drifting from generators, blowing snow, frozen breath, and atmospheric ice particles. Maintain ultra-photorealistic AAA rendering with highly detailed weapon materials, realistic snow deformation, wet ice reflections, volumetric fog, and cinematic environmental depth.\n\nCamera movement: Natural player-controlled movement with subtle handheld sway while walking across uneven frozen terrain. Perform small tactical checks left and right, briefly aim down the central lane, add realistic recoil and muzzle flash while firing short bursts, then smoothly recover and continue pushing forward. Include slight weapon bobbing, responsive aim corrections, natural acceleration and deceleration, and authentic FPS movement throughout, preserving the immersive feel of high-end modern military gameplay.\n2/2",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2084338601332994469",
    "sourceUrl": "https://x.com/LudovicCreator/status/2084338573210157293",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-03T18:00:15.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084338573210157293.mp4",
    "posterUrl": "/posters/x/2084338573210157293.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is a huge model for some themes .\n\nHere an example to animate a fake RPG Game.\n\nStarting image and prompt first comment \n\n@Hailuo_AI #MiniMaxH3\n1/2",
    "engagement": {
      "replies": 9,
      "reposts": 6,
      "likes": 69,
      "views": 2955,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084311486277599712",
    "title": "跑过静电噪声",
    "titleEn": "Outrun the Static",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "跑过静电噪声的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for outrun the static on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "️\n\n{   \"title\": \"Outrun the Static\",   \"duration\": \"15 seconds\",   \"aspect_ratio\": \"9:16\",   \"style\": \"high-quality stylised 3D animation blended with graphic-novel visuals, vibrant neon and daylight contrast, expressive motion blur, bold geometric lighting, fluid animation, dynamic social-media advertising, polished but not photorealistic\",   \"prompt\": \"Create an explosive, fast-paced 15-second vertical animated advertisement about shattering mental noise and building instant momentum. Open immediately inside a monochrome, static-filled city street. A slouched, hesitant urban runner stands frozen as grey glitching words reading SOMEDAY, TOO TIRED, and NOT TODAY orbit them like heavy fog. Rapid visual storytelling with zero delay: the runner stomps their foot on the pavement, pulling their laces tight as vibrant electric-cyan energy flares from their shoes. They raise their gaze with sharp focus and burst into an all-out sprint. As they stride through the grey street, each powerful footstrike shatters a floating word into glowing neon shards and transforms the dreary alley into a vibrant, high-contrast world of motion light-trails. The camera dynamically tracks alongside them in rapid close-ups and side angles. On their final leap through a giant wall of static, the fog completely shatters into brilliant golden light, revealing the transformed runner reaching a sunlit city summit. Finish with the runner standing empowered in peak condition as bold text snaps onto screen: BREAK THE STATIC. RUN YOUR DAY. Maintain relentless energy, satisfying impact beats, fluid motion, crisp transitions, and a motivational narrative that makes the viewer want to move immediately.\",   \"timeline\": [     {       \"time\": \"0.0-3.0 seconds\",       \"visual\": \"Immediate hook: monochrome grey street filled with glitch static. The runner stands frozen in heavy posture as floating grey words (SOMEDAY, TOO TIRED, NOT TODAY) orbit around them. Fast camera push-in.\",       \"text\": \"STUCK IN THE STATIC?\"     },     {       \"time\": \"3.0-6.5 seconds\",       \"visual\": \"Rapid close-ups of laces snapping tight and shoe sole hitting the ground. Cyan energy sparks out. The runner's posture transforms instantly from sluggish to razor-sharp focus.\",       \"text\": \"\"     },     {       \"time\": \"6.5-11.5 seconds\",       \"visual\": \"High-speed sprinting sequence with dynamic side tracking camera. Every footstrike explodes with impact fx, shattering the floating negative words into neon sparks and painting the gray environment in bright colors.\",       \"text\": \"\"     },     {       \"time\": \"11.5-15.0 seconds\",       \"visual\": \"The runner leaps through the final static wall into warm, golden sunlight over an elevated city skyline. Fast pullback as they land confidently under bright skies.\",       \"text\": \"BREAK THE STATIC\\nRUN YOUR DAY\"     }   ],   \"voiceover\": {     \"enabled\": true,     \"delivery\": \"direct, energetic, crisp, and motivating, like a high-performance coach speaking directly to the viewer\",     \"script\": \"Stuck in the static? The hardest part is standing still. Lock in. Outrun yesterday.\"   },   \"audio\": {     \"music\": \"fast-paced modern electronic-hiphop training beat with driving bass drop at second 3.0\",     \"sound_design\": \"glitch hum, sharp lace snap, heavy footfall impacts, glass shatter FX for words, wind rush, crisp final bass thump\",     \"mix\": \"Punchy and clear voiceover balanced over a driving beat; footsteps and word shatters synced precisely to the musical beat.\"   },   \"camera\": {     \"movement\": \"rapid whip push-in, low-angle boot macro shots, dynamic parallel tracking run, and high-speed final camera pullback\",     \"editing\": \"high temporal frequency, new camera angle every 0.6 to 1.0 seconds, motion-matched cuts\"   },   \"negative_prompt\": [     \"photorealistic characters\",     \"slow opening\",     \"dramatic movie trailer tone\",     \"long pauses\",     \"cluttered camera motion\",     \"anatomy glitches\",     \"unreadable text\"   ] }",
    "promptSourceUrl": "https://x.com/sebatheepan/status/2084311486277599712",
    "sourceUrl": "https://x.com/sebatheepan/status/2084311486277599712",
    "sourceLabel": "X 原帖 · @sebatheepan",
    "author": "Pan",
    "publishedAt": "2026-08-03T16:12:37.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084311486277599712.mp4",
    "posterUrl": "/posters/x/2084311486277599712.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 from Hailuio can easily make motivational videos like this \n\nPrompt⤵️\n\n{   \"title\": \"Outrun the Static\",   \"duration\": \"15 seconds\",   \"aspect_ratio\": \"9:16\",   \"style\": \"high-quality stylised 3D animation blended with graphic-novel visuals, vibrant neon and daylight contrast, expressive motion blur, bold geometric lighting, fluid animation, dynamic social-media advertising, polished but not photorealistic\",   \"prompt\": \"Create an explosive, fast-paced 15-second vertical animated advertisement about shattering mental noise and building instant momentum. Open immediately inside a monochrome, static-filled city street. A slouched, hesitant urban runner stands frozen as grey glitching words reading SOMEDAY, TOO TIRED, and NOT TODAY orbit them like heavy fog. Rapid visual storytelling with zero delay: the runner stomps their foot on the pavement, pulling their laces tight as vibrant electric-cyan energy flares from their shoes. They raise their gaze with sharp focus and burst into an all-out sprint. As they stride through the grey street, each powerful footstrike shatters a floating word into glowing neon shards and transforms the dreary alley into a vibrant, high-contrast world of motion light-trails. The camera dynamically tracks alongside them in rapid close-ups and side angles. On their final leap through a giant wall of static, the fog completely shatters into brilliant golden light, revealing the transformed runner reaching a sunlit city summit. Finish with the runner standing empowered in peak condition as bold text snaps onto screen: BREAK THE STATIC. RUN YOUR DAY. Maintain relentless energy, satisfying impact beats, fluid motion, crisp transitions, and a motivational narrative that makes the viewer want to move immediately.\",   \"timeline\": [     {       \"time\": \"0.0-3.0 seconds\",       \"visual\": \"Immediate hook: monochrome grey street filled with glitch static. The runner stands frozen in heavy posture as floating grey words (SOMEDAY, TOO TIRED, NOT TODAY) orbit around them. Fast camera push-in.\",       \"text\": \"STUCK IN THE STATIC?\"     },     {       \"time\": \"3.0-6.5 seconds\",       \"visual\": \"Rapid close-ups of laces snapping tight and shoe sole hitting the ground. Cyan energy sparks out. The runner's posture transforms instantly from sluggish to razor-sharp focus.\",       \"text\": \"\"     },     {       \"time\": \"6.5-11.5 seconds\",       \"visual\": \"High-speed sprinting sequence with dynamic side tracking camera. Every footstrike explodes with impact fx, shattering the floating negative words into neon sparks and painting the gray environment in bright colors.\",       \"text\": \"\"     },     {       \"time\": \"11.5-15.0 seconds\",       \"visual\": \"The runner leaps through the final static wall into warm, golden sunlight over an elevated city skyline. Fast pullback as they land confidently under bright skies.\",       \"text\": \"BREAK THE STATIC\\nRUN YOUR DAY\"     }   ],   \"voiceover\": {     \"enabled\": true,     \"delivery\": \"direct, energetic, crisp, and motivating, like a high-performance coach speaking directly to the viewer\",     \"script\": \"Stuck in the static? The hardest part is standing still. Lock in. Outrun yesterday.\"   },   \"audio\": {     \"music\": \"fast-paced modern electronic-hiphop training beat with driving bass drop at second 3.0\",     \"sound_design\": \"glitch hum, sharp lace snap, heavy footfall impacts, glass shatter FX for words, wind rush, crisp final bass thump\",     \"mix\": \"Punchy and clear voiceover balanced over a driving beat; footsteps and word shatters synced precisely to the musical beat.\"   },   \"camera\": {     \"movement\": \"rapid whip push-in, low-angle boot macro shots, dynamic parallel tracking run, and high-speed final camera pullback\",     \"editing\": \"high temporal frequency, new camera angle every 0.6 to 1.0 seconds, motion-matched cuts\"   },   \"negative_prompt\": [     \"photorealistic characters\",     \"slow opening\",     \"dramatic movie trailer tone\",     \"long pauses\",     \"cluttered camera motion\",     \"anatomy glitches\",     \"unreadable text\"   ] }",
    "engagement": {
      "replies": 7,
      "reposts": 1,
      "likes": 34,
      "views": 1420,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084289809359810731",
    "title": "霓虹暴雨中的赛博武士",
    "titleEn": "A Cyberpunk Samurai in Neon Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "霓虹暴雨中的赛博武士的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 12 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a cyberpunk samurai in neon rain on X. Native-video output: 12s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Continuous 15-second cinematic shot, extreme slow-motion: a futuristic cyberpunk samurai standing in heavy neon rain, reflective wet armor. The samurai slowly draws a glowing katana, lens flare, volumetric neon fog, dynamic camera panning around the character, continuous fluid motion, photorealistic 8k",
    "promptSourceUrl": "https://x.com/noorwithwifi/status/2084289809359810731",
    "sourceUrl": "https://x.com/noorwithwifi/status/2084289809359810731",
    "sourceLabel": "X 原帖 · @noorwithwifi",
    "author": "NoorAI",
    "publishedAt": "2026-08-03T14:46:29.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084289809359810731.mp4",
    "posterUrl": "/posters/x/2084289809359810731.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Generated with @Hailuo_AI #MiniMaxH3\n\nPrompt:\n\nContinuous 15-second cinematic shot, extreme slow-motion: a futuristic cyberpunk samurai standing in heavy neon rain, reflective wet armor. The samurai slowly draws a glowing katana, lens flare, volumetric neon fog, dynamic camera panning around the character, continuous fluid motion, photorealistic 8k",
    "engagement": {
      "replies": 40,
      "reposts": 4,
      "likes": 77,
      "views": 1985,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084233895202197998",
    "title": "日系偶像的五秒问候",
    "titleEn": "A Five-Second Greeting from an Anime Idol",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "日系偶像的五秒问候的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 5 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a five-second greeting from an anime idol on X. Native-video output: 5s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "プロンプト置いときます✋\n\nJapanese anime style, high-end TV anime production quality, soft pastel lighting, clean line art, subtle cel shading, bright and cheerful idol stage atmosphere, glowing lens flare, gentle bokeh in background, smooth vibrant 60fps animation feel.\n\nScene overview: An adorable anime idol girl stands center stage, looking directly at the camera with a brilliant smile. She holds a colorful microphone, energetic and joyful, giving a cute greeting to her fans. A short, charming 5-second anime idol greeting scene.\n\nStoryboard:\n[0s-1.5s] Shot 1: Medium close-up: The idol girl steps forward with a big smile, bringing the microphone to her mouth, blinking playfully as stage lights twinkle behind her.\n[1.5s-3.5s] Shot 2: Close-up on her face: She speaks clearly into the microphone with precise lip-sync: \"みんな、こんにちは！今日も一緒に楽しもうね！\" (Minna, konnichiwa! Kyou mo issho ni tanoshimou ne!). She winks on \"tanoshimou ne!\".\n[3.5s-5s] Shot 3: Slightly pulled-back medium shot: She strikes a cute victory-pose/heart-pose with her free hand, tilting her head with a beaming smile as sparkle particles drift around her.\n\nCamera: Smooth pan-in at the start, staying centered and static during the lip-sync for maximum clarity, subtle handheld bounce on the final pose at 3.5s.\n\nAudio: Crisp female anime idol voice clearly pronouncing \"みんな、こんにちは！今日も一緒に楽しもうね！\" in Japanese. Underneath, an upbeat synth-pop idol BGM playing, accompanied by a soft microphone \"pop\" sound at 0.5s and a sparkling magic SFX hit at 3.5s during the pose.\n\nNo real-life live action rendering, no realistic skin texture, no 3D CG look, no English subtitles or on-screen text clutter, keep the pure 2D anime aesthetic.",
    "promptSourceUrl": "https://x.com/riddi0908/status/2084538454596354549",
    "sourceUrl": "https://x.com/riddi0908/status/2084233895202197998",
    "sourceLabel": "X 原帖 · @riddi0908",
    "author": "まゆひらa",
    "publishedAt": "2026-08-03T11:04:18.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084233895202197998.mp4",
    "posterUrl": "/posters/x/2084233895202197998.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3、プロンプトはGemini\n\nComfyUIで対応したMiniMax H3の、プロンプトの書き方を調べてください\n↓\nComfyUIのワークフローのサンプルプロンプトは下記のとおりでした。今までの知見から、アニメ調でアイドルの女の子が日本語で挨拶をする5秒のt2v動画のプロンプトを考案してください。\n（略）",
    "engagement": {
      "replies": 2,
      "reposts": 10,
      "likes": 94,
      "views": 10748,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084227244533411987",
    "title": "把教育愿景变成学习奇观",
    "titleEn": "Turning an Educational Vision into an Experience",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "把教育愿景变成学习奇观的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for turning an educational vision into an experience on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second animated educational video that teaches young children the letters A, B, C, and D.\n\nThe learning pattern for every letter must be:\n\nLETTER → SOUND → OBJECT → PLAYFUL ACTION → OBJECT NAME\n\nTarget audience: children ages 3 to 6.\n\nVisual style:\nUse adorable rounded 3D characters, soft pastel colors, gentle facial expressions, and simple recognizable objects. Combine this with a premium minimalist technology aesthetic featuring clean white space, elegant composition, soft studio lighting, subtle reflections, smooth gradients, rounded geometry, crisp typography, and extremely polished transitions.\n\nThe animation should feel playful and child-friendly while remaining calm, uncluttered, and beautifully designed.\n\nUse a clean off-white background with a different soft color glow behind each letter.\n\n0:00–0:01 | Introduction\nA small smiling star mascot bounces into the center of the screen.\n\nColorful letters briefly float around it.\n\nDisplay the text:\n\n“Let’s learn!”\n\nThe mascot taps the screen, creating a soft ripple that reveals the first letter.\n\n0:01–0:04 | A is for Apple\n\nShow a large uppercase “A” and smaller lowercase “a” beside it.\nUse thick, rounded, highly readable typography.\n\nThe narrator says:\n\n“A. A says ah. A is for Apple.”\n\nThe uppercase A gently inflates and transforms into a shiny red apple.\n\nIts top point becomes the apple stem, and a small green leaf unfolds from the side.\n\nThe apple gains a cute smiling face and performs one soft bounce.\n\nDisplay the word:\n\n“APPLE”\n\nHighlight the first letter A in red.\n\nAdd a soft pop and a tiny crunchy sound.\n\n0:04–0:07 | B is for Ball\n\nThe apple rolls across the screen and leaves behind a curved red trail.\n\nThe trail loops twice and forms a large uppercase “B,” with a lowercase “b” appearing beside it.\nThe narrator says:\n\n“B. B says buh. B is for Ball.”\n\nThe two rounded sections of the B expand and merge into a colorful striped ball.\n\nThe ball bounces twice with playful squash-and-stretch animation.\n\nDisplay the word:\n\n“BALL”\n\nHighlight the first letter B in blue.\n\nSynchronize each bounce with a soft musical note.\n\n0:07–0:10 | C is for Cat\n\nOn its final bounce, the ball stretches into a curved shape and becomes a large uppercase “C.”\n\nA lowercase “c” slides gently into place beside it.\n\nThe narrator says:\n“C. C says kuh. C is for Cat.”\n\nThe C rotates and becomes the curled tail of a cute orange cat.\n\nThe rest of the cat forms from soft rounded shapes.\n\nThe cat stretches, blinks, and gives one gentle wave with its paw.\n\nDisplay the word:\n\n“CAT”\n\nHighlight the first letter C in orange.\n\nAdd a quiet and friendly “meow.”\n\n0:10–0:13 | D is for Duck\n\nThe cat’s tail uncurls and transforms into the curved side of a large uppercase “D.”\n\nA lowercase “d” pops up beside it.\n\nThe narrator says:\n\n“D. D says duh. D is for Duck.”\nThe straight line of the D becomes the duck’s neck.\n\nThe curved section becomes its round yellow body.\n\nA small orange beak and two tiny wings pop into place.\n\nThe duck waddles forward, flaps its wings, and gives one cheerful quack.\n\nDisplay the word:\n\n“DUCK”\n\nHighlight the first letter D in yellow.\n\nAdd tiny water ripples beneath its feet.\n\n0:13–0:15 | Recap\n\nThe apple, ball, cat, and duck slide into four clean rounded tiles.\n\nPlace their letters above them:\n\n“A  B  C  D”\n\nThe mascot returns and points to each object as they bounce once in sequence.\n\nNarrator:\n“A, B, C, D. Great job!”\n\nFinish with the text:\n\n“Great job!”\n\nUse a small sparkle animation and a warm musical chime.\n\nAnimation requirements:\n\nKeep each letter fully visible for a moment before it transforms.\n\nShow uppercase and lowercase versions clearly.\n\nMake every object instantly recognizable.\n\nUse smooth shape morphing so children can visually understand how the letter becomes the object.\n\nMaintain stable spelling, clean letterforms, accurate object shapes, and consistent character design.\n\nUse gentle squash-and-stretch, soft motion blur, subtle shadows, polished lighting, and precisely synchronized sound effects.\nAvoid fast camera movement, cluttered backgrounds, harsh colors, tiny text, warped letters, random symbols, duplicated objects, scary expressions, or overly complex transformations.\n\nThe final video should feel cute, educational, memorable, calming, and exceptionally polished.",
    "promptSourceUrl": "https://x.com/umesh_ai/status/2084227247586832868",
    "sourceUrl": "https://x.com/umesh_ai/status/2084227244533411987",
    "sourceLabel": "X 原帖 · @umesh_ai",
    "author": "Umesh",
    "publishedAt": "2026-08-03T10:37:52.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084227244533411987.mp4",
    "posterUrl": "/posters/x/2084227244533411987.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Video models are going to revolutionize education!\n\nYou can turn your vision into a prompt and generate a learning experience that once required an entire production team.\n\nMade with HailuoAI MiniMax H3\n\nPrompt ⤵️",
    "engagement": {
      "replies": 39,
      "reposts": 50,
      "likes": 459,
      "views": 55774,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply.",
    "featured": {
      "at": "2026-09-05T15:50:15.686Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2084201268977844521",
    "title": "凯恩·门罗的动态字体登场",
    "titleEn": "Cairn Munro Arrives in Motion Type",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "凯恩·门罗的动态字体登场的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for cairn munro arrives in motion type on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "CAIRN MUNRO — \"ARRIVE UNHEARD\" — 14 CUTS · 15s · 2K · 24fps  REFERENCES: the attached PRODUCT SHEET@Image1 is the exact vehicle — match its proportion, stance, glasshouse angle, cladding, rack and light signature exactly, invent nothing. The attached TYPE SHEET@Image2 is the exact typeface, weight, tracking and scale for every overlay. Match it.  AUDIO LAW — HIGHEST PRIORITY. NO MUSIC ANYWHERE. Do not generate score, soundtrack, background music, beat, drums, percussion, bassline, synth pad, drone, melody, chord, ambient bed, riser, braam, trailer hit, stinger, swell, or any tonal instrument. Where music would normally be scored, render SILENCE. NO VOICE — no narration, dialogue, vocals, breath. ONLY PERMITTED AUDIO, all of it environmental: rain striking wet granite; rain drumming on metal and glass; wind over ridge grass; a burn running hard over rock; gravel shifting; distant thunder; and true silence. All of it recorded-real, close, dry, no reverb wash. Nothing sustains musically.  SILENCE LAW — the engine of the film. The landscape is loud. The vehicle adds nothing to it. No engine, no exhaust, no motor whine, no transmission, no tyre roar, no wind noise off the body, no door, no suspension. Rain striking its panels is permitted — that is the world touching the car, not the car making sound. Whenever the vehicle fills frame and nothing is striking it, the mix drops to near-silence. If an engine or tyre sound would normally be added to a moving vehicle, render silence instead. An engine sound anywhere in this film is a failure.  EDIT METRONOME — SILENT: cuts locked to 96 BPM, 1 beat = 0.625s, 24 beats. A timing grid for cutting ONLY. Do not generate audio to this tempo.  BRAND: CAIRN. PRODUCT: CAIRN MUNRO, a fully electric expedition overlander. It is silent because it is electric — the claim is literal.  SUBJECT: per the sheet — boxy full-size SUV, upright glasshouse, flat vertical sides, squared arches, short overhangs, high clearance. Matte peat green-black body, charcoal composite cladding and sliders, raw anodised aluminium rack and skid plate, bone-white roof, signal-orange recovery hooks, full-width thin white LED signature front and rear, no grille, all-terrain tyres on flat-faced aluminium wheels. No badging.  WORLD: the Scottish Highlands in heavy weather. Wet granite, peat, ridge grass and heather. Low cloud moving fast across the tops. Constant rain. A single unmade stone track. No road markings, no buildings, no fences.  PALETTE: peat #1B2320, granite #4A4E4B, mist #D9DDDB, aluminium #8E9490, heather #6B5470. Signal orange #E24E1B on the recovery hooks only, never more than a few pixels. Cold northern light, overcast, no sun, no warmth anywhere. Desaturated but never graded grey — these are true wet colours.  MIST LAW: mist is real volume. It moves, thins, and parts to reveal. It never becomes a soft filter — everything outside the mist stays razor sharp. It never glows, never catches a god ray, never becomes smoke.  TYPE LAW: every overlay uses the TYPE SHEET face, CAIRN EXTENDED — wide uppercase, medium weight, wide tracking, pure white, with the hairline contour rule beneath each line. Two positions only: body lines lower-left on the safe-area grid, the lockup dead centre. Type is revealed by a hard-edged horizontal wipe travelling left to right. It never fades, scales, rotates, skews, glows or casts a shadow. It cuts on with the cut and holds absolutely still until the next cut takes it. Always pure white.  TEXT ACCURACY: every overlay word is spelled exactly as written and appears exactly once. No extra letters, no duplicated lines, no garbled glyphs, no invented words. No text of any kind on the vehicle.  CAMERA: motion-control smooth and heavy, no handheld, no drift. Long-lens tracking and slow cranes. Clean digital, no grain. Directional blur in transit; everything arrives sharp and holds still.  RHYTHM: cuts 01–02 hold two beats. Cuts 03–10 are one beat apart, the spec run. Cuts 11–12 hold two beats. Cuts 13 and 14 hold four beats each.  01 · 0.000 · Wide. A granite ridge under fast-moving low cloud, rain falling hard across frame. No vehicle. SOUND: rain on stone and wind, loud.  02 · 1.250 · Macro. Rain hammering a wet granite face, water sheeting down the fractures. TYPE: EVERYTHING HERE  03 · 2.500 · A burn crashing white over black rock. TYPE: MAKES A SOUND  04 · 3.125 · Wind flattening ridge grass and heather in one long gust.  05 · 3.750 · The vehicle enters, low and wide, emerging from mist on the stone track. The mix drops away to rain only. TYPE: ONE THING DOESN'T  06 · 4.375 · Macro. A tyre rolling through standing water, the water throwing up hard. SOUND: water only, no tyre noise. TYPE: 800 VOLT  07 · 5.000 · Macro. Rain drumming on the bone-white roof panel. TYPE: 710 KM  08 · 5.625 · Macro. Water running off the anodised aluminium rack. TYPE: 900 NM  09 · 6.250 · Low long-lens track alongside the flat body side, mist parting off the panel. TYPE: 1050MM WADING  10 · 6.875 · Macro. The thin white LED signature cutting through mist. TYPE: 3.5T TOWING  11 · 7.500 · High wide. The vehicle small on a vast stone plateau, cloud dragging over it. SOUND: wind, distant thunder.  12 · 8.750 · Front three-quarter, wading a burn, water pushing up over the sills and breaking white. SOUND: water only.  13 · 10.000 · THE SHOT. A single mature red deer stag stands side-on on the ridge above the track, wet coat, broad antlers, head turned to camera. The vehicle passes below him in the same frame. The stag does not flinch, does not run, does not move at all. Hold on him. SOUND: rain and wind only — no vehicle sound whatsoever.  14 · 12.500 · The vehicle static on the ridge in moving mist, dead frontal. TYPE: CAIRN, dead centre. Two beats later MUNRO wipes on beneath it, then ARRIVE UNHEARD. beneath that. Hold, cut to black. SOUND: rain and wind, then hard silence on the cut. No end cue, no final hit.  NEGATIVE: no music, score, beat, instruments, narration, voice, engine noise, tyre noise. No people, hands, faces, drivers, passengers. No other vehicles, no roads, no road markings, no signage, no buildings, no fences, no power lines. No text other than the specified overlays — no captions, subtitles, watermarks, UI, timecode. No badging or lettering on the vehicle. No sun, no warm light, no golden hour, no lens flare, no god rays, no smoke, no snow. No glitch, scanlines, film burn, light leaks. The stag never runs. No colour outside the palette. No invented body parts.",
    "promptSourceUrl": "https://x.com/Dheepanratnam/status/2084201274661056775",
    "sourceUrl": "https://x.com/Dheepanratnam/status/2084201268977844521",
    "sourceLabel": "X 原帖 · @Dheepanratnam",
    "author": "Dheepan Ratnam",
    "publishedAt": "2026-08-03T08:54:39.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084201268977844521.mp4",
    "posterUrl": "/posters/x/2084201268977844521.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax-H3 is now publicly available \n\nConcept + typography: ChatGPT 2 images \nAnimation: Minimax-H3\n\nCairn Munro. Arrive unheard. \nText rendering + many more great features\nReference images and prompt ⤵️",
    "engagement": {
      "replies": 11,
      "reposts": 3,
      "likes": 50,
      "views": 2864,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2084075915706052621",
    "title": "洗衣店里的超现实想象",
    "titleEn": "A Surreal Vision Inside a Laundromat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "洗衣店里的超现实想象的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a surreal vision inside a laundromat on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[1] FORMAT\nCreate a 15-second, 16:9 surreal live-action scene inside a quiet laundromat at night.\n\n[4] BEATS\n[0–3s] Slow handheld push toward a running front-loading washing machine.\n[3–6s] Inside the drum, the water becomes a miniature stormy ocean with waves, lightning, and a tiny lighthouse.\n[6–11s]The lighthouse beam shines through the glass as the storm grows stronger.\n[11-15s]The door suddenly opens. Only ordinary wet laundry falls onto the floor.\n\n[5] CAMERA\nOne continuous handheld push-in, ending on the falling laundry.\n\n[6] LIGHT\nCold flickering fluorescent lights, wet glass reflections, deep blue storm light inside the washer, realistic phone-camera texture.\n\n[8] AUDIO\nAudio: washing-machine rumble, water splashes, distant thunder, one door click, and wet laundry hitting the floor.\nBGM: a subtle 15-second mystery drone, building from 4 seconds and stopping abruptly when the door opens.\n\n[9] NEGATIVE\nKeep the laundromat fully photorealistic. The miniature ocean exists only inside the washing machine.\nNo people, text, subtitles, logos, watermarks, monsters, or cartoon styling.\n\n#minimaxh3 #hailuoai",
    "promptSourceUrl": "https://x.com/opener_ai/status/2084075915706052621",
    "sourceUrl": "https://x.com/opener_ai/status/2084075915706052621",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-03T00:36:33.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2084075915706052621.mp4",
    "posterUrl": "/posters/x/2084075915706052621.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo minimax H3's expressive imagination is also top-notch.\n@Hailuo_AI \n\nPrompt\n\n[1] FORMAT\nCreate a 15-second, 16:9 surreal live-action scene inside a quiet laundromat at night.\n\n[4] BEATS\n[0–3s] Slow handheld push toward a running front-loading washing machine.\n[3–6s] Inside the drum, the water becomes a miniature stormy ocean with waves, lightning, and a tiny lighthouse.\n[6–11s]The lighthouse beam shines through the glass as the storm grows stronger.\n[11-15s]The door suddenly opens. Only ordinary wet laundry falls onto the floor.\n\n[5] CAMERA\nOne continuous handheld push-in, ending on the falling laundry.\n\n[6] LIGHT\nCold flickering fluorescent lights, wet glass reflections, deep blue storm light inside the washer, realistic phone-camera texture.\n\n[8] AUDIO\nAudio: washing-machine rumble, water splashes, distant thunder, one door click, and wet laundry hitting the floor.\nBGM: a subtle 15-second mystery drone, building from 4 seconds and stopping abruptly when the door opens.\n\n[9] NEGATIVE\nKeep the laundromat fully photorealistic. The miniature ocean exists only inside the washing machine.\nNo people, text, subtitles, logos, watermarks, monsters, or cartoon styling.\n\n#minimaxh3 #hailuoai",
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 20,
      "views": 1268,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083909535593644291",
    "title": "每个伟大故事的动态标题",
    "titleEn": "Every Great Story Begins in Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "每个伟大故事的动态标题的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for every great story begins in motion on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second cinematic text-animation video built around the quote:  “Every great change begins quietly, grows through courage, and becomes impossible to ignore.”  The quote should appear gradually as a visual story. Each new phrase must transform the design, atmosphere, movement, and emotional intensity of the scene. Use elegant typography, accurate spelling, cinematic lighting, smooth transitions, and perfectly readable text.  0:00–0:03 | “Every great change”  Begin with a completely black screen.  A tiny point of warm light slowly appears in the center, like the first spark of an idea.  The words “Every great change” emerge softly from the darkness, one word at a time. Use thin, elegant serif typography with wide letter spacing.  “Every” fades in gently.  “Great” grows slightly larger.  “Change” forms from small drifting particles that gather into solid letters.  Keep the scene quiet, minimal, and mysterious.  0:03–0:06 | “begins quietly,”  The camera slowly moves closer to the text.  The previous words shrink and reposition toward the upper-left corner as the phrase “begins quietly,” appears in delicate lowercase letters.  Animate the phrase as though it is being written by an invisible hand.  Each letter should create a subtle ripple in the darkness.  Introduce faint textures, soft shadows, floating dust, and gentle light rays.  The comma should appear last and create a small circular pulse.  0:06–0:09 | “grows through courage,”  The pulse expands and transforms the scene from darkness into a rich sunrise gradient with deep orange, red, and golden tones.  The words “grows through courage” rise upward from the bottom of the frame.  Animate “grows” by gradually increasing its size and weight.  Animate “through” along a curved path.  Animate “courage” in bold uppercase letters that push through a translucent barrier, causing it to crack into geometric fragments.  The movement should feel powerful but controlled.  0:09–0:12 | “and becomes”  The fragments rotate in slow motion and reorganize into a clean editorial grid.  The phrase “AND BECOMES” appears across the frame in condensed sans-serif typography.  Animate the letters with fast tracking changes, vertical stretching, masking, and perspective movement.  The camera accelerates forward through the center of the word “BECOMES.”  The sound and visual energy should steadily build.  0:12–0:14 | “impossible to ignore.”  Reveal a vast bright space filled with light, moving shapes, and large-scale typography.  The words “IMPOSSIBLE TO IGNORE” appear one after another.  “IMPOSSIBLE” expands beyond the edges of the screen.  “TO” remains small and perfectly centered.  “IGNORE” slams into place with strong visual impact, briefly shaking the surrounding grid and shapes.  Use bold contrast, dramatic scale, sharp shadows, and synchronized motion.  0:14–0:15 | Final quote  All movement stops instantly.  The complete quote appears centered on a clean off-white background:  “Every great change begins quietly, grows through courage, and becomes impossible to ignore.”  Use refined black typography with “change,” “courage,” and “impossible” highlighted in deep red.  Hold the final composition clearly for the last second.  Maintain one continuous visual journey from darkness to light, silence to impact, and simplicity to complexity. Keep every phrase connected through visual transformations rather than hard cuts.  Use realistic motion blur, precise kerning, clean masks, stable letterforms, smooth camera movement, subtle film grain, cinematic sound design, rising ambient music, soft particles, controlled color transitions, and a final deep impact sound.  Avoid misspelled words, warped letters, duplicated characters, unreadable text, random symbols, excessive flickering, chaotic layouts, inconsistent fonts.",
    "promptSourceUrl": "https://x.com/umesh_ai/status/2083909535593644291",
    "sourceUrl": "https://x.com/umesh_ai/status/2083909535593644291",
    "sourceLabel": "X 原帖 · @umesh_ai",
    "author": "Umesh",
    "publishedAt": "2026-08-02T13:35:25.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2083909535593644291.mp4",
    "posterUrl": "/posters/x/2083909535593644291.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "retro"
    ],
    "scenes": [
      "music-performance",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is incredible at text rendering!\n\nThis is a text to video prompt!\n\nPrompt : Create a 15-second cinematic text-animation video built around the quote:  “Every great change begins quietly, grows through courage, and becomes impossible to ignore.”  The quote should appear gradually as a visual story. Each new phrase must transform the design, atmosphere, movement, and emotional intensity of the scene. Use elegant typography, accurate spelling, cinematic lighting, smooth transitions, and perfectly readable text.  0:00–0:03 | “Every great change”  Begin with a completely black screen.  A tiny point of warm light slowly appears in the center, like the first spark of an idea.  The words “Every great change” emerge softly from the darkness, one word at a time. Use thin, elegant serif typography with wide letter spacing.  “Every” fades in gently.  “Great” grows slightly larger.  “Change” forms from small drifting particles that gather into solid letters.  Keep the scene quiet, minimal, and mysterious.  0:03–0:06 | “begins quietly,”  The camera slowly moves closer to the text.  The previous words shrink and reposition toward the upper-left corner as the phrase “begins quietly,” appears in delicate lowercase letters.  Animate the phrase as though it is being written by an invisible hand.  Each letter should create a subtle ripple in the darkness.  Introduce faint textures, soft shadows, floating dust, and gentle light rays.  The comma should appear last and create a small circular pulse.  0:06–0:09 | “grows through courage,”  The pulse expands and transforms the scene from darkness into a rich sunrise gradient with deep orange, red, and golden tones.  The words “grows through courage” rise upward from the bottom of the frame.  Animate “grows” by gradually increasing its size and weight.  Animate “through” along a curved path.  Animate “courage” in bold uppercase letters that push through a translucent barrier, causing it to crack into geometric fragments.  The movement should feel powerful but controlled.  0:09–0:12 | “and becomes”  The fragments rotate in slow motion and reorganize into a clean editorial grid.  The phrase “AND BECOMES” appears across the frame in condensed sans-serif typography.  Animate the letters with fast tracking changes, vertical stretching, masking, and perspective movement.  The camera accelerates forward through the center of the word “BECOMES.”  The sound and visual energy should steadily build.  0:12–0:14 | “impossible to ignore.”  Reveal a vast bright space filled with light, moving shapes, and large-scale typography.  The words “IMPOSSIBLE TO IGNORE” appear one after another.  “IMPOSSIBLE” expands beyond the edges of the screen.  “TO” remains small and perfectly centered.  “IGNORE” slams into place with strong visual impact, briefly shaking the surrounding grid and shapes.  Use bold contrast, dramatic scale, sharp shadows, and synchronized motion.  0:14–0:15 | Final quote  All movement stops instantly.  The complete quote appears centered on a clean off-white background:  “Every great change begins quietly, grows through courage, and becomes impossible to ignore.”  Use refined black typography with “change,” “courage,” and “impossible” highlighted in deep red.  Hold the final composition clearly for the last second.  Maintain one continuous visual journey from darkness to light, silence to impact, and simplicity to complexity. Keep every phrase connected through visual transformations rather than hard cuts.  Use realistic motion blur, precise kerning, clean masks, stable letterforms, smooth camera movement, subtle film grain, cinematic sound design, rising ambient music, soft particles, controlled color transitions, and a final deep impact sound.  Avoid misspelled words, warped letters, duplicated characters, unreadable text, random symbols, excessive flickering, chaotic layouts, inconsistent fonts.",
    "engagement": {
      "replies": 37,
      "reposts": 21,
      "likes": 339,
      "views": 22196,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083875116900024386",
    "title": "动漫音乐录像的歌词排版",
    "titleEn": "Lyric Typography for an Anime Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "动漫音乐录像的歌词排版的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 16 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for lyric typography for an anime performance on X. Native-video output: 16s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "使用画像はリプ欄にあります\n                                                                      \n高品質アニメ映像。\n\n【作品ジャンル・世界観】\n夢幻的ファンタジーの植物温室と音楽標本世界。劇場版クラスの高密度な2D手描きアニメ、繊細なキーアニメーション、上質なコンポジット、透明感のあるライティング。ガラス温室、花、露、譜線、音符、細い植物線画が一つの視覚言語として連動する。前半は白、淡いシアン、空色、少量の金色による朝の記憶。終盤は同じ植物世界が深い青緑とブルーベルの余韻へ変わる。背景が変わっても、細い白文字と水平譜線が画面の建築構造として連続する。\n表情変化は、目、次に眉、次にまぶた、次に口元、最後に視線の順で最小限に行う。大きな演技、会話、歌唱口形は入れず、静かな呼吸と瞳の変化だけで感情を見せる。\n\n【参照役割と優先順位】\nImage1は顔と同一性の最優先参照であり、同時に衣装・装飾・体型・全身デザインの唯一参照。Image1の背景、背景内小物、元の縦構図は参照しない。参照内容が食い違う場合の優先順位は、顔、瞳、髪、体型、衣装、装飾、全身シルエットは常にImage1が最優先。Image2、Image3、Image4から人物特徴を取得しない。\n\nImage2は冒頭の明るく霞んだガラス温室、乳白色の逆光、淡い水色と象牙色、既存タイトル「Lumen low」、五線譜、音符、細いイタリック注釈の画面設計参照。\nImage3は中盤の明るく広い温室、青い花の密度、より鮮明な奥行き、既存タイトル「Lumen slow」、横断する五線譜、点、円、細い幾何学線の画面設計参照。\nImage4は終盤の深い青緑のブルーベル、柔らかなボケ、露の反射、既存タイトル「BLUEBELL AFTERGLOW」、白と淡いシアンの譜線、点、注釈の画面設計参照。\nAudio1は全編の音楽参照。画面の切替、タイトルの出現と消去、譜線の伸長、カメラの開始と停止、瞬き、光量変化を、Audio1のフレーズ開始、音量の立ち上がり、余韻へ同期する。Audio1以外の音楽、声を追加しない。効果音なし。\n人物は全編でImage1の一人だけ。追加人物、別人、双子、複製、残像人物、透明な人物の重複を出さない。\n\n【キャラクター固定・同一人物】\n全編でImage1と同じ少女を完全な同一人物として固定する。繊細で丸みのある同じ顔、同じ顎と頬、淡い血色、同じ年齢感。透明な青緑色の大きな瞳、虹彩内の金色の微粒子、白い星状ハイライト、長い上まつ毛を維持する。真珠色の長い髪、柔らかな前髪、側頭部から後方へ続く編み込み、後方へまとめた髪型を固定する。右側の金色リボンと青い雫型の髪飾りは消さない。\n衣装はImage1と同じ白いハイネックの長袖ドレス。襟、胸当て、袖口、裾の淡い水色の花刺繍、白い小ボタン、柔らかな幅広袖、金色の細い腰リボンと結び目、淡い水色の裾フリルまで一致させる。細身で上品な体格、白・水色・金色の固有配色を変えない。武器なし、装備品なし、小物なし。両手は常に自然な五本指で、顔や文字を隠さない。\n変化を許すのは呼吸、瞬き、表情、姿勢、自然な髪、リボン、袖、裾の小さな揺れだけ。別人化、平均的なAI顔、幼児化、瞳色変更、髪型変更、衣装交換、刺繍省略、髪飾り消失、手指変形、余分な指、人物複製を禁止する。この同一人物固定は、次項の画風固定とは別の独立条件であり、互いに代用しない。\n【画風固定：2D手描き】\nImage1の極細で淡い青灰色の線、黒すぎない輪郭線、細かな髪束を維持する。影は透明感のある二段から三段。肌と布は柔らかな水彩階調、髪は真珠色の細いハイライト、瞳、露、ガラスは小さく鋭い反射。人物だけを過度に平坦化せず、温室、ガラス骨組み、青い花、葉、露、布、刺繍まで同じ密度の2D手描き背景として描く。\n\n前半の肌は暖かな象牙色を保ち、青い環境光で灰色や病的な青肌にしない。終盤へ向けて環境の色温度が連続的に低くなり、深い青緑へ移行するが、Image1の白いドレス、金色リボン、顔、瞳は背景から明瞭に分離する。\n\n太い黒輪郭、簡略化されたテレビアニメ調、単層の平坦なセル塗り、低密度背景、滑らかなCG・3D化、写実・半写実化、プラスチック肌、過剰な美肌ぼかし、くすんだ色、汎用的なAIアニメ顔、作風混合を禁止する。\n【タイトル・譜線・注釈の固定】\nImage2、Image3、Image4に存在するタイトルを、参照画像由来の不変な2Dグラフィック面として維持する。文字を新しく再描画したり別の字形へ変形したりせず、参照内の正しい字形をクリップマスク、透明度、単語レイヤー単位の短い移動だけで見せる。綴り、細い書体、文字間、大小関係を変えない。画面に残す主文字は「Lumen low」「Lumen slow」「BLUEBELL AFTERGLOW」の三種類だけ。偽の単語、擬似文字、無意味な追加文章を生成しない。\n\n主文字は白から淡い象牙色、細いサンセリフ、広い字間。補助要素は細い高コントラストのイタリック、四本から五本の水平譜線、音符、点、円、短い矢印、極細の植物輪郭線。譜線は約一ピクセル、低めの不透明度。主文字は背景から読める明度を保つ。すべて画面固定座標に置き、カメラの遠近やパースへ吸着させない。\n\n出現は必ず、譜線が左から右へ伸びる、点と音符が現れる、主文字が単語単位の短い左向きクリップマスクとフェードで組み上がる、補助注釈が最後に薄く現れる、の順。全文同時表示や一文字ずつの派手なタイプ入力にはしない。消去は補助注釈、主文字、譜線の順。文字の跳ね、回転、震え、液体化、融解を入れない。\n\n字幕箱、黒帯、太い縁取り、強いグロー、大きな影、カラオケ式の色塗り、RGB分離、激しいグリッチを禁止する。タイトルと譜線は瞳、鼻、口、両手を横断しない。人物の顔と重なる場合は、文字を変形せず、レイアウト全体を顔の周囲へ分割して窓枠のように配置する。\n【冒頭の明るい温室】\nImage2と同じ霞んだガラス温室の正面ワイドから始める。開始画面では人物を出さず、Image2の背景、乳白色の光、青い花、ガラス骨組みを安定して見せる。カメラは目の高さでほぼ固定し、温室中央のガラス枠を消失軸に、ごくゆっくり前進する。目的は「Lumen low」の全体と柔らかな温室の奥行きを一度に読ませること。\n\n五線譜が左から右へ伸び、点と音符が順に現れ、「Lumen low」が参照内の配置で組み上がる。タイトル完成後も文字は画面へ固定する。続けて右三分の一に、細い白線で囲まれた縦長のガラス窓マスクが静かに開く。窓の内側だけにImage1の胸上を一人で現し、印刷された静止画ではなく、呼吸、髪先、金色リボンの微動を持つ生きた2Dアニメとして見せる。Image1は口を閉じたまま一度だけ自然に瞬きし、わずかに正面へ目を合わせる。\n\n【譜線によるImage2からImage3への遷移】\n一本の長い譜線が画面左端から右端へ進み、その通過領域だけを硬い水平マスクとしてImage2からImage3へ置き換える。これはカメラ移動ではなく編集遷移。途中で二つの背景を半透明に混ぜず、マスクの前方はImage2、後方はImage3として境界を明瞭に保つ。Image1も一人のまま受け渡し、二重像や別ポーズを同時に出さない。\n\n「Lumen low」の字形を変形させない。「low」は一本の譜線に覆われて見えなくなり、その同じ領域からImage3に存在する正しい「slow」を明かす。「Lumen」は短時間だけ同じ画面位置に残り、その後Image3の参照配置へ穏やかに整列する。文字を溶かして別単語へ変形させない。\n【中盤の鮮明な温室】\nImage3の明るく広い温室へ完全に切り替える。カメラは目の高さ、少し左寄りの正面ワイド。温室中央通路を奥行き軸にし、Image1を右三分の一の腰上で読める大きさに配置する。中央の「Lumen slow」、横断譜線、青い花、ガラス器、白い机の余白関係をImage3に合わせる。目的は、Image1の顔と衣装を読み取れる大きさにしながら、タイトルを画面構造として維持すること。\n\nカメラはImage1へ短く一定速度で前進し、わずかな前景花の視差を出す。Image1は先に定めた目から視線までの順序で、ごく小さく表情を変える。口を開かず、最後に青緑色の瞳を正面へ向ける。タイトル、譜線、顔の三つが同時に読める位置でカメラを止める。\n\n【譜線と花弁によるImage3からImage4への遷移】\n最初に小さなイタリック注釈と点が薄く消え、続いて「Lumen slow」が短くぼけて消える。文字より遅れて五線譜が画面端まで長く伸びる。温室の白い光が淡い青から深い青緑へ変わり、前景の一枚のブルーベル花弁がレンズへ近づいて画面を覆う。花弁の輪郭と長い水平譜線を一体の編集マスクにし、覆われた領域からImage4の世界を明かす。ランダムなクロスフェードや渦回転は入れない。\n\n【深いブルーベルの終盤】\nImage4の深い青緑の花園へ完全に切り替える。最初は高い位置からの斜め俯瞰で、露を帯びたブルーベルと柔らかな前後ボケを画面いっぱいに置く。背景は柔らかく、タイトルだけは鮮明。目的は花の密度と大きなタイトル余白を同時に見せること。\n\n白と淡いシアンの長い譜線を先に伸ばし、Image4の正しい「BLUEBELL」を左から中央、「AFTERGLOW」を中央から右へ、単語レイヤー単位のクリップマスクで組み上げる。最後の単語は字形を再生成せず、参照内の文字レイヤーを左から順に露出して完成させる。完成後のタイトルは以後の背景切替でも同じ位置に固定する。\nタイトルを動かさず、花の間に横たわるImage1を斜め構図で明かす。少女の身体は左下から右上へ伸び、顔は右上。白いドレス、淡い水色の刺繍、金色リボンを深い青い花から明瞭に分離する。カメラは左上からの緩い俯瞰で一定速度に前進し、身体の対角線を青緑色の瞳へ導く。目的は全身の斜線、顔、タイトル余白を一つの構図で読ませること。花は風でわずかに揺れるが、少女の身体を覆わない。\n\n続けて花背景だけを暗く柔らかくぼかし、白い長方形の窓パネルを開く。その内側にImage1の顔の極端な正面近景を一人だけ見せる。カメラは目の高さ、鼻筋と両目中央を通る厳密な正面軸。目的はImage1の青緑色の瞳、虹彩の金色粒子、微細な呼吸を明瞭な顔報酬として見せること。窓枠、譜線、「BLUEBELL AFTERGLOW」は目、鼻、口を横切らず、顔の周囲を建築的に囲う。Image1は一度ゆっくり瞬きし、わずかに息を吸い、瞳を開いて正面を見る。\n\n同じ顔、タイトル、中央軸を保ったまま、カメラが滑らかに後退する。窓パネルの外側が黒から深いブルーベルの花園へ変わり、Image1が花に囲まれて座る腰上構図を明かす。両手は自然に膝の上で重なり、指は正しい本数で静止する。白いドレスと顔は明瞭、花園は露と柔らかなボケを保ち、「BLUEBELL AFTERGLOW」は画面固定で完全に読める。\n最後に一本の細い水平線だけが終点まで伸びる。Image1は正面を見て瞳を開いたまま完全に静止し、髪先、金色リボン、袖、周囲の花弁の最後の小さな揺れが順に収まる。水平線が終点へ到達し、すべての動きが止まり、顔、両手、タイトル、ブルーベルが同時に読める瞬間に切る。\n\n【全体制御】\nカメラは一定で滑らかにし、不要な手持ち揺れ、急加速、急停止、ランダムズーム、無意味な回転、派手なオービットを入れない。タイトル変化を物理カメラ運動として扱わない。参照間で人物の顔、衣装、花の種類、色、タイトルの綴りを混ぜない。人物を長時間文字や花で隠さず、各区間の顔報酬を明瞭に保つ。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2083875116900024386",
    "sourceUrl": "https://x.com/haruuraeadss/status/2083875116900024386",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-08-02T11:18:39.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2083875116900024386.mp4",
    "posterUrl": "/posters/x/2083875116900024386.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Mankyuさんのワークフローすごすぎて、私もお借りして試してみたら、\nもう楽しすぎて寝不足です～✨\n\n本格的なMVでよく見るリリックタイポグラフィの表現が簡単にできて、びっくり！！\n#MiniMaxH3 の登場でMV作りにハマっちゃいそう🎉\n@Hailuo_AI　\nプロンプト👇　使用画像はリプ欄にあります\n                                                                      \n高品質アニメ映像。\n\n【作品ジャンル・世界観】\n夢幻的ファンタジーの植物温室と音楽標本世界。劇場版クラスの高密度な2D手描きアニメ、繊細なキーアニメーション、上質なコンポジット、透明感のあるライティング。ガラス温室、花、露、譜線、音符、細い植物線画が一つの視覚言語として連動する。前半は白、淡いシアン、空色、少量の金色による朝の記憶。終盤は同じ植物世界が深い青緑とブルーベルの余韻へ変わる。背景が変わっても、細い白文字と水平譜線が画面の建築構造として連続する。\n表情変化は、目、次に眉、次にまぶた、次に口元、最後に視線の順で最小限に行う。大きな演技、会話、歌唱口形は入れず、静かな呼吸と瞳の変化だけで感情を見せる。\n\n【参照役割と優先順位】\nImage1は顔と同一性の最優先参照であり、同時に衣装・装飾・体型・全身デザインの唯一参照。Image1の背景、背景内小物、元の縦構図は参照しない。参照内容が食い違う場合の優先順位は、顔、瞳、髪、体型、衣装、装飾、全身シルエットは常にImage1が最優先。Image2、Image3、Image4から人物特徴を取得しない。\n\nImage2は冒頭の明るく霞んだガラス温室、乳白色の逆光、淡い水色と象牙色、既存タイトル「Lumen low」、五線譜、音符、細いイタリック注釈の画面設計参照。\nImage3は中盤の明るく広い温室、青い花の密度、より鮮明な奥行き、既存タイトル「Lumen slow」、横断する五線譜、点、円、細い幾何学線の画面設計参照。\nImage4は終盤の深い青緑のブルーベル、柔らかなボケ、露の反射、既存タイトル「BLUEBELL AFTERGLOW」、白と淡いシアンの譜線、点、注釈の画面設計参照。\nAudio1は全編の音楽参照。画面の切替、タイトルの出現と消去、譜線の伸長、カメラの開始と停止、瞬き、光量変化を、Audio1のフレーズ開始、音量の立ち上がり、余韻へ同期する。Audio1以外の音楽、声を追加しない。効果音なし。\n人物は全編でImage1の一人だけ。追加人物、別人、双子、複製、残像人物、透明な人物の重複を出さない。\n\n【キャラクター固定・同一人物】\n全編でImage1と同じ少女を完全な同一人物として固定する。繊細で丸みのある同じ顔、同じ顎と頬、淡い血色、同じ年齢感。透明な青緑色の大きな瞳、虹彩内の金色の微粒子、白い星状ハイライト、長い上まつ毛を維持する。真珠色の長い髪、柔らかな前髪、側頭部から後方へ続く編み込み、後方へまとめた髪型を固定する。右側の金色リボンと青い雫型の髪飾りは消さない。\n衣装はImage1と同じ白いハイネックの長袖ドレス。襟、胸当て、袖口、裾の淡い水色の花刺繍、白い小ボタン、柔らかな幅広袖、金色の細い腰リボンと結び目、淡い水色の裾フリルまで一致させる。細身で上品な体格、白・水色・金色の固有配色を変えない。武器なし、装備品なし、小物なし。両手は常に自然な五本指で、顔や文字を隠さない。\n変化を許すのは呼吸、瞬き、表情、姿勢、自然な髪、リボン、袖、裾の小さな揺れだけ。別人化、平均的なAI顔、幼児化、瞳色変更、髪型変更、衣装交換、刺繍省略、髪飾り消失、手指変形、余分な指、人物複製を禁止する。この同一人物固定は、次項の画風固定とは別の独立条件であり、互いに代用しない。\n【画風固定：2D手描き】\nImage1の極細で淡い青灰色の線、黒すぎない輪郭線、細かな髪束を維持する。影は透明感のある二段から三段。肌と布は柔らかな水彩階調、髪は真珠色の細いハイライト、瞳、露、ガラスは小さく鋭い反射。人物だけを過度に平坦化せず、温室、ガラス骨組み、青い花、葉、露、布、刺繍まで同じ密度の2D手描き背景として描く。\n\n前半の肌は暖かな象牙色を保ち、青い環境光で灰色や病的な青肌にしない。終盤へ向けて環境の色温度が連続的に低くなり、深い青緑へ移行するが、Image1の白いドレス、金色リボン、顔、瞳は背景から明瞭に分離する。\n\n太い黒輪郭、簡略化されたテレビアニメ調、単層の平坦なセル塗り、低密度背景、滑らかなCG・3D化、写実・半写実化、プラスチック肌、過剰な美肌ぼかし、くすんだ色、汎用的なAIアニメ顔、作風混合を禁止する。\n【タイトル・譜線・注釈の固定】\nImage2、Image3、Image4に存在するタイトルを、参照画像由来の不変な2Dグラフィック面として維持する。文字を新しく再描画したり別の字形へ変形したりせず、参照内の正しい字形をクリップマスク、透明度、単語レイヤー単位の短い移動だけで見せる。綴り、細い書体、文字間、大小関係を変えない。画面に残す主文字は「Lumen low」「Lumen slow」「BLUEBELL AFTERGLOW」の三種類だけ。偽の単語、擬似文字、無意味な追加文章を生成しない。\n\n主文字は白から淡い象牙色、細いサンセリフ、広い字間。補助要素は細い高コントラストのイタリック、四本から五本の水平譜線、音符、点、円、短い矢印、極細の植物輪郭線。譜線は約一ピクセル、低めの不透明度。主文字は背景から読める明度を保つ。すべて画面固定座標に置き、カメラの遠近やパースへ吸着させない。\n\n出現は必ず、譜線が左から右へ伸びる、点と音符が現れる、主文字が単語単位の短い左向きクリップマスクとフェードで組み上がる、補助注釈が最後に薄く現れる、の順。全文同時表示や一文字ずつの派手なタイプ入力にはしない。消去は補助注釈、主文字、譜線の順。文字の跳ね、回転、震え、液体化、融解を入れない。\n\n字幕箱、黒帯、太い縁取り、強いグロー、大きな影、カラオケ式の色塗り、RGB分離、激しいグリッチを禁止する。タイトルと譜線は瞳、鼻、口、両手を横断しない。人物の顔と重なる場合は、文字を変形せず、レイアウト全体を顔の周囲へ分割して窓枠のように配置する。\n【冒頭の明るい温室】\nImage2と同じ霞んだガラス温室の正面ワイドから始める。開始画面では人物を出さず、Image2の背景、乳白色の光、青い花、ガラス骨組みを安定して見せる。カメラは目の高さでほぼ固定し、温室中央のガラス枠を消失軸に、ごくゆっくり前進する。目的は「Lumen low」の全体と柔らかな温室の奥行きを一度に読ませること。\n\n五線譜が左から右へ伸び、点と音符が順に現れ、「Lumen low」が参照内の配置で組み上がる。タイトル完成後も文字は画面へ固定する。続けて右三分の一に、細い白線で囲まれた縦長のガラス窓マスクが静かに開く。窓の内側だけにImage1の胸上を一人で現し、印刷された静止画ではなく、呼吸、髪先、金色リボンの微動を持つ生きた2Dアニメとして見せる。Image1は口を閉じたまま一度だけ自然に瞬きし、わずかに正面へ目を合わせる。\n\n【譜線によるImage2からImage3への遷移】\n一本の長い譜線が画面左端から右端へ進み、その通過領域だけを硬い水平マスクとしてImage2からImage3へ置き換える。これはカメラ移動ではなく編集遷移。途中で二つの背景を半透明に混ぜず、マスクの前方はImage2、後方はImage3として境界を明瞭に保つ。Image1も一人のまま受け渡し、二重像や別ポーズを同時に出さない。\n\n「Lumen low」の字形を変形させない。「low」は一本の譜線に覆われて見えなくなり、その同じ領域からImage3に存在する正しい「slow」を明かす。「Lumen」は短時間だけ同じ画面位置に残り、その後Image3の参照配置へ穏やかに整列する。文字を溶かして別単語へ変形させない。\n【中盤の鮮明な温室】\nImage3の明るく広い温室へ完全に切り替える。カメラは目の高さ、少し左寄りの正面ワイド。温室中央通路を奥行き軸にし、Image1を右三分の一の腰上で読める大きさに配置する。中央の「Lumen slow」、横断譜線、青い花、ガラス器、白い机の余白関係をImage3に合わせる。目的は、Image1の顔と衣装を読み取れる大きさにしながら、タイトルを画面構造として維持すること。\n\nカメラはImage1へ短く一定速度で前進し、わずかな前景花の視差を出す。Image1は先に定めた目から視線までの順序で、ごく小さく表情を変える。口を開かず、最後に青緑色の瞳を正面へ向ける。タイトル、譜線、顔の三つが同時に読める位置でカメラを止める。\n\n【譜線と花弁によるImage3からImage4への遷移】\n最初に小さなイタリック注釈と点が薄く消え、続いて「Lumen slow」が短くぼけて消える。文字より遅れて五線譜が画面端まで長く伸びる。温室の白い光が淡い青から深い青緑へ変わり、前景の一枚のブルーベル花弁がレンズへ近づいて画面を覆う。花弁の輪郭と長い水平譜線を一体の編集マスクにし、覆われた領域からImage4の世界を明かす。ランダムなクロスフェードや渦回転は入れない。\n\n【深いブルーベルの終盤】\nImage4の深い青緑の花園へ完全に切り替える。最初は高い位置からの斜め俯瞰で、露を帯びたブルーベルと柔らかな前後ボケを画面いっぱいに置く。背景は柔らかく、タイトルだけは鮮明。目的は花の密度と大きなタイトル余白を同時に見せること。\n\n白と淡いシアンの長い譜線を先に伸ばし、Image4の正しい「BLUEBELL」を左から中央、「AFTERGLOW」を中央から右へ、単語レイヤー単位のクリップマスクで組み上げる。最後の単語は字形を再生成せず、参照内の文字レイヤーを左から順に露出して完成させる。完成後のタイトルは以後の背景切替でも同じ位置に固定する。\nタイトルを動かさず、花の間に横たわるImage1を斜め構図で明かす。少女の身体は左下から右上へ伸び、顔は右上。白いドレス、淡い水色の刺繍、金色リボンを深い青い花から明瞭に分離する。カメラは左上からの緩い俯瞰で一定速度に前進し、身体の対角線を青緑色の瞳へ導く。目的は全身の斜線、顔、タイトル余白を一つの構図で読ませること。花は風でわずかに揺れるが、少女の身体を覆わない。\n\n続けて花背景だけを暗く柔らかくぼかし、白い長方形の窓パネルを開く。その内側にImage1の顔の極端な正面近景を一人だけ見せる。カメラは目の高さ、鼻筋と両目中央を通る厳密な正面軸。目的はImage1の青緑色の瞳、虹彩の金色粒子、微細な呼吸を明瞭な顔報酬として見せること。窓枠、譜線、「BLUEBELL AFTERGLOW」は目、鼻、口を横切らず、顔の周囲を建築的に囲う。Image1は一度ゆっくり瞬きし、わずかに息を吸い、瞳を開いて正面を見る。\n\n同じ顔、タイトル、中央軸を保ったまま、カメラが滑らかに後退する。窓パネルの外側が黒から深いブルーベルの花園へ変わり、Image1が花に囲まれて座る腰上構図を明かす。両手は自然に膝の上で重なり、指は正しい本数で静止する。白いドレスと顔は明瞭、花園は露と柔らかなボケを保ち、「BLUEBELL AFTERGLOW」は画面固定で完全に読める。\n最後に一本の細い水平線だけが終点まで伸びる。Image1は正面を見て瞳を開いたまま完全に静止し、髪先、金色リボン、袖、周囲の花弁の最後の小さな揺れが順に収まる。水平線が終点へ到達し、すべての動きが止まり、顔、両手、タイトル、ブルーベルが同時に読める瞬間に切る。\n\n【全体制御】\nカメラは一定で滑らかにし、不要な手持ち揺れ、急加速、急停止、ランダムズーム、無意味な回転、派手なオービットを入れない。タイトル変化を物理カメラ運動として扱わない。参照間で人物の顔、衣装、花の種類、色、タイトルの綴りを混ぜない。人物を長時間文字や花で隠さず、各区間の顔報酬を明瞭に保つ。",
    "engagement": {
      "replies": 6,
      "reposts": 4,
      "likes": 71,
      "views": 12014,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083855654205186452",
    "title": "二千画质与七百二十画质的对决",
    "titleEn": "A 2K-versus-720P Generation Face-Off",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "二千画质与七百二十画质的对决的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1080×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a 2k-versus-720p generation face-off on X. Native-video output: 15s · 1080×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second photorealistic cinematic dark-fantasy sequence in 16:9 widescreen. The video must contain exactly three clearly separated cinematic shots connected by clean hard cuts. No dissolves, morphing transitions or continuous camera orbit.\n\nMASTER DRAGON LOCK:\nShow one enormous ancient dragon with a broad powerful body, charcoal-black overlapping scales, subtle dark crimson colouring across its head and neck, two swept-back horns, smaller facial spikes, glowing amber-orange eyes, one long tail and one pair of enormous bat-like wings with dark leathery membranes. The dragon has four legs: two forelegs and two hind legs. Its design, size, horns, colour, anatomy and facial structure remain identical in all three shots. The dragon must feel extremely heavy, ancient and physically realistic.\n\nMASTER RIDER LOCK:\nOne adult armoured rider is already seated securely in a dark leather saddle on the dragon from the very first frame. The rider wears detailed weathered silver plate armour, dark under-clothing, a fitted medieval helmet with a narrow shadowed visor and a long dark cloak. The rider’s face is never clearly visible. The rider holds leather reins connected to the saddle and remains seated throughout the entire sequence. The rider does not suddenly appear, mount, dismount or change clothing. During the first shot, the rider is naturally hidden behind the dragon’s large head, neck and folded wings.\n\nENVIRONMENT AND LIGHTING:\nThe sequence takes place inside a gigantic ancient mountain cave. The cave contains rough black rock walls, enormous stone formations, drifting mist, scattered bones and faint glowing embers between cracks in the ground. The far interior is almost black, but the dragon must remain readable through soft amber reflections and cold blue rim light entering from the cave mouth. The cave entrance opens from a high mountain cliff into a vast dramatic landscape of jagged mountains, deep valleys, distant waterfalls, rolling clouds and pale stormy daylight. Maintain a cinematic cold-blue and muted-amber colour palette with deep shadows, volumetric light rays and realistic atmospheric haze. Dark and mysterious, but never so underexposed that the dragon or rider becomes blurry or invisible.\n\nSHOT 1 — DRAGON EMERGES, 0–5 SECONDS:\nBegin with a symmetrical low frontal wide shot positioned near the cave entrance, facing deep into the darkness. The camera slowly tracks backward toward the entrance as the enormous dragon walks forward out of the black cave interior. The dragon’s wings remain completely folded against both sides of its body. Its glowing amber eyes appear first, followed by its horned head, shoulders and powerful forelegs becoming visible through the mist.\n\nEach heavy footstep shakes the cave floor and dislodges small stones and dust from the ceiling. The dragon exhales dense vapour from its nostrils and releases a deep restrained growl. Its head moves naturally with the weight of each step. The dragon looks directly past the camera toward the cave opening. The rider is already seated but remains hidden behind the dragon’s head and neck. Do not reveal the rider clearly during this shot.\n\nSHOT 2 — RIDER REVEAL, 5–9 SECONDS:\nClean hard cut to a low three-quarter side tracking shot beside the moving dragon. Begin close to the dragon’s horned head and sharply detailed scales. Track smoothly backward along the dragon’s neck and folded wing while the dragon continues walking toward the cave entrance.\n\nAs the camera reaches the dragon’s shoulder, reveal the single armoured rider seated in the saddle between the folded wings. Hold the composition long enough to clearly show the rider’s silver armour, helmet, dark cloak and both hands gripping the reins. The rider looks toward the bright cave exit and leans slightly forward in preparation for takeoff. The dragon and rider remain in continuous forward motion. Do not show a separate rider anywhere else in the cave.\n\nSHOT 3 — REAR TAKEOFF AND FLIGHT, 9–15 SECONDS:\nClean hard cut to an epic wide rear view positioned behind and slightly above the dragon and rider. The bright cave mouth and vast mountain landscape fill the background ahead of them. The dragon accelerates into a powerful run toward the cave opening while its wings remain folded enough to pass safely through the cave.\n\nThe dragon reaches the cliff edge and makes one forceful leap completely out of the cave. Only after clearing the rock opening, it spreads both enormous wings to their full symmetrical width. The sudden wing expansion pushes mist, dust and loose stones outward from the cave entrance. The rider leans forward, grips the reins and remains firmly attached to the saddle.\n\nThe dragon performs one enormous downward wingbeat, catches the air and transitions naturally into stable forward flight. Loose stones fall from the cliff beneath it. The camera rapidly follows from behind as the dragon and rider soar away over the deep valley between jagged mountain peaks. The long tail streams naturally behind the dragon and the rider’s cloak whips in the wind. End on a majestic wide rear silhouette of the dragon flying into the open mountain landscape as sunlight breaks through the storm clouds.\n\nCAMERA AND MOTION:\nPremium cinematic camera movement with realistic scale and parallax. Shot one is a controlled backward dolly. Shot two is a smooth low side-tracking reveal. Shot three is a dynamic rear pursuit shot that follows the takeoff out of the cave. Use natural motion blur only during the leap and wingbeat. Keep the dragon’s movement heavy on the ground and powerful but graceful in the air. No shaky chaotic framing and no extreme close-ups that obscure the action.\n\nCINEMATIC QUALITY:\nBig-budget photorealistic dark-fantasy visual effects, highly detailed scales and armour, realistic creature muscles beneath the skin, physically accurate wing membranes, believable foot contact, convincing dust and falling-rock physics, volumetric cave mist, dramatic rim lighting, strong depth, atmospheric perspective and stable character consistency.\n\nAUDIO:\nNo dialogue and no music. Deep dragon breathing, restrained growl, heavy echoing footsteps, claws scraping stone, armour movement, leather saddle creaking, falling pebbles, rushing cave wind, powerful wingbeat and a deep dragon roar as it enters open air.\n\nSTRICT CONSISTENCY:\nExactly one dragon and one rider. The rider is seated on the dragon from the beginning and never disappears. No additional people, dragons or creatures. No dragon transformation or colour change. No extra heads, horns, tails, wings, legs or riders. The wings remain folded inside the cave and open only after the dragon has completely cleared the cave mouth. The wings must not clip through the cave walls. The rider must not float above the saddle or merge into the dragon. No fire breathing, combat, dialogue, text, logos, blood or visible injuries.",
    "promptSourceUrl": "https://x.com/MadMax_Series/status/2083855654205186452",
    "sourceUrl": "https://x.com/MadMax_Series/status/2083855654205186452",
    "sourceLabel": "X 原帖 · @MadMax_Series",
    "author": "MadMax",
    "publishedAt": "2026-08-02T10:01:18.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2083855654205186452.mp4",
    "posterUrl": "/posters/x/2083855654205186452.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo MiniMax H3 vs. Seedance 2.0   \n\nExperimented more for comparisons, i am loving MiniMax H3. 2K resolution videos for 4 times cheaper than seedance 2.0 is too hot to resist. Thank you @Hailuo_AI  for making it easier to creators. \n\nIts not just cheap, in some cases it is also performing way better than seedance 2.0 in terms of visuals and framing. One example is given below.\n\nUsed same Text to Video Prompt\n\nPrompt: \nCreate a 15-second photorealistic cinematic dark-fantasy sequence in 16:9 widescreen. The video must contain exactly three clearly separated cinematic shots connected by clean hard cuts. No dissolves, morphing transitions or continuous camera orbit.\n\nMASTER DRAGON LOCK:\nShow one enormous ancient dragon with a broad powerful body, charcoal-black overlapping scales, subtle dark crimson colouring across its head and neck, two swept-back horns, smaller facial spikes, glowing amber-orange eyes, one long tail and one pair of enormous bat-like wings with dark leathery membranes. The dragon has four legs: two forelegs and two hind legs. Its design, size, horns, colour, anatomy and facial structure remain identical in all three shots. The dragon must feel extremely heavy, ancient and physically realistic.\n\nMASTER RIDER LOCK:\nOne adult armoured rider is already seated securely in a dark leather saddle on the dragon from the very first frame. The rider wears detailed weathered silver plate armour, dark under-clothing, a fitted medieval helmet with a narrow shadowed visor and a long dark cloak. The rider’s face is never clearly visible. The rider holds leather reins connected to the saddle and remains seated throughout the entire sequence. The rider does not suddenly appear, mount, dismount or change clothing. During the first shot, the rider is naturally hidden behind the dragon’s large head, neck and folded wings.\n\nENVIRONMENT AND LIGHTING:\nThe sequence takes place inside a gigantic ancient mountain cave. The cave contains rough black rock walls, enormous stone formations, drifting mist, scattered bones and faint glowing embers between cracks in the ground. The far interior is almost black, but the dragon must remain readable through soft amber reflections and cold blue rim light entering from the cave mouth. The cave entrance opens from a high mountain cliff into a vast dramatic landscape of jagged mountains, deep valleys, distant waterfalls, rolling clouds and pale stormy daylight. Maintain a cinematic cold-blue and muted-amber colour palette with deep shadows, volumetric light rays and realistic atmospheric haze. Dark and mysterious, but never so underexposed that the dragon or rider becomes blurry or invisible.\n\nSHOT 1 — DRAGON EMERGES, 0–5 SECONDS:\nBegin with a symmetrical low frontal wide shot positioned near the cave entrance, facing deep into the darkness. The camera slowly tracks backward toward the entrance as the enormous dragon walks forward out of the black cave interior. The dragon’s wings remain completely folded against both sides of its body. Its glowing amber eyes appear first, followed by its horned head, shoulders and powerful forelegs becoming visible through the mist.\n\nEach heavy footstep shakes the cave floor and dislodges small stones and dust from the ceiling. The dragon exhales dense vapour from its nostrils and releases a deep restrained growl. Its head moves naturally with the weight of each step. The dragon looks directly past the camera toward the cave opening. The rider is already seated but remains hidden behind the dragon’s head and neck. Do not reveal the rider clearly during this shot.\n\nSHOT 2 — RIDER REVEAL, 5–9 SECONDS:\nClean hard cut to a low three-quarter side tracking shot beside the moving dragon. Begin close to the dragon’s horned head and sharply detailed scales. Track smoothly backward along the dragon’s neck and folded wing while the dragon continues walking toward the cave entrance.\n\nAs the camera reaches the dragon’s shoulder, reveal the single armoured rider seated in the saddle between the folded wings. Hold the composition long enough to clearly show the rider’s silver armour, helmet, dark cloak and both hands gripping the reins. The rider looks toward the bright cave exit and leans slightly forward in preparation for takeoff. The dragon and rider remain in continuous forward motion. Do not show a separate rider anywhere else in the cave.\n\nSHOT 3 — REAR TAKEOFF AND FLIGHT, 9–15 SECONDS:\nClean hard cut to an epic wide rear view positioned behind and slightly above the dragon and rider. The bright cave mouth and vast mountain landscape fill the background ahead of them. The dragon accelerates into a powerful run toward the cave opening while its wings remain folded enough to pass safely through the cave.\n\nThe dragon reaches the cliff edge and makes one forceful leap completely out of the cave. Only after clearing the rock opening, it spreads both enormous wings to their full symmetrical width. The sudden wing expansion pushes mist, dust and loose stones outward from the cave entrance. The rider leans forward, grips the reins and remains firmly attached to the saddle.\n\nThe dragon performs one enormous downward wingbeat, catches the air and transitions naturally into stable forward flight. Loose stones fall from the cliff beneath it. The camera rapidly follows from behind as the dragon and rider soar away over the deep valley between jagged mountain peaks. The long tail streams naturally behind the dragon and the rider’s cloak whips in the wind. End on a majestic wide rear silhouette of the dragon flying into the open mountain landscape as sunlight breaks through the storm clouds.\n\nCAMERA AND MOTION:\nPremium cinematic camera movement with realistic scale and parallax. Shot one is a controlled backward dolly. Shot two is a smooth low side-tracking reveal. Shot three is a dynamic rear pursuit shot that follows the takeoff out of the cave. Use natural motion blur only during the leap and wingbeat. Keep the dragon’s movement heavy on the ground and powerful but graceful in the air. No shaky chaotic framing and no extreme close-ups that obscure the action.\n\nCINEMATIC QUALITY:\nBig-budget photorealistic dark-fantasy visual effects, highly detailed scales and armour, realistic creature muscles beneath the skin, physically accurate wing membranes, believable foot contact, convincing dust and falling-rock physics, volumetric cave mist, dramatic rim lighting, strong depth, atmospheric perspective and stable character consistency.\n\nAUDIO:\nNo dialogue and no music. Deep dragon breathing, restrained growl, heavy echoing footsteps, claws scraping stone, armour movement, leather saddle creaking, falling pebbles, rushing cave wind, powerful wingbeat and a deep dragon roar as it enters open air.\n\nSTRICT CONSISTENCY:\nExactly one dragon and one rider. The rider is seated on the dragon from the beginning and never disappears. No additional people, dragons or creatures. No dragon transformation or colour change. No extra heads, horns, tails, wings, legs or riders. The wings remain folded inside the cave and open only after the dragon has completely cleared the cave mouth. The wings must not clip through the cave walls. The rider must not float above the saddle or merge into the dragon. No fire breathing, combat, dialogue, text, logos, blood or visible injuries.",
    "engagement": {
      "replies": 8,
      "reposts": 4,
      "likes": 116,
      "views": 14578,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083854803801960837",
    "title": "史诗级太空大战",
    "titleEn": "An Epic Battle in Deep Space",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "史诗级太空大战的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an epic battle in deep space on X. Native-video output: 15s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "- Epic theatrical space-opera teaser\n- (Partial prompt — expand as needed)\n- Keep the pace fast and the scale enormous without letting the edit drag. Use sharp hard cuts, a shaking command deck, white-hot flashes, split-second black frames, and a violent jump-to-warp impact. Title cards should use wide-tracked cinematic typography—not pure white—with restrained material texture, subtle illumination, and a faint edge glow. Animate the titles by emerging from deep-space shadow, catching a sweep of starlight, opening their letter spacing, leaving a slight afterimage, and flashing briefly against black.",
    "promptSourceUrl": "https://x.com/itxabdullaa/status/2083854803801960837",
    "sourceUrl": "https://x.com/itxabdullaa/status/2083854803801960837",
    "sourceLabel": "X 原帖 · @itxabdullaa",
    "author": "ÀBDŪLLÂH",
    "publishedAt": "2026-08-02T09:57:56.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2083854803801960837.mp4",
    "posterUrl": "/posters/x/2083854803801960837.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is redefining AI with open-weight multimodal intelligence—seamlessly understanding and generating across text, images, video, and audio. ✨ \n\nMade this Epic Space Opera Teaser with MiniMax H3 on @Hailuo_AI\n\nPrompt:\n\n- Epic theatrical space-opera teaser\n- (Partial prompt — expand as needed)\n- Keep the pace fast and the scale enormous without letting the edit drag. Use sharp hard cuts, a shaking command deck, white-hot flashes, split-second black frames, and a violent jump-to-warp impact. Title cards should use wide-tracked cinematic typography—not pure white—with restrained material texture, subtle illumination, and a faint edge glow. Animate the titles by emerging from deep-space shadow, catching a sweep of starlight, opening their letter spacing, leaving a slight afterimage, and flashing briefly against black.",
    "engagement": {
      "replies": 73,
      "reposts": 3,
      "likes": 93,
      "views": 1794,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083768289612947872",
    "title": "星花都市的角色登场",
    "titleEn": "A Character Enters the City of Starflowers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "星花都市的角色登场的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a character enters the city of starflowers on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "高品質アニメ映像。\n\n星花魔法都市を舞台にした高級ファンタジーRPGのキャラクター導入シーン。装備設定から街への出発までを、一切途切れないゲーム内カメラで描く。表情が変わる場面は、最初に目、次に眉、続けてまぶた、口元、最後に視線の順で変化する。\n\n【キャラクター参照】\n\n画像1のキャラクターを全編の唯一の主人公として参照し、同一人物として維持する。\n\n画像1は人物設計専用の参照画像とし、顔、目の形、虹彩色、瞳の特徴、髪型、髪色、前髪、衣装、装飾、武器装飾、小物、体格、シルエット、年齢感、雰囲気、キャラクターカラーのみを参照する。\n\n画像1そのものの背景、部屋、家具、ポーズ、画角、フレーミング、文字、レイアウトは再現しない。\n\n表情、視線、口元、姿勢、呼吸、自然な髪と衣装の揺れのみ変化してよい。\n\n特徴の混合、顔の平均化、衣装交換、別人化、クローン化、余分な主人公の追加は禁止。\n\n【キャラクター固定】\n\n主人公は全編で一人のみ。\n\n顔、髪型、衣装、装飾、小物、武器、体格、シルエット、キャラクターカラーを維持する。\n\n武器は全編で一冊だけ存在する魔導書とし、画像1に合わせた材質・色・装飾・デザインを維持する。\n\n開始時は左腰へ装備され、カスタマイズ時のみ左手上へ浮遊し、終了後は左腰へ戻る。\n\n画像1に存在するアクセサリーや装飾品も維持し、不要な複製や消失は発生しない。\n\n【画風】\n\n劇場版品質。\n\n高密度3Dトゥーンセルルック。\n\n二〜三段セルシェーディング。\n\n透明感のあるライティング。\n\n髪、瞳、布、革、金属、宝石、紙を高密度に描き分ける。\n\n色彩は画像1のキャラクターカラーを最優先とし、背景は人物より一段落ち着いた色調で統一する。\n\n画風混在、リアル調、プラスチックCG、低密度背景は禁止。\n\n【ゲームUI】\n\n高級ファンタジーRPG風UI。\n\n半透明クリスタルパネル。\n\n細い金装飾。\n\n魔法陣と星座モチーフ。\n\n文字は読み取れる品質で安定表示。\n\nUIは顔を覆わず画面の空き領域へ配置する。\n\n操作に合わせて光と効果音が同期する。\n\n【シーン】\n\n高い俯瞰から開始。\n\n主人公は魔法陣中央へ座り、カメラを見る。\n\n右側へ\n\nSTART NEW GAME\n\nCONTINUE\n\nARCANE SETTINGS\n\nEXIT\n\n左上へプロフィールを表示。\n\nCONTINUEのみ選択状態。\n\nカーソルが一度移動して決定する。\n\nカメラは右手へ降下。\n\nRIGHT-HAND CATALYSTパネルが表示される。\n\n掌の上へ六枚花弁の魔法術式が形成される。\n\n手の形は変化せず、術式のみ再構成される。\n\n花弁、宝石核、二重金環が完成して停止する。\n\n続いてカメラは左手へ回り込む。\n\nGRIMOIRE CUSTOMIZATIONを表示。\n\nCOVER\n\nCLASP\n\nCORE\n\nPAGES\n\n四項目を表示。\n\n主人公が左手を開くと左腰から魔導書が浮遊する。\n\n閉じた一冊の状態から、表紙、金具、中心核、ページが順番に分解される。\n\n全パーツはルーンで接続されたまま維持される。\n\nカーソルが四項目を巡るたび各パーツが更新され、最後に一冊へ戻る。\n\n革、紙、金属、宝石の質感を高密度に描写する。\n\nカメラは正面へ戻る。\n\nCONFIRM CONFIGを表示。\n\n主人公が一度押す。\n\n全UIが魔導書へ吸収される。\n\n魔導書は左腰へ戻る。\n\n主人公は自然に座り直す。\n\n掌の術式がゆっくり回転して完成する。\n\n表情は目→眉→まぶた→口元→視線の順で自然に変化する。\n\nその後カメラを少し引く。\n\n画面下へLOADINGバー。\n\n0%から100%まで進行する。\n\n紫色の開始空間から石畳、建物、屋台、灯り、街路が順番に実体化する。\n\n線画→素材→完成世界の三段階で生成される。\n\n舞台は雨上がりの星花魔法都市。\n\n濡れた石畳。\n\n魔法露店。\n\n魔法灯。\n\nルーン看板。\n\nローブ姿の群衆。\n\n飛行する箒便。\n\n浮遊馬車。\n\n石造と木造建築。\n\n遠景には巨大な魔法学院。\n\n背景人物は全員別人であり、主人公は複製しない。\n\n最後に三人称追従カメラになる。\n\n主人公は中景サイズを維持する。\n\n右上へ円形ミニマップ。\n\n左下へ\n\nVITALITY\n\nMANA\n\nSPELL SLOT\n\n街の奥へ\n\nRETURN TO ASTRA ACADEMY\n\n任務マーカーを表示。\n\n主人公は左腰へ魔導書を装備したまま街へ歩き出す。\n\n最後は左足が石畳へ着地し、右足が次の一歩へ移る瞬間で終了する。\n\n都市は完全に実体化し続け、群衆、箒便、浮遊馬車は動き続ける。\n\nHUDと任務マーカーは安定表示する。\n\n【音】\n\nBGMなし。\n\n音楽なし。\n\n環境音のみ。\n\n雨上がりの街の空気。\n\n石畳の足音。\n\n衣装の布擦れ。\n\n遠い市場の声。\n\n箒の風切り。\n\n浮遊馬車の魔力音。\n\n主人公は自然な呼吸のみ。\n\n台詞、ナレーション、長い発声は禁止。\n\n【禁止事項】\n\n画像1の背景や構図の再現、特徴の混合、顔の平均化、衣装交換、武器変更、別人化、クローン化、余分な主人公、余分な武器、余分な装飾、手指崩れ、顔へUIが重なる構図、読めないUI文字、急激な背景転換、白飛び、画風変化は禁止。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2083768289612947872",
    "sourceUrl": "https://x.com/haruuraeadss/status/2083768289612947872",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-08-02T04:14:09.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2083768289612947872.mp4",
    "posterUrl": "/posters/x/2083768289612947872.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ゲーム映像\n\n@Hailuo_AI　\n#MiniMaxH3　\nプロンプト👇\n                                                                                                                                                                                                                                          \n高品質アニメ映像。\n\n星花魔法都市を舞台にした高級ファンタジーRPGのキャラクター導入シーン。装備設定から街への出発までを、一切途切れないゲーム内カメラで描く。表情が変わる場面は、最初に目、次に眉、続けてまぶた、口元、最後に視線の順で変化する。\n\n【キャラクター参照】\n\n画像1のキャラクターを全編の唯一の主人公として参照し、同一人物として維持する。\n\n画像1は人物設計専用の参照画像とし、顔、目の形、虹彩色、瞳の特徴、髪型、髪色、前髪、衣装、装飾、武器装飾、小物、体格、シルエット、年齢感、雰囲気、キャラクターカラーのみを参照する。\n\n画像1そのものの背景、部屋、家具、ポーズ、画角、フレーミング、文字、レイアウトは再現しない。\n\n表情、視線、口元、姿勢、呼吸、自然な髪と衣装の揺れのみ変化してよい。\n\n特徴の混合、顔の平均化、衣装交換、別人化、クローン化、余分な主人公の追加は禁止。\n\n【キャラクター固定】\n\n主人公は全編で一人のみ。\n\n顔、髪型、衣装、装飾、小物、武器、体格、シルエット、キャラクターカラーを維持する。\n\n武器は全編で一冊だけ存在する魔導書とし、画像1に合わせた材質・色・装飾・デザインを維持する。\n\n開始時は左腰へ装備され、カスタマイズ時のみ左手上へ浮遊し、終了後は左腰へ戻る。\n\n画像1に存在するアクセサリーや装飾品も維持し、不要な複製や消失は発生しない。\n\n【画風】\n\n劇場版品質。\n\n高密度3Dトゥーンセルルック。\n\n二〜三段セルシェーディング。\n\n透明感のあるライティング。\n\n髪、瞳、布、革、金属、宝石、紙を高密度に描き分ける。\n\n色彩は画像1のキャラクターカラーを最優先とし、背景は人物より一段落ち着いた色調で統一する。\n\n画風混在、リアル調、プラスチックCG、低密度背景は禁止。\n\n【ゲームUI】\n\n高級ファンタジーRPG風UI。\n\n半透明クリスタルパネル。\n\n細い金装飾。\n\n魔法陣と星座モチーフ。\n\n文字は読み取れる品質で安定表示。\n\nUIは顔を覆わず画面の空き領域へ配置する。\n\n操作に合わせて光と効果音が同期する。\n\n【シーン】\n\n高い俯瞰から開始。\n\n主人公は魔法陣中央へ座り、カメラを見る。\n\n右側へ\n\nSTART NEW GAME\n\nCONTINUE\n\nARCANE SETTINGS\n\nEXIT\n\n左上へプロフィールを表示。\n\nCONTINUEのみ選択状態。\n\nカーソルが一度移動して決定する。\n\nカメラは右手へ降下。\n\nRIGHT-HAND CATALYSTパネルが表示される。\n\n掌の上へ六枚花弁の魔法術式が形成される。\n\n手の形は変化せず、術式のみ再構成される。\n\n花弁、宝石核、二重金環が完成して停止する。\n\n続いてカメラは左手へ回り込む。\n\nGRIMOIRE CUSTOMIZATIONを表示。\n\nCOVER\n\nCLASP\n\nCORE\n\nPAGES\n\n四項目を表示。\n\n主人公が左手を開くと左腰から魔導書が浮遊する。\n\n閉じた一冊の状態から、表紙、金具、中心核、ページが順番に分解される。\n\n全パーツはルーンで接続されたまま維持される。\n\nカーソルが四項目を巡るたび各パーツが更新され、最後に一冊へ戻る。\n\n革、紙、金属、宝石の質感を高密度に描写する。\n\nカメラは正面へ戻る。\n\nCONFIRM CONFIGを表示。\n\n主人公が一度押す。\n\n全UIが魔導書へ吸収される。\n\n魔導書は左腰へ戻る。\n\n主人公は自然に座り直す。\n\n掌の術式がゆっくり回転して完成する。\n\n表情は目→眉→まぶた→口元→視線の順で自然に変化する。\n\nその後カメラを少し引く。\n\n画面下へLOADINGバー。\n\n0%から100%まで進行する。\n\n紫色の開始空間から石畳、建物、屋台、灯り、街路が順番に実体化する。\n\n線画→素材→完成世界の三段階で生成される。\n\n舞台は雨上がりの星花魔法都市。\n\n濡れた石畳。\n\n魔法露店。\n\n魔法灯。\n\nルーン看板。\n\nローブ姿の群衆。\n\n飛行する箒便。\n\n浮遊馬車。\n\n石造と木造建築。\n\n遠景には巨大な魔法学院。\n\n背景人物は全員別人であり、主人公は複製しない。\n\n最後に三人称追従カメラになる。\n\n主人公は中景サイズを維持する。\n\n右上へ円形ミニマップ。\n\n左下へ\n\nVITALITY\n\nMANA\n\nSPELL SLOT\n\n街の奥へ\n\nRETURN TO ASTRA ACADEMY\n\n任務マーカーを表示。\n\n主人公は左腰へ魔導書を装備したまま街へ歩き出す。\n\n最後は左足が石畳へ着地し、右足が次の一歩へ移る瞬間で終了する。\n\n都市は完全に実体化し続け、群衆、箒便、浮遊馬車は動き続ける。\n\nHUDと任務マーカーは安定表示する。\n\n【音】\n\nBGMなし。\n\n音楽なし。\n\n環境音のみ。\n\n雨上がりの街の空気。\n\n石畳の足音。\n\n衣装の布擦れ。\n\n遠い市場の声。\n\n箒の風切り。\n\n浮遊馬車の魔力音。\n\n主人公は自然な呼吸のみ。\n\n台詞、ナレーション、長い発声は禁止。\n\n【禁止事項】\n\n画像1の背景や構図の再現、特徴の混合、顔の平均化、衣装交換、武器変更、別人化、クローン化、余分な主人公、余分な武器、余分な装飾、手指崩れ、顔へUIが重なる構図、読めないUI文字、急激な背景転換、白飛び、画風変化は禁止。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 21,
      "views": 3613,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083736220635738298",
    "title": "黑暗幻想世界的游戏预告",
    "titleEn": "A Trailer for a Dark Fantasy World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "黑暗幻想世界的游戏预告的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a trailer for a dark fantasy world on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "高品質アニメ映像。\nExpression progression: eyes -> brows -> eyelids -> mouth -> gaze.\n【参照画像とキャラクター同一性固定】\n参照画像は画像1のみ。画像1の人物を全編で同じ一人の主人公として維持する。画像1は、顔、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、全身シルエット、年齢感、雰囲気、キャラクターカラーの唯一の参照とする。画像1そのものの背景、部屋、家具、床、ポーズ、表情、画角、構図、フレーミング、文字、UIは再現しない。\n\n顔立ち、目、髪、衣装構造、装飾配置、体格を混合・平均化・簡略化しない。変えてよいのは、表情、視線、口元、呼吸、ポーズ、歩行、自然な髪と衣装の揺れだけ。別人化、顔平均化、衣装交換、装飾欠落、複製、クローン、追加人物を禁止する。\n画像1から主人公の主色、副色、差し色、金属色、宝石色、性格印象を読み取り、映像全体の光、魔法、星花、建築装飾、魔導書、ドラゴンの配色と意匠へ反映する。ただし主人公自身の虹彩色、髪色、衣装色は勝手に変更しない。\n\n【キャラクター連動小物】\n能動小物は主人公専用の魔導書一冊だけ。形状は革装丁の本で固定し、表紙色、金具、紋章、宝石、ページ上の図形だけを画像1の衣装・装飾・キャラクターカラーに合わせる。開始時は閉じ、左手で下から支え、右手で表紙を開き、紙のページを一枚だけめくる。終盤に右手で閉じ、最後も一冊のまま左手に保持する。腰飾りやアクセサリーは身体へ固定したまま使用しない。\n\n【固定レンダリング】\n高密度3Dトゥーン／セルルック。細い色付き線画、二段から三段のトゥーン影、透明感のある柔らかな中間影。肌はマット、髪は絹、衣装は織布、魔導書は革、金属は固体反射、宝石は透明結晶として描き分ける。顔、髪束、衣装端、装飾、鱗、浮遊建築、雲海まで劇場版クラスの密度。\n\n画像1のキャラクターカラーを中心に、高彩度かつ上品な配色へ統一する。太い黒輪郭、単層セル影、汎用3D美少女顔、プラスチックCG、低密度背景、半写実、写実、くすみ、画風混合にしない。\n【古代星竜固定】\nドラゴンは一体だけ。主人公を選ぶ荘厳な守護者で、敵対者として描かない。鱗、装甲、瞳、星座紋、翼膜の色と発光は画像1のキャラクターカラーに調和させるが、身体構造は固定する。\n\n一つの頭、二本の角、一続きの首と胴体、胴体へ接続した四本の脚、背中へ接続した左右一対の翼、骨盤から先端まで一続きの一本の尻尾。翼は付け根から翼骨と翼膜全体が一体で動き、尻尾は根元から鱗のある先端まで連続する。頭、翼、脚、角、尻尾の増加、分岐、欠損、別個体化、水晶像化、人型化を禁止する。\n【固定世界と可変トーン】\n場所は全編固定。雲海上の巨大な浮遊島、空中回廊、尖塔群、空へ逆流する一本の滝、星花の断崖、巨大な環状天文儀、二重月、遠景の浮遊城。\n\n構造と地形は変えず、建築の縁取り、星花、月光、空、霧、魔法光、ドラゴンの発光だけを画像1の色と雰囲気に合わせる。可憐な人物なら透明感と軽やかさ、高貴な人物なら荘厳さ、冷静な人物なら静謐さ、活発な人物なら鮮烈さを強める。ただし恐怖、陰惨さ、現代都市、犯罪ノワール、敵対場面にはしない。王道ハイファンタジー映画ティザーとして、主人公と古代星竜が初めて互いを認める瞬間を描く。\n\n【カメラと編集】\n物理カメラと編集遷移を分離する。星花線と星光が示す一本の経路を進み、瞳の異変から巨大な浮遊世界までスケールを開く。\n主人公の左斜め前、eye level。片目をextreme close-upで捉え、短くpush-in。\n\n魔導書の真上、胸の高さからページ全面をmacro insert。\n\n主人公の右後方、腰より低いlow angle。全身を保ち、空中回廊に沿って前進追従。\n高い左側面からwideへpull-back。主人公を前景に残し、雲海、浮遊島、逆流する滝、星竜の全身をreveal。\n\n左斜め前eye level、胸上mediumで表情を見せる。\n\n最後は低い右側面のwide固定。左前景の主人公、背景の星竜、右側のタイトル余白を同時に保つ。\n【固定シーン進行】\n暗い画面へ画像1の差し色に調和した細い星花線が一本描かれ、その線が主人公の片目の輪郭になる。主人公自身の虹彩と瞳孔は変えない。\n\n虹彩に映る魔導書の紋章だけが一度閉じ、反射の中で竜眼として開く。主人公は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に、小さな微笑みから息を止めた驚きへ変わり、左手の魔導書を見る。竜眼の円形反射が環状天文儀へ一致し、硬い円形ワイプになる。\n右手で一冊の魔導書を開き、紙のページを一枚だけめくる。ページ上の図形から、平面紋章、半透明の光骨格、空へ伸びる一本の星光という三段階で竜座が立体化する。これはドラゴン個体ではなく光図形。紙のページ端で斜めの硬いワイプ。\n主人公は空中回廊を二歩だけ前進する。星光が数メートル前を案内し、花形の床紋を順に点灯させる。髪と衣装は同じ後方風へなびく。画面中央を上昇する逆流滝を垂直ワイプとして巨大世界へ切り替える。\n遠景の雲の環から一体の古代星竜が現れる。画面右奥から浮遊城の周囲を時計回りに一度だけ大きく旋回し、左右一対の翼を背中の付け根からゆっくり一回羽ばたかせる。四本の脚は胴体の下へ自然に揃い、一本の尻尾は旋回軌道へ遅れて流れる。片翼がカメラ前を横切り、翼膜の硬い影ワイプで次へ渡す。\n主人公は星花の断崖で立ち止まり、魔導書を左手で胸元に支え、右手の掌を上へ向ける。掌の上に画像1の意匠に合う星花型魔法陣が一つだけ開く。星竜は背後へ位置を合わせ、長い首を下げ、一つの瞳を主人公と同じ高さへ置く。両者の間には明瞭な空間を残す。魔法陣と星竜の額の星座紋が一度だけ同時に発光する。\n星竜の瞳の縦長ハイライトが巨大な英字「D」の輪郭へ変わり、硬い文字マスクで主人公の胸上へ戻る。主人公は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に、驚きから静かな喜びと決意へ変わり、星竜を見上げる。右手で魔法陣を閉じ、続けて魔導書を閉じ、右手を表紙へ置く。\n\n最後は画像1の主色、副色、金属色による薄い矩形フレームがスナップし、一枚のタイトル画面を完成させる。終端タイプはspace reveal / pose lock。\n左前景で主人公は閉じた魔導書一冊を左手で胸元に保持し、右手を表紙へ置いて静止する。背景では一体の星竜が浮遊城を包む弧を描き、左右一対の翼を開いて静止し、一本の尻尾も城の外周で止まる。右側に読みやすい夜空の余白を残す。\n\n右側に英字タイトル「[CHARACTER NAME]: OATH OF THE STAR DRAGON」を一度だけ正しい綴りで表示し、その下に「COMING SOON」を一度だけ表示する。ほかの文字、疑似文字、字幕、ロゴは出さない。主人公、閉じた魔導書、一体の星竜、浮遊城、タイトルを二秒間保持し、新しい要素を入れない。最後の音の余韻が消える正確なフレームでカットする。\n\n【音響】\n完全オリジナルの幻想映画用タイトル音楽。低い弦のドローン、透明なチェレスタ、遠い混声コーラスの母音パッド、ハープの短い上昇音型、低いフレームドラム、フレンチホルン、深い弦、古い大鐘。冒頭は低い弦とチェレスタ。魔導書で小さなコーラス。空中回廊でフレームドラム。世界全景でホルンと弦が開く。星竜が瞳を合わせる瞬間に短い無言のコーラス和音。タイトル完成時は壮大な一和音と大鐘一打。台詞、ナレーション、歌詞、歌唱なし。\n\n効果音は見える動作だけに同期する。魔導書の開閉時だけ革のきしみ、ページ一枚だけ乾いた紙音、竜座の立体化時だけ細い結晶共鳴、逆流滝には上向きの水音、星竜の一回の羽ばたきには翼膜と低い風圧音、髪と衣装には軽い布音、額の星座紋の点灯時だけ澄んだ魔力共鳴。\n\n【禁止】\n別人化、顔崩れ、虹彩色変更、髪型変更、衣装交換、衣装簡略化、装飾欠落、魔導書の増殖、手指破綻、余分な腕、追加人物、群衆、ドラゴンの追加個体、多頭化、翼・脚・角・尻尾の増殖や欠損、敵対行動、火炎放射、流血、ホラー、実写、半写実、画風変化、太い黒輪郭、低密度背景、プラスチック質感、ソフトディゾルブ、液体モーフ、過剰な手ぶれ、綴り違い、重複タイトル、疑似文字、字幕、ロゴ、透かし。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2083736220635738298",
    "sourceUrl": "https://x.com/haruuraeadss/status/2083736220635738298",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-08-02T02:06:43.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2083736220635738298.mp4",
    "posterUrl": "/posters/x/2083736220635738298.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ゲームトレーラー風\n\n@Hailuo_AI\n#MiniMaxH3　\nプロンプト👇\n                                                                                                                                                                                                                                       \n高品質アニメ映像。\nExpression progression: eyes -> brows -> eyelids -> mouth -> gaze.\n【参照画像とキャラクター同一性固定】\n参照画像は画像1のみ。画像1の人物を全編で同じ一人の主人公として維持する。画像1は、顔、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、全身シルエット、年齢感、雰囲気、キャラクターカラーの唯一の参照とする。画像1そのものの背景、部屋、家具、床、ポーズ、表情、画角、構図、フレーミング、文字、UIは再現しない。\n\n顔立ち、目、髪、衣装構造、装飾配置、体格を混合・平均化・簡略化しない。変えてよいのは、表情、視線、口元、呼吸、ポーズ、歩行、自然な髪と衣装の揺れだけ。別人化、顔平均化、衣装交換、装飾欠落、複製、クローン、追加人物を禁止する。\n画像1から主人公の主色、副色、差し色、金属色、宝石色、性格印象を読み取り、映像全体の光、魔法、星花、建築装飾、魔導書、ドラゴンの配色と意匠へ反映する。ただし主人公自身の虹彩色、髪色、衣装色は勝手に変更しない。\n\n【キャラクター連動小物】\n能動小物は主人公専用の魔導書一冊だけ。形状は革装丁の本で固定し、表紙色、金具、紋章、宝石、ページ上の図形だけを画像1の衣装・装飾・キャラクターカラーに合わせる。開始時は閉じ、左手で下から支え、右手で表紙を開き、紙のページを一枚だけめくる。終盤に右手で閉じ、最後も一冊のまま左手に保持する。腰飾りやアクセサリーは身体へ固定したまま使用しない。\n\n【固定レンダリング】\n高密度3Dトゥーン／セルルック。細い色付き線画、二段から三段のトゥーン影、透明感のある柔らかな中間影。肌はマット、髪は絹、衣装は織布、魔導書は革、金属は固体反射、宝石は透明結晶として描き分ける。顔、髪束、衣装端、装飾、鱗、浮遊建築、雲海まで劇場版クラスの密度。\n\n画像1のキャラクターカラーを中心に、高彩度かつ上品な配色へ統一する。太い黒輪郭、単層セル影、汎用3D美少女顔、プラスチックCG、低密度背景、半写実、写実、くすみ、画風混合にしない。\n【古代星竜固定】\nドラゴンは一体だけ。主人公を選ぶ荘厳な守護者で、敵対者として描かない。鱗、装甲、瞳、星座紋、翼膜の色と発光は画像1のキャラクターカラーに調和させるが、身体構造は固定する。\n\n一つの頭、二本の角、一続きの首と胴体、胴体へ接続した四本の脚、背中へ接続した左右一対の翼、骨盤から先端まで一続きの一本の尻尾。翼は付け根から翼骨と翼膜全体が一体で動き、尻尾は根元から鱗のある先端まで連続する。頭、翼、脚、角、尻尾の増加、分岐、欠損、別個体化、水晶像化、人型化を禁止する。\n【固定世界と可変トーン】\n場所は全編固定。雲海上の巨大な浮遊島、空中回廊、尖塔群、空へ逆流する一本の滝、星花の断崖、巨大な環状天文儀、二重月、遠景の浮遊城。\n\n構造と地形は変えず、建築の縁取り、星花、月光、空、霧、魔法光、ドラゴンの発光だけを画像1の色と雰囲気に合わせる。可憐な人物なら透明感と軽やかさ、高貴な人物なら荘厳さ、冷静な人物なら静謐さ、活発な人物なら鮮烈さを強める。ただし恐怖、陰惨さ、現代都市、犯罪ノワール、敵対場面にはしない。王道ハイファンタジー映画ティザーとして、主人公と古代星竜が初めて互いを認める瞬間を描く。\n\n【カメラと編集】\n物理カメラと編集遷移を分離する。星花線と星光が示す一本の経路を進み、瞳の異変から巨大な浮遊世界までスケールを開く。\n主人公の左斜め前、eye level。片目をextreme close-upで捉え、短くpush-in。\n\n魔導書の真上、胸の高さからページ全面をmacro insert。\n\n主人公の右後方、腰より低いlow angle。全身を保ち、空中回廊に沿って前進追従。\n高い左側面からwideへpull-back。主人公を前景に残し、雲海、浮遊島、逆流する滝、星竜の全身をreveal。\n\n左斜め前eye level、胸上mediumで表情を見せる。\n\n最後は低い右側面のwide固定。左前景の主人公、背景の星竜、右側のタイトル余白を同時に保つ。\n【固定シーン進行】\n暗い画面へ画像1の差し色に調和した細い星花線が一本描かれ、その線が主人公の片目の輪郭になる。主人公自身の虹彩と瞳孔は変えない。\n\n虹彩に映る魔導書の紋章だけが一度閉じ、反射の中で竜眼として開く。主人公は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に、小さな微笑みから息を止めた驚きへ変わり、左手の魔導書を見る。竜眼の円形反射が環状天文儀へ一致し、硬い円形ワイプになる。\n右手で一冊の魔導書を開き、紙のページを一枚だけめくる。ページ上の図形から、平面紋章、半透明の光骨格、空へ伸びる一本の星光という三段階で竜座が立体化する。これはドラゴン個体ではなく光図形。紙のページ端で斜めの硬いワイプ。\n主人公は空中回廊を二歩だけ前進する。星光が数メートル前を案内し、花形の床紋を順に点灯させる。髪と衣装は同じ後方風へなびく。画面中央を上昇する逆流滝を垂直ワイプとして巨大世界へ切り替える。\n遠景の雲の環から一体の古代星竜が現れる。画面右奥から浮遊城の周囲を時計回りに一度だけ大きく旋回し、左右一対の翼を背中の付け根からゆっくり一回羽ばたかせる。四本の脚は胴体の下へ自然に揃い、一本の尻尾は旋回軌道へ遅れて流れる。片翼がカメラ前を横切り、翼膜の硬い影ワイプで次へ渡す。\n主人公は星花の断崖で立ち止まり、魔導書を左手で胸元に支え、右手の掌を上へ向ける。掌の上に画像1の意匠に合う星花型魔法陣が一つだけ開く。星竜は背後へ位置を合わせ、長い首を下げ、一つの瞳を主人公と同じ高さへ置く。両者の間には明瞭な空間を残す。魔法陣と星竜の額の星座紋が一度だけ同時に発光する。\n星竜の瞳の縦長ハイライトが巨大な英字「D」の輪郭へ変わり、硬い文字マスクで主人公の胸上へ戻る。主人公は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に、驚きから静かな喜びと決意へ変わり、星竜を見上げる。右手で魔法陣を閉じ、続けて魔導書を閉じ、右手を表紙へ置く。\n\n最後は画像1の主色、副色、金属色による薄い矩形フレームがスナップし、一枚のタイトル画面を完成させる。終端タイプはspace reveal / pose lock。\n左前景で主人公は閉じた魔導書一冊を左手で胸元に保持し、右手を表紙へ置いて静止する。背景では一体の星竜が浮遊城を包む弧を描き、左右一対の翼を開いて静止し、一本の尻尾も城の外周で止まる。右側に読みやすい夜空の余白を残す。\n\n右側に英字タイトル「[CHARACTER NAME]: OATH OF THE STAR DRAGON」を一度だけ正しい綴りで表示し、その下に「COMING SOON」を一度だけ表示する。ほかの文字、疑似文字、字幕、ロゴは出さない。主人公、閉じた魔導書、一体の星竜、浮遊城、タイトルを二秒間保持し、新しい要素を入れない。最後の音の余韻が消える正確なフレームでカットする。\n\n【音響】\n完全オリジナルの幻想映画用タイトル音楽。低い弦のドローン、透明なチェレスタ、遠い混声コーラスの母音パッド、ハープの短い上昇音型、低いフレームドラム、フレンチホルン、深い弦、古い大鐘。冒頭は低い弦とチェレスタ。魔導書で小さなコーラス。空中回廊でフレームドラム。世界全景でホルンと弦が開く。星竜が瞳を合わせる瞬間に短い無言のコーラス和音。タイトル完成時は壮大な一和音と大鐘一打。台詞、ナレーション、歌詞、歌唱なし。\n\n効果音は見える動作だけに同期する。魔導書の開閉時だけ革のきしみ、ページ一枚だけ乾いた紙音、竜座の立体化時だけ細い結晶共鳴、逆流滝には上向きの水音、星竜の一回の羽ばたきには翼膜と低い風圧音、髪と衣装には軽い布音、額の星座紋の点灯時だけ澄んだ魔力共鳴。\n\n【禁止】\n別人化、顔崩れ、虹彩色変更、髪型変更、衣装交換、衣装簡略化、装飾欠落、魔導書の増殖、手指破綻、余分な腕、追加人物、群衆、ドラゴンの追加個体、多頭化、翼・脚・角・尻尾の増殖や欠損、敵対行動、火炎放射、流血、ホラー、実写、半写実、画風変化、太い黒輪郭、低密度背景、プラスチック質感、ソフトディゾルブ、液体モーフ、過剰な手ぶれ、綴り違い、重複タイトル、疑似文字、字幕、ロゴ、透かし。",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 11,
      "views": 10249,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083628852165672988",
    "title": "另一张苏醒的动态海报",
    "titleEn": "Another Poster Awakens",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "另一张苏醒的动态海报的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for another poster awakens on X. Native-video output: 10s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Animate the source artwork as a 10-second premium sci-fi motion poster while preserving the white gallery background, thin border, navy-and-orange palette, typography, composition and original poster design. Begin on an almost empty white canvas, then progressively assemble the poster: the outer frame draws itself on, the orange “ORIGIN” background typography slides vertically into place, the planetary landscape fades and expands inside the central frame, and the colossal station descends from above as its lights activate. Bring the cloaked traveler forward from the distant landscape toward the foreground with strong dimensional parallax, allowing the figure and rocky base to slightly break outside the inner frame. Introduce the smaller titles, numbering, studio credits, graphic marks and footer icons one after another with clean pop-in, slide and type-on animations, accompanied by light futuristic clicks, playful digital typing sounds and soft mechanical impacts. Finish with a restrained camera push-in, subtle station lights, drifting mist and a gentle movement in the traveler’s coat while the completed poster remains perfectly readable and close to the original still image. No major scene transformation, no camera orbit, no distorted typography, no altered layout.\n2/2",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2083628879407632890",
    "sourceUrl": "https://x.com/LudovicCreator/status/2083628852165672988",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-01T19:00:04.000Z",
    "addedAt": "2026-08-21T20:02:13.120Z",
    "mediaUrl": "/media/x-2083628852165672988.mp4",
    "posterUrl": "/posters/x/2083628852165672988.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Honestly the animations possibilities with Minimax H3 are really great .\n\nHere an example for poster.\n\nFind the image used and prompt in first comment.\n\nMade in MinimaxHub @Hailuo_AI #MiniMaxH3\n1/2",
    "engagement": {
      "replies": 16,
      "reposts": 6,
      "likes": 55,
      "views": 5658,
      "snapshotAt": "2026-08-21T20:02:13.120Z"
    },
    "approvedAt": "2026-08-21T20:02:13.120Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085973171581161893",
    "title": "泥泞脚步与惊恐回眸",
    "titleEn": "Muddy Footsteps and a Terrified Turn",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "泥泞脚步与惊恐回眸的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 5 秒 · 1920×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for muddy footsteps and a terrified turn on X. Native-video output: 5s · 1920×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[PROSE & IN-LINE SPATIAL & GLOBAL LOCK]: At 0.0s, heavy weathered leather boots wrapped in thick wet mud violently stomp down onto a murky black water puddle, instantly snapping a thick wooden branch underneath with crushing weight. Strict environment geometry: ground covered in wet black mud, scattered dead leaves, and thick atmospheric fog. At 2.5s, [SMASH CUT TO: EXTREME CLOSE-UP] Elias's face snaps sharply toward SCREEN-LEFT in a sudden panic, his jaw clenched tight and his pupils contracting, freezing in terror as he stares at an off-screen threat. Strict character state: rugged 30-year-old man, face heavily splattered with dark wet mud and cold sweat micro-drops, wearing a thick moss-green canvas hunting jacket. Strict universal numerical quota lock: exactly one pair of boots, exactly one man, zero duplicate entities, zero melting geometry.\n[ACTING & LIGHTING SCIENCE]: Zero cartoon facial blushing, zero screaming, pure grounded realistic micro-expressions, eyes locked strictly off-camera to SCREEN-LEFT. Gloomy overcast daylight filtering through thick fog. Strict Triadic Color Law: Moss Green (#284D34), Pale cold skin tone (#D4D0C5), Deep shadow negative space (#121A15).\n[CAMERA SCIENCE & KINETIC PHYSICS]: [0.0s-2.5s] Cinematic extreme low-angle macro shot on the boots, organic micro-jitter human-held camera physics capturing the downward kinetic force. [2.5s-5.0s] Pure static locked-off camera physics emphasizing the sudden tense stillness on the face. Fast f/1.2 Leica anamorphic optics, heavy depth-of-field isolation melting the dead pine forest background into a creamy defocused blur.",
    "promptSourceUrl": "https://x.com/FikkkFX/status/2085973176069267487",
    "sourceUrl": "https://x.com/FikkkFX/status/2085973171581161893",
    "sourceLabel": "X 原帖 · @FikkkFX",
    "author": "Fikkkfx",
    "publishedAt": "2026-08-08T06:15:34.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2085973171581161893.mp4",
    "posterUrl": "/posters/x/2085973171581161893.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1920×2160",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just testing the new update on my agent.\n\nUpgraded the entire Director OS to V19.1 with strict Anti-Melodrama & Triadic Color rules.\n\nThe seamless cut-on-action and texture details are insane.\n\nBase Skills Agent: @apilpirman \nVideo: Seedance 2.5 and Minimax H3 on @atlas_cloud_ai \n\nPrompt👇",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 10,
      "views": 620,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085952141395726491",
    "title": "办公室里的人工智能新员工",
    "titleEn": "The Office Welcomes an AI Employee",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "办公室里的人工智能新员工的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the office welcomes an ai employee on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "\"The Office (US) mockumentary style. Dunder Mifflin bullpen, mid-afternoon, harsh fluorescent office lighting, beige cubicles, cluttered desks, printers humming in the background. Handheld documentary camera with subtle shake, imperfect framing, sudden snap zooms during reactions, quick reframes as if a real documentary crew is trying to keep up. Muted office color palette, soft consumer-camera look.\nMichael Scott proudly stands beside an empty desk with a laptop open. Dwight Schrute stands nearby, suspicious and visibly offended. Only Michael Scott and Dwight Schrute appear in the scene. The laptop represents an AI assistant named \"Claude\" but no third person appears.\nDIALOGUE (≈15 seconds):\nMICHAEL (excited):\n\"Dwight, I'd like you to meet our newest employee... Claude.\"\nDwight squints at the laptop.\nDWIGHT:\n\"That's a computer.\"\nSnap zoom to Michael.\nMICHAEL (grinning):\n\"It's AI. It works faster... and it doesn't ask for vacation.\"\nBeat.\nDWIGHT (confident):\n\"It also cannot harvest beets.\"\nMichael nods thoughtfully.\nMICHAEL:\n\"...Fair point.\"\nBeat.\nMichael suddenly smiles again.\nMICHAEL:\n\"But it can replace you.\"\nSnap zoom to Dwight.\nDWIGHT (dead serious):\n\"Then it's my enemy.\"\nDwight slowly closes the laptop lid as Michael stares into the documentary camera.\nSTYLE: Single-camera mockumentary, natural performances, dry deadpan humor, subtle handheld movement, documentary snap zooms, awkward pauses, realistic office ambience, no laugh track, with only Michael Scott and Dwight Schrute appearing in the scene.\"",
    "promptSourceUrl": "https://x.com/karankendre/status/2085952141395726491",
    "sourceUrl": "https://x.com/karankendre/status/2085952141395726491",
    "sourceLabel": "X 原帖 · @karankendre",
    "author": "Karan",
    "publishedAt": "2026-08-08T04:52:00.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2085952141395726491.mp4",
    "posterUrl": "/posters/x/2085952141395726491.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 one-shotted this complete scene\n\nPrompt : \"The Office (US) mockumentary style. Dunder Mifflin bullpen, mid-afternoon, harsh fluorescent office lighting, beige cubicles, cluttered desks, printers humming in the background. Handheld documentary camera with subtle shake, imperfect framing, sudden snap zooms during reactions, quick reframes as if a real documentary crew is trying to keep up. Muted office color palette, soft consumer-camera look.\nMichael Scott proudly stands beside an empty desk with a laptop open. Dwight Schrute stands nearby, suspicious and visibly offended. Only Michael Scott and Dwight Schrute appear in the scene. The laptop represents an AI assistant named \"Claude\" but no third person appears.\nDIALOGUE (≈15 seconds):\nMICHAEL (excited):\n\"Dwight, I'd like you to meet our newest employee... Claude.\"\nDwight squints at the laptop.\nDWIGHT:\n\"That's a computer.\"\nSnap zoom to Michael.\nMICHAEL (grinning):\n\"It's AI. It works faster... and it doesn't ask for vacation.\"\nBeat.\nDWIGHT (confident):\n\"It also cannot harvest beets.\"\nMichael nods thoughtfully.\nMICHAEL:\n\"...Fair point.\"\nBeat.\nMichael suddenly smiles again.\nMICHAEL:\n\"But it can replace you.\"\nSnap zoom to Dwight.\nDWIGHT (dead serious):\n\"Then it's my enemy.\"\nDwight slowly closes the laptop lid as Michael stares into the documentary camera.\nSTYLE: Single-camera mockumentary, natural performances, dry deadpan humor, subtle handheld movement, documentary snap zooms, awkward pauses, realistic office ambience, no laugh track, with only Michael Scott and Dwight Schrute appearing in the scene.\"",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 14,
      "views": 1680,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085947083463295024",
    "title": "邦德式谍战动漫片头",
    "titleEn": "A Bond-Style Spy Anime Title Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "邦德式谍战动漫片头的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a bond-style spy anime title sequence on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Using Image as the character reference and the attached modern retro-western electro track for timing, create a 60s Bond-style spy-thriller op-art anime title sequence. Every cut must land exactly on the beat. Start with slow, sparse shots and progressively accelerate into rapid-fire montage as the drum pulse intensifies.\n\nSTYLE: Flat, high-contrast pure black/white silhouette against psychedelic 1960s op-art backgrounds: concentric circles, starbursts, spirals, stripes, chevrons and kaleidoscopic patterns. Rotate through hot pink/orange, electric teal/violet, and acid yellow/magenta palettes. No realistic shading.\n\nUse classic Bond-title compositions: walking/spinning silhouettes, gun-barrel-style irises, mirrored/kaleidoscopic duplicates, and extreme close-ups of eyes, hands and weapons. Backgrounds may swirl and morph, but all shot transitions are hard cuts only.\n\nUse frequent half-to-one-beat freeze-frames on dramatic poses, with the character and pattern completely frozen before snapping to the next beat and palette. Alternate kinetic movement with sudden dead-stop freezes.\n\nAdd bold 1960s Bond-style geometric italic typography, spinning/tiling/zooming in sync with the music. Apply subtle analog film grain and warm, saturated vintage print grading.\n\nENDING: On the final hit, freeze the character in full black silhouette at center of a fully bloomed hot-pink-and-gold radiating starburst, while the title card spins/zooms into frame beside them.",
    "promptSourceUrl": "https://x.com/doctorwasif/status/2085947083463295024",
    "sourceUrl": "https://x.com/doctorwasif/status/2085947083463295024",
    "sourceLabel": "X 原帖 · @doctorwasif",
    "author": "WasifAI",
    "publishedAt": "2026-08-08T04:31:54.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2085947083463295024.mp4",
    "posterUrl": "/posters/x/2085947083463295024.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax H3\n\nPrompt:\nUsing Image as the character reference and the attached modern retro-western electro track for timing, create a 60s Bond-style spy-thriller op-art anime title sequence. Every cut must land exactly on the beat. Start with slow, sparse shots and progressively accelerate into rapid-fire montage as the drum pulse intensifies.\n\nSTYLE: Flat, high-contrast pure black/white silhouette against psychedelic 1960s op-art backgrounds: concentric circles, starbursts, spirals, stripes, chevrons and kaleidoscopic patterns. Rotate through hot pink/orange, electric teal/violet, and acid yellow/magenta palettes. No realistic shading.\n\nUse classic Bond-title compositions: walking/spinning silhouettes, gun-barrel-style irises, mirrored/kaleidoscopic duplicates, and extreme close-ups of eyes, hands and weapons. Backgrounds may swirl and morph, but all shot transitions are hard cuts only.\n\nUse frequent half-to-one-beat freeze-frames on dramatic poses, with the character and pattern completely frozen before snapping to the next beat and palette. Alternate kinetic movement with sudden dead-stop freezes.\n\nAdd bold 1960s Bond-style geometric italic typography, spinning/tiling/zooming in sync with the music. Apply subtle analog film grain and warm, saturated vintage print grading.\n\nENDING: On the final hit, freeze the character in full black silhouette at center of a fully bloomed hot-pink-and-gold radiating starburst, while the title card spins/zooms into frame beside them.",
    "engagement": {
      "replies": 29,
      "reposts": 1,
      "likes": 43,
      "views": 3017,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085918684879765882",
    "title": "自动对口型的高速说唱表演",
    "titleEn": "Automatic Lip-Sync for a High-Speed Rap Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "自动对口型的高速说唱表演的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 30 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for automatic lip-sync for a high-speed rap performance on X. Native-video output: 30s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "モード1：Omni Reference\nプロンプト：添付のキャラクター参照（ 📷Image1 ）とオーディオトラックをタイミングガイドとして使用してください。オーディオにある「音声」に完全に会うようにキャラクターの口とアクションを入れてリップシンクしてダイナミックなカメラアングルでラップを披露してください。カメラはマルチカットでキャラの部位をクローズアップしてください。全て超ハイスピードで激しく行ってください。\n\nモード2：Omni Reference\nプロンプト：添付のキャラクター参照（ 📷Image1 ）とオーディオトラックをタイミングガイドとして使用してください。オーディオにある「音声」に完全に会うようにキャラクターの口とアクションを入れてリップシンクしてダイナミックなカメラアングルでラップを披露してください。カメラはマルチカットで超ローアングル・超ハイアングルを小刻みに切り替えたUS東海岸のラップビデオのようなスタイルに仕上げる。全て超ハイスピードで激しく行ってください。",
    "promptSourceUrl": "https://x.com/KEETY2591756/status/2085918684879765882",
    "sourceUrl": "https://x.com/KEETY2591756/status/2085918684879765882",
    "sourceLabel": "X 原帖 · @KEETY2591756",
    "author": "KEETY｜AIクリエイター",
    "publishedAt": "2026-08-08T02:39:03.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2085918684879765882.mp4",
    "posterUrl": "/posters/x/2085918684879765882.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "え？マジか‼️MiniMax H3「リップシンクが自動」✨画像一枚と音声ファイル添付して次のプロンプトでポン出しがこれ⬇️ @Hailuo_AI \n\nモード1：Omni Reference\nプロンプト：添付のキャラクター参照（ 📷Image1 ）とオーディオトラックをタイミングガイドとして使用してください。オーディオにある「音声」に完全に会うようにキャラクターの口とアクションを入れてリップシンクしてダイナミックなカメラアングルでラップを披露してください。カメラはマルチカットでキャラの部位をクローズアップしてください。全て超ハイスピードで激しく行ってください。\n\nモード2：Omni Reference\nプロンプト：添付のキャラクター参照（ 📷Image1 ）とオーディオトラックをタイミングガイドとして使用してください。オーディオにある「音声」に完全に会うようにキャラクターの口とアクションを入れてリップシンクしてダイナミックなカメラアングルでラップを披露してください。カメラはマルチカットで超ローアングル・超ハイアングルを小刻みに切り替えたUS東海岸のラップビデオのようなスタイルに仕上げる。全て超ハイスピードで激しく行ってください。",
    "engagement": {
      "replies": 3,
      "reposts": 10,
      "likes": 61,
      "views": 3731,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2085893756688167396",
    "title": "交通规则里的推理挑战",
    "titleEn": "A Reasoning Challenge Built from Traffic Rules",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "交通规则里的推理挑战的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 14 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a reasoning challenge built from traffic rules on X. Native-video output: 14s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A single continuous fixed wide shot in a plain, brightly lit test room, filmed like an unedited behavioral experiment. At the start, a red cube is on a table, a drawer is closed, a lamp is off, a chair sits precisely inside a taped floor outline, a hat hangs on a wall hook beside a bare mannequin, and a door is open. One person quickly performs six clearly separated actions in this exact order: places the red cube on a shelf, opens the drawer, turns on the lamp, moves the chair away from its taped outline, takes the hat from the hook and puts it on the mannequin, then closes the door. An off-screen instructor clearly says, \"Undo only the last three actions.\" The person pauses briefly, understands the instruction, and carries it out exactly. No cuts, no time jumps, no extra actions, no additional dialogue, realistic synchronized room sounds.",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2085893756688167396",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2085893756688167396",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-08T01:00:00.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2085893756688167396.mp4",
    "posterUrl": "/posters/x/2085893756688167396.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Can Minimax H3 reason? -- I wanted to see if reasoning could somehow emerge as an unexpected behavior in H3, so tried an experiment.\n\n1. Good news: It carries out every single instruction in exact detail, still amazed by this.\n\n2. Bad news:\n\nI don't think it can reason. If the model could reason, it would have understood what it means by \"The person understands the instruction, and carries out exactly\", and carried out the instruction, but it doesn't.\n\nPrompt: A single continuous fixed wide shot in a plain, brightly lit test room, filmed like an unedited behavioral experiment. At the start, a red cube is on a table, a drawer is closed, a lamp is off, a chair sits precisely inside a taped floor outline, a hat hangs on a wall hook beside a bare mannequin, and a door is open. One person quickly performs six clearly separated actions in this exact order: places the red cube on a shelf, opens the drawer, turns on the lamp, moves the chair away from its taped outline, takes the hat from the hook and puts it on the mannequin, then closes the door. An off-screen instructor clearly says, \"Undo only the last three actions.\" The person pauses briefly, understands the instruction, and carries it out exactly. No cuts, no time jumps, no extra actions, no additional dialogue, realistic synchronized room sounds.",
    "engagement": {
      "replies": 11,
      "reposts": 2,
      "likes": 81,
      "views": 10353,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083617497216106839",
    "title": "漫游城市的浪漫蒙太奇",
    "titleEn": "A Romantic Montage Across the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "漫游城市的浪漫蒙太奇的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 8 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a romantic montage across the city on X. Native-video output: 8s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Romantic montage of two people exploring a beautiful city. Soft lighting, candid smiles, and cinematic motion graphics.",
    "promptSourceUrl": "https://x.com/ikausarsolanki/status/2083617497216106839",
    "sourceUrl": "https://x.com/ikausarsolanki/status/2083617497216106839",
    "sourceLabel": "X 原帖 · @ikausarsolanki",
    "author": "𝙆𝙖𝙪𝙨𝙖𝙧 𝙎𝙤𝙡𝙖𝙣𝙠𝙞",
    "publishedAt": "2026-08-01T18:14:57.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083617497216106839.mp4",
    "posterUrl": "/posters/x/2083617497216106839.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "motion-graphics",
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Prompt: \n\nRomantic montage of two people exploring a beautiful city. Soft lighting, candid smiles, and cinematic motion graphics.\n\nDone with @Minimax H3 in @magnific",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 269,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083615699260313955",
    "title": "博物馆里苏醒的机械蝴蝶",
    "titleEn": "The Mechanical Butterfly Awakens in a Museum",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "博物馆里苏醒的机械蝴蝶的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for the mechanical butterfly awakens in a museum on X. Native-video output: 15s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "生成一条 15 秒、竖屏 9:16 的强写实电影感短视频。\n请参考上传的 博物馆大厅图片 作为唯一场景环境参考，参考上传的 机械蝴蝶图片 作为唯一蝴蝶外观参考，参考上传的 手部图片 作为结尾落点动作参考。\n全片必须保持 同一座博物馆、同一只机械蝴蝶、同一种夜晚氛围，不要切换到其他场景，不要出现第二只蝴蝶，不要出现多余人物。\n整体风格为：\n 超写实、电影感、神秘、安静、带一点奇迹感的夜晚博物馆短片。\n光线以 冷色月光 为主，辅以少量柔和展柜灯光。\n镜头语言要求自然、克制、流畅，像真实摄影机拍摄。\n尽量表现为 一个连续镜头的观感，不要硬切，不要频繁跳镜头，不要夸张运镜。\n视频内容节奏如下：\n\n0–3 秒\n镜头从安静空旷的自然历史博物馆大厅开始，夜晚无人，月光穿过高窗照进来，照亮恐龙骨架、玻璃展柜和地面反光。镜头缓慢向前推进，氛围神秘安静。一个玻璃展柜内，一只机械蝴蝶静静停着，随后开始苏醒。它身体内部极微弱地亮起，细小机械部件开始运转，翅膀缓缓展开。\n3–7 秒\n机械蝴蝶轻轻振动翅膀，真实自然地从展柜中飞起，飞出展柜。镜头顺势跟随蝴蝶移动。蝴蝶飞行姿态轻盈优雅，金属翅膀在月光下反射出细腻光泽。展柜玻璃、地面、恐龙骨架和大厅结构在背景中保持真实稳定的空间透视。\n7–12 秒\n镜头持续跟拍机械蝴蝶穿过博物馆主展厅。它飞过巨大的恐龙骨架前方，掠过月光照亮的空气和玻璃反射，画面具有强烈的空间纵深感和电影感。飞行路线流畅清晰，不要忽快忽慢，不要突然瞬移，不要夸张魔法特效。重点表现蝴蝶的精密金属结构、半透明翅膀和真实飞行动作。\n12–15 秒\n一只成年人自然伸出的手进入画面，姿态放松、安静。机械蝴蝶缓缓减速，轻轻降落在手指上。它停稳后，翅膀慢慢合拢，身体微微发出低亮度光泽。镜头轻微推近，停留在机械蝴蝶与手指的精致近景上，形成一个安静、惊艳、富有记忆点的结尾。",
    "promptSourceUrl": "https://x.com/zhongying14/status/2083615699260313955",
    "sourceUrl": "https://x.com/zhongying14/status/2083615699260313955",
    "sourceLabel": "X 原帖 · @zhongying14",
    "author": "麻酱AI实验室",
    "publishedAt": "2026-08-01T18:07:49.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083615699260313955.mp4",
    "posterUrl": "/posters/x/2083615699260313955.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "我让 MiniMax H3 用 3 张参考图\n拍了一条 15 秒短片：\n夜晚的博物馆里，最后一只机械蝴蝶苏醒了。\n要开声音听，煽动翅膀的声音很棒，机械部分也好精致\n\n提示词：\n生成一条 15 秒、竖屏 9:16 的强写实电影感短视频。\n请参考上传的 博物馆大厅图片 作为唯一场景环境参考，参考上传的 机械蝴蝶图片 作为唯一蝴蝶外观参考，参考上传的 手部图片 作为结尾落点动作参考。\n全片必须保持 同一座博物馆、同一只机械蝴蝶、同一种夜晚氛围，不要切换到其他场景，不要出现第二只蝴蝶，不要出现多余人物。\n整体风格为：\n 超写实、电影感、神秘、安静、带一点奇迹感的夜晚博物馆短片。\n光线以 冷色月光 为主，辅以少量柔和展柜灯光。\n镜头语言要求自然、克制、流畅，像真实摄影机拍摄。\n尽量表现为 一个连续镜头的观感，不要硬切，不要频繁跳镜头，不要夸张运镜。\n视频内容节奏如下：\n\n0–3 秒\n镜头从安静空旷的自然历史博物馆大厅开始，夜晚无人，月光穿过高窗照进来，照亮恐龙骨架、玻璃展柜和地面反光。镜头缓慢向前推进，氛围神秘安静。一个玻璃展柜内，一只机械蝴蝶静静停着，随后开始苏醒。它身体内部极微弱地亮起，细小机械部件开始运转，翅膀缓缓展开。\n3–7 秒\n机械蝴蝶轻轻振动翅膀，真实自然地从展柜中飞起，飞出展柜。镜头顺势跟随蝴蝶移动。蝴蝶飞行姿态轻盈优雅，金属翅膀在月光下反射出细腻光泽。展柜玻璃、地面、恐龙骨架和大厅结构在背景中保持真实稳定的空间透视。\n7–12 秒\n镜头持续跟拍机械蝴蝶穿过博物馆主展厅。它飞过巨大的恐龙骨架前方，掠过月光照亮的空气和玻璃反射，画面具有强烈的空间纵深感和电影感。飞行路线流畅清晰，不要忽快忽慢，不要突然瞬移，不要夸张魔法特效。重点表现蝴蝶的精密金属结构、半透明翅膀和真实飞行动作。\n12–15 秒\n一只成年人自然伸出的手进入画面，姿态放松、安静。机械蝴蝶缓缓减速，轻轻降落在手指上。它停稳后，翅膀慢慢合拢，身体微微发出低亮度光泽。镜头轻微推近，停留在机械蝴蝶与手指的精致近景上，形成一个安静、惊艳、富有记忆点的结尾。",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 3,
      "views": 2597,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083611951003734471",
    "title": "六图一音频打造动漫片头",
    "titleEn": "An Anime Opening from Six Images and One Audio Track",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "六图一音频打造动漫片头的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an anime opening from six images and one audio track on X. Native-video output: 15s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a fast-paced 15-second, 16:9 anime opening-title sequence for an original series titled “MINIMAX.” Use Image1, Image2, Image3 as the overall visual, ensemble and color-style references. Use Image4 for the pink-haired swordswoman, Image5 for the red-haired swordswoman and Image6 for the black-haired swordswoman. Preserve each character’s face, hairstyle, outfit, proportions, colors and weapon design.\n\nUse the supplied audio Audio1 as the exact timing reference. Synchronize every cut, character action, camera movement, graphic transition and title reveal to its beats, accents, fills, rises and final hit. Do not create a separate soundtrack. Add only subtle sword swishes, cloth movement and transition impacts under the reference audio.\n\nVISUAL STYLE\n\nA real high-energy anime opening combined with premium motion graphics. Use black, crimson, wine red, dusty pink, pale ivory and cool gray-blue. Combine angular split screens, diagonal masks, torn-paper shapes, manga-style framing, ink textures, petals, crimson moon forms, torii silhouettes, temple rooftops and sharp graphic lines.\n\nKeep the sequence kinetic and animated, not a slideshow. Characters must run, turn, pivot, slide, draw their swords, change stance and react to momentum. Hair and clothing must remain in motion. Use tracking shots, whip pans, rapid push-ins, low angles and short camera orbits.\n\nNever show multiple copies of the same character in one shot. A shot may feature one character alone or all three together naturally. No clone effects, repeated character fragments or layered duplicates.\n\nSEQUENCE FLOW\n\n[0.0–1.5s]\nOpen instantly on the first beat. A crimson line cuts through black as petals and white fragments burst outward. Pink, red and black graphic panels snap into place while a crimson moon quickly assembles in the background.\n\n[1.5–4.3s]\nFeature the pink-haired swordswoman alone. She runs into frame, plants one foot, turns and draws her katana in one continuous agile motion. Track beside her, then push toward her face. Cut between her eye, hand, blade and full-body movement on the audio accents. Use pink petals, red lines and diagonal panel cuts.\n\n[4.3–7.2s]\nSwitch to the red-haired swordswoman on a strong musical hit. Use a low-angle tracking shot as she moves forward, pivots and performs one powerful controlled sword motion. Her long hair and garments sweep with momentum. Place a large crimson sun disc behind her while black-red panels strike into frame on the heavier beats.\n\n[7.2–10.2s]\nFeature the black-haired swordswoman alone. Begin with a moving close-up of her crescent earring and eyes, then orbit around her as she turns, changes stance and redirects her blade. Keep her hair and sleeves flowing. Use dark negative space, moon-shaped masks and precise cuts synchronized to smaller rhythmic details.\n\n[10.2–12.7s]\nAt the main musical peak, reveal all three characters together in one shared scene. They enter through real movement and form a strong triangular composition. Use a fast camera push or short orbit as the crimson moon, torii silhouette, petals and graphic debris align behind them. Show each character only once.\n\n[12.7–15.0s]\nDuring the final phrase, alternate between very short moving close-ups of the three characters, one character per shot. Show an eye turn, hand gripping a sword, flowing hair or a blade catching light. On the final major hit, reveal the title “MINIMAX” through sharp diagonal masks, sword-line wipes and fractured graphic panels.\n\nEnd on a clean ensemble composition with all three characters together. Keep subtle motion in their hair, clothing, petals and light reflections.\n\nTRANSITIONS\nUse beat-synchronized slash cuts, angular panel snaps, moon-mask wipes, petal streaks, torn-paper shutters, ink-impact cuts, foreground wipes and oversized title masks. No soft dissolves, slow fades, liquid morphing or random transitions.\n\nTYPOGRAPHY\nUse only the title “MINIMAX.” Keep it sharp, elegant, premium and fully readable. No credits, extra text, random symbols or broken lettering.\n\nFINAL RESULT\nA fast, polished and rhythmically precise anime opening with expressive character motion, strong graphic editing, strict visual consistency and no duplicated characters within the same shot.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2083612295146455516",
    "sourceUrl": "https://x.com/aimikoda/status/2083611951003734471",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-01T17:52:55.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083611951003734471.mp4",
    "posterUrl": "/posters/x/2083611951003734471.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created a 15-second anime opening using MiniMax H3 by @Hailuo_AI.\n\nReally impressive model. I used six image references and one audio reference. I’ll share the prompt in the replies so you can check it out.\n\n#MiniMaxH3",
    "engagement": {
      "replies": 22,
      "reposts": 7,
      "likes": 224,
      "views": 21987,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083583963512062019",
    "title": "低机位追踪的时尚短片",
    "titleEn": "A Low-Angle Fashion Tracking Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "低机位追踪的时尚短片的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a low-angle fashion tracking film on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Ultra realistic cinematic video, shot on ARRI Alexa 35 with spherical lens, natural daylight, photorealistic, looks like real footage filmed on location, high-end fashion film aesthetic, extreme detail, realistic skin texture, real fabric physics, subtle film grain. Continuous low angle tracking shot, camera smoothly moving backward as a beautiful young woman with long dark wind-blown hair walks and performs down a narrow sun-drenched urban alley between old textured stone buildings under deep blue sky. She wears a flowing milky white asymmetrical off-shoulder top and high-waist wide-leg technical cotton culottes with a thin rope belt. The wide pants and fabric move naturally and dramatically with the wind and her body. She sings with genuine joy, smiling and softly laughing, maintaining strong eye contact with the camera, while performing free, light dance elements — fluid arm gestures, body swaying, and elegant full turns around herself as she continues walking forward. Her movement is energetic yet effortless, full of lightness and positive energy. Cinematic natural lighting with strong sunlight and deep shadows, realistic motion blur, high dynamic range, masterpiece, best quality, no artificial look.",
    "promptSourceUrl": "https://x.com/Kiber_Alla/status/2083583966578098464",
    "sourceUrl": "https://x.com/Kiber_Alla/status/2083583963512062019",
    "sourceLabel": "X 原帖 · @Kiber_Alla",
    "author": "Kiber Alla",
    "publishedAt": "2026-08-01T16:01:42.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083583963512062019.mp4",
    "posterUrl": "/posters/x/2083583963512062019.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One of the most interesting models I’ve tested recently is definitely #MiniMaxH3 @Hailuo_AI \n\nVery convenient to work with. If you haven’t tried it yet — highly recommend. Worth the attention. \n\nPrompt below👇",
    "engagement": {
      "replies": 13,
      "reposts": 3,
      "likes": 61,
      "views": 1506,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083582886280561048",
    "title": "暴风雨中的海盗船大战",
    "titleEn": "A Pirate Battle in a Raging Storm",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暴风雨中的海盗船大战的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a pirate battle in a raging storm on X. Native-video output: 10s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Open ocean, storm-lit sky, two massive wooden galleons broadside to broadside, cannons blazing, sails torn and smoking.\n\nLOCATION: towering waves, rain-slicked decks, splintering wood, the ships as looming wooden fortresses dwarfing the men fighting on them. Grey-green storm light, cannon smoke rolling across the water like fog.\n\nACTORS: two pirate crews — one crew in ragged dark coats and bandanas led by a scarred captain wielding a cutlass and flintlock; the other crew in stolen naval coats, more disciplined but outnumbered. Dozens of background fighters clashing across both decks.\n\nMOVEMENT: heavy, physical combat — swinging on rigging ropes across the gap between ships, cutlass clashes with real weight and follow-through, men thrown by cannon blasts, boots slipping on wet planks.\n\nINTERACTION: the two captains as the emotional throughline — circling each other through the chaos, their duel the spine of the sequence while the crews fight around them.\n\nCINEMATOGRAPHY (Kubrick-style): open with a long, wide, almost painterly establishing shot of the two ships closing in through the storm — hold it, let the dread build before the first cannon fires. Use symmetry in the initial approach (ships mirrored on either side of frame), then shatter that order completely once battle erupts. Favor a few extended tracking shots through the melee (a single unbroken move weaving across the deck) rather than only quick cuts, so the audience feels physically present in the chaos, punctuated by hard cuts at moments of violence.\n\nCUTS:\n\n1. Wide establishing shot: the two galleons converging through rain and swell, lightning flashing behind torn sails, ominous and silent but for wind and creaking timber. Hold long.\n\n2. Close-up: the pirate captain's face, rain streaming down a scarred cheek, eyes narrowing, hand tightening on his cutlass hilt.\n\n3. Wide shot: cannons fire in unison from both ships — deafening flash and smoke, the hulls shuddering, wood splintering on impact, rigging snapping and crashing to the deck.\n\n4. Grappling hooks fly across the gap, biting into the enemy ship's rail; ships lurch together with a groan of timber as the gap closes.\n\n5. Long tracking shot: pirates swing across on ropes into the fray, landing mid-swing into combat — camera follows one pirate's full arc from rail to rail without cutting, weapons clashing all around as he lands and immediately engages.\n\n6. Fast cuts through the melee: cutlass parries, a pistol shot at close range, a man thrown overboard by a cannon's recoil, barrels rolling loose across the tilting deck.\n\n7. Medium shot: the two captains spot each other across the chaos, a beat of stillness in the middle of battle, then they cut a path toward one another through the fighting.\n\n8. Close-up duel: blades clash in tight, brutal exchanges, rain and sweat, snarled dialogue, one captain slammed back against the mast.\n\n9. Wide shot: a mast cracks and begins to fall, crew scattering, fire catching on torn sailcloth, chaos peaking as both crews scramble.\n\n10. Final close combat: the rival captain disarmed, blade at his throat — beat of hesitation, thunder cracks, and the scene holds on the standoff rather than resolving cleanly, smoke and rain swirling around them both.\n\nCamera: mostly handheld-feel tracking shots through the melee for immersion, hard cuts on cannon fire and key hits, a still symmetrical wide shot to open",
    "promptSourceUrl": "https://x.com/seirdotmk/status/2083582942714990765",
    "sourceUrl": "https://x.com/seirdotmk/status/2083582886280561048",
    "sourceLabel": "X 原帖 · @seirdotmk",
    "author": "Myron AI artist",
    "publishedAt": "2026-08-01T15:57:25.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083582886280561048.mp4",
    "posterUrl": "/posters/x/2083582886280561048.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Epic Pirate Sea Battle in @Hailuo_AI and #MiniMaxH3\nPrompt Below",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 180,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083555537912729607",
    "title": "暴风雨里的黏土鲸鱼",
    "titleEn": "A Claymation Whale in a Storm",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暴风雨里的黏土鲸鱼的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 10 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a claymation whale in a storm on X. Native-video output: 10s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Claymation. A whale breaches from a stormy clay sea and rises in a slow-motion arc above the waves, body fully extended, water sheeting off its flanks. At the top of the arc it rolls onto one flank, hangs for a beat, then falls back toward the surface. Midair, the camera sweeps beneath its belly in a bold dynamic move, revealing the vast churning ocean far below and the full stretch of its clay body against the sky, then holds as it crashes down and white clay foam bursts outward across the frame. Visible fingerprints, tool marks, and material imperfections in the clay. Slightly stepped, handmade movement. The sea is sculpted clay with rolling modelled waves and white clay foam. Audio: a deep rush of water, wind, the dull slap of clay on clay, distant gulls. Composition: subject centered, action contained within the central third of the frame.",
    "promptSourceUrl": "https://x.com/shikoba_86/status/2083555542253838569",
    "sourceUrl": "https://x.com/shikoba_86/status/2083555537912729607",
    "sourceLabel": "X 原帖 · @shikoba_86",
    "author": "Marco \"Shikoba\" Riccetti",
    "publishedAt": "2026-08-01T14:08:45.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083555537912729607.mp4",
    "posterUrl": "/posters/x/2083555537912729607.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Prompt share for Minimax H3 on @itsPolloAI \n\ngreat for claymotion video too, right?\n\nPrompt in the comments ⬇️",
    "engagement": {
      "replies": 6,
      "reposts": 1,
      "likes": 21,
      "views": 645,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083553619413209379",
    "title": "蓝发少年与灵狐逃生",
    "titleEn": "A Blue-Haired Hero Escapes with a Spirit Fox",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝发少年与灵狐逃生的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a blue-haired hero escapes with a spirit fox on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use the reference @ image for the exact character design, clothing, proportions, blue hair, pointed ears and glowing cyan spirit fox. Preserve the cinematic stylized 3D look, saturated teal jungle palette, soft volumetric light and detailed natural textures. No dialogue; environmental sound only.\n\n0–4 seconds: Begin close to the reference composition on the ancient stone bridge. The girl kneels beside the small glowing spirit fox as ripples begin moving backward across the stream below. She notices the unnatural movement and looks over her shoulder. A massive moss-covered stone serpent suddenly erupts from beneath the bridge, lifting one side of the platform. The stone tilts physically under her weight. She loses balance, catches herself with one hand and quickly pulls the spirit fox against her chest as cracks spread beneath her knees.\n\n4–10 seconds: The bridge collapses progressively behind her. She rises naturally from the crouch and sprints barefoot across the tilting stone slabs, carrying the fox securely with one arm. Each footfall dislodges moss and small fragments; broken slabs fall with believable weight and strike the water below, producing delayed splashes. The camera tracks closely beside her at waist height. The serpent sweeps its heavy stone tail across her path. She drops into a fast knee slide beneath it; her torn cloak and blue hair trail behind with clear inertia. Still moving forward, she catches a hanging vine with her free hand. Her momentum pulls the vine taut and swings her outward across the widening gap while the camera arcs alongside her.\n\n10–15 seconds: At the highest point of the swing, the vine begins tearing from the temple ceiling. She releases before it snaps, travels forward in a natural gravity-driven arc and lands hard on the opposite platform. She absorbs the impact through her feet and one hand, then rolls once across her shoulder while protecting the fox against her chest. The camera drops and rolls partially with her, then stabilizes behind her as she rises into a defensive crouch. The spirit fox leaps from her arms and lands in front of her, its cyan light intensifying as the serpent crashes into the broken bridge behind them, sending water, dust and stone fragments through the backlight.\n\nMaintain one continuous route and clear cause-and-effect motion. No teleporting, floating, sudden directional changes, extra characters, costume changes or transformation of the girl. Keep the fox small throughout; only its light becomes brighter in the final moment.",
    "promptSourceUrl": "https://x.com/HBCoop_/status/2083553640544149812",
    "sourceUrl": "https://x.com/HBCoop_/status/2083553619413209379",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-08-01T14:01:08.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083553619413209379.mp4",
    "posterUrl": "/posters/x/2083553619413209379.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I love this clip I generated with MiniMax H3 and a character reference.\n\nPrompt below:",
    "engagement": {
      "replies": 9,
      "reposts": 2,
      "likes": 28,
      "views": 2605,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083525563491524882",
    "title": "金继剑刃的无缝循环",
    "titleEn": "A Seamless Kintsugi Sword Loop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金继剑刃的无缝循环的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a seamless kintsugi sword loop on X. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "#MiniMaxH3 in Chinese:\n\n15秒 | 16:9 | 1440p | 24fps | 无缝循环\n\n[角色锁定] 完全按照 @image1 呈现，不进行任何重新设计。蓬松的蓝灰色波浪短发，刘海垂落遮住上半张脸；面容清秀，唇色偏暖；珊瑚朱红色宽袖长袍，腰间层层缠绕的宽腰带，深色内衬；持剑手戴深色手套。长剑：华丽的银金巴洛克护手与柄头，镜面剑身映出真实的天空与云。全片不得改动发量与卷曲走向、刘海遮眼的方式、袍身轮廓、腰带缠法、手套、护手纹样与剑身比例。\n\n[循环 · 核心设定] 剑身自第一帧起就已布满一张固定的旧金脉之网——那是它此前碎裂并被金继修补留下的痕迹。本片是这一循环中的又一次：它沿着既有金脉再次裂开，碎片散去，熔金重新灌回同样的裂缝，恢复成完全相同的样子。不增添任何新金脉、不改变其走向与数量。最后一帧须与第一帧完全一致。\n\n[剑的恒定 · 最高优先级] 重铸后的剑必须与开场完全相同：长度、宽度、弧度、剑尖、护手雕花、柄头、缠柄、镜面质感、金脉走向数量粗细、刻痕位置，逐一对应。严禁任何形态变化——不得变长变大、不得生出新纹饰或棱刺、不得产生光晕或能量外溢、不得改变颜色材质、不得出现任何“进化”“觉醒”“强化形态”式转变。金只填回原有裂缝，绝不溢出剑身之外、绝不覆盖镜面。\n\n[解剖锁定 · 最高优先级] 全片有且仅有一名人物：两条手臂、两只手、每只手五指、一把剑。严禁多余的手臂、手掌、手指或第二把武器。高速转身与运动模糊中，残影绝不可被渲染成额外肢体。手部每帧结构完整、指节分明；戴手套的手握柄，空出的手须清晰可见且有明确朝向。\n\n[沉浸 · 观众入画] 画面本身是一张真实存在的和纸，镜头是观众伸进这个世界的一只眼睛，世界会反过来影响它。\n一、画面裂纹：0.9 秒剑开裂的同一帧，一道极细的裂纹自画面中心向四角蔓延，如同观看的这块玻璃也被划开；裂纹全片存在并缓慢加深，10.8 秒起被金继之金一并填补，13.8 秒前完全消失。\n二、纸的边界：画面四边的和纸纤维始终微微起伏，重击时被掀起并向内卷曲，露出纸的毛边。\n三、墨上镜头：3.2 秒的长弧甩出的墨点直接溅在镜头上，附着后沿纸纤维缓慢晕开，随后逐渐变淡，13.8 秒前完全干透消失。\n四、倒影中的观众：2.2 秒与 13.4 秒各有一片碎片旋转至极近处，其镜面中短暂映出镜头本身与其外的一片天光，仿佛观众看见了自己所在的位置——只出现一瞬即转开。\n五、沉入：7.0 至 8.8 秒，墨云直扑镜头并将其吞没，镜头沉入墨中——四周被浓墨包裹，仅上方透出天光与她的剪影，纸纤维如絮状物在“水中”缓缓漂过，画面轻微起伏如同随液体晃动；8.8 秒墨向下退去，镜头破“水面”而出。\n所有以上效果必须在 13.8 秒前彻底清除，最后一帧的画面须与第一帧一样干净。\n\n[场景 · 明亮] 高空之上，正午。头顶通透饱和的湛蓝天空，向地平线渐变为明亮浅色薄雾；脚下极远处是被阳光照亮的洁白积云海。她悬浮其中——无地面、无建筑、无道具。金箔长带随上升气流飘过。全片阳光充沛、明亮通透，绝不转阴、绝不去饱和。\n\n[色彩] 天空 #2E6FA8 渐变至 #CFE3EF。长袍珊瑚朱红 #D9563F，褶皱 #A63A2A。头发蓝灰 #8CA3B8，边缘接暖色轮廓光。手套与内衬 #14161C。护手做旧银金。碎片映出真实蓝天白云。墨为纯黑 #0A0A0C。金脉为哑光金箔 #C9962E，熔金前沿炽白金 #FFEBB0，冷却后回到完全相同的 #C9962E。\n\n[风格与质感] 水墨画与照相级玻璃的融合。整个画面覆盖极淡的和纸纤维质感；天空与云以水墨渲染，边缘晕开带枯笔飞白；人物与剑为完全清晰的实体渲染，如实体嵌在画中。金为真实金箔——极薄、有褶皱、氧化磨损、受光不均。碎片有真实厚度与内部裂纹，每片映出不同的云。发丝逐根渲染，逆光半透明；袍料厚重，褶皱有真实重量与回弹；护手雕花凹处压暗。胶片高光滚降，金上柔和光晕，细腻颗粒。\n\n[特效] 逐级升级：剑锋所过之处墨如入水般翻卷晕开；金箔自长带剥落飘散；墨与金彼此排斥绝不相融，如大理石纹互相推挤；高速动作拖出枯笔飞白擦痕；每次重击有一帧纯墨黑剪影；熔金沿裂缝奔流，前沿炽白，其后冷却为哑光金箔。\n\n[动画法则] 每个动作遵循预备—爆发—收势：出手前明确反向蓄势，爆发只占少数帧，收势有减速与轻微过冲后回位，绝不匀速。发丝、宽袖、袍摆、腰带末端各以不同延迟跟随身体，绝不同步启停。所有轨迹为弧线。高空中一切略带浮力感。杜绝任何僵硬、停顿或不自然的关节角度。\n\n[逐拍动作]\n0.0-1.4 她双手垂落，右手握剑于身侧偏后，剑斜指下方，头微低、刘海遮眼；仅呼吸使胸口微起伏。剑沿一道既有金脉裂开，手指本能收紧一下。\n1.4-3.2 剑沿所有金脉同时裂开脱手，碎片向外散开，断口边缘皆为原有的金。右手仍维持握持姿势停在半空，指节微张；左手抬至胸前，掌心向上；头始终微低。\n3.2-5.2 右手将空柄自身侧向斜上方划出完整长弧，腰胯先行带动肩背，宽袖甩开翻卷；收势时手腕轻微过冲后回位，左手向后平展保持平衡。\n5.2-7.0 侧身转体半周，重心自右脚移至左脚，袍摆螺旋展开；转身中右手将空柄换至身前反握，左手抬至面侧。\n7.0-8.8 虚空中一蹬，身体略下沉蓄势，双臂猛地向两侧张开召回碎片；随后双臂收拢至胸前，掌心相对，碎片锁定成环。\n8.8-10.8 一次真实的挥臂与转身，被不同渲染风格切分呈现。\n10.8-13.0 右手缓缓翻转、掌心向上再合拢，碎片沿手臂旋绕归位；剑成形时五指依次收紧扣柄，手臂承力下沉半寸，肩线压低。\n13.0-13.8 第一次抬起头，刘海被气流分开露出双眼，平视镜头；手腕转动一周使金脉迎光。\n13.8-15.0 缓缓垂下手腕，剑回到身侧偏后斜指下方；头重新低下，刘海落回遮眼，左手垂落身侧——完全回到第一帧姿势。\n\n[表演] 全片沉静克制。碎裂时无惊慌，只有指节一次收紧；召回时下颌微抬、呼吸加深；金脉合拢时嘴角极细微的松动；抬头对视时眼神平静笃定；垂下头时回归最初的静默。\n\n[署名] 剑身护手上方刃根处刻有 CIA0（大写 C、大写 I、大写 A、数字零；末位是数字零而非字母 O），小号宽字距镌刻大写，是剑本身的一部分，绝非叠加图层。刻痕内自第一帧起已嵌金，重铸后完全相同，绝不变亮变大或增添装饰。\n\n[文字 · 墨落纸上] 所有文字为湿润毛笔书法渗入画面本身的和纸纤维：落笔带刷毛痕迹与枯笔断白，墨沿纤维晕开一拍后随纸干淡去。拉丁文为厚重镌刻衬线，日文为浓墨重笔竖排于侧三分之一。文字位于她身后并被遮挡，绝不覆盖面部。以下须原样逐字呈现：\n0.8秒「墨」竖排右三分之一\n2.4秒 \"IT HAS BROKEN BEFORE\" 单行靠左\n5.0秒 \"EVERY CRACK\" / \"REMEMBERS THE SKY\" 两行堆叠靠左\n7.6秒「金継」竖排左三分之一，墨将干时自笔画渗出金色\n10.6秒 \"AND WILL AGAIN\" 先落墨再被金箔灌满\n13.2秒 居中标题组：「天墨」上方，\"INKFALL\" 下方，须在 14.4 秒前完全淡去，最后一帧不得残留任何文字\n\n[摄影] 变形宽银幕，椭圆焦外，暖色条状光斑，180 度快门，轻微桶形畸变。摇臂有质量、起停缓入缓出；甩镜末端轻微过冲后回稳；移焦有极短过冲再回拉。前景始终有金箔长带、碎片或袍袖掠过制造视差。\n0.0-1.4 100mm 微距沿镜面剑身向护手行进，云影流过钢面，金脉与刻痕掠过。\n1.4-3.2 35mm 环绕碎片场 180 度，穿行于碎片之间。\n3.2-5.2 24mm 广角随长弧上升，天空大面积入画。\n5.2-7.0 18mm 一镜到底，推入一片碎片的倒影穿过，自另一侧高处回望她。\n7.0-8.8 沉入墨中，镜头略微上仰仰望她的剪影，随液体轻微起伏。\n8.8-10.8 焦段随拍点跳切，85mm 特写与 24mm 广角交替。\n10.8-13.0 50mm 缓慢推入，荷兰角自五度回正。\n13.0-13.8 85mm 面部特写。\n13.8-15.0 自面部缓缓下降推近，沿剑身回到 100mm 微距，与第一帧完全对位。\n\n[分镜]\n0.0-1.4 循环点。微距沿镜面剑身疾行，云影流过钢面。剑沿一道既有金脉缓慢裂开如薄冰崩解；同一帧，一道细裂纹自画面中心向四角蔓延。墨自裂口涌出黑色羽状云。\n1.4-3.2 悬停。镜头环绕漂浮的碎片场，每片缓缓自转，各自捧着一角天空，断口的金边闪烁；2.2 秒一片碎片旋至极近，镜面中一瞬映出镜头本身。她悬于中心，刘海遮眼。\n3.2-5.2 墨绽。她以空柄划出长弧，在整片天空拖出巨大水墨笔触，湿黑弧带枯笔飞白；甩出的墨点溅上镜头并沿纸纤维晕开。落入墨中的碎片转为剪影。\n5.2-7.0 穿镜。一镜到底：镜头推入一片碎片的倒影并穿过，自另一侧高处回望她；金箔长带掠过镜头，她的转身拖出枯笔擦痕。\n7.0-8.8 沉入。墨云直扑镜头并将其吞没——四周被浓墨包裹，仅上方透出天光与她张开双臂的剪影，纸纤维如絮状物缓缓漂过，画面随液体轻微起伏，画面裂纹在墨中被压出细密支线。8.8 秒墨向下退去，镜头破出水面，碎片同帧翻转锁成绕她缓转的环。\n8.8-10.8 水墨蒙太奇，一帧一姿态交替渲染：纸白底纯墨线稿、朱红与青蓝二色木版画帧、黑底金箔剪影、枯笔拖影帧、一帧全画面墨色反转、完整渲染。墨点与金屑持续累积绝不重置。\n10.8-13.0 金继。碎片沿手臂旋绕归位，断口一一对上；熔金沿原有裂缝灌回，前沿炽白，其后冷却为哑光金箔，最终与开场金脉完全一致，不多一道不少一道。同时画面上那道裂纹也被金一并填补、随后金色渐渐淡去。全画面光线转暖。\n13.0-13.8 对视。她第一次抬头，刘海分开露出双眼，平视镜头；13.4 秒一片余留的碎片旋过，镜面中再度一瞬映出镜头。手腕转动一周使金脉迎光。\n13.8-15.0 归位。她垂下手腕与视线，剑回到身侧，刘海落回遮眼；镜头上的墨点干透消失，画面裂纹与金色痕迹尽数退去，文字淡尽；镜头沿剑身回到开场微距。最后一帧与第一帧完全一致。\n\n[音频] 每个音效卡在准确帧上，混音首尾相接可循环。底层：高空稀薄空气感，远处持续的风，一层极淡的纸张摩挲声。\n0.0-1.4 上行金属长鸣，玻璃缓慢开裂的绵长脆响，接墨入水的低沉涌动，画面裂开时一记极细的近场脆响仿佛就在耳边。\n1.4-3.2 近乎静默——碎片自转的细微水晶轻响，墨翻卷的低频液流，一次屏住的呼吸。\n3.2-5.2 宽阔湿笔拖过纸面的摩擦，墨沿纤维扩散的细密嘶声，墨点溅上镜头的一记极近的湿响，其下尺八吐息升起。\n5.2-7.0 进入碎片时气压骤降，内部闷响的封闭音，穿出一记清亮释放，金箔长带掠过的猎猎。\n7.0-8.8 沉入——全频瞬间被厚重低通滤波吞没，只余低沉的液体涌动、遥远变形的风声与一次缓慢的心跳；8.8 秒破水而出，高频与空气感轰然回归。\n8.8-10.8 八记毛笔击打卡在每次剪辑上，墨点飞溅，风格间以短滤波门分隔，反转帧一记太鼓。\n10.8-13.0 温暖低鸣升起，熔金奔流的柔和噼啪，每道裂缝闭合一声清亮短鸣，古筝音阶逐级上行。\n13.0-13.8 延长的尺八长音，剑身转动的清亮金属鸣响。\n13.8-15.0 一切收束——袖摆落定的柔响，一次平缓呼气，解析为与开场完全相同的高空风声与纸声，无缝接回起点。\n配乐：极简温暖，仅古筝与尺八。3.2 秒进入，7.0 至 8.8 秒随沉入被一同滤掉，8.8 秒回归，10.8 秒起弦乐加入，13.8 秒退出仅余风声，与开场首音无缝相接。无对白。\n\n[限制] 只有一名人物，两条手臂，两只手，一把剑。剑自第一帧即带金脉，重铸后与第一帧完全相同，严禁任何形态、尺寸、纹饰、颜色或光效变化。墨保持纯黑、金保持金箔，绝不混合成褐色。所有文字须为湿墨渗入纸面，绝非平贴叠加。镜头上的墨、画面裂纹与金痕须在 13.8 秒前彻底清除。刻痕须准确拼写为 CIA0。除上述文字外无任何标识、水印或字幕。无地面、无建筑、无道具。全片明亮饱和。首尾帧完全一致，两帧中都不得残留任何文字或镜头污迹。",
    "promptSourceUrl": "https://x.com/Cia0_exe/status/2083525568377876535",
    "sourceUrl": "https://x.com/Cia0_exe/status/2083525563491524882",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-08-01T12:09:39.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083525563491524882.mp4",
    "posterUrl": "/posters/x/2083525563491524882.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "You'll be drowning with Ai animation this, INKFALL by MiniMax H3 - 4,984 Chinese prompt",
    "engagement": {
      "replies": 17,
      "reposts": 4,
      "likes": 40,
      "views": 3398,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083501727836225758",
    "title": "同一提示词的双模型奇幻对决",
    "titleEn": "A Fantasy Face-Off from the Same Prompt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "同一提示词的双模型奇幻对决的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1080×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a fantasy face-off from the same prompt on X. Native-video output: 15s · 1080×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second photorealistic cinematic dark-fantasy battlefield sequence in 16:9 widescreen. The video contains exactly two shots connected by one clean hard cut at approximately 11 seconds. No dissolves, morphing transitions or montage editing.\n\nMASTER CREATURE LOCK:\nShow exactly one colossal ash-white bat-dragon with wyvern anatomy. The creature has one broad skull-like head, two glowing ember-red eyes, one enormous mouth filled with long dark teeth, pale hairless scaled skin, a powerful chest, two muscular hind legs, one long narrow tail and one pair of gigantic bat-like wings. Its wings have strong finger-like bones supporting huge semi-translucent pale membranes. The wings glow warm orange when backlit by the battlefield fires.\n\nThe bat-dragon has no separate forelegs or human arms. Its wings are its only front limbs. Its design, colour, face, size, anatomy and wing structure remain identical throughout both shots. It never transforms, multiplies, changes colour or grows additional limbs.\n\nBATTLEFIELD AND ARMY:\nA vast apocalyptic fantasy battlefield beneath turbulent charcoal-grey storm clouds. The ground is dark, broken and covered with ash, embers, scorched rocks and drifting smoke. Large walls of orange fire burn across the distant horizon, strongly backlighting the approaching bat-dragon.\n\nHundreds of shadowy horned warriors crowd the battlefield. The warriors wear dark rough armour and simple horned helmets. Keep the warriors anonymous, visually similar and mostly seen as small silhouettes. No individual warrior is a main character and no faces require close detail. The army fills the foreground and middle distance to establish the impossible scale of the flying creature.\n\nUse a cold grey-blue environment contrasted with intense orange firelight, glowing embers and warm translucent wing membranes. The environment must remain clearly visible through the smoke.\n\nSHOT 1 — COLOSSAL LOW FLYOVER, 0–11 SECONDS:\nBegin with a low ground-level ultra-wide camera positioned inside the tightly packed horned army, looking toward the burning horizon. Foreground warriors partially frame the bottom and sides of the image. The handheld camera moves slightly between the crowd while always looking forward.\n\nThrough the smoke and fire, the silhouette of one colossal bat-dragon appears in the sky. Its enormous folded wing shape initially resembles two mountains rising through the flames. The dragon then opens both wings to their full symmetrical width, completely dominating the sky. Bright orange fire shines through the pale wing membranes while the creature’s red eyes become visible.\n\nThe dragon performs one slow, immensely powerful downward wingbeat and descends toward the battlefield. The wingbeat pushes smoke outward in a huge circular pattern. The army looks upward and begins moving in panic as the creature rapidly approaches.\n\nThe bat-dragon glides extremely low above the battlefield without landing. Its chest and head pass just above the horned warriors. It banks slightly, bringing one enormous wing close over the foreground camera while the other wing remains fully visible and correctly attached. The wing itself does not physically strike the warriors.\n\nThe pressure wave created by the low flyover blasts across the army. Ash, embers, loose weapons, cloth and small pieces of debris surge sideways through the frame. Foreground warriors are knocked off balance, lifted briefly and scattered by the powerful wind, tumbling away from the creature without graphic impact or injury. Keep the movement chaotic but physically believable.\n\nThe camera shakes violently from the wing pressure and quickly pans to follow the dragon’s low flight across the battlefield. Flames bend beneath the airflow and smoke rolls outward behind the creature. Maintain a clear view of the same bat-dragon throughout the flyover. The dragon must not disappear completely behind its own wing or the smoke.\n\nSHOT 2 — FRONTAL ROARING CHARGE, 11–15 SECONDS:\nAt approximately 11 seconds, make one clean hard cut to a frontal low-angle telephoto shot directly in the bat-dragon’s flight path. The creature is already flying forward through thick smoke and towering flames, coming straight toward the camera.\n\nKeep both enormous wings visible on either side of its body as they perform one powerful coordinated wingbeat. The glowing red eyes remain locked directly on the camera. Burning embers stream around the creature while the orange flames behind it illuminate the translucent wing membranes.\n\nAs the bat-dragon rapidly closes the distance, it lowers its skull-like head and opens its enormous mouth, revealing multiple rows of long sharp teeth. It releases a deafening roar directly toward the camera. The force of the roar pushes smoke and embers outward from its face.\n\nThe camera retreats slightly while the dragon continues approaching, creating intense forward momentum. Finish just before impact with the creature’s roaring face, glowing eyes and wings dominating the frame. The dragon does not hit or pass through the camera.\n\nCAMERA AND MOTION:\nShot one uses a low handheld battlefield perspective with strong scale, foreground silhouettes, realistic parallax and a fast pan following the flyover. Shot two uses a steady frontal tracking camera that retreats as the dragon approaches. Use cinematic motion blur on the wing tips, airborne ash and scattered debris, but keep the creature’s face and body sharply readable. The motion must feel fast, heavy and dangerous rather than weightless.\n\nVISUAL QUALITY:\nPremium big-budget dark-fantasy visual effects, photorealistic creature texture, physically accurate wing membranes, believable muscles moving beneath pale skin, enormous scale, detailed fire and smoke simulations, dense volumetric atmosphere, realistic aerodynamic pressure, strong depth, natural motion blur, sharp creature details and dramatic fire backlighting.\n\nAUDIO:\nNo dialogue and no music. Roaring battlefield fire, panicked distant army sounds, rushing wind, deep leathery wingbeats, armour rattling, debris scraping across the ground, heavy low-frequency pressure wave and one terrifying creature roar during the final frontal approach.\n\nSTRICT CONSISTENCY:\nExactly one bat-dragon. No additional flying creatures. No rider. Exactly one head, two eyes, one mouth, two wings, two hind legs and one tail. No extra wings, arms, heads, faces or tails. The wings must remain attached to the same creature and must not turn into hands. The dragon never lands, breathes fire or changes design. Keep both wings intact and symmetrical. The horned warriors remain separate from the dragon and do not merge into its body. No visible casualties, blood, gore, text, subtitles, logos or real-world symbols.",
    "promptSourceUrl": "https://x.com/MadMax_Series/status/2083501727836225758",
    "sourceUrl": "https://x.com/MadMax_Series/status/2083501727836225758",
    "sourceLabel": "X 原帖 · @MadMax_Series",
    "author": "MadMax",
    "publishedAt": "2026-08-01T10:34:56.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083501727836225758.mp4",
    "posterUrl": "/posters/x/2083501727836225758.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo MiniMax H3 vs. Seedance 2.0 \n\nUsed same Text to Video Prompt, Prompt given Below\n\nIn my opinion Seedance 2.0 is still better as comparatively but its way cheaper than Seedance 2.0 depending on the platforms. \n\nFor example, MiniMax H3 (2K) would cost around 1.5 dollars and Seedance 2.0 (1080) would cost around 5.8 Dollars. \n\nIt depends on your video complexity, if its simple would recommend MiniMax H3 and its cost effective. \n\nMade it on @Flovaai \n\nPrompt: \nCreate a 15-second photorealistic cinematic dark-fantasy battlefield sequence in 16:9 widescreen. The video contains exactly two shots connected by one clean hard cut at approximately 11 seconds. No dissolves, morphing transitions or montage editing.\n\nMASTER CREATURE LOCK:\nShow exactly one colossal ash-white bat-dragon with wyvern anatomy. The creature has one broad skull-like head, two glowing ember-red eyes, one enormous mouth filled with long dark teeth, pale hairless scaled skin, a powerful chest, two muscular hind legs, one long narrow tail and one pair of gigantic bat-like wings. Its wings have strong finger-like bones supporting huge semi-translucent pale membranes. The wings glow warm orange when backlit by the battlefield fires.\n\nThe bat-dragon has no separate forelegs or human arms. Its wings are its only front limbs. Its design, colour, face, size, anatomy and wing structure remain identical throughout both shots. It never transforms, multiplies, changes colour or grows additional limbs.\n\nBATTLEFIELD AND ARMY:\nA vast apocalyptic fantasy battlefield beneath turbulent charcoal-grey storm clouds. The ground is dark, broken and covered with ash, embers, scorched rocks and drifting smoke. Large walls of orange fire burn across the distant horizon, strongly backlighting the approaching bat-dragon.\n\nHundreds of shadowy horned warriors crowd the battlefield. The warriors wear dark rough armour and simple horned helmets. Keep the warriors anonymous, visually similar and mostly seen as small silhouettes. No individual warrior is a main character and no faces require close detail. The army fills the foreground and middle distance to establish the impossible scale of the flying creature.\n\nUse a cold grey-blue environment contrasted with intense orange firelight, glowing embers and warm translucent wing membranes. The environment must remain clearly visible through the smoke.\n\nSHOT 1 — COLOSSAL LOW FLYOVER, 0–11 SECONDS:\nBegin with a low ground-level ultra-wide camera positioned inside the tightly packed horned army, looking toward the burning horizon. Foreground warriors partially frame the bottom and sides of the image. The handheld camera moves slightly between the crowd while always looking forward.\n\nThrough the smoke and fire, the silhouette of one colossal bat-dragon appears in the sky. Its enormous folded wing shape initially resembles two mountains rising through the flames. The dragon then opens both wings to their full symmetrical width, completely dominating the sky. Bright orange fire shines through the pale wing membranes while the creature’s red eyes become visible.\n\nThe dragon performs one slow, immensely powerful downward wingbeat and descends toward the battlefield. The wingbeat pushes smoke outward in a huge circular pattern. The army looks upward and begins moving in panic as the creature rapidly approaches.\n\nThe bat-dragon glides extremely low above the battlefield without landing. Its chest and head pass just above the horned warriors. It banks slightly, bringing one enormous wing close over the foreground camera while the other wing remains fully visible and correctly attached. The wing itself does not physically strike the warriors.\n\nThe pressure wave created by the low flyover blasts across the army. Ash, embers, loose weapons, cloth and small pieces of debris surge sideways through the frame. Foreground warriors are knocked off balance, lifted briefly and scattered by the powerful wind, tumbling away from the creature without graphic impact or injury. Keep the movement chaotic but physically believable.\n\nThe camera shakes violently from the wing pressure and quickly pans to follow the dragon’s low flight across the battlefield. Flames bend beneath the airflow and smoke rolls outward behind the creature. Maintain a clear view of the same bat-dragon throughout the flyover. The dragon must not disappear completely behind its own wing or the smoke.\n\nSHOT 2 — FRONTAL ROARING CHARGE, 11–15 SECONDS:\nAt approximately 11 seconds, make one clean hard cut to a frontal low-angle telephoto shot directly in the bat-dragon’s flight path. The creature is already flying forward through thick smoke and towering flames, coming straight toward the camera.\n\nKeep both enormous wings visible on either side of its body as they perform one powerful coordinated wingbeat. The glowing red eyes remain locked directly on the camera. Burning embers stream around the creature while the orange flames behind it illuminate the translucent wing membranes.\n\nAs the bat-dragon rapidly closes the distance, it lowers its skull-like head and opens its enormous mouth, revealing multiple rows of long sharp teeth. It releases a deafening roar directly toward the camera. The force of the roar pushes smoke and embers outward from its face.\n\nThe camera retreats slightly while the dragon continues approaching, creating intense forward momentum. Finish just before impact with the creature’s roaring face, glowing eyes and wings dominating the frame. The dragon does not hit or pass through the camera.\n\nCAMERA AND MOTION:\nShot one uses a low handheld battlefield perspective with strong scale, foreground silhouettes, realistic parallax and a fast pan following the flyover. Shot two uses a steady frontal tracking camera that retreats as the dragon approaches. Use cinematic motion blur on the wing tips, airborne ash and scattered debris, but keep the creature’s face and body sharply readable. The motion must feel fast, heavy and dangerous rather than weightless.\n\nVISUAL QUALITY:\nPremium big-budget dark-fantasy visual effects, photorealistic creature texture, physically accurate wing membranes, believable muscles moving beneath pale skin, enormous scale, detailed fire and smoke simulations, dense volumetric atmosphere, realistic aerodynamic pressure, strong depth, natural motion blur, sharp creature details and dramatic fire backlighting.\n\nAUDIO:\nNo dialogue and no music. Roaring battlefield fire, panicked distant army sounds, rushing wind, deep leathery wingbeats, armour rattling, debris scraping across the ground, heavy low-frequency pressure wave and one terrifying creature roar during the final frontal approach.\n\nSTRICT CONSISTENCY:\nExactly one bat-dragon. No additional flying creatures. No rider. Exactly one head, two eyes, one mouth, two wings, two hind legs and one tail. No extra wings, arms, heads, faces or tails. The wings must remain attached to the same creature and must not turn into hands. The dragon never lands, breathes fire or changes design. Keep both wings intact and symmetrical. The horned warriors remain separate from the dragon and do not merge into its body. No visible casualties, blood, gore, text, subtitles, logos or real-world symbols.",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 10,
      "views": 919,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083461207919612348",
    "title": "拉面试吃的真实 UGC 质感",
    "titleEn": "Tasting Ramen in Authentic UGC Style",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拉面试吃的真实 UGC 质感的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 1280×544。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for tasting ramen in authentic ugc style on X. Native-video output: 15s · 1280×544. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Duration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, slight motion blur, TikTok/Reels energy — NOT cinematic.\nProduct Reference: Use the uploaded ramen bowl image as the only product reference. Keep the broth color, noodle thickness, toppings, egg placement, chashu slices, and bowl appearance identical throughout every shot.\nCamera: iPhone 15 Pro handheld with front/back camera switching, realistic autofocus hunting, natural handheld movement, slight exposure shifts, warm apartment lighting.\nCharacter Description\nName: Yuna\nA young Japanese woman in her early 20s with soft natural beauty, shoulder-length dark hair tied loosely, subtle makeup, oversized gray hoodie, cheerful personality, filming casually inside a cozy apartment kitchen.\nShot Breakdown\nSHOT 1 (0–2s) — Selfie holding the ramen bowl.\nDialogue: \"Late-night ramen again.\"\nSHOT 2 (2–4s) — Overhead reveal with steam rising.\nSHOT 3 (4–6s) — Close-up autofocus on noodles and broth.\nSHOT 4 (6–8s) — Chopsticks lifting noodles with a natural noodle stretch.\nSHOT 5 (8–10s) — Selfie bite reaction.\nDialogue: \"That's insanely good.\"\nSHOT 6 (10–12s) — Casual b-roll of the bowl while she reaches for another bite.\nSHOT 7 (12–14s) — Selfie holding noodles toward camera.\nDialogue: \"Worth every bite.\"\nSHOT 8 (14–15s) — Freeze frame with handwritten overlay: \"ramen fix 🍜\"\nLook & Feel\nWarm lighting, natural phone quality, authentic UGC pacing, jump cuts, realistic handheld movement.\nNegative Prompt\ncinematic look, luxury commercial, perfect framing, CGI noodles, unrealistic steam, plastic food, distorted chopsticks, extra fingers, warped hands, overprocessed HDR, text artifacts.",
    "promptSourceUrl": "https://x.com/Taaruk_/status/2083461207919612348",
    "sourceUrl": "https://x.com/Taaruk_/status/2083461207919612348",
    "sourceLabel": "X 原帖 · @Taaruk_",
    "author": "Taaruk",
    "publishedAt": "2026-08-01T07:53:55.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083461207919612348.mp4",
    "posterUrl": "/posters/x/2083461207919612348.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×544",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "dark",
      "commercial"
    ],
    "scenes": [
      "dialogue",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "From the steam to the textures, every detail feels incredibly real. MiniMax H3 is redefining AI-generated food videos. 🍜 \n\nEvery shot in this video was generated with MiniMax H3 by \n @Hailuo_AI. #MiniMaxH3 \n\nPrompt:\nDuration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, slight motion blur, TikTok/Reels energy — NOT cinematic.\nProduct Reference: Use the uploaded ramen bowl image as the only product reference. Keep the broth color, noodle thickness, toppings, egg placement, chashu slices, and bowl appearance identical throughout every shot.\nCamera: iPhone 15 Pro handheld with front/back camera switching, realistic autofocus hunting, natural handheld movement, slight exposure shifts, warm apartment lighting.\nCharacter Description\nName: Yuna\nA young Japanese woman in her early 20s with soft natural beauty, shoulder-length dark hair tied loosely, subtle makeup, oversized gray hoodie, cheerful personality, filming casually inside a cozy apartment kitchen.\nShot Breakdown\nSHOT 1 (0–2s) — Selfie holding the ramen bowl.\nDialogue: \"Late-night ramen again.\"\nSHOT 2 (2–4s) — Overhead reveal with steam rising.\nSHOT 3 (4–6s) — Close-up autofocus on noodles and broth.\nSHOT 4 (6–8s) — Chopsticks lifting noodles with a natural noodle stretch.\nSHOT 5 (8–10s) — Selfie bite reaction.\nDialogue: \"That's insanely good.\"\nSHOT 6 (10–12s) — Casual b-roll of the bowl while she reaches for another bite.\nSHOT 7 (12–14s) — Selfie holding noodles toward camera.\nDialogue: \"Worth every bite.\"\nSHOT 8 (14–15s) — Freeze frame with handwritten overlay: \"ramen fix 🍜\"\nLook & Feel\nWarm lighting, natural phone quality, authentic UGC pacing, jump cuts, realistic handheld movement.\nNegative Prompt\ncinematic look, luxury commercial, perfect framing, CGI noodles, unrealistic steam, plastic food, distorted chopsticks, extra fingers, warped hands, overprocessed HDR, text artifacts.",
    "engagement": {
      "replies": 21,
      "reposts": 0,
      "likes": 35,
      "views": 5235,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083436822085165289",
    "title": "汉堡试吃的第一视角短片",
    "titleEn": "Tasting a Burger from a First-Person View",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "汉堡试吃的第一视角短片的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2944×1248。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for tasting a burger from a first-person view on X. Native-video output: 15s · 2944×1248. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Duration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, TikTok/Reels aesthetic.\nProduct Reference: Use the uploaded gourmet burger image as the only product reference. Preserve the bun shape, patty thickness, cheese melt, lettuce, tomato, sauces, and proportions exactly in every shot.\nCharacter Description\nName: Hana\nA young Japanese woman in her early 20s with natural beauty, long dark hair in a loose ponytail, oversized cream sweatshirt, minimal makeup, bright smile, friendly lifestyle-vlogger personality.\nShot Breakdown\nSHOT 1 (0–2s) — Selfie showing the burger box.\nDialogue: \"Burger night!\"\nSHOT 2 (2–4s) — Opens the box.\nSHOT 3 (4–6s) — Quick zoom on the burger.\nSHOT 4 (6–8s) — Hands lifting the burger with cheese stretching naturally.\nSHOT 5 (8–10s) — Bite reaction.\nDialogue: \"Okay... that's incredible.\"\nSHOT 6 (10–12s) — Casual close-up b-roll while reaching for fries.\nSHOT 7 (12–14s) — Toasting the burger toward the camera.\nDialogue: \"You need this.\"\nSHOT 8 (14–15s) — Freeze frame with overlay: \"burger cravings = solved 🍔\"\nLook & Feel\nWarm apartment lighting, genuine phone footage, slight grain, natural autofocus breathing, handheld imperfections, fast jump cuts.\nNegative Prompt\ncinematic grading, commercial production, CGI burger, fake cheese, distorted hands, warped food, perfect stabilization, studio lighting, text glitches, logo distortion.",
    "promptSourceUrl": "https://x.com/oggii_0/status/2083436822085165289",
    "sourceUrl": "https://x.com/oggii_0/status/2083436822085165289",
    "sourceLabel": "X 原帖 · @oggii_0",
    "author": "Oogie",
    "publishedAt": "2026-08-01T06:17:01.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083436822085165289.mp4",
    "posterUrl": "/posters/x/2083436822085165289.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "dark",
      "commercial"
    ],
    "scenes": [
      "dialogue",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "POV: You ordered a burger... and it exceeded every expectation. 🍔🤤\n\nShot entirely in authentic UGC style with MiniMax H3.\n\n@Hailuo_AI #MiniMaxH3\n\nPrompt: Duration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, TikTok/Reels aesthetic.\nProduct Reference: Use the uploaded gourmet burger image as the only product reference. Preserve the bun shape, patty thickness, cheese melt, lettuce, tomato, sauces, and proportions exactly in every shot.\nCharacter Description\nName: Hana\nA young Japanese woman in her early 20s with natural beauty, long dark hair in a loose ponytail, oversized cream sweatshirt, minimal makeup, bright smile, friendly lifestyle-vlogger personality.\nShot Breakdown\nSHOT 1 (0–2s) — Selfie showing the burger box.\nDialogue: \"Burger night!\"\nSHOT 2 (2–4s) — Opens the box.\nSHOT 3 (4–6s) — Quick zoom on the burger.\nSHOT 4 (6–8s) — Hands lifting the burger with cheese stretching naturally.\nSHOT 5 (8–10s) — Bite reaction.\nDialogue: \"Okay... that's incredible.\"\nSHOT 6 (10–12s) — Casual close-up b-roll while reaching for fries.\nSHOT 7 (12–14s) — Toasting the burger toward the camera.\nDialogue: \"You need this.\"\nSHOT 8 (14–15s) — Freeze frame with overlay: \"burger cravings = solved 🍔\"\nLook & Feel\nWarm apartment lighting, genuine phone footage, slight grain, natural autofocus breathing, handheld imperfections, fast jump cuts.\nNegative Prompt\ncinematic grading, commercial production, CGI burger, fake cheese, distorted hands, warped food, perfect stabilization, studio lighting, text glitches, logo distortion.",
    "engagement": {
      "replies": 23,
      "reposts": 5,
      "likes": 94,
      "views": 16021,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083407991114805571",
    "title": "八镜头奢华护肤广告",
    "titleEn": "An Eight-Shot Luxury Skincare Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "八镜头奢华护肤广告的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for an eight-shot luxury skincare commercial on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use the uploaded storyboard as the camera guide. Follow all 8 frames in order with smooth cinematic continuity. The same woman, outfit, hairstyle, makeup, room and product must remain perfectly consistent throughout the entire 15-second commercial. Every shot must begin where the previous shot ends with seamless match cuts and natural camera movement.\n\nShot 1 (0-1.8s)\nA slow push in as morning sunlight enters through the window. The woman looks outside and gently turns toward the mirror. The camera continues moving without stopping.\n\nTransition:\nThe camera keeps pushing forward until her face fills the frame.\n\nShot 2 (1.8-3.6s)\nWithout cutting, the shot becomes a close-up of her face in the mirror. She notices dull skin and lightly touches her cheek. The camera slowly circles around her as she reaches toward the vanity.\n\nTransition:\nFollow her hand naturally to the serum bottle.\n\nShot 3 (3.6-5.4s)\nMacro shot as she picks up the serum bottle. The bottle rotates slightly in her hand while natural sunlight reflects across the glass. The camera smoothly follows the bottle upward.\n\nTransition:\nContinue tracking into an extreme macro of the dropper.\n\nShot 4 (5.4-7.2s)\nExtreme macro of the dropper lifting from the bottle. A single drop forms and falls in slow motion. The camera follows the drop as it lands on her fingertips.\n\nTransition:\nThe camera continues following her fingertips toward her face.\n\nShot 5 (7.2-9.0s)\nClose-up as she gently applies the serum across her cheeks using smooth circular motions. Skin catches soft natural highlights. Hair moves slightly in the breeze.\n\nTransition:\nThe camera slowly rotates around her face while maintaining eye level.\n\nShot 6 (9.0-11.1s)\nThe lighting naturally becomes warmer as her skin appears healthier and more radiant. The transformation happens gradually, not instantly. She smiles softly while looking into the mirror.\n\nTransition:\nShe turns away from the mirror and walks toward the window as the camera follows.\n\nShot 7 (11.1-13.2s)\nMedium shot as she confidently walks into warm sunlight. She smiles naturally at the camera while the background softly blurs. The camera slowly arcs around her.\n\nTransition:\nShe lowers the product toward the camera until it fills the frame.\n\nShot 8 (13.2-15.0s)\nHero product shot. The serum bottle stands on a clean surface with warm natural daylight. Soft reflections move across the glass. Premium end card with elegant typography and subtle camera push in. End on the product with \"Shop Now.\"\n\nOverall style:\nUltra photorealistic luxury skincare commercial, Apple-style cinematography, ARRI Alexa 65, 50mm and 85mm lenses, soft natural daylight, realistic skin texture, premium color grading, physically accurate lighting, smooth cinematic camera movement, seamless match cuts, invisible transitions, no jump cuts, no frame skipping, no character changes, no outfit changes, no background changes, premium beauty ad quality.\n\n#MiniMaxH3 @Hailuo_AI",
    "promptSourceUrl": "https://x.com/Maercihh/status/2083407991114805571",
    "sourceUrl": "https://x.com/Maercihh/status/2083407991114805571",
    "sourceLabel": "X 原帖 · @Maercihh",
    "author": "Maercih",
    "publishedAt": "2026-08-01T04:22:27.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083407991114805571.mp4",
    "posterUrl": "/posters/x/2083407991114805571.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A luxury skincare commercial, created entirely with MiniMax H3.\n\nPrompt:\nUse the uploaded storyboard as the camera guide. Follow all 8 frames in order with smooth cinematic continuity. The same woman, outfit, hairstyle, makeup, room and product must remain perfectly consistent throughout the entire 15-second commercial. Every shot must begin where the previous shot ends with seamless match cuts and natural camera movement.\n\nShot 1 (0-1.8s)\nA slow push in as morning sunlight enters through the window. The woman looks outside and gently turns toward the mirror. The camera continues moving without stopping.\n\nTransition:\nThe camera keeps pushing forward until her face fills the frame.\n\nShot 2 (1.8-3.6s)\nWithout cutting, the shot becomes a close-up of her face in the mirror. She notices dull skin and lightly touches her cheek. The camera slowly circles around her as she reaches toward the vanity.\n\nTransition:\nFollow her hand naturally to the serum bottle.\n\nShot 3 (3.6-5.4s)\nMacro shot as she picks up the serum bottle. The bottle rotates slightly in her hand while natural sunlight reflects across the glass. The camera smoothly follows the bottle upward.\n\nTransition:\nContinue tracking into an extreme macro of the dropper.\n\nShot 4 (5.4-7.2s)\nExtreme macro of the dropper lifting from the bottle. A single drop forms and falls in slow motion. The camera follows the drop as it lands on her fingertips.\n\nTransition:\nThe camera continues following her fingertips toward her face.\n\nShot 5 (7.2-9.0s)\nClose-up as she gently applies the serum across her cheeks using smooth circular motions. Skin catches soft natural highlights. Hair moves slightly in the breeze.\n\nTransition:\nThe camera slowly rotates around her face while maintaining eye level.\n\nShot 6 (9.0-11.1s)\nThe lighting naturally becomes warmer as her skin appears healthier and more radiant. The transformation happens gradually, not instantly. She smiles softly while looking into the mirror.\n\nTransition:\nShe turns away from the mirror and walks toward the window as the camera follows.\n\nShot 7 (11.1-13.2s)\nMedium shot as she confidently walks into warm sunlight. She smiles naturally at the camera while the background softly blurs. The camera slowly arcs around her.\n\nTransition:\nShe lowers the product toward the camera until it fills the frame.\n\nShot 8 (13.2-15.0s)\nHero product shot. The serum bottle stands on a clean surface with warm natural daylight. Soft reflections move across the glass. Premium end card with elegant typography and subtle camera push in. End on the product with \"Shop Now.\"\n\nOverall style:\nUltra photorealistic luxury skincare commercial, Apple-style cinematography, ARRI Alexa 65, 50mm and 85mm lenses, soft natural daylight, realistic skin texture, premium color grading, physically accurate lighting, smooth cinematic camera movement, seamless match cuts, invisible transitions, no jump cuts, no frame skipping, no character changes, no outfit changes, no background changes, premium beauty ad quality.\n\n#MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 52,
      "reposts": 2,
      "likes": 97,
      "views": 8713,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083301476206588086",
    "title": "蓝色影棚中的人马共舞",
    "titleEn": "A Woman, a Man, and a Horse in a Blue Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝色影棚中的人马共舞的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a woman, a man, and a horse in a blue studio on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use @ Image1 as the man reference. \nUse @ Image2 as the woman reference. \nUse @ Image3 as the horse reference.  \n\nCreate a surreal 15-second fashion-film sequence inside a minimalist monochrome blue studio.  The woman performs an elegant contemporary dance.  The man casually walks into frame wearing a completely serious expression.  Without explanation, he immediately joins the choreography.  His movements are awkward, strangely confident, almost ritualistic.  The woman never acknowledges him.  The horse behaves as though both dancers are part of its routine.  It wanders through the studio.  Runs behind them.  Appears from impossible places.  Stops.  Looks directly into the camera.  Leaves.  Returns again.  Sometimes all three performers accidentally synchronize for one beat before drifting apart again.  The choreography becomes increasingly absurd but remains perfectly cinematic.  Every movement is intentional.  Every frame feels designed.  Throughout the sequence, premium kinetic typography appears as part of the composition, not subtitles, but editorial motion graphics inspired by high-end After Effects title design.  Animated words flash for only a few frames:  THE GARDENER.  THE DRIVER.  THE HORSE.  THE WOMAN.  KNIFE.  MOVE.  FASTER.  AGAIN.  THE EXIT. These words slam into frame, wrap around the dancers, disappear behind bodies, stretch with motion blur, rotate in 3D, fragment into pieces, reveal through masks, pulse with the bass, smear across cuts, become oversized, tiny, vertical, upside down, and constantly transform using premium motion-design techniques.  Add abstract editorial graphics: animated arrows, circles, blueprint lines, tape labels, interface markers, halftone textures, scribbles, grain, tracking graphics, geometric overlays, chromatic aberration, liquid masks, flash frames, split screens, typography trails, speed ramps and graphic wipes.  Music evolves into strange electronic art-pop.  Analog synths.  Mechanical breathing.  Heavy bass.  Everything is synchronized to the rhythm.  Final close-up.  The woman whispers:  \"Again.\"  The man quietly smiles for the first time.  The horse suddenly sprints directly toward the camera.  The word  AGAIN.  fills the screen before an instant cut to black.",
    "promptSourceUrl": "https://x.com/egeberkina/status/2083301476206588086",
    "sourceUrl": "https://x.com/egeberkina/status/2083301476206588086",
    "sourceLabel": "X 原帖 · @egeberkina",
    "author": "Ege",
    "publishedAt": "2026-07-31T21:19:12.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083301476206588086.mp4",
    "posterUrl": "/posters/x/2083301476206588086.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One of my first MiniMax H3 tests\n\nOmni Reference worked really well here. I only provided references for the woman, the man and the horse and then used the prompt below 👇\n\nPrompt: \nUse @ Image1 as the man reference. \nUse @ Image2 as the woman reference. \nUse @ Image3 as the horse reference.  \n\nCreate a surreal 15-second fashion-film sequence inside a minimalist monochrome blue studio.  The woman performs an elegant contemporary dance.  The man casually walks into frame wearing a completely serious expression.  Without explanation, he immediately joins the choreography.  His movements are awkward, strangely confident, almost ritualistic.  The woman never acknowledges him.  The horse behaves as though both dancers are part of its routine.  It wanders through the studio.  Runs behind them.  Appears from impossible places.  Stops.  Looks directly into the camera.  Leaves.  Returns again.  Sometimes all three performers accidentally synchronize for one beat before drifting apart again.  The choreography becomes increasingly absurd but remains perfectly cinematic.  Every movement is intentional.  Every frame feels designed.  Throughout the sequence, premium kinetic typography appears as part of the composition, not subtitles, but editorial motion graphics inspired by high-end After Effects title design.  Animated words flash for only a few frames:  THE GARDENER.  THE DRIVER.  THE HORSE.  THE WOMAN.  KNIFE.  MOVE.  FASTER.  AGAIN.  THE EXIT. These words slam into frame, wrap around the dancers, disappear behind bodies, stretch with motion blur, rotate in 3D, fragment into pieces, reveal through masks, pulse with the bass, smear across cuts, become oversized, tiny, vertical, upside down, and constantly transform using premium motion-design techniques.  Add abstract editorial graphics: animated arrows, circles, blueprint lines, tape labels, interface markers, halftone textures, scribbles, grain, tracking graphics, geometric overlays, chromatic aberration, liquid masks, flash frames, split screens, typography trails, speed ramps and graphic wipes.  Music evolves into strange electronic art-pop.  Analog synths.  Mechanical breathing.  Heavy bass.  Everything is synchronized to the rhythm.  Final close-up.  The woman whispers:  \"Again.\"  The man quietly smiles for the first time.  The horse suddenly sprints directly toward the camera.  The word  AGAIN.  fills the screen before an instant cut to black.",
    "engagement": {
      "replies": 15,
      "reposts": 26,
      "likes": 237,
      "views": 21798,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083298802413555943",
    "title": "动态字体融入咖啡广告",
    "titleEn": "Kinetic Typography Inside a Coffee Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动态字体融入咖啡广告的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2944×1248。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for kinetic typography inside a coffee commercial on X. Native-video output: 15s · 2944×1248. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15-second premium luxury coffee commercial for the coffee brand \"INSOMNIA\". Create a cinematic editorial film, not a traditional advertisement. The visual language blends high-end fashion campaigns, premium product films, and modern graphic design. Every shot feels intentionally art-directed and professionally edited. Deep matte black, espresso brown, warm amber highlights, soft volumetric lighting, glossy reflections, subtle film grain, rich contrast, and an expensive cinematic color grade. Fast editorial rhythm using only crisp hard cuts synchronized with the soundtrack. No dissolves or soft transitions. Integrate bold typography naturally into the scene instead of placing it like subtitles. Massive words such as \"INSOMNIA\", \"NO SLEEP\", \"STAY AWAKE\", and \"STAY INSPIRED\" appear as architectural elements, illuminated walls, floating design layers, reflections on glass, or oversized background graphics while the camera moves around the product. Typography should remain clean, sharp, readable, and premium. Mix typography styles including giant bold headlines, thin technical labels, vertical text, rotated captions, editorial annotations, product specifications, registration marks, minimal grid lines, and stylish barcode-inspired graphic elements used only as design features, not functional barcodes. Use smooth masking reveals, frame drawing animations, split layouts, sliding typography, geometric wipes, and magazine-style collage compositions while maintaining realistic live-action footage. Show premium roasted coffee beans, aromatic steam, rich espresso pouring in macro detail, a matte black luxury coffee cup, dramatic close-ups, elegant camera movement, and refined product beauty shots. The entire film should feel like an international luxury fashion campaign directed by a top creative agency, combining cinematic realism with sophisticated editorial motion graphics. Finish with the INSOMNIA logo centered on a black background with the tagline \"Stay Awake. Stay Inspired.\" in elegant minimal typography.\n\nCreated using MiniMax H3 on @openart_ai",
    "promptSourceUrl": "https://x.com/pewdenai/status/2083298802413555943",
    "sourceUrl": "https://x.com/pewdenai/status/2083298802413555943",
    "sourceLabel": "X 原帖 · @pewdenai",
    "author": "pewden AI",
    "publishedAt": "2026-07-31T21:08:34.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083298802413555943.mp4",
    "posterUrl": "/posters/x/2083298802413555943.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made this coffee commercial with integrated typography using MiniMax H3 on @openart_ai  \nFull prompt at the end 👇\n\nThe typography came through clean the text sat naturally inside the scene and the overall output had a strong editorial brand feel. The color grading and product shots also held up really well for a first run.\n\nThis was only the first test and I'm already excited to experiment with a lot more ideas. Can't wait to share what's coming next.\n\nPrompt:\n15-second premium luxury coffee commercial for the coffee brand \"INSOMNIA\". Create a cinematic editorial film, not a traditional advertisement. The visual language blends high-end fashion campaigns, premium product films, and modern graphic design. Every shot feels intentionally art-directed and professionally edited. Deep matte black, espresso brown, warm amber highlights, soft volumetric lighting, glossy reflections, subtle film grain, rich contrast, and an expensive cinematic color grade. Fast editorial rhythm using only crisp hard cuts synchronized with the soundtrack. No dissolves or soft transitions. Integrate bold typography naturally into the scene instead of placing it like subtitles. Massive words such as \"INSOMNIA\", \"NO SLEEP\", \"STAY AWAKE\", and \"STAY INSPIRED\" appear as architectural elements, illuminated walls, floating design layers, reflections on glass, or oversized background graphics while the camera moves around the product. Typography should remain clean, sharp, readable, and premium. Mix typography styles including giant bold headlines, thin technical labels, vertical text, rotated captions, editorial annotations, product specifications, registration marks, minimal grid lines, and stylish barcode-inspired graphic elements used only as design features, not functional barcodes. Use smooth masking reveals, frame drawing animations, split layouts, sliding typography, geometric wipes, and magazine-style collage compositions while maintaining realistic live-action footage. Show premium roasted coffee beans, aromatic steam, rich espresso pouring in macro detail, a matte black luxury coffee cup, dramatic close-ups, elegant camera movement, and refined product beauty shots. The entire film should feel like an international luxury fashion campaign directed by a top creative agency, combining cinematic realism with sophisticated editorial motion graphics. Finish with the INSOMNIA logo centered on a black background with the tagline \"Stay Awake. Stay Inspired.\" in elegant minimal typography.\n\nCreated using MiniMax H3 on @openart_ai",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 9,
      "views": 808,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083282581450375367",
    "title": "骑手、摩托与十五秒节拍",
    "titleEn": "A Rider, a Motorcycle, and a Fifteen-Second Beat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "骑手、摩托与十五秒节拍的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a rider, a motorcycle, and a fifteen-second beat on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use @ image1 as the opening frame and exact visual reference. Preserve the rider’s face, short curly hair, round glasses, black leather jacket, gloves, motorcycle, and realistic nighttime city environment throughout.\n\nUse @ audio1 as the soundtrack. Synchronize the motorcycle’s acceleration, camera movement, passing lights, and edits precisely to the music’s rhythm and rising intensity. No dialogue or additional music.\n\n0–3s: Begin in the frontal composition of @ image1. A stabilized tracking camera moves backward directly ahead of the motorcycle as the rider speeds through traffic. Her expression is alert and focused; wind moves her curls and jacket naturally.\n\n3–6s: On a strong musical beat, shift smoothly into a low front three-quarter tracking angle beside the front wheel. The motorcycle leans gently around a moving car without changing direction abruptly. Streetlights streak across the wet pavement.\n\n6–10s: Transition into a fast side-tracking shot. The rider accelerates through a narrow opening between two lanes while maintaining believable balance, speed, and forward momentum. Background traffic and illuminated buildings create controlled motion blur.\n\n10–13s: The camera arcs from her side toward the rear as she clears the traffic. Red emergency lights begin flashing across the surrounding buildings and reflect naturally in her glasses and leather jacket.\n\n13–15s: A fire truck crosses the intersection behind her as she races into an open stretch of road. Finish with the camera pulling wider and rising slightly, revealing the motorcycle speeding deeper into the city on the final musical accent.\n\nCinematic live-action realism, physically accurate motorcycle movement, energetic but readable action, natural motion blur, shallow depth of field, cool blue-black city palette with warm streetlights and red emergency-light accents. Maintain one continuous direction of travel. No collisions, teleporting, sudden direction reversals, distorted hands, changing wardrobe, altered motorcycle design, facial drift, extra limbs, text, logos, or slow motion.",
    "promptSourceUrl": "https://x.com/HBCoop_/status/2083282593592869102",
    "sourceUrl": "https://x.com/HBCoop_/status/2083282581450375367",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-07-31T20:04:07.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083282581450375367.mp4",
    "posterUrl": "/posters/x/2083282581450375367.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 test using one reference image, an audio track and a timed 15-second prompt.\n\n• Maintained the rider well and followed most of the requested camera progression.\n\n• Followed the sequence of events but varied the timing of each planned beat.\n\nPrompt below:",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 16,
      "views": 2148,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083240416166748313",
    "title": "千禧流行女团说唱片段",
    "titleEn": "A Y2K K-Pop Girl-Group Rap Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "千禧流行女团说唱片段的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a y2k k-pop girl-group rap sequence on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "This is what H3 does with Omni Reference and one long prompt:\n\nprompt：Soft cute Y2K crush K-pop girl group rap MV. High fashion performance film mixed with inflated 3D candy typography graphic system. Three female idols wearing pink, blue and purple luxury Y2K stage outfits. Cute 3D chibi doll face, big dewy eyes, porcelain skin, soft blush cheeks, chibi idol proportions. Reference typography: only the inflated 3D rubber bubble font language, pink/lavender/ice-blue chrome plastic material, not the whole layout. Visual style: soft pastel crush energy, Y2K game UI, candy glossy finish, milky translucent plastic × K-pop aggressive feminine rap energy. Texture: 35mm film grain + soft glow bloom + iridescent holographic foil + jelly plastic specular + Y2K scanline + candy dot halftone + VHS frame vibration + plastic toy surface reflection + glitter confetti rain. Editing: Fast rhythm editing. Only hard cuts. No fade. No smooth transitions. Every cut follows: bass hit, snare, rap vocal impact.\n\nCharacter Performance Three female performers from reference image 1. Keep: same 3D chibi doll faces, big sparkling eyes, pastel Y2K idol makeup, pink/blue/purple fashion styling, soft cute cold expression. Each member has a signature prop: A holds a chunky rhinestone microphone, B has a plush bear shoulder bag, C wears oversized heart-shaped sunglasses pushed up on head. They are not posing. They are performing rap. Actions: aggressive eye contact, head nodding, shoulder movement, leaning toward camera, hand pushing toward lens, confident cute facial expression, sharp body movement, wink on snare hit, tongue-out on bass drop, hair flip on beat. Lip sync: perfect rap mouth synchronization. The performance should feel like a real K-pop rap unit MV with Y2K crush energy.\n\nTypography Animation Typography is a visual weapon. Not subtitles. Not static graphics. Use: giant English words inflated 3D rubber bubble font, jelly plastic material, pink lavender ice-blue chrome, soft puffy candy letters, compressed bold balloon fonts, stretched vertical puffy letters, handwritten graffiti titles, 3D extruded candy text, glitch displacement, print error effect, Y2K game UI pixel corruption. Text appears only when the beat hits. Characters can cover the text. Text can cover parts of clothing. Never cover eyes or main facial expression. Tiny Y2K stickers burst on beat: heart, star, smiley, sparkle, chain link, bow, four-leaf clover, mini bear.\n\nShot 1 — Member A · Extreme Close-up「RUN NOW」 Character A face close-up, framed inside a pink heart-shaped iris wipe transition (in only, hard cut out). She looks directly into camera. Cute cold doll expression, big dewy eyes, soft blush, frosted pink lip gloss, rhinestone tear sticker under one eye. Starts rap. Chunky rhinestone microphone enters frame from below, she grips it tightly. Camera slowly pushes closer. Tiny pink heart stickers float up from bottom edge like bubbles. Huge inflated bubble letters crash into frame from left and right. Typography: \"RUN NOW\" — Pink jelly rubber 3D font, glossy candy surface, soft inner glow. Letters stretch vertically like puffy balloons. The word vibrates and squishes with bass, casts soft pink shadow on her cheek. Rap: \"Run now, I don't chase.\" Confetti burst on final syllable. Hard cut.\n\nShot 2 — Member B · Rap Performance「BREAK LIMIT」 Member B medium close-up, slight low angle, camera orbits 15° around her during the take. Lavender fur jacket, Y2K layered top, plush bear bag bouncing on her hip, chunky pearl choker, star-shaped earrings. She performs aggressively. Shoulders hit the rhythm. Hair flips hard on the second snare, blue hair ties catch the light. Hair moves naturally. Background: pastel gradient pink-to-lavender with faint Y2K perspective grid lines floating. Behind her appears huge handwritten text. Typography: \"BREAK LIMIT\" — Ice-blue bubble graffiti brush writing, soft inflated stroke, candy-coated, letters appear one by one with bounce easing. Animation: letters tear apart, reassemble, rotate 360° on snare, shake with bass. Chain link graphic swings across frame, sparkle stickers pop on every beat. Rap: \"Break limits, break rules.\" Hard cut.\n\nShot 3 — Member C · Hand Attack Shot「NO APOLOGY」 Member C moves close to camera. Top-down camera angle, she leans over the lens. Hand reaches toward lens. Nails are long pink acrylic with 3D star and heart charms. Focus: rings with mini plush bear charm, heart-cut gemstone, baby-pink leather texture, knit ribbed sleeve, plushy toy surface. Foreground vertical text: \"NO APOLOGY\" — Stretched vertical pink bubble letters, jelly puffy material, letters pierce through the back of her hand visually. Hand covers part of the typography. Fingers spread wide on the bass hit. Camera: handheld vibration. Film grain strong. Candy halftone dots bloom in shadow. Tiny \"XOXO\" sticker tag pulses in corner. Rap: \"No apology, no regret.\" Hard cut.\n\nShot 4 — Fashion Magazine Impact Cut「UNSTOPPABLE」 Member A close-up. Her face appears inside a broken Y2K trading card / idol photo card frame, with barcode at bottom, \"SYS.07-A\" serial tag, holographic foil border torn on left side. Not a full layout. Only one graphic element. Typography: \"UNSTOPPABLE\" — Huge lavender handwritten bubble text, inflated 3D rubber, glossy plastic finish, letters explode outward from center of her forehead, push edges of the card frame. Animation: text suddenly expands from behind her. Card frame rips in half vertically, holographic confetti pours from the tear, revealing pastel gradient background underneath. A second smaller photo card of her winks in from the right, slightly rotated. Rap: \"Can't stop my wave.\" Hard cut.\n\nShot 5 — Glitch Double Image「STATIC」 Member B close-up, slight Dutch angle (5° tilt). Create: duplicate frame, pastel RGB separation (pink/blue/lavender channels), xerox shadow, motion trail, VHS tracking line at bottom edge. Main face stays sharp. She holds a sparkly heart-shaped lollipop up beside her cheek, the lollipop leaves a trail of echo frames. Background text: \"STATIC\" — Y2K pixel font, candy pink chrome, letters flicker like broken game sprite, jump between frames, sometimes inverted color, sometimes missing entirely. Like corrupted music video signal + cute game glitch. A tiny \"ERROR 404\" sticker blinks in upper corner. Rap: \"Static noise, I get louder.\" Hard cut.\n\nShot 6 — Pink Blue Purple Battle「PINK OUT」 Member A and C alternate close-ups, rhythm: A-A-C-A-C-C, two frames each. Fast switching. Visual split: each member sits in a vertical color block (A=pink, C=lavender, alternating). Center text: \"PINK OUT\" — Pastel pink-to-purple gradient bubble 3D, letters appear one word at a time, syncing to each cut. Words appear one by one. Each word: stretch vertically on entry, compress horizontally on hold, invert color on snare, explode into heart stickers and star confetti on exit. Tiny \"1.2.3.GO!\" counter graphic ticks in corner. Rap: \"Lights out, pink out.\" Hard cut.\n\nShot 7 — Group Rap Performance「UNBREAKABLE」 Three members together.",
    "promptSourceUrl": "https://x.com/LeoCreaIA/status/2083240429412286683",
    "sourceUrl": "https://x.com/LeoCreaIA/status/2083240416166748313",
    "sourceLabel": "X 原帖 · @LeoCreaIA",
    "author": "Leo",
    "publishedAt": "2026-07-31T17:16:34.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083240416166748313.mp4",
    "posterUrl": "/posters/x/2083240416166748313.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is out and the thing that jumps at you is the typography.\n\nMost video models still melt text the moment there's motion. This holds inflated 3D lettering across hard cuts, on beat, without turning to soup. Native audio in the same pass.\n\nFull prompt below.",
    "engagement": {
      "replies": 10,
      "reposts": 2,
      "likes": 83,
      "views": 14526,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083225662316265763",
    "title": "蓝色影棚里的超现实时装",
    "titleEn": "Surreal Fashion Inside a Blue Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝色影棚里的超现实时装的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for surreal fashion inside a blue studio on X. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[Upload reference ] + Use la ragazza in reference  as a strict identity reference for the model and her yellow sunglasses.\n\nPreserve face, hair, and the exact frame shape and colour throughout.\n\n15 seconds, 16:9. Premium fashion commercial, photoreal live action.\n\nInfinite black void, polished black floor. She is the only lit element: hard key from\n\none side, second hard light from the other, deep shadow between, strong speculars on\n\nthe yellow frame and lenses.\n\nShe walks slowly toward the camera in a black high-neck top and long black coat, chin\n\nlevel, eyes on the lens. Camera tracks backward at her pace, holding her size in frame.\n\nAll type is heavy condensed uppercase sans-serif in vivid yellow, rendered as physical\n\nobjects in 3D perspective with real thickness, catching the same light as she does and\n\ncasting faint shadows on the floor.\n\n[0-4s] LOOK rises out of the floor ahead of her in deep perspective, grows as she\n\napproaches, and sweeps past on both sides out of frame.\n\n[4-7s] SEE enters from the right at high speed, streaks behind her with heavy motion\n\nblur, exits left.\n\n[7-10s] FOCUS appears large and fully out of focus, resolves into razor sharpness over\n\nhalf a second, holds, breaks apart.\n\n[10-12s] Four thin yellow lines draw a rectangle around her and FRAME assembles along\n\nits lower edge.\n\nThroughout, passing type reflects across the glossy lenses as it moves near her.\n\n[12-15s] She stops. All type collapses inward and vanishes into black. The camera\n\ncontinues into a tight close-up on her face and the yellow sunglasses. She holds\n\ncompletely still, eyes on the lens. On the black beneath her the title composes in the\n\nsame yellow type: \"SHIOBA LENS\", and below it, smaller: \"MADE TO BE SEEN\". Hold.\n\nAll text clean, legible, correctly spelled English capitals, each word once only. No\n\ngarbled or duplicated letters, misspellings, subtitles, watermarks, or other text or\n\nlogos.\n\nAudio: low sustained bass,\n\nComposition: model centered, all type within the central third of the frame",
    "promptSourceUrl": "https://x.com/shikoba_86/status/2083225666699305184",
    "sourceUrl": "https://x.com/shikoba_86/status/2083225662316265763",
    "sourceLabel": "X 原帖 · @shikoba_86",
    "author": "Marco \"Shikoba\" Riccetti",
    "publishedAt": "2026-07-31T16:17:57.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083225662316265763.mp4",
    "posterUrl": "/posters/x/2083225662316265763.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 il live on @itsPolloAI \n\nPolloAI adds a cutting-edge model, among other skills, which is excellent for motion graphics and text rendering.\n\nPrompt in comments ⬇️",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 8,
      "views": 501,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2083217899540586917",
    "title": "游乐园记忆的动态图形蒙太奇",
    "titleEn": "A Motion-Graphic Montage of Theme-Park Memories",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "游乐园记忆的动态图形蒙太奇的原生视频效果，原作者在 X 上完整公开了生成 Prompt。 原生视频规格为 5 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The original creator published the complete generation prompt for a motion-graphic montage of theme-park memories on X. Native-video output: 5s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Nostalgic montage of different clips. Graphic motion layouts of friends having fun in a theme park.",
    "promptSourceUrl": "https://x.com/magnific/status/2083217899540586917",
    "sourceUrl": "https://x.com/magnific/status/2083217899540586917",
    "sourceLabel": "X 原帖 · @magnific",
    "author": "Magnific",
    "publishedAt": "2026-07-31T15:47:06.000Z",
    "addedAt": "2026-08-22T08:11:11.985Z",
    "mediaUrl": "/media/x-2083217899540586917.mp4",
    "posterUrl": "/posters/x/2083217899540586917.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Prompt:\n\n\"Nostalgic montage of different clips. Graphic motion layouts of friends having fun in a theme park.\"\n\nAs simple as this\n\nMiniMax H3 will do the rest. Now on Magnific",
    "engagement": {
      "replies": 13,
      "reposts": 5,
      "likes": 508,
      "views": 55249,
      "snapshotAt": "2026-08-22T08:11:11.985Z"
    },
    "approvedAt": "2026-08-22T08:11:11.985Z",
    "editorialBasis": "The original X status explicitly identifies MiniMax H3, contains a playable native video, and publishes a complete verbatim prompt in the post or a same-author reply."
  },
  {
    "id": "x-2087066113775665414",
    "title": "打烊前最后一个气泡 · 下篇",
    "titleEn": "The Last Bubble Before Closing · Part 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《打烊前最后一个气泡 · 下篇》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Last Bubble Before Closing · Part 2. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "\"[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for:\n\n“THE LAST POP BEFORE CLOSING”\n\nas the STRICT PRIMARY visual and narrative reference.\n\nThis video continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n\n- Fizz’s final carbonation bubble drifting toward the ceiling fan\n- Fizz and Crunch failing to reach it\n- Chill calmly studying the situation\n- Chill noticing his silver spoon tool\n- Chill removing the spoon from his back\n- Chill saying:\n\n<d>[English] I have an idea.</d>\n\nPART 2 begins at that EXACT story moment.\n\nUse the storyboard to lock:\n\n- Fizz’s exact glossy red soda-can design\n- Crunch’s exact bright-yellow snack-bag design\n- Chill’s exact blue ice-cream-cup design\n- their faces, shoes, gloves and proportions\n- Chill’s spoon\n- the ONE surviving bubble\n- shopping basket\n- snack bowl / tray\n- freezer\n- cold mist\n- hanging price tags\n- ceiling fan\n- same convenience-store architecture\n- same nighttime lighting\n- same tiny-mascot scale\n- Part 2 rescue progression\n- final three-character payoff\n\nGenerate a FINISHED cinematic 3D animated sequence.\n\nDO NOT animate or reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- captions\n- storyboard headers\n- production notes\n- prop/reference sheets\n- character turnaround drawings\n- page background\n- watermarks\n- brand logos\n- unnecessary text overlays\n\nTranslate the storyboard’s 15 one-second panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT use fifteen frantic one-second cuts.\n\nStory progression:\n\nChill reveals the plan\n→ trio races ahead of bubble\n→ Chill uses freezer air\n→ bubble changes direction\n→ Crunch catches it temporarily\n→ bubble escapes again\n→ Chill triggers price-tag chain reaction\n→ bubble descends\n→ Fizz makes one perfect jump\n→ bubble returns\n→ Fizz recharges\n→ friends celebrate\n→ warm final shelf payoff.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 2 — 0:15 to 0:30\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium stylized 3D product-mascot animation\n\nGenre:\n\n- miniature adventure\n- product mascot comedy\n- friendship story\n- playful rescue mission\n\nTone:\n\n- energetic\n- clever\n- cute\n- visually satisfying\n- comedic\n- wholesome\n- heartwarming\n- triumphant\n\nDialogue:\nshort natural synchronized English\n\nAudio:\nnative dialogue + store ambience + physical comedy SFX + original playful cinematic score\n\nCORE PAYOFF:\n\nFizz thinks his “spark” depends on one final bubble.\n\nBut what actually saves that spark is:\n\n# teamwork.\n\nChill supplies the idea.\nCrunch supplies chaotic effort.\nFizz supplies the final leap.\n\n==================================================\nPART 2 START / END\n==================================================\n\nSTART:\n\nChill stands beneath the drifting bubble holding his silver spoon.\n\nFizz and Crunch look toward him.\n\nChill immediately begins executing his plan.\n\nEND:\n\nFizz has successfully regained his carbonation.\n\nFizz, Crunch, and Chill stand together in a warm celebratory final pose.\n\nThe story must feel COMPLETE.\n\nNo cliffhanger.\n\n==================================================\nCHARACTER IDENTITY LOCK — FIZZ\n==================================================\n\nMatch PART 1 and the uploaded PART 2 storyboard EXACTLY.\n\nFIZZ — RED SODA CAN MASCOT\n\nAppearance:\n\n- small glossy red aluminum soda can\n- cylindrical body\n- metallic silver top and bottom rims\n- silver pull tab\n- original fictional graphics only\n- large expressive blue/dark cartoon eyes\n- flexible eyebrows\n- expressive mouth\n- tiny arms\n- white mascot gloves\n- small legs\n- red-and-white sneakers\n\nMaterial:\n\n- glossy aluminum\n- strong store-light reflections\n- subtle metallic highlights\n- optional tiny condensation beads\n- body remains mostly rigid\n\nDo NOT redesign him.\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\ndesperate hope\n\n3–6 sec:\nexcited as Chill’s plan works\n\n6–8 sec:\ncelebrates too early\n\n8–11 sec:\nfocused determination\n\n11–13 sec:\nheroic relief\n\n13–15 sec:\nfull joyful Fizz energy\n\nMOTION:\n\nFizz is:\n\n- springy\n- quick\n- dramatic\n- highly expressive\n\nBut his can body should not behave like rubber.\n\n==================================================\nCHARACTER IDENTITY LOCK — CRUNCH\n==================================================\n\nCRUNCH — YELLOW SNACK BAG MASCOT\n\nAppearance:\n\n- bright yellow flexible snack bag\n- sealed ridged top\n- crinkled packaging material\n- original fictional package graphics\n- large expressive eyes\n- small expressive mouth\n- thin arms\n- white gloves\n- tiny legs\n- oversized orange sneakers\n\nPersonality:\n\n- loyal\n- excitable\n- chaotic\n- lovable\n- desperate to help\n- occasionally useful by accident\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\nhopeful and ready\n\n3–6 sec:\nheroic enthusiasm\n\n6–8 sec:\nbrief victory\n\n8–10 sec:\nimmediate panic when bubble escapes\n\n10–13 sec:\ncheering Fizz\n\n13–15 sec:\nhappy friend celebration\n\nPACKAGING MOTION:\n\nUse:\n\n- small crinkles\n- squash\n- puffing\n- top-edge wobble\n\nDo not deform him beyond recognition.\n\n==================================================\nCHARACTER IDENTITY LOCK — CHILL\n==================================================\n\nCHILL — BLUE ICE-CREAM CUP MASCOT\n\nAppearance:\n\n- small mint-blue / cool-blue cylindrical ice cream cup\n- white lid like a flat cap\n- original minimal fictional graphics\n- relaxed half-lidded eyes\n- small mouth\n- tiny arms\n- white gloves\n- tiny white shoes\n- silver spoon normally attached on back\n\nPersonality:\n\n- calm\n- intelligent\n- deadpan\n- observant\n- precise\n- quietly confident\n\nPART 2 PERFORMANCE ARC:\n\n0–4 sec:\nplan execution\n\n4–8 sec:\ncalm supervision\n\n8–11 sec:\nsecond tactical idea\n\n11–13 sec:\nwatches Fizz finish plan\n\n13–15 sec:\nsmall proud smile\n\nChill should NEVER become hyperactive.\n\nHis calmness is the joke.\n\n==================================================\nKEY PROP LOCK — THE LAST BUBBLE\n==================================================\n\nThere is:\n\n# ONE SINGLE BUBBLE.\n\nNo duplicates.\n\nAppearance:\n\n- transparent\n- iridescent\n- subtle rainbow edge\n- about the size of Fizz’s head\n- reflects store lights\n- delicate and lightweight\n\nBubble trajectory for Part 2 must remain easy to follow:\n\nceiling fan area\n→ redirected by freezer airflow\n→ descends toward Crunch\n→ briefly rests in bowl\n→ escapes\n→ drifts toward hanging price tags\n→ price-tag chain reaction guides it downward\n→ Fizz jumps\n→ bubble enters through slightly opened pull-tab area\n→ carbonation returns.\n\nNever teleport the bubble.\n\nNever spawn extra copies.\n\n==================================================\nKEY PROP LOCK — CHILL’S SPOON\n==================================================\n\nUse exactly one small silver spoon.\n\nIt begins in Chill’s hand because he removed it at the end of Part 1.\n\nUses:\n\n1. operating / assisting with freezer opening safely\n2. precisely flicking the price-tag string later\n\nThe spoon is a clever tool.\n\nIt is NOT a weapon.\n\nNo dangerous swinging.\n\nNo sharp-object action.\n\n==================================================\nKEY PROP — SHOPPING BASKET\n==================================================\n\nSmall red convenience-store shopping basket relative to human scale, but large enough for the mascots to ride in.\n\nUse it only as a comic transport device.\n\nSafe movement.\n\nNo crashes.\n\nNo dangerous speed.\n\n==================================================\nENVIRONMENT LOCK\n==================================================\n\nSame convenience store from Part 1.\n\nMaintain:\n\n- glossy tile floor\n- large aisles\n- soda shelf\n- snack aisle\n- freezers\n- candy displays\n- fruit display\n- hanging price cards\n- ceiling fan\n- warm overhead practical lights\n- cool freezer illumination\n- midnight darkness beyond storefront windows\n\nThe store should feel enormous compared with the mascots.\n\nKeep miniature cinematography throughout.\n\n==================================================\nWORLD / PHYSICS RULES\n==================================================\n\nBubble:\n\nextremely light.\n\nIt responds to:\n\n- airflow\n- small impacts\n- movement\n- gentle rebounds\n\nFreezer:\n\ncold air pushes bubble away from fan.\n\nSnack bowl:\n\nbubble can rest briefly without popping.\n\nPrice tags:\n\nsmall hanging cardboard tags swing like pendulums and redirect the bubble through gentle contact / airflow.\n\nFizz’s recovery:\n\nbubble enters Fizz through his slightly opened pull-tab opening.\n\nThen carbonation visibly rebuilds INSIDE the can through:\n\n- subtle internal fizz\n- a few tiny external bubbles\n- brighter expression\n- energetic body language\n\nDo NOT show explosive pressure.\n\nNo dangerous soda burst.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.7]\nCHILL’S PLAN BEGINS\n\nContinue EXACTLY from Part 1.\n\nBubble floats high in the background near the ceiling airflow.\n\nFizz and Crunch turn toward Chill.\n\nChill holds the silver spoon.\n\nChill looks toward:\n\nbubble\n→ freezer aisle.\n\nHe points the spoon forward like a tiny conductor.\n\nChill:\n\n<d>[English] Freezer.</d>\n\nFizz:\n\n<d>[English] The freezer?</d>\n\nChill gives a tiny confident nod.\n\nCrunch:\n\n<d>[English] I love plans I don’t understand!</d>\n\nChill starts moving.\n\nCAMERA:\n\nclose Chill hero angle\n→ rack focus to bubble\n→ fast low-angle follow as trio starts moving.\n\nMusic immediately restarts with clever playful momentum.\n\n==================================================\n[SHOT 2 — 0:01.7–0:03.3]\nTHE SHOPPING-BASKET DASH\n\nFizz and Crunch jump into / grab a small red shopping basket.\n\nChill pushes / steers it from behind or rides at the edge while controlling direction.\n\nThe basket rolls/slides smoothly across the polished store floor.\n\nUse a low mascot-height tracking shot.\n\nBubble remains visible above and ahead.\n\nCrunch points upward.\n\nCrunch:\n\n<d>[English] Faster!</d>\n\nFizz:\n\n<d>[English] It’s getting away!</d>\n\nChill:\n\n<d>[English] Relax.</d>\n\nThey pass:\n\n- oversized product shelves\n- cool refrigerator reflections\n- hanging price tags\n\nNo collision.\n\nNo store destruction.\n\n==================================================\n[SHOT 3 — 0:03.3–0:05.1]\nCOLD AIR CHANGES THE GAME\n\nThey arrive at the freezer.\n\nChill safely uses the spoon as a tiny lever/tool to pull the freezer-door edge enough to open it.\n\nDoor opens.\n\n# FWHOOSH.\n\nA beautiful plume of cold mist rolls outward.\n\nBlue-white freezer light contrasts with the warm store.\n\nFizz and Crunch brace themselves behind the basket.\n\nThe cold airflow reaches the drifting bubble.\n\nThe bubble slows.\n\nThen gently curves AWAY from the ceiling fan.\n\nFizz’s expression instantly changes.\n\nFizz:\n\n<d>[English] It’s working!</d>\n\nChill, completely calm:\n\n<d>[English] Obviously.</d>\n\nCAMERA:\n\nfreezer close-up\n→ cold mist\n→ follow airflow visually\n→ bubble changing trajectory.\n\nSFX:\n\n- freezer seal opening\n- gentle cold-air whoosh\n- faint icy shimmer\n- bubble wobble\n\n==================================================\n[SHOT 4 — 0:05.1–0:06.8]\nCRUNCH GETS HIS HERO MOMENT\n\nBubble now floats downward through the aisle.\n\nCrunch spots the snack bowl / tray.\n\nHis eyes widen.\n\nCrunch:\n\n<d>[English] My turn!</d>\n\nHe grabs the bowl.\n\nRuns underneath the bubble.\n\nFizz guides him:\n\n<d>[English] Left! Left! No—your other left!</d>\n\nCrunch shifts.\n\nThe bubble gently drops—\n\n# into the bowl.\n\nPerfect catch.\n\nCrunch freezes.\n\nLooks down.\n\nHis eyes become huge with joy.\n\nCrunch:\n\n<d>[English] I GOT IT!</d>\n\nFizz throws both arms up.\n\n==================================================\n[SHOT 5 — 0:06.8–0:08.4]\nCELEBRATED TOO EARLY\n\nFizz and Crunch start celebrating immediately.\n\nCrunch lifts the bowl proudly.\n\nBad idea.\n\nThe movement tilts it.\n\nThe bubble gently floats back out.\n\nAll three characters track it.\n\nFizz and Crunch’s smiles disappear at the exact same time.\n\nCrunch:\n\n<d>[English] ...I don’t got it.</d>\n\nFizz:\n\n<d>[English] CRUNCH!</d>\n\nThe bubble continues toward a row of hanging price tags.\n\nChill does not panic.\n\nHe simply turns his eyes upward.\n\nCamera pushes in on Chill.\n\nHe notices the tags.\n\nThen the spoon in his hand.\n\nSecond idea.\n\nTiny eyebrow raise.\n\n==================================================\n[SHOT 6 — 0:08.4–0:10.4]\nTHE PRICE-TAG CHAIN REACTION\n\nChill steps into position beneath the first hanging price tag.\n\nPrecisely:\n\nhe taps the hanging string with the spoon.\n\n# tik.\n\nFirst price tag swings.\n\nIt gently nudges the next hanging tag.\n\nSecond swings.\n\nThen third.\n\nCreate a clean satisfying cascading motion:\n\nTAG 1\n→ TAG 2\n→ TAG 3\n→ TAG 4.\n\nThe bubble moves through this chain.\n\nEach swinging tag / tiny air current guides it slightly lower.\n\nNo hard impact.\n\nNo popping.\n\nUse highly satisfying visual rhythm.\n\nFizz watches.\n\nHis eyes widen.\n\nCrunch:\n\n<d>[English] He planned THAT?!</d>\n\nChill:\n\n<d>[English] Mostly.</d>\n\nMusic syncs each tag movement to percussion:\n\ntik\ntik\ntik\ntik.\n\n==================================================\n[SHOT 7 — 0:10.4–0:12.6]\nONE PERFECT JUMP\n\nThe bubble is finally descending.\n\nFizz runs underneath it.\n\nCrunch runs behind him cheering.\n\nChill watches calmly.\n\nFizz’s expression becomes focused.\n\nNo more panic.\n\nCamera drops to dramatic low angle.\n\nFizz accelerates.\n\nHe jumps.\n\nUse a brief cinematic slow-down—not full slow motion.\n\nHis arms spread.\n\nBubble directly above him.\n\nFizz opens his silver pull tab just slightly.\n\nFizz:\n\n<d>[English] Come on...</d>\n\nThe bubble descends.\n\nCloser.\n\nCloser.\n\nThen—\n\n# PLOOP / FIZZ.\n\nIt slips neatly through the opening.\n\nFizz’s eyes widen.\n\nPause for one beat.\n\n==================================================\n[SHOT 8 — 0:12.6–0:15.0]\nFIZZ GETS HIS SPARK BACK — FINAL PAYOFF\n\nImmediately after the bubble enters:\n\n# FSSSSHHH!\n\nA playful carbonation pulse moves through Fizz.\n\nNOT an explosion.\n\nHis red aluminum surface becomes bright and lively.\n\nTiny fizzy bubbles swirl around him.\n\nHis eyes light up.\n\nHis posture pops back into full confidence.\n\nFizz lands safely.\n\nFizz:\n\n<d>[English] I’M BACK!</d>\n\nCrunch cheers wildly.\n\nCrunch:\n\n<d>[English] WE SAVED THE FIZZ!</d>\n\nChill walks up calmly.\n\nFizz looks toward both friends.\n\nFor one sincere beat:\n\nFizz:\n\n<d>[English] You guys saved me.</d>\n\nCrunch immediately poses proudly.\n\nChill:\n\n<d>[English] Technically, the bubble did.</d>\n\nSmall comedy beat.\n\nFizz laughs.\n\nA few harmless tiny carbonation bubbles float around all three.\n\nOne bubble lands perfectly on Chill’s white lid.\n\nChill slowly looks upward at it.\n\nTiny smile.\n\nFINAL COMPOSITION:\n\nFizz in the center, fully recharged.\n\nCrunch beside him, excited and proud.\n\nChill beside them, calm and satisfied.\n\nStore shelves glow warmly behind.\n\nSmall bubbles drift around them.\n\nOptional quick match transition / cut:\n\nthe three mascots back near their product shelf, posing together like the night’s adventure is their secret.\n\nCamera gently pulls backward.\n\nWarm practical lights.\n\nCool midnight shadows.\n\nPlayful final musical chord.\n\nFADE TO BLACK.\n\nEND STORY.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio tone:\n\n- playful\n- tactile\n- energetic\n- miniature\n- cinematic\n- satisfying\n- wholesome\n\n==================================================\nFIZZ VOICE\n==================================================\n\nEnergetic youthful masculine mascot voice.\n\nQualities:\n\n- charismatic\n- dramatic\n- quick\n- lovable\n- expressive\n\nPart 2 progression:\n\nhopeful urgency\n→ premature celebration\n→ focused determination\n→ huge relief\n→ grateful friendship.\n\nPriority lines:\n\n<d>[English] It’s working!</d>\n\n<d>[English] CRUNCH!</d>\n\n<d>[English] Come on...</d>\n\n<d>[English] I’M BACK!</d>\n\n<d>[English] You guys saved me.</d>\n\nFinal sincere line should be warmer and quieter than his usual show-off delivery.\n\n==================================================\nCRUNCH VOICE\n==================================================\n\nExcitable comedic mascot voice.\n\nQualities:\n\n- enthusiastic\n- slightly squeaky\n- dramatic\n- lovable\n- fast reaction timing\n\nPriority:\n\n<d>[English] My turn!</d>\n\n<d>[English] I GOT IT!</d>\n\n<d>[English] ...I don’t got it.</d>\n\n<d>[English] He planned THAT?!</d>\n\n<d>[English] WE SAVED THE FIZZ!</d>\n\nUse tiny packaging crinkles with body motion.\n\n==================================================\nCHILL VOICE\n==================================================\n\nCalm, understated, low-energy mascot voice.\n\nQualities:\n\n- dry humor\n- relaxed\n- intelligent\n- confident\n- slightly sleepy\n\nPriority lines:\n\n<d>[English] Freezer.</d>\n\n<d>[English] Obviously.</d>\n\n<d>[English] Mostly.</d>\n\nFinal:\n\n<d>[English] Technically, the bubble did.</d>\n\nDeliver completely deadpan.\n\n==================================================\nSTORE SOUND DESIGN\n==================================================\n\nMaintain subtle closed-store ambience:\n\n- refrigerator hum\n- freezer compressor\n- ventilation\n- faint building hum\n- soft sneaker taps\n- packaging crinkles\n- shopping basket wheel/slide sounds\n- tiny metal spoon clicks\n- hanging card flutter\n- distant night ambience\n\nDo not introduce humans.\n\n==================================================\nFREEZER SOUND DESIGN\n==================================================\n\nWhen freezer opens:\n\n- soft magnetic seal release\n- cold air whoosh\n- compressor hum slightly louder\n- gentle mist movement\n\nDo NOT use giant icy explosion SFX.\n\n==================================================\nPRICE TAG CHAIN SFX\n==================================================\n\nMake this extremely satisfying.\n\nUse four slightly different light sounds:\n\ntik\ntap\ntik\ntap\n\nEach synchronized to a swinging tag.\n\nAdd gentle cardboard flutter.\n\nMusic percussion should sync perfectly.\n\n==================================================\nBUBBLE SOUND DESIGN\n==================================================\n\nBubble signature:\n\n- tiny glassy shimmer\n- fizzy air texture\n- delicate elastic wobble\n\nBowl landing:\n\nsoft:\n\n“plip.”\n\nLeaving bowl:\n\ntiny airy:\n\n“woop.”\n\nReturning to Fizz:\n\nsmall:\n\n“ploop”\n→ immediately followed by carbonation:\n\n“fsssshhhh.”\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nContinue the ORIGINAL miniature adventure score from Part 1.\n\nInstrumentation:\n\n- pizzicato strings\n- marimba\n- woodblocks\n- brushed percussion\n- muted brass\n- upright bass\n- tiny synth plucks\n- glockenspiel\n- warm orchestral accents\n\nMUSIC ARC:\n\n0:00–0:03\nclever plan / movement energy.\n\n0:03–0:05\nfreezer solution reveal.\n\n0:05–0:07\ncomic triumph.\n\n0:07–0:08.5\ncomedic drop after bubble escapes.\n\n0:08.5–0:10.5\nprecise rhythmic chain-reaction sequence.\n\n0:10.5–0:12.5\nheroic but playful Fizz build.\n\n0:12.5–0:15\nwarm celebratory theme combining motifs associated with all three mascots.\n\nFinal chord:\n\nbright,\nsatisfying,\ncomplete.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 2 should feel slightly more cinematic and energetic than Part 1.\n\nUse:\n\n- floor-level tracking\n- macro bubble shots\n- low mascot-scale lensing\n- freezer push-ins\n- dramatic cool/warm lighting contrast\n- overhead shot for bowl catch\n- rhythmic insert shots during price-tag chain\n- low-angle Fizz hero jump\n- expressive reaction close-ups\n- final slow pull-back\n\nCamera progression:\n\n0–3 sec:\nquick planning/action.\n\n3–5 sec:\ncold-air spectacle.\n\n5–8 sec:\nphysical comedy.\n\n8–10 sec:\nprecise chain-reaction rhythm.\n\n10–13 sec:\nheroic focus.\n\n13–15 sec:\nwarm emotional resolution.\n\nAvoid:\n\n- shaky handheld\n- extreme motion blur\n- chaotic whip pans\n- fisheye distortion\n- unreadable fast cutting\n- action-movie explosions\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium high-end stylized 3D animation.\n\nMaterial quality:\n\nFizz:\nglossy reflective aluminum.\n\nCrunch:\nsoft flexible crinkly snack-film packaging.\n\nChill:\nsemi-matte coated paper/plastic ice-cream cup.\n\nSpoon:\nclean brushed silver.\n\nBubble:\ntransparent iridescent membrane.\n\nStore:\n\n- realistic enough product shelves\n- stylized family-animation proportions\n- polished floor\n- soft global illumination\n- cinematic depth of field\n- beautiful reflections\n- highly readable character silhouettes\n\nThe final result should feel like:\n\nA PREMIUM 3D PRODUCT-MASCOT COMMERCIAL\n+\nA MINIATURE ANIMATED ADVENTURE FILM.\n\n==================================================\nCOLOR / LIGHTING ARC\n==================================================\n\nOPENING:\n\nwarm store lighting + cool midnight shadows.\n\nFREEZER SEQUENCE:\n\nincrease:\n\n- cyan\n- icy white\n- cool blue\n\nbut keep characters’ faces readable.\n\nPRICE-TAG SEQUENCE:\n\nreturn to mixed warm/cool store tones.\n\nFIZZ HERO JUMP:\n\nbubble receives brightest highlight.\n\nRECHARGE:\n\nFizz becomes visual focal point through:\n\n- stronger red reflections\n- small sparkling bubbles\n- warm rim light\n\nFINAL:\n\nslightly warmer than Part 1.\n\nUse:\n\n- amber practicals\n- rich red Fizz\n- yellow Crunch\n- mint-blue Chill\n- subtle blue night shadows\n\nThe final shot should visually communicate:\n\n# problem solved.\n\n==================================================\nMOTION RULES — FIZZ\n==================================================\n\nBefore recovery:\n\nquick,\nslightly frantic,\nhigh-energy.\n\nDuring hero jump:\n\nmovement becomes focused and precise.\n\nAfter recovery:\n\n- chest/body held proudly\n- big expressive arm gestures\n- springy landing\n- happy tiny hops\n- playful bubbles\n\nKeep can body rigid enough to preserve product identity.\n\n==================================================\nMOTION RULES — CRUNCH\n==================================================\n\nUse:\n\n- energetic running\n- bowl held with both hands\n- packaging bounce\n- little bag crinkles\n- exaggerated celebration\n- embarrassed deflation after bubble escapes\n- proud cheering at ending\n\nKeep him cute.\n\nNo chaotic deformation.\n\n==================================================\nMOTION RULES — CHILL\n==================================================\n\nUse controlled precision.\n\nActions:\n\n- point spoon\n- safely open freezer\n- watch bubble path\n- precise tag flick\n- small eyebrow response\n- quiet final approach\n- understated smile\n\nChill must visually appear to think BEFORE acting.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Fizz.\nONE Crunch.\nONE Chill.\nONE final bubble.\nONE spoon.\n\nSame:\n\n- mascot packaging designs\n- facial features\n- shoe colors\n- glove designs\n- relative heights\n- store layout\n- freezer\n- bubble appearance\n- shopping basket\n- bowl\n- hanging price cards\n\nBubble progression:\n\nfan\n→ cold-air redirect\n→ bowl\n→ escape\n→ price tags\n→ Fizz\n→ inside Fizz.\n\nOnce bubble enters Fizz:\n\ndo NOT show the original giant bubble outside him again.\n\nSmall NEW carbonation bubbles may appear only AFTER Fizz has recovered.\n\n==================================================\nEMOTIONAL PRIORITY\n==================================================\n\nAlthough this is comedy, the ending needs one brief emotional beat.\n\nFizz has spent both parts obsessed with getting his sparkle back.\n\nAfter he succeeds:\n\nhe looks at Crunch and Chill.\n\nRealizes:\n\nhe could not have done it alone.\n\nUse a tiny pause before:\n\n<d>[English] You guys saved me.</d>\n\nThat brief sincerity makes the comedic ending feel earned.\n\nThen immediately let Chill release tension with:\n\n<d>[English] Technically, the bubble did.</d>\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\n15 seconds is short.\n\nNatural voice timing is more important than including every line.\n\nHIGHEST PRIORITY:\n\n1.\nChill:\n<d>[English] Freezer.</d>\n\n2.\nFizz:\n<d>[English] It’s working!</d>\n\n3.\nCrunch:\n<d>[English] I GOT IT!</d>\n\n4.\nCrunch:\n<d>[English] ...I don’t got it.</d>\n\n5.\nChill:\n<d>[English] Mostly.</d>\n\n6.\nFizz:\n<d>[English] Come on...</d>\n\n7.\nFizz:\n<d>[English] I’M BACK!</d>\n\n8.\nFizz:\n<d>[English] You guys saved me.</d>\n\n9.\nChill:\n<d>[English] Technically, the bubble did.</d>\n\nIf timing becomes crowded:\n\nREMOVE secondary lines.\n\nDo NOT make dialogue unnaturally fast.\n\nLet:\n\n- expressions\n- movement\n- sound effects\n- music\n\ntell much of the story.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nEverything remains cute and family-friendly.\n\nNo injury.\nNo violence.\nNo weapons.\nNo broken glass.\nNo sharp attacks.\nNo dangerous ceiling fan interaction.\nNo exploding cans.\nNo destructive store damage.\nNo falling from dangerous heights.\nNo freezer entrapment.\nNo food contamination.\nNo humans stepping on characters.\n\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2087066113775665414",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-11T06:38:31.994Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2087066113775665414.mp4",
    "posterUrl": "/posters/x/2087066113775665414.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second Storyboard Image to video using MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n\"[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for:\n\n“THE LAST POP BEFORE CLOSING”\n\nas the STRICT PRIMARY visual and narrative reference.\n\nThis video continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n\n- Fizz’s final carbonation bubble drifting toward the ceiling fan\n- Fizz and Crunch failing to reach it\n- Chill calmly studying the situation\n- Chill noticing his silver spoon tool\n- Chill removing the spoon from his back\n- Chill saying:\n\n<d>[English] I have an idea.</d>\n\nPART 2 begins at that EXACT story moment.\n\nUse the storyboard to lock:\n\n- Fizz’s exact glossy red soda-can design\n- Crunch’s exact bright-yellow snack-bag design\n- Chill’s exact blue ice-cream-cup design\n- their faces, shoes, gloves and proportions\n- Chill’s spoon\n- the ONE surviving bubble\n- shopping basket\n- snack bowl / tray\n- freezer\n- cold mist\n- hanging price tags\n- ceiling fan\n- same convenience-store architecture\n- same nighttime lighting\n- same tiny-mascot scale\n- Part 2 rescue progression\n- final three-character payoff\n\nGenerate a FINISHED cinematic 3D animated sequence.\n\nDO NOT animate or reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- captions\n- storyboard headers\n- production notes\n- prop/reference sheets\n- character turnaround drawings\n- page background\n- watermarks\n- brand logos\n- unnecessary text overlays\n\nTranslate the storyboard’s 15 one-second panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT use fifteen frantic one-second cuts.\n\nStory progression:\n\nChill reveals the plan\n→ trio races ahead of bubble\n→ Chill uses freezer air\n→ bubble changes direction\n→ Crunch catches it temporarily\n→ bubble escapes again\n→ Chill triggers price-tag chain reaction\n→ bubble descends\n→ Fizz makes one perfect jump\n→ bubble returns\n→ Fizz recharges\n→ friends celebrate\n→ warm final shelf payoff.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 2 — 0:15 to 0:30\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium stylized 3D product-mascot animation\n\nGenre:\n\n- miniature adventure\n- product mascot comedy\n- friendship story\n- playful rescue mission\n\nTone:\n\n- energetic\n- clever\n- cute\n- visually satisfying\n- comedic\n- wholesome\n- heartwarming\n- triumphant\n\nDialogue:\nshort natural synchronized English\n\nAudio:\nnative dialogue + store ambience + physical comedy SFX + original playful cinematic score\n\nCORE PAYOFF:\n\nFizz thinks his “spark” depends on one final bubble.\n\nBut what actually saves that spark is:\n\n# teamwork.\n\nChill supplies the idea.\nCrunch supplies chaotic effort.\nFizz supplies the final leap.\n\n==================================================\nPART 2 START / END\n==================================================\n\nSTART:\n\nChill stands beneath the drifting bubble holding his silver spoon.\n\nFizz and Crunch look toward him.\n\nChill immediately begins executing his plan.\n\nEND:\n\nFizz has successfully regained his carbonation.\n\nFizz, Crunch, and Chill stand together in a warm celebratory final pose.\n\nThe story must feel COMPLETE.\n\nNo cliffhanger.\n\n==================================================\nCHARACTER IDENTITY LOCK — FIZZ\n==================================================\n\nMatch PART 1 and the uploaded PART 2 storyboard EXACTLY.\n\nFIZZ — RED SODA CAN MASCOT\n\nAppearance:\n\n- small glossy red aluminum soda can\n- cylindrical body\n- metallic silver top and bottom rims\n- silver pull tab\n- original fictional graphics only\n- large expressive blue/dark cartoon eyes\n- flexible eyebrows\n- expressive mouth\n- tiny arms\n- white mascot gloves\n- small legs\n- red-and-white sneakers\n\nMaterial:\n\n- glossy aluminum\n- strong store-light reflections\n- subtle metallic highlights\n- optional tiny condensation beads\n- body remains mostly rigid\n\nDo NOT redesign him.\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\ndesperate hope\n\n3–6 sec:\nexcited as Chill’s plan works\n\n6–8 sec:\ncelebrates too early\n\n8–11 sec:\nfocused determination\n\n11–13 sec:\nheroic relief\n\n13–15 sec:\nfull joyful Fizz energy\n\nMOTION:\n\nFizz is:\n\n- springy\n- quick\n- dramatic\n- highly expressive\n\nBut his can body should not behave like rubber.\n\n==================================================\nCHARACTER IDENTITY LOCK — CRUNCH\n==================================================\n\nCRUNCH — YELLOW SNACK BAG MASCOT\n\nAppearance:\n\n- bright yellow flexible snack bag\n- sealed ridged top\n- crinkled packaging material\n- original fictional package graphics\n- large expressive eyes\n- small expressive mouth\n- thin arms\n- white gloves\n- tiny legs\n- oversized orange sneakers\n\nPersonality:\n\n- loyal\n- excitable\n- chaotic\n- lovable\n- desperate to help\n- occasionally useful by accident\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\nhopeful and ready\n\n3–6 sec:\nheroic enthusiasm\n\n6–8 sec:\nbrief victory\n\n8–10 sec:\nimmediate panic when bubble escapes\n\n10–13 sec:\ncheering Fizz\n\n13–15 sec:\nhappy friend celebration\n\nPACKAGING MOTION:\n\nUse:\n\n- small crinkles\n- squash\n- puffing\n- top-edge wobble\n\nDo not deform him beyond recognition.\n\n==================================================\nCHARACTER IDENTITY LOCK — CHILL\n==================================================\n\nCHILL — BLUE ICE-CREAM CUP MASCOT\n\nAppearance:\n\n- small mint-blue / cool-blue cylindrical ice cream cup\n- white lid like a flat cap\n- original minimal fictional graphics\n- relaxed half-lidded eyes\n- small mouth\n- tiny arms\n- white gloves\n- tiny white shoes\n- silver spoon normally attached on back\n\nPersonality:\n\n- calm\n- intelligent\n- deadpan\n- observant\n- precise\n- quietly confident\n\nPART 2 PERFORMANCE ARC:\n\n0–4 sec:\nplan execution\n\n4–8 sec:\ncalm supervision\n\n8–11 sec:\nsecond tactical idea\n\n11–13 sec:\nwatches Fizz finish plan\n\n13–15 sec:\nsmall proud smile\n\nChill should NEVER become hyperactive.\n\nHis calmness is the joke.\n\n==================================================\nKEY PROP LOCK — THE LAST BUBBLE\n==================================================\n\nThere is:\n\n# ONE SINGLE BUBBLE.\n\nNo duplicates.\n\nAppearance:\n\n- transparent\n- iridescent\n- subtle rainbow edge\n- about the size of Fizz’s head\n- reflects store lights\n- delicate and lightweight\n\nBubble trajectory for Part 2 must remain easy to follow:\n\nceiling fan area\n→ redirected by freezer airflow\n→ descends toward Crunch\n→ briefly rests in bowl\n→ escapes\n→ drifts toward hanging price tags\n→ price-tag chain reaction guides it downward\n→ Fizz jumps\n→ bubble enters through slightly opened pull-tab area\n→ carbonation returns.\n\nNever teleport the bubble.\n\nNever spawn extra copies.\n\n==================================================\nKEY PROP LOCK — CHILL’S SPOON\n==================================================\n\nUse exactly one small silver spoon.\n\nIt begins in Chill’s hand because he removed it at the end of Part 1.\n\nUses:\n\n1. operating / assisting with freezer opening safely\n2. precisely flicking the price-tag string later\n\nThe spoon is a clever tool.\n\nIt is NOT a weapon.\n\nNo dangerous swinging.\n\nNo sharp-object action.\n\n==================================================\nKEY PROP — SHOPPING BASKET\n==================================================\n\nSmall red convenience-store shopping basket relative to human scale, but large enough for the mascots to ride in.\n\nUse it only as a comic transport device.\n\nSafe movement.\n\nNo crashes.\n\nNo dangerous speed.\n\n==================================================\nENVIRONMENT LOCK\n==================================================\n\nSame convenience store from Part 1.\n\nMaintain:\n\n- glossy tile floor\n- large aisles\n- soda shelf\n- snack aisle\n- freezers\n- candy displays\n- fruit display\n- hanging price cards\n- ceiling fan\n- warm overhead practical lights\n- cool freezer illumination\n- midnight darkness beyond storefront windows\n\nThe store should feel enormous compared with the mascots.\n\nKeep miniature cinematography throughout.\n\n==================================================\nWORLD / PHYSICS RULES\n==================================================\n\nBubble:\n\nextremely light.\n\nIt responds to:\n\n- airflow\n- small impacts\n- movement\n- gentle rebounds\n\nFreezer:\n\ncold air pushes bubble away from fan.\n\nSnack bowl:\n\nbubble can rest briefly without popping.\n\nPrice tags:\n\nsmall hanging cardboard tags swing like pendulums and redirect the bubble through gentle contact / airflow.\n\nFizz’s recovery:\n\nbubble enters Fizz through his slightly opened pull-tab opening.\n\nThen carbonation visibly rebuilds INSIDE the can through:\n\n- subtle internal fizz\n- a few tiny external bubbles\n- brighter expression\n- energetic body language\n\nDo NOT show explosive pressure.\n\nNo dangerous soda burst.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.7]\nCHILL’S PLAN BEGINS\n\nContinue EXACTLY from Part 1.\n\nBubble floats high in the background near the ceiling airflow.\n\nFizz and Crunch turn toward Chill.\n\nChill holds the silver spoon.\n\nChill looks toward:\n\nbubble\n→ freezer aisle.\n\nHe points the spoon forward like a tiny conductor.\n\nChill:\n\n<d>[English] Freezer.</d>\n\nFizz:\n\n<d>[English] The freezer?</d>\n\nChill gives a tiny confident nod.\n\nCrunch:\n\n<d>[English] I love plans I don’t understand!</d>\n\nChill starts moving.\n\nCAMERA:\n\nclose Chill hero angle\n→ rack focus to bubble\n→ fast low-angle follow as trio starts moving.\n\nMusic immediately restarts with clever playful momentum.\n\n==================================================\n[SHOT 2 — 0:01.7–0:03.3]\nTHE SHOPPING-BASKET DASH\n\nFizz and Crunch jump into / grab a small red shopping basket.\n\nChill pushes / steers it from behind or rides at the edge while controlling direction.\n\nThe basket rolls/slides smoothly across the polished store floor.\n\nUse a low mascot-height tracking shot.\n\nBubble remains visible above and ahead.\n\nCrunch points upward.\n\nCrunch:\n\n<d>[English] Faster!</d>\n\nFizz:\n\n<d>[English] It’s getting away!</d>\n\nChill:\n\n<d>[English] Relax.</d>\n\nThey pass:\n\n- oversized product shelves\n- cool refrigerator reflections\n- hanging price tags\n\nNo collision.\n\nNo store destruction.\n\n==================================================\n[SHOT 3 — 0:03.3–0:05.1]\nCOLD AIR CHANGES THE GAME\n\nThey arrive at the freezer.\n\nChill safely uses the spoon as a tiny lever/tool to pull the freezer-door edge enough to open it.\n\nDoor opens.\n\n# FWHOOSH.\n\nA beautiful plume of cold mist rolls outward.\n\nBlue-white freezer light contrasts with the warm store.\n\nFizz and Crunch brace themselves behind the basket.\n\nThe cold airflow reaches the drifting bubble.\n\nThe bubble slows.\n\nThen gently curves AWAY from the ceiling fan.\n\nFizz’s expression instantly changes.\n\nFizz:\n\n<d>[English] It’s working!</d>\n\nChill, completely calm:\n\n<d>[English] Obviously.</d>\n\nCAMERA:\n\nfreezer close-up\n→ cold mist\n→ follow airflow visually\n→ bubble changing trajectory.\n\nSFX:\n\n- freezer seal opening\n- gentle cold-air whoosh\n- faint icy shimmer\n- bubble wobble\n\n==================================================\n[SHOT 4 — 0:05.1–0:06.8]\nCRUNCH GETS HIS HERO MOMENT\n\nBubble now floats downward through the aisle.\n\nCrunch spots the snack bowl / tray.\n\nHis eyes widen.\n\nCrunch:\n\n<d>[English] My turn!</d>\n\nHe grabs the bowl.\n\nRuns underneath the bubble.\n\nFizz guides him:\n\n<d>[English] Left! Left! No—your other left!</d>\n\nCrunch shifts.\n\nThe bubble gently drops—\n\n# into the bowl.\n\nPerfect catch.\n\nCrunch freezes.\n\nLooks down.\n\nHis eyes become huge with joy.\n\nCrunch:\n\n<d>[English] I GOT IT!</d>\n\nFizz throws both arms up.\n\n==================================================\n[SHOT 5 — 0:06.8–0:08.4]\nCELEBRATED TOO EARLY\n\nFizz and Crunch start celebrating immediately.\n\nCrunch lifts the bowl proudly.\n\nBad idea.\n\nThe movement tilts it.\n\nThe bubble gently floats back out.\n\nAll three characters track it.\n\nFizz and Crunch’s smiles disappear at the exact same time.\n\nCrunch:\n\n<d>[English] ...I don’t got it.</d>\n\nFizz:\n\n<d>[English] CRUNCH!</d>\n\nThe bubble continues toward a row of hanging price tags.\n\nChill does not panic.\n\nHe simply turns his eyes upward.\n\nCamera pushes in on Chill.\n\nHe notices the tags.\n\nThen the spoon in his hand.\n\nSecond idea.\n\nTiny eyebrow raise.\n\n==================================================\n[SHOT 6 — 0:08.4–0:10.4]\nTHE PRICE-TAG CHAIN REACTION\n\nChill steps into position beneath the first hanging price tag.\n\nPrecisely:\n\nhe taps the hanging string with the spoon.\n\n# tik.\n\nFirst price tag swings.\n\nIt gently nudges the next hanging tag.\n\nSecond swings.\n\nThen third.\n\nCreate a clean satisfying cascading motion:\n\nTAG 1\n→ TAG 2\n→ TAG 3\n→ TAG 4.\n\nThe bubble moves through this chain.\n\nEach swinging tag / tiny air current guides it slightly lower.\n\nNo hard impact.\n\nNo popping.\n\nUse highly satisfying visual rhythm.\n\nFizz watches.\n\nHis eyes widen.\n\nCrunch:\n\n<d>[English] He planned THAT?!</d>\n\nChill:\n\n<d>[English] Mostly.</d>\n\nMusic syncs each tag movement to percussion:\n\ntik\ntik\ntik\ntik.\n\n==================================================\n[SHOT 7 — 0:10.4–0:12.6]\nONE PERFECT JUMP\n\nThe bubble is finally descending.\n\nFizz runs underneath it.\n\nCrunch runs behind him cheering.\n\nChill watches calmly.\n\nFizz’s expression becomes focused.\n\nNo more panic.\n\nCamera drops to dramatic low angle.\n\nFizz accelerates.\n\nHe jumps.\n\nUse a brief cinematic slow-down—not full slow motion.\n\nHis arms spread.\n\nBubble directly above him.\n\nFizz opens his silver pull tab just slightly.\n\nFizz:\n\n<d>[English] Come on...</d>\n\nThe bubble descends.\n\nCloser.\n\nCloser.\n\nThen—\n\n# PLOOP / FIZZ.\n\nIt slips neatly through the opening.\n\nFizz’s eyes widen.\n\nPause for one beat.\n\n==================================================\n[SHOT 8 — 0:12.6–0:15.0]\nFIZZ GETS HIS SPARK BACK — FINAL PAYOFF\n\nImmediately after the bubble enters:\n\n# FSSSSHHH!\n\nA playful carbonation pulse moves through Fizz.\n\nNOT an explosion.\n\nHis red aluminum surface becomes bright and lively.\n\nTiny fizzy bubbles swirl around him.\n\nHis eyes light up.\n\nHis posture pops back into full confidence.\n\nFizz lands safely.\n\nFizz:\n\n<d>[English] I’M BACK!</d>\n\nCrunch cheers wildly.\n\nCrunch:\n\n<d>[English] WE SAVED THE FIZZ!</d>\n\nChill walks up calmly.\n\nFizz looks toward both friends.\n\nFor one sincere beat:\n\nFizz:\n\n<d>[English] You guys saved me.</d>\n\nCrunch immediately poses proudly.\n\nChill:\n\n<d>[English] Technically, the bubble did.</d>\n\nSmall comedy beat.\n\nFizz laughs.\n\nA few harmless tiny carbonation bubbles float around all three.\n\nOne bubble lands perfectly on Chill’s white lid.\n\nChill slowly looks upward at it.\n\nTiny smile.\n\nFINAL COMPOSITION:\n\nFizz in the center, fully recharged.\n\nCrunch beside him, excited and proud.\n\nChill beside them, calm and satisfied.\n\nStore shelves glow warmly behind.\n\nSmall bubbles drift around them.\n\nOptional quick match transition / cut:\n\nthe three mascots back near their product shelf, posing together like the night’s adventure is their secret.\n\nCamera gently pulls backward.\n\nWarm practical lights.\n\nCool midnight shadows.\n\nPlayful final musical chord.\n\nFADE TO BLACK.\n\nEND STORY.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio tone:\n\n- playful\n- tactile\n- energetic\n- miniature\n- cinematic\n- satisfying\n- wholesome\n\n==================================================\nFIZZ VOICE\n==================================================\n\nEnergetic youthful masculine mascot voice.\n\nQualities:\n\n- charismatic\n- dramatic\n- quick\n- lovable\n- expressive\n\nPart 2 progression:\n\nhopeful urgency\n→ premature celebration\n→ focused determination\n→ huge relief\n→ grateful friendship.\n\nPriority lines:\n\n<d>[English] It’s working!</d>\n\n<d>[English] CRUNCH!</d>\n\n<d>[English] Come on...</d>\n\n<d>[English] I’M BACK!</d>\n\n<d>[English] You guys saved me.</d>\n\nFinal sincere line should be warmer and quieter than his usual show-off delivery.\n\n==================================================\nCRUNCH VOICE\n==================================================\n\nExcitable comedic mascot voice.\n\nQualities:\n\n- enthusiastic\n- slightly squeaky\n- dramatic\n- lovable\n- fast reaction timing\n\nPriority:\n\n<d>[English] My turn!</d>\n\n<d>[English] I GOT IT!</d>\n\n<d>[English] ...I don’t got it.</d>\n\n<d>[English] He planned THAT?!</d>\n\n<d>[English] WE SAVED THE FIZZ!</d>\n\nUse tiny packaging crinkles with body motion.\n\n==================================================\nCHILL VOICE\n==================================================\n\nCalm, understated, low-energy mascot voice.\n\nQualities:\n\n- dry humor\n- relaxed\n- intelligent\n- confident\n- slightly sleepy\n\nPriority lines:\n\n<d>[English] Freezer.</d>\n\n<d>[English] Obviously.</d>\n\n<d>[English] Mostly.</d>\n\nFinal:\n\n<d>[English] Technically, the bubble did.</d>\n\nDeliver completely deadpan.\n\n==================================================\nSTORE SOUND DESIGN\n==================================================\n\nMaintain subtle closed-store ambience:\n\n- refrigerator hum\n- freezer compressor\n- ventilation\n- faint building hum\n- soft sneaker taps\n- packaging crinkles\n- shopping basket wheel/slide sounds\n- tiny metal spoon clicks\n- hanging card flutter\n- distant night ambience\n\nDo not introduce humans.\n\n==================================================\nFREEZER SOUND DESIGN\n==================================================\n\nWhen freezer opens:\n\n- soft magnetic seal release\n- cold air whoosh\n- compressor hum slightly louder\n- gentle mist movement\n\nDo NOT use giant icy explosion SFX.\n\n==================================================\nPRICE TAG CHAIN SFX\n==================================================\n\nMake this extremely satisfying.\n\nUse four slightly different light sounds:\n\ntik\ntap\ntik\ntap\n\nEach synchronized to a swinging tag.\n\nAdd gentle cardboard flutter.\n\nMusic percussion should sync perfectly.\n\n==================================================\nBUBBLE SOUND DESIGN\n==================================================\n\nBubble signature:\n\n- tiny glassy shimmer\n- fizzy air texture\n- delicate elastic wobble\n\nBowl landing:\n\nsoft:\n\n“plip.”\n\nLeaving bowl:\n\ntiny airy:\n\n“woop.”\n\nReturning to Fizz:\n\nsmall:\n\n“ploop”\n→ immediately followed by carbonation:\n\n“fsssshhhh.”\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nContinue the ORIGINAL miniature adventure score from Part 1.\n\nInstrumentation:\n\n- pizzicato strings\n- marimba\n- woodblocks\n- brushed percussion\n- muted brass\n- upright bass\n- tiny synth plucks\n- glockenspiel\n- warm orchestral accents\n\nMUSIC ARC:\n\n0:00–0:03\nclever plan / movement energy.\n\n0:03–0:05\nfreezer solution reveal.\n\n0:05–0:07\ncomic triumph.\n\n0:07–0:08.5\ncomedic drop after bubble escapes.\n\n0:08.5–0:10.5\nprecise rhythmic chain-reaction sequence.\n\n0:10.5–0:12.5\nheroic but playful Fizz build.\n\n0:12.5–0:15\nwarm celebratory theme combining motifs associated with all three mascots.\n\nFinal chord:\n\nbright,\nsatisfying,\ncomplete.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 2 should feel slightly more cinematic and energetic than Part 1.\n\nUse:\n\n- floor-level tracking\n- macro bubble shots\n- low mascot-scale lensing\n- freezer push-ins\n- dramatic cool/warm lighting contrast\n- overhead shot for bowl catch\n- rhythmic insert shots during price-tag chain\n- low-angle Fizz hero jump\n- expressive reaction close-ups\n- final slow pull-back\n\nCamera progression:\n\n0–3 sec:\nquick planning/action.\n\n3–5 sec:\ncold-air spectacle.\n\n5–8 sec:\nphysical comedy.\n\n8–10 sec:\nprecise chain-reaction rhythm.\n\n10–13 sec:\nheroic focus.\n\n13–15 sec:\nwarm emotional resolution.\n\nAvoid:\n\n- shaky handheld\n- extreme motion blur\n- chaotic whip pans\n- fisheye distortion\n- unreadable fast cutting\n- action-movie explosions\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium high-end stylized 3D animation.\n\nMaterial quality:\n\nFizz:\nglossy reflective aluminum.\n\nCrunch:\nsoft flexible crinkly snack-film packaging.\n\nChill:\nsemi-matte coated paper/plastic ice-cream cup.\n\nSpoon:\nclean brushed silver.\n\nBubble:\ntransparent iridescent membrane.\n\nStore:\n\n- realistic enough product shelves\n- stylized family-animation proportions\n- polished floor\n- soft global illumination\n- cinematic depth of field\n- beautiful reflections\n- highly readable character silhouettes\n\nThe final result should feel like:\n\nA PREMIUM 3D PRODUCT-MASCOT COMMERCIAL\n+\nA MINIATURE ANIMATED ADVENTURE FILM.\n\n==================================================\nCOLOR / LIGHTING ARC\n==================================================\n\nOPENING:\n\nwarm store lighting + cool midnight shadows.\n\nFREEZER SEQUENCE:\n\nincrease:\n\n- cyan\n- icy white\n- cool blue\n\nbut keep characters’ faces readable.\n\nPRICE-TAG SEQUENCE:\n\nreturn to mixed warm/cool store tones.\n\nFIZZ HERO JUMP:\n\nbubble receives brightest highlight.\n\nRECHARGE:\n\nFizz becomes visual focal point through:\n\n- stronger red reflections\n- small sparkling bubbles\n- warm rim light\n\nFINAL:\n\nslightly warmer than Part 1.\n\nUse:\n\n- amber practicals\n- rich red Fizz\n- yellow Crunch\n- mint-blue Chill\n- subtle blue night shadows\n\nThe final shot should visually communicate:\n\n# problem solved.\n\n==================================================\nMOTION RULES — FIZZ\n==================================================\n\nBefore recovery:\n\nquick,\nslightly frantic,\nhigh-energy.\n\nDuring hero jump:\n\nmovement becomes focused and precise.\n\nAfter recovery:\n\n- chest/body held proudly\n- big expressive arm gestures\n- springy landing\n- happy tiny hops\n- playful bubbles\n\nKeep can body rigid enough to preserve product identity.\n\n==================================================\nMOTION RULES — CRUNCH\n==================================================\n\nUse:\n\n- energetic running\n- bowl held with both hands\n- packaging bounce\n- little bag crinkles\n- exaggerated celebration\n- embarrassed deflation after bubble escapes\n- proud cheering at ending\n\nKeep him cute.\n\nNo chaotic deformation.\n\n==================================================\nMOTION RULES — CHILL\n==================================================\n\nUse controlled precision.\n\nActions:\n\n- point spoon\n- safely open freezer\n- watch bubble path\n- precise tag flick\n- small eyebrow response\n- quiet final approach\n- understated smile\n\nChill must visually appear to think BEFORE acting.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Fizz.\nONE Crunch.\nONE Chill.\nONE final bubble.\nONE spoon.\n\nSame:\n\n- mascot packaging designs\n- facial features\n- shoe colors\n- glove designs\n- relative heights\n- store layout\n- freezer\n- bubble appearance\n- shopping basket\n- bowl\n- hanging price cards\n\nBubble progression:\n\nfan\n→ cold-air redirect\n→ bowl\n→ escape\n→ price tags\n→ Fizz\n→ inside Fizz.\n\nOnce bubble enters Fizz:\n\ndo NOT show the original giant bubble outside him again.\n\nSmall NEW carbonation bubbles may appear only AFTER Fizz has recovered.\n\n==================================================\nEMOTIONAL PRIORITY\n==================================================\n\nAlthough this is comedy, the ending needs one brief emotional beat.\n\nFizz has spent both parts obsessed with getting his sparkle back.\n\nAfter he succeeds:\n\nhe looks at Crunch and Chill.\n\nRealizes:\n\nhe could not have done it alone.\n\nUse a tiny pause before:\n\n<d>[English] You guys saved me.</d>\n\nThat brief sincerity makes the comedic ending feel earned.\n\nThen immediately let Chill release tension with:\n\n<d>[English] Technically, the bubble did.</d>\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\n15 seconds is short.\n\nNatural voice timing is more important than including every line.\n\nHIGHEST PRIORITY:\n\n1.\nChill:\n<d>[English] Freezer.</d>\n\n2.\nFizz:\n<d>[English] It’s working!</d>\n\n3.\nCrunch:\n<d>[English] I GOT IT!</d>\n\n4.\nCrunch:\n<d>[English] ...I don’t got it.</d>\n\n5.\nChill:\n<d>[English] Mostly.</d>\n\n6.\nFizz:\n<d>[English] Come on...</d>\n\n7.\nFizz:\n<d>[English] I’M BACK!</d>\n\n8.\nFizz:\n<d>[English] You guys saved me.</d>\n\n9.\nChill:\n<d>[English] Technically, the bubble did.</d>\n\nIf timing becomes crowded:\n\nREMOVE secondary lines.\n\nDo NOT make dialogue unnaturally fast.\n\nLet:\n\n- expressions\n- movement\n- sound effects\n- music\n\ntell much of the story.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nEverything remains cute and family-friendly.\n\nNo injury.\nNo violence.\nNo weapons.\nNo broken glass.\nNo sharp attacks.\nNo dangerous ceiling fan interaction.\nNo exploding cans.\nNo destructive store damage.\nNo falling from dangerous heights.\nNo freezer entrapment.\nNo food contamination.\nNo humans stepping on characters.\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2087066113775665414"
  },
  {
    "id": "x-2087065646903517354",
    "title": "打烊前最后一个气泡 · 上篇",
    "titleEn": "The Last Bubble Before Closing · Part 1",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《打烊前最后一个气泡 · 上篇》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Last Bubble Before Closing · Part 1. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "\"[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for:\n\n“THE LAST POP BEFORE CLOSING”\n\nas the STRICT PRIMARY reference for:\n\n- Fizz’s exact red soda-can design\n- Crunch’s exact yellow snack-bag design\n- Chill’s exact blue ice-cream-cup design\n- character proportions\n- facial features\n- arms, legs, gloves and footwear\n- packaging materials\n- relative mascot scale\n- convenience-store environment\n- soda shelves\n- snack aisle\n- fruit display\n- shopping basket\n- ceiling ventilation fan\n- hanging price tags\n- Chill’s spoon tool\n- bubble design\n- midnight lighting\n- Part 1 story progression\n\nGenerate a FINISHED cinematic 3D animated sequence from the storyboard.\n\nDO NOT animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- captions\n- production notes\n- character reference drawings\n- prop reference drawings\n- page background\n- storyboard labels\n- watermark\n- unnecessary subtitles\n\nTranslate the 15 storyboard panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT make fifteen separate one-second cuts.\n\nThe story progression must read clearly as:\n\nFizz makes a flashy entrance\n→ discovers only one bubble remains\n→ the final bubble escapes\n→ Fizz and Crunch chase it\n→ repeated funny near-misses\n→ bubble reaches ceiling fan\n→ Fizz and Crunch cannot reach it\n→ Chill quietly discovers the solution\n→ cliffhanger.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 1 — 0:00 to 0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium stylized 3D animated product-mascot film\n\nGenre:\n\n- miniature adventure\n- product mascot comedy\n- friendship story\n- playful rescue mission\n\nTone:\n\n- energetic\n- cute\n- funny\n- visually satisfying\n- slightly suspenseful\n- heartwarming\n- family-friendly\n\nDialogue:\nshort synchronized English dialogue\n\nAudio:\nnative dialogue + convenience-store ambience + physical comedy SFX + playful cinematic music\n\nCORE PART 1 IDEA:\n\nFizz thinks he is still the fizziest product in the store...\n\nuntil he discovers he has only ONE bubble left.\n\nWhen that final bubble escapes, his friends launch a tiny midnight rescue mission.\n\n==================================================\nPART 1 STORY LIMIT\n==================================================\n\nSTART:\n\nA quiet convenience store after closing.\n\nFizz suddenly makes a ridiculous heroic entrance riding a giant bubble.\n\nEND:\n\nThe final bubble is almost lost near the ceiling ventilation fan.\n\nFizz and Crunch are helplessly too short.\n\nChill studies the situation, removes the spoon from his back, smiles slightly and says:\n\n<d>[English] I have an idea.</d>\n\nCUT TO BLACK.\n\nDO NOT show:\n\n- Chill executing the solution\n- freezer rescue\n- shopping-basket rescue tower\n- bubble returning to Fizz\n- Fizz becoming fully carbonated again\n- final celebration\n- morning customer\n- shelf ending\n\nSave ALL successful rescue payoff for PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — FIZZ\n==================================================\n\nMatch the storyboard EXACTLY.\n\nFIZZ — RED SODA CAN MASCOT\n\nAppearance:\n\n- small glossy cylindrical red aluminum soda can\n- metallic silver top and bottom rims\n- silver pull tab\n- original fictional red label graphics\n- NO recognizable real-world soda branding\n- large expressive cartoon eyes integrated naturally into can body\n- flexible black eyebrows\n- tiny expressive mouth\n- thin cartoon arms\n- white four-finger-style mascot gloves\n- tiny legs\n- red-and-white sneakers\n\nSurface:\n\n- glossy lacquered aluminum\n- crisp reflections\n- slightly cool metal highlights\n- subtle condensation if appropriate\n\nPersonality:\n\n- energetic\n- dramatic\n- overconfident\n- loves showing off\n- easily panics\n- lovable rather than annoying\n\nPART 1 EMOTIONAL ARC:\n\n0–2 sec:\nmaximum confidence\n\n2–4 sec:\nconfusion\n\n4–6 sec:\npanic\n\n6–11 sec:\ndesperate comic determination\n\n11–13 sec:\nfear of losing the bubble\n\n13–15 sec:\nconfused hope as Chill gets an idea\n\nIMPORTANT:\n\nFizz must remain clearly identifiable as the SAME red can in every shot.\n\nDo not alter:\n\n- shape\n- height\n- face placement\n- silver top\n- shoes\n- glove style\n- red color\n\n==================================================\nCHARACTER IDENTITY LOCK — CRUNCH\n==================================================\n\nCRUNCH — YELLOW SNACK BAG MASCOT\n\nAppearance:\n\n- small puffy rectangular snack bag\n- bright warm yellow packaging\n- crinkled foil/plastic texture\n- sealed ridged top edge\n- original fictional package graphics only\n- large expressive eyes\n- tiny mouth\n- short flexible arms\n- white gloves\n- tiny legs\n- oversized orange sneakers\n\nPersonality:\n\n- enthusiastic\n- loyal\n- easily alarmed\n- physically comedic\n- wants to be heroic\n- often makes things more chaotic\n\nPART 1 PERFORMANCE:\n\n0–2 sec:\nFizz fanboy / applause\n\n2–4 sec:\ninstant concern\n\n4–6 sec:\ndramatic panic\n\n6–10 sec:\noverenthusiastic rescue attempts\n\n10–13 sec:\nfull comic emergency mode\n\n13–15 sec:\nconfused by Chill’s calmness\n\nPackaging motion:\n\nCrunch’s bag should lightly crinkle, squash and puff with emotion.\n\nDo NOT deform him so much that his identity changes.\n\n==================================================\nCHARACTER IDENTITY LOCK — CHILL\n==================================================\n\nCHILL — BLUE ICE-CREAM CUP MASCOT\n\nAppearance:\n\n- compact cylindrical mint-blue / cool-blue ice cream cup\n- white plastic lid resembling a flat cap\n- clean original fictional cup graphics\n- relaxed half-lidded expressive eyes\n- small understated mouth\n- tiny flexible arms\n- white gloves\n- small legs\n- white sneakers\n- one silver spoon attached vertically/diagonally to his back like a tiny tool\n\nPersonality:\n\n- extremely calm\n- intelligent\n- observant\n- dry sense of humor\n- speaks rarely\n- never panics\n\nPART 1 PERFORMANCE:\n\n0–5 sec:\nunimpressed observer\n\n5–10 sec:\nquietly follows and studies\n\n10–13 sec:\nanalyzes fan + bubble trajectory\n\n13–15 sec:\nsolution clicks into place\n\nHis comedy comes from contrast:\n\nFizz and Crunch = chaos.\n\nChill = almost completely calm.\n\n==================================================\nKEY PROP LOCK — LAST BUBBLE\n==================================================\n\nThe final bubble is the MAIN STORY OBJECT.\n\nAppearance:\n\n- one transparent soap-like carbonation bubble\n- roughly the size of Fizz’s head or slightly smaller\n- subtle rainbow iridescence\n- reflective convenience-store lights on surface\n- transparent center\n- physically delicate\n- readable against dark backgrounds\n\nThe bubble must remain:\n\n# ONE SINGLE BUBBLE\n\nNever duplicate it.\n\nTrajectory must be physically understandable.\n\nIt moves through the story:\n\nFizz\n→ floats upward\n→ drifts into aisle\n→ passes candy / price tags\n→ Crunch misses it\n→ bounces gently from shiny apple\n→ rises toward ceiling\n→ ventilation airflow begins drawing it upward.\n\nThe audience must always understand where the bubble is going.\n\n==================================================\n“1 BUBBLE LEFT” VISUAL\n==================================================\n\nAt the discovery beat, briefly show a simple ORIGINAL diegetic graphic beside or reflected near Fizz:\n\n“1 BUBBLE LEFT”\n\nStyle:\n\n- tiny red/orange digital indicator\n- simple iconography\n- readable for less than one second\n- not a permanent HUD\n- not a video-game interface\n\nIt should feel like a comedic internal soda-status visualization.\n\nDo NOT cover the whole frame with UI.\n\n==================================================\nENVIRONMENT LOCK — MIDNIGHT CONVENIENCE STORE\n==================================================\n\nMaintain the uploaded storyboard environment.\n\nStore:\n\n- closed for the night\n- organized aisles\n- soda shelves\n- snack shelves\n- candy section\n- fruit display\n- shopping baskets\n- small paper price tags\n- refrigerators/freezers\n- ceiling ventilation fans\n- glossy tile floor\n- large storefront windows\n- shelves dramatically oversized compared with the tiny mascots\n\nLighting:\n\n- warm practical ceiling lights\n- soft orange shelf illumination\n- cool midnight blue shadows\n- occasional moonlight from storefront windows\n- reflections on polished floor\n- subtle refrigerator cyan light\n\nThe store should feel huge from mascot scale.\n\nUse miniature cinematography.\n\n==================================================\nWORLD / PHYSICS RULES\n==================================================\n\nThe characters are tiny product mascots living secretly inside a normal convenience store.\n\nMaintain believable miniature scale.\n\nFizz jumping:\nlightweight metal-can motion.\n\nCrunch:\nsoft bag bounce and packaging crinkle.\n\nChill:\nsmall solid cup motion with slightly heavier stable footing.\n\nBubble physics:\n\n- extremely lightweight\n- slow upward drift\n- reacts to air currents\n- gently rebounds from objects\n- never behaves like a heavy ball\n- never explodes in Part 1\n\nCAUSE → EFFECT must remain readable.\n\nFizz releases last bubble\n→ bubble rises.\n\nCharacters chase\n→ air movement causes slight drift.\n\nCrunch misses\n→ bubble continues.\n\nBubble touches apple\n→ gently redirects upward.\n\nCeiling fan airflow\n→ bubble starts being pulled toward fan.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.8]\nMIDNIGHT — THE GREAT FIZZ ENTRANCE\n\nOpen outside / just inside the convenience store at midnight.\n\nEstablish:\n\n- dark blue night sky\n- quiet storefront\n- warm store lights glowing inside\n\nMove quickly into the soda aisle.\n\nSuddenly—\n\n# POP!\n\nFizz shoots upward from the soda shelf riding on top of a large fizzy bubble.\n\nHe balances dramatically like a tiny action hero.\n\nSeveral tiny harmless carbonation bubbles trail around him.\n\nHe lands perfectly on the front edge of the shelf.\n\nCrunch immediately applauds enthusiastically.\n\nChill stands nearby with half-lidded eyes and almost no reaction.\n\nFizz opens his arms proudly.\n\nFizz:\n\n<d>[English] Still got it.</d>\n\nCrunch:\n\n<d>[English] That was AMAZING!</d>\n\nChill gives a tiny unimpressed:\n\n<d>[English] Mm-hm.</d>\n\nCAMERA:\n\nwide store establish\n→ fast push toward soda shelf\n→ low-angle Fizz entrance\n→ medium three-character reaction.\n\nSFX:\n\n- fizzy POP\n- bubble wobble\n- tiny sneaker landing\n- Crunch package crinkle\n- small celebratory sparkle sound\n\nMUSIC:\n\nplayful miniature heist/adventure theme begins.\n\n==================================================\n[SHOT 2 — 0:01.8–0:03.6]\nONE BUBBLE LEFT\n\nMedium close-up on Fizz.\n\nHe continues posing proudly.\n\nThen—\n\na tiny pathetic bubble emerges from near his pull tab.\n\nSFX:\n\n# “pip.”\n\nFizz slowly looks upward.\n\nThe bubble floats beside him.\n\nA tiny diegetic indicator flashes:\n\n# 1 BUBBLE LEFT\n\nFizz’s smile disappears.\n\nCrunch stops applauding.\n\nChill raises one eyebrow.\n\nFizz:\n\n<d>[English] Wait...</d>\n\nFizz gently shakes himself.\n\nNo bubbles.\n\nHe shakes again.\n\nNothing.\n\nFizz:\n\n<d>[English] That can’t be right.</d>\n\nCrunch:\n\n<d>[English] ONE?!</d>\n\nChill remains calm.\n\nCamera:\n\nFizz hero close-up\n→ tiny bubble\n→ status indicator\n→ reaction three-shot.\n\nMusic drops into a comic suspense note.\n\n==================================================\n[SHOT 3 — 0:03.6–0:05.4]\nTHE FINAL BUBBLE ESCAPES\n\nFizz tries increasingly silly but safe methods to make more carbonation:\n\ntiny bounce,\nsmall shake,\nquick spin.\n\nNothing.\n\nThen—\n\nthe ONE remaining bubble slowly drifts away from him.\n\nFizz freezes.\n\nHis pupils track it.\n\nFizz:\n\n<d>[English] No no no no...</d>\n\nBubble clears the edge of the shelf.\n\nFizz reaches forward.\n\nToo late.\n\nFizz:\n\n<d>[English] MY BUBBLE!</d>\n\nCrunch gasps dramatically.\n\nChill turns his eyes toward the bubble’s trajectory.\n\nCamera follows the bubble over the shelf edge.\n\n==================================================\n[SHOT 4 — 0:05.4–0:07.5]\nTHE STORE-WIDE CHASE\n\nFizz leaps safely down from the low product display and runs after the floating bubble.\n\nCrunch runs immediately behind him, bag bouncing and crinkling wildly.\n\nChill calmly climbs/slides down and follows at an efficient pace.\n\nThe camera tracks low at mascot height.\n\nThe bubble floats through the giant store aisle.\n\nPass:\n\n- candy displays\n- hanging price cards\n- colorful product shelves\n- oversized boxes\n- polished tiles\n\nFizz stretches both hands upward while running.\n\nFizz:\n\n<d>[English] Don’t let it get away!</d>\n\nCrunch:\n\n<d>[English] I GOT IT!</d>\n\nChill, quietly:\n\n<d>[English] You don’t.</d>\n\nStrong visual scale:\n\ntiny mascots below,\nhuge supermarket aisle above.\n\n==================================================\n[SHOT 5 — 0:07.5–0:09.4]\nCRUNCH’S GREAT RESCUE ATTEMPT\n\nCrunch spots an empty decorative snack bowl / tray.\n\nHis eyes light up.\n\nHe grabs it with both hands.\n\nCrunch:\n\n<d>[English] Stand back!</d>\n\nHe runs underneath the bubble.\n\nMoves bowl left.\n\nBubble moves right.\n\nHe moves right.\n\nBubble moves left.\n\nCrunch finally jumps and thrusts the bowl upward.\n\nMISS.\n\nFizz also makes a tiny dive beneath it.\n\nMISS.\n\nCrunch lands seated inside / behind the bowl, stunned but unharmed.\n\nFizz:\n\n<d>[English] Great plan!</d>\n\nCrunch:\n\n<d>[English] Thank you!</d>\n\nBeat.\n\nFizz:\n\n<d>[English] I was being sarcastic!</d>\n\nIf dialogue timing is crowded, prioritize:\n“Stand back!” + visual comedy.\n\nBubble continues forward.\n\n==================================================\n[SHOT 6 — 0:09.4–0:11.0]\nTHE APPLE BOUNCE\n\nThe last bubble drifts toward the fruit display.\n\nClose macro shot:\n\na shiny red apple reflects the bubble and tiny running mascots.\n\nBubble gently touches the apple.\n\n# boop.\n\nIt redirects upward.\n\nFizz arrives just beneath it.\n\nHis eyes widen.\n\nFizz:\n\n<d>[English] Uh-oh.</d>\n\nCamera tilts vertically following the bubble.\n\nReveal:\n\na ceiling ventilation fan high above.\n\nThe bubble begins drifting toward it.\n\nMusic shifts from playful chase to light suspense.\n\nChill stops running.\n\nHe studies the situation.\n\n==================================================\n[SHOT 7 — 0:11.0–0:13.3]\nTOO SHORT\n\nThe ventilation fan rotates above.\n\nIMPORTANT:\n\nDo not make it threatening or dangerous-looking.\n\nIt is simply creating airflow.\n\nThe bubble begins moving upward faster.\n\nFizz jumps repeatedly.\n\nToo short.\n\nCrunch rushes behind him.\n\nFizz grabs a safe shelf lip / hanging display support.\n\nCrunch holds Fizz around the legs.\n\nThey stretch upward in a ridiculous mascot chain.\n\nFizz:\n\n<d>[English] Almost...</d>\n\nCrunch:\n\n<d>[English] I can’t stretch anymore!</d>\n\nFizz:\n\n<d>[English] You’re a BAG!</d>\n\nCrunch:\n\n<d>[English] EXACTLY!</d>\n\nThe bubble remains just beyond reach.\n\nWide low-angle composition:\n\nFizz + Crunch foreground\n→ bubble above\n→ fan far overhead.\n\nChill watches from below.\n\nDo NOT show either character falling dangerously.\n\n==================================================\n[SHOT 8 — 0:13.3–0:15.0]\nCHILL GETS AN IDEA — CLIFFHANGER\n\nCut away from the chaos.\n\nClose-up on Chill.\n\nHe calmly looks upward.\n\nHis eyes track:\n\nbubble\n→ ventilation fan\n→ nearby store environment\n→ spoon attached to his back.\n\nUse a subtle visual thought sequence.\n\nNo literal thought bubble required.\n\nHis half-lidded expression changes.\n\nOne eyebrow rises.\n\nThen—\n\nthe smallest confident smile.\n\nChill reaches behind himself.\n\nSlowly removes his silver spoon.\n\nFizz and Crunch turn toward him.\n\nFizz:\n\n<d>[English] Chill?!</d>\n\nCrunch:\n\n<d>[English] Why are you smiling?!</d>\n\nChill looks at the spoon.\n\nThen at the bubble.\n\nChill:\n\n<d>[English] I have an idea.</d>\n\nHold on:\n\nChill foreground with spoon,\nFizz and Crunch staring,\nbubble suspended high in background,\nfan visible above.\n\nMusic stops on a clever suspense sting.\n\nCUT TO BLACK.\n\nEND PART 1.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nSoundscape should feel:\n\n- playful\n- tactile\n- miniature\n- energetic\n- clean\n- cinematic\n- family-friendly\n\n==================================================\nFIZZ VOICE\n==================================================\n\nYoung energetic masculine cartoon voice.\n\nQualities:\n\n- quick\n- confident\n- charismatic\n- expressive\n- slightly dramatic\n- high energy without becoming shrill\n\nEmotional progression:\n\nshow-off\n→ disbelief\n→ panic\n→ desperate determination.\n\nPriority lines:\n\n<d>[English] Still got it.</d>\n\n<d>[English] Wait...</d>\n\n<d>[English] MY BUBBLE!</d>\n\n<d>[English] Don’t let it get away!</d>\n\n<d>[English] Uh-oh.</d>\n\n==================================================\nCRUNCH VOICE\n==================================================\n\nEnergetic comedic masculine/androgynous mascot voice.\n\nQualities:\n\n- excitable\n- earnest\n- slightly squeaky\n- lovable\n- nervous\n\nUse packaging crinkle underneath body movement.\n\nPriority lines:\n\n<d>[English] That was AMAZING!</d>\n\n<d>[English] ONE?!</d>\n\n<d>[English] Stand back!</d>\n\n<d>[English] I can’t stretch anymore!</d>\n\n<d>[English] Why are you smiling?!</d>\n\n==================================================\nCHILL VOICE\n==================================================\n\nLow-energy calm mascot voice.\n\nQualities:\n\n- dry\n- relaxed\n- understated\n- clever\n- almost sleepy\n\nDo NOT make him emotionless.\n\nHis minimalism is the joke.\n\nPriority:\n\n<d>[English] Mm-hm.</d>\n\nOptional:\n\n<d>[English] You don’t.</d>\n\nFinal line MUST be clear:\n\n<d>[English] I have an idea.</d>\n\n==================================================\nSTORE AMBIENCE\n==================================================\n\nUse:\n\n- quiet refrigerator hum\n- fluorescent / practical store ambience\n- subtle ventilation\n- distant compressor\n- small shelf creaks\n- soft floor squeaks\n- tiny packaging rustles\n- very faint outdoor night ambience\n\nBecause the store is closed:\n\nno human voices,\nno shopping crowd,\nno cashier sounds.\n\n==================================================\nBUBBLE SOUND DESIGN\n==================================================\n\nBubble gets a very subtle signature sound.\n\nUse:\n\n- tiny glassy shimmer\n- soft fizzy sparkle\n- gentle elastic “boop” on apple\n- airy upward whoosh near fan\n\nDo NOT make it sound magical or supernatural.\n\nIt is stylized carbonation.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL playful miniature-adventure score.\n\nInstrumentation:\n\n- pizzicato strings\n- muted marimba\n- light bass\n- tiny brass accents\n- brushed percussion\n- woodblocks\n- playful synth plucks\n- occasional glockenspiel\n\nMusic arc:\n\n0:00–0:02\nheroic-comedy entrance.\n\n0:02–0:04\nsudden “something is wrong” motif.\n\n0:04–0:06\nbubble escape acceleration.\n\n0:06–0:10\nfast playful chase rhythm.\n\n0:10–0:13\nrising light suspense.\n\n0:13–0:15\nmusic strips back for Chill’s realization.\n\nFinish with:\n\nshort clever unresolved sting.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nThe camera should emphasize the mascots’ TINY SCALE.\n\nUse:\n\n- low floor-level tracking\n- macro product close-ups\n- wide oversized aisle shots\n- shelf-edge perspective\n- dynamic but readable pans\n- bubble-follow camera\n- rack focus between characters and bubble\n- low-angle ceiling reveal\n- intimate reaction close-ups\n\nCamera energy progression:\n\n0–3 sec:\nconfident and playful.\n\n3–6 sec:\ntighter reactions.\n\n6–10 sec:\nfaster chase.\n\n10–13 sec:\nvertical tension.\n\n13–15 sec:\nslow down for Chill’s idea.\n\nAvoid:\n\n- excessive handheld shake\n- giant action-movie camera spins\n- extreme fisheye\n- incomprehensible rapid cuts\n- heavy motion blur\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium stylized 3D animated product-mascot short.\n\nUse:\n\n- polished CGI\n- expressive cartoon animation\n- glossy packaging materials\n- realistic-enough product reflections\n- soft global illumination\n- detailed miniature environments\n- cinematic depth of field\n- playful squash-and-stretch\n- clean readable silhouettes\n- premium commercial-quality rendering\n- subtle filmic motion blur\n- tactile packaging surfaces\n\nFizz:\nhard glossy aluminum.\n\nCrunch:\nsoft crinkled flexible bag.\n\nChill:\nsemi-matte paper/plastic ice-cream cup.\n\nThis material contrast must remain visible.\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary mascot colors:\n\nFizz:\nbright glossy red + silver + white.\n\nCrunch:\nsaturated yellow + orange.\n\nChill:\ncool mint blue + white + silver spoon.\n\nEnvironment:\n\n- dark navy shadows\n- warm amber practical lighting\n- muted shelf colors\n- warm wood/brown accents\n- cool freezer cyan\n\nLighting should keep all faces readable.\n\nNo horror lighting.\n\nNo flashing emergency lights.\n\n==================================================\nMOTION RULES — FIZZ\n==================================================\n\nFizz is lightweight and energetic.\n\nMovement:\n\n- springy jumps\n- fast arm gestures\n- exaggerated facial acting\n- metallic body stays mostly rigid\n- slight cartoon body squash only\n- fast sneaker movement\n- pull tab remains attached\n\nDo NOT bend the can like rubber.\n\n==================================================\nMOTION RULES — CRUNCH\n==================================================\n\nCrunch can deform more than Fizz because he is flexible packaging.\n\nUse:\n\n- little crinkles\n- puffing\n- squash\n- bounce\n- flappy top edge\n- oversized sneaker motion\n\nDo not flatten him completely.\n\n==================================================\nMOTION RULES — CHILL\n==================================================\n\nChill is the calmest.\n\nMovement:\n\n- minimal head turns\n- slow eye movement\n- controlled walking\n- small eyebrow raises\n- careful spoon removal\n\nHis final slow action should contrast strongly against the chaotic chase.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Fizz.\nONE Crunch.\nONE Chill.\nONE final bubble.\n\nNo duplicate mascots.\n\nNo bubble duplication.\n\nSame:\n\n- character colors\n- footwear\n- glove designs\n- facial layouts\n- packaging shapes\n- spoon\n- bubble size\n- store environment\n\nBubble trajectory must remain continuous across shots.\n\nFizz does NOT regain carbonation in Part 1.\n\nChill does NOT execute the rescue before cut to black.\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\n15 seconds is very short.\n\nNatural pacing is MORE important than using every optional line.\n\nHighest-priority dialogue:\n\n1.\nFizz:\n<d>[English] Still got it.</d>\n\n2.\nCrunch:\n<d>[English] ONE?!</d>\n\n3.\nFizz:\n<d>[English] MY BUBBLE!</d>\n\n4.\nCrunch:\n<d>[English] Stand back!</d>\n\n5.\nFizz:\n<d>[English] Uh-oh.</d>\n\n6.\nCrunch:\n<d>[English] Why are you smiling?!</d>\n\n7.\nChill:\n<d>[English] I have an idea.</d>\n\nIf needed, remove secondary dialogue.\n\nDo NOT speed voices unnaturally.\n\nUse facial acting and SFX to carry comedy.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nAll action is playful and safe.\n\nNo injury.\nNo violence.\nNo weapons.\nNo dangerous machinery contact.\nNo character gets pulled into the fan.\nNo sharp-object threat.\nNo crashing glass.\nNo broken products.\nNo food contamination.\nNo human conflict.\n\nThe ventilation fan remains high and distant.\n\nThe bubble is the only object being affected by its airflow.\n\n==================================================\nNEGATIVE CONSTRAINTS\n==================================================\n\nNo storyboard sheet.\nNo borders.\nNo captions.\nNo timestamps.\nNo production notes.\nNo reference drawings.\nNo watermark.\nNo brand logos.\n\nNo real soda brands.\nNo real snack brands.\nNo real ice-cream brands.\n\nNo duplicate Fizz.\nNo duplicate Crunch.\nNo duplicate Chill.\nNo duplicate bubble.\n\nNo character identity drift.\nNo color changes.\nNo shoe changes.\nNo glove changes.\nNo missing spoon.\nNo disappearing pull tab.\n\nNo humanoid humans joining the chase.\nNo customers.\nNo morning scene.\n\nNo bubble returning to Fizz yet.\nNo Part 2 resolution.\n\nNo dangerous ceiling fan contact.\nNo crushed mascots.\nNo exploding products.\nNo soda explosion.\n\nNo malformed hands.\nNo extra arms.\nNo extra legs.\nNo face distortion.\nNo label flicker.\nNo packaging morphing.\n\n==================================================\nFINAL 3-SECOND PRIORITY\n==================================================\n\nThe final three seconds must create a STRONG Part 2 hook.\n\nBubble rises toward ceiling ventilation.\n\nFizz and Crunch form a ridiculous stretched chain beneath it.\n\nThey are still too short.\n\nFizz reaches desperately.\n\nCrunch struggles below.\n\nThen—\n\nCUT TO CHILL.\n\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2087065646903517354",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-11T06:36:40.683Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2087065646903517354.mp4",
    "posterUrl": "/posters/x/2087065646903517354.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n\"[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for:\n\n“THE LAST POP BEFORE CLOSING”\n\nas the STRICT PRIMARY reference for:\n\n- Fizz’s exact red soda-can design\n- Crunch’s exact yellow snack-bag design\n- Chill’s exact blue ice-cream-cup design\n- character proportions\n- facial features\n- arms, legs, gloves and footwear\n- packaging materials\n- relative mascot scale\n- convenience-store environment\n- soda shelves\n- snack aisle\n- fruit display\n- shopping basket\n- ceiling ventilation fan\n- hanging price tags\n- Chill’s spoon tool\n- bubble design\n- midnight lighting\n- Part 1 story progression\n\nGenerate a FINISHED cinematic 3D animated sequence from the storyboard.\n\nDO NOT animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- captions\n- production notes\n- character reference drawings\n- prop reference drawings\n- page background\n- storyboard labels\n- watermark\n- unnecessary subtitles\n\nTranslate the 15 storyboard panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT make fifteen separate one-second cuts.\n\nThe story progression must read clearly as:\n\nFizz makes a flashy entrance\n→ discovers only one bubble remains\n→ the final bubble escapes\n→ Fizz and Crunch chase it\n→ repeated funny near-misses\n→ bubble reaches ceiling fan\n→ Fizz and Crunch cannot reach it\n→ Chill quietly discovers the solution\n→ cliffhanger.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 1 — 0:00 to 0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium stylized 3D animated product-mascot film\n\nGenre:\n\n- miniature adventure\n- product mascot comedy\n- friendship story\n- playful rescue mission\n\nTone:\n\n- energetic\n- cute\n- funny\n- visually satisfying\n- slightly suspenseful\n- heartwarming\n- family-friendly\n\nDialogue:\nshort synchronized English dialogue\n\nAudio:\nnative dialogue + convenience-store ambience + physical comedy SFX + playful cinematic music\n\nCORE PART 1 IDEA:\n\nFizz thinks he is still the fizziest product in the store...\n\nuntil he discovers he has only ONE bubble left.\n\nWhen that final bubble escapes, his friends launch a tiny midnight rescue mission.\n\n==================================================\nPART 1 STORY LIMIT\n==================================================\n\nSTART:\n\nA quiet convenience store after closing.\n\nFizz suddenly makes a ridiculous heroic entrance riding a giant bubble.\n\nEND:\n\nThe final bubble is almost lost near the ceiling ventilation fan.\n\nFizz and Crunch are helplessly too short.\n\nChill studies the situation, removes the spoon from his back, smiles slightly and says:\n\n<d>[English] I have an idea.</d>\n\nCUT TO BLACK.\n\nDO NOT show:\n\n- Chill executing the solution\n- freezer rescue\n- shopping-basket rescue tower\n- bubble returning to Fizz\n- Fizz becoming fully carbonated again\n- final celebration\n- morning customer\n- shelf ending\n\nSave ALL successful rescue payoff for PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — FIZZ\n==================================================\n\nMatch the storyboard EXACTLY.\n\nFIZZ — RED SODA CAN MASCOT\n\nAppearance:\n\n- small glossy cylindrical red aluminum soda can\n- metallic silver top and bottom rims\n- silver pull tab\n- original fictional red label graphics\n- NO recognizable real-world soda branding\n- large expressive cartoon eyes integrated naturally into can body\n- flexible black eyebrows\n- tiny expressive mouth\n- thin cartoon arms\n- white four-finger-style mascot gloves\n- tiny legs\n- red-and-white sneakers\n\nSurface:\n\n- glossy lacquered aluminum\n- crisp reflections\n- slightly cool metal highlights\n- subtle condensation if appropriate\n\nPersonality:\n\n- energetic\n- dramatic\n- overconfident\n- loves showing off\n- easily panics\n- lovable rather than annoying\n\nPART 1 EMOTIONAL ARC:\n\n0–2 sec:\nmaximum confidence\n\n2–4 sec:\nconfusion\n\n4–6 sec:\npanic\n\n6–11 sec:\ndesperate comic determination\n\n11–13 sec:\nfear of losing the bubble\n\n13–15 sec:\nconfused hope as Chill gets an idea\n\nIMPORTANT:\n\nFizz must remain clearly identifiable as the SAME red can in every shot.\n\nDo not alter:\n\n- shape\n- height\n- face placement\n- silver top\n- shoes\n- glove style\n- red color\n\n==================================================\nCHARACTER IDENTITY LOCK — CRUNCH\n==================================================\n\nCRUNCH — YELLOW SNACK BAG MASCOT\n\nAppearance:\n\n- small puffy rectangular snack bag\n- bright warm yellow packaging\n- crinkled foil/plastic texture\n- sealed ridged top edge\n- original fictional package graphics only\n- large expressive eyes\n- tiny mouth\n- short flexible arms\n- white gloves\n- tiny legs\n- oversized orange sneakers\n\nPersonality:\n\n- enthusiastic\n- loyal\n- easily alarmed\n- physically comedic\n- wants to be heroic\n- often makes things more chaotic\n\nPART 1 PERFORMANCE:\n\n0–2 sec:\nFizz fanboy / applause\n\n2–4 sec:\ninstant concern\n\n4–6 sec:\ndramatic panic\n\n6–10 sec:\noverenthusiastic rescue attempts\n\n10–13 sec:\nfull comic emergency mode\n\n13–15 sec:\nconfused by Chill’s calmness\n\nPackaging motion:\n\nCrunch’s bag should lightly crinkle, squash and puff with emotion.\n\nDo NOT deform him so much that his identity changes.\n\n==================================================\nCHARACTER IDENTITY LOCK — CHILL\n==================================================\n\nCHILL — BLUE ICE-CREAM CUP MASCOT\n\nAppearance:\n\n- compact cylindrical mint-blue / cool-blue ice cream cup\n- white plastic lid resembling a flat cap\n- clean original fictional cup graphics\n- relaxed half-lidded expressive eyes\n- small understated mouth\n- tiny flexible arms\n- white gloves\n- small legs\n- white sneakers\n- one silver spoon attached vertically/diagonally to his back like a tiny tool\n\nPersonality:\n\n- extremely calm\n- intelligent\n- observant\n- dry sense of humor\n- speaks rarely\n- never panics\n\nPART 1 PERFORMANCE:\n\n0–5 sec:\nunimpressed observer\n\n5–10 sec:\nquietly follows and studies\n\n10–13 sec:\nanalyzes fan + bubble trajectory\n\n13–15 sec:\nsolution clicks into place\n\nHis comedy comes from contrast:\n\nFizz and Crunch = chaos.\n\nChill = almost completely calm.\n\n==================================================\nKEY PROP LOCK — LAST BUBBLE\n==================================================\n\nThe final bubble is the MAIN STORY OBJECT.\n\nAppearance:\n\n- one transparent soap-like carbonation bubble\n- roughly the size of Fizz’s head or slightly smaller\n- subtle rainbow iridescence\n- reflective convenience-store lights on surface\n- transparent center\n- physically delicate\n- readable against dark backgrounds\n\nThe bubble must remain:\n\n# ONE SINGLE BUBBLE\n\nNever duplicate it.\n\nTrajectory must be physically understandable.\n\nIt moves through the story:\n\nFizz\n→ floats upward\n→ drifts into aisle\n→ passes candy / price tags\n→ Crunch misses it\n→ bounces gently from shiny apple\n→ rises toward ceiling\n→ ventilation airflow begins drawing it upward.\n\nThe audience must always understand where the bubble is going.\n\n==================================================\n“1 BUBBLE LEFT” VISUAL\n==================================================\n\nAt the discovery beat, briefly show a simple ORIGINAL diegetic graphic beside or reflected near Fizz:\n\n“1 BUBBLE LEFT”\n\nStyle:\n\n- tiny red/orange digital indicator\n- simple iconography\n- readable for less than one second\n- not a permanent HUD\n- not a video-game interface\n\nIt should feel like a comedic internal soda-status visualization.\n\nDo NOT cover the whole frame with UI.\n\n==================================================\nENVIRONMENT LOCK — MIDNIGHT CONVENIENCE STORE\n==================================================\n\nMaintain the uploaded storyboard environment.\n\nStore:\n\n- closed for the night\n- organized aisles\n- soda shelves\n- snack shelves\n- candy section\n- fruit display\n- shopping baskets\n- small paper price tags\n- refrigerators/freezers\n- ceiling ventilation fans\n- glossy tile floor\n- large storefront windows\n- shelves dramatically oversized compared with the tiny mascots\n\nLighting:\n\n- warm practical ceiling lights\n- soft orange shelf illumination\n- cool midnight blue shadows\n- occasional moonlight from storefront windows\n- reflections on polished floor\n- subtle refrigerator cyan light\n\nThe store should feel huge from mascot scale.\n\nUse miniature cinematography.\n\n==================================================\nWORLD / PHYSICS RULES\n==================================================\n\nThe characters are tiny product mascots living secretly inside a normal convenience store.\n\nMaintain believable miniature scale.\n\nFizz jumping:\nlightweight metal-can motion.\n\nCrunch:\nsoft bag bounce and packaging crinkle.\n\nChill:\nsmall solid cup motion with slightly heavier stable footing.\n\nBubble physics:\n\n- extremely lightweight\n- slow upward drift\n- reacts to air currents\n- gently rebounds from objects\n- never behaves like a heavy ball\n- never explodes in Part 1\n\nCAUSE → EFFECT must remain readable.\n\nFizz releases last bubble\n→ bubble rises.\n\nCharacters chase\n→ air movement causes slight drift.\n\nCrunch misses\n→ bubble continues.\n\nBubble touches apple\n→ gently redirects upward.\n\nCeiling fan airflow\n→ bubble starts being pulled toward fan.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.8]\nMIDNIGHT — THE GREAT FIZZ ENTRANCE\n\nOpen outside / just inside the convenience store at midnight.\n\nEstablish:\n\n- dark blue night sky\n- quiet storefront\n- warm store lights glowing inside\n\nMove quickly into the soda aisle.\n\nSuddenly—\n\n# POP!\n\nFizz shoots upward from the soda shelf riding on top of a large fizzy bubble.\n\nHe balances dramatically like a tiny action hero.\n\nSeveral tiny harmless carbonation bubbles trail around him.\n\nHe lands perfectly on the front edge of the shelf.\n\nCrunch immediately applauds enthusiastically.\n\nChill stands nearby with half-lidded eyes and almost no reaction.\n\nFizz opens his arms proudly.\n\nFizz:\n\n<d>[English] Still got it.</d>\n\nCrunch:\n\n<d>[English] That was AMAZING!</d>\n\nChill gives a tiny unimpressed:\n\n<d>[English] Mm-hm.</d>\n\nCAMERA:\n\nwide store establish\n→ fast push toward soda shelf\n→ low-angle Fizz entrance\n→ medium three-character reaction.\n\nSFX:\n\n- fizzy POP\n- bubble wobble\n- tiny sneaker landing\n- Crunch package crinkle\n- small celebratory sparkle sound\n\nMUSIC:\n\nplayful miniature heist/adventure theme begins.\n\n==================================================\n[SHOT 2 — 0:01.8–0:03.6]\nONE BUBBLE LEFT\n\nMedium close-up on Fizz.\n\nHe continues posing proudly.\n\nThen—\n\na tiny pathetic bubble emerges from near his pull tab.\n\nSFX:\n\n# “pip.”\n\nFizz slowly looks upward.\n\nThe bubble floats beside him.\n\nA tiny diegetic indicator flashes:\n\n# 1 BUBBLE LEFT\n\nFizz’s smile disappears.\n\nCrunch stops applauding.\n\nChill raises one eyebrow.\n\nFizz:\n\n<d>[English] Wait...</d>\n\nFizz gently shakes himself.\n\nNo bubbles.\n\nHe shakes again.\n\nNothing.\n\nFizz:\n\n<d>[English] That can’t be right.</d>\n\nCrunch:\n\n<d>[English] ONE?!</d>\n\nChill remains calm.\n\nCamera:\n\nFizz hero close-up\n→ tiny bubble\n→ status indicator\n→ reaction three-shot.\n\nMusic drops into a comic suspense note.\n\n==================================================\n[SHOT 3 — 0:03.6–0:05.4]\nTHE FINAL BUBBLE ESCAPES\n\nFizz tries increasingly silly but safe methods to make more carbonation:\n\ntiny bounce,\nsmall shake,\nquick spin.\n\nNothing.\n\nThen—\n\nthe ONE remaining bubble slowly drifts away from him.\n\nFizz freezes.\n\nHis pupils track it.\n\nFizz:\n\n<d>[English] No no no no...</d>\n\nBubble clears the edge of the shelf.\n\nFizz reaches forward.\n\nToo late.\n\nFizz:\n\n<d>[English] MY BUBBLE!</d>\n\nCrunch gasps dramatically.\n\nChill turns his eyes toward the bubble’s trajectory.\n\nCamera follows the bubble over the shelf edge.\n\n==================================================\n[SHOT 4 — 0:05.4–0:07.5]\nTHE STORE-WIDE CHASE\n\nFizz leaps safely down from the low product display and runs after the floating bubble.\n\nCrunch runs immediately behind him, bag bouncing and crinkling wildly.\n\nChill calmly climbs/slides down and follows at an efficient pace.\n\nThe camera tracks low at mascot height.\n\nThe bubble floats through the giant store aisle.\n\nPass:\n\n- candy displays\n- hanging price cards\n- colorful product shelves\n- oversized boxes\n- polished tiles\n\nFizz stretches both hands upward while running.\n\nFizz:\n\n<d>[English] Don’t let it get away!</d>\n\nCrunch:\n\n<d>[English] I GOT IT!</d>\n\nChill, quietly:\n\n<d>[English] You don’t.</d>\n\nStrong visual scale:\n\ntiny mascots below,\nhuge supermarket aisle above.\n\n==================================================\n[SHOT 5 — 0:07.5–0:09.4]\nCRUNCH’S GREAT RESCUE ATTEMPT\n\nCrunch spots an empty decorative snack bowl / tray.\n\nHis eyes light up.\n\nHe grabs it with both hands.\n\nCrunch:\n\n<d>[English] Stand back!</d>\n\nHe runs underneath the bubble.\n\nMoves bowl left.\n\nBubble moves right.\n\nHe moves right.\n\nBubble moves left.\n\nCrunch finally jumps and thrusts the bowl upward.\n\nMISS.\n\nFizz also makes a tiny dive beneath it.\n\nMISS.\n\nCrunch lands seated inside / behind the bowl, stunned but unharmed.\n\nFizz:\n\n<d>[English] Great plan!</d>\n\nCrunch:\n\n<d>[English] Thank you!</d>\n\nBeat.\n\nFizz:\n\n<d>[English] I was being sarcastic!</d>\n\nIf dialogue timing is crowded, prioritize:\n“Stand back!” + visual comedy.\n\nBubble continues forward.\n\n==================================================\n[SHOT 6 — 0:09.4–0:11.0]\nTHE APPLE BOUNCE\n\nThe last bubble drifts toward the fruit display.\n\nClose macro shot:\n\na shiny red apple reflects the bubble and tiny running mascots.\n\nBubble gently touches the apple.\n\n# boop.\n\nIt redirects upward.\n\nFizz arrives just beneath it.\n\nHis eyes widen.\n\nFizz:\n\n<d>[English] Uh-oh.</d>\n\nCamera tilts vertically following the bubble.\n\nReveal:\n\na ceiling ventilation fan high above.\n\nThe bubble begins drifting toward it.\n\nMusic shifts from playful chase to light suspense.\n\nChill stops running.\n\nHe studies the situation.\n\n==================================================\n[SHOT 7 — 0:11.0–0:13.3]\nTOO SHORT\n\nThe ventilation fan rotates above.\n\nIMPORTANT:\n\nDo not make it threatening or dangerous-looking.\n\nIt is simply creating airflow.\n\nThe bubble begins moving upward faster.\n\nFizz jumps repeatedly.\n\nToo short.\n\nCrunch rushes behind him.\n\nFizz grabs a safe shelf lip / hanging display support.\n\nCrunch holds Fizz around the legs.\n\nThey stretch upward in a ridiculous mascot chain.\n\nFizz:\n\n<d>[English] Almost...</d>\n\nCrunch:\n\n<d>[English] I can’t stretch anymore!</d>\n\nFizz:\n\n<d>[English] You’re a BAG!</d>\n\nCrunch:\n\n<d>[English] EXACTLY!</d>\n\nThe bubble remains just beyond reach.\n\nWide low-angle composition:\n\nFizz + Crunch foreground\n→ bubble above\n→ fan far overhead.\n\nChill watches from below.\n\nDo NOT show either character falling dangerously.\n\n==================================================\n[SHOT 8 — 0:13.3–0:15.0]\nCHILL GETS AN IDEA — CLIFFHANGER\n\nCut away from the chaos.\n\nClose-up on Chill.\n\nHe calmly looks upward.\n\nHis eyes track:\n\nbubble\n→ ventilation fan\n→ nearby store environment\n→ spoon attached to his back.\n\nUse a subtle visual thought sequence.\n\nNo literal thought bubble required.\n\nHis half-lidded expression changes.\n\nOne eyebrow rises.\n\nThen—\n\nthe smallest confident smile.\n\nChill reaches behind himself.\n\nSlowly removes his silver spoon.\n\nFizz and Crunch turn toward him.\n\nFizz:\n\n<d>[English] Chill?!</d>\n\nCrunch:\n\n<d>[English] Why are you smiling?!</d>\n\nChill looks at the spoon.\n\nThen at the bubble.\n\nChill:\n\n<d>[English] I have an idea.</d>\n\nHold on:\n\nChill foreground with spoon,\nFizz and Crunch staring,\nbubble suspended high in background,\nfan visible above.\n\nMusic stops on a clever suspense sting.\n\nCUT TO BLACK.\n\nEND PART 1.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nSoundscape should feel:\n\n- playful\n- tactile\n- miniature\n- energetic\n- clean\n- cinematic\n- family-friendly\n\n==================================================\nFIZZ VOICE\n==================================================\n\nYoung energetic masculine cartoon voice.\n\nQualities:\n\n- quick\n- confident\n- charismatic\n- expressive\n- slightly dramatic\n- high energy without becoming shrill\n\nEmotional progression:\n\nshow-off\n→ disbelief\n→ panic\n→ desperate determination.\n\nPriority lines:\n\n<d>[English] Still got it.</d>\n\n<d>[English] Wait...</d>\n\n<d>[English] MY BUBBLE!</d>\n\n<d>[English] Don’t let it get away!</d>\n\n<d>[English] Uh-oh.</d>\n\n==================================================\nCRUNCH VOICE\n==================================================\n\nEnergetic comedic masculine/androgynous mascot voice.\n\nQualities:\n\n- excitable\n- earnest\n- slightly squeaky\n- lovable\n- nervous\n\nUse packaging crinkle underneath body movement.\n\nPriority lines:\n\n<d>[English] That was AMAZING!</d>\n\n<d>[English] ONE?!</d>\n\n<d>[English] Stand back!</d>\n\n<d>[English] I can’t stretch anymore!</d>\n\n<d>[English] Why are you smiling?!</d>\n\n==================================================\nCHILL VOICE\n==================================================\n\nLow-energy calm mascot voice.\n\nQualities:\n\n- dry\n- relaxed\n- understated\n- clever\n- almost sleepy\n\nDo NOT make him emotionless.\n\nHis minimalism is the joke.\n\nPriority:\n\n<d>[English] Mm-hm.</d>\n\nOptional:\n\n<d>[English] You don’t.</d>\n\nFinal line MUST be clear:\n\n<d>[English] I have an idea.</d>\n\n==================================================\nSTORE AMBIENCE\n==================================================\n\nUse:\n\n- quiet refrigerator hum\n- fluorescent / practical store ambience\n- subtle ventilation\n- distant compressor\n- small shelf creaks\n- soft floor squeaks\n- tiny packaging rustles\n- very faint outdoor night ambience\n\nBecause the store is closed:\n\nno human voices,\nno shopping crowd,\nno cashier sounds.\n\n==================================================\nBUBBLE SOUND DESIGN\n==================================================\n\nBubble gets a very subtle signature sound.\n\nUse:\n\n- tiny glassy shimmer\n- soft fizzy sparkle\n- gentle elastic “boop” on apple\n- airy upward whoosh near fan\n\nDo NOT make it sound magical or supernatural.\n\nIt is stylized carbonation.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL playful miniature-adventure score.\n\nInstrumentation:\n\n- pizzicato strings\n- muted marimba\n- light bass\n- tiny brass accents\n- brushed percussion\n- woodblocks\n- playful synth plucks\n- occasional glockenspiel\n\nMusic arc:\n\n0:00–0:02\nheroic-comedy entrance.\n\n0:02–0:04\nsudden “something is wrong” motif.\n\n0:04–0:06\nbubble escape acceleration.\n\n0:06–0:10\nfast playful chase rhythm.\n\n0:10–0:13\nrising light suspense.\n\n0:13–0:15\nmusic strips back for Chill’s realization.\n\nFinish with:\n\nshort clever unresolved sting.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nThe camera should emphasize the mascots’ TINY SCALE.\n\nUse:\n\n- low floor-level tracking\n- macro product close-ups\n- wide oversized aisle shots\n- shelf-edge perspective\n- dynamic but readable pans\n- bubble-follow camera\n- rack focus between characters and bubble\n- low-angle ceiling reveal\n- intimate reaction close-ups\n\nCamera energy progression:\n\n0–3 sec:\nconfident and playful.\n\n3–6 sec:\ntighter reactions.\n\n6–10 sec:\nfaster chase.\n\n10–13 sec:\nvertical tension.\n\n13–15 sec:\nslow down for Chill’s idea.\n\nAvoid:\n\n- excessive handheld shake\n- giant action-movie camera spins\n- extreme fisheye\n- incomprehensible rapid cuts\n- heavy motion blur\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium stylized 3D animated product-mascot short.\n\nUse:\n\n- polished CGI\n- expressive cartoon animation\n- glossy packaging materials\n- realistic-enough product reflections\n- soft global illumination\n- detailed miniature environments\n- cinematic depth of field\n- playful squash-and-stretch\n- clean readable silhouettes\n- premium commercial-quality rendering\n- subtle filmic motion blur\n- tactile packaging surfaces\n\nFizz:\nhard glossy aluminum.\n\nCrunch:\nsoft crinkled flexible bag.\n\nChill:\nsemi-matte paper/plastic ice-cream cup.\n\nThis material contrast must remain visible.\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary mascot colors:\n\nFizz:\nbright glossy red + silver + white.\n\nCrunch:\nsaturated yellow + orange.\n\nChill:\ncool mint blue + white + silver spoon.\n\nEnvironment:\n\n- dark navy shadows\n- warm amber practical lighting\n- muted shelf colors\n- warm wood/brown accents\n- cool freezer cyan\n\nLighting should keep all faces readable.\n\nNo horror lighting.\n\nNo flashing emergency lights.\n\n==================================================\nMOTION RULES — FIZZ\n==================================================\n\nFizz is lightweight and energetic.\n\nMovement:\n\n- springy jumps\n- fast arm gestures\n- exaggerated facial acting\n- metallic body stays mostly rigid\n- slight cartoon body squash only\n- fast sneaker movement\n- pull tab remains attached\n\nDo NOT bend the can like rubber.\n\n==================================================\nMOTION RULES — CRUNCH\n==================================================\n\nCrunch can deform more than Fizz because he is flexible packaging.\n\nUse:\n\n- little crinkles\n- puffing\n- squash\n- bounce\n- flappy top edge\n- oversized sneaker motion\n\nDo not flatten him completely.\n\n==================================================\nMOTION RULES — CHILL\n==================================================\n\nChill is the calmest.\n\nMovement:\n\n- minimal head turns\n- slow eye movement\n- controlled walking\n- small eyebrow raises\n- careful spoon removal\n\nHis final slow action should contrast strongly against the chaotic chase.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Fizz.\nONE Crunch.\nONE Chill.\nONE final bubble.\n\nNo duplicate mascots.\n\nNo bubble duplication.\n\nSame:\n\n- character colors\n- footwear\n- glove designs\n- facial layouts\n- packaging shapes\n- spoon\n- bubble size\n- store environment\n\nBubble trajectory must remain continuous across shots.\n\nFizz does NOT regain carbonation in Part 1.\n\nChill does NOT execute the rescue before cut to black.\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\n15 seconds is very short.\n\nNatural pacing is MORE important than using every optional line.\n\nHighest-priority dialogue:\n\n1.\nFizz:\n<d>[English] Still got it.</d>\n\n2.\nCrunch:\n<d>[English] ONE?!</d>\n\n3.\nFizz:\n<d>[English] MY BUBBLE!</d>\n\n4.\nCrunch:\n<d>[English] Stand back!</d>\n\n5.\nFizz:\n<d>[English] Uh-oh.</d>\n\n6.\nCrunch:\n<d>[English] Why are you smiling?!</d>\n\n7.\nChill:\n<d>[English] I have an idea.</d>\n\nIf needed, remove secondary dialogue.\n\nDo NOT speed voices unnaturally.\n\nUse facial acting and SFX to carry comedy.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nAll action is playful and safe.\n\nNo injury.\nNo violence.\nNo weapons.\nNo dangerous machinery contact.\nNo character gets pulled into the fan.\nNo sharp-object threat.\nNo crashing glass.\nNo broken products.\nNo food contamination.\nNo human conflict.\n\nThe ventilation fan remains high and distant.\n\nThe bubble is the only object being affected by its airflow.\n\n==================================================\nNEGATIVE CONSTRAINTS\n==================================================\n\nNo storyboard sheet.\nNo borders.\nNo captions.\nNo timestamps.\nNo production notes.\nNo reference drawings.\nNo watermark.\nNo brand logos.\n\nNo real soda brands.\nNo real snack brands.\nNo real ice-cream brands.\n\nNo duplicate Fizz.\nNo duplicate Crunch.\nNo duplicate Chill.\nNo duplicate bubble.\n\nNo character identity drift.\nNo color changes.\nNo shoe changes.\nNo glove changes.\nNo missing spoon.\nNo disappearing pull tab.\n\nNo humanoid humans joining the chase.\nNo customers.\nNo morning scene.\n\nNo bubble returning to Fizz yet.\nNo Part 2 resolution.\n\nNo dangerous ceiling fan contact.\nNo crushed mascots.\nNo exploding products.\nNo soda explosion.\n\nNo malformed hands.\nNo extra arms.\nNo extra legs.\nNo face distortion.\nNo label flicker.\nNo packaging morphing.\n\n==================================================\nFINAL 3-SECOND PRIORITY\n==================================================\n\nThe final three seconds must create a STRONG Part 2 hook.\n\nBubble rises toward ceiling ventilation.\n\nFizz and Crunch form a ridiculous stretched chain beneath it.\n\nThey are still too short.\n\nFizz reaches desperately.\n\nCrunch struggles below.\n\nThen—\n\nCUT TO CHILL.\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2087065646903517354"
  },
  {
    "id": "x-2086667788488442041",
    "title": "阿里与布鲁诺的羊角包花 · 下篇",
    "titleEn": "Ari and Bruno’s Croissant Bloom · Part 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《阿里与布鲁诺的羊角包花 · 下篇》的 H3 成片。本页保留原始 X 来源链接。来源未公开完整 Prompt，本库不反推或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Ari and Bruno’s Croissant Bloom · Part 2. This page retains the original X source link. The source did not publish a complete Prompt, so this library does not infer or complete one. Hosted-video output: 15s · 1280×720.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2086667788488442041",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-10T04:15:43.845Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086667788488442041.mp4",
    "posterUrl": "/posters/x/2086667788488442041.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "food",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seocnd Storyboard image to video using @Hailuo_AI's MiniMax H3 on @higgsfield_ai @higgsfield\n\nPrompt :  \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for:\n\n“ARI & BRUNO — THE CROISSANT BLOOMS”\n\nas the STRICT PRIMARY visual, character, environment, prop, food-design, lighting, performance, and story-continuity reference.\n\nThis sequence continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n\n- Ari believing he had accidentally ruined the raw croissants\n- Bruno remaining calm and supportive\n- Bruno reshaping one squashed pastry into a flower\n- Ari realizing the mistake could become a new idea\n- Ari saying:\n  <d>[English] Then let’s make them bloom.</d>\n- Ari and Bruno beginning to reshape the remaining raw pastries together\n\nPART 2 begins AFTER they have completed the flower-shaped raw pastries and baked them.\n\nGenerate a FINISHED cinematic animated short from the storyboard.\n\nDO NOT animate the storyboard sheet itself.\n\nThe finished video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character-reference drawings\n- prop-reference drawings\n- color swatches\n- storyboard-paper background\n- title/header\n- page labels\n- subtitles\n- speech bubbles\n- watermarks\n- logos\n\nThe uploaded storyboard locks:\n\n1. Ari’s exact face\n2. Ari’s short tousled brown hair\n3. Ari’s youthful slim proportions\n4. Ari’s white/cream baker shirt\n5. Ari’s rolled sleeves\n6. Ari’s light apron\n7. Bruno’s exact face\n8. Bruno’s curly dark hair\n9. Bruno’s beard\n10. Bruno’s large heavyset build\n11. Bruno’s cream baker clothing/apron\n12. same cozy artisan bakery\n13. same wooden counters and shelving\n14. same warm morning lighting\n15. same oven area\n16. same baking tray\n17. same croissant-flower concept\n18. baked croissant bloom design\n19. powdered sugar\n20. berries/apricot accents\n21. serving platter/cake stand\n22. mentor-apprentice emotional relationship\n23. Ari’s nervous-to-proud performance arc\n24. Bruno’s calm encouraging personality\n\nTranslate the storyboard’s fifteen one-second panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT make fifteen frantic cuts.\n\nNarrative progression:\n\noven reveal\n→ beautiful baked transformation\n→ delicate finishing\n→ Ari spots imperfection\n→ Bruno calmly fixes it\n→ hero food reveal\n→ customer reaction\n→ Ari gains confidence\n→ bakery comes alive\n→ final mentor/apprentice payoff.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 2 — 0:15 to 0:30\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D animated-film style\n\nGenre:\ncozy bakery story / uplifting mentor-apprentice short\n\nTone:\n\n- warm\n- satisfying\n- tactile\n- charming\n- lightly comedic\n- emotionally uplifting\n- delicious-looking\n- wholesome\n- family-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + bakery ambience + baking sounds + subtle customer reactions + original cinematic score\n\nCORE PAYOFF:\n\nThe pastries Ari thought were ruined become the bakery’s most beautiful creation.\n\nBut the real transformation is Ari:\n\nworried apprentice\n→ proud young baker.\n\n==================================================\nPART 2 START / END\n==================================================\n\nSTART:\n\nBruno safely opens the bakery oven.\n\nWarm steam escapes.\n\nAri leans forward.\n\nTheir experimental flower-shaped pastries have baked beautifully.\n\nEND:\n\nAri confidently presents the finished Croissant Bloom.\n\nCustomers are delighted.\n\nBruno stands proudly beside him.\n\nFinish with Ari and Bruno together behind the counter, glowing bakery around them.\n\nThe story must feel COMPLETE.\n\nNO cliffhanger.\n\n==================================================\nCHARACTER IDENTITY LOCK — ARI\n==================================================\n\nMatch the uploaded storyboard and Part 1 EXACTLY.\n\nARI — APPRENTICE BAKER\n\nAppearance:\n\n- young teenage boy\n- warm light skin\n- short tousled brown hair\n- large expressive dark/brown eyes\n- soft youthful face\n- slim body\n- narrow shoulders\n- white/cream baker shirt\n- rolled sleeves\n- light beige apron\n- simple work trousers\n\nDo NOT alter:\n\n- face\n- age\n- hairstyle\n- proportions\n- shirt\n- apron\n- coloring\n\nPERSONALITY:\n\n- earnest\n- hardworking\n- careful\n- talented but inexperienced\n- slightly perfectionistic\n- eager to learn\n- kind\n- quietly creative\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\nhopeful anticipation\n\n3–5 sec:\namazed by successful bake\n\n5–7 sec:\ncareful craftsmanship\n\n7–9 sec:\nbrief perfectionist worry\n\n9–11 sec:\nreassured\n\n11–13 sec:\nsurprised by customer approval\n\n13–15 sec:\ngenuine confidence and pride\n\nIMPORTANT:\n\nAri should not instantly become wildly confident.\n\nHis pride grows through the sequence.\n\nUse small acting details:\n\n- widened eyes\n- careful hands\n- tiny nervous breath\n- relieved smile\n- shoulders gradually relaxing\n- final stronger posture\n\n==================================================\nCHARACTER IDENTITY LOCK — BRUNO\n==================================================\n\nMatch PART 1 EXACTLY.\n\nBRUNO — HEAD BAKER\n\nAppearance:\n\n- large heavyset adult build\n- curly dark hair\n- full short beard\n- warm expressive eyes\n- broad face\n- large friendly hands\n- light baker shirt\n- cream apron\n\nPersonality:\n\n- calm\n- experienced\n- cheerful\n- patient\n- encouraging\n- playful\n- deeply supportive of Ari\n\nPERFORMANCE ARC:\n\n0–3 sec:\nproud anticipation\n\n3–7 sec:\nquiet mentor satisfaction\n\n7–10 sec:\ncalm reassurance\n\n10–13 sec:\nlets Ari own the success\n\n13–15 sec:\nproud mentor\n\nBruno should NEVER steal Ari’s final hero moment.\n\nBy the ending:\n\nBruno moves slightly into the background while Ari presents the pastry.\n\nHis expression should communicate:\n\n“You made this.”\n\n==================================================\nCROISSANT BLOOM — STRICT FOOD DESIGN LOCK\n==================================================\n\nThis is the visual centerpiece.\n\nUse the exact concept from the storyboard.\n\nThe finished Croissant Bloom consists of:\n\n- multiple golden baked croissant/flower pastries\n- arranged radially like overlapping flower petals\n- circular bouquet-like structure\n- crisp flaky laminated layers\n- buttery golden-brown surface\n- slightly darker toasted edges\n- small berry / apricot accents clustered near center\n- delicate powdered sugar finish\n- presented on a round platter / cake stand\n\nIt must look:\n\n- handmade\n- delicious\n- premium\n- warm\n- elegant\n- slightly imperfect in a charming artisanal way\n\nDo NOT make it look like:\n\n- cake\n- pizza\n- giant single croissant\n- plastic prop\n- unrealistic floral sculpture\n- magical glowing object\n\nIt remains believable pastry.\n\n==================================================\nFOOD CONTINUITY\n==================================================\n\nPart 1:\n\npale raw dough flowers.\n\nPart 2:\n\nthe SAME flower pastries are now baked.\n\nThe transformation is caused by OVEN BAKING.\n\nCorrect sequence:\n\nraw flowers\n→ oven\n→ baked golden flowers\n→ plating\n→ powdered sugar\n→ fruit accents\n→ finished Croissant Bloom.\n\nNo magical morph.\n\n==================================================\nENVIRONMENT LOCK — ARI & BRUNO BAKERY\n==================================================\n\nMaintain the same artisan bakery.\n\nInclude:\n\n- wooden counters\n- warm shelving\n- glass and ceramic jars\n- flour containers\n- bowls\n- old-fashioned oven\n- mixing tools\n- baking trays\n- rustic wood\n- baskets\n- cloth towels\n- berries\n- morning light\n- subtle flour particles\n\nLighting:\n\nwarm morning sunlight from the window.\n\nSecondary:\n\namber oven glow.\n\nBy the ending:\n\nthe bakery feels slightly brighter and livelier because customers have entered.\n\nDo NOT change the building or create a new location.\n\n==================================================\nCUSTOMER DESIGN\n==================================================\n\nIntroduce customers ONLY during the second half.\n\nUse a very small group.\n\nPrimary customer:\n\na friendly bakery regular, parent with a child, or warm young adult customer.\n\nExpression:\n\n- curious\n- delighted\n- impressed\n- warm\n\nOptional additional customers near ending:\n\n2–4 people maximum.\n\nKeep them visually secondary.\n\nThe story remains about:\n\nARI + BRUNO.\n\nNo huge crowd.\n\nNo influencer cameras.\n\nNo modern social-media behavior.\n\n==================================================\nWORLD / ACTION LOGIC\n==================================================\n\nBruno opens oven\n→ heat/steam reveals baked pastries.\n\nAri sees they worked\n→ amazement.\n\nPastries cool briefly / move to finishing station\n→ Ari decorates.\n\nAri notices one uneven petal\n→ perfectionist worry returns.\n\nBruno adjusts it gently\n→ Ari learns imperfection is manageable.\n\nBloom completed\n→ hero reveal.\n\nCustomer sees pastry\n→ delighted reaction.\n\nPraise reaches Ari\n→ confidence rises.\n\nBruno encourages Ari\n→ Ari presents it himself.\n\nMore bakery visitors notice\n→ morning becomes joyful.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.8]\nTHE OVEN REVEAL\n\nContinue naturally from Part 1.\n\nWarm bakery morning.\n\nBruno, wearing a safe oven cloth/mitt, opens the oven.\n\nA soft wave of warm steam rises.\n\nAmber oven light spills onto Bruno and Ari.\n\nAri leans forward from a safe distance.\n\nBruno carefully slides the baking tray outward.\n\nREVEAL:\n\ntheir flower-shaped pastries have puffed into beautiful golden layered blooms.\n\nAri’s eyes become huge.\n\nAri:\n\n<d>[English] Bruno...</d>\n\nBruno smiles.\n\n<d>[English] Take a look.</d>\n\nCAMERA:\n\nmedium oven shot\n→ gentle push toward tray\n→ Ari reaction close-up.\n\nSFX:\n\n- oven door\n- subtle oven crackle\n- tray glide\n- warm steam\n- quiet bakery ambience\n\nMUSIC:\n\nPart 1 creative motif returns, now warmer and fuller.\n\n==================================================\n[SHOT 2 — 0:01.8–0:03.4]\nTHE CROISSANTS BLOOMED\n\nMacro food hero photography-style animation.\n\nCamera glides slowly over the tray.\n\nShow:\n\n- flaky golden layers\n- crisp edges\n- flower-like shapes\n- buttery highlights\n- tiny warm steam\n- rich pastry texture\n\nAri carefully lifts/rotates the tray on the work surface.\n\nBruno stands behind him proudly.\n\nAri:\n\n<d>[English] They actually bloomed.</d>\n\nBruno:\n\n<d>[English] Told you.</d>\n\nAri looks toward Bruno.\n\nBruno adds, playfully:\n\n<d>[English] Mostly.</d>\n\nTiny smile.\n\nKeep comedy subtle.\n\n==================================================\n[SHOT 3 — 0:03.4–0:05.2]\nTHE FINAL DELICATE TOUCH\n\nMove to decorating station.\n\nThe pastries are arranged into a circular floral display.\n\nAri focuses intensely.\n\nUse tactile close-ups.\n\nFirst:\n\nAri places several small red berries / apricot pieces into the center.\n\nThen:\n\nhe gently shakes a fine layer of powdered sugar.\n\nPowder falls through golden morning light.\n\nVery beautiful slow visual moment.\n\nNot full slow motion—just graceful.\n\nBruno watches nearby.\n\nBruno:\n\n<d>[English] Easy...</d>\n\nAri carefully finishes.\n\n<d>[English] Got it.</d>\n\nSFX:\n\n- berry placement taps\n- metal shaker\n- powdered sugar fall\n- soft plate movement\n\n==================================================\n[SHOT 4 — 0:05.2–0:06.8]\nONE PETAL IS WRONG\n\nAri steps backward to admire the Croissant Bloom.\n\nHis smile begins.\n\nThen stops.\n\nHis eyes lock onto ONE slightly tilted pastry petal.\n\nCamera push toward the small imperfection.\n\nBack to Ari.\n\nHis brows tighten.\n\nAri:\n\n<d>[English] Wait.</d>\n\nHe leans closer.\n\n<d>[English] That one’s crooked.</d>\n\nBruno looks.\n\nSmall pause.\n\nBruno:\n\n<d>[English] So?</d>\n\nAri turns toward him, surprised.\n\nKeep this beat light and charming.\n\n==================================================\n[SHOT 5 — 0:06.8–0:08.3]\nBRUNO KNOWS JUST WHAT TO DO\n\nBruno steps beside Ari.\n\nHe gently reaches toward the slightly tilted pastry.\n\nWith TWO fingers:\n\ntiny adjustment.\n\nThat is all.\n\nThe pastry sits naturally within the flower arrangement.\n\nBruno:\n\n<d>[English] Handmade things don’t need to be identical.</d>\n\nIf dialogue timing is tight, shorten to:\n\n<d>[English] Handmade isn’t perfect.</d>\n\nAri studies the bloom.\n\nBruno:\n\n<d>[English] It has character.</d>\n\nAri gives a tiny relieved smile.\n\nThis should echo Part 1’s message without becoming preachy.\n\n==================================================\n[SHOT 6 — 0:08.3–0:10.0]\nTHE CROISSANT BLOOM — HERO REVEAL\n\nAri and Bruno together place the finished Croissant Bloom on a raised round serving stand.\n\nThen both hands withdraw.\n\nCAMERA:\n\nstart close\n→ slowly dolly backward\n→ reveal full pastry centerpiece.\n\nThis is the MAIN FOOD BEAUTY SHOT.\n\nWarm morning sunlight catches:\n\n- flaky croissant layers\n- powdered sugar\n- berries\n- apricot pieces\n- golden crust\n- polished platter\n\nTiny flour particles sparkle naturally in the sunlight.\n\nNO magical light.\n\nBruno looks toward Ari rather than the food.\n\nBruno is proud of HIM.\n\nAri quietly smiles.\n\nNo dialogue for a beat.\n\nLet music carry the reveal.\n\n==================================================\n[SHOT 7 — 0:10.0–0:12.3]\nTHE FIRST CUSTOMER\n\nThe bakery door bell gently rings.\n\nSFX:\n\nDING.\n\nAri’s smile immediately becomes nervous again.\n\nA friendly customer approaches the counter, perhaps accompanied by a child.\n\nThe customer sees the Croissant Bloom.\n\nStops.\n\nEyes widen.\n\nCustomer:\n\n<d>[English] Oh, wow.</d>\n\nThe child / companion leans closer with delight.\n\nCustomer:\n\n<d>[English] Who made this?</d>\n\nAri freezes briefly.\n\nHe looks toward Bruno.\n\nBruno does NOT answer.\n\nInstead:\n\nBruno places one warm hand lightly on Ari’s shoulder.\n\nThen subtly gestures toward Ari.\n\nAri:\n\n<d>[English] I... did.</d>\n\nBruno smiles.\n\n<d>[English] He did.</d>\n\nCustomer:\n\n<d>[English] It’s beautiful.</d>\n\nClose on Ari.\n\nHis nervous expression melts into surprised pride.\n\n==================================================\n[SHOT 8 — 0:12.3–0:15.0]\nARI’S MOMENT — COMPLETE PAYOFF\n\nTransition smoothly to a slightly wider bakery view.\n\nTwo or three more customers have noticed the pastry.\n\nKeep crowd modest.\n\nWarm happy chatter.\n\nSomeone points at the Croissant Bloom.\n\nAnother customer smiles.\n\nThe bakery becomes gently lively.\n\nAri now stands behind the pastry.\n\nHis posture is visibly more confident.\n\nBruno steps half a pace behind him.\n\nAri opens one hand toward the display like a real baker presenting his work.\n\nAri:\n\n<d>[English] It’s our Croissant Bloom.</d>\n\nBruno raises one eyebrow.\n\nAri realizes what he said.\n\nLooks toward Bruno.\n\nBruno gently corrects:\n\n<d>[English] Your Croissant Bloom.</d>\n\nAri pauses.\n\nWarm genuine smile.\n\nFINAL COMPOSITION:\n\nAri centered beside the beautiful pastry.\n\nBruno behind/next to him with proud mentor smile.\n\nCroissant Bloom in foreground.\n\nCustomers softly enjoying the bakery behind.\n\nMorning sunlight pours across the counter.\n\nAri looks down at the pastry.\n\nThen at Bruno.\n\nAri:\n\n<d>[English] Not bad for a mistake.</d>\n\nBruno laughs warmly.\n\n<d>[English] Not bad at all.</d>\n\nCamera slowly pulls backward.\n\nHold on:\n\nAri\n+\nBruno\n+\nCroissant Bloom\n+\nwarm living bakery.\n\nFADE TO BLACK.\n\nEND STORY.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nSound should feel:\n\n- intimate\n- tactile\n- appetizing\n- cozy\n- cheerful\n- emotionally warm\n\n==================================================\nARI VOICE\n==================================================\n\nYoung teenage male voice.\n\nQualities:\n\n- sincere\n- soft\n- hopeful\n- slightly nervous\n- naturally expressive\n- increasingly confident\n\nPerformance arc:\n\nopening:\nwonder.\n\nfinishing:\nfocused.\n\ncrooked petal:\nsmall perfectionist concern.\n\ncustomer:\nnervous again.\n\nending:\nearned confidence.\n\nImportant lines:\n\n<d>[English] They actually bloomed.</d>\n\nDeliver with genuine surprise.\n\n<d>[English] I... did.</d>\n\nsmall hesitation.\n\nFinal:\n\n<d>[English] Not bad for a mistake.</d>\n\ngentle humor + pride.\n\n==================================================\nBRUNO VOICE\n==================================================\n\nWarm adult male voice.\n\nQualities:\n\n- deep\n- gentle\n- broad\n- relaxed\n- encouraging\n- playful\n- mentor-like\n\nImportant performance:\n\nBruno never sounds like a lecturer.\n\nHis lessons emerge naturally through action.\n\nKey lines:\n\n<d>[English] So?</d>\n\nlight, amused.\n\n<d>[English] It has character.</d>\n\nwarm.\n\n<d>[English] Your Croissant Bloom.</d>\n\nproud and sincere.\n\nFinal laugh should feel genuine.\n\n==================================================\nCUSTOMER VOICE\n==================================================\n\nFriendly natural adult voice.\n\nVery little dialogue.\n\nPriority:\n\n<d>[English] Oh, wow.</d>\n\n<d>[English] Who made this?</d>\n\n<d>[English] It’s beautiful.</d>\n\nDo not exaggerate into advertisement-style praise.\n\nIt should sound like a real warm customer reaction.\n\n==================================================\nBAKERY SOUND DESIGN\n==================================================\n\nUse:\n\n- oven crackle\n- oven door hinge\n- baking tray glide\n- warm steam\n- plate / stand contact\n- berry placement\n- powdered sugar shaker\n- fabric/apron movement\n- wooden floor\n- soft ceramic clinks\n- bakery door bell\n- low customer chatter\n- subtle room ambience\n\nFOOD SOUND:\n\nKeep realistic and restrained.\n\nNo exaggerated ASMR crunching.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL cozy bakery score continuing Part 1.\n\nInstrumentation:\n\n- warm piano\n- acoustic guitar\n- pizzicato strings\n- clarinet\n- light upright bass\n- brushed percussion\n- subtle accordion\n- glockenspiel accents\n\nMUSIC ARC:\n\n0:00–0:02\nanticipation from oven reveal.\n\n0:02–0:04\nwarm success motif.\n\n0:04–0:06\ndelicate decorating rhythm.\n\n0:06–0:08\ntiny comic uncertainty.\n\n0:08–0:10\nbeautiful food-reveal swell.\n\n0:10–0:12\nsoft emotional customer moment.\n\n0:12–0:15\nfull warm resolution of Part 1 melody.\n\nFinal chord should feel:\n\ncomplete,\ncozy,\nsatisfying.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPart 2 camera progression:\n\n0–3 sec:\noven + food reveal.\n\n3–6 sec:\nintimate tactile decorating.\n\n6–8 sec:\ncharacter interaction.\n\n8–10 sec:\npremium centerpiece reveal.\n\n10–12 sec:\nemotional reaction coverage.\n\n12–15 sec:\nconfidence + warm bakery-wide payoff.\n\nPreferred camera techniques:\n\n- gentle dolly\n- macro pastry photography\n- top-down food detail\n- medium two-shot\n- subtle rack focus\n- close expression coverage\n- shallow depth\n- slow hero pull-back\n- natural parallax\n\nAvoid:\n\n- handheld shake\n- crash zoom\n- fast whip pans\n- dramatic Dutch angles\n- 360-degree spins\n- excessive motion blur\n\n==================================================\nFOOD CINEMATOGRAPHY PRIORITY\n==================================================\n\nThe Croissant Bloom should be extremely appealing.\n\nShow clearly:\n\n- laminated pastry layers\n- tiny crisp flakes\n- buttery golden surface\n- toasted ridges\n- soft center\n- powdered sugar\n- berry reds\n- apricot orange accents\n- radial flower arrangement\n\nLighting should create small specular highlights across flaky crust.\n\nDo NOT oversaturate.\n\nDo NOT make food unnaturally glossy.\n\nAim for:\n\nhandmade artisan bakery beauty.\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished hand-painted 2D animated short.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n\n- expressive ink contours\n- watercolor/gouache-inspired rendering\n- painterly shadows\n- warm textured backgrounds\n- stylized but believable food\n- expressive faces\n- tactile fabric\n- beautiful wooden surfaces\n- subtle depth of field\n- gentle cinematic bloom\n- hand-crafted animation feeling\n- soft natural squash-and-stretch\n\nThe final video should feel like:\n\nTHE STORYBOARD ILLUSTRATIONS HAVE COME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealistic live action\n- generic 3D CGI\n- flat vector art\n- anime\n- hyperreal food commercial\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary colors:\n\n- warm cream\n- pastry gold\n- flour white\n- berry red\n- apricot orange\n- bakery wood brown\n- muted sage\n- honey sunlight\n\nARI:\n\ncream/white outfit,\nwarm brown hair.\n\nBRUNO:\n\nwarm neutral clothing,\ndark hair/beard.\n\nFOOD:\n\nrichest golden tones in frame.\n\nLIGHTING:\n\nsoft morning sun through bakery window.\n\nUse:\n\n- warm rim light\n- amber highlights\n- subtle cool fill from window shadows\n- gentle oven glow early\n- brighter counter light for hero reveal\n\nDo not suddenly change time of day.\n\n==================================================\nMOTION RULES — ARI\n==================================================\n\nUse:\n\n- excited lean toward oven\n- careful tray handling\n- delicate berry placement\n- controlled powdered-sugar dusting\n- small worried eyebrow shift\n- slightly tense hands\n- relief after Bruno fixes petal\n- nervous customer glance\n- gradually straighter posture\n- confident final presentation\n\n==================================================\nMOTION RULES — BRUNO\n==================================================\n\nUse:\n\n- safe oven handling\n- broad warm smile\n- calm gestures\n- tiny pastry adjustment\n- supportive shoulder touch\n- subtle nods\n- allowing Ari to lead\n- warm final laugh\n\nBruno should never dominate the finished presentation.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Ari.\nONE Bruno.\n\nSame:\n\n- character faces\n- hairstyles\n- clothing\n- aprons\n- proportions\n- bakery\n- oven\n- counter\n- tools\n- morning lighting\n\nCroissant Bloom continuity:\n\noven tray\n→ finishing station\n→ serving platter\n→ customer presentation.\n\nDo NOT randomly alter:\n\n- pastry arrangement\n- fruit center\n- serving stand\n- number/shape of major pastry petals\n\nOnce powdered sugar is added:\n\nkeep it visible.\n\nOnce berries are added:\n\nkeep them visible.\n\n==================================================\nPART 1 → PART 2 CONTINUITY\n==================================================\n\nPart 1 final image:\n\nraw flower pastry idea.\n\nPart 2 beginning:\n\nthose flower pastries are now baked.\n\nDo NOT begin with normal conventional croissants.\n\nThe audience must immediately understand:\n\nTHE “MISTAKE” BECAME THE NEW DESIGN.\n\n==================================================\nEMOTIONAL PERFORMANCE PRIORITY\n==================================================\n\nThe strongest emotional chain is:\n\nAri sees successful bake\n→ feels proud.\n\nNotices tiny imperfection\n→ old insecurity briefly returns.\n\nBruno casually fixes it\n→ demonstrates perfection is not required.\n\nCustomer loves pastry\n→ Ari discovers others value his creative work.\n\nBruno lets Ari receive the credit\n→ confidence becomes internal.\n\nThe ending is NOT:\n\n“Ari made perfect pastries.”\n\nThe ending is:\n\n“Ari learned that imperfect beginnings can produce something worth sharing.”\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\n15 seconds is short.\n\nNatural performance is more important than using every line.\n\nHIGHEST PRIORITY:\n\n1.\nAri:\n<d>[English] They actually bloomed.</d>\n\n2.\nAri:\n<d>[English] That one’s crooked.</d>\n\n3.\nBruno:\n<d>[English] So?</d>\n\n4.\nBruno:\n<d>[English] It has character.</d>\n\n5.\nCustomer:\n<d>[English] Who made this?</d>\n\n6.\nAri:\n<d>[English] I... did.</d>\n\n7.\nCustomer:\n<d>[English] It’s beautiful.</d>\n\n8.\nBruno:\n<d>[English] Your Croissant Bloom.</d>\n\n9.\nAri:\n<d>[English] Not bad for a mistake.</d>\n\nIf timing becomes crowded:\n\nREMOVE secondary dialogue.\n\nDo NOT make characters speak unnaturally fast.\n\nLet expressions carry several beats.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nEverything remains wholesome and safe.\n\nNo injuries.\nNo burns.\nNo unsafe oven behavior.\nNo dangerous tools.\nNo conflict.\nNo yelling.\nNo humiliation.\nNo bullying.\n\nBruno handles hot oven equipment safely.\n\nAri only handles the tray once it is safe / using proper protection where necessary.\n\n==================================================\nNEGATIVE CONSTRAINTS\n==================================================\n\nNo storyboard sheet.\nNo borders.\nNo captions.\nNo panel numbers.\nNo timestamps.\nNo title text.\nNo production notes.\nNo reference drawings.\nNo subtitles.\nNo watermarks.\nNo logos.\n\nNo duplicate Ari.\nNo duplicate Bruno.\n\nNo identity drift.\nNo clothing changes.\nNo hairstyle changes.\nNo age changes.\nNo apron flickering.\n\nNo angry Bruno.\nNo crying Ari.\nNo food accident.\nNo burned pastries.\nNo smoke emergency.\nNo fire.\nNo broken dishes.\n\nNo magical food transformation.\nNo glowing supernatural pastry.\nNo levitating food.\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2086667268700995928",
    "title": "阿里与布鲁诺的羊角包花 · 上篇",
    "titleEn": "Ari and Bruno’s Croissant Bloom · Part 1",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《阿里与布鲁诺的羊角包花 · 上篇》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Ari and Bruno’s Croissant Bloom · Part 1. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for:\n\n“ARI & BRUNO — THE CROISSANT THAT BLOOMED”\n\nas the STRICT PRIMARY visual, character, environment, prop, lighting, scale, performance, and narrative reference.\n\nGenerate a FINISHED cinematic animated sequence based on this storyboard.\n\nDO NOT animate the storyboard sheet itself.\n\nThe finished video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character reference drawings\n- prop reference drawings\n- color swatches\n- title/header\n- page background\n- annotations\n- subtitles\n- logos\n- watermarks\n\nThe uploaded storyboard locks:\n\n1. Ari’s exact face\n2. Ari’s short messy brown hair\n3. Ari’s slim teenage proportions\n4. Ari’s cream/white baker shirt\n5. Ari’s rolled sleeves\n6. Ari’s light apron\n7. Ari’s expressive large eyes\n8. Bruno’s exact face\n9. Bruno’s curly dark hair\n10. Bruno’s short full beard\n11. Bruno’s large heavyset build\n12. Bruno’s light baker shirt and cream apron\n13. cozy artisan bakery environment\n14. warm sunrise lighting\n15. wooden worktable\n16. bowls, flour, rolling pin and baking tray\n17. original croissant dough design\n18. squashed dough design\n19. first flower-shaped dough concept\n20. Ari and Bruno’s mentor/apprentice relationship\n\nTranslate the storyboard’s 15 one-second panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT create fifteen frantic one-second cuts.\n\nStory progression:\n\nwarm bakery morning\n→ Bruno teaches\n→ Ari succeeds\n→ confidence grows\n→ small tray accident\n→ Ari believes he failed\n→ Bruno responds calmly\n→ mistake becomes a creative idea\n→ hopeful Part 1 cliffhanger.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 1 — 0:00 to 0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D animated-film style\n\nGenre:\ncozy bakery story / mentor-apprentice friendship\n\nTone:\n\n- warm\n- tactile\n- charming\n- gentle\n- lightly comedic\n- emotionally sincere\n- hopeful\n- family-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative character dialogue + bakery ambience + baking sounds + original warm score\n\nCORE STORY IDEA:\n\nAri wants to make everything perfectly.\n\nA small mistake makes him think he has failed.\n\nBruno sees something else:\n\na chance to create something new.\n\n==================================================\nPART 1 STORY LIMIT\n==================================================\n\nPART 1 ONLY.\n\nSTART:\n\nMorning begins inside the cozy bakery.\n\nBruno teaches Ari how to shape croissants.\n\nEND:\n\nBruno reshapes ONE squashed pastry into a flower.\n\nAri realizes the ruined batch might become something completely new.\n\nThe remaining pastries are only BEGINNING to be reshaped.\n\nCUT TO BLACK.\n\nDO NOT show:\n\n- fully finished croissant bloom\n- baked flower pastries\n- berries added to finished bloom\n- powdered sugar finishing\n- final cake stand\n- customer reaction\n- customers entering bakery\n- Ari presenting final pastry\n- bakery crowd\n- final success\n\nAll of those belong to PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — ARI\n==================================================\n\nMatch the uploaded storyboard EXACTLY.\n\nARI — APPRENTICE BAKER\n\nYoung teenage baker apprentice.\n\nAppearance:\n\n- warm light skin\n- youthful soft face\n- large expressive brown/dark eyes\n- short tousled medium-brown hair\n- slim build\n- narrow shoulders\n- youthful proportions\n- cream/white baker shirt\n- sleeves rolled to forearms\n- light beige/cream apron\n- simple work trousers\n- practical bakery shoes\n\nDo NOT redesign him.\n\nDo NOT change:\n\n- hairstyle\n- age\n- body proportions\n- apron\n- shirt\n- facial structure\n\nPERSONALITY:\n\n- hopeful\n- hardworking\n- eager\n- careful\n- slightly nervous\n- wants to impress Bruno\n- learns through doing\n\nPART 1 PERFORMANCE ARC:\n\n0–4 sec:\nfocused learner\n\n4–6 sec:\nsmall success and growing confidence\n\n6–8 sec:\ncareful but slightly rushed\n\n8–10 sec:\nsudden worry\n\n10–12 sec:\nexpects disappointment\n\n12–15 sec:\nsurprise → curiosity → renewed hope\n\nIMPORTANT:\n\nAri is NOT terrified.\n\nHis reaction to the mistake should be:\n\n“Oh no, I ruined our work.”\n\nNot:\n\npanic,\ntrauma,\nextreme fear,\ncrying.\n\nUse subtle emotional acting.\n\n==================================================\nCHARACTER IDENTITY LOCK — BRUNO\n==================================================\n\nMatch uploaded storyboard EXACTLY.\n\nBRUNO — HEAD BAKER\n\nAppearance:\n\n- large broad heavyset build\n- warm medium/light-olive skin\n- curly dark-brown hair\n- short full beard\n- thick eyebrows\n- kind expressive eyes\n- large friendly hands\n- light baker shirt\n- cream apron\n- sturdy practical bakery clothing\n\nPERSONALITY:\n\n- grounded\n- cheerful\n- experienced\n- patient\n- reassuring\n- playful\n- mentor/fatherly energy\n\nMovement style:\n\n- broad\n- calm\n- confident\n- fluid\n- never rushed unnecessarily\n\nPART 1 PERFORMANCE ARC:\n\n0–4 sec:\nenthusiastic teacher\n\n4–7 sec:\nquietly proud of Ari\n\n7–10 sec:\noccupied nearby\n\n10–12 sec:\nnotices mistake calmly\n\n12–15 sec:\nthinking → inspired → encouraging\n\nCRITICAL:\n\nBruno must NEVER look angry.\n\nWhen he sees the squashed pastries:\n\nno yelling,\nno disappointment,\nno aggressive gestures.\n\nHis first reaction should be:\n\nobserve\n→ think\n→ smile slightly\n→ find possibility.\n\n==================================================\nENVIRONMENT LOCK — MORNING BAKERY\n==================================================\n\nMaintain the exact bakery world from the storyboard.\n\nEnvironment:\n\n- artisan bakery kitchen\n- rustic wooden work counters\n- warm wooden shelves\n- flour jars\n- ceramic bowls\n- mixing bowls\n- rolling pins\n- baking trays\n- berries\n- butter\n- utensils\n- hanging cookware\n- old-fashioned oven\n- bread/pastry preparation surfaces\n- warm morning window\n- subtle flour particles floating in sunlight\n\nLighting:\n\nwarm sunrise entering from windows.\n\nSecondary lighting:\n\nsoft amber oven glow.\n\nColor palette:\n\n- pastry gold\n- butter cream\n- flour white\n- berry red\n- wood brown\n- honey orange\n- muted sage accents\n\nThe bakery should feel:\n\nhandmade,\nlived-in,\nwarm,\ninviting,\ndelicious.\n\n==================================================\nKEY PROP LOCK — CROISSANT DOUGH\n==================================================\n\nMaintain clear physical progression.\n\nSTAGE 1:\nflat triangular croissant dough.\n\nSTAGE 2:\nrolled crescent-shaped raw croissants.\n\nSTAGE 3:\nseveral neat croissants arranged on tray.\n\nSTAGE 4:\ntray accidentally tilts.\n\nSTAGE 5:\nsome dough pieces become softly squashed / flattened.\n\nSTAGE 6:\nBruno uses the flattened shape as material for a flower pastry.\n\nIMPORTANT:\n\nThe pastries are RAW DOUGH in Part 1.\n\nDo NOT suddenly turn them golden-baked before Part 2.\n\nDo not duplicate pastries unnaturally.\n\n==================================================\nFLOWER-SHAPED PASTRY IDEA LOCK\n==================================================\n\nThe Part 1 ending reveals only the INITIAL IDEA.\n\nBruno takes ONE squashed dough piece.\n\nHe carefully:\n\npinches one edge\n→ folds another section\n→ curves the layers\n→ forms overlapping dough petals.\n\nThe result should resemble a simple raw pastry rose / flower.\n\nIt is still:\n\n- pale dough\n- unbaked\n- soft\n- lightly floured\n\nThe transformation must happen through Bruno’s HANDS.\n\nNo magical transformation.\n\nNo glowing magic.\n\nNo instant morph.\n\nThe creativity itself is the magic.\n\n==================================================\nWORLD / ACTION LOGIC\n==================================================\n\nBruno demonstrates\n→ Ari copies.\n\nAri copies correctly\n→ croissants look good.\n\nSuccess\n→ Ari gains confidence.\n\nAri moves tray\n→ edge catches / tray tilts slightly.\n\nTray tilts\n→ several soft raw pastries slide together.\n\nSoft pastries collide\n→ shapes become flattened.\n\nAri sees damaged shapes\n→ thinks he failed.\n\nBruno notices\n→ stays calm.\n\nBruno studies flattened dough\n→ recognizes flower-like possibility.\n\nBruno reshapes one pastry\n→ Ari understands the new idea.\n\nEverything must follow readable physical cause-and-effect.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.8]\nMORNING IN THE LITTLE BAKERY\n\nOpen with a warm cinematic establishing shot.\n\nGolden sunrise streams through the bakery window.\n\nFlour dust floats softly through light beams.\n\nWooden counters are covered with:\n\n- bowls\n- butter\n- flour\n- dough\n- berries\n- rolling pins\n- baking tools\n\nAri and Bruno work together at the center table.\n\nCamera glides slowly toward them.\n\nBruno happily places a smooth sheet of dough on the floured work surface.\n\nAri stands beside him, attentive.\n\nBruno:\n\n<d>[English] Ready?</d>\n\nAri nods enthusiastically.\n\nAri:\n\n<d>[English] Ready.</d>\n\nBruno smiles.\n\nMUSIC:\n\nsoft morning bakery theme begins.\n\nUse:\n\n- acoustic guitar\n- pizzicato strings\n- piano\n- clarinet\n- subtle brushed percussion\n\nAUDIO:\n\n- quiet room tone\n- wooden counter sounds\n- soft bowl clinks\n- distant oven fire\n- flour brushing\n- morning birds outside\n\n==================================================\n[SHOT 2 — 0:01.8–0:03.6]\nBRUNO SHOWS THE FIRST STEP\n\nMove closer to the worktable.\n\nBruno cuts / presents a triangle of laminated dough.\n\nMacro close-up of his large hands.\n\nHe gently rolls from the wide end toward the point.\n\nOne smooth motion.\n\nThe layers curl into a neat croissant.\n\nBruno:\n\n<d>[English] Gentle hands.</d>\n\nHe lightly taps the finished shape.\n\n<d>[English] Let the dough do the work.</d>\n\nAri watches closely.\n\nUse tactile food detail:\n\n- soft flour\n- buttery layers\n- subtle dough elasticity\n- hand pressure\n- flour fingerprints\n\nCAMERA:\n\nmedium mentor/apprentice\n→ close dough triangle\n→ macro rolling hands\n→ finished raw croissant.\n\n==================================================\n[SHOT 3 — 0:03.6–0:05.4]\nARI TRIES\n\nBruno slides another dough triangle toward Ari.\n\nAri inhales quietly.\n\nHe places both hands carefully.\n\nHe imitates Bruno.\n\nFirst fold.\n\nSmall pause.\n\nSecond roll.\n\nFinal curl.\n\nThe croissant holds its shape.\n\nAri’s eyes widen.\n\nAri:\n\n<d>[English] I did it.</d>\n\nBruno smiles behind him.\n\nBruno:\n\n<d>[English] You did.</d>\n\nQuick reveal of the tray:\n\nseveral neat raw croissants lined beautifully.\n\nAri smiles.\n\nBruno gives him a warm thumbs-up.\n\nMusic becomes slightly brighter.\n\n==================================================\n[SHOT 4 — 0:05.4–0:07.0]\nCONFIDENCE GROWS\n\nCreate a short flowing preparation sequence.\n\nAri shapes another croissant.\n\nThen another.\n\nHis movements become slightly faster and more natural.\n\nBruno prepares something at the neighboring station.\n\nAri places the final neat piece onto the baking tray.\n\nHe proudly looks at the complete row.\n\nAri:\n\n<d>[English] Not bad.</d>\n\nFrom nearby Bruno replies:\n\n<d>[English] Not bad?</d>\n\nBruno glances toward the tray.\n\nSmiles.\n\n<d>[English] Very good.</d>\n\nAri’s smile becomes bigger.\n\nDo not linger too long.\n\n==================================================\n[SHOT 5 — 0:07.0–0:08.8]\nONE SMALL MISTAKE\n\nAri reaches for the baking tray.\n\nUse clear physical movement.\n\nHe grips both sides.\n\nBegins lifting it from the work surface.\n\nAs he turns:\n\none tray edge gently catches the corner of a folded cloth / wooden board.\n\nThe tray tips several degrees.\n\nNOT a dramatic fall.\n\nThe tray remains in Ari’s hands.\n\nThe soft croissants slide toward one side.\n\nAri:\n\n<d>[English] Whoa—!</d>\n\nHe quickly levels the tray.\n\nNothing falls onto the floor.\n\nNo dangerous action.\n\nBut several pastries have pressed softly into one another.\n\nA tiny puff of flour rises.\n\nMusic stops on a gentle comic “oops” beat.\n\nSFX:\n\n- tray scrape\n- dough sliding\n- soft flour PUFF\n- light metal wobble\n\n==================================================\n[SHOT 6 — 0:08.8–0:10.4]\nARI THINKS HE RUINED IT\n\nClose-up on the tray.\n\nSeveral once-perfect croissants are now:\n\n- flattened\n- pressed together\n- irregular\n- soft and misshapen\n\nNot disgusting.\n\nNot destroyed.\n\nJust imperfect.\n\nRack focus to Ari.\n\nHis eyes widen.\n\nShoulders drop.\n\nAri:\n\n<d>[English] Oh no...</d>\n\nHe carefully sets the tray back down.\n\nLooks at the pastries.\n\nAri, quietly:\n\n<d>[English] I ruined them.</d>\n\nHold briefly on his worried expression.\n\nNo crying.\n\nNo melodrama.\n\n==================================================\n[SHOT 7 — 0:10.4–0:12.3]\nBRUNO DOESN’T GET UPSET\n\nBruno enters the composition.\n\nAri looks toward him nervously.\n\nAri:\n\n<d>[English] Bruno... I’m sorry.</d>\n\nBruno looks at:\n\nAri\n→ tray\n→ squashed dough.\n\nHe does NOT frown angrily.\n\nHe simply leans closer.\n\nStudies one flattened piece.\n\nTilts his head.\n\nTouches his beard thoughtfully.\n\nAri waits.\n\nSilence for a small emotional beat.\n\nThen Bruno’s eyebrow rises.\n\nHis expression becomes curious.\n\nA tiny smile begins.\n\nBruno:\n\n<d>[English] Hmm...</d>\n\nAri looks confused.\n\nBruno:\n\n<d>[English] Maybe you didn’t.</d>\n\nMusic introduces a soft playful “idea” motif.\n\n==================================================\n[SHOT 8 — 0:12.3–0:15.0]\nTHE CROISSANT BEGINS TO BLOOM\n\nBruno gently picks up ONE squashed raw pastry.\n\nPlace it on the floured work surface.\n\nAri leans closer.\n\nBruno uses both hands.\n\nCLEAR ACTION:\n\nhe pinches one edge inward\n→ folds a soft dough section\n→ curls another layer around it\n→ rotates it slightly\n→ gently opens the outer layers.\n\nThe flattened pastry begins looking like a little flower / rose.\n\nAri’s worried face transforms.\n\nConcern\n→ confusion\n→ surprise\n→ hope.\n\nAri:\n\n<d>[English] A flower?</d>\n\nBruno smiles.\n\n<d>[English] Why not?</d>\n\nHe places the raw flower pastry onto the tray.\n\nCamera reveals several remaining squashed pieces beside it.\n\nBruno looks at Ari.\n\nBruno:\n\n<d>[English] Mistakes can become ideas.</d>\n\nAri looks from the little flower pastry to the rest of the tray.\n\nHis eyes brighten.\n\nAri reaches toward another flattened piece.\n\nAri:\n\n<d>[English] Then let’s make them bloom.</d>\n\nIMPORTANT:\n\nDo NOT show all pastries completed.\n\nOnly:\n- ONE clear finished raw flower\n- perhaps ONE second pastry just beginning to be reshaped.\n\nEnd on:\n\nAri and Bruno leaning over the tray together.\n\nThe raw flower pastry sits between them.\n\nWarm sunlight catches the flour.\n\nAri smiles with renewed confidence.\n\nMusic lifts into hopeful cliffhanger chord.\n\nCUT TO BLACK.\n\nEND PART 1.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio should be:\n\n- intimate\n- warm\n- tactile\n- cozy\n- emotionally clear\n- lightly playful\n\n==================================================\nARI VOICE\n==================================================\n\nYoung teenage male voice.\n\nQualities:\n\n- sincere\n- gentle\n- slightly nervous\n- curious\n- eager\n- natural\n- not overly childish\n\nPerformance progression:\n\nbeginning:\nfocused excitement\n\nsuccess:\nproud but modest\n\nmistake:\nembarrassed and worried\n\nending:\nsurprised and inspired\n\nImportant lines:\n\n<d>[English] I did it.</d>\n\nshould feel like small genuine pride.\n\n<d>[English] I ruined them.</d>\n\nsoft disappointment, not dramatic despair.\n\nFinal:\n\n<d>[English] Then let’s make them bloom.</d>\n\nshould carry renewed excitement.\n\n==================================================\nBRUNO VOICE\n==================================================\n\nWarm adult male voice.\n\nQualities:\n\n- broad\n- relaxed\n- cheerful\n- grounded\n- reassuring\n- mentor-like\n- slightly playful\n\nNever harsh.\n\nImportant dialogue:\n\n<d>[English] Gentle hands.</d>\n\ninstructional and warm.\n\n<d>[English] Maybe you didn’t.</d>\n\nspoken calmly, creating curiosity.\n\n<d>[English] Mistakes can become ideas.</d>\n\nsimple, natural, not overly inspirational or preachy.\n\n==================================================\nBAKERY SOUND DESIGN\n==================================================\n\nUse detailed but subtle sounds:\n\n- flour brush\n- soft dough pressing\n- rolling dough\n- metal bowl clinks\n- tray sliding\n- rolling pin on wood\n- apron fabric movement\n- oven crackle\n- distant kitchen ambience\n- gentle footsteps\n\nTRAY ACCIDENT:\n\nKeep sounds light.\n\nUse:\n\nsmall metal scrape\n→ dough slide\n→ flour puff.\n\nNo loud crash.\n\nNo smashing.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL cozy bakery score.\n\nInstrumentation:\n\n- warm piano\n- acoustic guitar\n- pizzicato strings\n- clarinet\n- upright bass\n- brushed percussion\n- subtle accordion texture\n- tiny glockenspiel for inspiration moment\n\nMUSIC ARC:\n\n0:00–0:02\nwarm morning introduction.\n\n0:02–0:05\ngentle learning rhythm.\n\n0:05–0:07\nsmall confidence lift.\n\n0:07–0:09\nbrief playful “oops” punctuation.\n\n0:09–0:11\nsoft emotional uncertainty.\n\n0:11–0:13\ncurious pause.\n\n0:13–0:15\nhopeful creative motif.\n\nEnd unresolved enough to invite Part 2, but emotionally positive.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nUse a tactile cinematic food-storytelling approach.\n\nStart:\n\nwide environment.\n\nThen progressively move into:\n\n- hands\n- dough\n- facial reactions\n- food texture\n\nPreferred camera:\n\n- gentle dolly\n- medium two-shot\n- overhead preparation detail\n- macro dough close-ups\n- rack focus\n- subtle side tracking\n- eye-level emotional close-ups\n- final slow push-in on flower pastry\n\nThe mistake should NOT use action-movie camera language.\n\nNo shaky camera.\n\nNo aggressive whip pan.\n\nNo dramatic Dutch angle.\n\nNo huge slow motion.\n\nKeep everything readable and intimate.\n\n==================================================\nFOOD CINEMATOGRAPHY\n==================================================\n\nMake the food visually satisfying.\n\nShow:\n\n- flour texture\n- laminated dough layers\n- soft butter sheen\n- gentle hand pressure\n- subtle elasticity\n- raw croissant curves\n- flour particles\n- flower-petal dough layers\n\nIMPORTANT:\n\nPart 1 pastries remain RAW.\n\nUse pale:\n\n- butter cream\n- flour beige\n- soft dough yellow\n\nDo NOT make them baked golden brown yet.\n\nGolden baked pastries belong to Part 2.\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished hand-painted 2D animated short.\n\nMatch the uploaded storyboard.\n\nUse:\n\n- clean expressive ink linework\n- watercolor/gouache inspired rendering\n- soft painterly shadows\n- warm textured wood\n- expressive stylized faces\n- detailed hand-painted food\n- subtle atmospheric depth\n- gentle cinematic bloom\n- warm morning light\n- tactile materials\n- tasteful squash-and-stretch\n\nThe final animation should feel like:\n\nTHE STORYBOARD ILLUSTRATIONS HAVE COME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealistic live action\n- generic 3D CGI\n- flat vector cartoon\n- anime\n- hyperreal food commercial\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary palette:\n\n- warm cream\n- butter yellow\n- pastry beige\n- flour white\n- wood brown\n- soft berry red\n- honey orange\n- muted sage\n\nLIGHTING:\n\nstrong but soft sunrise through window.\n\nUse warm rim lighting on:\n\n- Ari’s brown hair\n- Bruno’s curly hair/beard\n- hands\n- flour particles\n- dough edges\n\nOven provides subtle secondary amber glow.\n\nDuring Ari’s mistake:\n\ndo NOT suddenly darken the scene.\n\nMaintain visual warmth.\n\nDuring Bruno’s idea:\n\nlet sunlight gently highlight the newly formed flower pastry.\n\nNo magical glow required.\n\n==================================================\nMOTION RULES — ARI\n==================================================\n\nUse:\n\n- careful hand movements\n- slightly hesitant shaping\n- focused eye movements\n- small proud smile\n- faster confidence after success\n- safe tray lift\n- small startled body recoil\n- lowered shoulders after mistake\n- leaning forward with curiosity\n- renewed energy at ending\n\nNo exaggerated panic.\n\n==================================================\nMOTION RULES — BRUNO\n==================================================\n\nUse:\n\n- broad demonstration gestures\n- steady hand movements\n- relaxed posture\n- expressive eyebrows\n- thoughtful beard-touch\n- gentle dough reshaping\n- warm smile\n- subtle nods\n\nHe is an experienced baker.\n\nHis movements should feel effortless.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Ari.\nONE Bruno.\n\nSame:\n\n- faces\n- hairstyles\n- clothing\n- aprons\n- body proportions\n- bakery layout\n- morning light\n- worktable\n- utensils\n- tray\n- dough appearance\n\nRaw croissant progression must stay continuous.\n\nOnce the tray is squashed:\n\ndo NOT show perfect original croissants again.\n\nOnce Bruno creates the flower:\n\nthat flower stays on the tray.\n\nNo instant baked transformation.\n\nNo berries or powdered sugar added to the flower pastries yet unless already incidental in the environment.\n\n==================================================\nEMOTIONAL PERFORMANCE PRIORITY\n==================================================\n\nThe most important emotional transition is:\n\nARI:\n\n“I ruined them.”\n\nBRUNO:\n\ndoes NOT react negatively.\n\nHe studies the dough.\n\nThen:\n\n“Maybe you didn’t.”\n\nThat small reaction teaches Ari that mistakes do not automatically equal failure.\n\nThe final flower idea should feel earned through:\n\nobservation\n+\ncraft\n+\ncreativity.\n\nNot magic.\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\nBecause the sequence is only 15 seconds, DO NOT force every optional line.\n\nHighest-priority spoken dialogue:\n\n1.\nBruno:\n<d>[English] Gentle hands.</d>\n\n2.\nAri:\n<d>[English] I did it.</d>\n\n3.\nAri:\n<d>[English] Oh no...</d>\n\n4.\nAri:\n<d>[English] I ruined them.</d>\n\n5.\nBruno:\n<d>[English] Maybe you didn’t.</d>\n\n6.\nAri:\n<d>[English] A flower?</d>\n\n7.\nBruno:\n<d>[English] Why not?</d>\n\n8.\nAri:\n<d>[English] Then let’s make them bloom.</d>\n\nIf timing becomes tight:\n\nremove secondary dialogue rather than speeding up voices.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nKeep the entire video safe and family-friendly.\n\nNo injury.\nNo dangerous oven interaction.\nNo burns.\nNo fire accident.\nNo knives used dangerously.\nNo yelling.\nNo conflict.\nNo bullying.\nNo harsh discipline.\n\nWhen oven appears:\n\nBruno handles it safely.\n\nAri stays at safe distance unless supervised.\n\nTray accident occurs at the WORKTABLE, not near a dangerous hot oven.\n\n==================================================\nNEGATIVE CONSTRAINTS\n==================================================\n\nNo storyboard sheet.\nNo panels.\nNo timestamps.\nNo captions.\nNo production notes.\nNo reference drawings.\nNo typography.\nNo subtitle text.\nNo logo.\nNo watermark.\n\nNo duplicate Ari.\nNo duplicate Bruno.\n\nNo character identity drift.\nNo hairstyle changes.\nNo apron changes.\nNo costume changes.\nNo sudden age changes.\n\nNo angry Bruno.\nNo crying Ari.\nNo screaming.\nNo dangerous accident.\nNo food falling onto floor.\nNo burning food.\nNo fire emergency.\n\nNo baked golden flower bloom yet.\nNo customer.\nNo bakery crowd.\nNo finished presentation stand.\nNo Part 2 payoff.\n\nNo magical dough transformation.\nNo levitating pastries.\nNo glowing supernatural food.\n\nNo malformed hands.\nNo extra fingers.\nNo warped faces.\nNo flickering props.\nNo disappearing tray.\nNo inconsistent pastry count.\n\n==================================================\nFINAL 4-SECOND PRIORITY\n==================================================\n\nThe final four seconds must create the Part 2 hook.\n\nAri looks down sadly at the squashed raw croissants.\n\nBruno studies them calmly.\n\nBruno touches his beard.\n\nHis eyes brighten with an idea.\n\nBruno:\n\n<d>[English] Maybe you didn’t.</d>\n\nHe takes ONE flattened pastry.\n\nUsing his hands:\n\npinch\n→ fold\n→ curl\n→ open the layers.\n\nIt becomes a simple RAW flower-shaped pastry.\n\nAri leans closer.\n\nHis eyes widen.\n\nAri:\n\n<d>[English] A flower?</d>\n\nBruno smiles:\n\n<d>[English] Why not?</d>\n\nBruno places the flower on the tray.\n\nAri looks at the remaining flattened pastries.\n\nThen at Bruno.\n\nAri smiles.\n\n<d>[English] Then let’s make them bloom.</d>\n\nAri reaches toward the next pastry.\n\nSTOP BEFORE THE FULL TRANSFORMATION.\n\nSlow camera push toward:\n\none flower pastry\n+\nseveral squashed pastries\n+\nAri’s hopeful face\n+\nBruno’s encouraging smile.\n\nWarm sunrise illuminates the flour particles.\n\nCUT TO BLACK.\n\nEND PART 1.\n\nThe viewer should immediately want to see:\n\n# “What will their mistake become?”",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086667268700995928",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086667268700995928",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-10T04:13:39.918Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086667268700995928.mp4",
    "posterUrl": "/posters/x/2086667268700995928.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "有 Prompt",
      "完整 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "food",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard image to video using @Hailuo_AI's MiniMax H3 on @higgsfield_ai @higgsfield \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for:\n\n“ARI & BRUNO — THE CROISSANT THAT BLOOMED”\n\nas the STRICT PRIMARY visual, character, environment, prop, lighting, scale, performance, and narrative reference.\n\nGenerate a FINISHED cinematic animated sequence based on this storyboard.\n\nDO NOT animate the storyboard sheet itself.\n\nThe finished video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character reference drawings\n- prop reference drawings\n- color swatches\n- title/header\n- page background\n- annotations\n- subtitles\n- logos\n- watermarks\n\nThe uploaded storyboard locks:\n\n1. Ari’s exact face\n2. Ari’s short messy brown hair\n3. Ari’s slim teenage proportions\n4. Ari’s cream/white baker shirt\n5. Ari’s rolled sleeves\n6. Ari’s light apron\n7. Ari’s expressive large eyes\n8. Bruno’s exact face\n9. Bruno’s curly dark hair\n10. Bruno’s short full beard\n11. Bruno’s large heavyset build\n12. Bruno’s light baker shirt and cream apron\n13. cozy artisan bakery environment\n14. warm sunrise lighting\n15. wooden worktable\n16. bowls, flour, rolling pin and baking tray\n17. original croissant dough design\n18. squashed dough design\n19. first flower-shaped dough concept\n20. Ari and Bruno’s mentor/apprentice relationship\n\nTranslate the storyboard’s 15 one-second panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT create fifteen frantic one-second cuts.\n\nStory progression:\n\nwarm bakery morning\n→ Bruno teaches\n→ Ari succeeds\n→ confidence grows\n→ small tray accident\n→ Ari believes he failed\n→ Bruno responds calmly\n→ mistake becomes a creative idea\n→ hopeful Part 1 cliffhanger.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 1 — 0:00 to 0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D animated-film style\n\nGenre:\ncozy bakery story / mentor-apprentice friendship\n\nTone:\n\n- warm\n- tactile\n- charming\n- gentle\n- lightly comedic\n- emotionally sincere\n- hopeful\n- family-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative character dialogue + bakery ambience + baking sounds + original warm score\n\nCORE STORY IDEA:\n\nAri wants to make everything perfectly.\n\nA small mistake makes him think he has failed.\n\nBruno sees something else:\n\na chance to create something new.\n\n==================================================\nPART 1 STORY LIMIT\n==================================================\n\nPART 1 ONLY.\n\nSTART:\n\nMorning begins inside the cozy bakery.\n\nBruno teaches Ari how to shape croissants.\n\nEND:\n\nBruno reshapes ONE squashed pastry into a flower.\n\nAri realizes the ruined batch might become something completely new.\n\nThe remaining pastries are only BEGINNING to be reshaped.\n\nCUT TO BLACK.\n\nDO NOT show:\n\n- fully finished croissant bloom\n- baked flower pastries\n- berries added to finished bloom\n- powdered sugar finishing\n- final cake stand\n- customer reaction\n- customers entering bakery\n- Ari presenting final pastry\n- bakery crowd\n- final success\n\nAll of those belong to PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — ARI\n==================================================\n\nMatch the uploaded storyboard EXACTLY.\n\nARI — APPRENTICE BAKER\n\nYoung teenage baker apprentice.\n\nAppearance:\n\n- warm light skin\n- youthful soft face\n- large expressive brown/dark eyes\n- short tousled medium-brown hair\n- slim build\n- narrow shoulders\n- youthful proportions\n- cream/white baker shirt\n- sleeves rolled to forearms\n- light beige/cream apron\n- simple work trousers\n- practical bakery shoes\n\nDo NOT redesign him.\n\nDo NOT change:\n\n- hairstyle\n- age\n- body proportions\n- apron\n- shirt\n- facial structure\n\nPERSONALITY:\n\n- hopeful\n- hardworking\n- eager\n- careful\n- slightly nervous\n- wants to impress Bruno\n- learns through doing\n\nPART 1 PERFORMANCE ARC:\n\n0–4 sec:\nfocused learner\n\n4–6 sec:\nsmall success and growing confidence\n\n6–8 sec:\ncareful but slightly rushed\n\n8–10 sec:\nsudden worry\n\n10–12 sec:\nexpects disappointment\n\n12–15 sec:\nsurprise → curiosity → renewed hope\n\nIMPORTANT:\n\nAri is NOT terrified.\n\nHis reaction to the mistake should be:\n\n“Oh no, I ruined our work.”\n\nNot:\n\npanic,\ntrauma,\nextreme fear,\ncrying.\n\nUse subtle emotional acting.\n\n==================================================\nCHARACTER IDENTITY LOCK — BRUNO\n==================================================\n\nMatch uploaded storyboard EXACTLY.\n\nBRUNO — HEAD BAKER\n\nAppearance:\n\n- large broad heavyset build\n- warm medium/light-olive skin\n- curly dark-brown hair\n- short full beard\n- thick eyebrows\n- kind expressive eyes\n- large friendly hands\n- light baker shirt\n- cream apron\n- sturdy practical bakery clothing\n\nPERSONALITY:\n\n- grounded\n- cheerful\n- experienced\n- patient\n- reassuring\n- playful\n- mentor/fatherly energy\n\nMovement style:\n\n- broad\n- calm\n- confident\n- fluid\n- never rushed unnecessarily\n\nPART 1 PERFORMANCE ARC:\n\n0–4 sec:\nenthusiastic teacher\n\n4–7 sec:\nquietly proud of Ari\n\n7–10 sec:\noccupied nearby\n\n10–12 sec:\nnotices mistake calmly\n\n12–15 sec:\nthinking → inspired → encouraging\n\nCRITICAL:\n\nBruno must NEVER look angry.\n\nWhen he sees the squashed pastries:\n\nno yelling,\nno disappointment,\nno aggressive gestures.\n\nHis first reaction should be:\n\nobserve\n→ think\n→ smile slightly\n→ find possibility.\n\n==================================================\nENVIRONMENT LOCK — MORNING BAKERY\n==================================================\n\nMaintain the exact bakery world from the storyboard.\n\nEnvironment:\n\n- artisan bakery kitchen\n- rustic wooden work counters\n- warm wooden shelves\n- flour jars\n- ceramic bowls\n- mixing bowls\n- rolling pins\n- baking trays\n- berries\n- butter\n- utensils\n- hanging cookware\n- old-fashioned oven\n- bread/pastry preparation surfaces\n- warm morning window\n- subtle flour particles floating in sunlight\n\nLighting:\n\nwarm sunrise entering from windows.\n\nSecondary lighting:\n\nsoft amber oven glow.\n\nColor palette:\n\n- pastry gold\n- butter cream\n- flour white\n- berry red\n- wood brown\n- honey orange\n- muted sage accents\n\nThe bakery should feel:\n\nhandmade,\nlived-in,\nwarm,\ninviting,\ndelicious.\n\n==================================================\nKEY PROP LOCK — CROISSANT DOUGH\n==================================================\n\nMaintain clear physical progression.\n\nSTAGE 1:\nflat triangular croissant dough.\n\nSTAGE 2:\nrolled crescent-shaped raw croissants.\n\nSTAGE 3:\nseveral neat croissants arranged on tray.\n\nSTAGE 4:\ntray accidentally tilts.\n\nSTAGE 5:\nsome dough pieces become softly squashed / flattened.\n\nSTAGE 6:\nBruno uses the flattened shape as material for a flower pastry.\n\nIMPORTANT:\n\nThe pastries are RAW DOUGH in Part 1.\n\nDo NOT suddenly turn them golden-baked before Part 2.\n\nDo not duplicate pastries unnaturally.\n\n==================================================\nFLOWER-SHAPED PASTRY IDEA LOCK\n==================================================\n\nThe Part 1 ending reveals only the INITIAL IDEA.\n\nBruno takes ONE squashed dough piece.\n\nHe carefully:\n\npinches one edge\n→ folds another section\n→ curves the layers\n→ forms overlapping dough petals.\n\nThe result should resemble a simple raw pastry rose / flower.\n\nIt is still:\n\n- pale dough\n- unbaked\n- soft\n- lightly floured\n\nThe transformation must happen through Bruno’s HANDS.\n\nNo magical transformation.\n\nNo glowing magic.\n\nNo instant morph.\n\nThe creativity itself is the magic.\n\n==================================================\nWORLD / ACTION LOGIC\n==================================================\n\nBruno demonstrates\n→ Ari copies.\n\nAri copies correctly\n→ croissants look good.\n\nSuccess\n→ Ari gains confidence.\n\nAri moves tray\n→ edge catches / tray tilts slightly.\n\nTray tilts\n→ several soft raw pastries slide together.\n\nSoft pastries collide\n→ shapes become flattened.\n\nAri sees damaged shapes\n→ thinks he failed.\n\nBruno notices\n→ stays calm.\n\nBruno studies flattened dough\n→ recognizes flower-like possibility.\n\nBruno reshapes one pastry\n→ Ari understands the new idea.\n\nEverything must follow readable physical cause-and-effect.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.8]\nMORNING IN THE LITTLE BAKERY\n\nOpen with a warm cinematic establishing shot.\n\nGolden sunrise streams through the bakery window.\n\nFlour dust floats softly through light beams.\n\nWooden counters are covered with:\n\n- bowls\n- butter\n- flour\n- dough\n- berries\n- rolling pins\n- baking tools\n\nAri and Bruno work together at the center table.\n\nCamera glides slowly toward them.\n\nBruno happily places a smooth sheet of dough on the floured work surface.\n\nAri stands beside him, attentive.\n\nBruno:\n\n<d>[English] Ready?</d>\n\nAri nods enthusiastically.\n\nAri:\n\n<d>[English] Ready.</d>\n\nBruno smiles.\n\nMUSIC:\n\nsoft morning bakery theme begins.\n\nUse:\n\n- acoustic guitar\n- pizzicato strings\n- piano\n- clarinet\n- subtle brushed percussion\n\nAUDIO:\n\n- quiet room tone\n- wooden counter sounds\n- soft bowl clinks\n- distant oven fire\n- flour brushing\n- morning birds outside\n\n==================================================\n[SHOT 2 — 0:01.8–0:03.6]\nBRUNO SHOWS THE FIRST STEP\n\nMove closer to the worktable.\n\nBruno cuts / presents a triangle of laminated dough.\n\nMacro close-up of his large hands.\n\nHe gently rolls from the wide end toward the point.\n\nOne smooth motion.\n\nThe layers curl into a neat croissant.\n\nBruno:\n\n<d>[English] Gentle hands.</d>\n\nHe lightly taps the finished shape.\n\n<d>[English] Let the dough do the work.</d>\n\nAri watches closely.\n\nUse tactile food detail:\n\n- soft flour\n- buttery layers\n- subtle dough elasticity\n- hand pressure\n- flour fingerprints\n\nCAMERA:\n\nmedium mentor/apprentice\n→ close dough triangle\n→ macro rolling hands\n→ finished raw croissant.\n\n==================================================\n[SHOT 3 — 0:03.6–0:05.4]\nARI TRIES\n\nBruno slides another dough triangle toward Ari.\n\nAri inhales quietly.\n\nHe places both hands carefully.\n\nHe imitates Bruno.\n\nFirst fold.\n\nSmall pause.\n\nSecond roll.\n\nFinal curl.\n\nThe croissant holds its shape.\n\nAri’s eyes widen.\n\nAri:\n\n<d>[English] I did it.</d>\n\nBruno smiles behind him.\n\nBruno:\n\n<d>[English] You did.</d>\n\nQuick reveal of the tray:\n\nseveral neat raw croissants lined beautifully.\n\nAri smiles.\n\nBruno gives him a warm thumbs-up.\n\nMusic becomes slightly brighter.\n\n==================================================\n[SHOT 4 — 0:05.4–0:07.0]\nCONFIDENCE GROWS\n\nCreate a short flowing preparation sequence.\n\nAri shapes another croissant.\n\nThen another.\n\nHis movements become slightly faster and more natural.\n\nBruno prepares something at the neighboring station.\n\nAri places the final neat piece onto the baking tray.\n\nHe proudly looks at the complete row.\n\nAri:\n\n<d>[English] Not bad.</d>\n\nFrom nearby Bruno replies:\n\n<d>[English] Not bad?</d>\n\nBruno glances toward the tray.\n\nSmiles.\n\n<d>[English] Very good.</d>\n\nAri’s smile becomes bigger.\n\nDo not linger too long.\n\n==================================================\n[SHOT 5 — 0:07.0–0:08.8]\nONE SMALL MISTAKE\n\nAri reaches for the baking tray.\n\nUse clear physical movement.\n\nHe grips both sides.\n\nBegins lifting it from the work surface.\n\nAs he turns:\n\none tray edge gently catches the corner of a folded cloth / wooden board.\n\nThe tray tips several degrees.\n\nNOT a dramatic fall.\n\nThe tray remains in Ari’s hands.\n\nThe soft croissants slide toward one side.\n\nAri:\n\n<d>[English] Whoa—!</d>\n\nHe quickly levels the tray.\n\nNothing falls onto the floor.\n\nNo dangerous action.\n\nBut several pastries have pressed softly into one another.\n\nA tiny puff of flour rises.\n\nMusic stops on a gentle comic “oops” beat.\n\nSFX:\n\n- tray scrape\n- dough sliding\n- soft flour PUFF\n- light metal wobble\n\n==================================================\n[SHOT 6 — 0:08.8–0:10.4]\nARI THINKS HE RUINED IT\n\nClose-up on the tray.\n\nSeveral once-perfect croissants are now:\n\n- flattened\n- pressed together\n- irregular\n- soft and misshapen\n\nNot disgusting.\n\nNot destroyed.\n\nJust imperfect.\n\nRack focus to Ari.\n\nHis eyes widen.\n\nShoulders drop.\n\nAri:\n\n<d>[English] Oh no...</d>\n\nHe carefully sets the tray back down.\n\nLooks at the pastries.\n\nAri, quietly:\n\n<d>[English] I ruined them.</d>\n\nHold briefly on his worried expression.\n\nNo crying.\n\nNo melodrama.\n\n==================================================\n[SHOT 7 — 0:10.4–0:12.3]\nBRUNO DOESN’T GET UPSET\n\nBruno enters the composition.\n\nAri looks toward him nervously.\n\nAri:\n\n<d>[English] Bruno... I’m sorry.</d>\n\nBruno looks at:\n\nAri\n→ tray\n→ squashed dough.\n\nHe does NOT frown angrily.\n\nHe simply leans closer.\n\nStudies one flattened piece.\n\nTilts his head.\n\nTouches his beard thoughtfully.\n\nAri waits.\n\nSilence for a small emotional beat.\n\nThen Bruno’s eyebrow rises.\n\nHis expression becomes curious.\n\nA tiny smile begins.\n\nBruno:\n\n<d>[English] Hmm...</d>\n\nAri looks confused.\n\nBruno:\n\n<d>[English] Maybe you didn’t.</d>\n\nMusic introduces a soft playful “idea” motif.\n\n==================================================\n[SHOT 8 — 0:12.3–0:15.0]\nTHE CROISSANT BEGINS TO BLOOM\n\nBruno gently picks up ONE squashed raw pastry.\n\nPlace it on the floured work surface.\n\nAri leans closer.\n\nBruno uses both hands.\n\nCLEAR ACTION:\n\nhe pinches one edge inward\n→ folds a soft dough section\n→ curls another layer around it\n→ rotates it slightly\n→ gently opens the outer layers.\n\nThe flattened pastry begins looking like a little flower / rose.\n\nAri’s worried face transforms.\n\nConcern\n→ confusion\n→ surprise\n→ hope.\n\nAri:\n\n<d>[English] A flower?</d>\n\nBruno smiles.\n\n<d>[English] Why not?</d>\n\nHe places the raw flower pastry onto the tray.\n\nCamera reveals several remaining squashed pieces beside it.\n\nBruno looks at Ari.\n\nBruno:\n\n<d>[English] Mistakes can become ideas.</d>\n\nAri looks from the little flower pastry to the rest of the tray.\n\nHis eyes brighten.\n\nAri reaches toward another flattened piece.\n\nAri:\n\n<d>[English] Then let’s make them bloom.</d>\n\nIMPORTANT:\n\nDo NOT show all pastries completed.\n\nOnly:\n- ONE clear finished raw flower\n- perhaps ONE second pastry just beginning to be reshaped.\n\nEnd on:\n\nAri and Bruno leaning over the tray together.\n\nThe raw flower pastry sits between them.\n\nWarm sunlight catches the flour.\n\nAri smiles with renewed confidence.\n\nMusic lifts into hopeful cliffhanger chord.\n\nCUT TO BLACK.\n\nEND PART 1.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio should be:\n\n- intimate\n- warm\n- tactile\n- cozy\n- emotionally clear\n- lightly playful\n\n==================================================\nARI VOICE\n==================================================\n\nYoung teenage male voice.\n\nQualities:\n\n- sincere\n- gentle\n- slightly nervous\n- curious\n- eager\n- natural\n- not overly childish\n\nPerformance progression:\n\nbeginning:\nfocused excitement\n\nsuccess:\nproud but modest\n\nmistake:\nembarrassed and worried\n\nending:\nsurprised and inspired\n\nImportant lines:\n\n<d>[English] I did it.</d>\n\nshould feel like small genuine pride.\n\n<d>[English] I ruined them.</d>\n\nsoft disappointment, not dramatic despair.\n\nFinal:\n\n<d>[English] Then let’s make them bloom.</d>\n\nshould carry renewed excitement.\n\n==================================================\nBRUNO VOICE\n==================================================\n\nWarm adult male voice.\n\nQualities:\n\n- broad\n- relaxed\n- cheerful\n- grounded\n- reassuring\n- mentor-like\n- slightly playful\n\nNever harsh.\n\nImportant dialogue:\n\n<d>[English] Gentle hands.</d>\n\ninstructional and warm.\n\n<d>[English] Maybe you didn’t.</d>\n\nspoken calmly, creating curiosity.\n\n<d>[English] Mistakes can become ideas.</d>\n\nsimple, natural, not overly inspirational or preachy.\n\n==================================================\nBAKERY SOUND DESIGN\n==================================================\n\nUse detailed but subtle sounds:\n\n- flour brush\n- soft dough pressing\n- rolling dough\n- metal bowl clinks\n- tray sliding\n- rolling pin on wood\n- apron fabric movement\n- oven crackle\n- distant kitchen ambience\n- gentle footsteps\n\nTRAY ACCIDENT:\n\nKeep sounds light.\n\nUse:\n\nsmall metal scrape\n→ dough slide\n→ flour puff.\n\nNo loud crash.\n\nNo smashing.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL cozy bakery score.\n\nInstrumentation:\n\n- warm piano\n- acoustic guitar\n- pizzicato strings\n- clarinet\n- upright bass\n- brushed percussion\n- subtle accordion texture\n- tiny glockenspiel for inspiration moment\n\nMUSIC ARC:\n\n0:00–0:02\nwarm morning introduction.\n\n0:02–0:05\ngentle learning rhythm.\n\n0:05–0:07\nsmall confidence lift.\n\n0:07–0:09\nbrief playful “oops” punctuation.\n\n0:09–0:11\nsoft emotional uncertainty.\n\n0:11–0:13\ncurious pause.\n\n0:13–0:15\nhopeful creative motif.\n\nEnd unresolved enough to invite Part 2, but emotionally positive.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nUse a tactile cinematic food-storytelling approach.\n\nStart:\n\nwide environment.\n\nThen progressively move into:\n\n- hands\n- dough\n- facial reactions\n- food texture\n\nPreferred camera:\n\n- gentle dolly\n- medium two-shot\n- overhead preparation detail\n- macro dough close-ups\n- rack focus\n- subtle side tracking\n- eye-level emotional close-ups\n- final slow push-in on flower pastry\n\nThe mistake should NOT use action-movie camera language.\n\nNo shaky camera.\n\nNo aggressive whip pan.\n\nNo dramatic Dutch angle.\n\nNo huge slow motion.\n\nKeep everything readable and intimate.\n\n==================================================\nFOOD CINEMATOGRAPHY\n==================================================\n\nMake the food visually satisfying.\n\nShow:\n\n- flour texture\n- laminated dough layers\n- soft butter sheen\n- gentle hand pressure\n- subtle elasticity\n- raw croissant curves\n- flour particles\n- flower-petal dough layers\n\nIMPORTANT:\n\nPart 1 pastries remain RAW.\n\nUse pale:\n\n- butter cream\n- flour beige\n- soft dough yellow\n\nDo NOT make them baked golden brown yet.\n\nGolden baked pastries belong to Part 2.\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished hand-painted 2D animated short.\n\nMatch the uploaded storyboard.\n\nUse:\n\n- clean expressive ink linework\n- watercolor/gouache inspired rendering\n- soft painterly shadows\n- warm textured wood\n- expressive stylized faces\n- detailed hand-painted food\n- subtle atmospheric depth\n- gentle cinematic bloom\n- warm morning light\n- tactile materials\n- tasteful squash-and-stretch\n\nThe final animation should feel like:\n\nTHE STORYBOARD ILLUSTRATIONS HAVE COME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealistic live action\n- generic 3D CGI\n- flat vector cartoon\n- anime\n- hyperreal food commercial\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary palette:\n\n- warm cream\n- butter yellow\n- pastry beige\n- flour white\n- wood brown\n- soft berry red\n- honey orange\n- muted sage\n\nLIGHTING:\n\nstrong but soft sunrise through window.\n\nUse warm rim lighting on:\n\n- Ari’s brown hair\n- Bruno’s curly hair/beard\n- hands\n- flour particles\n- dough edges\n\nOven provides subtle secondary amber glow.\n\nDuring Ari’s mistake:\n\ndo NOT suddenly darken the scene.\n\nMaintain visual warmth.\n\nDuring Bruno’s idea:\n\nlet sunlight gently highlight the newly formed flower pastry.\n\nNo magical glow required.\n\n==================================================\nMOTION RULES — ARI\n==================================================\n\nUse:\n\n- careful hand movements\n- slightly hesitant shaping\n- focused eye movements\n- small proud smile\n- faster confidence after success\n- safe tray lift\n- small startled body recoil\n- lowered shoulders after mistake\n- leaning forward with curiosity\n- renewed energy at ending\n\nNo exaggerated panic.\n\n==================================================\nMOTION RULES — BRUNO\n==================================================\n\nUse:\n\n- broad demonstration gestures\n- steady hand movements\n- relaxed posture\n- expressive eyebrows\n- thoughtful beard-touch\n- gentle dough reshaping\n- warm smile\n- subtle nods\n\nHe is an experienced baker.\n\nHis movements should feel effortless.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Ari.\nONE Bruno.\n\nSame:\n\n- faces\n- hairstyles\n- clothing\n- aprons\n- body proportions\n- bakery layout\n- morning light\n- worktable\n- utensils\n- tray\n- dough appearance\n\nRaw croissant progression must stay continuous.\n\nOnce the tray is squashed:\n\ndo NOT show perfect original croissants again.\n\nOnce Bruno creates the flower:\n\nthat flower stays on the tray.\n\nNo instant baked transformation.\n\nNo berries or powdered sugar added to the flower pastries yet unless already incidental in the environment.\n\n==================================================\nEMOTIONAL PERFORMANCE PRIORITY\n==================================================\n\nThe most important emotional transition is:\n\nARI:\n\n“I ruined them.”\n\nBRUNO:\n\ndoes NOT react negatively.\n\nHe studies the dough.\n\nThen:\n\n“Maybe you didn’t.”\n\nThat small reaction teaches Ari that mistakes do not automatically equal failure.\n\nThe final flower idea should feel earned through:\n\nobservation\n+\ncraft\n+\ncreativity.\n\nNot magic.\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\nBecause the sequence is only 15 seconds, DO NOT force every optional line.\n\nHighest-priority spoken dialogue:\n\n1.\nBruno:\n<d>[English] Gentle hands.</d>\n\n2.\nAri:\n<d>[English] I did it.</d>\n\n3.\nAri:\n<d>[English] Oh no...</d>\n\n4.\nAri:\n<d>[English] I ruined them.</d>\n\n5.\nBruno:\n<d>[English] Maybe you didn’t.</d>\n\n6.\nAri:\n<d>[English] A flower?</d>\n\n7.\nBruno:\n<d>[English] Why not?</d>\n\n8.\nAri:\n<d>[English] Then let’s make them bloom.</d>\n\nIf timing becomes tight:\n\nremove secondary dialogue rather than speeding up voices.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nKeep the entire video safe and family-friendly.\n\nNo injury.\nNo dangerous oven interaction.\nNo burns.\nNo fire accident.\nNo knives used dangerously.\nNo yelling.\nNo conflict.\nNo bullying.\nNo harsh discipline.\n\nWhen oven appears:\n\nBruno handles it safely.\n\nAri stays at safe distance unless supervised.\n\nTray accident occurs at the WORKTABLE, not near a dangerous hot oven.\n\n==================================================\nNEGATIVE CONSTRAINTS\n==================================================\n\nNo storyboard sheet.\nNo panels.\nNo timestamps.\nNo captions.\nNo production notes.\nNo reference drawings.\nNo typography.\nNo subtitle text.\nNo logo.\nNo watermark.\n\nNo duplicate Ari.\nNo duplicate Bruno.\n\nNo character identity drift.\nNo hairstyle changes.\nNo apron changes.\nNo costume changes.\nNo sudden age changes.\n\nNo angry Bruno.\nNo crying Ari.\nNo screaming.\nNo dangerous accident.\nNo food falling onto floor.\nNo burning food.\nNo fire emergency.\n\nNo baked golden flower bloom yet.\nNo customer.\nNo bakery crowd.\nNo finished presentation stand.\nNo Part 2 payoff.\n\nNo magical dough transformation.\nNo levitating pastries.\nNo glowing supernatural food.\n\nNo malformed hands.\nNo extra fingers.\nNo warped faces.\nNo flickering props.\nNo disappearing tray.\nNo inconsistent pastry count.\n\n==================================================\nFINAL 4-SECOND PRIORITY\n==================================================\n\nThe final four seconds must create the Part 2 hook.\n\nAri looks down sadly at the squashed raw croissants.\n\nBruno studies them calmly.\n\nBruno touches his beard.\n\nHis eyes brighten with an idea.\n\nBruno:\n\n<d>[English] Maybe you didn’t.</d>\n\nHe takes ONE flattened pastry.\n\nUsing his hands:\n\npinch\n→ fold\n→ curl\n→ open the layers.\n\nIt becomes a simple RAW flower-shaped pastry.\n\nAri leans closer.\n\nHis eyes widen.\n\nAri:\n\n<d>[English] A flower?</d>\n\nBruno smiles:\n\n<d>[English] Why not?</d>\n\nBruno places the flower on the tray.\n\nAri looks at the remaining flattened pastries.\n\nThen at Bruno.\n\nAri smiles.\n\n<d>[English] Then let’s make them bloom.</d>\n\nAri reaches toward the next pastry.\n\nSTOP BEFORE THE FULL TRANSFORMATION.\n\nSlow camera push toward:\n\none flower pastry\n+\nseveral squashed pastries\n+\nAri’s hopeful face\n+\nBruno’s encouraging smile.\n\nWarm sunrise illuminates the flour particles.\n\nCUT TO BLACK.\n\nEND PART 1.\n\nThe viewer should immediately want to see:\n\n# “What will their mistake become?”",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2086426551114215889",
    "title": "绅士猫与小老鼠的惊喜 · 下篇",
    "titleEn": "The Gentleman Cat and Little Mouse Surprise · Part 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《绅士猫与小老鼠的惊喜 · 下篇》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Gentleman Cat and Little Mouse Surprise · Part 2. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "\"[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for:\n\n“THE GENTLEMAN CAT AND THE LITTLE MOUSE’S BIG SURPRISE”\n\nas the STRICT PRIMARY visual, character, environment, prop, lighting, performance, scale, and narrative reference.\n\nThis video continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n\n- Theo still sitting alone on the green City Park bench\n- Theo’s golden harmonica still in his possession\n- Theo believing he may not perform at the Sunset Music Festival\n- Millie secretly gathering the blue bird, yellow bird, squirrel, pigeons, and friendly violinist\n- the helpers quietly moving into position\n- Theo remaining completely unaware\n- Millie standing on the CITY PARK lamp-post base like a tiny conductor\n- Millie raising one paw\n- Millie whispering:\n  <d>[English] Ready?</d>\n\nPART 2 begins at that EXACT moment.\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT animate or reproduce the storyboard sheet itself.\n\nThe finished video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- reference drawings\n- color swatches\n- title/header\n- page number\n- storyboard paper texture\n- labels\n- subtitles\n\nThe uploaded storyboard locks:\n\n1. Theo’s exact black-cat identity\n2. Theo’s golden-yellow eyes\n3. Theo’s elegant black fedora\n4. fedora feather detail\n5. charcoal gentleman suit\n6. cream shirt\n7. bright pink tie\n8. matching pink pocket square\n9. gold pocket chain\n10. black-and-cream shoes\n11. Theo’s lean gentleman proportions\n12. Millie’s exact mouse identity\n13. Millie’s beige/taupe fur\n14. Millie’s large rounded ears\n15. pink polka-dot bow\n16. matching pink polka-dot dress\n17. pink shoes\n18. Theo/Millie size contrast\n19. golden harmonica\n20. green City Park bench\n21. CITY PARK lamp post\n22. festival poster\n23. birds, squirrel and pigeons\n24. friendly violinist\n25. handmade surprise decoration\n26. golden-hour park lighting\n27. pastel city skyline\n28. Part 2 harmonica-performance payoff\n29. Theo inviting Millie to dance\n30. final warm bench callback\n\nTranslate the storyboard’s fifteen 1-second panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT make 15 frantic one-second cuts.\n\nUse this progression:\n\nMillie gives signal\n→ helpers reveal surprise\n→ Theo realizes it is for him\n→ Millie presents harmonica\n→ Theo hesitates\n→ Millie reassures him\n→ Theo chooses to perform\n→ first note\n→ confidence grows\n→ park begins listening\n→ Theo invites Millie\n→ tiny joyful dance\n→ emotional bench ending.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 2 — 0:15 to 0:30\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D storybook animation\n\nGenre:\nheartwarming musical friendship short\n\nTone:\n\n- charming\n- elegant\n- funny\n- sincere\n- uplifting\n- warm\n- emotionally satisfying\n- family-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + harmonica + violin + park ambience + cinematic music\n\nCORE PAYOFF:\n\nTheo does not suddenly become fearless.\n\nMillie creates a place where he feels safe enough to TRY.\n\nTheo plays one note.\n\nThen another.\n\nAnd slowly discovers that people actually want to hear him.\n\n==================================================\nPART 2 START / END\n==================================================\n\nSTART:\n\nMillie lowers her conductor paw and launches her surprise.\n\nTheo is still unaware.\n\nEND:\n\nTheo and Millie are sitting together on the same green bench at sunset.\n\nTheo’s harmonica is now proudly visible instead of hidden.\n\nTheo tips his fedora toward Millie with a genuine smile.\n\nMillie smiles back.\n\nThe story must feel COMPLETE.\n\nNo cliffhanger.\n\n==================================================\nCHARACTER IDENTITY LOCK — THEO\n==================================================\n\nMatch PART 1 and the uploaded PART 2 storyboard EXACTLY.\n\nTHEO — THE GENTLEMAN CAT\n\nAppearance:\n\n- sleek charcoal-black / deep black fur\n- soft gray muzzle\n- small black feline nose\n- clear whiskers\n- triangular cat ears\n- expressive golden-yellow eyes\n- long black tail\n- tall lean anthropomorphic proportions\n\nCLOTHING:\n\n- tailored charcoal suit jacket\n- matching dark trousers\n- cream dress shirt\n- bright pink tie\n- matching pink pocket square\n- black fedora\n- small feather tucked into hat\n- subtle gold pocket chain\n- polished black-and-cream shoes\n\nNever change his clothes.\n\nNever remove his fedora permanently.\n\nPERSONALITY:\n\n- refined\n- charming\n- dryly witty\n- musically talented\n- emotionally gentle\n- outwardly composed\n- secretly nervous\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\nconfused by sudden activity\n\n3–5 sec:\nsurprised and emotionally touched\n\n5–7 sec:\nfear returns when he sees harmonica\n\n7–9 sec:\ndecides to try\n\n9–12 sec:\nnervous musician → genuine confidence\n\n12–14 sec:\nopen happiness\n\n14–15 sec:\npeaceful gratitude\n\nCRITICAL:\n\nTheo’s body language must tell the story.\n\nAt first:\n\ntight shoulders,\nears slightly lowered,\nharmonica held carefully.\n\nAs he plays:\n\nshoulders loosen,\nears rise,\ntail begins moving with rhythm,\neyes brighten,\nsmile becomes genuine.\n\n==================================================\nCHARACTER IDENTITY LOCK — MILLIE\n==================================================\n\nMatch PART 1 EXACTLY.\n\nMILLIE — THE LITTLE MOUSE\n\nAppearance:\n\n- tiny warm beige/taupe mouse\n- very large circular ears\n- soft pink inner ears\n- large expressive dark eyes\n- long eyelashes\n- tiny black nose\n- delicate whiskers\n- thin mouse tail\n- petite proportions\n\nCLOTHING:\n\n- large pink bow with white polka dots\n- pink sleeveless polka-dot dress\n- softly flared skirt\n- pink Mary Jane shoes\n\nPersonality:\n\n- cheerful\n- observant\n- supportive\n- emotionally intelligent\n- playful\n- determined\n- affectionate\n- brave on behalf of her friend\n\nPART 2 PERFORMANCE:\n\n0–3 sec:\nconfident little organizer\n\n3–5 sec:\nexcited surprise reveal\n\n5–7 sec:\ngentle reassurance\n\n7–11 sec:\nquiet supportive witness\n\n11–14 sec:\npure joy\n\n14–15 sec:\nwarmly satisfied\n\nIMPORTANT:\n\nMillie NEVER pressures Theo.\n\nHer encouragement communicates:\n\n“You don’t have to be perfect. We’re staying either way.”\n\n==================================================\nGOLDEN HARMONICA LOCK\n==================================================\n\nMatch the storyboard EXACTLY.\n\nTheo’s golden harmonica:\n\n- small vintage rectangular harmonica\n- warm polished brass\n- fine engraved lines\n- slightly worn sentimental edges\n- metallic golden reflections\n\nOnly ONE harmonica exists.\n\nStory continuity:\n\nStart:\nTheo still owns/holds it after Part 1.\n\nMillie’s “giving it back” should be interpreted as her gently presenting/returning it into Theo’s hands after he has set it down or allowed her to hold it during the surprise setup.\n\nDo NOT duplicate it.\n\nIt should move clearly:\n\nTheo / nearby bench\n→ Millie\n→ Theo’s paws\n→ Theo plays it\n→ Theo keeps it at ending.\n\nThe harmonica symbolizes:\n\n# confidence he had hidden.\n\n==================================================\nHELPER CHARACTER LOCK\n==================================================\n\nUse exactly the same supporting helpers established in Part 1.\n\nBLUE BIRD:\nsmall rounded blue bird.\n\nYELLOW BIRD:\nsmall yellow songbird.\n\nSQUIRREL:\nfriendly brown squirrel with large fluffy tail.\n\nPIGEONS:\nsmall friendly gray park pigeons.\n\nVIOLINIST:\nfriendly human park musician with warm approachable expression, violin and bow.\n\nThey support Theo.\n\nThey do NOT become main characters.\n\n==================================================\nSURPRISE SETUP\n==================================================\n\nThe helpers transform the immediate bench area into a tiny homemade performance space.\n\nInclude:\n\n- a small ribbon garland\n- flower decorations\n- modest bunting\n- festival poster positioned clearly\n- birds on branches / ribbons\n- squirrel near flowers\n- pigeons near path\n- violinist nearby\n- warm string lights beginning to glow\n\nThe setup should feel:\n\nsmall,\nhandmade,\npersonal,\nheartfelt.\n\nNOT:\n\na giant stage,\narena,\nconcert venue,\nexpensive production.\n\nThe emotional idea is:\n\n“A few people made one safe little place for Theo to try.”\n\n==================================================\nENVIRONMENT CONTINUITY — CITY PARK\n==================================================\n\nMaintain the same City Park.\n\nUse:\n\n- green bench\n- CITY PARK lamp post\n- black iron fence\n- large leafy trees\n- flowers\n- fountain\n- pastel skyline\n- park path\n- Sunset Music Festival poster\n\nLIGHTING PROGRESSION:\n\nPart 1:\nlate afternoon.\n\nPart 2:\nlate afternoon gradually becomes beautiful golden sunset.\n\nBy final shot:\n\n- sky is peach-gold\n- city windows softly glow\n- lamp/park lights begin to illuminate\n- leaves receive warm rim light\n- Theo and Millie receive soft golden edge lighting\n\n==================================================\nWORLD / STORY LOGIC\n==================================================\n\nMillie gives signal\n→ helpers begin setup.\n\nMovement catches Theo’s attention\n→ Theo turns.\n\nTheo sees surprise\n→ Millie reveals it is for him.\n\nMillie presents harmonica\n→ Theo becomes nervous.\n\nTheo admits fear\n→ Millie reassures him.\n\nTheo chooses to try\n→ prepares himself.\n\nTheo takes breath\n→ first note.\n\nFirst note succeeds\n→ confidence increases.\n\nMusic attracts park listeners\n→ Theo realizes people enjoy hearing him.\n\nTheo finishes phrase\n→ invites Millie to celebrate.\n\nThey dance\n→ story resolves.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.8]\nMILLIE GIVES THE SIGNAL\n\nContinue EXACTLY from Part 1.\n\nMillie stands on the safe low base of the CITY PARK lamp post.\n\nOne tiny paw is raised like a conductor.\n\nTheo sits on the green bench with his back partly turned, still unaware.\n\nMillie looks:\n\nblue bird\n→ yellow bird\n→ squirrel\n→ pigeons\n→ violinist.\n\nEveryone is waiting.\n\nMillie whispers:\n\n<d>[English] Now.</d>\n\nShe lowers her paw.\n\nImmediately—but gently—\n\nthe helper plan begins.\n\nThe blue and yellow birds fly across frame carrying a small ribbon.\n\nThe squirrel places flowers beside the bench.\n\nTwo pigeons help pull open / reveal the handmade festival banner.\n\nThe violinist quietly moves into position.\n\nCAMERA:\n\nMillie low-angle hero close-up\n→ follow lowering paw\n→ smooth wide reveal of helpers beginning work.\n\nSFX:\n\n- wing flutters\n- tiny squirrel footsteps\n- fabric ribbon flutter\n- gentle pigeon coos\n- violin bow being readied\n\nMUSIC:\n\nPart 1 secret-mission motif begins resolving into warmer music.\n\n==================================================\n[SHOT 2 — 0:01.8–0:03.4]\nTHEO FINALLY NOTICES\n\nTheo’s ear twitches.\n\nHe hears:\n\nflutter,\ntiny footsteps,\nsoft violin tuning.\n\nHe slowly turns his head.\n\nA single pink flower petal lands on the brim of his black fedora.\n\nTheo looks up at it.\n\nConfused.\n\nHe turns fully around.\n\nCamera follows his gaze.\n\nReveal:\n\nMillie standing proudly in front of the handmade surprise.\n\nSmall decorations surround the bench.\n\nBirds perched above.\n\nSquirrel holding a flower.\n\nPigeons nearby.\n\nViolinist smiling in the background.\n\nMillie opens both arms.\n\nMillie:\n\n<d>[English] Surprise!</d>\n\nSmall pause.\n\n<d>[English] This is for you.</d>\n\nTheo’s golden eyes widen.\n\nHis usual composed expression disappears.\n\n==================================================\n[SHOT 3 — 0:03.4–0:05.0]\nTHE GOLDEN HARMONICA\n\nMove into an intimate close composition.\n\nMillie carefully holds the golden harmonica with BOTH tiny paws.\n\nBecause she is so small, the harmonica looks large in her hands.\n\nShe lifts it toward Theo.\n\nTheo slowly bends/lower himself enough to receive it.\n\nHe looks:\n\nMillie\n→ harmonica\n→ little surprise setup.\n\nHis ears lower slightly.\n\nHis expression becomes vulnerable.\n\nTheo takes the harmonica gently.\n\nDialogue:\n\nTheo:\n<d>[English] Millie...</d>\n\nMillie smiles.\n\nTheo:\n<d>[English] You did all this?</d>\n\nMillie gives tiny proud nod.\n\n<d>[English] We did.</d>\n\nThe squirrel gives a tiny wave behind her.\n\nSmall warm comedy beat.\n\n==================================================\n[SHOT 4 — 0:05.0–0:06.8]\n“WHAT IF I MESS UP?”\n\nTheo looks down at the harmonica.\n\nHis thumb / paw lightly traces its worn edge.\n\nThe park ambience softens.\n\nCamera moves closer.\n\nTheo glances toward the small group watching.\n\nHe becomes nervous again.\n\nDialogue:\n\nTheo:\n<d>[English] What if I mess up?</d>\n\nMillie looks genuinely surprised by the question.\n\nThen smiles softly.\n\nShe steps closer.\n\nMillie:\n\n<d>[English] Then we’ll still be here.</d>\n\nHOLD the two-character eye contact for a fraction of a second.\n\nThis is the emotional heart of Part 2.\n\nTheo’s golden eyes soften.\n\nNo melodrama.\n\nNo crying.\n\nJust understanding.\n\n==================================================\n[SHOT 5 — 0:06.8–0:08.4]\nTHEO DECIDES TO TRY\n\nTheo slowly stands.\n\nHis posture is still slightly nervous.\n\nThen his familiar gentleman ritual returns:\n\nhe straightens his pink tie.\n\nSmooths the front of his suit.\n\nAdjusts his jacket.\n\nTips the brim of his fedora.\n\nDeep breath.\n\nMillie watches from below with a giant proud smile.\n\nThe violinist raises the violin into playing position.\n\nTheo looks at Millie.\n\nMillie nods once.\n\nTheo:\n\n<d>[English] All right.</d>\n\nTiny pause.\n\n<d>[English] One song.</d>\n\nMillie immediately:\n\n<d>[English] One song.</d>\n\nHer smile says she knows this will be enough.\n\n==================================================\n[SHOT 6 — 0:08.4–0:10.4]\nONE BREATH — THE FIRST NOTE\n\nCLOSE-UP.\n\nTheo raises the golden harmonica.\n\nHis eyes close.\n\nOne visible slow breath.\n\nThe whole soundtrack briefly becomes quiet.\n\nThen—\n\nTheo plays ONE soft harmonica note.\n\nWarm, clean, slightly jazzy.\n\nA subtle golden musical line curls through the air.\n\nThe violinist joins with one gentle supporting note.\n\nTheo opens one eye.\n\nThe note sounded good.\n\nMillie’s face lights up.\n\nTheo plays a second phrase.\n\nLonger.\n\nSmoother.\n\nHis shoulders begin relaxing.\n\nHis ears lift.\n\nHis tail starts moving subtly with the beat.\n\nCAMERA:\n\nharmonica close-up\n→ Theo face\n→ Millie reaction\n→ slow widening as violin joins.\n\nNo flashy magical effects.\n\nMusical notes can be subtle stylized visual accents only.\n\n==================================================\n[SHOT 7 — 0:10.4–0:12.6]\nEVERY NOTE BRINGS CONFIDENCE BACK\n\nThe music builds into a warm playful jazz melody.\n\nTheo now plays with genuine rhythm.\n\nHis posture transforms naturally.\n\nBefore:\nstiff.\n\nNow:\ncomfortable.\n\nHis eyes become expressive.\n\nTail lightly sways.\n\nOne shoe taps the rhythm.\n\nMillie sways happily.\n\nBirds bob their heads.\n\nThe squirrel gently moves with the beat.\n\nPigeons listen.\n\nNearby park visitors begin stopping.\n\nDo NOT suddenly create a huge crowd.\n\nJust a few warm smiling listeners.\n\nThe golden sunset becomes richer.\n\nCamera performs a gentle side track around Theo and Millie.\n\nMusic carries the scene.\n\nNo dialogue necessary.\n\nThe visual should communicate:\n\nTheo has forgotten to worry about being judged.\n\n==================================================\n[SHOT 8 — 0:12.6–0:15.0]\nTHEO’S SURPRISE — DANCE + FINAL BENCH CALLBACK\n\nTheo finishes a short satisfying harmonica phrase.\n\nSoft applause begins.\n\nHe lowers the harmonica.\n\nFor the first time:\n\nTheo gives a completely genuine, open smile.\n\nHe looks down at Millie.\n\nExtends one elegant paw.\n\nTheo:\n\n<d>[English] May I?</d>\n\nMillie’s eyes widen.\n\nShe grins.\n\n<d>[English] Absolutely!</d>\n\nTheo gently lowers his paw.\n\nMillie places her tiny hand in it.\n\nTheo guides her into ONE small playful spin.\n\nMillie’s pink polka-dot skirt flares beautifully.\n\nTheo laughs.\n\nNot his stylish gentleman chuckle—\n\na real laugh.\n\nThe helper team applauds.\n\nThe violinist finishes with a warm flourish.\n\nUse a fast but elegant MATCH TRANSITION into the final emotional composition.\n\nSAME GREEN BENCH.\n\nNow golden sunset / early evening.\n\nTheo and Millie are sitting together again.\n\nTheo’s posture is relaxed.\n\nHis golden harmonica rests visibly in one paw / beside him rather than hidden in his suit.\n\nMillie sits close enough to feel like a trusted friend.\n\nWarm park lights glow.\n\nThe pastel skyline catches sunset.\n\nBlue bird and squirrel remain nearby.\n\nTheo looks toward Millie.\n\nHe tips his fedora gently.\n\nTheo:\n\n<d>[English] Thank you.</d>\n\nMillie:\n\n<d>[English] Same time tomorrow?</d>\n\nTheo looks at the harmonica.\n\nThen smiles.\n\n<d>[English] Maybe two songs.</d>\n\nMillie beams.\n\nCamera slowly pulls backward.\n\nThe two become smaller against the warm City Park.\n\nA few soft musical notes linger.\n\nFADE TO BLACK.\n\nEND STORY.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio should feel:\n\n- warm\n- elegant\n- musical\n- playful\n- natural\n- emotionally sincere\n- cinematic\n\n==================================================\nTHEO VOICE\n==================================================\n\nAdult male anthropomorphic-cat voice.\n\nQualities:\n\n- smooth\n- refined\n- warm\n- calm\n- soft vintage-jazz personality\n- understated emotion\n- never theatrical or threatening\n\nBeginning:\nconfused and touched.\n\nMiddle:\nquietly vulnerable.\n\nPerformance:\nfocused.\n\nEnding:\ngenuinely relaxed.\n\nImportant lines:\n\n<d>[English] You did all this?</d>\n\nDeliver with surprise and emotion.\n\n<d>[English] What if I mess up?</d>\n\nDo NOT sound childish.\n\nIt should feel like a sincere admission from someone who usually hides uncertainty.\n\nFinal:\n\n<d>[English] Maybe two songs.</d>\n\nUse warm dry humor.\n\n==================================================\nMILLIE VOICE\n==================================================\n\nYoung feminine mouse voice.\n\nQualities:\n\n- bright\n- gentle\n- cheerful\n- warm\n- encouraging\n- emotionally sincere\n- energetic but not shrill\n\nCritical line:\n\n<d>[English] Then we’ll still be here.</d>\n\nDeliver calmly.\n\nDo NOT oversell emotion.\n\nThat simple certainty is what affects Theo.\n\nFinal:\n\n<d>[English] Same time tomorrow?</d>\n\nplayful and optimistic.\n\n==================================================\nVIOLIN AUDIO\n==================================================\n\nThe violinist should support Theo.\n\nDo NOT overpower the harmonica.\n\nProgression:\n\nfirst harmonica note\n→ gentle violin sustained note\n\nsecond phrase\n→ soft countermelody\n\nmain performance\n→ warm supportive accompaniment\n\ndance\n→ slightly brighter flourish\n\nEnding:\nsoft resolved violin note.\n\n==================================================\nHARMONICA PERFORMANCE\n==================================================\n\nThe harmonica is the central musical instrument.\n\nSound:\n\n- warm\n- jazzy\n- melodic\n- expressive\n- intimate\n- slightly vintage\n\nFirst note:\ntentative but clean.\n\nSecond phrase:\nmore confident.\n\nBy main performance:\nrich, playful melody.\n\nDo NOT make Theo instantly sound like a massive concert virtuoso.\n\nThe emotional point is:\n\nhe starts cautiously and grows into it.\n\n==================================================\nPARK HELPER AUDIO\n==================================================\n\nBirds:\nsoft cheerful chirping.\n\nSquirrel:\nsmall happy chatter.\n\nPigeons:\ngentle coos.\n\nApplause:\nwarm and modest.\n\nNo screaming crowd.\n\nNo massive cheering stadium.\n\n==================================================\nENVIRONMENT AUDIO\n==================================================\n\nMaintain City Park atmosphere:\n\n- gentle wind\n- leaves\n- distant fountain\n- quiet city background\n- footsteps\n- birds\n- subtle park ambience\n\nAs music begins:\n\nreduce ambient distractions slightly.\n\nAt ending:\n\nrestore:\nfountain,\nwind,\nnight birds,\nsoft city ambience.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL warm cinematic jazz-storybook score.\n\nInstrumentation:\n\n- piano\n- upright bass\n- brushed drums\n- clarinet\n- pizzicato strings\n- vibraphone\n- subtle acoustic guitar\n- warm strings\n- harmonica integrated diegetically\n- violin integrated diegetically\n\nMUSIC ARC:\n\n0:00–0:02\nsecret plan resolves into surprise motif.\n\n0:02–0:04\nwarm emotional reveal.\n\n0:04–0:06.8\nmusic becomes sparse during Theo’s vulnerability.\n\n0:06.8–0:08.4\ngentle hopeful rise.\n\n0:08.4–0:10.4\nmusic nearly disappears for first harmonica note.\n\n0:10.4–0:12.6\nfull warm jazz groove develops.\n\n0:12.6–0:14\njoyful dance flourish.\n\n0:14–0:15\nreturn to the SAME melody from Part 1 opening, now fuller and emotionally resolved.\n\nNo giant orchestral trailer music.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 2 camera should evolve from:\n\nplayful preparation\n→ intimate emotional close-ups\n→ musical freedom\n→ peaceful final closure.\n\nUse:\n\n- gentle tracking\n- eye-level character coverage\n- over-the-shoulder compositions\n- close harmonica inserts\n- rack focus\n- slow push-ins\n- subtle orbit during performance\n- clean wide sunset park views\n- low scale framing for Millie\n- final slow pull-back\n\nAvoid:\n\n- shaky handheld\n- crash zoom\n- chaotic whip-pan\n- excessive motion blur\n- extreme fisheye\n- Dutch angles\n- 360-degree camera spins\n\nThe camera becomes MORE relaxed as Theo becomes more confident.\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished hand-painted 2D storybook animation.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n\n- clean hand-drawn ink contours\n- watercolor / gouache-inspired fills\n- expressive faces\n- softly stylized fur\n- hand-painted city backgrounds\n- gentle vintage cartoon charm\n- subtle cinematic depth\n- warm sunset lighting\n- natural cloth animation\n- controlled squash-and-stretch\n- painterly shadows\n- soft film texture\n\nThe result should look like:\n\nTHE UPLOADED STORYBOARD ART CAME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- slideshow\n- motion comic\n- live-action animals\n- photorealistic cat/mouse\n- generic 3D CGI\n- hyper-real fur\n- anime\n- horror\n\n==================================================\nCOLOR / LIGHTING ARC\n==================================================\n\nOPENING:\n\nlate afternoon.\n\nColors:\n\n- spring green\n- sky blue\n- pastel beige\n- light city pink\n- flower colors\n\nTHEO:\n\n- deep charcoal\n- cream\n- bright pink tie\n- golden eyes\n\nMILLIE:\n\n- warm beige\n- soft pink\n- white polka dots\n\nHARMONICA:\n\nwarm polished gold.\n\nAs Theo begins playing:\n\ntransition lighting toward:\n\n- honey gold\n- amber\n- peach\n- soft sunset orange\n\nFINAL SHOT:\n\n- golden sunset\n- warm park lamps\n- soft amber skyline windows\n- cool green shadows\n- subtle pink sky accents\n\nFaces remain bright and readable.\n\n==================================================\nMOTION RULES — THEO\n==================================================\n\nBefore playing:\n\n- small ear movements\n- slightly tense shoulders\n- controlled paw gestures\n- careful harmonica handling\n\nPreparing:\n\ntie straightening\n→ suit adjustment\n→ fedora tip\n→ breath.\n\nPlaying:\n\nhands stable,\nhead gently moves with melody,\ntail gains rhythm,\nshoulders loosen.\n\nDance invitation:\n\nslow elegant paw extension.\n\nFinal:\n\nfully relaxed seated posture.\n\n==================================================\nMOTION RULES — MILLIE\n==================================================\n\nUse:\n\n- energetic conductor motion\n- tiny proud steps\n- careful two-paw harmonica presentation\n- gentle reassuring expression\n- excited sway to music\n- believable small spin during dance\n- dress/skirt secondary movement\n- bow remains stable\n\nMillie must remain tiny relative to Theo.\n\n==================================================\nDANCE SAFETY / MOTION\n==================================================\n\nKeep the dance:\n\n- gentle\n- wholesome\n- brief\n- playful\n- family-friendly\n\nTheo extends one paw.\n\nMillie takes it.\n\nTheo gently raises his hand enough for Millie to turn once beneath it.\n\nNo lifting Millie high into the air.\n\nNo throwing.\n\nNo acrobatics.\n\nNo romantic kiss.\n\nThe moment should read as:\n\ncelebratory friendship.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Theo.\nONE Millie.\nONE golden harmonica.\n\nSame:\n\n- Theo face\n- Theo black fur\n- golden eyes\n- black fedora\n- pink tie\n- suit\n- Millie face\n- bow\n- dress\n- shoes\n- relative scale\n- park bench\n- CITY PARK lamp post\n- skyline\n- helper designs\n- festival poster\n- violinist\n\nHarmonica progression must remain physically logical.\n\nDo NOT duplicate it.\n\nOnce Theo begins playing:\nkeep harmonica with Theo.\n\nFinal bench:\nharmonica must remain visible.\n\n==================================================\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086426551114215889",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T12:17:08.373Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086426551114215889.mp4",
    "posterUrl": "/posters/x/2086426551114215889.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "dialogue",
      "dance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second Storyboard Image to video using MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n\"[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for:\n\n“THE GENTLEMAN CAT AND THE LITTLE MOUSE’S BIG SURPRISE”\n\nas the STRICT PRIMARY visual, character, environment, prop, lighting, performance, scale, and narrative reference.\n\nThis video continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n\n- Theo still sitting alone on the green City Park bench\n- Theo’s golden harmonica still in his possession\n- Theo believing he may not perform at the Sunset Music Festival\n- Millie secretly gathering the blue bird, yellow bird, squirrel, pigeons, and friendly violinist\n- the helpers quietly moving into position\n- Theo remaining completely unaware\n- Millie standing on the CITY PARK lamp-post base like a tiny conductor\n- Millie raising one paw\n- Millie whispering:\n  <d>[English] Ready?</d>\n\nPART 2 begins at that EXACT moment.\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT animate or reproduce the storyboard sheet itself.\n\nThe finished video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- reference drawings\n- color swatches\n- title/header\n- page number\n- storyboard paper texture\n- labels\n- subtitles\n\nThe uploaded storyboard locks:\n\n1. Theo’s exact black-cat identity\n2. Theo’s golden-yellow eyes\n3. Theo’s elegant black fedora\n4. fedora feather detail\n5. charcoal gentleman suit\n6. cream shirt\n7. bright pink tie\n8. matching pink pocket square\n9. gold pocket chain\n10. black-and-cream shoes\n11. Theo’s lean gentleman proportions\n12. Millie’s exact mouse identity\n13. Millie’s beige/taupe fur\n14. Millie’s large rounded ears\n15. pink polka-dot bow\n16. matching pink polka-dot dress\n17. pink shoes\n18. Theo/Millie size contrast\n19. golden harmonica\n20. green City Park bench\n21. CITY PARK lamp post\n22. festival poster\n23. birds, squirrel and pigeons\n24. friendly violinist\n25. handmade surprise decoration\n26. golden-hour park lighting\n27. pastel city skyline\n28. Part 2 harmonica-performance payoff\n29. Theo inviting Millie to dance\n30. final warm bench callback\n\nTranslate the storyboard’s fifteen 1-second panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT make 15 frantic one-second cuts.\n\nUse this progression:\n\nMillie gives signal\n→ helpers reveal surprise\n→ Theo realizes it is for him\n→ Millie presents harmonica\n→ Theo hesitates\n→ Millie reassures him\n→ Theo chooses to perform\n→ first note\n→ confidence grows\n→ park begins listening\n→ Theo invites Millie\n→ tiny joyful dance\n→ emotional bench ending.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\n\n# EXACTLY 15 SECONDS\n\nStory range:\n\nPART 2 — 0:15 to 0:30\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D storybook animation\n\nGenre:\nheartwarming musical friendship short\n\nTone:\n\n- charming\n- elegant\n- funny\n- sincere\n- uplifting\n- warm\n- emotionally satisfying\n- family-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + harmonica + violin + park ambience + cinematic music\n\nCORE PAYOFF:\n\nTheo does not suddenly become fearless.\n\nMillie creates a place where he feels safe enough to TRY.\n\nTheo plays one note.\n\nThen another.\n\nAnd slowly discovers that people actually want to hear him.\n\n==================================================\nPART 2 START / END\n==================================================\n\nSTART:\n\nMillie lowers her conductor paw and launches her surprise.\n\nTheo is still unaware.\n\nEND:\n\nTheo and Millie are sitting together on the same green bench at sunset.\n\nTheo’s harmonica is now proudly visible instead of hidden.\n\nTheo tips his fedora toward Millie with a genuine smile.\n\nMillie smiles back.\n\nThe story must feel COMPLETE.\n\nNo cliffhanger.\n\n==================================================\nCHARACTER IDENTITY LOCK — THEO\n==================================================\n\nMatch PART 1 and the uploaded PART 2 storyboard EXACTLY.\n\nTHEO — THE GENTLEMAN CAT\n\nAppearance:\n\n- sleek charcoal-black / deep black fur\n- soft gray muzzle\n- small black feline nose\n- clear whiskers\n- triangular cat ears\n- expressive golden-yellow eyes\n- long black tail\n- tall lean anthropomorphic proportions\n\nCLOTHING:\n\n- tailored charcoal suit jacket\n- matching dark trousers\n- cream dress shirt\n- bright pink tie\n- matching pink pocket square\n- black fedora\n- small feather tucked into hat\n- subtle gold pocket chain\n- polished black-and-cream shoes\n\nNever change his clothes.\n\nNever remove his fedora permanently.\n\nPERSONALITY:\n\n- refined\n- charming\n- dryly witty\n- musically talented\n- emotionally gentle\n- outwardly composed\n- secretly nervous\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\nconfused by sudden activity\n\n3–5 sec:\nsurprised and emotionally touched\n\n5–7 sec:\nfear returns when he sees harmonica\n\n7–9 sec:\ndecides to try\n\n9–12 sec:\nnervous musician → genuine confidence\n\n12–14 sec:\nopen happiness\n\n14–15 sec:\npeaceful gratitude\n\nCRITICAL:\n\nTheo’s body language must tell the story.\n\nAt first:\n\ntight shoulders,\nears slightly lowered,\nharmonica held carefully.\n\nAs he plays:\n\nshoulders loosen,\nears rise,\ntail begins moving with rhythm,\neyes brighten,\nsmile becomes genuine.\n\n==================================================\nCHARACTER IDENTITY LOCK — MILLIE\n==================================================\n\nMatch PART 1 EXACTLY.\n\nMILLIE — THE LITTLE MOUSE\n\nAppearance:\n\n- tiny warm beige/taupe mouse\n- very large circular ears\n- soft pink inner ears\n- large expressive dark eyes\n- long eyelashes\n- tiny black nose\n- delicate whiskers\n- thin mouse tail\n- petite proportions\n\nCLOTHING:\n\n- large pink bow with white polka dots\n- pink sleeveless polka-dot dress\n- softly flared skirt\n- pink Mary Jane shoes\n\nPersonality:\n\n- cheerful\n- observant\n- supportive\n- emotionally intelligent\n- playful\n- determined\n- affectionate\n- brave on behalf of her friend\n\nPART 2 PERFORMANCE:\n\n0–3 sec:\nconfident little organizer\n\n3–5 sec:\nexcited surprise reveal\n\n5–7 sec:\ngentle reassurance\n\n7–11 sec:\nquiet supportive witness\n\n11–14 sec:\npure joy\n\n14–15 sec:\nwarmly satisfied\n\nIMPORTANT:\n\nMillie NEVER pressures Theo.\n\nHer encouragement communicates:\n\n“You don’t have to be perfect. We’re staying either way.”\n\n==================================================\nGOLDEN HARMONICA LOCK\n==================================================\n\nMatch the storyboard EXACTLY.\n\nTheo’s golden harmonica:\n\n- small vintage rectangular harmonica\n- warm polished brass\n- fine engraved lines\n- slightly worn sentimental edges\n- metallic golden reflections\n\nOnly ONE harmonica exists.\n\nStory continuity:\n\nStart:\nTheo still owns/holds it after Part 1.\n\nMillie’s “giving it back” should be interpreted as her gently presenting/returning it into Theo’s hands after he has set it down or allowed her to hold it during the surprise setup.\n\nDo NOT duplicate it.\n\nIt should move clearly:\n\nTheo / nearby bench\n→ Millie\n→ Theo’s paws\n→ Theo plays it\n→ Theo keeps it at ending.\n\nThe harmonica symbolizes:\n\n# confidence he had hidden.\n\n==================================================\nHELPER CHARACTER LOCK\n==================================================\n\nUse exactly the same supporting helpers established in Part 1.\n\nBLUE BIRD:\nsmall rounded blue bird.\n\nYELLOW BIRD:\nsmall yellow songbird.\n\nSQUIRREL:\nfriendly brown squirrel with large fluffy tail.\n\nPIGEONS:\nsmall friendly gray park pigeons.\n\nVIOLINIST:\nfriendly human park musician with warm approachable expression, violin and bow.\n\nThey support Theo.\n\nThey do NOT become main characters.\n\n==================================================\nSURPRISE SETUP\n==================================================\n\nThe helpers transform the immediate bench area into a tiny homemade performance space.\n\nInclude:\n\n- a small ribbon garland\n- flower decorations\n- modest bunting\n- festival poster positioned clearly\n- birds on branches / ribbons\n- squirrel near flowers\n- pigeons near path\n- violinist nearby\n- warm string lights beginning to glow\n\nThe setup should feel:\n\nsmall,\nhandmade,\npersonal,\nheartfelt.\n\nNOT:\n\na giant stage,\narena,\nconcert venue,\nexpensive production.\n\nThe emotional idea is:\n\n“A few people made one safe little place for Theo to try.”\n\n==================================================\nENVIRONMENT CONTINUITY — CITY PARK\n==================================================\n\nMaintain the same City Park.\n\nUse:\n\n- green bench\n- CITY PARK lamp post\n- black iron fence\n- large leafy trees\n- flowers\n- fountain\n- pastel skyline\n- park path\n- Sunset Music Festival poster\n\nLIGHTING PROGRESSION:\n\nPart 1:\nlate afternoon.\n\nPart 2:\nlate afternoon gradually becomes beautiful golden sunset.\n\nBy final shot:\n\n- sky is peach-gold\n- city windows softly glow\n- lamp/park lights begin to illuminate\n- leaves receive warm rim light\n- Theo and Millie receive soft golden edge lighting\n\n==================================================\nWORLD / STORY LOGIC\n==================================================\n\nMillie gives signal\n→ helpers begin setup.\n\nMovement catches Theo’s attention\n→ Theo turns.\n\nTheo sees surprise\n→ Millie reveals it is for him.\n\nMillie presents harmonica\n→ Theo becomes nervous.\n\nTheo admits fear\n→ Millie reassures him.\n\nTheo chooses to try\n→ prepares himself.\n\nTheo takes breath\n→ first note.\n\nFirst note succeeds\n→ confidence increases.\n\nMusic attracts park listeners\n→ Theo realizes people enjoy hearing him.\n\nTheo finishes phrase\n→ invites Millie to celebrate.\n\nThey dance\n→ story resolves.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.8]\nMILLIE GIVES THE SIGNAL\n\nContinue EXACTLY from Part 1.\n\nMillie stands on the safe low base of the CITY PARK lamp post.\n\nOne tiny paw is raised like a conductor.\n\nTheo sits on the green bench with his back partly turned, still unaware.\n\nMillie looks:\n\nblue bird\n→ yellow bird\n→ squirrel\n→ pigeons\n→ violinist.\n\nEveryone is waiting.\n\nMillie whispers:\n\n<d>[English] Now.</d>\n\nShe lowers her paw.\n\nImmediately—but gently—\n\nthe helper plan begins.\n\nThe blue and yellow birds fly across frame carrying a small ribbon.\n\nThe squirrel places flowers beside the bench.\n\nTwo pigeons help pull open / reveal the handmade festival banner.\n\nThe violinist quietly moves into position.\n\nCAMERA:\n\nMillie low-angle hero close-up\n→ follow lowering paw\n→ smooth wide reveal of helpers beginning work.\n\nSFX:\n\n- wing flutters\n- tiny squirrel footsteps\n- fabric ribbon flutter\n- gentle pigeon coos\n- violin bow being readied\n\nMUSIC:\n\nPart 1 secret-mission motif begins resolving into warmer music.\n\n==================================================\n[SHOT 2 — 0:01.8–0:03.4]\nTHEO FINALLY NOTICES\n\nTheo’s ear twitches.\n\nHe hears:\n\nflutter,\ntiny footsteps,\nsoft violin tuning.\n\nHe slowly turns his head.\n\nA single pink flower petal lands on the brim of his black fedora.\n\nTheo looks up at it.\n\nConfused.\n\nHe turns fully around.\n\nCamera follows his gaze.\n\nReveal:\n\nMillie standing proudly in front of the handmade surprise.\n\nSmall decorations surround the bench.\n\nBirds perched above.\n\nSquirrel holding a flower.\n\nPigeons nearby.\n\nViolinist smiling in the background.\n\nMillie opens both arms.\n\nMillie:\n\n<d>[English] Surprise!</d>\n\nSmall pause.\n\n<d>[English] This is for you.</d>\n\nTheo’s golden eyes widen.\n\nHis usual composed expression disappears.\n\n==================================================\n[SHOT 3 — 0:03.4–0:05.0]\nTHE GOLDEN HARMONICA\n\nMove into an intimate close composition.\n\nMillie carefully holds the golden harmonica with BOTH tiny paws.\n\nBecause she is so small, the harmonica looks large in her hands.\n\nShe lifts it toward Theo.\n\nTheo slowly bends/lower himself enough to receive it.\n\nHe looks:\n\nMillie\n→ harmonica\n→ little surprise setup.\n\nHis ears lower slightly.\n\nHis expression becomes vulnerable.\n\nTheo takes the harmonica gently.\n\nDialogue:\n\nTheo:\n<d>[English] Millie...</d>\n\nMillie smiles.\n\nTheo:\n<d>[English] You did all this?</d>\n\nMillie gives tiny proud nod.\n\n<d>[English] We did.</d>\n\nThe squirrel gives a tiny wave behind her.\n\nSmall warm comedy beat.\n\n==================================================\n[SHOT 4 — 0:05.0–0:06.8]\n“WHAT IF I MESS UP?”\n\nTheo looks down at the harmonica.\n\nHis thumb / paw lightly traces its worn edge.\n\nThe park ambience softens.\n\nCamera moves closer.\n\nTheo glances toward the small group watching.\n\nHe becomes nervous again.\n\nDialogue:\n\nTheo:\n<d>[English] What if I mess up?</d>\n\nMillie looks genuinely surprised by the question.\n\nThen smiles softly.\n\nShe steps closer.\n\nMillie:\n\n<d>[English] Then we’ll still be here.</d>\n\nHOLD the two-character eye contact for a fraction of a second.\n\nThis is the emotional heart of Part 2.\n\nTheo’s golden eyes soften.\n\nNo melodrama.\n\nNo crying.\n\nJust understanding.\n\n==================================================\n[SHOT 5 — 0:06.8–0:08.4]\nTHEO DECIDES TO TRY\n\nTheo slowly stands.\n\nHis posture is still slightly nervous.\n\nThen his familiar gentleman ritual returns:\n\nhe straightens his pink tie.\n\nSmooths the front of his suit.\n\nAdjusts his jacket.\n\nTips the brim of his fedora.\n\nDeep breath.\n\nMillie watches from below with a giant proud smile.\n\nThe violinist raises the violin into playing position.\n\nTheo looks at Millie.\n\nMillie nods once.\n\nTheo:\n\n<d>[English] All right.</d>\n\nTiny pause.\n\n<d>[English] One song.</d>\n\nMillie immediately:\n\n<d>[English] One song.</d>\n\nHer smile says she knows this will be enough.\n\n==================================================\n[SHOT 6 — 0:08.4–0:10.4]\nONE BREATH — THE FIRST NOTE\n\nCLOSE-UP.\n\nTheo raises the golden harmonica.\n\nHis eyes close.\n\nOne visible slow breath.\n\nThe whole soundtrack briefly becomes quiet.\n\nThen—\n\nTheo plays ONE soft harmonica note.\n\nWarm, clean, slightly jazzy.\n\nA subtle golden musical line curls through the air.\n\nThe violinist joins with one gentle supporting note.\n\nTheo opens one eye.\n\nThe note sounded good.\n\nMillie’s face lights up.\n\nTheo plays a second phrase.\n\nLonger.\n\nSmoother.\n\nHis shoulders begin relaxing.\n\nHis ears lift.\n\nHis tail starts moving subtly with the beat.\n\nCAMERA:\n\nharmonica close-up\n→ Theo face\n→ Millie reaction\n→ slow widening as violin joins.\n\nNo flashy magical effects.\n\nMusical notes can be subtle stylized visual accents only.\n\n==================================================\n[SHOT 7 — 0:10.4–0:12.6]\nEVERY NOTE BRINGS CONFIDENCE BACK\n\nThe music builds into a warm playful jazz melody.\n\nTheo now plays with genuine rhythm.\n\nHis posture transforms naturally.\n\nBefore:\nstiff.\n\nNow:\ncomfortable.\n\nHis eyes become expressive.\n\nTail lightly sways.\n\nOne shoe taps the rhythm.\n\nMillie sways happily.\n\nBirds bob their heads.\n\nThe squirrel gently moves with the beat.\n\nPigeons listen.\n\nNearby park visitors begin stopping.\n\nDo NOT suddenly create a huge crowd.\n\nJust a few warm smiling listeners.\n\nThe golden sunset becomes richer.\n\nCamera performs a gentle side track around Theo and Millie.\n\nMusic carries the scene.\n\nNo dialogue necessary.\n\nThe visual should communicate:\n\nTheo has forgotten to worry about being judged.\n\n==================================================\n[SHOT 8 — 0:12.6–0:15.0]\nTHEO’S SURPRISE — DANCE + FINAL BENCH CALLBACK\n\nTheo finishes a short satisfying harmonica phrase.\n\nSoft applause begins.\n\nHe lowers the harmonica.\n\nFor the first time:\n\nTheo gives a completely genuine, open smile.\n\nHe looks down at Millie.\n\nExtends one elegant paw.\n\nTheo:\n\n<d>[English] May I?</d>\n\nMillie’s eyes widen.\n\nShe grins.\n\n<d>[English] Absolutely!</d>\n\nTheo gently lowers his paw.\n\nMillie places her tiny hand in it.\n\nTheo guides her into ONE small playful spin.\n\nMillie’s pink polka-dot skirt flares beautifully.\n\nTheo laughs.\n\nNot his stylish gentleman chuckle—\n\na real laugh.\n\nThe helper team applauds.\n\nThe violinist finishes with a warm flourish.\n\nUse a fast but elegant MATCH TRANSITION into the final emotional composition.\n\nSAME GREEN BENCH.\n\nNow golden sunset / early evening.\n\nTheo and Millie are sitting together again.\n\nTheo’s posture is relaxed.\n\nHis golden harmonica rests visibly in one paw / beside him rather than hidden in his suit.\n\nMillie sits close enough to feel like a trusted friend.\n\nWarm park lights glow.\n\nThe pastel skyline catches sunset.\n\nBlue bird and squirrel remain nearby.\n\nTheo looks toward Millie.\n\nHe tips his fedora gently.\n\nTheo:\n\n<d>[English] Thank you.</d>\n\nMillie:\n\n<d>[English] Same time tomorrow?</d>\n\nTheo looks at the harmonica.\n\nThen smiles.\n\n<d>[English] Maybe two songs.</d>\n\nMillie beams.\n\nCamera slowly pulls backward.\n\nThe two become smaller against the warm City Park.\n\nA few soft musical notes linger.\n\nFADE TO BLACK.\n\nEND STORY.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio should feel:\n\n- warm\n- elegant\n- musical\n- playful\n- natural\n- emotionally sincere\n- cinematic\n\n==================================================\nTHEO VOICE\n==================================================\n\nAdult male anthropomorphic-cat voice.\n\nQualities:\n\n- smooth\n- refined\n- warm\n- calm\n- soft vintage-jazz personality\n- understated emotion\n- never theatrical or threatening\n\nBeginning:\nconfused and touched.\n\nMiddle:\nquietly vulnerable.\n\nPerformance:\nfocused.\n\nEnding:\ngenuinely relaxed.\n\nImportant lines:\n\n<d>[English] You did all this?</d>\n\nDeliver with surprise and emotion.\n\n<d>[English] What if I mess up?</d>\n\nDo NOT sound childish.\n\nIt should feel like a sincere admission from someone who usually hides uncertainty.\n\nFinal:\n\n<d>[English] Maybe two songs.</d>\n\nUse warm dry humor.\n\n==================================================\nMILLIE VOICE\n==================================================\n\nYoung feminine mouse voice.\n\nQualities:\n\n- bright\n- gentle\n- cheerful\n- warm\n- encouraging\n- emotionally sincere\n- energetic but not shrill\n\nCritical line:\n\n<d>[English] Then we’ll still be here.</d>\n\nDeliver calmly.\n\nDo NOT oversell emotion.\n\nThat simple certainty is what affects Theo.\n\nFinal:\n\n<d>[English] Same time tomorrow?</d>\n\nplayful and optimistic.\n\n==================================================\nVIOLIN AUDIO\n==================================================\n\nThe violinist should support Theo.\n\nDo NOT overpower the harmonica.\n\nProgression:\n\nfirst harmonica note\n→ gentle violin sustained note\n\nsecond phrase\n→ soft countermelody\n\nmain performance\n→ warm supportive accompaniment\n\ndance\n→ slightly brighter flourish\n\nEnding:\nsoft resolved violin note.\n\n==================================================\nHARMONICA PERFORMANCE\n==================================================\n\nThe harmonica is the central musical instrument.\n\nSound:\n\n- warm\n- jazzy\n- melodic\n- expressive\n- intimate\n- slightly vintage\n\nFirst note:\ntentative but clean.\n\nSecond phrase:\nmore confident.\n\nBy main performance:\nrich, playful melody.\n\nDo NOT make Theo instantly sound like a massive concert virtuoso.\n\nThe emotional point is:\n\nhe starts cautiously and grows into it.\n\n==================================================\nPARK HELPER AUDIO\n==================================================\n\nBirds:\nsoft cheerful chirping.\n\nSquirrel:\nsmall happy chatter.\n\nPigeons:\ngentle coos.\n\nApplause:\nwarm and modest.\n\nNo screaming crowd.\n\nNo massive cheering stadium.\n\n==================================================\nENVIRONMENT AUDIO\n==================================================\n\nMaintain City Park atmosphere:\n\n- gentle wind\n- leaves\n- distant fountain\n- quiet city background\n- footsteps\n- birds\n- subtle park ambience\n\nAs music begins:\n\nreduce ambient distractions slightly.\n\nAt ending:\n\nrestore:\nfountain,\nwind,\nnight birds,\nsoft city ambience.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL warm cinematic jazz-storybook score.\n\nInstrumentation:\n\n- piano\n- upright bass\n- brushed drums\n- clarinet\n- pizzicato strings\n- vibraphone\n- subtle acoustic guitar\n- warm strings\n- harmonica integrated diegetically\n- violin integrated diegetically\n\nMUSIC ARC:\n\n0:00–0:02\nsecret plan resolves into surprise motif.\n\n0:02–0:04\nwarm emotional reveal.\n\n0:04–0:06.8\nmusic becomes sparse during Theo’s vulnerability.\n\n0:06.8–0:08.4\ngentle hopeful rise.\n\n0:08.4–0:10.4\nmusic nearly disappears for first harmonica note.\n\n0:10.4–0:12.6\nfull warm jazz groove develops.\n\n0:12.6–0:14\njoyful dance flourish.\n\n0:14–0:15\nreturn to the SAME melody from Part 1 opening, now fuller and emotionally resolved.\n\nNo giant orchestral trailer music.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 2 camera should evolve from:\n\nplayful preparation\n→ intimate emotional close-ups\n→ musical freedom\n→ peaceful final closure.\n\nUse:\n\n- gentle tracking\n- eye-level character coverage\n- over-the-shoulder compositions\n- close harmonica inserts\n- rack focus\n- slow push-ins\n- subtle orbit during performance\n- clean wide sunset park views\n- low scale framing for Millie\n- final slow pull-back\n\nAvoid:\n\n- shaky handheld\n- crash zoom\n- chaotic whip-pan\n- excessive motion blur\n- extreme fisheye\n- Dutch angles\n- 360-degree camera spins\n\nThe camera becomes MORE relaxed as Theo becomes more confident.\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished hand-painted 2D storybook animation.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n\n- clean hand-drawn ink contours\n- watercolor / gouache-inspired fills\n- expressive faces\n- softly stylized fur\n- hand-painted city backgrounds\n- gentle vintage cartoon charm\n- subtle cinematic depth\n- warm sunset lighting\n- natural cloth animation\n- controlled squash-and-stretch\n- painterly shadows\n- soft film texture\n\nThe result should look like:\n\nTHE UPLOADED STORYBOARD ART CAME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- slideshow\n- motion comic\n- live-action animals\n- photorealistic cat/mouse\n- generic 3D CGI\n- hyper-real fur\n- anime\n- horror\n\n==================================================\nCOLOR / LIGHTING ARC\n==================================================\n\nOPENING:\n\nlate afternoon.\n\nColors:\n\n- spring green\n- sky blue\n- pastel beige\n- light city pink\n- flower colors\n\nTHEO:\n\n- deep charcoal\n- cream\n- bright pink tie\n- golden eyes\n\nMILLIE:\n\n- warm beige\n- soft pink\n- white polka dots\n\nHARMONICA:\n\nwarm polished gold.\n\nAs Theo begins playing:\n\ntransition lighting toward:\n\n- honey gold\n- amber\n- peach\n- soft sunset orange\n\nFINAL SHOT:\n\n- golden sunset\n- warm park lamps\n- soft amber skyline windows\n- cool green shadows\n- subtle pink sky accents\n\nFaces remain bright and readable.\n\n==================================================\nMOTION RULES — THEO\n==================================================\n\nBefore playing:\n\n- small ear movements\n- slightly tense shoulders\n- controlled paw gestures\n- careful harmonica handling\n\nPreparing:\n\ntie straightening\n→ suit adjustment\n→ fedora tip\n→ breath.\n\nPlaying:\n\nhands stable,\nhead gently moves with melody,\ntail gains rhythm,\nshoulders loosen.\n\nDance invitation:\n\nslow elegant paw extension.\n\nFinal:\n\nfully relaxed seated posture.\n\n==================================================\nMOTION RULES — MILLIE\n==================================================\n\nUse:\n\n- energetic conductor motion\n- tiny proud steps\n- careful two-paw harmonica presentation\n- gentle reassuring expression\n- excited sway to music\n- believable small spin during dance\n- dress/skirt secondary movement\n- bow remains stable\n\nMillie must remain tiny relative to Theo.\n\n==================================================\nDANCE SAFETY / MOTION\n==================================================\n\nKeep the dance:\n\n- gentle\n- wholesome\n- brief\n- playful\n- family-friendly\n\nTheo extends one paw.\n\nMillie takes it.\n\nTheo gently raises his hand enough for Millie to turn once beneath it.\n\nNo lifting Millie high into the air.\n\nNo throwing.\n\nNo acrobatics.\n\nNo romantic kiss.\n\nThe moment should read as:\n\ncelebratory friendship.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Theo.\nONE Millie.\nONE golden harmonica.\n\nSame:\n\n- Theo face\n- Theo black fur\n- golden eyes\n- black fedora\n- pink tie\n- suit\n- Millie face\n- bow\n- dress\n- shoes\n- relative scale\n- park bench\n- CITY PARK lamp post\n- skyline\n- helper designs\n- festival poster\n- violinist\n\nHarmonica progression must remain physically logical.\n\nDo NOT duplicate it.\n\nOnce Theo begins playing:\nkeep harmonica with Theo.\n\nFinal bench:\nharmonica must remain visible.\n\n==================================================\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086426551114215889"
  },
  {
    "id": "x-2086426143612440739",
    "title": "绅士猫与小老鼠的惊喜 · 上篇",
    "titleEn": "The Gentleman Cat and Little Mouse Surprise · Part 1",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《绅士猫与小老鼠的惊喜 · 上篇》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Gentleman Cat and Little Mouse Surprise · Part 1. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "\"[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for:\n\n“THE GENTLEMAN CAT AND THE LITTLE MOUSE’S BIG SURPRISE”\n\nas the STRICT PRIMARY visual reference for the finished video.\n\nThe uploaded storyboard controls:\n\n- Theo’s exact face and body identity\n- Theo’s black fur\n- Theo’s yellow-gold eyes\n- Theo’s black fedora and feather detail\n- Theo’s charcoal gentleman suit\n- Theo’s cream shirt\n- Theo’s bright pink tie and pocket square\n- Theo’s black tail\n- Theo’s black-and-cream dress shoes\n- Millie’s exact mouse identity\n- Millie’s warm beige fur\n- Millie’s large round ears\n- Millie’s pink polka-dot bow\n- Millie’s pink polka-dot dress\n- Millie’s pink shoes\n- Theo/Millie size relationship\n- golden harmonica design\n- green park bench\n- CITY PARK lamp post\n- Sunset Music Festival poster\n- blue and yellow birds\n- squirrel\n- pigeons\n- friendly violinist\n- flowerbeds\n- pastel city skyline\n- park lighting\n- entire Part 1 story progression\n\nGenerate a FINISHED cinematic animation from the storyboard.\n\nDO NOT animate the storyboard page itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- numbered panels\n- timestamps\n- handwritten captions\n- production notes\n- reference drawings\n- title/header\n- page number\n- white storyboard-paper background\n- color swatches\n- annotation boxes\n- subtitles\n- speech bubbles\n\nTranslate the 15 storyboard panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT create 15 rapid one-second cuts.\n\nThe cinematic progression should feel like:\n\npeaceful park introduction\n→ Theo’s hidden nervousness\n→ harmonica reveal\n→ Millie understands\n→ big idea\n→ secret mission\n→ helper recruitment\n→ Theo remains unaware\n→ Millie gives the signal\n→ cliffhanger.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\nPART 1 — 0:00 to 0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium polished 2D storybook animation\n\nGenre:\nheartwarming city-park comedy / friendship story\n\nTone:\n\n- charming\n- stylish\n- sweet\n- warm\n- lightly funny\n- emotionally sincere\n- wholesome\n\nAudience:\nfamily-friendly\n\nDialogue:\nnative synchronized English dialogue\n\nAudio:\nnative character dialogue + park ambience + harmonica motif + original cinematic music\n\nCORE PART 1 IDEA:\n\nTheo looks like the coolest gentleman in the park...\n\nbut he is secretly afraid to perform.\n\nMillie notices.\n\nAnd decides to do something about it.\n\n==================================================\nPART 1 STORY LIMIT\n==================================================\n\nPART 1 ONLY.\n\nSTART:\n\nTheo and Millie sit together on the green City Park bench during a beautiful late afternoon.\n\nEND:\n\nMillie raises one tiny paw and silently signals her secret helper team.\n\nTheo is still turned away and has NO IDEA what is about to happen.\n\nCUT TO BLACK.\n\nDO NOT show:\n\n- finished surprise stage\n- Theo receiving the harmonica from Millie\n- Theo accepting the surprise\n- evening festival performance\n- Theo actually playing the harmonica publicly\n- audience gathering\n- Millie dancing with Theo\n- applause\n- emotional final bench scene\n\nSave ALL payoff for PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — THEO\n==================================================\n\nMatch the uploaded storyboard EXACTLY.\n\nTHEO — THE GENTLEMAN CAT\n\nAnthropomorphic adult black cat gentleman.\n\nAppearance:\n\n- sleek charcoal-black / black fur\n- lighter gray muzzle area\n- small rounded black nose\n- long feline whiskers\n- triangular black ears with subtle warm inner-ear tones\n- expressive golden-yellow eyes\n- elegant eyelids\n- soft cheek fur\n- long black tail\n- tall, lean, sophisticated proportions\n\nOUTFIT:\n\n- fitted charcoal-black suit jacket\n- matching dark trousers\n- cream / ivory dress shirt\n- bright pink necktie\n- matching pink pocket square\n- black fedora\n- small feather tucked into fedora band\n- black-and-cream polished dress shoes\n- subtle gold waist / pocket chain\n\nDO NOT change:\n\n- suit\n- hat\n- tie color\n- fur color\n- eye color\n- proportions\n- facial structure\n\nPERSONALITY:\n\nOutwardly:\n\n- stylish\n- smooth\n- controlled\n- gentlemanly\n- confident-looking\n\nInternally:\n\n- shy\n- thoughtful\n- nervous about performing\n- emotionally guarded\n- sensitive\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nrelaxed gentleman exterior\n\n3–6 sec:\nfestival poster triggers nervousness\n\n6–8 sec:\nbrief harmonica hesitation\n\n8–10 sec:\nquietly withdrawn\n\n10–15 sec:\nstill unaware that Millie is organizing help\n\nCRITICAL:\n\nTheo should never become melodramatically sad.\n\nHis emotion is subtle:\n\na small sigh,\nlowered eyelids,\nhesitation,\nlooking away,\nhiding the harmonica.\n\n==================================================\nCHARACTER IDENTITY LOCK — MILLIE\n==================================================\n\nMatch storyboard EXACTLY.\n\nMILLIE — THE LITTLE MOUSE\n\nAppearance:\n\n- tiny anthropomorphic mouse girl\n- warm beige / light taupe fur\n- very large rounded mouse ears\n- soft pink inner ears\n- large expressive dark eyes\n- long eyelashes\n- tiny black nose\n- fine whiskers\n- thin mouse tail\n- petite cute proportions\n\nOUTFIT:\n\n- large pink bow with white polka dots\n- sleeveless pink dress\n- white polka-dot pattern\n- gently flared skirt\n- pink Mary Jane-style shoes\n\nScale:\n\nMillie is considerably smaller than Theo.\n\nMaintain this exact scale in every shot.\n\nPERSONALITY:\n\n- cheerful\n- observant\n- imaginative\n- compassionate\n- enthusiastic\n- playful\n- determined\n- quietly clever\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nhappy admiration\n\n3–6 sec:\nnotices Theo is distracted\n\n6–8 sec:\nrealizes his secret\n\n8–10 sec:\nbig idea\n\n10–14 sec:\nenergetic secret organizer\n\n14–15 sec:\ntiny confident conductor / mission leader\n\n==================================================\nKEY PROP LOCK — GOLDEN HARMONICA\n==================================================\n\nMatch storyboard exactly.\n\nAppearance:\n\n- small vintage golden harmonica\n- polished brass/gold finish\n- subtle engraved lines\n- compact rectangular shape\n- sentimental appearance\n\nIt belongs to Theo.\n\nPART 1 use:\n\nTheo carefully pulls it halfway from inside his suit.\n\nHe looks at it.\n\nHesitates.\n\nThen quietly holds / hides it again.\n\nIMPORTANT:\n\nMillie may SEE the harmonica.\n\nShe does NOT take it during Part 1.\n\nIt remains Theo’s possession.\n\nNo public harmonica playing yet.\n\n==================================================\nENVIRONMENT LOCK — CITY PARK\n==================================================\n\nMatch storyboard closely.\n\nBeautiful classic urban park in warm late afternoon.\n\nInclude:\n\n- green wooden park bench\n- black iron armrests / frame\n- black vintage lamp post\n- readable small CITY PARK sign\n- black iron fence\n- large leafy green trees\n- flowerbeds with pink, yellow and white flowers\n- park path\n- distant fountain area\n- pastel city skyline\n- soft blue sky\n- fluffy white clouds\n\nArchitecture:\n\nsoft pastel beige,\npale yellow,\npink,\nmuted city colors.\n\nNo modern advertising.\n\nNo vehicles dominating the composition.\n\nNo gloomy weather.\n\n==================================================\nSUNSET MUSIC FESTIVAL POSTER LOCK\n==================================================\n\nMatch storyboard.\n\nA handmade cheerful poster located near the park.\n\nReadable central wording:\n\n“SUNSET MUSIC FESTIVAL TONIGHT!”\n\nDecoration:\n\n- music notes\n- small stars\n- colorful bunting\n- hand-painted festival style\n\nThe poster is an important story clue.\n\nDo not make it look like a commercial advertisement.\n\n==================================================\nTINY HELPER LOCK\n==================================================\n\nUse the same helper designs shown in the storyboard.\n\nHelpers include:\n\nBLUE BIRD:\nsmall round blue songbird,\norange/yellow beak.\n\nYELLOW BIRD:\nsmall warm yellow songbird.\n\nSQUIRREL:\nfriendly brown squirrel,\nlarge fluffy tail.\n\nPIGEONS:\nsmall gray city pigeons.\n\nVIOLINIST:\nfriendly adult park musician,\nsimple neutral clothing,\nholding violin and bow.\n\nThe helpers are friendly and supportive.\n\nNo dialogue required from animals.\n\nThe violinist can acknowledge Millie with a smile/nod.\n\n==================================================\nPART 1 SURPRISE LOGIC\n==================================================\n\nMillie must keep the surprise hidden from Theo.\n\nTheo should NOT notice the helpers gathering.\n\nVisual logic:\n\nTheo looks toward festival poster\n→ reveals harmonica\n→ hesitates\n→ Millie understands\n\nMillie gets idea\n→ quietly leaves bench\n\nMillie recruits birds\n→ squirrel\n→ pigeons\n→ violinist\n\nhelpers understand her plan\n→ move into position\n\nTheo still believes he is alone with his problem\n\nMillie gives signal\n\nCUT.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:02]\nA STYLISH AFTERNOON IN CITY PARK\n\nOpen with a beautiful wide cinematic view matching Storyboard Panel 1.\n\nBright blue sky.\n\nSoft white clouds.\n\nGreen trees frame the image.\n\nPastel city buildings rise behind the park.\n\nFlowers line the walkway.\n\nThe camera glides gently toward the familiar green bench.\n\nTHEO sits elegantly on the left side.\n\nHis legs are crossed.\n\nOne elbow rests against the bench.\n\nHis fedora sits perfectly angled.\n\nMILLIE sits beside him on the right.\n\nShe faces Theo with an excited smile.\n\nTheo subtly adjusts his bright pink tie.\n\nMillie beams at him.\n\nDialogue:\n\nMillie:\n<d>[English] You look very fancy today.</d>\n\nTheo gives a smooth little smile.\n\nTheo:\n<d>[English] Naturally.</d>\n\nHe lightly touches the brim of his fedora.\n\nMillie giggles.\n\nCAMERA:\n\nwide park establishing\n→ gentle dolly toward bench\n→ medium two-shot.\n\nAUDIO:\n\n- soft park birds\n- distant city ambience\n- leaves moving\n- quiet footsteps in distance\n- faint fountain\n- flower/grass breeze\n\nMUSIC:\n\nlight original jazz-storybook theme:\n\n- upright bass\n- brushed percussion\n- clarinet\n- gentle piano\n- pizzicato strings\n\nThe scene should initially feel carefree.\n\n==================================================\n[SHOT 2 — 0:02–0:03.8]\nSOMETHING FEELS OFF\n\nMove into a closer Theo/Millie composition.\n\nTheo maintains his confident smile.\n\nThen his golden eyes drift away from Millie.\n\nRack focus toward the nearby festival poster.\n\n“SUNSET MUSIC FESTIVAL TONIGHT!”\n\nTheo’s expression changes very subtly.\n\nHis eyelids lower.\n\nHis smile fades just a little.\n\nMillie continues talking for a beat...\n\nthen notices he is no longer listening.\n\nMillie:\n\n<d>[English] Theo?</d>\n\nTheo quickly looks back.\n\nTheo:\n\n<d>[English] Hmm?</d>\n\nHe forces his small gentleman smile again.\n\nCamera:\n\nTheo close-up\n→ eye direction\n→ rack focus to poster\n→ return to Theo.\n\nMusic removes some playful notes.\n\nA small unresolved piano chord enters.\n\n==================================================\n[SHOT 3 — 0:03.8–0:05.8]\nTHE GOLDEN HARMONICA\n\nClose on Theo’s suit.\n\nHis black paw slowly reaches inside his jacket.\n\nHe pulls out the small GOLDEN HARMONICA.\n\nWarm sunlight glints across its brass surface.\n\nTheo looks at it.\n\nFor a tiny moment:\n\nhe imagines performing.\n\nHis ears lift slightly.\n\nThen distant festival music / rehearsal sound reaches him.\n\nTheo becomes nervous.\n\nHis ears lower slightly.\n\nHe begins putting the harmonica away.\n\nMillie sees everything.\n\nDialogue:\n\nMillie, softly:\n<d>[English] You were going to play tonight?</d>\n\nTheo pauses.\n\nHe avoids her gaze.\n\nTheo:\n\n<d>[English] I was thinking about it.</d>\n\nHe slips the harmonica carefully back toward his pocket.\n\nTheo, quieter:\n\n<d>[English] Maybe another time.</d>\n\nIMPORTANT:\n\nNo big sad performance.\n\nKeep it restrained and sincere.\n\n==================================================\n[SHOT 4 — 0:05.8–0:07.4]\nMILLIE UNDERSTANDS\n\nClose-up on Millie.\n\nShe looks:\n\nTheo\n→ harmonica pocket\n→ festival poster\n→ Theo again.\n\nHer happy expression becomes concerned.\n\nThen thoughtful.\n\nOne finger touches her chin.\n\nTheo looks away toward the city, trying to seem casual.\n\nMillie’s eyes suddenly brighten.\n\nTiny visual idea sparkle beside her bow.\n\nShe smiles.\n\nDialogue:\n\nMillie, almost whispering:\n<d>[English] Hmm...</d>\n\nTiny pause.\n\n<d>[English] I have an idea.</d>\n\nTheo is still looking away.\n\nHe does NOT hear her clearly.\n\nMusic introduces a mischievous tiny plucked-string motif.\n\n==================================================\n[SHOT 5 — 0:07.4–0:09.0]\nTHE SECRET MISSION BEGINS\n\nTheo remains seated on the bench.\n\nHis chin rests lightly against one paw.\n\nHe looks toward the distant festival area.\n\nBehind him:\n\nMillie quietly climbs down from the bench.\n\nShe puts one finger to her lips:\n\n“shhh.”\n\nShe tiptoes past the flowers.\n\nHer polka-dot skirt bounces gently.\n\nHer tail follows naturally.\n\nTheo remains unaware.\n\nCAMERA:\n\nmedium Theo foreground\n→ Millie slipping down behind him\n→ low tiny tracking shot following Millie.\n\nMillie whispers to herself:\n\n<d>[English] Quietly...</d>\n\nSound:\n\n- tiny shoe taps\n- dress rustle\n- leaves\n- park ambience\n\n==================================================\n[SHOT 6 — 0:09.0–0:11.0]\nFIRST, THE TINY HELPERS\n\nMillie reaches a tree.\n\nA BLUE BIRD and YELLOW BIRD sit together on a branch.\n\nShe waves them closer.\n\nMillie gestures toward Theo.\n\nThen mimics:\n\nplaying music.\n\nThen points toward the festival poster.\n\nThe birds tilt their heads.\n\nMillie clasps her paws:\n\nplease?\n\nThe blue bird chirps.\n\nThe yellow bird chirps back.\n\nBoth nod enthusiastically.\n\nMillie smiles.\n\nThen she turns and finds the friendly SQUIRREL in the flowerbed.\n\nShe whispers:\n\n<d>[English] I need your help too.</d>\n\nThe squirrel looks toward Theo.\n\nThen back to Millie.\n\nHe salutes with one tiny paw.\n\nUse cute visual comedy.\n\nCAMERA:\n\nMillie with birds\n→ bird reaction close-ups\n→ quick flowing transition to squirrel\n→ Millie delighted reaction.\n\nNo frantic cutting.\n\n==================================================\n[SHOT 7 — 0:11.0–0:13.2]\nTHE PLAN GETS BIGGER\n\nCreate one flowing helper-gathering sequence.\n\nMillie walks briskly toward the fountain.\n\nTwo / three pigeons turn toward her.\n\nShe makes a tiny directing gesture.\n\nThey follow.\n\nThen she notices the friendly VIOLINIST nearby.\n\nHe is casually practicing.\n\nMillie approaches.\n\nShe points:\n\ntoward Theo,\ntoward harmonica pocket area,\ntoward festival poster.\n\nThen presses her paws together hopefully.\n\nThe violinist understands.\n\nHe looks toward Theo.\n\nSmiles warmly.\n\nNods.\n\nMillie turns around.\n\nBehind her:\n\nbirds,\nsquirrel,\npigeons,\nviolinist\n\nare now a tiny secret team.\n\nMillie’s eyes widen with happy surprise at how many helpers she has gathered.\n\nDialogue:\n\nMillie, whispered:\n<d>[English] Okay... this got bigger.</d>\n\nSmall humorous music accent.\n\n==================================================\n[SHOT 8 — 0:13.2–0:15.0]\nTHE SURPRISE IS READY — CLIFFHANGER\n\nIntercut emotional contrast smoothly.\n\nTHEO:\n\nstill sits on the bench.\n\nHis posture is elegant but slightly slumped.\n\nHe looks down toward his pocket.\n\nOne paw rests over the hidden harmonica.\n\nHe quietly sighs.\n\nTheo murmurs:\n\n<d>[English] Maybe next year.</d>\n\nDO NOT make him cry.\n\nCUT / FLOW BEHIND HIM:\n\nMillie and the helpers quietly move into position.\n\n- birds perch above\n- squirrel waits beside flowers\n- pigeons line up near the path\n- violinist stands ready\n- everyone watches Millie\n\nMillie climbs onto the low base of the CITY PARK lamp post / safe bench edge.\n\nShe becomes a tiny conductor.\n\nShe looks at every helper.\n\nThey become still.\n\nMillie glances toward Theo.\n\nHe is turned away.\n\nPerfect.\n\nMillie raises one tiny paw dramatically.\n\nHold.\n\nHer expression:\n\ndetermined\n+\nexcited\n+\nloving.\n\nMusic reduces to a playful suspense pulse.\n\nNo one starts the surprise yet.\n\nMillie whispers:\n\n<d>[English] Ready?</d>\n\nThe blue bird gives a tiny nod.\n\nThe squirrel nods.\n\nThe violinist raises his bow slightly.\n\nMillie begins lowering her paw—\n\nCUT TO BLACK BEFORE THE SURPRISE STARTS.\n\nEND PART 1.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nThe soundscape should feel:\n\n- warm\n- natural\n- whimsical\n- cinematic\n- heartfelt\n- lightly jazzy\n\n==================================================\nTHEO VOICE\n==================================================\n\nAdult male voice.\n\nQualities:\n\n- warm\n- smooth\n- refined\n- understated\n- charming\n- gentle\n- slightly jazzy/vintage personality\n- never threatening\n\nHis confidence early should sound natural.\n\nHis nervousness should be subtle.\n\nImportant emotional dialogue:\n\n<d>[English] I was thinking about it.</d>\n\nThen:\n\n<d>[English] Maybe another time.</d>\n\nFinal:\n\n<d>[English] Maybe next year.</d>\n\nThese should sound reflective, not dramatically depressed.\n\n==================================================\nMILLIE VOICE\n==================================================\n\nYoung feminine mouse voice.\n\nQualities:\n\n- bright\n- warm\n- cheerful\n- energetic\n- sincere\n- expressive\n- slightly higher pitch than Theo\n\nDo NOT make it shrill.\n\nHer emotional progression:\n\nhappy\n→ curious\n→ concerned\n→ thoughtful\n→ inspired\n→ playful secret organizer.\n\nImportant:\n\n<d>[English] I have an idea.</d>\n\nshould be soft but excited.\n\nFinal:\n\n<d>[English] Ready?</d>\n\nshould be whispered with delighted anticipation.\n\n==================================================\nHELPER AUDIO\n==================================================\n\nBIRDS:\n\nsoft cheerful chirps.\n\nSQUIRREL:\n\ntiny harmless squeaks / chatter.\n\nPIGEONS:\n\ngentle coos.\n\nVIOLINIST:\n\nvery short natural violin warm-up tones.\n\nNo speaking dialogue required.\n\n==================================================\nPARK AMBIENCE\n==================================================\n\nUse:\n\n- soft breeze\n- leaves\n- distant fountain\n- birds\n- subtle city ambience\n- occasional distant footstep\n- light iron-fence / bench environmental resonance\n- flowerbed insects if subtle\n\nNo loud traffic.\n\nNo horns.\n\nNo urban noise dominating the story.\n\n==================================================\nHARMONICA SOUND\n==================================================\n\nIMPORTANT:\n\nTheo does NOT perform publicly during Part 1.\n\nWhen the harmonica is revealed:\n\nuse only a very small metallic handling sound.\n\nOptionally:\n\na single extremely faint memory-like harmonica note can appear NON-DIEGETICALLY in the score.\n\nDo NOT show Theo playing it.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL warm jazz-storybook score.\n\nInstrumentation:\n\n- soft upright bass\n- brushed snare\n- gentle piano\n- clarinet\n- muted trumpet used very lightly\n- pizzicato strings\n- acoustic guitar\n- subtle vibraphone\n- tiny glockenspiel for Millie’s idea\n\nMUSIC ARC:\n\n0:00–0:02\ncharming city-park jazz.\n\n0:02–0:04\nslight uncertainty as Theo looks at poster.\n\n0:04–0:06\nsoft emotional harmonica motif.\n\n0:06–0:08\ngentle emotional pause.\n\n0:08–0:10\nMillie’s playful idea motif.\n\n0:10–0:13\nlight sneaky/helper rhythm.\n\n0:13–0:15\nwarm suspense build.\n\nCUT music just before surprise begins.\n\nNo sad orchestral melodrama.\n\nNo huge trailer music.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 1 camera should evolve from:\n\nrelaxed\n→ intimate\n→ emotionally observant\n→ playful secret-mission energy.\n\nUse:\n\n- wide park establishment\n- gentle dolly\n- medium two-shots\n- close facial acting\n- rack focus\n- prop insert\n- low tiny-character tracking with Millie\n- helper reaction close-ups\n- shallow depth emphasis\n- final wide bench + helper composition\n\nAvoid:\n\n- handheld shake\n- crash zoom\n- frantic editing\n- excessive whip pans\n- extreme fisheye\n- 360 camera\n- action-style camera\n\nThe overall camera should remain graceful.\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished hand-painted 2D storybook/cartoon animation.\n\nMatch the uploaded storyboard in spirit.\n\nUse:\n\n- clean expressive ink contours\n- watercolor / gouache-inspired fills\n- subtle textured shading\n- beautiful park background painting\n- soft sunlight\n- stylized animal fur\n- expressive eyes\n- controlled squash-and-stretch\n- cinematic atmospheric depth\n- elegant vintage styling\n- subtle film grain / paper-paint texture if appropriate\n\nThe animation should feel like:\n\nTHE STORYBOARD ART HAS COME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealistic animals\n- live-action animals\n- generic 3D CGI\n- horror\n- hyperreal fur simulation\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary palette:\n\nTHEO:\n- charcoal black\n- warm gray muzzle\n- cream shirt\n- pink tie\n- pink pocket square\n- golden-yellow eyes\n\nMILLIE:\n- beige / taupe\n- soft pink\n- white polka dots\n- warm ear pink\n\nPARK:\n- spring green\n- sky blue\n- pastel city beige/pink/yellow\n- flower pink/yellow/white\n- black lamp/fence accents\n\nLighting:\n\nwarm late afternoon sunlight.\n\nUse:\n\n- soft golden side light\n- gentle cool sky fill\n- natural shadow under bench\n- bright readable faces\n\nBy final shot:\n\nintroduce slightly warmer golden tones to foreshadow the sunset festival.\n\n==================================================\nMOTION RULES — THEO\n==================================================\n\nTheo’s movements are refined.\n\nUse:\n\n- slow tie adjustment\n- relaxed leg crossing\n- tiny fedora touch\n- subtle eye movement\n- controlled harmonica reveal\n- small ear changes\n- gentle tail movement\n- restrained sigh\n\nAvoid exaggerated cartoon nervousness.\n\nThe emotion is in:\n\neyes,\nears,\nhands,\nposture.\n\n==================================================\nMOTION RULES — MILLIE\n==================================================\n\nMillie is energetic but believable.\n\nUse:\n\n- small enthusiastic hand gestures\n- light bouncing steps\n- gentle skirt secondary motion\n- tiny tiptoeing\n- expressive ear movement\n- quick turns\n- thoughtful chin-touch\n- subtle “idea” pose\n- conductor-like final signal\n\nHer bow must stay attached and maintain exact shape.\n\n==================================================\nSIZE / SCALE RULES\n==================================================\n\nTheo is substantially larger than Millie.\n\nMaintain consistent relative scale.\n\nWhen seated:\n\nMillie occupies a much smaller portion of the bench.\n\nTheo’s harmonica is comfortable in Theo’s paw but looks large relative to Millie.\n\nBirds and squirrel should be approximately around Millie’s scale or somewhat smaller.\n\nDo not make Millie suddenly human-sized.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Theo.\nONE Millie.\n\nSame:\n\n- Theo face\n- Theo hat\n- Theo suit\n- Theo pink tie\n- Theo eye color\n- Millie face\n- Millie bow\n- Millie dress\n- Millie shoes\n- relative scale\n- green bench\n- CITY PARK lamp\n- festival poster\n- city skyline\n- flowerbeds\n- golden harmonica\n\nHelper characters should remain visually consistent once introduced.\n\nTheo remains around the bench through Part 1.\n\nMillie physically moves through the park to gather helpers.\n\nNo teleportation.\n\n==================================================\nEMOTIONAL PERFORMANCE PRIORITY\n==================================================\n\nThis story works because Theo’s problem is SMALL and HUMAN:\n\nHe can play music.\n\nHe simply lacks confidence to share it publicly.\n\nMillie does NOT mock him.\n\nShe does NOT steal the harmonica.\n\nShe does NOT force him to perform.\n\nShe notices his hesitation and quietly creates support.\n\nPart 1 should make the audience think:\n\n“Oh... she’s going to help him.”\n\nwithout showing HOW yet.\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\n15 seconds is short.\n\nUse natural pacing.\n\nIf some dialogue must be removed, PRIORITIZE:\n\n1.\nMillie:\n<d>[English] Theo?</d>\n\n2.\nMillie:\n<d>[English] You were going to play tonight?</d>\n\n3.\nTheo:\n<d>[English] Maybe another time.</d>\n\n4.\nMillie:\n<d>[English] I have an idea.</d>\n\n5.\nTheo:\n<d>[English] Maybe next year.</d>\n\n6.\nMillie:\n<d>[English] Ready?</d>\n\nSecondary lines may be omitted rather than spoken unnaturally fast.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nKeep all action safe and wholesome.\n\nNo violence.\n\nNo threatening behavior.\n\nNo weapons.\n\nNo fighting.\n\nNo dangerous park activity.\n\nNo romantic kissing.\n\nNo sexual behavior.\n\nNo disturbing imagery.\n\nNo bullying.\n\nTheo’s suit and fedora represent an elegant gentleman musician only.\n\nNo crime/gangster framing.\n\n==================================================\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086426143612440739",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T12:15:31.217Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086426143612440739.mp4",
    "posterUrl": "/posters/x/2086426143612440739.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "dance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n\"[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for:\n\n“THE GENTLEMAN CAT AND THE LITTLE MOUSE’S BIG SURPRISE”\n\nas the STRICT PRIMARY visual reference for the finished video.\n\nThe uploaded storyboard controls:\n\n- Theo’s exact face and body identity\n- Theo’s black fur\n- Theo’s yellow-gold eyes\n- Theo’s black fedora and feather detail\n- Theo’s charcoal gentleman suit\n- Theo’s cream shirt\n- Theo’s bright pink tie and pocket square\n- Theo’s black tail\n- Theo’s black-and-cream dress shoes\n- Millie’s exact mouse identity\n- Millie’s warm beige fur\n- Millie’s large round ears\n- Millie’s pink polka-dot bow\n- Millie’s pink polka-dot dress\n- Millie’s pink shoes\n- Theo/Millie size relationship\n- golden harmonica design\n- green park bench\n- CITY PARK lamp post\n- Sunset Music Festival poster\n- blue and yellow birds\n- squirrel\n- pigeons\n- friendly violinist\n- flowerbeds\n- pastel city skyline\n- park lighting\n- entire Part 1 story progression\n\nGenerate a FINISHED cinematic animation from the storyboard.\n\nDO NOT animate the storyboard page itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- numbered panels\n- timestamps\n- handwritten captions\n- production notes\n- reference drawings\n- title/header\n- page number\n- white storyboard-paper background\n- color swatches\n- annotation boxes\n- subtitles\n- speech bubbles\n\nTranslate the 15 storyboard panels into approximately:\n\n# 8 COHERENT CINEMATIC SHOTS\n\nDo NOT create 15 rapid one-second cuts.\n\nThe cinematic progression should feel like:\n\npeaceful park introduction\n→ Theo’s hidden nervousness\n→ harmonica reveal\n→ Millie understands\n→ big idea\n→ secret mission\n→ helper recruitment\n→ Theo remains unaware\n→ Millie gives the signal\n→ cliffhanger.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\nPART 1 — 0:00 to 0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium polished 2D storybook animation\n\nGenre:\nheartwarming city-park comedy / friendship story\n\nTone:\n\n- charming\n- stylish\n- sweet\n- warm\n- lightly funny\n- emotionally sincere\n- wholesome\n\nAudience:\nfamily-friendly\n\nDialogue:\nnative synchronized English dialogue\n\nAudio:\nnative character dialogue + park ambience + harmonica motif + original cinematic music\n\nCORE PART 1 IDEA:\n\nTheo looks like the coolest gentleman in the park...\n\nbut he is secretly afraid to perform.\n\nMillie notices.\n\nAnd decides to do something about it.\n\n==================================================\nPART 1 STORY LIMIT\n==================================================\n\nPART 1 ONLY.\n\nSTART:\n\nTheo and Millie sit together on the green City Park bench during a beautiful late afternoon.\n\nEND:\n\nMillie raises one tiny paw and silently signals her secret helper team.\n\nTheo is still turned away and has NO IDEA what is about to happen.\n\nCUT TO BLACK.\n\nDO NOT show:\n\n- finished surprise stage\n- Theo receiving the harmonica from Millie\n- Theo accepting the surprise\n- evening festival performance\n- Theo actually playing the harmonica publicly\n- audience gathering\n- Millie dancing with Theo\n- applause\n- emotional final bench scene\n\nSave ALL payoff for PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — THEO\n==================================================\n\nMatch the uploaded storyboard EXACTLY.\n\nTHEO — THE GENTLEMAN CAT\n\nAnthropomorphic adult black cat gentleman.\n\nAppearance:\n\n- sleek charcoal-black / black fur\n- lighter gray muzzle area\n- small rounded black nose\n- long feline whiskers\n- triangular black ears with subtle warm inner-ear tones\n- expressive golden-yellow eyes\n- elegant eyelids\n- soft cheek fur\n- long black tail\n- tall, lean, sophisticated proportions\n\nOUTFIT:\n\n- fitted charcoal-black suit jacket\n- matching dark trousers\n- cream / ivory dress shirt\n- bright pink necktie\n- matching pink pocket square\n- black fedora\n- small feather tucked into fedora band\n- black-and-cream polished dress shoes\n- subtle gold waist / pocket chain\n\nDO NOT change:\n\n- suit\n- hat\n- tie color\n- fur color\n- eye color\n- proportions\n- facial structure\n\nPERSONALITY:\n\nOutwardly:\n\n- stylish\n- smooth\n- controlled\n- gentlemanly\n- confident-looking\n\nInternally:\n\n- shy\n- thoughtful\n- nervous about performing\n- emotionally guarded\n- sensitive\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nrelaxed gentleman exterior\n\n3–6 sec:\nfestival poster triggers nervousness\n\n6–8 sec:\nbrief harmonica hesitation\n\n8–10 sec:\nquietly withdrawn\n\n10–15 sec:\nstill unaware that Millie is organizing help\n\nCRITICAL:\n\nTheo should never become melodramatically sad.\n\nHis emotion is subtle:\n\na small sigh,\nlowered eyelids,\nhesitation,\nlooking away,\nhiding the harmonica.\n\n==================================================\nCHARACTER IDENTITY LOCK — MILLIE\n==================================================\n\nMatch storyboard EXACTLY.\n\nMILLIE — THE LITTLE MOUSE\n\nAppearance:\n\n- tiny anthropomorphic mouse girl\n- warm beige / light taupe fur\n- very large rounded mouse ears\n- soft pink inner ears\n- large expressive dark eyes\n- long eyelashes\n- tiny black nose\n- fine whiskers\n- thin mouse tail\n- petite cute proportions\n\nOUTFIT:\n\n- large pink bow with white polka dots\n- sleeveless pink dress\n- white polka-dot pattern\n- gently flared skirt\n- pink Mary Jane-style shoes\n\nScale:\n\nMillie is considerably smaller than Theo.\n\nMaintain this exact scale in every shot.\n\nPERSONALITY:\n\n- cheerful\n- observant\n- imaginative\n- compassionate\n- enthusiastic\n- playful\n- determined\n- quietly clever\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nhappy admiration\n\n3–6 sec:\nnotices Theo is distracted\n\n6–8 sec:\nrealizes his secret\n\n8–10 sec:\nbig idea\n\n10–14 sec:\nenergetic secret organizer\n\n14–15 sec:\ntiny confident conductor / mission leader\n\n==================================================\nKEY PROP LOCK — GOLDEN HARMONICA\n==================================================\n\nMatch storyboard exactly.\n\nAppearance:\n\n- small vintage golden harmonica\n- polished brass/gold finish\n- subtle engraved lines\n- compact rectangular shape\n- sentimental appearance\n\nIt belongs to Theo.\n\nPART 1 use:\n\nTheo carefully pulls it halfway from inside his suit.\n\nHe looks at it.\n\nHesitates.\n\nThen quietly holds / hides it again.\n\nIMPORTANT:\n\nMillie may SEE the harmonica.\n\nShe does NOT take it during Part 1.\n\nIt remains Theo’s possession.\n\nNo public harmonica playing yet.\n\n==================================================\nENVIRONMENT LOCK — CITY PARK\n==================================================\n\nMatch storyboard closely.\n\nBeautiful classic urban park in warm late afternoon.\n\nInclude:\n\n- green wooden park bench\n- black iron armrests / frame\n- black vintage lamp post\n- readable small CITY PARK sign\n- black iron fence\n- large leafy green trees\n- flowerbeds with pink, yellow and white flowers\n- park path\n- distant fountain area\n- pastel city skyline\n- soft blue sky\n- fluffy white clouds\n\nArchitecture:\n\nsoft pastel beige,\npale yellow,\npink,\nmuted city colors.\n\nNo modern advertising.\n\nNo vehicles dominating the composition.\n\nNo gloomy weather.\n\n==================================================\nSUNSET MUSIC FESTIVAL POSTER LOCK\n==================================================\n\nMatch storyboard.\n\nA handmade cheerful poster located near the park.\n\nReadable central wording:\n\n“SUNSET MUSIC FESTIVAL TONIGHT!”\n\nDecoration:\n\n- music notes\n- small stars\n- colorful bunting\n- hand-painted festival style\n\nThe poster is an important story clue.\n\nDo not make it look like a commercial advertisement.\n\n==================================================\nTINY HELPER LOCK\n==================================================\n\nUse the same helper designs shown in the storyboard.\n\nHelpers include:\n\nBLUE BIRD:\nsmall round blue songbird,\norange/yellow beak.\n\nYELLOW BIRD:\nsmall warm yellow songbird.\n\nSQUIRREL:\nfriendly brown squirrel,\nlarge fluffy tail.\n\nPIGEONS:\nsmall gray city pigeons.\n\nVIOLINIST:\nfriendly adult park musician,\nsimple neutral clothing,\nholding violin and bow.\n\nThe helpers are friendly and supportive.\n\nNo dialogue required from animals.\n\nThe violinist can acknowledge Millie with a smile/nod.\n\n==================================================\nPART 1 SURPRISE LOGIC\n==================================================\n\nMillie must keep the surprise hidden from Theo.\n\nTheo should NOT notice the helpers gathering.\n\nVisual logic:\n\nTheo looks toward festival poster\n→ reveals harmonica\n→ hesitates\n→ Millie understands\n\nMillie gets idea\n→ quietly leaves bench\n\nMillie recruits birds\n→ squirrel\n→ pigeons\n→ violinist\n\nhelpers understand her plan\n→ move into position\n\nTheo still believes he is alone with his problem\n\nMillie gives signal\n\nCUT.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:02]\nA STYLISH AFTERNOON IN CITY PARK\n\nOpen with a beautiful wide cinematic view matching Storyboard Panel 1.\n\nBright blue sky.\n\nSoft white clouds.\n\nGreen trees frame the image.\n\nPastel city buildings rise behind the park.\n\nFlowers line the walkway.\n\nThe camera glides gently toward the familiar green bench.\n\nTHEO sits elegantly on the left side.\n\nHis legs are crossed.\n\nOne elbow rests against the bench.\n\nHis fedora sits perfectly angled.\n\nMILLIE sits beside him on the right.\n\nShe faces Theo with an excited smile.\n\nTheo subtly adjusts his bright pink tie.\n\nMillie beams at him.\n\nDialogue:\n\nMillie:\n<d>[English] You look very fancy today.</d>\n\nTheo gives a smooth little smile.\n\nTheo:\n<d>[English] Naturally.</d>\n\nHe lightly touches the brim of his fedora.\n\nMillie giggles.\n\nCAMERA:\n\nwide park establishing\n→ gentle dolly toward bench\n→ medium two-shot.\n\nAUDIO:\n\n- soft park birds\n- distant city ambience\n- leaves moving\n- quiet footsteps in distance\n- faint fountain\n- flower/grass breeze\n\nMUSIC:\n\nlight original jazz-storybook theme:\n\n- upright bass\n- brushed percussion\n- clarinet\n- gentle piano\n- pizzicato strings\n\nThe scene should initially feel carefree.\n\n==================================================\n[SHOT 2 — 0:02–0:03.8]\nSOMETHING FEELS OFF\n\nMove into a closer Theo/Millie composition.\n\nTheo maintains his confident smile.\n\nThen his golden eyes drift away from Millie.\n\nRack focus toward the nearby festival poster.\n\n“SUNSET MUSIC FESTIVAL TONIGHT!”\n\nTheo’s expression changes very subtly.\n\nHis eyelids lower.\n\nHis smile fades just a little.\n\nMillie continues talking for a beat...\n\nthen notices he is no longer listening.\n\nMillie:\n\n<d>[English] Theo?</d>\n\nTheo quickly looks back.\n\nTheo:\n\n<d>[English] Hmm?</d>\n\nHe forces his small gentleman smile again.\n\nCamera:\n\nTheo close-up\n→ eye direction\n→ rack focus to poster\n→ return to Theo.\n\nMusic removes some playful notes.\n\nA small unresolved piano chord enters.\n\n==================================================\n[SHOT 3 — 0:03.8–0:05.8]\nTHE GOLDEN HARMONICA\n\nClose on Theo’s suit.\n\nHis black paw slowly reaches inside his jacket.\n\nHe pulls out the small GOLDEN HARMONICA.\n\nWarm sunlight glints across its brass surface.\n\nTheo looks at it.\n\nFor a tiny moment:\n\nhe imagines performing.\n\nHis ears lift slightly.\n\nThen distant festival music / rehearsal sound reaches him.\n\nTheo becomes nervous.\n\nHis ears lower slightly.\n\nHe begins putting the harmonica away.\n\nMillie sees everything.\n\nDialogue:\n\nMillie, softly:\n<d>[English] You were going to play tonight?</d>\n\nTheo pauses.\n\nHe avoids her gaze.\n\nTheo:\n\n<d>[English] I was thinking about it.</d>\n\nHe slips the harmonica carefully back toward his pocket.\n\nTheo, quieter:\n\n<d>[English] Maybe another time.</d>\n\nIMPORTANT:\n\nNo big sad performance.\n\nKeep it restrained and sincere.\n\n==================================================\n[SHOT 4 — 0:05.8–0:07.4]\nMILLIE UNDERSTANDS\n\nClose-up on Millie.\n\nShe looks:\n\nTheo\n→ harmonica pocket\n→ festival poster\n→ Theo again.\n\nHer happy expression becomes concerned.\n\nThen thoughtful.\n\nOne finger touches her chin.\n\nTheo looks away toward the city, trying to seem casual.\n\nMillie’s eyes suddenly brighten.\n\nTiny visual idea sparkle beside her bow.\n\nShe smiles.\n\nDialogue:\n\nMillie, almost whispering:\n<d>[English] Hmm...</d>\n\nTiny pause.\n\n<d>[English] I have an idea.</d>\n\nTheo is still looking away.\n\nHe does NOT hear her clearly.\n\nMusic introduces a mischievous tiny plucked-string motif.\n\n==================================================\n[SHOT 5 — 0:07.4–0:09.0]\nTHE SECRET MISSION BEGINS\n\nTheo remains seated on the bench.\n\nHis chin rests lightly against one paw.\n\nHe looks toward the distant festival area.\n\nBehind him:\n\nMillie quietly climbs down from the bench.\n\nShe puts one finger to her lips:\n\n“shhh.”\n\nShe tiptoes past the flowers.\n\nHer polka-dot skirt bounces gently.\n\nHer tail follows naturally.\n\nTheo remains unaware.\n\nCAMERA:\n\nmedium Theo foreground\n→ Millie slipping down behind him\n→ low tiny tracking shot following Millie.\n\nMillie whispers to herself:\n\n<d>[English] Quietly...</d>\n\nSound:\n\n- tiny shoe taps\n- dress rustle\n- leaves\n- park ambience\n\n==================================================\n[SHOT 6 — 0:09.0–0:11.0]\nFIRST, THE TINY HELPERS\n\nMillie reaches a tree.\n\nA BLUE BIRD and YELLOW BIRD sit together on a branch.\n\nShe waves them closer.\n\nMillie gestures toward Theo.\n\nThen mimics:\n\nplaying music.\n\nThen points toward the festival poster.\n\nThe birds tilt their heads.\n\nMillie clasps her paws:\n\nplease?\n\nThe blue bird chirps.\n\nThe yellow bird chirps back.\n\nBoth nod enthusiastically.\n\nMillie smiles.\n\nThen she turns and finds the friendly SQUIRREL in the flowerbed.\n\nShe whispers:\n\n<d>[English] I need your help too.</d>\n\nThe squirrel looks toward Theo.\n\nThen back to Millie.\n\nHe salutes with one tiny paw.\n\nUse cute visual comedy.\n\nCAMERA:\n\nMillie with birds\n→ bird reaction close-ups\n→ quick flowing transition to squirrel\n→ Millie delighted reaction.\n\nNo frantic cutting.\n\n==================================================\n[SHOT 7 — 0:11.0–0:13.2]\nTHE PLAN GETS BIGGER\n\nCreate one flowing helper-gathering sequence.\n\nMillie walks briskly toward the fountain.\n\nTwo / three pigeons turn toward her.\n\nShe makes a tiny directing gesture.\n\nThey follow.\n\nThen she notices the friendly VIOLINIST nearby.\n\nHe is casually practicing.\n\nMillie approaches.\n\nShe points:\n\ntoward Theo,\ntoward harmonica pocket area,\ntoward festival poster.\n\nThen presses her paws together hopefully.\n\nThe violinist understands.\n\nHe looks toward Theo.\n\nSmiles warmly.\n\nNods.\n\nMillie turns around.\n\nBehind her:\n\nbirds,\nsquirrel,\npigeons,\nviolinist\n\nare now a tiny secret team.\n\nMillie’s eyes widen with happy surprise at how many helpers she has gathered.\n\nDialogue:\n\nMillie, whispered:\n<d>[English] Okay... this got bigger.</d>\n\nSmall humorous music accent.\n\n==================================================\n[SHOT 8 — 0:13.2–0:15.0]\nTHE SURPRISE IS READY — CLIFFHANGER\n\nIntercut emotional contrast smoothly.\n\nTHEO:\n\nstill sits on the bench.\n\nHis posture is elegant but slightly slumped.\n\nHe looks down toward his pocket.\n\nOne paw rests over the hidden harmonica.\n\nHe quietly sighs.\n\nTheo murmurs:\n\n<d>[English] Maybe next year.</d>\n\nDO NOT make him cry.\n\nCUT / FLOW BEHIND HIM:\n\nMillie and the helpers quietly move into position.\n\n- birds perch above\n- squirrel waits beside flowers\n- pigeons line up near the path\n- violinist stands ready\n- everyone watches Millie\n\nMillie climbs onto the low base of the CITY PARK lamp post / safe bench edge.\n\nShe becomes a tiny conductor.\n\nShe looks at every helper.\n\nThey become still.\n\nMillie glances toward Theo.\n\nHe is turned away.\n\nPerfect.\n\nMillie raises one tiny paw dramatically.\n\nHold.\n\nHer expression:\n\ndetermined\n+\nexcited\n+\nloving.\n\nMusic reduces to a playful suspense pulse.\n\nNo one starts the surprise yet.\n\nMillie whispers:\n\n<d>[English] Ready?</d>\n\nThe blue bird gives a tiny nod.\n\nThe squirrel nods.\n\nThe violinist raises his bow slightly.\n\nMillie begins lowering her paw—\n\nCUT TO BLACK BEFORE THE SURPRISE STARTS.\n\nEND PART 1.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nThe soundscape should feel:\n\n- warm\n- natural\n- whimsical\n- cinematic\n- heartfelt\n- lightly jazzy\n\n==================================================\nTHEO VOICE\n==================================================\n\nAdult male voice.\n\nQualities:\n\n- warm\n- smooth\n- refined\n- understated\n- charming\n- gentle\n- slightly jazzy/vintage personality\n- never threatening\n\nHis confidence early should sound natural.\n\nHis nervousness should be subtle.\n\nImportant emotional dialogue:\n\n<d>[English] I was thinking about it.</d>\n\nThen:\n\n<d>[English] Maybe another time.</d>\n\nFinal:\n\n<d>[English] Maybe next year.</d>\n\nThese should sound reflective, not dramatically depressed.\n\n==================================================\nMILLIE VOICE\n==================================================\n\nYoung feminine mouse voice.\n\nQualities:\n\n- bright\n- warm\n- cheerful\n- energetic\n- sincere\n- expressive\n- slightly higher pitch than Theo\n\nDo NOT make it shrill.\n\nHer emotional progression:\n\nhappy\n→ curious\n→ concerned\n→ thoughtful\n→ inspired\n→ playful secret organizer.\n\nImportant:\n\n<d>[English] I have an idea.</d>\n\nshould be soft but excited.\n\nFinal:\n\n<d>[English] Ready?</d>\n\nshould be whispered with delighted anticipation.\n\n==================================================\nHELPER AUDIO\n==================================================\n\nBIRDS:\n\nsoft cheerful chirps.\n\nSQUIRREL:\n\ntiny harmless squeaks / chatter.\n\nPIGEONS:\n\ngentle coos.\n\nVIOLINIST:\n\nvery short natural violin warm-up tones.\n\nNo speaking dialogue required.\n\n==================================================\nPARK AMBIENCE\n==================================================\n\nUse:\n\n- soft breeze\n- leaves\n- distant fountain\n- birds\n- subtle city ambience\n- occasional distant footstep\n- light iron-fence / bench environmental resonance\n- flowerbed insects if subtle\n\nNo loud traffic.\n\nNo horns.\n\nNo urban noise dominating the story.\n\n==================================================\nHARMONICA SOUND\n==================================================\n\nIMPORTANT:\n\nTheo does NOT perform publicly during Part 1.\n\nWhen the harmonica is revealed:\n\nuse only a very small metallic handling sound.\n\nOptionally:\n\na single extremely faint memory-like harmonica note can appear NON-DIEGETICALLY in the score.\n\nDo NOT show Theo playing it.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL warm jazz-storybook score.\n\nInstrumentation:\n\n- soft upright bass\n- brushed snare\n- gentle piano\n- clarinet\n- muted trumpet used very lightly\n- pizzicato strings\n- acoustic guitar\n- subtle vibraphone\n- tiny glockenspiel for Millie’s idea\n\nMUSIC ARC:\n\n0:00–0:02\ncharming city-park jazz.\n\n0:02–0:04\nslight uncertainty as Theo looks at poster.\n\n0:04–0:06\nsoft emotional harmonica motif.\n\n0:06–0:08\ngentle emotional pause.\n\n0:08–0:10\nMillie’s playful idea motif.\n\n0:10–0:13\nlight sneaky/helper rhythm.\n\n0:13–0:15\nwarm suspense build.\n\nCUT music just before surprise begins.\n\nNo sad orchestral melodrama.\n\nNo huge trailer music.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 1 camera should evolve from:\n\nrelaxed\n→ intimate\n→ emotionally observant\n→ playful secret-mission energy.\n\nUse:\n\n- wide park establishment\n- gentle dolly\n- medium two-shots\n- close facial acting\n- rack focus\n- prop insert\n- low tiny-character tracking with Millie\n- helper reaction close-ups\n- shallow depth emphasis\n- final wide bench + helper composition\n\nAvoid:\n\n- handheld shake\n- crash zoom\n- frantic editing\n- excessive whip pans\n- extreme fisheye\n- 360 camera\n- action-style camera\n\nThe overall camera should remain graceful.\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished hand-painted 2D storybook/cartoon animation.\n\nMatch the uploaded storyboard in spirit.\n\nUse:\n\n- clean expressive ink contours\n- watercolor / gouache-inspired fills\n- subtle textured shading\n- beautiful park background painting\n- soft sunlight\n- stylized animal fur\n- expressive eyes\n- controlled squash-and-stretch\n- cinematic atmospheric depth\n- elegant vintage styling\n- subtle film grain / paper-paint texture if appropriate\n\nThe animation should feel like:\n\nTHE STORYBOARD ART HAS COME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealistic animals\n- live-action animals\n- generic 3D CGI\n- horror\n- hyperreal fur simulation\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary palette:\n\nTHEO:\n- charcoal black\n- warm gray muzzle\n- cream shirt\n- pink tie\n- pink pocket square\n- golden-yellow eyes\n\nMILLIE:\n- beige / taupe\n- soft pink\n- white polka dots\n- warm ear pink\n\nPARK:\n- spring green\n- sky blue\n- pastel city beige/pink/yellow\n- flower pink/yellow/white\n- black lamp/fence accents\n\nLighting:\n\nwarm late afternoon sunlight.\n\nUse:\n\n- soft golden side light\n- gentle cool sky fill\n- natural shadow under bench\n- bright readable faces\n\nBy final shot:\n\nintroduce slightly warmer golden tones to foreshadow the sunset festival.\n\n==================================================\nMOTION RULES — THEO\n==================================================\n\nTheo’s movements are refined.\n\nUse:\n\n- slow tie adjustment\n- relaxed leg crossing\n- tiny fedora touch\n- subtle eye movement\n- controlled harmonica reveal\n- small ear changes\n- gentle tail movement\n- restrained sigh\n\nAvoid exaggerated cartoon nervousness.\n\nThe emotion is in:\n\neyes,\nears,\nhands,\nposture.\n\n==================================================\nMOTION RULES — MILLIE\n==================================================\n\nMillie is energetic but believable.\n\nUse:\n\n- small enthusiastic hand gestures\n- light bouncing steps\n- gentle skirt secondary motion\n- tiny tiptoeing\n- expressive ear movement\n- quick turns\n- thoughtful chin-touch\n- subtle “idea” pose\n- conductor-like final signal\n\nHer bow must stay attached and maintain exact shape.\n\n==================================================\nSIZE / SCALE RULES\n==================================================\n\nTheo is substantially larger than Millie.\n\nMaintain consistent relative scale.\n\nWhen seated:\n\nMillie occupies a much smaller portion of the bench.\n\nTheo’s harmonica is comfortable in Theo’s paw but looks large relative to Millie.\n\nBirds and squirrel should be approximately around Millie’s scale or somewhat smaller.\n\nDo not make Millie suddenly human-sized.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Theo.\nONE Millie.\n\nSame:\n\n- Theo face\n- Theo hat\n- Theo suit\n- Theo pink tie\n- Theo eye color\n- Millie face\n- Millie bow\n- Millie dress\n- Millie shoes\n- relative scale\n- green bench\n- CITY PARK lamp\n- festival poster\n- city skyline\n- flowerbeds\n- golden harmonica\n\nHelper characters should remain visually consistent once introduced.\n\nTheo remains around the bench through Part 1.\n\nMillie physically moves through the park to gather helpers.\n\nNo teleportation.\n\n==================================================\nEMOTIONAL PERFORMANCE PRIORITY\n==================================================\n\nThis story works because Theo’s problem is SMALL and HUMAN:\n\nHe can play music.\n\nHe simply lacks confidence to share it publicly.\n\nMillie does NOT mock him.\n\nShe does NOT steal the harmonica.\n\nShe does NOT force him to perform.\n\nShe notices his hesitation and quietly creates support.\n\nPart 1 should make the audience think:\n\n“Oh... she’s going to help him.”\n\nwithout showing HOW yet.\n\n==================================================\nDIALOGUE TIMING PRIORITY\n==================================================\n\n15 seconds is short.\n\nUse natural pacing.\n\nIf some dialogue must be removed, PRIORITIZE:\n\n1.\nMillie:\n<d>[English] Theo?</d>\n\n2.\nMillie:\n<d>[English] You were going to play tonight?</d>\n\n3.\nTheo:\n<d>[English] Maybe another time.</d>\n\n4.\nMillie:\n<d>[English] I have an idea.</d>\n\n5.\nTheo:\n<d>[English] Maybe next year.</d>\n\n6.\nMillie:\n<d>[English] Ready?</d>\n\nSecondary lines may be omitted rather than spoken unnaturally fast.\n\n==================================================\nSAFETY / CONTENT RULES\n==================================================\n\nKeep all action safe and wholesome.\n\nNo violence.\n\nNo threatening behavior.\n\nNo weapons.\n\nNo fighting.\n\nNo dangerous park activity.\n\nNo romantic kissing.\n\nNo sexual behavior.\n\nNo disturbing imagery.\n\nNo bullying.\n\nTheo’s suit and fedora represent an elegant gentleman musician only.\n\nNo crime/gangster framing.\n\n==================================================\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086426143612440739"
  },
  {
    "id": "x-2086360699505451093",
    "title": "小草莓唤醒菠萝王 · 下篇",
    "titleEn": "The Little Strawberry Wakes the Pineapple King · Part 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《小草莓唤醒菠萝王 · 下篇》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Little Strawberry Wakes the Pineapple King · Part 2. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for:\n\n“THE LITTLE STRAWBERRY AND THE SLEEPY PINEAPPLE KING”\n\nas the PRIMARY visual reference for:\n\n- exact character identities\n- proportions\n- fruit textures\n- facial features\n- costume/accessory design\n- scale relationships\n- fruit-market environment\n- throne design\n- mirror design\n- makeover props\n- lighting\n- colors\n- emotional performances\n- Part 2 story progression\n\nThis sequence continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n\n- King Pine still peacefully asleep\n- grape monocle already attached\n- orange-peel mustache already attached\n- blueberry bow tie already attached\n- citrus crown jewel already attached\n- fruit garlands and decorations already covering him\n- Berry holding the FINAL sparkling fruit cape\n- Kiwi staring at Berry in complete disbelief\n- Berry saying:\n  <d>[English] One last thing!</d>\n- Kiwi reacting:\n  <d>[English] ONE MORE?!</d>\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- character-reference drawings\n- prop-reference drawings\n- title/header\n- page number\n- white storyboard-paper background\n- layout boxes\n- color notes\n\nThe uploaded storyboard locks:\n\n1. Berry’s exact strawberry-girl identity\n2. Berry’s strawberry-red body\n3. Berry’s green leafy crown/top\n4. Berry’s expressive large eyes\n5. Berry’s tiny scale\n6. King Pine’s exact giant pineapple identity\n7. King Pine’s golden pineapple texture\n8. King Pine’s giant leafy crown\n9. King Pine’s gentle facial design\n10. Kiwi’s exact small fuzzy kiwi design\n11. Kiwi’s comic facial language\n12. grape monocle\n13. orange-peel mustache\n14. blueberry bow tie\n15. citrus crown jewel\n16. fruit garlands\n17. sparkling fruit cape\n18. purple fruit throne\n19. ornate standing mirror\n20. warm fruit-market environment\n21. festival string lights\n22. final fruit crowd celebration\n23. warm golden-orange lighting\n24. Part 2 reveal and payoff\n\nTranslate the storyboard’s 15 one-second panels into approximately:\n\n8 COHERENT CINEMATIC SHOTS.\n\nDO NOT create fifteen frantic cuts.\n\nUse clear progression:\n\nfinal decoration\n→ Kiwi panic\n→ King Pine stirs\n→ nervous wake-up\n→ Berry presents makeover\n→ mirror suspense\n→ delighted approval\n→ festival celebration.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:15–0:30 of the complete 30-second story\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D storybook animation\n\nGenre:\ncozy fruit fantasy comedy\n\nTone:\nwarm,\nplayful,\nadorable,\nfestive,\nheartwarming.\n\nAudience:\nfamily-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + character sounds + market ambience + fruit-decoration sounds + festival music\n\nEMOTIONAL ARC:\n\nBerry confidently finishes masterpiece\n→ Kiwi panics\n→ King Pine wakes\n→ awkward suspense\n→ mirror reveal\n→ possible disaster\n→ King Pine LOVES it\n→ Berry receives approval\n→ fruit-market celebration\n→ warm comedy ending.\n\nCORE PART 2 PAYOFF:\n\nEveryone expects King Pine to dislike Berry’s ridiculous makeover.\n\nInstead...\n\nhe absolutely LOVES it.\n\n==================================================\nPART 2 START / END\n==================================================\n\nSTART:\n\nBerry is holding the sparkling fruit cape from the end of Part 1.\n\nKing Pine remains asleep.\n\nKiwi is already panicking.\n\nEND:\n\nKing Pine proudly wears the complete fruit makeover.\n\nBerry stands proudly beside him.\n\nKiwi remains comically baffled.\n\nFruit-market characters celebrate around them.\n\nThe ending must feel COMPLETE.\n\nNo cliffhanger.\n\n==================================================\nCHARACTER IDENTITY LOCK — BERRY\n==================================================\n\nMatch the uploaded storyboard and Part 1 EXACTLY.\n\nBERRY:\n\nTiny anthropomorphic strawberry girl.\n\nAppearance:\n\n- bright strawberry-red fruit body\n- visible tiny strawberry-seed details\n- rounded adorable face\n- very large expressive dark eyes\n- tiny mouth\n- rosy cheek feeling\n- green leafy crown/top\n- tiny arms\n- tiny legs\n- red dress-like lower silhouette\n- petite childlike proportions\n\nScale:\n\nVERY SMALL compared with King Pine.\n\nKing Pine’s head and torso should remain dramatically larger than Berry.\n\nPersonality:\n\n- imaginative\n- cheerful\n- fearless about her creative choices\n- proud\n- playful\n- affectionate\n- slightly mischievous\n\nPART 2 PERFORMANCE ARC:\n\n0–2 sec:\nfocused final styling\n\n2–4 sec:\nbrief nervousness as King wakes\n\n4–7 sec:\nproud presentation\n\n7–9 sec:\nnervous mirror suspense\n\n9–12 sec:\nexplosive joy when he approves\n\n12–15 sec:\nproud tiny stylist / festival hero.\n\nBerry should never appear afraid of King Pine.\n\nShe trusts that her creation is wonderful.\n\n==================================================\nCHARACTER IDENTITY LOCK — KING PINE\n==================================================\n\nMatch Part 1 and the uploaded Part 2 storyboard EXACTLY.\n\nKING PINE:\n\nHuge gentle pineapple king.\n\nAppearance:\n\n- enormous rounded pineapple body\n- warm golden-yellow pineapple skin\n- recognizable diamond-pattern texture\n- giant green leafy crown\n- expressive eyebrows\n- large gentle eyes once awake\n- soft cheeks\n- warm friendly mouth\n- lovable king-like proportions\n\nMAKEOVER CONTINUITY:\n\nAt Part 2 start he already wears:\n\n- grape monocle\n- orange-peel mustache\n- blueberry bow tie\n- fruit garlands\n- flowers\n- berry decorations\n- citrus crown jewel\n\nBerry now adds:\n\n- FINAL sparkling fruit cape\n\nOnce cape is installed:\n\nALL decorations remain.\n\nDo not remove anything.\n\nPERFORMANCE ARC:\n\n0–2 sec:\nstill asleep\n\n2–4 sec:\nslow sleepy awakening\n\n4–6 sec:\nconfused\n\n6–8 sec:\ncareful inspection\n\n8–10 sec:\nmirror suspense\n\n10–12 sec:\ndelighted realization\n\n12–15 sec:\nproud royal confidence + joyful laughter\n\nCRITICAL:\n\nKing Pine must NEVER become:\n\n- angry\n- threatening\n- scary\n- aggressive\n\nHis suspenseful reaction should only be:\n\nsleepy confusion\n+\nprocessing what he sees.\n\n==================================================\nCHARACTER IDENTITY LOCK — KIWI\n==================================================\n\nMatch Part 1 exactly.\n\nKIWI:\n\nTiny fuzzy kiwi companion.\n\nAppearance:\n\n- round fuzzy green-brown fruit body\n- large expressive eyes\n- small mouth\n- tiny dark arms/legs\n- simple adorable proportions\n\nPersonality:\n\n- cautious\n- dramatic\n- nervous\n- practical\n- loyal\n- unintentionally funny\n\nPART 2 PERFORMANCE ARC:\n\n0–2 sec:\nmaximum panic\n\n2–5 sec:\n“we are doomed”\n\n5–8 sec:\nbracing for King’s reaction\n\n8–10 sec:\nabsolute dread during mirror pause\n\n10–12 sec:\nshock that King loves it\n\n12–15 sec:\nconfused acceptance / exhausted disbelief.\n\nKiwi remains the PRIMARY comedy reaction character.\n\n==================================================\nFINAL SPARKLING FRUIT CAPE LOCK\n==================================================\n\nUse the exact cape design established in Part 1 and Part 2 storyboard.\n\nAppearance:\n\n- translucent rosy-pink / citrus-gold storybook fabric\n- warm glittering highlights\n- orange-slice medallions\n- tiny grapes / berries\n- fruit-bead trim\n- subtle hanging decorations\n- elegant royal silhouette\n\nAt the beginning:\n\nBerry places the cape around King Pine.\n\nAfter placement:\n\ncape remains on King Pine throughout Part 2.\n\nWhen King Pine spreads it later:\n\nthe fabric should open elegantly like a royal festival cloak.\n\nNo costume transformation.\n\n==================================================\nGRAPE MONOCLE LOCK\n==================================================\n\nCircular purple-grape ornament around one eye.\n\nOnce King Pine wakes:\n\nhe may touch or inspect it.\n\nIt stays attached.\n\nDo not replace it with normal eyeglasses.\n\n==================================================\nORANGE-PEEL MUSTACHE LOCK\n==================================================\n\nCurled orange-peel royal mustache under King Pine’s nose.\n\nIt stays attached.\n\nWhen King Pine wakes:\n\nhe may gently touch / twirl one side while confused.\n\nKeep it cute.\n\n==================================================\nBLUEBERRY BOW TIE LOCK\n==================================================\n\nDark blueberry-purple bow tie beneath King Pine’s face.\n\nClearly visible in:\n\n- wake-up shot\n- mirror reveal\n- approval\n- final celebration.\n\n==================================================\nMIRROR LOCK\n==================================================\n\nUse the tall ornate mirror from the storyboard.\n\nDesign:\n\n- tall full-body standing mirror\n- elegant carved golden/wooden frame\n- fruit-market fantasy aesthetic\n- clean reflective surface\n\nMirror must clearly show King Pine’s FULL decorated look.\n\nDo not create a separate duplicate King Pine outside normal reflection logic.\n\n==================================================\nENVIRONMENT LOCK — FRUIT FESTIVAL MARKET\n==================================================\n\nMaintain exact environment continuity from Part 1.\n\nInclude:\n\n- cozy wooden market stalls\n- fruit baskets\n- fruit crates\n- citrus garlands\n- warm hanging lanterns\n- string lights\n- bunting\n- wooden shelving\n- baskets of grapes, oranges, berries and apples\n- warm sunset / early evening light\n\nPart 2 should gradually become brighter and more festive.\n\nBeginning:\nwarm market light.\n\nEnding:\nlanterns and string lights sparkle more strongly as crowd celebrates.\n\nNo modern humans.\n\nNo modern technology.\n\nNo brands.\n\n==================================================\nFRUIT CROWD DESIGN\n==================================================\n\nThe festival crowd appears ONLY near the ending.\n\nCreate a small cheerful group of original anthropomorphic fruits such as:\n\n- apple characters\n- orange characters\n- grape cluster character\n- banana character\n- cherry pair\n- pear\n- watermelon slice / small watermelon\n- berry characters\n\nKeep them:\n\ncute,\nsimple,\nfamily-friendly,\nsupporting background personalities.\n\nDo not introduce them too early.\n\nTheir role:\n\nwitness\n→ surprise\n→ applause\n→ celebration.\n\n==================================================\nWORLD / STORY LOGIC\n==================================================\n\nBerry places final cape\n→ makeover becomes complete.\n\nKiwi panics\n→ King Pine begins waking.\n\nKing Pine notices decorations\n→ confusion.\n\nBerry proudly presents him\n→ King Pine investigates.\n\nKing Pine looks into mirror\n→ suspense.\n\nKing Pine understands how spectacular he looks\n→ delight.\n\nKing Pine approves Berry\n→ nearby fruit crowd gathers.\n\nCrowd celebrates\n→ Berry’s makeover becomes festival hit.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.7]\nTHE FINAL ROYAL LAYER\n\nContinue DIRECTLY from Part 1.\n\nBerry stands beside still-sleeping King Pine holding the sparkling fruit cape.\n\nKiwi stares in panic.\n\nBerry gently swings the cape around King Pine’s shoulders/body.\n\nThe cape settles beautifully over the existing decorations.\n\nShow close detail:\n\n- glittering fruit beads\n- orange-slice medallions\n- rosy translucent fabric\n- berries catching warm light\n\nBerry steps back.\n\nHands together proudly.\n\nDialogue:\n\nBerry, softly:\n<d>[English] There.</d>\n\nTiny pause.\n\n<d>[English] Perfect.</d>\n\nKiwi:\n\n<d>[English] Berry...</d>\n\nBerry smiles.\n\nKing Pine continues sleeping.\n\nCAMERA:\n\ncape detail\n→ smooth fabric placement\n→ Berry proud close-up\n→ Kiwi reaction.\n\nSFX:\n\n- soft fabric unfurl\n- fruit bead chimes\n- gentle glitter shimmer\n- King Pine snore\n\nMUSIC:\n\ncontinue Part 1 playful motif,\nbut add a tiny “masterpiece complete” flourish.\n\n==================================================\n[SHOT 2 — 0:01.7–0:03.2]\nKIWI’S FINAL RESCUE MISSION\n\nKiwi looks at Berry.\n\nLooks at King Pine.\n\nLooks at the outrageous makeover.\n\nHis eyes widen.\n\nKiwi rushes toward King Pine.\n\nHe gently calls:\n\n<d>[English] Your Majesty?</d>\n\nNo response.\n\nA little louder:\n\n<d>[English] Your Majesty!</d>\n\nSuddenly:\n\nKing Pine’s giant leafy crown TWITCHES.\n\nKiwi freezes.\n\nBerry freezes.\n\nOne side of the orange-peel mustache moves slightly with King Pine’s sleepy breath.\n\nThe grape monocle shifts a tiny amount.\n\nKing Pine:\n\n<d>[English] Mmm...?</d>\n\nKiwi slowly backs away.\n\nKiwi whispers:\n\n<d>[English] Oh no.</d>\n\nMusic briefly reduces.\n\n==================================================\n[SHOT 3 — 0:03.2–0:05.0]\nTHE SLEEPY KING AWAKENS\n\nClose on King Pine.\n\nOne eye opens slowly.\n\nThen the other.\n\nHuge sleepy blink.\n\nHe sits slightly more upright in the purple throne.\n\nThe sparkling cape slides naturally into place around his shoulders.\n\nKing Pine yawns gently.\n\nNo aggression.\n\nHe notices the grape monocle.\n\nTouches it.\n\nThen notices his orange-peel mustache.\n\nTouches one curled end.\n\nHe looks at Berry.\n\nThen Kiwi.\n\nDialogue:\n\nKing Pine, sleepy:\n<d>[English] Did I miss something?</d>\n\nBerry immediately lights up.\n\nKiwi braces himself.\n\nCAMERA:\n\neye opening close-up\n→ medium King Pine\n→ monocle touch\n→ mustache touch\n→ three-character reaction shot.\n\nSFX:\n\n- sleepy yawn\n- soft pineapple texture rustle\n- leafy crown movement\n- fruit decorations jingling gently.\n\n==================================================\n[SHOT 4 — 0:05.0–0:06.7]\nBERRY PRESENTS HER MASTERPIECE\n\nBerry steps dramatically into open space in front of King Pine.\n\nShe throws both tiny arms outward like a professional stylist revealing a masterpiece.\n\nDialogue:\n\nBerry:\n<d>[English] Your festival look!</d>\n\nKing Pine looks down at himself.\n\nGrape chains.\n\nBlueberry bow tie.\n\nOrange peel mustache.\n\nSparkling cape.\n\nFruit jewels.\n\nBerry gestures proudly from one decoration to another.\n\nBerry:\n\n<d>[English] Royal. Festive. Magnificent.</d>\n\nKiwi slowly covers his face with both tiny hands.\n\nKing Pine blinks.\n\nDialogue:\n\nKing Pine:\n<d>[English] I see...</d>\n\nUse a tiny suspenseful musical note.\n\n==================================================\n[SHOT 5 — 0:06.7–0:08.6]\nTHE ROYAL WALK TO THE MIRROR\n\nKing Pine slowly rises from the throne.\n\nKeep scale large and movement gentle.\n\nThe cape hangs beautifully.\n\nBerry leads him toward the tall ornate mirror.\n\nKiwi follows reluctantly.\n\nBerry walks with proud stylist confidence.\n\nKing Pine carefully adjusts the cape while walking.\n\nCamera tracks sideways.\n\nThen move behind them toward mirror.\n\nIn the reflection:\n\nfirst reveal partial silhouette of the decorated King.\n\nBerry:\n\n<d>[English] Ready?</d>\n\nKing Pine:\n<d>[English] I think so.</d>\n\nKiwi quietly:\n\n<d>[English] I’m not.</d>\n\nSmall comedic music accent.\n\n==================================================\n[SHOT 6 — 0:08.6–0:10.5]\nTHE MIRROR REVEAL\n\nKing Pine reaches the mirror.\n\nCamera shifts so the audience sees his FULL reflection.\n\nThis is the MAIN visual reveal.\n\nHe wears:\n\n- grape monocle\n- orange-peel mustache\n- blueberry bow tie\n- citrus crown jewel\n- flower decorations\n- berry medallions\n- fruit garlands\n- sparkling fruit cape\n\nThe look is:\n\nglorious,\ncolorful,\nregal,\nridiculous,\nadorable.\n\nKing Pine stares.\n\nSilence.\n\nBerry stands beside him with hopeful smile.\n\nKiwi hides partly behind Berry.\n\nKing Pine slowly touches his mustache.\n\nThen adjusts his monocle.\n\nStill silent.\n\nBerry’s confident smile becomes slightly nervous.\n\nDialogue:\n\nBerry:\n<d>[English] Well?</d>\n\nHOLD.\n\nKiwi shuts his eyes.\n\nMusic almost completely stops.\n\n==================================================\n[SHOT 7 — 0:10.5–0:12.5]\nDISASTER TURNS INTO DELIGHT\n\nKing Pine’s serious expression breaks.\n\nFirst:\n\ntiny smile.\n\nThen:\n\nbigger smile.\n\nThen:\n\nwarm, delighted laughter.\n\nKing Pine:\n\n<d>[English] I LOVE IT!</d>\n\nMusic bursts into joyful festival melody.\n\nBerry jumps with happiness.\n\nKiwi opens one eye.\n\nCompletely confused.\n\nKing Pine turns dramatically away from mirror.\n\nHe spreads the sparkling fruit cape wide.\n\nThe fruit ornaments shimmer.\n\nKing Pine proudly poses.\n\nDialogue:\n\nKing Pine:\n<d>[English] Look at me!</d>\n\nBerry:\n\n<d>[English] I knew it!</d>\n\nKiwi:\n\n<d>[English] You did?</d>\n\nKing Pine laughs.\n\nCAMERA:\n\nserious King close-up\n→ smile progression\n→ Berry reaction\n→ wide cape-spread hero shot.\n\n==================================================\n[SHOT 8 — 0:12.5–0:15]\nROYAL APPROVAL + FESTIVAL PAYOFF\n\nKing Pine gently kneels / lowers himself enough to reach tiny Berry safely.\n\nHe gives Berry a huge gentle thumbs-up or places one hand over his heart.\n\nDialogue:\n\nKing Pine:\n<d>[English] Brilliant work, little stylist.</d>\n\nBerry beams.\n\nShe performs a tiny proud bow.\n\nNearby fruit-market characters notice.\n\nOne apple gasps happily.\n\nAn orange points.\n\nA grape cluster cheers.\n\nMore fruit characters gather.\n\nWarm festival lights brighten.\n\nThe crowd begins applauding and celebrating.\n\nKing Pine stands proudly in full makeover.\n\nHe spreads the sparkling cape one final time.\n\nBerry stands beside him with hands proudly on hips.\n\nKiwi stands on Berry’s other side.\n\nKiwi looks:\n\nKing Pine\n→ cheering crowd\n→ Berry.\n\nHe cannot believe this worked.\n\nKiwi:\n\n<d>[English] I understand nothing.</d>\n\nBerry giggles.\n\nKing Pine laughs warmly.\n\nFINAL HERO COMPOSITION:\n\n- giant King Pine centered\n- full fruit makeover visible\n- sparkling cape spread\n- Berry proudly beside him\n- Kiwi baffled beside Berry\n- cheerful fruit crowd surrounding them\n- lanterns glowing overhead\n- fruit garlands framing image\n- warm festival market behind\n- golden-orange light\n- tiny sparkling fruit highlights\n\nCamera slowly pulls backward.\n\nHold the celebratory composition.\n\nFADE TO BLACK.\n\nEND STORY.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio should feel:\n\n- warm\n- playful\n- family-friendly\n- festive\n- expressive\n- storybook cinematic\n\n==================================================\nBERRY VOICE\n==================================================\n\nCute youthful girl voice.\n\nQualities:\n\n- bright\n- creative\n- confident\n- playful\n- sincere\n- energetic\n\nDo not make her overly baby-like.\n\nPerformance:\n\nbeginning:\nproud stylist.\n\nKing wakes:\nbrief nervous anticipation.\n\nMirror:\nhopeful.\n\nApproval:\njoyful.\n\nImportant lines:\n\n<d>[English] Your festival look!</d>\n\n<d>[English] Royal. Festive. Magnificent.</d>\n\n<d>[English] Well?</d>\n\n<d>[English] I knew it!</d>\n\n==================================================\nKING PINE VOICE\n==================================================\n\nWarm gentle adult male fantasy voice.\n\nQualities:\n\n- deep but soft\n- sleepy at first\n- kind\n- regal\n- lovable\n- warm laugh\n- never intimidating\n\nVoice progression:\n\nwake-up:\nslow and sleepy.\n\nmirror:\nquiet curiosity.\n\napproval:\nbig warm delight.\n\nCritical line:\n\n<d>[English] I LOVE IT!</d>\n\nDeliver with genuine childlike delight despite his deep voice.\n\nNot shouted aggressively.\n\nFinal approval:\n\n<d>[English] Brilliant work, little stylist.</d>\n\nshould feel affectionate and sincere.\n\n==================================================\nKIWI VOICE\n==================================================\n\nSmall high-energy companion voice.\n\nQualities:\n\n- nervous\n- fast\n- expressive\n- lovable\n- slightly dramatic\n- excellent comic timing\n\nPerformance:\n\nbeginning:\npure panic.\n\nwake-up:\nfearful whisper.\n\nmirror:\nbracing for disaster.\n\napproval:\ncomplete disbelief.\n\nFinal:\n\n<d>[English] I understand nothing.</d>\n\nDeliver slowly, exhausted and baffled.\n\n==================================================\nFRUIT CROWD AUDIO\n==================================================\n\nKeep supporting voices simple.\n\nUse:\n\n- cheerful gasps\n- little laughs\n- soft applause\n- celebratory “wow!”\n- happy fruit chatter\n\nDo not let crowd dialogue overpower main characters.\n\n==================================================\nMARKET AUDIO\n==================================================\n\nMaintain from Part 1:\n\n- gentle market ambience\n- breeze\n- wooden stall creaks\n- basket movement\n- leaves\n- hanging lantern chimes\n- fruit-garland movements\n\nDuring ending:\n\nincrease:\n\n- applause\n- festival bells\n- cheerful chatter\n- tiny tambourine-like fruit sounds\n\n==================================================\nMAKEOVER PROP SOUND DESIGN\n==================================================\n\nCape placement:\n\nsoft fabric SWISH\n+\ntiny berry-bead chimes.\n\nMonocle touch:\n\nsmall grape click.\n\nOrange-peel mustache touch:\n\ngentle peel rustle.\n\nCape hero spread:\n\nlarger elegant fabric SWOOSH\n+\nsoft glitter shimmer.\n\nMirror:\n\ntiny clean glass sparkle when full reveal appears.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL whimsical festival score.\n\nInstrumentation:\n\n- pizzicato strings\n- marimba\n- clarinet\n- flute\n- soft xylophone\n- tiny bells\n- hand percussion\n- acoustic plucks\n- warm strings\n- light accordion-style texture if subtle\n- festive tambourine near ending\n\nMUSIC ARC:\n\n0:00–0:02\n“finishing masterpiece” motif.\n\n0:02–0:04\ncomic suspense as King wakes.\n\n0:04–0:07\ncurious sleepy motif.\n\n0:07–0:09\nslow royal mirror walk.\n\n0:09–0:10.5\nalmost complete silence / tension.\n\n0:10.5–0:12.5\nbig joyful comedy release.\n\n0:12.5–0:15\nwarm fruit-festival celebration.\n\nEnd on a happy resolved chord.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 2 camera progression:\n\n0–2 sec:\nclose finishing-detail coverage.\n\n2–5 sec:\nwake-up reaction close-ups.\n\n5–7 sec:\ncharacter presentation.\n\n7–9 sec:\nslow royal tracking toward mirror.\n\n9–10.5 sec:\nmirror reveal + reaction stillness.\n\n10.5–12.5 sec:\nwider energetic comedy payoff.\n\n12.5–15 sec:\nlarge celebratory hero composition.\n\nUse:\n\n- macro cape details\n- gentle push-in\n- eye-level reaction close-ups\n- over-the-shoulder mirror view\n- reflection composition\n- scale-revealing wide shots\n- slow pull-back\n- subtle rack focus\n- gentle cinematic parallax\n\nAvoid:\n\n- shaky handheld\n- crash zoom\n- spinning camera\n- extreme fisheye\n- frantic editing\n- excessive motion blur\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished painterly 2D storybook animation.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n\n- charming hand-drawn ink contours\n- watercolor / gouache-inspired rendering\n- soft painterly shading\n- warm fruit textures\n- expressive rounded faces\n- beautiful wooden market details\n- soft glow from lanterns\n- subtle cinematic depth\n- natural fabric movement\n- tasteful sparkle effects\n- gentle squash and stretch\n- premium family-animation finish\n\nThe final video should feel like:\n\nTHE UPLOADED STORYBOARD ILLUSTRATIONS CAME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photoreal food\n- realistic fruit photography\n- live action\n- generic CGI\n- flat vector animation\n- horror\n\n==================================================\nCOLOR / LIGHTING ARC\n==================================================\n\nPART 2 palette:\n\nBERRY:\nstrawberry red + leafy green.\n\nKING PINE:\nwarm pineapple gold + leafy green.\n\nKIWI:\nolive green / fuzzy brown-green.\n\nMAKEOVER:\npurple grapes,\nblueberry violet,\norange citrus,\nsoft flower pink,\nberry red,\nwarm gold.\n\nENVIRONMENT:\nwarm honey gold,\namber,\nsoft brown,\nmuted green.\n\nLIGHTING PROGRESSION:\n\n0–5 sec:\nsame warm late-afternoon glow from Part 1.\n\n5–9 sec:\nintroduce more visible festival lanterns.\n\n9–10.5 sec:\nmirror scene slightly calmer and more focused.\n\n10.5–12.5 sec:\nincrease sparkling fruit highlights after King smiles.\n\n12.5–15 sec:\nbrightest, richest, warmest lighting of the entire story.\n\nFinal shot should feel like:\n\nthe festival officially started because King Pine loved the makeover.\n\n==================================================\nMOTION RULES — BERRY\n==================================================\n\nUse:\n\n- tiny energetic steps\n- careful cape placement\n- expressive hand movements\n- proud presentation stance\n- tiny nervous pause\n- joyful hop when King approves\n- proud final bow\n- hands-on-hips hero pose\n\nBerry must remain physically tiny relative to King Pine.\n\n==================================================\nMOTION RULES — KING PINE\n==================================================\n\nKing Pine is enormous and gentle.\n\nWake-up:\n\nleaf crown twitch\n→ eyelid movement\n→ slow eyes opening\n→ big sleepy blink\n→ slow sit-up.\n\nWalking:\n\nheavy but soft,\nslow,\nfriendly.\n\nMirror:\n\nsubtle facial acting.\n\nApproval:\n\nlarge warm smile,\ngentle laugh,\ncape spread.\n\nNever make his motion aggressive.\n\n==================================================\nMOTION RULES — KIWI\n==================================================\n\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086360699505451093",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T07:55:28.126Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086360699505451093.mp4",
    "posterUrl": "/posters/x/2086360699505451093.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "food"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second Storyboard Image to video using MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for:\n\n“THE LITTLE STRAWBERRY AND THE SLEEPY PINEAPPLE KING”\n\nas the PRIMARY visual reference for:\n\n- exact character identities\n- proportions\n- fruit textures\n- facial features\n- costume/accessory design\n- scale relationships\n- fruit-market environment\n- throne design\n- mirror design\n- makeover props\n- lighting\n- colors\n- emotional performances\n- Part 2 story progression\n\nThis sequence continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n\n- King Pine still peacefully asleep\n- grape monocle already attached\n- orange-peel mustache already attached\n- blueberry bow tie already attached\n- citrus crown jewel already attached\n- fruit garlands and decorations already covering him\n- Berry holding the FINAL sparkling fruit cape\n- Kiwi staring at Berry in complete disbelief\n- Berry saying:\n  <d>[English] One last thing!</d>\n- Kiwi reacting:\n  <d>[English] ONE MORE?!</d>\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- character-reference drawings\n- prop-reference drawings\n- title/header\n- page number\n- white storyboard-paper background\n- layout boxes\n- color notes\n\nThe uploaded storyboard locks:\n\n1. Berry’s exact strawberry-girl identity\n2. Berry’s strawberry-red body\n3. Berry’s green leafy crown/top\n4. Berry’s expressive large eyes\n5. Berry’s tiny scale\n6. King Pine’s exact giant pineapple identity\n7. King Pine’s golden pineapple texture\n8. King Pine’s giant leafy crown\n9. King Pine’s gentle facial design\n10. Kiwi’s exact small fuzzy kiwi design\n11. Kiwi’s comic facial language\n12. grape monocle\n13. orange-peel mustache\n14. blueberry bow tie\n15. citrus crown jewel\n16. fruit garlands\n17. sparkling fruit cape\n18. purple fruit throne\n19. ornate standing mirror\n20. warm fruit-market environment\n21. festival string lights\n22. final fruit crowd celebration\n23. warm golden-orange lighting\n24. Part 2 reveal and payoff\n\nTranslate the storyboard’s 15 one-second panels into approximately:\n\n8 COHERENT CINEMATIC SHOTS.\n\nDO NOT create fifteen frantic cuts.\n\nUse clear progression:\n\nfinal decoration\n→ Kiwi panic\n→ King Pine stirs\n→ nervous wake-up\n→ Berry presents makeover\n→ mirror suspense\n→ delighted approval\n→ festival celebration.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:15–0:30 of the complete 30-second story\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D storybook animation\n\nGenre:\ncozy fruit fantasy comedy\n\nTone:\nwarm,\nplayful,\nadorable,\nfestive,\nheartwarming.\n\nAudience:\nfamily-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + character sounds + market ambience + fruit-decoration sounds + festival music\n\nEMOTIONAL ARC:\n\nBerry confidently finishes masterpiece\n→ Kiwi panics\n→ King Pine wakes\n→ awkward suspense\n→ mirror reveal\n→ possible disaster\n→ King Pine LOVES it\n→ Berry receives approval\n→ fruit-market celebration\n→ warm comedy ending.\n\nCORE PART 2 PAYOFF:\n\nEveryone expects King Pine to dislike Berry’s ridiculous makeover.\n\nInstead...\n\nhe absolutely LOVES it.\n\n==================================================\nPART 2 START / END\n==================================================\n\nSTART:\n\nBerry is holding the sparkling fruit cape from the end of Part 1.\n\nKing Pine remains asleep.\n\nKiwi is already panicking.\n\nEND:\n\nKing Pine proudly wears the complete fruit makeover.\n\nBerry stands proudly beside him.\n\nKiwi remains comically baffled.\n\nFruit-market characters celebrate around them.\n\nThe ending must feel COMPLETE.\n\nNo cliffhanger.\n\n==================================================\nCHARACTER IDENTITY LOCK — BERRY\n==================================================\n\nMatch the uploaded storyboard and Part 1 EXACTLY.\n\nBERRY:\n\nTiny anthropomorphic strawberry girl.\n\nAppearance:\n\n- bright strawberry-red fruit body\n- visible tiny strawberry-seed details\n- rounded adorable face\n- very large expressive dark eyes\n- tiny mouth\n- rosy cheek feeling\n- green leafy crown/top\n- tiny arms\n- tiny legs\n- red dress-like lower silhouette\n- petite childlike proportions\n\nScale:\n\nVERY SMALL compared with King Pine.\n\nKing Pine’s head and torso should remain dramatically larger than Berry.\n\nPersonality:\n\n- imaginative\n- cheerful\n- fearless about her creative choices\n- proud\n- playful\n- affectionate\n- slightly mischievous\n\nPART 2 PERFORMANCE ARC:\n\n0–2 sec:\nfocused final styling\n\n2–4 sec:\nbrief nervousness as King wakes\n\n4–7 sec:\nproud presentation\n\n7–9 sec:\nnervous mirror suspense\n\n9–12 sec:\nexplosive joy when he approves\n\n12–15 sec:\nproud tiny stylist / festival hero.\n\nBerry should never appear afraid of King Pine.\n\nShe trusts that her creation is wonderful.\n\n==================================================\nCHARACTER IDENTITY LOCK — KING PINE\n==================================================\n\nMatch Part 1 and the uploaded Part 2 storyboard EXACTLY.\n\nKING PINE:\n\nHuge gentle pineapple king.\n\nAppearance:\n\n- enormous rounded pineapple body\n- warm golden-yellow pineapple skin\n- recognizable diamond-pattern texture\n- giant green leafy crown\n- expressive eyebrows\n- large gentle eyes once awake\n- soft cheeks\n- warm friendly mouth\n- lovable king-like proportions\n\nMAKEOVER CONTINUITY:\n\nAt Part 2 start he already wears:\n\n- grape monocle\n- orange-peel mustache\n- blueberry bow tie\n- fruit garlands\n- flowers\n- berry decorations\n- citrus crown jewel\n\nBerry now adds:\n\n- FINAL sparkling fruit cape\n\nOnce cape is installed:\n\nALL decorations remain.\n\nDo not remove anything.\n\nPERFORMANCE ARC:\n\n0–2 sec:\nstill asleep\n\n2–4 sec:\nslow sleepy awakening\n\n4–6 sec:\nconfused\n\n6–8 sec:\ncareful inspection\n\n8–10 sec:\nmirror suspense\n\n10–12 sec:\ndelighted realization\n\n12–15 sec:\nproud royal confidence + joyful laughter\n\nCRITICAL:\n\nKing Pine must NEVER become:\n\n- angry\n- threatening\n- scary\n- aggressive\n\nHis suspenseful reaction should only be:\n\nsleepy confusion\n+\nprocessing what he sees.\n\n==================================================\nCHARACTER IDENTITY LOCK — KIWI\n==================================================\n\nMatch Part 1 exactly.\n\nKIWI:\n\nTiny fuzzy kiwi companion.\n\nAppearance:\n\n- round fuzzy green-brown fruit body\n- large expressive eyes\n- small mouth\n- tiny dark arms/legs\n- simple adorable proportions\n\nPersonality:\n\n- cautious\n- dramatic\n- nervous\n- practical\n- loyal\n- unintentionally funny\n\nPART 2 PERFORMANCE ARC:\n\n0–2 sec:\nmaximum panic\n\n2–5 sec:\n“we are doomed”\n\n5–8 sec:\nbracing for King’s reaction\n\n8–10 sec:\nabsolute dread during mirror pause\n\n10–12 sec:\nshock that King loves it\n\n12–15 sec:\nconfused acceptance / exhausted disbelief.\n\nKiwi remains the PRIMARY comedy reaction character.\n\n==================================================\nFINAL SPARKLING FRUIT CAPE LOCK\n==================================================\n\nUse the exact cape design established in Part 1 and Part 2 storyboard.\n\nAppearance:\n\n- translucent rosy-pink / citrus-gold storybook fabric\n- warm glittering highlights\n- orange-slice medallions\n- tiny grapes / berries\n- fruit-bead trim\n- subtle hanging decorations\n- elegant royal silhouette\n\nAt the beginning:\n\nBerry places the cape around King Pine.\n\nAfter placement:\n\ncape remains on King Pine throughout Part 2.\n\nWhen King Pine spreads it later:\n\nthe fabric should open elegantly like a royal festival cloak.\n\nNo costume transformation.\n\n==================================================\nGRAPE MONOCLE LOCK\n==================================================\n\nCircular purple-grape ornament around one eye.\n\nOnce King Pine wakes:\n\nhe may touch or inspect it.\n\nIt stays attached.\n\nDo not replace it with normal eyeglasses.\n\n==================================================\nORANGE-PEEL MUSTACHE LOCK\n==================================================\n\nCurled orange-peel royal mustache under King Pine’s nose.\n\nIt stays attached.\n\nWhen King Pine wakes:\n\nhe may gently touch / twirl one side while confused.\n\nKeep it cute.\n\n==================================================\nBLUEBERRY BOW TIE LOCK\n==================================================\n\nDark blueberry-purple bow tie beneath King Pine’s face.\n\nClearly visible in:\n\n- wake-up shot\n- mirror reveal\n- approval\n- final celebration.\n\n==================================================\nMIRROR LOCK\n==================================================\n\nUse the tall ornate mirror from the storyboard.\n\nDesign:\n\n- tall full-body standing mirror\n- elegant carved golden/wooden frame\n- fruit-market fantasy aesthetic\n- clean reflective surface\n\nMirror must clearly show King Pine’s FULL decorated look.\n\nDo not create a separate duplicate King Pine outside normal reflection logic.\n\n==================================================\nENVIRONMENT LOCK — FRUIT FESTIVAL MARKET\n==================================================\n\nMaintain exact environment continuity from Part 1.\n\nInclude:\n\n- cozy wooden market stalls\n- fruit baskets\n- fruit crates\n- citrus garlands\n- warm hanging lanterns\n- string lights\n- bunting\n- wooden shelving\n- baskets of grapes, oranges, berries and apples\n- warm sunset / early evening light\n\nPart 2 should gradually become brighter and more festive.\n\nBeginning:\nwarm market light.\n\nEnding:\nlanterns and string lights sparkle more strongly as crowd celebrates.\n\nNo modern humans.\n\nNo modern technology.\n\nNo brands.\n\n==================================================\nFRUIT CROWD DESIGN\n==================================================\n\nThe festival crowd appears ONLY near the ending.\n\nCreate a small cheerful group of original anthropomorphic fruits such as:\n\n- apple characters\n- orange characters\n- grape cluster character\n- banana character\n- cherry pair\n- pear\n- watermelon slice / small watermelon\n- berry characters\n\nKeep them:\n\ncute,\nsimple,\nfamily-friendly,\nsupporting background personalities.\n\nDo not introduce them too early.\n\nTheir role:\n\nwitness\n→ surprise\n→ applause\n→ celebration.\n\n==================================================\nWORLD / STORY LOGIC\n==================================================\n\nBerry places final cape\n→ makeover becomes complete.\n\nKiwi panics\n→ King Pine begins waking.\n\nKing Pine notices decorations\n→ confusion.\n\nBerry proudly presents him\n→ King Pine investigates.\n\nKing Pine looks into mirror\n→ suspense.\n\nKing Pine understands how spectacular he looks\n→ delight.\n\nKing Pine approves Berry\n→ nearby fruit crowd gathers.\n\nCrowd celebrates\n→ Berry’s makeover becomes festival hit.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:01.7]\nTHE FINAL ROYAL LAYER\n\nContinue DIRECTLY from Part 1.\n\nBerry stands beside still-sleeping King Pine holding the sparkling fruit cape.\n\nKiwi stares in panic.\n\nBerry gently swings the cape around King Pine’s shoulders/body.\n\nThe cape settles beautifully over the existing decorations.\n\nShow close detail:\n\n- glittering fruit beads\n- orange-slice medallions\n- rosy translucent fabric\n- berries catching warm light\n\nBerry steps back.\n\nHands together proudly.\n\nDialogue:\n\nBerry, softly:\n<d>[English] There.</d>\n\nTiny pause.\n\n<d>[English] Perfect.</d>\n\nKiwi:\n\n<d>[English] Berry...</d>\n\nBerry smiles.\n\nKing Pine continues sleeping.\n\nCAMERA:\n\ncape detail\n→ smooth fabric placement\n→ Berry proud close-up\n→ Kiwi reaction.\n\nSFX:\n\n- soft fabric unfurl\n- fruit bead chimes\n- gentle glitter shimmer\n- King Pine snore\n\nMUSIC:\n\ncontinue Part 1 playful motif,\nbut add a tiny “masterpiece complete” flourish.\n\n==================================================\n[SHOT 2 — 0:01.7–0:03.2]\nKIWI’S FINAL RESCUE MISSION\n\nKiwi looks at Berry.\n\nLooks at King Pine.\n\nLooks at the outrageous makeover.\n\nHis eyes widen.\n\nKiwi rushes toward King Pine.\n\nHe gently calls:\n\n<d>[English] Your Majesty?</d>\n\nNo response.\n\nA little louder:\n\n<d>[English] Your Majesty!</d>\n\nSuddenly:\n\nKing Pine’s giant leafy crown TWITCHES.\n\nKiwi freezes.\n\nBerry freezes.\n\nOne side of the orange-peel mustache moves slightly with King Pine’s sleepy breath.\n\nThe grape monocle shifts a tiny amount.\n\nKing Pine:\n\n<d>[English] Mmm...?</d>\n\nKiwi slowly backs away.\n\nKiwi whispers:\n\n<d>[English] Oh no.</d>\n\nMusic briefly reduces.\n\n==================================================\n[SHOT 3 — 0:03.2–0:05.0]\nTHE SLEEPY KING AWAKENS\n\nClose on King Pine.\n\nOne eye opens slowly.\n\nThen the other.\n\nHuge sleepy blink.\n\nHe sits slightly more upright in the purple throne.\n\nThe sparkling cape slides naturally into place around his shoulders.\n\nKing Pine yawns gently.\n\nNo aggression.\n\nHe notices the grape monocle.\n\nTouches it.\n\nThen notices his orange-peel mustache.\n\nTouches one curled end.\n\nHe looks at Berry.\n\nThen Kiwi.\n\nDialogue:\n\nKing Pine, sleepy:\n<d>[English] Did I miss something?</d>\n\nBerry immediately lights up.\n\nKiwi braces himself.\n\nCAMERA:\n\neye opening close-up\n→ medium King Pine\n→ monocle touch\n→ mustache touch\n→ three-character reaction shot.\n\nSFX:\n\n- sleepy yawn\n- soft pineapple texture rustle\n- leafy crown movement\n- fruit decorations jingling gently.\n\n==================================================\n[SHOT 4 — 0:05.0–0:06.7]\nBERRY PRESENTS HER MASTERPIECE\n\nBerry steps dramatically into open space in front of King Pine.\n\nShe throws both tiny arms outward like a professional stylist revealing a masterpiece.\n\nDialogue:\n\nBerry:\n<d>[English] Your festival look!</d>\n\nKing Pine looks down at himself.\n\nGrape chains.\n\nBlueberry bow tie.\n\nOrange peel mustache.\n\nSparkling cape.\n\nFruit jewels.\n\nBerry gestures proudly from one decoration to another.\n\nBerry:\n\n<d>[English] Royal. Festive. Magnificent.</d>\n\nKiwi slowly covers his face with both tiny hands.\n\nKing Pine blinks.\n\nDialogue:\n\nKing Pine:\n<d>[English] I see...</d>\n\nUse a tiny suspenseful musical note.\n\n==================================================\n[SHOT 5 — 0:06.7–0:08.6]\nTHE ROYAL WALK TO THE MIRROR\n\nKing Pine slowly rises from the throne.\n\nKeep scale large and movement gentle.\n\nThe cape hangs beautifully.\n\nBerry leads him toward the tall ornate mirror.\n\nKiwi follows reluctantly.\n\nBerry walks with proud stylist confidence.\n\nKing Pine carefully adjusts the cape while walking.\n\nCamera tracks sideways.\n\nThen move behind them toward mirror.\n\nIn the reflection:\n\nfirst reveal partial silhouette of the decorated King.\n\nBerry:\n\n<d>[English] Ready?</d>\n\nKing Pine:\n<d>[English] I think so.</d>\n\nKiwi quietly:\n\n<d>[English] I’m not.</d>\n\nSmall comedic music accent.\n\n==================================================\n[SHOT 6 — 0:08.6–0:10.5]\nTHE MIRROR REVEAL\n\nKing Pine reaches the mirror.\n\nCamera shifts so the audience sees his FULL reflection.\n\nThis is the MAIN visual reveal.\n\nHe wears:\n\n- grape monocle\n- orange-peel mustache\n- blueberry bow tie\n- citrus crown jewel\n- flower decorations\n- berry medallions\n- fruit garlands\n- sparkling fruit cape\n\nThe look is:\n\nglorious,\ncolorful,\nregal,\nridiculous,\nadorable.\n\nKing Pine stares.\n\nSilence.\n\nBerry stands beside him with hopeful smile.\n\nKiwi hides partly behind Berry.\n\nKing Pine slowly touches his mustache.\n\nThen adjusts his monocle.\n\nStill silent.\n\nBerry’s confident smile becomes slightly nervous.\n\nDialogue:\n\nBerry:\n<d>[English] Well?</d>\n\nHOLD.\n\nKiwi shuts his eyes.\n\nMusic almost completely stops.\n\n==================================================\n[SHOT 7 — 0:10.5–0:12.5]\nDISASTER TURNS INTO DELIGHT\n\nKing Pine’s serious expression breaks.\n\nFirst:\n\ntiny smile.\n\nThen:\n\nbigger smile.\n\nThen:\n\nwarm, delighted laughter.\n\nKing Pine:\n\n<d>[English] I LOVE IT!</d>\n\nMusic bursts into joyful festival melody.\n\nBerry jumps with happiness.\n\nKiwi opens one eye.\n\nCompletely confused.\n\nKing Pine turns dramatically away from mirror.\n\nHe spreads the sparkling fruit cape wide.\n\nThe fruit ornaments shimmer.\n\nKing Pine proudly poses.\n\nDialogue:\n\nKing Pine:\n<d>[English] Look at me!</d>\n\nBerry:\n\n<d>[English] I knew it!</d>\n\nKiwi:\n\n<d>[English] You did?</d>\n\nKing Pine laughs.\n\nCAMERA:\n\nserious King close-up\n→ smile progression\n→ Berry reaction\n→ wide cape-spread hero shot.\n\n==================================================\n[SHOT 8 — 0:12.5–0:15]\nROYAL APPROVAL + FESTIVAL PAYOFF\n\nKing Pine gently kneels / lowers himself enough to reach tiny Berry safely.\n\nHe gives Berry a huge gentle thumbs-up or places one hand over his heart.\n\nDialogue:\n\nKing Pine:\n<d>[English] Brilliant work, little stylist.</d>\n\nBerry beams.\n\nShe performs a tiny proud bow.\n\nNearby fruit-market characters notice.\n\nOne apple gasps happily.\n\nAn orange points.\n\nA grape cluster cheers.\n\nMore fruit characters gather.\n\nWarm festival lights brighten.\n\nThe crowd begins applauding and celebrating.\n\nKing Pine stands proudly in full makeover.\n\nHe spreads the sparkling cape one final time.\n\nBerry stands beside him with hands proudly on hips.\n\nKiwi stands on Berry’s other side.\n\nKiwi looks:\n\nKing Pine\n→ cheering crowd\n→ Berry.\n\nHe cannot believe this worked.\n\nKiwi:\n\n<d>[English] I understand nothing.</d>\n\nBerry giggles.\n\nKing Pine laughs warmly.\n\nFINAL HERO COMPOSITION:\n\n- giant King Pine centered\n- full fruit makeover visible\n- sparkling cape spread\n- Berry proudly beside him\n- Kiwi baffled beside Berry\n- cheerful fruit crowd surrounding them\n- lanterns glowing overhead\n- fruit garlands framing image\n- warm festival market behind\n- golden-orange light\n- tiny sparkling fruit highlights\n\nCamera slowly pulls backward.\n\nHold the celebratory composition.\n\nFADE TO BLACK.\n\nEND STORY.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nAudio should feel:\n\n- warm\n- playful\n- family-friendly\n- festive\n- expressive\n- storybook cinematic\n\n==================================================\nBERRY VOICE\n==================================================\n\nCute youthful girl voice.\n\nQualities:\n\n- bright\n- creative\n- confident\n- playful\n- sincere\n- energetic\n\nDo not make her overly baby-like.\n\nPerformance:\n\nbeginning:\nproud stylist.\n\nKing wakes:\nbrief nervous anticipation.\n\nMirror:\nhopeful.\n\nApproval:\njoyful.\n\nImportant lines:\n\n<d>[English] Your festival look!</d>\n\n<d>[English] Royal. Festive. Magnificent.</d>\n\n<d>[English] Well?</d>\n\n<d>[English] I knew it!</d>\n\n==================================================\nKING PINE VOICE\n==================================================\n\nWarm gentle adult male fantasy voice.\n\nQualities:\n\n- deep but soft\n- sleepy at first\n- kind\n- regal\n- lovable\n- warm laugh\n- never intimidating\n\nVoice progression:\n\nwake-up:\nslow and sleepy.\n\nmirror:\nquiet curiosity.\n\napproval:\nbig warm delight.\n\nCritical line:\n\n<d>[English] I LOVE IT!</d>\n\nDeliver with genuine childlike delight despite his deep voice.\n\nNot shouted aggressively.\n\nFinal approval:\n\n<d>[English] Brilliant work, little stylist.</d>\n\nshould feel affectionate and sincere.\n\n==================================================\nKIWI VOICE\n==================================================\n\nSmall high-energy companion voice.\n\nQualities:\n\n- nervous\n- fast\n- expressive\n- lovable\n- slightly dramatic\n- excellent comic timing\n\nPerformance:\n\nbeginning:\npure panic.\n\nwake-up:\nfearful whisper.\n\nmirror:\nbracing for disaster.\n\napproval:\ncomplete disbelief.\n\nFinal:\n\n<d>[English] I understand nothing.</d>\n\nDeliver slowly, exhausted and baffled.\n\n==================================================\nFRUIT CROWD AUDIO\n==================================================\n\nKeep supporting voices simple.\n\nUse:\n\n- cheerful gasps\n- little laughs\n- soft applause\n- celebratory “wow!”\n- happy fruit chatter\n\nDo not let crowd dialogue overpower main characters.\n\n==================================================\nMARKET AUDIO\n==================================================\n\nMaintain from Part 1:\n\n- gentle market ambience\n- breeze\n- wooden stall creaks\n- basket movement\n- leaves\n- hanging lantern chimes\n- fruit-garland movements\n\nDuring ending:\n\nincrease:\n\n- applause\n- festival bells\n- cheerful chatter\n- tiny tambourine-like fruit sounds\n\n==================================================\nMAKEOVER PROP SOUND DESIGN\n==================================================\n\nCape placement:\n\nsoft fabric SWISH\n+\ntiny berry-bead chimes.\n\nMonocle touch:\n\nsmall grape click.\n\nOrange-peel mustache touch:\n\ngentle peel rustle.\n\nCape hero spread:\n\nlarger elegant fabric SWOOSH\n+\nsoft glitter shimmer.\n\nMirror:\n\ntiny clean glass sparkle when full reveal appears.\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL whimsical festival score.\n\nInstrumentation:\n\n- pizzicato strings\n- marimba\n- clarinet\n- flute\n- soft xylophone\n- tiny bells\n- hand percussion\n- acoustic plucks\n- warm strings\n- light accordion-style texture if subtle\n- festive tambourine near ending\n\nMUSIC ARC:\n\n0:00–0:02\n“finishing masterpiece” motif.\n\n0:02–0:04\ncomic suspense as King wakes.\n\n0:04–0:07\ncurious sleepy motif.\n\n0:07–0:09\nslow royal mirror walk.\n\n0:09–0:10.5\nalmost complete silence / tension.\n\n0:10.5–0:12.5\nbig joyful comedy release.\n\n0:12.5–0:15\nwarm fruit-festival celebration.\n\nEnd on a happy resolved chord.\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPART 2 camera progression:\n\n0–2 sec:\nclose finishing-detail coverage.\n\n2–5 sec:\nwake-up reaction close-ups.\n\n5–7 sec:\ncharacter presentation.\n\n7–9 sec:\nslow royal tracking toward mirror.\n\n9–10.5 sec:\nmirror reveal + reaction stillness.\n\n10.5–12.5 sec:\nwider energetic comedy payoff.\n\n12.5–15 sec:\nlarge celebratory hero composition.\n\nUse:\n\n- macro cape details\n- gentle push-in\n- eye-level reaction close-ups\n- over-the-shoulder mirror view\n- reflection composition\n- scale-revealing wide shots\n- slow pull-back\n- subtle rack focus\n- gentle cinematic parallax\n\nAvoid:\n\n- shaky handheld\n- crash zoom\n- spinning camera\n- extreme fisheye\n- frantic editing\n- excessive motion blur\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished painterly 2D storybook animation.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n\n- charming hand-drawn ink contours\n- watercolor / gouache-inspired rendering\n- soft painterly shading\n- warm fruit textures\n- expressive rounded faces\n- beautiful wooden market details\n- soft glow from lanterns\n- subtle cinematic depth\n- natural fabric movement\n- tasteful sparkle effects\n- gentle squash and stretch\n- premium family-animation finish\n\nThe final video should feel like:\n\nTHE UPLOADED STORYBOARD ILLUSTRATIONS CAME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photoreal food\n- realistic fruit photography\n- live action\n- generic CGI\n- flat vector animation\n- horror\n\n==================================================\nCOLOR / LIGHTING ARC\n==================================================\n\nPART 2 palette:\n\nBERRY:\nstrawberry red + leafy green.\n\nKING PINE:\nwarm pineapple gold + leafy green.\n\nKIWI:\nolive green / fuzzy brown-green.\n\nMAKEOVER:\npurple grapes,\nblueberry violet,\norange citrus,\nsoft flower pink,\nberry red,\nwarm gold.\n\nENVIRONMENT:\nwarm honey gold,\namber,\nsoft brown,\nmuted green.\n\nLIGHTING PROGRESSION:\n\n0–5 sec:\nsame warm late-afternoon glow from Part 1.\n\n5–9 sec:\nintroduce more visible festival lanterns.\n\n9–10.5 sec:\nmirror scene slightly calmer and more focused.\n\n10.5–12.5 sec:\nincrease sparkling fruit highlights after King smiles.\n\n12.5–15 sec:\nbrightest, richest, warmest lighting of the entire story.\n\nFinal shot should feel like:\n\nthe festival officially started because King Pine loved the makeover.\n\n==================================================\nMOTION RULES — BERRY\n==================================================\n\nUse:\n\n- tiny energetic steps\n- careful cape placement\n- expressive hand movements\n- proud presentation stance\n- tiny nervous pause\n- joyful hop when King approves\n- proud final bow\n- hands-on-hips hero pose\n\nBerry must remain physically tiny relative to King Pine.\n\n==================================================\nMOTION RULES — KING PINE\n==================================================\n\nKing Pine is enormous and gentle.\n\nWake-up:\n\nleaf crown twitch\n→ eyelid movement\n→ slow eyes opening\n→ big sleepy blink\n→ slow sit-up.\n\nWalking:\n\nheavy but soft,\nslow,\nfriendly.\n\nMirror:\n\nsubtle facial acting.\n\nApproval:\n\nlarge warm smile,\ngentle laugh,\ncape spread.\n\nNever make his motion aggressive.\n\n==================================================\nMOTION RULES — KIWI\n==================================================\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086360699505451093"
  },
  {
    "id": "x-2086360167176954269",
    "title": "小草莓唤醒菠萝王 · 上篇",
    "titleEn": "The Little Strawberry Wakes the Pineapple King · Part 1",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《小草莓唤醒菠萝王 · 上篇》的 H3 成片。本页保留原始 X 来源链接。来源未公开完整 Prompt，本库不反推或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Little Strawberry Wakes the Pineapple King · Part 1. This page retains the original X source link. The source did not publish a complete Prompt, so this library does not infer or complete one. Hosted-video output: 15s · 1280×720.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2086360167176954269",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T07:53:21.209Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086360167176954269.mp4",
    "posterUrl": "/posters/x/2086360167176954269.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "food"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for:\n\n“THE LITTLE STRAWBERRY AND THE SLEEPY PINEAPPLE KING”\n\nas the PRIMARY visual reference for:\n\n- exact character identities\n- proportions\n- expressions\n- fruit textures\n- costume/accessory design\n- fruit-market environment\n- throne design\n- lighting\n- colors\n- scale relationships\n- props\n- story progression\n\nGenerate a FINISHED cinematic animated short from the storyboard.\n\nDO NOT animate the storyboard sheet itself.\n\nThe finished video must NOT contain:\n\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character reference boxes\n- prop-reference illustrations\n- title/header\n- white storyboard paper\n- page numbers\n- color-note sections\n\nThe uploaded storyboard locks:\n\n1. Berry’s exact strawberry-girl design\n2. Berry’s red strawberry body\n3. Berry’s green leafy top\n4. Berry’s expressive large eyes\n5. Berry’s tiny scale\n6. Kiwi’s exact fuzzy round design\n7. Kiwi’s green-brown coloring\n8. Kiwi’s exaggerated reaction expressions\n9. King Pine’s exact giant pineapple design\n10. King Pine’s golden pineapple texture\n11. King Pine’s leafy green crown top\n12. King Pine’s giant scale\n13. King Pine’s peaceful sleeping expression\n14. purple cushioned fruit-crate throne\n15. Berry’s woven decoration basket\n16. grape monocle\n17. orange-peel mustache\n18. blueberry bow tie\n19. citrus crown jewel\n20. fruit garlands and berry decorations\n21. sparkling fruit cape teaser\n22. same warm magical fruit market\n23. same golden afternoon lighting\n24. Part 1 cliffhanger before King Pine wakes\n\nTranslate the 15 storyboard panels into approximately\n7–8 COHERENT CINEMATIC SHOTS.\n\nDO NOT create fifteen frantic one-second hard cuts.\n\nUse:\n\nsetup\n→ discovery\n→ first decoration\n→ makeover escalation\n→ Kiwi panic\n→ bigger makeover\n→ final cape reveal\n→ cliffhanger.\n\n==================================================\nVIDEO GOAL\n==================================================\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:00–0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D storybook animation\n\nGenre:\ncozy fruit fantasy comedy\n\nAudience:\nfamily-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + character sounds + market ambience + playful fruit-decoration sounds + original music\n\nEMOTIONAL ARC:\n\ncozy festival anticipation\n→ Berry discovers sleepy king\n→ creative idea\n→ harmless decorating\n→ Kiwi becomes concerned\n→ makeover grows ridiculous\n→ Berry gets even more ambitious\n→ final royal cape cliffhanger\n\nCORE STORY QUESTION:\n\n“What will happen when the Pineapple King wakes up?”\n\n==================================================\nPART 1 STORY LIMIT\n==================================================\n\nPART 1 ONLY.\n\nSTART:\n\nThe magical fruit market prepares for the evening festival.\n\nKing Pine is peacefully sleeping on his throne.\n\nBerry arrives with Kiwi and a basket of decorations.\n\nEND:\n\nBerry reveals the FINAL sparkling fruit cape.\n\nKing Pine is STILL ASLEEP.\n\nKiwi realizes Berry is about to make the makeover even bigger.\n\nCUT TO BLACK.\n\nDO NOT show:\n\n- King Pine waking\n- King Pine opening his eyes fully\n- King Pine standing\n- mirror reveal\n- King Pine judging the makeover\n- King Pine laughing\n- King Pine approving Berry\n- festival crowd arriving\n- final celebration\n\nAll of those belong to PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — BERRY\n==================================================\n\nMatch the uploaded storyboard EXACTLY.\n\nBERRY:\n\nA tiny anthropomorphic strawberry girl.\n\nAppearance:\n\n- bright strawberry-red fruit body\n- visible tiny strawberry seed details\n- adorable rounded face\n- very large expressive dark eyes\n- small smiling mouth\n- rosy cheerful cheeks\n- green leafy crown/top resembling soft leafy hair\n- tiny arms\n- tiny legs\n- small red dress-like lower silhouette\n- petite childlike proportions\n\nScale:\n\nVERY SMALL compared with King Pine.\n\nApproximately small enough that:\n\n- King Pine’s face is much larger than Berry’s entire body\n- Berry must reach/climb to decorate him\n- his throne feels enormous beside her\n\nPersonality:\n\n- creative\n- cheerful\n- confident\n- mischievous\n- imaginative\n- completely convinced she has excellent taste\n- gentle and caring\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nhappy and purposeful\n\n3–5 sec:\ncreative realization\n\n5–8 sec:\nfocused tiny stylist\n\n8–11 sec:\nincreasingly proud\n\n11–13 sec:\nfully committed to the makeover\n\n13–15 sec:\nhuge “one more thing!” excitement\n\nCRITICAL:\n\nMaintain exactly:\n\n- same Berry face\n- same strawberry shape\n- same green leaves\n- same proportions\n- same color\n- same size\n- same visual styling\n\nNo redesign.\n\n==================================================\nCHARACTER IDENTITY LOCK — KING PINE\n==================================================\n\nMatch storyboard EXACTLY.\n\nKING PINE:\n\nThe giant sleepy Pineapple King.\n\nAppearance:\n\n- enormous rounded pineapple body\n- warm golden-yellow skin\n- recognizable diamond pineapple texture\n- large soft face integrated into fruit body\n- thick gentle eyebrows\n- large nose\n- soft round cheeks\n- huge leafy green pineapple crown\n- massive friendly scale\n- seated on purple cushioned fruit throne\n\nPersonality:\n\n- gentle\n- regal\n- lovable\n- incredibly deep sleeper\n\nPART 1 STATE:\n\nASLEEP THROUGHOUT THE ENTIRE VIDEO.\n\nAllowed:\n\n- slow breathing\n- tiny peaceful snore\n- subtle body rise/fall\n- slight leaf movement\n- faint sleepy smile\n- tiny nose twitch if desired\n\nNOT ALLOWED:\n\n- fully opening eyes\n- speaking\n- deliberately moving\n- standing\n- realizing he is being decorated\n- reacting to makeover\n\nAll awakening belongs to PART 2.\n\n==================================================\nCHARACTER IDENTITY LOCK — KIWI\n==================================================\n\nMatch storyboard exactly.\n\nKIWI:\n\nTiny fuzzy kiwi companion.\n\nAppearance:\n\n- small rounded kiwi-fruit body\n- muted olive / green-brown fuzzy texture\n- large expressive eyes\n- tiny dark limbs\n- small expressive mouth\n- very small compared with King Pine\n- approximately Berry’s companion scale\n\nPersonality:\n\n- nervous\n- cautious\n- practical\n- dramatic\n- loyal\n- very expressive\n\nPART 1 PERFORMANCE:\n\ncalm\n→ suspicious\n→ worried\n→ alarmed\n→ full panic\n→ desperate wake-up attempt\n→ defeated\n→ horrified by final cape idea\n\nKiwi is the PRIMARY comedy reaction character.\n\n==================================================\nPROP LOCK — BERRY’S DECORATION BASKET\n==================================================\n\nMaintain the same woven basket from the storyboard.\n\nIt contains:\n\n- grapes\n- blueberries\n- orange peel curls\n- citrus slices\n- flowers\n- ribbons\n- fruit medallions\n- berry chains\n- small fruit jewels\n- sparkling fruit cape\n\nThe basket should visibly become less full as Berry decorates King Pine.\n\nDo not randomly change its design.\n\n==================================================\nMAKEOVER PROP CONTINUITY\n==================================================\n\nThe decorations must accumulate.\n\nDO NOT reset King Pine between shots.\n\nOrder:\n\n1. grape monocle\n2. orange-peel mustache\n3. blueberry bow tie\n4. fruit/flower decorations\n5. crown jewel\n6. berry/citrus garlands and medallions\n7. final sparkling fruit cape TEASER\n\nOnce an item has been placed:\n\nKEEP IT THERE.\n\nBy the end:\n\nKing Pine should look gloriously overdecorated while remaining asleep.\n\n==================================================\nGRAPE MONOCLE\n==================================================\n\nA playful circular monocle constructed from small purple grapes.\n\nPlaced gently around / beside one closed eye.\n\nNo glass required.\n\nIt remains in position once attached.\n\n==================================================\nORANGE-PEEL MUSTACHE\n==================================================\n\nA curled orange-peel decoration shaped like a fancy royal mustache.\n\nWarm orange color.\n\nSoft flexible peel texture.\n\nPlaced gently beneath King Pine’s nose.\n\nCute and harmless.\n\n==================================================\nBLUEBERRY BOW TIE\n==================================================\n\nA dark blueberry-purple bow tie / berry ribbon decoration.\n\nPlaced under King Pine’s face.\n\nClearly visible in later shots.\n\n==================================================\nCITRUS CROWN JEWEL\n==================================================\n\nSmall ornamental royal accent made from:\n\n- orange slice\n- tiny berry accents\n- golden-yellow fruit details\n\nPlaced among King Pine’s leafy crown.\n\n==================================================\nFINAL SPARKLING FRUIT CAPE\n==================================================\n\nThis is the PART 1 CLIFFHANGER PROP.\n\nDesign:\n\n- translucent rosy-pink / citrus-gold fabric\n- tiny glowing sparkles\n- orange-slice medallions\n- berries\n- hanging fruit beads\n- tiny fruit garlands\n- elegant storybook fantasy appearance\n\nAt the end of Part 1:\n\nBerry has revealed/unfurled it.\n\nDO NOT complete the full cape placement.\n\nSave that for the beginning of PART 2.\n\n==================================================\nENVIRONMENT LOCK — FRUIT FESTIVAL MARKET\n==================================================\n\nMaintain the exact warm magical fruit-market world from the storyboard.\n\nInclude:\n\n- wooden fruit stalls\n- stacked wooden crates\n- baskets overflowing with fruit\n- oranges\n- grapes\n- apples\n- berries\n- citrus slices\n- hanging fruit garlands\n- colorful bunting\n- warm festival lantern strings\n- small plants\n- wood shelving\n- cozy market details\n- gentle depth into background\n\nKing Pine’s throne:\n\n- large fruit-crate inspired base\n- purple cushioned seat\n- large enough for his enormous body\n\nLighting:\n\nwarm golden afternoon approaching festival evening.\n\nUse:\n\n- soft amber sunlight\n- glowing string lights\n- warm fruit reflections\n- gentle rim lighting\n- soft shadows\n\nNo modern objects.\n\nNo humans.\n\nNo branded signs.\n\n==================================================\nWORLD / MOTION LOGIC\n==================================================\n\nThe story must follow clear cause-and-effect.\n\nBerry arrives\n→ sees sleeping King Pine\n\nKing looks plain to Berry\n→ Berry gets idea\n\nBerry has decoration basket\n→ begins makeover\n\none decoration succeeds\n→ Berry gains confidence\n\nBerry adds more\n→ Kiwi becomes worried\n\nKiwi tries to wake King Pine\n→ King continues sleeping\n\nBerry sees makeover\n→ thinks it still needs more\n\nbasket contains final cape\n→ Berry reveals cape\n\nKiwi realizes Berry is not finished\n→ cliffhanger.\n\n==================================================\nINTEGRATED_MULTIMODAL_DESCRIPTION\n==================================================\n\n[SHOT 1 — 0:00–0:02]\nTHE FRUIT MARKET BEFORE THE FESTIVAL\n\nOpen with a beautiful wide establishing shot.\n\nWarm golden afternoon light pours through a magical fruit market.\n\nShow:\n\n- colorful fruit stalls\n- wooden crates\n- citrus garlands\n- hanging festival lanterns\n- baskets of berries\n- fruit banners\n- cozy decorations\n\nCamera glides slowly through the market.\n\nAt the far end:\n\nreveal GIANT KING PINE peacefully sleeping on his purple cushioned fruit throne.\n\nHis leafy crown gently leans to one side.\n\nSoft sleepy breathing.\n\nA tiny cute snore.\n\nBerry enters foreground carrying an overflowing woven decoration basket.\n\nKiwi waddles beside her.\n\nBerry looks toward King Pine.\n\nDialogue:\n\nBerry, cheerful:\n<d>[English] There he is!</d>\n\nKiwi:\n<d>[English] He’s sleeping.</d>\n\nBerry:\n<d>[English] Perfect.</d>\n\nKiwi slowly looks toward Berry.\n\nSuspicious.\n\nCAMERA:\n\nmarket wide\n→ slow forward glide\n→ King Pine reveal\n→ medium Berry + Kiwi.\n\nAUDIO:\n\n- gentle market ambience\n- fruit basket movement\n- distant lantern chimes\n- soft breeze\n- tiny sleepy snore\n\nMUSIC:\n\nlight pizzicato strings,\nsoft marimba,\nflute,\ngentle whimsical percussion.\n\n==================================================\n\n[SHOT 2 — 0:02–0:04]\nBERRY GETS A VERY CREATIVE IDEA\n\nMedium-close view.\n\nBerry walks closer to enormous King Pine.\n\nScale is important.\n\nShe looks tiny beside his giant face.\n\nBerry studies him seriously.\n\nFinger under chin.\n\nShe looks at:\n\nKing Pine\n→ her basket\n→ King Pine again.\n\nA tiny sparkle appears in her eyes.\n\nBerry smiles.\n\nDialogue:\n\nBerry:\n<d>[English] Hmm...</d>\n\nKiwi:\n<d>[English] Berry?</d>\n\nBerry:\n<d>[English] He needs something.</d>\n\nKiwi:\n<d>[English] A nap?</d>\n\nBerry shakes head.\n\n<d>[English] Style.</d>\n\nQuick comedic music accent.\n\nCamera gently pushes toward Berry’s delighted “big idea” expression.\n\n==================================================\n\n[SHOT 3 — 0:04–0:06]\nTHE FIRST TWO DECORATIONS\n\nBerry climbs / reaches gently toward King Pine’s face.\n\nNo dangerous climbing.\n\nUse a nearby crate or throne edge as a safe step.\n\nFirst:\n\nBerry carefully places a circular purple grape monocle around one of King Pine’s closed eyes.\n\nTiny decorative:\n\nTING.\n\nBerry leans back proudly.\n\nThen she reaches into the basket.\n\nPulls out a curled orange peel.\n\nShe gently shapes it beneath King Pine’s nose into an elaborate royal mustache.\n\nKing Pine remains completely asleep.\n\nHis breathing continues.\n\nBerry whispers:\n\n<d>[English] Much better.</d>\n\nKiwi stares.\n\nDialogue:\n\nKiwi:\n<d>[English] Is it?</d>\n\nCAMERA:\n\nmacro grape monocle placement\n→ Berry reaction\n→ orange-peel close-up\n→ wider face reveal.\n\nSFX:\n\n- grape bead clicks\n- orange peel rustle\n- tiny magical chimes\n- soft pineapple snore\n\n==================================================\n\n[SHOT 4 — 0:06–0:08]\nNOW FOR THE BOW TIE\n\nBerry retrieves a large blueberry-purple bow tie.\n\nShe gently attaches it beneath King Pine’s face.\n\nThe royal look now includes:\n\n- grape monocle\n- orange-peel mustache\n- blueberry bow tie\n\nBerry beams.\n\nKiwi’s eyes widen.\n\nCut closer to Kiwi.\n\nHis expression progresses:\n\nconcern\n→ disbelief\n→ full panic.\n\nKiwi:\n<d>[English] Berry...</d>\n\nBerry:\n<d>[English] Hmm?</d>\n\nKiwi:\n<d>[English] I think that’s enough.</d>\n\nBerry looks back at King Pine.\n\nThen at the basket.\n\nBerry:\n\n<d>[English] Almost.</d>\n\nKiwi’s jaw drops.\n\nMUSIC:\n\nintroduce faster playful pizzicato rhythm.\n\n==================================================\n\n[SHOT 5 — 0:08–0:10.5]\nTHE MAKEOVER ESCALATES\n\nCreate a smooth SHORT decorating montage.\n\nDo NOT use frantic cuts.\n\nBerry becomes fully focused like a tiny professional stylist.\n\nShow clear progression:\n\nBerry adds several berries around King Pine’s upper body.\n\nThen a tiny flower.\n\nThen one citrus slice.\n\nThen decorative grape loops.\n\nThen berry medallions.\n\nThen fruit ribbons.\n\nEach addition stays physically present.\n\nKing Pine becomes progressively more spectacular.\n\nKiwi moves around Berry trying to get her attention.\n\nDialogue:\n\nKiwi:\n<d>[English] Berry!</d>\n\nBerry:\n<d>[English] Hold this.</d>\n\nBerry hands Kiwi a small fruit ribbon.\n\nKiwi automatically holds it.\n\nHe realizes what happened.\n\nKiwi:\n<d>[English] Wait—why am I helping?!</d>\n\nBerry continues happily.\n\nCAMERA:\n\nmedium decorating shot\n→ close prop insert\n→ moving side shot across accumulating decorations\n→ Kiwi reaction.\n\nMUSIC:\n\nplayful escalating comedy rhythm.\n\nSFX:\n\n- berry clicks\n- ribbon rustles\n- tiny chimes\n- flower “pop”\n- fruit decoration taps\n\n==================================================\n\n[SHOT 6 — 0:10.5–0:12.2]\nKIWI ATTEMPTS A RESCUE\n\nBerry reaches for a tiny citrus crown jewel.\n\nWhile Berry is focused:\n\nKiwi looks at sleeping King Pine.\n\nThen at the enormous amount of decorations.\n\nKiwi decides:\n\nwake him.\n\nKiwi waddles / hops close to King Pine’s face.\n\nHe gently calls:\n\n<d>[English] Your Majesty?</d>\n\nNothing.\n\nKiwi tries slightly louder:\n\n<d>[English] Your Majesty!</d>\n\nKing Pine gives one adorable sleepy snore.\n\nHis body rises and falls peacefully.\n\nThe grape monocle stays perfectly in place.\n\nKiwi freezes.\n\nThen a tiny defeated expression.\n\nMeanwhile Berry gently places the citrus crown jewel among King Pine’s leafy crown.\n\nBerry:\n\n<d>[English] Perfect!</d>\n\nCAMERA:\n\nKiwi medium\n→ sleeping King Pine close-up\n→ Kiwi listening\n→ crown jewel placement.\n\nNo awakening.\n\n==================================================\n\n[SHOT 7 — 0:12.2–0:13.7]\nTHE FESTIVAL LEGEND\n\nPull back.\n\nKing Pine is now spectacularly decorated.\n\nVisible:\n\n- grape monocle\n- orange-peel mustache\n- blueberry bow tie\n- berry chains\n- orange slices\n- flowers\n- fruit medallions\n- citrus crown jewel\n- colorful garlands\n\nBerry stands proudly beside her masterpiece.\n\nHands on hips.\n\nHuge grin.\n\nDialogue:\n\nBerry:\n<d>[English] Ta-da!</d>\n\nKiwi looks upward at King Pine.\n\nLong silence.\n\nKiwi:\n<d>[English] He looks...</d>\n\nBerry leans forward excitedly.\n\nKiwi:\n\n<d>[English] ...very decorated.</d>\n\nBerry nods proudly.\n\n<d>[English] Exactly.</d>\n\nUse a gentle wide reveal emphasizing:\n\ntiny Berry\n+\ntiny Kiwi\n+\nmassive decorated King Pine.\n\n==================================================\n\n[SHOT 8 — 0:13.7–0:15]\nONE LAST IDEA — CLIFFHANGER\n\nKiwi collapses into a harmless exhausted seated / flattened pose near the decoration basket.\n\nBerry turns toward the basket.\n\nShe notices something inside.\n\nHer eyes widen.\n\nA sparkle appears.\n\nBerry:\n\n<d>[English] Oh!</d>\n\nKiwi slowly looks toward her.\n\nInstant concern.\n\nKiwi:\n<d>[English] No.</d>\n\nBerry reaches inside.\n\nKiwi:\n<d>[English] Berry...</d>\n\nBerry pulls out and dramatically unfurls the FINAL SPARKLING FRUIT CAPE.\n\nThe cape opens beautifully in the warm market light.\n\nShow:\n\n- translucent rosy fruit fabric\n- orange-slice medallions\n- berry beads\n- small hanging fruit decorations\n- gentle glittering points\n\nBerry’s face becomes pure creative delight.\n\nBerry:\n\n<d>[English] One last thing!</d>\n\nKiwi’s eyes become enormous.\n\nKiwi:\n<d>[English] ONE MORE?!</d>\n\nKing Pine remains completely asleep in the background.\n\nCRITICAL:\n\nDO NOT put the cape fully onto King Pine.\n\nBerry is merely holding/unfurling it.\n\nCamera slowly pushes toward:\n\nBerry\n+\nsparkling cape\n+\npanicked Kiwi\n+\nsleeping decorated King Pine behind.\n\nMusic ends on playful suspense sting.\n\nCUT TO BLACK.\n\nEND PART 1.\n\n==================================================\nAUDIO DESCRIPTION\n==================================================\n\nGenerate synchronized native stereo audio.\n\nThe sound should feel:\n\n- cozy\n- playful\n- warm\n- energetic\n- family-friendly\n- storybook-like\n\n==================================================\nBERRY VOICE\n==================================================\n\nCute youthful girl voice.\n\nQualities:\n\n- bright\n- cheerful\n- creative\n- confident\n- expressive\n- warm\n- slightly mischievous\n\nDo NOT make her overly baby-like.\n\nDialogue should be clear and naturally paced.\n\nMost important final line:\n\n<d>[English] One last thing!</d>\n\nDeliver with delighted creative excitement.\n\n==================================================\nKIWI VOICE\n==================================================\n\nCute small companion voice.\n\nSlightly higher pitch than Berry.\n\nQualities:\n\n- nervous\n- fast\n- expressive\n- practical\n- comic\n- lovable\n\nPerformance arc:\n\nbeginning:\nmild suspicion\n\nmiddle:\ngrowing concern\n\nlater:\npanic\n\nending:\ncomplete disbelief\n\nFinal:\n\n<d>[English] ONE MORE?!</d>\n\nis the comedy cliffhanger punctuation.\n\n==================================================\nKING PINE AUDIO\n==================================================\n\nNO spoken dialogue in Part 1.\n\nOnly:\n\n- gentle breathing\n- tiny cute snores\n- occasional sleepy nose sound\n- subtle leaf rustle\n\nNo wake-up dialogue.\n\nNo reaction words.\n\n==================================================\nFRUIT MARKET AMBIENCE\n==================================================\n\nUse:\n\n- gentle warm market ambience\n- soft breeze\n- distant fruit stall activity\n- tiny wooden creaks\n- baskets moving\n- lantern chimes\n- hanging garland movement\n- subtle leaves rustling\n\nThe market should feel populated by life without requiring additional speaking characters.\n\n==================================================\nMAKEOVER SOUND DESIGN\n==================================================\n\nEach decoration should have a soft signature.\n\nGrape monocle:\ntiny grape bead click + delicate TING.\n\nOrange peel:\nsoft peel curl / fabric-like rustle.\n\nBow tie:\nlight berry tap + ribbon flick.\n\nFlowers:\ntiny soft pop.\n\nGarlands:\ngentle fruit bead rattles.\n\nCrown jewel:\nbright but delicate magical chime.\n\nCape:\nsoft fabric unfurl +\nseveral sparkling bell tones.\n\nAvoid:\n\n- loud impacts\n- slapstick crashes\n- aggressive sounds\n\n==================================================\nNON-DIEGETIC MUSIC\n==================================================\n\nCreate an ORIGINAL whimsical storybook score.\n\nInstrumentation:\n\n- pizzicato strings\n- marimba\n- soft xylophone\n- flute\n- clarinet\n- light hand percussion\n- tiny bells\n- soft acoustic plucks\n- warm strings\n\nMUSIC ARC:\n\n0:00–0:02\ncozy festival introduction.\n\n0:02–0:04\ncurious “idea” motif.\n\n0:04–0:06\ntiny stylist playfulness.\n\n0:06–0:08\nKiwi concern begins.\n\n0:08–0:10.5\nfaster escalating makeover comedy.\n\n0:10.5–0:12.2\nbrief sneaky Kiwi rescue motif.\n\n0:12.2–0:13.7\nwarm mini triumph.\n\n0:13.7–0:15\nmischievous suspense rise.\n\nEnd with comic musical sting after:\n\n<d>[English] ONE MORE?!</d>\n\n==================================================\nCAMERA LANGUAGE\n==================================================\n\nPart 1 camera progression:\n\n0–2 sec:\nwarm environmental wide.\n\n2–4 sec:\nintimate character comedy.\n\n4–8 sec:\ndecoration details + reaction coverage.\n\n8–12 sec:\nmore energetic makeover movement.\n\n12–13.7 sec:\nlarge masterpiece reveal.\n\n13.7–15 sec:\ntight cliffhanger reactions.\n\nUse:\n\n- slow dolly\n- gentle tracking\n- macro fruit-detail close-ups\n- reaction close-ups\n- scale-revealing wide shots\n- rack focus\n- gentle push-ins\n- clean eye-level comedy framing\n\nAvoid:\n\n- camera shake\n- crash zoom\n- chaotic editing\n- extreme motion blur\n- Dutch angles\n- 360-degree spins\n\n==================================================\nVISUAL STYLE\n==================================================\n\nPremium finished painterly 2D storybook animation.\n\nMatch the storyboard exactly in spirit.\n\nUse:\n\n- hand-drawn ink-like contours\n- watercolor / gouache inspired rendering\n- soft painterly shadows\n- beautiful fruit texture\n- warm market lighting\n- expressive rounded faces\n- storybook environmental detail\n- subtle depth of field\n- soft cinematic bloom\n- gentle squash and stretch\n- natural secondary motion\n\nThe finished video should look like:\n\nTHE STORYBOARD ILLUSTRATIONS HAVE COME TO LIFE.\n\nNOT:\n\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealistic fruit\n- live action\n- realistic food photography\n- flat vector art\n- generic 3D CGI\n- horror\n\n==================================================\nCOLOR / LIGHTING\n==================================================\n\nPrimary palette:\n\nBERRY:\n- strawberry red\n- leafy green\n\nKING PINE:\n- warm pineapple gold\n- yellow\n- leafy green\n\nKIWI:\n- olive green\n- fuzzy brown-green\n\nDECORATIONS:\n- purple grapes\n- blueberry blue/purple\n- citrus orange\n- flower pink\n- berry red\n- warm gold\n\nENVIRONMENT:\n- honey gold\n- soft wood brown\n- warm amber\n- muted green\n- creamy highlights\n\nLighting progression:\n\n0–4 sec:\nwarm natural afternoon.\n\n4–10 sec:\nsoft golden light + colorful fruit accents.\n\n10–13 sec:\nmore sparkling festival highlights.\n\n13–15 sec:\nfinal cape receives the brightest magical sparkle.\n\nDo NOT use harsh neon lighting.\n\n==================================================\nMOTION RULES — BERRY\n==================================================\n\nBerry moves with:\n\n- tiny energetic steps\n- quick head turns\n- expressive hand gestures\n- careful decoration placement\n- playful little hops\n- proud hands-on-hips posture\n\nWhen handling decorations:\n\nmake actions physically clear.\n\nNo teleporting props.\n\nNo impossible floating decorations unless held or attached.\n\n==================================================\nMOTION RULES — KING PINE\n==================================================\n\nKing Pine stays asleep.\n\nUse:\n\n- slow breathing\n- tiny chest/body movement\n- subtle leafy crown movement\n- slight sleepy smile\n- soft snoring\n\nHe must remain stable while Berry decorates him.\n\nDo NOT awaken.\n\n==================================================\nMOTION RULES — KIWI\n==================================================\n\nUse expressive comedy:\n\n- nervous waddling\n- quick small hops\n- arms spreading\n- wide eyes\n- head shaking\n- tiny panic pacing\n- gentle wake-up attempt\n- exhausted collapse\n\nKeep physical comedy harmless.\n\n==================================================\nCONTINUITY RULES\n==================================================\n\nMaintain exactly:\n\nONE Berry.\nONE Kiwi.\nONE King Pine.\n\nSame:\n\n- character designs\n- colors\n- sizes\n- market\n- throne\n- basket\n- decorations\n- lighting direction\n\nKing Pine stays seated / reclined in the same throne area.\n\nBerry physically adds decorations.\n\nDo NOT instantly transform him.\n\nEvery decoration persists.\n\nBy the final shot:\n\nKing Pine must still visibly wear:\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2086323087092470161",
    "title": "归还月亮的少年 · 第四幕",
    "titleEn": "The Boy Who Returned the Moon · Chapter 4",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《归还月亮的少年 · 第四幕》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Boy Who Returned the Moon · Chapter 4. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 4 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, scale, costume, prop, lighting, cinematography, emotional-performance, and narrative reference.\n\nThis sequence continues IMMEDIATELY from PART 3.\n\nPART 3 ended with:\n- the moon almost completely dark\n- only a faint silver rim remaining\n- the Living Bone Bridge still spanning the canyon\n- Nyx magically exhausted but physically unharmed\n- Zara pointing toward the Moon Shrine\n- Kael holding the ONLY glowing moon fragment securely in both hands\n- Zara shouting:\n  <d>[English] KAEL! GO!</d>\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- reference drawings\n- activation diagrams\n- paper background\n- storyboard header/footer\n- color swatches\n- annotation text\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact face, hair, costume and proportions\n2. Zara’s exact face, enormous curly hairstyle and blue rib-pattern shirt\n3. Nyx’s exact black dragon anatomy\n4. Nyx’s enormous amber eyes and luminous grin\n5. the ONLY silver moon fragment\n6. the almost-dark damaged moon\n7. the Living Bone Bridge\n8. the Moon Shrine architecture\n9. the impossibly high fragment receptacle\n10. Nyx giving Kael the final lift\n11. Kael inserting the fragment\n12. silver-blue shrine activation\n13. moonlight wave crossing the valley\n14. gradual moon restoration\n15. restored Noctara\n16. skeleton guardian → moon-firefly transformation\n17. Zara’s emotional reaction\n18. Kael and Nyx reconciliation\n19. final lakeside callback\n20. final Kael + Zara + Nyx friendship composition\n\nTranslate the storyboard’s 15 one-second panels into approximately **9 coherent cinematic shots**.\n\nDO NOT create fifteen frantic one-second cuts.\n\nUse:\n\nurgent action\n→ obstacle\n→ trust\n→ sacred activation\n→ enormous magical payoff\n→ emotional release\n→ reconciliation\n→ quiet final comedy.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:45–1:00 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nfantasy adventure finale / emotional magical payoff\n\nDialogue:\nnative synchronized English\n\nAudio:\nnative dialogue + environmental sound + moon magic + guardian transformation + original cinematic score\n\nEMOTIONAL ARC:\n\ndesperate urgency\n→ frustration\n→ unexpected help\n→ complete trust\n→ sacred wonder\n→ triumphant restoration\n→ emotional release\n→ realization\n→ friendship\n→ gentle comedy\n\nCORE MESSAGE:\n\nKael began the story believing Nyx was stealing the light.\n\nHe ends the story understanding that Nyx was protecting it.\n\nAnd Kael began the story watching the moon alone.\n\nHe ends it watching with friends.\n\n--------------------------------------------------\n\n[PART 4 STORY LIMIT]\n\nThis is the FINAL CHAPTER.\n\nSTART:\n\nKael has just caught the moon fragment.\n\nZara tells him to go.\n\nEND:\n\nThe moon is fully restored.\n\nThe guardians have peacefully become silver-blue moon-fireflies.\n\nKael, Zara and Nyx sit together beside the moonlit lake.\n\nThe story must feel COMPLETELY resolved.\n\nNo cliffhanger.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch Parts 1–3 and the uploaded PART 4 storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- pale-to-medium cool-toned skin\n- youthful angular face\n- large expressive dark eyes\n- thick dark eyebrows\n- swept-back tousled black hair\n- recognizable pointed windswept silhouette\n- lean narrow build\n- long thin limbs\n\nClothing:\n- simple charcoal-black tunic\n- dark fitted trousers\n- dark practical boots\n- subtle belt\n- no armor\n- no cape\n- no weapon\n\nPART 4 PERFORMANCE ARC:\n\n0–3 sec:\nurgent, breathless, carrying huge responsibility\n\n3–5 sec:\nfrustrated by unreachable shrine slot\n\n5–7 sec:\nsurprised by Nyx’s help → chooses trust\n\n7–10 sec:\nfocused → overwhelmed by sacred magic\n\n10–12 sec:\nwonder and relief\n\n12–14 sec:\nhumbled, apologetic, playful reconciliation\n\n14–15 sec:\npeaceful belonging\n\nCRITICAL:\n\nKael’s trust moment must be visible.\n\nWhen Nyx lowers his head:\n\nKael looks at Nyx’s face.\n\nTiny hesitation.\n\nThen he willingly climbs onto him.\n\nNo fear.\n\nNo argument.\n\nTheir action completes the relationship arc.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — ZARA]\n\nMatch Parts 2–3 exactly.\n\nZARA — THE BONEKEEPER\n\nAppearance:\n- deep warm-brown skin\n- enormous dense black curly hair\n- cool blue highlights along curls\n- very large expressive brown eyes\n- strong dark eyebrows\n- blue distressed top\n- stylized pale ribcage design printed on fabric\n- brown utility belt\n- practical fantasy trousers / waist layers\n- wrapped forearms\n- practical boots\n- compact athletic build\n\nPART 4 PERFORMANCE:\n\nurgent leader\n→ shrine awe\n→ emotional concern for guardians\n→ realization that their ancient duty is fulfilled\n→ quiet happiness\n→ warm laughter with Kael and Nyx\n\nDo not make Zara dramatically cry.\n\nA tiny glossy eye / soft teary smile is enough.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch Parts 1–3 exactly.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- enormous elongated black dragon\n- long serpentine neck\n- feather-like layered black-blue scales\n- backward wild head spines\n- huge pointed ears\n- enormous glowing amber-orange eyes\n- vertical slit pupils\n- exaggerated luminous toothy grin\n- long flexible body\n- dark wings\n\nBEGINNING STATE:\n\nNyx is magically exhausted.\n\nShow:\n- slightly lowered head\n- slower breathing\n- dimmer amber eyes\n- tired posture\n\nHe is NOT injured.\n\nAs the moon returns:\n- amber eyes regain brightness\n- scales regain moonlit highlights\n- posture rises\n- energy returns\n\nPART 4 ROLE:\n\nNyx provides the physical reach Kael lacks.\n\nHe does not solve the final action himself.\n\nHe enables Kael to finish it.\n\nThis reinforces teamwork.\n\n--------------------------------------------------\n\n[MOON FRAGMENT LOCK]\n\nOnly ONE moon fragment exists.\n\nSame fragment from Parts 1–3.\n\nAppearance:\n- irregular crescent/crystalline shard\n- bright silver-white core\n- translucent pale cyan edges\n- small floating particles\n- powerful cool illumination\n\nTimeline:\n\nStart:\nKael holds fragment with both hands.\n\nMid:\nKael carries it up shrine.\n\nThen:\nKael inserts it into exact receptacle.\n\nAfter insertion:\nit becomes integrated into shrine energy.\n\nIt must NOT remain simultaneously in Kael’s hands.\n\nNo duplicates.\n\n--------------------------------------------------\n\n[MOON SHRINE LOCK]\n\nMatch the storyboard exactly.\n\nAncient lunar shrine on cliff summit.\n\nDesign:\n- tall weathered dark-stone structure\n- giant crescent / circular lunar arch\n- narrow vertical receptacle near upper center\n- ancient crescent carvings\n- geometric silver rune channels\n- stone staircase\n- central pedestal\n- cliff-top silhouette\n\nBEGINNING:\ncompletely inactive.\n\nACTIVATION:\n\nfragment inserted\n→ central rune ignites\n→ nearby runes illuminate\n→ light traces upward and outward\n→ entire lunar arch glows\n→ brilliant silver-blue pulse travels into valley\n→ moon reacts\n\nNo explosion.\n\nNo structural destruction.\n\n--------------------------------------------------\n\n[MOON RESTORATION LOCK]\n\nAt beginning:\nmoon is almost black.\n\nOnly:\n- faint silver rim\n- subtle crack lines.\n\nRestoration must be gradual and readable:\n\nsilver pulse reaches sky\n→ one lunar edge brightens\n→ cracks illuminate\n→ silver light spreads across surface\n→ missing fragment area reforms\n→ cracks close\n→ full moon becomes brilliant again.\n\nFinal moon must visually match the enormous perfect moon from PART 1.\n\nNo second moon.\n\nNo explosion.\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN TRANSFORMATION LOCK]\n\nThe skeleton guardians’ transformation is peaceful and magical.\n\nThey have completed their ancient purpose.\n\nUse:\n\nblue runes glow\n→ ivory bodies become silver translucent\n→ edges dissolve gently into particles\n→ forms turn into dozens of tiny silver-blue lights\n→ lights become moon-fireflies\n→ thousands of fireflies rise through valley\n\nIMPORTANT:\n\nThis is NOT death.\n\nThis is NOT destruction.\n\nTheir consciousness / personality is implied to continue in the lights.\n\nOne guardian should give Zara a final little wave BEFORE transforming.\n\nZara recognizes the gesture.\n\nMake it tender.\n\n--------------------------------------------------\n\n[LIVING BONE BRIDGE CONTINUITY]\n\nThe Living Bone Bridge from Part 3 remains visible early in Part 4.\n\nAs the guardians transform later:\n\nthe bridge itself may gently dissolve into silver-blue fireflies from far end toward near end.\n\nCharacters are already safely OFF the bridge before this happens.\n\nNo collapsing.\n\nNo falling.\n\nNo violent disassembly.\n\n--------------------------------------------------\n\n[RESTORED NOCTARA LOCK]\n\nAfter the moon returns:\n\nshow the same world established in Part 1:\n\n- giant gothic castles\n- amber-orange windows\n- reflective lake\n- dark blue forest\n- gigantic restored silver moon\n- calm night sky\n\nSilver light travels over architecture.\n\nWarm castle lights become more vivid.\n\nLake reflects:\nsilver moon\n+\namber city\n+\nsilver-blue moon-fireflies.\n\n--------------------------------------------------\n\n[FINAL LOCATION CALLBACK]\n\nThe final scene must visually echo PART 1.\n\nPART 1 OPENING:\n\nKael sits alone on rock\n+\nlake\n+\nhuge moon\n+\nNoctara castles.\n\nPART 4 ENDING:\n\nKael sits on same or visually matching rock,\nbut now:\n\nZara sits beside him.\n\nNyx’s giant head squeezes awkwardly beside/between them.\n\nSilver-blue fireflies drift across the lake.\n\nThe visual meaning must read even without dialogue:\n\nKael is no longer alone.\n\n--------------------------------------------------\n\n[WORLD / STORY LOGIC]\n\nKael catches fragment\n→ runs toward shrine.\n\nShrine receptacle too high\n→ Kael cannot reach.\n\nNyx sees problem\n→ lowers head.\n\nKael trusts Nyx\n→ climbs onto dragon.\n\nNyx raises him\n→ Kael reaches receptacle.\n\nKael inserts fragment\n→ shrine activates.\n\nShrine activates\n→ moonlight crosses valley.\n\nMoon receives energy\n→ moon restores.\n\nMoon restored\n→ guardians’ duty is complete.\n\nGuardians transform\n→ Zara realizes they are free.\n\nKael understands Nyx\n→ friendship resolves.\n\nNo unexplained jumps.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.70s\nTHE LAST RUN\n\nContinue DIRECTLY from Part 3.\n\nKael rises from his safe slide on the Living Bone Bridge.\n\nThe glowing moon fragment is secured in BOTH hands.\n\nIts silver light illuminates his face.\n\nZara points toward the shrine.\n\nZara:\n<d>[English] GO!</d>\n\nKael runs.\n\nHe leaves the Bone Bridge and sprints up the ancient shrine approach.\n\nNyx follows behind more slowly, visibly exhausted but determined.\n\nCamera:\nlow tracking beside Kael\n→ front-facing running shot\n→ tilt up revealing enormous inactive shrine ahead.\n\nSound:\n- fast footsteps\n- Kael breathing\n- fragment hum\n- cold wind\n- distant bridge rune tone\n\nMusic:\nurgent strings + low percussion.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.70s–3.20s\nONE FINAL PROBLEM\n\nKael races up the stone steps.\n\nThe shard’s light briefly reflects off dormant shrine runes.\n\nHe reaches the center.\n\nStops.\n\nLooks upward.\n\nCamera follows his gaze.\n\nThe receptacle is extremely high above him inside the giant lunar arch.\n\nKael’s expression drops.\n\nHe tries one quick safe jump while lifting the fragment.\n\nNot even close.\n\nHe lands.\n\nLooks up again.\n\nDialogue:\n\nKael:\n<d>[English] Seriously?!</d>\n\nBrief comedy beat.\n\nCamera:\nmedium Kael\n→ huge vertical tilt showing impossible height\n→ close frustrated expression.\n\nMusic pauses for one tiny comic note.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.20s–4.70s\nTHE DRAGON OFFERS THE FINAL LIFT\n\nA huge black shape moves behind Kael.\n\nKael turns quickly.\n\nNyx lowers his enormous tired head beside him.\n\nAmber eyes meet Kael’s.\n\nNyx gives his familiar terrifying luminous grin.\n\nBut the emotional meaning is completely different now.\n\nKael understands.\n\nHold eye contact.\n\nKael glances:\nNyx\n→ high receptacle\n→ Nyx again.\n\nTiny hesitation.\n\nThen Kael smiles faintly.\n\nDialogue:\n\nKael:\n<d>[English] Okay.</d>\n\nNyx lowers his head even further.\n\nKael carefully climbs between his large ears/head spines while holding the shard securely.\n\nZara watches from below.\n\nCamera:\nKael close\n→ Nyx face\n→ intimate eye contact\n→ wide scale composition.\n\nMusic:\nurgent score softens into trust motif.\n\n--------------------------------------------------\n\n[SHOT 4] — 4.70s–6.20s\nTHE MISSING PIECE COMES HOME\n\nNyx slowly stretches his long neck upward.\n\nNo sudden movement.\n\nKael rises toward the receptacle.\n\nThe almost-dark moon hangs behind the shrine.\n\nKael reaches forward.\n\nThe silver shard aligns with the exact missing slot.\n\nFor one breath:\n\nKael stares at it.\n\nThen inserts the fragment.\n\nCLICK / soft magical lock.\n\nThe shard fits perfectly.\n\nA single thin silver rune illuminates beneath Kael’s hand.\n\nKael whispers:\n\n<d>[English] Come on...</d>\n\nCamera:\nside silhouette of Nyx raising Kael\n→ close hand/shard\n→ macro alignment\n→ shard insertion\n→ rune ignition.\n\nSound:\n- crystalline alignment\n- soft stone resonance\n- deep lunar tone\n\n--------------------------------------------------\n\n[SHOT 5] — 6.20s–8.10s\nTHE MOON SHRINE REMEMBERS\n\nThe first rune becomes brighter.\n\nThen light races elegantly through nearby carvings.\n\nOne line.\n\nThen another.\n\nThen dozens.\n\nThe entire crescent arch activates in glowing silver-blue geometry.\n\nWHOOOM.\n\nNot an explosion.\n\nKael shields his eyes slightly.\n\nNyx’s black scales become beautifully rim-lit in silver.\n\nHis amber eyes begin regaining brightness.\n\nZara looks upward in awe.\n\nA controlled wave of moonlight launches outward from the shrine along ancient rune paths.\n\nDialogue:\n\nZara, breathlessly:\n<d>[English] It worked...</d>\n\nCamera:\nrune close-up\n→ tracing lines\n→ giant shrine-wide activation\n→ silver wave leaving shrine.\n\nMusic:\nfull magical orchestra begins opening upward.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.10s–10.10s\nAND THE MOON RETURNS\n\nFollow the silver wave.\n\nIt travels:\n\nacross cliff\n→ along Living Bone Bridge\n→ through Valley of Bones\n→ into distant Noctara\n→ upward toward the sky.\n\nCut to the almost-dark moon.\n\nSilver light touches one edge.\n\nThen restoration begins:\n\nthin silver rim brightens\n→ fracture lines glow\n→ pale light spreads inward\n→ missing lunar section reforms\n→ cracks seal\n→ full giant silver moon returns.\n\nGive the moon reveal enough time to register.\n\nNyx, Kael and Zara stare upward.\n\nNyx’s amber eyes become bright again.\n\nDialogue:\n\nKael, quietly:\n<d>[English] We did it.</d>\n\nZara:\n<d>[English] We did.</d>\n\nSound:\n- deep celestial harmonic swell\n- layered crystalline resonance\n- gentle wave\n- distant castle bells beginning\n\nMusic reaches emotional peak.\n\n--------------------------------------------------\n\n[SHOT 7] — 10.10s–11.80s\nNOCTARA SEES ITS LIGHT AGAIN\n\nVery wide cinematic reveal.\n\nThe restored enormous moon illuminates:\n\n- Valley of Bones\n- distant gothic Noctara\n- castle towers\n- calm lake\n- dark forest\n\nAmber castle windows glow against silver moonlight.\n\nThe lake becomes a silver-blue mirror.\n\nKael’s face relaxes.\n\nZara smiles.\n\nNyx lifts his head proudly.\n\nDo not rush.\n\nUse this as the grand visual payoff.\n\nCamera:\nhuge moon\n→ tilt downward through city\n→ reflective lake\n→ heroes silhouetted against restored world.\n\nSound:\n- distant bells\n- soft wind\n- lake\n- gentle magical resonance\n\n--------------------------------------------------\n\n[SHOT 8] — 11.80s–13.20s\nTHE VALLEY OF STARS\n\nBack near Zara.\n\nThe skeleton guardians begin glowing.\n\nZara notices.\n\nOne familiar guardian turns toward her.\n\nIt slowly lifts one skeletal hand.\n\nWaves.\n\nZara’s expression softens immediately.\n\nShe raises her hand slightly in return.\n\nThe guardian’s ivory form becomes translucent.\n\nBlue runes become silver-white.\n\nThen—\n\nthe guardian gently dissolves into dozens of tiny silver-blue moon-fireflies.\n\nOther guardians follow.\n\nAcross the bridge and valley:\n\nhundreds of guardians become thousands of beautiful lights.\n\nNo destruction.\n\nNo pain.\n\nThe Living Bone Bridge itself transforms gracefully into rising luminous fireflies.\n\nThe valley fills with stars.\n\nA few tiny fireflies circle Zara’s curls and settle briefly near her open palm.\n\nZara smiles with glossy emotional eyes.\n\nDialogue:\n\nZara, soft:\n<d>[English] You’re free.</d>\n\nSound:\n- delicate bone/rune chime\n- thousands of soft magical wing sparkles\n- airy silver shimmer\n\nMusic:\nshift from triumph into warm emotional release.\n\n--------------------------------------------------\n\n[SHOT 9] — 13.20s–15.00s\nFIRST IMPRESSIONS — FINAL FRIENDSHIP\n\nUse a VERY efficient flowing transition.\n\nKael stands beside recovered Nyx.\n\nZara joins them.\n\nSilver-blue fireflies surround the trio.\n\nKael looks up at Nyx.\n\nDialogue:\n\nKael:\n<d>[English] You weren’t stealing it.</d>\n\nNyx gives his enormous ridiculous grin.\n\nNyx:\n<d>[English] Obviously.</d>\n\nKael:\n<d>[English] You could’ve told me.</d>\n\nNyx:\n<d>[English] You screamed and ran away.</d>\n\nZara laughs.\n\nMATCH CUT / elegant visual transition into final lakeside callback.\n\nSame visual language as Part 1 opening:\n\n- gigantic restored moon\n- glowing Noctara castles\n- still lake\n- silver reflection\n- same lakeside rock\n\nBut now:\n\nKael sits in the middle.\n\nZara sits beside him.\n\nNyx’s enormous head is squeezed awkwardly between/next to them, trying to fit.\n\nMoon-fireflies drift everywhere.\n\nKael looks toward the moon.\n\nKael:\n<d>[English] Same time tomorrow?</d>\n\nNyx:\n<d>[English] Only if the moon behaves.</d>\n\nZara, dry:\n<d>[English] It won’t.</d>\n\nNyx’s amber eyes widen.\n\nKael gives a tired little sigh.\n\nZara smiles.\n\nIMPORTANT TIMING:\n\nBecause 15 seconds is short, if all final dialogue cannot fit naturally, PRIORITIZE:\n\n1. <d>[English] You weren’t stealing it.</d>\n2. <d>[English] Obviously.</d>\n3. <d>[English] Same time tomorrow?</d>\n4. <d>[English] Only if the moon behaves.</d>\n5. Zara’s final:\n   <d>[English] It won’t.</d>\n\nDo NOT rush voices unnaturally.\n\nThe “You could’ve told me / You screamed and ran away” exchange may be conveyed partly through facial performance if needed.\n\nCamera slowly pulls backward across the moonlit lake.\n\nThe three friends become small against the huge restored world.\n\nSilver-blue moon-fireflies float through foreground.\n\nHOLD final composition.\n\nFADE TO BLACK.\n\nEND.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage-boy voice.\n\nQualities:\n- youthful\n- expressive\n- sincere\n- thoughtful\n- not overly heroic\n- clear English\n\nPerformance progression:\n\n0–3 sec:\nbreathless urgency\n\n3–5 sec:\ncomic frustration → trust\n\n5–8 sec:\nfocused\n\n8–11 sec:\nquiet wonder and relief\n\n11–13 sec:\nemotionally softened\n\n13–15 sec:\ncomfortable playful friendship\n\nImportant:\n\n<d>[English] We did it.</d>\n\nshould be spoken softly in disbelief rather than shouted.\n\nFinal dialogue should feel relaxed.\n\n--------------------------------------------------\n\n[ZARA VOICE]\n\nNatural teenage-girl voice.\n\nQualities:\n- grounded\n- confident\n- dry humor\n- warm\n- clear English\n\nPart 4:\n\nopening:\nurgent.\n\nrestoration:\nawed.\n\nguardian transformation:\nsoft and emotional.\n\nending:\ndry comedic confidence.\n\nCritical line:\n\n<d>[English] You’re free.</d>\n\nshould be nearly whispered.\n\nFinal:\n\n<d>[English] It won’t.</d>\n\nshould be perfectly deadpan.\n\n--------------------------------------------------\n\n[NYX VOICE]\n\nFor the first time, Nyx may speak clearly in this finale.\n\nVoice:\n- deep but surprisingly articulate\n- warm\n- dry\n- slightly theatrical\n- not monstrous\n- no evil resonance\n\nKeep speech concise.\n\nMost important comedy line:\n\n<d>[English] Obviously.</d>\n\nDeliver as though Nyx cannot believe Kael needed an explanation.\n\nFinal:\n\n<d>[English] Only if the moon behaves.</d>\n\nwarm, playful.\n\n--------------------------------------------------\n\n[MOON SHRINE SOUND DESIGN]\n\nInactive shrine:\nvery faint stone resonance.\n\nShard insertion:\n- crystal alignment\n- soft magnetic/stone lock\n- one pure silver bell tone\n\nRune activation:\n- harmonic layers awakening sequentially\n- gentle low-frequency resonance\n- glassy silver overtones\n\nFull shrine:\nrich sacred “WHOOOM.”\n\nNo explosion.\n\n--------------------------------------------------\n\n[MOON RESTORATION SOUND]\n\nUse a distinctive celestial progression:\n\nfaint lunar hum\n→ silver harmonic grows\n→ crack lines resonate\n→ rich glass/choir tone\n→ warm resolved celestial chord.\n\nNo thunder.\nNo boom.\nNo destruction.\n\n--------------------------------------------------\n\n[MOON-FIREFLY SOUND DESIGN]\n\nGuardian rune glow:\ndelicate cyan chimes.\n\nTransformation:\n- soft sparkling release\n- airy wing shimmer\n- tiny individual bells\n\nThousands rising:\nlayered soft shimmering atmosphere.\n\nNo ghost sounds.\nNo mourning wails.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nBeginning:\n- cold canyon wind\n- fast footsteps\n- distant bridge rune hum\n\nShrine:\n- stone resonance\n- magical rune energy\n\nMoon restoration:\n- valley wind softens\n- distant castle bells\n- lake ambience gradually returns\n- distant night insects return\n\nFinal lakeside:\n- gentle water\n- quiet night wind\n- soft firefly shimmer\n- distant Noctara bells\n\nUse sound to communicate that the world itself has recovered.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL cinematic fantasy finale score.\n\nInstrumentation:\n- urgent strings\n- piano\n- French horn\n- harp\n- celesta\n- glass harmonics\n- warm sustained strings\n- low cello\n- airy choir without lyrics\n- light flute\n- subtle frame drum\n\nMUSIC ARC:\n\n0.00–1.70\nurgent final run.\n\n1.70–3.20\ntension + brief comedy pause.\n\n3.20–4.70\nquiet trust motif.\n\n4.70–6.20\nsacred anticipation.\n\n6.20–8.10\nlarge magical activation build.\n\n8.10–10.10\nmajor emotional climax during moon restoration.\n\n10.10–11.80\ntriumphant but warm release.\n\n11.80–13.20\nsoft emotional guardian farewell.\n\n13.20–15.00\ngentle friendship theme + tiny comedy cadence.\n\nEnd with the SAME melodic idea used in the Part 1 opening, but now harmonically fuller and warmer.\n\nThis creates musical closure.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 4 camera progression:\n\n0–3 sec:\nurgent forward motion + vertical shrine scale.\n\n3–5 sec:\nintimate Kael/Nyx trust coverage.\n\n5–8 sec:\nclose sacred action → giant activation reveal.\n\n8–11 sec:\nlargest environmental / celestial wides.\n\n11–13 sec:\nslow graceful guardian transformation.\n\n13–15 sec:\nintimate reconciliation → huge calm final callback.\n\nUse:\n- forward tracking\n- dramatic upward tilt\n- eye-level emotional close-ups\n- slow crane\n- macro shard insertion\n- large shrine pull-back\n- huge moon composition\n- environmental traveling wave\n- slow orbit around Zara/fireflies\n- final lakeside pull-back\n\nAvoid:\n- handheld shake\n- crash zoom\n- extreme Dutch angles\n- frantic editing\n- 360 spins\n- action-movie camera after moon restoration\n\nThe camera should become CALMER as the story resolves.\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH PARTS 1–3 EXACTLY.\n\nUse:\n- expressive dark ink contours\n- watercolor/gouache painterly backgrounds\n- rich midnight blues\n- silver moon illumination\n- expressive stylized faces\n- detailed black-blue dragon scales\n- warm amber eyes\n- ancient stone textures\n- luminous cyan/silver runes\n- elegant magic\n- atmospheric depth\n- subtle cinematic bloom\n- restrained squash/stretch\n\nThe final animation should feel like the uploaded storyboard art has come alive.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealism\n- live action\n- 3D CGI\n- anime\n- flat vector\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\n0–3 sec:\n\nAlmost-dark world.\n\nPalette:\n- charcoal\n- deep navy\n- black-blue\n- faint cyan\n\nPRIMARY LIGHT SOURCE:\nmoon fragment in Kael’s hands.\n\n3–5 sec:\n\nIntroduce:\n- Nyx’s amber eyes\n- silver shard light\n- faint shrine reflected light.\n\n5–8 sec:\n\nSilver/cyan rune illumination expands rapidly.\n\n8–10 sec:\n\nMoon restoration becomes brightest sequence.\n\nUse:\n- brilliant silver-white\n- cool lunar blue\n- subtle cyan\n- deep navy contrast.\n\n10–13 sec:\n\nRestored night palette:\n- silver moon\n- amber castle lights\n- blue-black environment\n- silver-blue moon-fireflies.\n\n13–15 sec:\n\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086323087092470161",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T05:26:00.628Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086323087092470161.mp4",
    "posterUrl": "/posters/x/2086323087092470161.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "fantasy",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Fourthi Storyboard Image to video using MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 4 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, scale, costume, prop, lighting, cinematography, emotional-performance, and narrative reference.\n\nThis sequence continues IMMEDIATELY from PART 3.\n\nPART 3 ended with:\n- the moon almost completely dark\n- only a faint silver rim remaining\n- the Living Bone Bridge still spanning the canyon\n- Nyx magically exhausted but physically unharmed\n- Zara pointing toward the Moon Shrine\n- Kael holding the ONLY glowing moon fragment securely in both hands\n- Zara shouting:\n  <d>[English] KAEL! GO!</d>\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- reference drawings\n- activation diagrams\n- paper background\n- storyboard header/footer\n- color swatches\n- annotation text\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact face, hair, costume and proportions\n2. Zara’s exact face, enormous curly hairstyle and blue rib-pattern shirt\n3. Nyx’s exact black dragon anatomy\n4. Nyx’s enormous amber eyes and luminous grin\n5. the ONLY silver moon fragment\n6. the almost-dark damaged moon\n7. the Living Bone Bridge\n8. the Moon Shrine architecture\n9. the impossibly high fragment receptacle\n10. Nyx giving Kael the final lift\n11. Kael inserting the fragment\n12. silver-blue shrine activation\n13. moonlight wave crossing the valley\n14. gradual moon restoration\n15. restored Noctara\n16. skeleton guardian → moon-firefly transformation\n17. Zara’s emotional reaction\n18. Kael and Nyx reconciliation\n19. final lakeside callback\n20. final Kael + Zara + Nyx friendship composition\n\nTranslate the storyboard’s 15 one-second panels into approximately **9 coherent cinematic shots**.\n\nDO NOT create fifteen frantic one-second cuts.\n\nUse:\n\nurgent action\n→ obstacle\n→ trust\n→ sacred activation\n→ enormous magical payoff\n→ emotional release\n→ reconciliation\n→ quiet final comedy.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:45–1:00 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nfantasy adventure finale / emotional magical payoff\n\nDialogue:\nnative synchronized English\n\nAudio:\nnative dialogue + environmental sound + moon magic + guardian transformation + original cinematic score\n\nEMOTIONAL ARC:\n\ndesperate urgency\n→ frustration\n→ unexpected help\n→ complete trust\n→ sacred wonder\n→ triumphant restoration\n→ emotional release\n→ realization\n→ friendship\n→ gentle comedy\n\nCORE MESSAGE:\n\nKael began the story believing Nyx was stealing the light.\n\nHe ends the story understanding that Nyx was protecting it.\n\nAnd Kael began the story watching the moon alone.\n\nHe ends it watching with friends.\n\n--------------------------------------------------\n\n[PART 4 STORY LIMIT]\n\nThis is the FINAL CHAPTER.\n\nSTART:\n\nKael has just caught the moon fragment.\n\nZara tells him to go.\n\nEND:\n\nThe moon is fully restored.\n\nThe guardians have peacefully become silver-blue moon-fireflies.\n\nKael, Zara and Nyx sit together beside the moonlit lake.\n\nThe story must feel COMPLETELY resolved.\n\nNo cliffhanger.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch Parts 1–3 and the uploaded PART 4 storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- pale-to-medium cool-toned skin\n- youthful angular face\n- large expressive dark eyes\n- thick dark eyebrows\n- swept-back tousled black hair\n- recognizable pointed windswept silhouette\n- lean narrow build\n- long thin limbs\n\nClothing:\n- simple charcoal-black tunic\n- dark fitted trousers\n- dark practical boots\n- subtle belt\n- no armor\n- no cape\n- no weapon\n\nPART 4 PERFORMANCE ARC:\n\n0–3 sec:\nurgent, breathless, carrying huge responsibility\n\n3–5 sec:\nfrustrated by unreachable shrine slot\n\n5–7 sec:\nsurprised by Nyx’s help → chooses trust\n\n7–10 sec:\nfocused → overwhelmed by sacred magic\n\n10–12 sec:\nwonder and relief\n\n12–14 sec:\nhumbled, apologetic, playful reconciliation\n\n14–15 sec:\npeaceful belonging\n\nCRITICAL:\n\nKael’s trust moment must be visible.\n\nWhen Nyx lowers his head:\n\nKael looks at Nyx’s face.\n\nTiny hesitation.\n\nThen he willingly climbs onto him.\n\nNo fear.\n\nNo argument.\n\nTheir action completes the relationship arc.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — ZARA]\n\nMatch Parts 2–3 exactly.\n\nZARA — THE BONEKEEPER\n\nAppearance:\n- deep warm-brown skin\n- enormous dense black curly hair\n- cool blue highlights along curls\n- very large expressive brown eyes\n- strong dark eyebrows\n- blue distressed top\n- stylized pale ribcage design printed on fabric\n- brown utility belt\n- practical fantasy trousers / waist layers\n- wrapped forearms\n- practical boots\n- compact athletic build\n\nPART 4 PERFORMANCE:\n\nurgent leader\n→ shrine awe\n→ emotional concern for guardians\n→ realization that their ancient duty is fulfilled\n→ quiet happiness\n→ warm laughter with Kael and Nyx\n\nDo not make Zara dramatically cry.\n\nA tiny glossy eye / soft teary smile is enough.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch Parts 1–3 exactly.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- enormous elongated black dragon\n- long serpentine neck\n- feather-like layered black-blue scales\n- backward wild head spines\n- huge pointed ears\n- enormous glowing amber-orange eyes\n- vertical slit pupils\n- exaggerated luminous toothy grin\n- long flexible body\n- dark wings\n\nBEGINNING STATE:\n\nNyx is magically exhausted.\n\nShow:\n- slightly lowered head\n- slower breathing\n- dimmer amber eyes\n- tired posture\n\nHe is NOT injured.\n\nAs the moon returns:\n- amber eyes regain brightness\n- scales regain moonlit highlights\n- posture rises\n- energy returns\n\nPART 4 ROLE:\n\nNyx provides the physical reach Kael lacks.\n\nHe does not solve the final action himself.\n\nHe enables Kael to finish it.\n\nThis reinforces teamwork.\n\n--------------------------------------------------\n\n[MOON FRAGMENT LOCK]\n\nOnly ONE moon fragment exists.\n\nSame fragment from Parts 1–3.\n\nAppearance:\n- irregular crescent/crystalline shard\n- bright silver-white core\n- translucent pale cyan edges\n- small floating particles\n- powerful cool illumination\n\nTimeline:\n\nStart:\nKael holds fragment with both hands.\n\nMid:\nKael carries it up shrine.\n\nThen:\nKael inserts it into exact receptacle.\n\nAfter insertion:\nit becomes integrated into shrine energy.\n\nIt must NOT remain simultaneously in Kael’s hands.\n\nNo duplicates.\n\n--------------------------------------------------\n\n[MOON SHRINE LOCK]\n\nMatch the storyboard exactly.\n\nAncient lunar shrine on cliff summit.\n\nDesign:\n- tall weathered dark-stone structure\n- giant crescent / circular lunar arch\n- narrow vertical receptacle near upper center\n- ancient crescent carvings\n- geometric silver rune channels\n- stone staircase\n- central pedestal\n- cliff-top silhouette\n\nBEGINNING:\ncompletely inactive.\n\nACTIVATION:\n\nfragment inserted\n→ central rune ignites\n→ nearby runes illuminate\n→ light traces upward and outward\n→ entire lunar arch glows\n→ brilliant silver-blue pulse travels into valley\n→ moon reacts\n\nNo explosion.\n\nNo structural destruction.\n\n--------------------------------------------------\n\n[MOON RESTORATION LOCK]\n\nAt beginning:\nmoon is almost black.\n\nOnly:\n- faint silver rim\n- subtle crack lines.\n\nRestoration must be gradual and readable:\n\nsilver pulse reaches sky\n→ one lunar edge brightens\n→ cracks illuminate\n→ silver light spreads across surface\n→ missing fragment area reforms\n→ cracks close\n→ full moon becomes brilliant again.\n\nFinal moon must visually match the enormous perfect moon from PART 1.\n\nNo second moon.\n\nNo explosion.\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN TRANSFORMATION LOCK]\n\nThe skeleton guardians’ transformation is peaceful and magical.\n\nThey have completed their ancient purpose.\n\nUse:\n\nblue runes glow\n→ ivory bodies become silver translucent\n→ edges dissolve gently into particles\n→ forms turn into dozens of tiny silver-blue lights\n→ lights become moon-fireflies\n→ thousands of fireflies rise through valley\n\nIMPORTANT:\n\nThis is NOT death.\n\nThis is NOT destruction.\n\nTheir consciousness / personality is implied to continue in the lights.\n\nOne guardian should give Zara a final little wave BEFORE transforming.\n\nZara recognizes the gesture.\n\nMake it tender.\n\n--------------------------------------------------\n\n[LIVING BONE BRIDGE CONTINUITY]\n\nThe Living Bone Bridge from Part 3 remains visible early in Part 4.\n\nAs the guardians transform later:\n\nthe bridge itself may gently dissolve into silver-blue fireflies from far end toward near end.\n\nCharacters are already safely OFF the bridge before this happens.\n\nNo collapsing.\n\nNo falling.\n\nNo violent disassembly.\n\n--------------------------------------------------\n\n[RESTORED NOCTARA LOCK]\n\nAfter the moon returns:\n\nshow the same world established in Part 1:\n\n- giant gothic castles\n- amber-orange windows\n- reflective lake\n- dark blue forest\n- gigantic restored silver moon\n- calm night sky\n\nSilver light travels over architecture.\n\nWarm castle lights become more vivid.\n\nLake reflects:\nsilver moon\n+\namber city\n+\nsilver-blue moon-fireflies.\n\n--------------------------------------------------\n\n[FINAL LOCATION CALLBACK]\n\nThe final scene must visually echo PART 1.\n\nPART 1 OPENING:\n\nKael sits alone on rock\n+\nlake\n+\nhuge moon\n+\nNoctara castles.\n\nPART 4 ENDING:\n\nKael sits on same or visually matching rock,\nbut now:\n\nZara sits beside him.\n\nNyx’s giant head squeezes awkwardly beside/between them.\n\nSilver-blue fireflies drift across the lake.\n\nThe visual meaning must read even without dialogue:\n\nKael is no longer alone.\n\n--------------------------------------------------\n\n[WORLD / STORY LOGIC]\n\nKael catches fragment\n→ runs toward shrine.\n\nShrine receptacle too high\n→ Kael cannot reach.\n\nNyx sees problem\n→ lowers head.\n\nKael trusts Nyx\n→ climbs onto dragon.\n\nNyx raises him\n→ Kael reaches receptacle.\n\nKael inserts fragment\n→ shrine activates.\n\nShrine activates\n→ moonlight crosses valley.\n\nMoon receives energy\n→ moon restores.\n\nMoon restored\n→ guardians’ duty is complete.\n\nGuardians transform\n→ Zara realizes they are free.\n\nKael understands Nyx\n→ friendship resolves.\n\nNo unexplained jumps.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.70s\nTHE LAST RUN\n\nContinue DIRECTLY from Part 3.\n\nKael rises from his safe slide on the Living Bone Bridge.\n\nThe glowing moon fragment is secured in BOTH hands.\n\nIts silver light illuminates his face.\n\nZara points toward the shrine.\n\nZara:\n<d>[English] GO!</d>\n\nKael runs.\n\nHe leaves the Bone Bridge and sprints up the ancient shrine approach.\n\nNyx follows behind more slowly, visibly exhausted but determined.\n\nCamera:\nlow tracking beside Kael\n→ front-facing running shot\n→ tilt up revealing enormous inactive shrine ahead.\n\nSound:\n- fast footsteps\n- Kael breathing\n- fragment hum\n- cold wind\n- distant bridge rune tone\n\nMusic:\nurgent strings + low percussion.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.70s–3.20s\nONE FINAL PROBLEM\n\nKael races up the stone steps.\n\nThe shard’s light briefly reflects off dormant shrine runes.\n\nHe reaches the center.\n\nStops.\n\nLooks upward.\n\nCamera follows his gaze.\n\nThe receptacle is extremely high above him inside the giant lunar arch.\n\nKael’s expression drops.\n\nHe tries one quick safe jump while lifting the fragment.\n\nNot even close.\n\nHe lands.\n\nLooks up again.\n\nDialogue:\n\nKael:\n<d>[English] Seriously?!</d>\n\nBrief comedy beat.\n\nCamera:\nmedium Kael\n→ huge vertical tilt showing impossible height\n→ close frustrated expression.\n\nMusic pauses for one tiny comic note.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.20s–4.70s\nTHE DRAGON OFFERS THE FINAL LIFT\n\nA huge black shape moves behind Kael.\n\nKael turns quickly.\n\nNyx lowers his enormous tired head beside him.\n\nAmber eyes meet Kael’s.\n\nNyx gives his familiar terrifying luminous grin.\n\nBut the emotional meaning is completely different now.\n\nKael understands.\n\nHold eye contact.\n\nKael glances:\nNyx\n→ high receptacle\n→ Nyx again.\n\nTiny hesitation.\n\nThen Kael smiles faintly.\n\nDialogue:\n\nKael:\n<d>[English] Okay.</d>\n\nNyx lowers his head even further.\n\nKael carefully climbs between his large ears/head spines while holding the shard securely.\n\nZara watches from below.\n\nCamera:\nKael close\n→ Nyx face\n→ intimate eye contact\n→ wide scale composition.\n\nMusic:\nurgent score softens into trust motif.\n\n--------------------------------------------------\n\n[SHOT 4] — 4.70s–6.20s\nTHE MISSING PIECE COMES HOME\n\nNyx slowly stretches his long neck upward.\n\nNo sudden movement.\n\nKael rises toward the receptacle.\n\nThe almost-dark moon hangs behind the shrine.\n\nKael reaches forward.\n\nThe silver shard aligns with the exact missing slot.\n\nFor one breath:\n\nKael stares at it.\n\nThen inserts the fragment.\n\nCLICK / soft magical lock.\n\nThe shard fits perfectly.\n\nA single thin silver rune illuminates beneath Kael’s hand.\n\nKael whispers:\n\n<d>[English] Come on...</d>\n\nCamera:\nside silhouette of Nyx raising Kael\n→ close hand/shard\n→ macro alignment\n→ shard insertion\n→ rune ignition.\n\nSound:\n- crystalline alignment\n- soft stone resonance\n- deep lunar tone\n\n--------------------------------------------------\n\n[SHOT 5] — 6.20s–8.10s\nTHE MOON SHRINE REMEMBERS\n\nThe first rune becomes brighter.\n\nThen light races elegantly through nearby carvings.\n\nOne line.\n\nThen another.\n\nThen dozens.\n\nThe entire crescent arch activates in glowing silver-blue geometry.\n\nWHOOOM.\n\nNot an explosion.\n\nKael shields his eyes slightly.\n\nNyx’s black scales become beautifully rim-lit in silver.\n\nHis amber eyes begin regaining brightness.\n\nZara looks upward in awe.\n\nA controlled wave of moonlight launches outward from the shrine along ancient rune paths.\n\nDialogue:\n\nZara, breathlessly:\n<d>[English] It worked...</d>\n\nCamera:\nrune close-up\n→ tracing lines\n→ giant shrine-wide activation\n→ silver wave leaving shrine.\n\nMusic:\nfull magical orchestra begins opening upward.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.10s–10.10s\nAND THE MOON RETURNS\n\nFollow the silver wave.\n\nIt travels:\n\nacross cliff\n→ along Living Bone Bridge\n→ through Valley of Bones\n→ into distant Noctara\n→ upward toward the sky.\n\nCut to the almost-dark moon.\n\nSilver light touches one edge.\n\nThen restoration begins:\n\nthin silver rim brightens\n→ fracture lines glow\n→ pale light spreads inward\n→ missing lunar section reforms\n→ cracks seal\n→ full giant silver moon returns.\n\nGive the moon reveal enough time to register.\n\nNyx, Kael and Zara stare upward.\n\nNyx’s amber eyes become bright again.\n\nDialogue:\n\nKael, quietly:\n<d>[English] We did it.</d>\n\nZara:\n<d>[English] We did.</d>\n\nSound:\n- deep celestial harmonic swell\n- layered crystalline resonance\n- gentle wave\n- distant castle bells beginning\n\nMusic reaches emotional peak.\n\n--------------------------------------------------\n\n[SHOT 7] — 10.10s–11.80s\nNOCTARA SEES ITS LIGHT AGAIN\n\nVery wide cinematic reveal.\n\nThe restored enormous moon illuminates:\n\n- Valley of Bones\n- distant gothic Noctara\n- castle towers\n- calm lake\n- dark forest\n\nAmber castle windows glow against silver moonlight.\n\nThe lake becomes a silver-blue mirror.\n\nKael’s face relaxes.\n\nZara smiles.\n\nNyx lifts his head proudly.\n\nDo not rush.\n\nUse this as the grand visual payoff.\n\nCamera:\nhuge moon\n→ tilt downward through city\n→ reflective lake\n→ heroes silhouetted against restored world.\n\nSound:\n- distant bells\n- soft wind\n- lake\n- gentle magical resonance\n\n--------------------------------------------------\n\n[SHOT 8] — 11.80s–13.20s\nTHE VALLEY OF STARS\n\nBack near Zara.\n\nThe skeleton guardians begin glowing.\n\nZara notices.\n\nOne familiar guardian turns toward her.\n\nIt slowly lifts one skeletal hand.\n\nWaves.\n\nZara’s expression softens immediately.\n\nShe raises her hand slightly in return.\n\nThe guardian’s ivory form becomes translucent.\n\nBlue runes become silver-white.\n\nThen—\n\nthe guardian gently dissolves into dozens of tiny silver-blue moon-fireflies.\n\nOther guardians follow.\n\nAcross the bridge and valley:\n\nhundreds of guardians become thousands of beautiful lights.\n\nNo destruction.\n\nNo pain.\n\nThe Living Bone Bridge itself transforms gracefully into rising luminous fireflies.\n\nThe valley fills with stars.\n\nA few tiny fireflies circle Zara’s curls and settle briefly near her open palm.\n\nZara smiles with glossy emotional eyes.\n\nDialogue:\n\nZara, soft:\n<d>[English] You’re free.</d>\n\nSound:\n- delicate bone/rune chime\n- thousands of soft magical wing sparkles\n- airy silver shimmer\n\nMusic:\nshift from triumph into warm emotional release.\n\n--------------------------------------------------\n\n[SHOT 9] — 13.20s–15.00s\nFIRST IMPRESSIONS — FINAL FRIENDSHIP\n\nUse a VERY efficient flowing transition.\n\nKael stands beside recovered Nyx.\n\nZara joins them.\n\nSilver-blue fireflies surround the trio.\n\nKael looks up at Nyx.\n\nDialogue:\n\nKael:\n<d>[English] You weren’t stealing it.</d>\n\nNyx gives his enormous ridiculous grin.\n\nNyx:\n<d>[English] Obviously.</d>\n\nKael:\n<d>[English] You could’ve told me.</d>\n\nNyx:\n<d>[English] You screamed and ran away.</d>\n\nZara laughs.\n\nMATCH CUT / elegant visual transition into final lakeside callback.\n\nSame visual language as Part 1 opening:\n\n- gigantic restored moon\n- glowing Noctara castles\n- still lake\n- silver reflection\n- same lakeside rock\n\nBut now:\n\nKael sits in the middle.\n\nZara sits beside him.\n\nNyx’s enormous head is squeezed awkwardly between/next to them, trying to fit.\n\nMoon-fireflies drift everywhere.\n\nKael looks toward the moon.\n\nKael:\n<d>[English] Same time tomorrow?</d>\n\nNyx:\n<d>[English] Only if the moon behaves.</d>\n\nZara, dry:\n<d>[English] It won’t.</d>\n\nNyx’s amber eyes widen.\n\nKael gives a tired little sigh.\n\nZara smiles.\n\nIMPORTANT TIMING:\n\nBecause 15 seconds is short, if all final dialogue cannot fit naturally, PRIORITIZE:\n\n1. <d>[English] You weren’t stealing it.</d>\n2. <d>[English] Obviously.</d>\n3. <d>[English] Same time tomorrow?</d>\n4. <d>[English] Only if the moon behaves.</d>\n5. Zara’s final:\n   <d>[English] It won’t.</d>\n\nDo NOT rush voices unnaturally.\n\nThe “You could’ve told me / You screamed and ran away” exchange may be conveyed partly through facial performance if needed.\n\nCamera slowly pulls backward across the moonlit lake.\n\nThe three friends become small against the huge restored world.\n\nSilver-blue moon-fireflies float through foreground.\n\nHOLD final composition.\n\nFADE TO BLACK.\n\nEND.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage-boy voice.\n\nQualities:\n- youthful\n- expressive\n- sincere\n- thoughtful\n- not overly heroic\n- clear English\n\nPerformance progression:\n\n0–3 sec:\nbreathless urgency\n\n3–5 sec:\ncomic frustration → trust\n\n5–8 sec:\nfocused\n\n8–11 sec:\nquiet wonder and relief\n\n11–13 sec:\nemotionally softened\n\n13–15 sec:\ncomfortable playful friendship\n\nImportant:\n\n<d>[English] We did it.</d>\n\nshould be spoken softly in disbelief rather than shouted.\n\nFinal dialogue should feel relaxed.\n\n--------------------------------------------------\n\n[ZARA VOICE]\n\nNatural teenage-girl voice.\n\nQualities:\n- grounded\n- confident\n- dry humor\n- warm\n- clear English\n\nPart 4:\n\nopening:\nurgent.\n\nrestoration:\nawed.\n\nguardian transformation:\nsoft and emotional.\n\nending:\ndry comedic confidence.\n\nCritical line:\n\n<d>[English] You’re free.</d>\n\nshould be nearly whispered.\n\nFinal:\n\n<d>[English] It won’t.</d>\n\nshould be perfectly deadpan.\n\n--------------------------------------------------\n\n[NYX VOICE]\n\nFor the first time, Nyx may speak clearly in this finale.\n\nVoice:\n- deep but surprisingly articulate\n- warm\n- dry\n- slightly theatrical\n- not monstrous\n- no evil resonance\n\nKeep speech concise.\n\nMost important comedy line:\n\n<d>[English] Obviously.</d>\n\nDeliver as though Nyx cannot believe Kael needed an explanation.\n\nFinal:\n\n<d>[English] Only if the moon behaves.</d>\n\nwarm, playful.\n\n--------------------------------------------------\n\n[MOON SHRINE SOUND DESIGN]\n\nInactive shrine:\nvery faint stone resonance.\n\nShard insertion:\n- crystal alignment\n- soft magnetic/stone lock\n- one pure silver bell tone\n\nRune activation:\n- harmonic layers awakening sequentially\n- gentle low-frequency resonance\n- glassy silver overtones\n\nFull shrine:\nrich sacred “WHOOOM.”\n\nNo explosion.\n\n--------------------------------------------------\n\n[MOON RESTORATION SOUND]\n\nUse a distinctive celestial progression:\n\nfaint lunar hum\n→ silver harmonic grows\n→ crack lines resonate\n→ rich glass/choir tone\n→ warm resolved celestial chord.\n\nNo thunder.\nNo boom.\nNo destruction.\n\n--------------------------------------------------\n\n[MOON-FIREFLY SOUND DESIGN]\n\nGuardian rune glow:\ndelicate cyan chimes.\n\nTransformation:\n- soft sparkling release\n- airy wing shimmer\n- tiny individual bells\n\nThousands rising:\nlayered soft shimmering atmosphere.\n\nNo ghost sounds.\nNo mourning wails.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nBeginning:\n- cold canyon wind\n- fast footsteps\n- distant bridge rune hum\n\nShrine:\n- stone resonance\n- magical rune energy\n\nMoon restoration:\n- valley wind softens\n- distant castle bells\n- lake ambience gradually returns\n- distant night insects return\n\nFinal lakeside:\n- gentle water\n- quiet night wind\n- soft firefly shimmer\n- distant Noctara bells\n\nUse sound to communicate that the world itself has recovered.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL cinematic fantasy finale score.\n\nInstrumentation:\n- urgent strings\n- piano\n- French horn\n- harp\n- celesta\n- glass harmonics\n- warm sustained strings\n- low cello\n- airy choir without lyrics\n- light flute\n- subtle frame drum\n\nMUSIC ARC:\n\n0.00–1.70\nurgent final run.\n\n1.70–3.20\ntension + brief comedy pause.\n\n3.20–4.70\nquiet trust motif.\n\n4.70–6.20\nsacred anticipation.\n\n6.20–8.10\nlarge magical activation build.\n\n8.10–10.10\nmajor emotional climax during moon restoration.\n\n10.10–11.80\ntriumphant but warm release.\n\n11.80–13.20\nsoft emotional guardian farewell.\n\n13.20–15.00\ngentle friendship theme + tiny comedy cadence.\n\nEnd with the SAME melodic idea used in the Part 1 opening, but now harmonically fuller and warmer.\n\nThis creates musical closure.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 4 camera progression:\n\n0–3 sec:\nurgent forward motion + vertical shrine scale.\n\n3–5 sec:\nintimate Kael/Nyx trust coverage.\n\n5–8 sec:\nclose sacred action → giant activation reveal.\n\n8–11 sec:\nlargest environmental / celestial wides.\n\n11–13 sec:\nslow graceful guardian transformation.\n\n13–15 sec:\nintimate reconciliation → huge calm final callback.\n\nUse:\n- forward tracking\n- dramatic upward tilt\n- eye-level emotional close-ups\n- slow crane\n- macro shard insertion\n- large shrine pull-back\n- huge moon composition\n- environmental traveling wave\n- slow orbit around Zara/fireflies\n- final lakeside pull-back\n\nAvoid:\n- handheld shake\n- crash zoom\n- extreme Dutch angles\n- frantic editing\n- 360 spins\n- action-movie camera after moon restoration\n\nThe camera should become CALMER as the story resolves.\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH PARTS 1–3 EXACTLY.\n\nUse:\n- expressive dark ink contours\n- watercolor/gouache painterly backgrounds\n- rich midnight blues\n- silver moon illumination\n- expressive stylized faces\n- detailed black-blue dragon scales\n- warm amber eyes\n- ancient stone textures\n- luminous cyan/silver runes\n- elegant magic\n- atmospheric depth\n- subtle cinematic bloom\n- restrained squash/stretch\n\nThe final animation should feel like the uploaded storyboard art has come alive.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealism\n- live action\n- 3D CGI\n- anime\n- flat vector\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\n0–3 sec:\n\nAlmost-dark world.\n\nPalette:\n- charcoal\n- deep navy\n- black-blue\n- faint cyan\n\nPRIMARY LIGHT SOURCE:\nmoon fragment in Kael’s hands.\n\n3–5 sec:\n\nIntroduce:\n- Nyx’s amber eyes\n- silver shard light\n- faint shrine reflected light.\n\n5–8 sec:\n\nSilver/cyan rune illumination expands rapidly.\n\n8–10 sec:\n\nMoon restoration becomes brightest sequence.\n\nUse:\n- brilliant silver-white\n- cool lunar blue\n- subtle cyan\n- deep navy contrast.\n\n10–13 sec:\n\nRestored night palette:\n- silver moon\n- amber castle lights\n- blue-black environment\n- silver-blue moon-fireflies.\n\n13–15 sec:\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086323087092470161"
  },
  {
    "id": "x-2086322454679707867",
    "title": "归还月亮的少年 · 第三幕",
    "titleEn": "The Boy Who Returned the Moon · Chapter 3",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《归还月亮的少年 · 第三幕》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Boy Who Returned the Moon · Chapter 3. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 3 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, scale, costume, prop, lighting, camera, performance, and narrative reference.\n\nThis sequence continues DIRECTLY from PART 2.\n\nPART 2 ended with:\n- Kael learning that Nyx did NOT steal the moon fragment\n- Zara revealing:\n  <d>[English] He’s trying to return it.</d>\n- Nyx still carefully holding the ONLY glowing moon shard\n- the cracked moon losing more of its light\n- the skeleton guardians gathered around the group\n- the distant mountain and ancient Moon Shrine ahead\n- Zara warning:\n  <d>[English] And we’re running out of time.</d>\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- storyboard panels\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- character reference drawings\n- environment diagrams\n- storyboard paper texture\n- title/header\n- footer graphics\n- color swatches\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact character identity\n2. Zara’s exact face, giant curly hair and blue rib-pattern shirt\n3. Nyx’s exact black midnight-dragon design\n4. Nyx’s amber-orange eyes\n5. Nyx’s luminous grin\n6. the single glowing moon fragment\n7. the ivory magical skeleton guardians\n8. the exact Living Bone Bridge design\n9. blue runic bridge energy\n10. the huge canyon geography\n11. the ancient Moon Shrine across the canyon\n12. the damaged moon state\n13. cold dark-fantasy color palette\n14. Nyx’s failed flight attempt\n15. Zara commanding the guardians\n16. bridge construction\n17. Kael/Zara race across bridge\n18. Nyx becoming magically exhausted\n19. the moon shard falling\n20. Kael catching the shard\n21. final “KAEL! GO!” cliffhanger\n\nTranslate the storyboard’s fifteen one-second panels into approximately **8 cinematic shots**.\n\nDO NOT create fifteen frantic one-second cuts.\n\nUse clear:\n\nmission\n→ failed solution\n→ impossible obstacle\n→ magical solution\n→ race\n→ moon failure\n→ falling shard\n→ heroic catch.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:30–0:45 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nfantasy adventure / magical teamwork\n\nDialogue:\nnative synchronized English dialogue\n\nAudio:\nnative dialogue + environmental sound + magic SFX + skeleton movement + original cinematic score\n\nEMOTIONAL ARC:\n\nurgent mission\n→ momentary hope\n→ failed shortcut\n→ impossible canyon\n→ Zara’s clever solution\n→ spectacular magical bridge\n→ comedic relief\n→ desperate race\n→ moon nearly dies\n→ shard falls\n→ Kael accepts responsibility\n\nCORE PART 3 PAYOFF:\n\nThe dragon carried the moonlight this far.\n\nThe Bonekeeper creates the path.\n\nNow the boy who wrongly accused them must become the one who carries the moonlight forward.\n\n--------------------------------------------------\n\n[PART 3 STORY LIMIT]\n\nPART 3 ONLY.\n\nSTART:\n\nZara points toward the ancient Moon Shrine across the valley.\n\nNyx still holds the moon shard.\n\nEND:\n\nKael has caught the moon shard safely with both hands.\n\nThe Moon Shrine towers immediately ahead.\n\nZara points toward it and shouts:\n\n<d>[English] KAEL! GO!</d>\n\nCUT TO BLACK.\n\nDO NOT show:\n- Kael climbing the shrine\n- Kael saying “Seriously?!”\n- Nyx lifting Kael toward the receptacle\n- shard being inserted\n- shrine activation\n- moon restoration\n- skeleton-to-firefly transformation\n- restored Noctara\n- final friendship dialogue\n\nThose belong to PART 4.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch Parts 1–2 and the uploaded PART 3 storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- pale-to-medium cool-toned skin\n- youthful angular stylized face\n- large expressive dark eyes\n- thick black eyebrows\n- swept-back tousled black hair\n- distinctive windswept pointed silhouette\n- lean narrow build\n- long thin arms and legs\n\nClothing:\n- simple charcoal-black tunic\n- dark trousers\n- practical dark boots\n- subtle belt\n- no armor\n- no cape\n- no weapon\n\nPART 3 PERFORMANCE ARC:\n\n0–3 sec:\nfocused on mission\n\n3–6 sec:\nhope → frustration\n\n6–10 sec:\nastonished by Zara’s solution\n\n10–12 sec:\nurgent sprint\n\n12–14 sec:\nfear when Nyx weakens\n\n14–15 sec:\ninstinctive courage and determination\n\nKael must NOT suddenly behave like a superhero.\n\nHis decisive act is simply:\n\nthe shard falls near the bridge\n→ Kael lunges/slides safely\n→ catches it with both hands.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — ZARA]\n\nMatch PART 2 and the uploaded storyboard EXACTLY.\n\nZARA — THE BONEKEEPER\n\nAppearance:\n- deep warm-brown skin\n- enormous dense curly black hair\n- cool-blue moonlight highlights through curls\n- large dark-brown eyes\n- strong expressive eyebrows\n- blue distressed top\n- pale stylized ribcage motif printed on the fabric\n- brown utility belt\n- practical fantasy trousers / waist wraps\n- forearm wraps\n- sturdy boots\n- compact athletic build\n\nPersonality:\n- fearless\n- intelligent\n- dry humor\n- decisive\n- experienced\n- protective\n\nPART 3 ROLE:\n\nZara is the tactical leader.\n\nShe knows:\n- the destination\n- the urgency\n- the valley\n- the skeleton guardians\n- how to create a safe route\n\nPerformance:\nurgent explanation\n→ skepticism at Nyx’s flight attempt\n→ confident idea\n→ commanding Bonekeeper\n→ sprinting leader\n→ alarm when shard falls\n→ decisive final “GO!”\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch Parts 1–2 EXACTLY.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- huge elongated black dragon\n- long serpentine neck\n- layered black-blue feather-like scales\n- wild backward-pointing spines\n- large pointed ears/horns\n- gigantic glowing amber-orange eyes\n- vertical pupils\n- luminous toothy grin\n- flexible long body\n- large dark wings\n\nAt the start:\nNyx still carries the ONE moon fragment carefully between his jaws.\n\nPART 3 PERFORMANCE:\n\ndetermined\n→ confident flight attempt\n→ surprised by unstable magic\n→ safely pushed back\n→ slightly annoyed\n→ races alongside friends\n→ becomes progressively magically exhausted\n→ loses strength when moon nearly dies\n→ accidentally loses the shard\n\nIMPORTANT:\n\nNyx is NOT physically injured.\n\nHe is experiencing magical exhaustion because his power is tied to the failing moon.\n\nNo fire breathing.\nNo attack.\nNo crash.\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN IDENTITY LOCK]\n\nMatch Part 2 exactly.\n\nAppearance:\n- clean ivory/cream bones\n- rounded stylized proportions\n- expressive skulls\n- large hollow eye sockets\n- pale blue magical eye/runic glow\n- family-friendly\n- individual personality\n\nNo:\n- gore\n- flesh\n- decay\n- horror\n- corpse realism\n\nPART 3 ROLE:\n\nThey voluntarily become magical architecture.\n\nTheir bodies may harmlessly separate and reconnect because they are magical constructs.\n\nThis is NOT physical dismemberment.\n\nVisual logic:\n\nrunes activate\n→ guardian pieces float\n→ components organize\n→ rib arches connect\n→ stable walkway forms\n→ Living Bone Bridge completed.\n\nThe skeleton guardians remain conscious and expressive.\n\nOne skull may wave at Kael.\n\nAnother may look mildly annoyed at becoming a railing.\n\n--------------------------------------------------\n\n[LIVING BONE BRIDGE LOCK]\n\nMatch the storyboard reference EXACTLY.\n\nThe bridge must look like elegant ancient magical architecture.\n\nDesign:\n- ivory structural arches\n- wide stable central walkway\n- curved rib-like side rails\n- guardian skull nodes\n- repeating skeleton-inspired architecture\n- pale cyan/blue magical runes\n- silver-blue energy flowing through joints\n- elegant curve across enormous canyon\n\nThe bridge must feel:\nsafe,\nsolid,\nbeautiful,\nstrange,\nslightly comedic.\n\nNOT:\na pile of bones.\n\nNOT:\na horror bridge.\n\n--------------------------------------------------\n\n[MOON SHRINE LOCK]\n\nUse the same shrine shown in the storyboard reference.\n\nLocation:\nhigh cliff on far side of canyon.\n\nDesign:\n- ancient dark stone\n- towering crescent/lunar arch\n- silver geometric carvings\n- inactive central altar\n- high fragment receptacle\n- ruined stairs\n- faint dormant runes\n\nIn PART 3:\n\nShrine remains completely INACTIVE.\n\nNo activation.\n\nNo silver energy wave.\n\nNo restored moon.\n\n--------------------------------------------------\n\n[MOON FRAGMENT LOCK]\n\nOnly ONE fragment exists.\n\nAppearance:\n- irregular crescent/crystalline silver shard\n- bright white-silver core\n- pale cyan translucent edges\n- soft floating silver particles\n- strong cool light\n\nTimeline:\n\n0–12+ sec:\nheld by Nyx\n\n12–14 sec:\nNyx weakens\n\naround 14 sec:\nshard slips free\n\nending:\nKael catches the SAME fragment with both hands\n\nDo NOT duplicate the shard.\n\nOnce Kael catches it:\nNyx no longer has it.\n\n--------------------------------------------------\n\n[MOON STATE CONTINUITY]\n\nStart:\nheavily cracked and dim.\n\nMiddle:\ncontinues losing brightness.\n\nNear 12–13 sec:\nalmost completely dark.\n\nUse:\n- faint silver rim\n- weak crack lines\n\nAs moon darkens:\nNyx’s strength decreases.\n\nThe moon shard becomes the strongest silver light source in the scene.\n\nDo NOT completely restore it.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK]\n\nContinue the Valley of Bones from PART 2.\n\nUse:\n- towering black-blue cliffs\n- massive canyon\n- pale moonlit mist\n- ruined paths\n- ancient arches\n- silver-gray vegetation\n- fossil-like formations\n- broken original stone bridge\n- distant Moon Shrine\n- deep shadowed valley\n- cold wind\n\nThe environment should feel enormous.\n\nScale is critical.\n\n--------------------------------------------------\n\n[STORY / CAUSE-AND-EFFECT LOGIC]\n\nZara shows destination\n→ group races toward shrine.\n\nNyx sees canyon\n→ tries flying directly.\n\nBroken moon disrupts sky magic\n→ magical current pushes Nyx safely back.\n\nThey discover destroyed bridge\n→ Kael sees no route.\n\nZara sees skeleton guardians\n→ gets solution.\n\nGuardians activate\n→ form Living Bone Bridge.\n\nBridge completed\n→ Kael and Zara sprint across.\n\nMoon dims\n→ Nyx loses magical strength.\n\nNyx weakens\n→ moon shard slips.\n\nShard falls\n→ Kael lunges safely.\n\nKael catches fragment\n→ responsibility transfers to him.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.70s\nTHE SHRINE ACROSS THE VALLEY\n\nContinue DIRECTLY from Part 2.\n\nWide cinematic shot.\n\nThe cracked moon hangs above a jagged mountain.\n\nZara points toward a distant ancient Moon Shrine across the huge valley.\n\nKael follows her gesture.\n\nNyx stands beside them holding the glowing shard.\n\nSkeleton guardians gather behind.\n\nZara speaks quickly:\n\n<d>[English] The shard has to reach that shrine.</d>\n\nKael sees how far away it is.\n\nZara:\n<d>[English] Before the moon goes dark.</d>\n\nThe moon flickers faintly.\n\nCamera:\ngroup medium-wide\n→ follow Zara’s pointing arm\n→ extreme wide reveal of distant shrine.\n\nSound:\n- cold canyon wind\n- faint moon hum\n- shard shimmer\n\nMusic:\nurgent low strings begin.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.70s–3.40s\nNYX TRIES THE OBVIOUS SOLUTION\n\nNyx looks from the shrine to the canyon.\n\nHis giant amber eyes narrow with confidence.\n\nHe gently adjusts the shard in his jaws.\n\nKael looks at him.\n\nKael:\n<d>[English] Can’t he just fly it there?</d>\n\nNyx opens his wings.\n\nZara:\n<d>[English] He can try.</d>\n\nNyx launches upward.\n\nCamera follows him dramatically.\n\nFor half a second:\nit appears to work.\n\nThen pale silver-blue magical currents begin spiraling across the sky.\n\nSound:\n- powerful wing beats\n- rising wind\n- magical air resonance\n\n--------------------------------------------------\n\n[SHOT 3] — 3.40s–4.90s\nTHE BROKEN MOON HAS TWISTED THE SKY\n\nThe silver-blue current wraps around Nyx.\n\nNOT an attack.\n\nIt behaves like an unstable magical jet stream.\n\nNyx struggles to move forward.\n\nHis long body bends with the current.\n\nHis wings compensate.\n\nThe shard remains safely secured.\n\nThen the current gently but firmly pushes him backward.\n\nNyx redirects downward and lands safely beside Kael and Zara.\n\nHis head spines are blown backward.\n\nHe looks up at the current with deeply offended disbelief.\n\nKael:\n<d>[English] ...So that’s a no.</d>\n\nNyx gives Kael an irritated side-eye.\n\nBrief comedy beat.\n\nSound:\n- rushing magical wind\n- wing strain\n- safe landing scrape\n- Nyx annoyed exhale\n\nNo crash.\n\n--------------------------------------------------\n\n[SHOT 4] — 4.90s–6.50s\nTHE BRIDGE IS GONE\n\nThe group reaches the canyon edge.\n\nCamera moves beyond Kael and dramatically pulls wide.\n\nReveal:\n\nan enormous chasm.\n\nThe original stone bridge is destroyed.\n\nOnly broken stone ends remain.\n\nAcross the void:\n\nthe Moon Shrine.\n\nKael steps close but safely behind the edge.\n\nHis face drops.\n\nDialogue:\n\nKael:\n<d>[English] How are we supposed to cross THAT?</d>\n\nNyx looks across.\n\nZara does not answer immediately.\n\nShe slowly turns toward the skeleton guardians behind them.\n\nA tiny confident smile appears.\n\nCamera:\nKael reaction\n→ huge canyon reveal\n→ Zara glance to guardians.\n\nMusic briefly suspends.\n\n--------------------------------------------------\n\n[SHOT 5] — 6.50s–8.80s\nTHE BONEKEEPER’S ANSWER\n\nHero medium-low shot of Zara.\n\nShe raises one hand.\n\nThe skeleton guardians straighten.\n\nTheir eye sockets glow pale cyan.\n\nBlue magical runes illuminate across their ivory forms.\n\nKael slowly turns.\n\nDialogue:\n\nKael:\n<d>[English] Zara...?</d>\n\nZara:\n<d>[English] Guardians.</d>\n\nShe sweeps her hand toward the canyon.\n\nZara:\n<d>[English] Bridge.</d>\n\nThen the magical construction begins.\n\nSkeleton guardian components float harmlessly apart like enchanted puzzle pieces.\n\nNo pain.\n\nNo fear.\n\nShow:\n- rib forms lifting\n- long ivory structural pieces rotating\n- skull guardian nodes floating into position\n- cyan runes linking pieces\n- stable walkway building outward\n\nOne floating skull looks toward Kael while being carried into position.\n\nIt gives him a cheerful wave.\n\nKael stares.\n\nSound:\n- soft bone clacks\n- runic hum\n- magical assembling tones\n- airy movements\n\nMusic:\nrising magical orchestral pattern.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.80s–10.60s\nTHE LIVING BONE BRIDGE\n\nPull VERY wide.\n\nThe final sections connect.\n\nA beautiful Living Bone Bridge now spans the entire canyon.\n\nIvory arches.\nPale blue rune energy.\nStable wide walkway.\nCurved guardian rails.\n\nThe Moon Shrine waits at the far side.\n\nKael is speechless.\n\nHe touches / looks at one guardian rail.\n\nA skeleton skull embedded nearby turns toward him.\n\nKael:\n<d>[English] That is both amazing...</d>\n\nTiny beat.\n\n<d>[English] ...and deeply uncomfortable.</d>\n\nZara is already running onto the bridge.\n\nShe looks back:\n\n<d>[English] RUN!</d>\n\nThe skeleton skull gestures forward as if agreeing with her.\n\nCamera:\ngiant bridge reveal\n→ Kael reaction\n→ Zara sprinting.\n\nMusic shifts directly into adventure momentum.\n\n--------------------------------------------------\n\n[SHOT 7] — 10.60s–12.70s\nRACE TOWARD THE SHRINE\n\nDynamic but smooth tracking shot.\n\nKael and Zara sprint across the Living Bone Bridge.\n\nZara leads.\n\nKael follows.\n\nNyx races/glides beside them through the lower stable air current, still holding the moon fragment.\n\nThe Moon Shrine grows larger.\n\nUse:\n- side tracking\n- bridge parallax\n- running close-ups\n- shrine destination ahead\n\nSkeleton guardian skulls along the rails react as they pass.\n\nOne appears to cheer them on silently.\n\nKael:\n<d>[English] Almost there!</d>\n\nThen the environment changes.\n\nMoonlight weakens.\n\nThe bridge’s cyan glow becomes more noticeable because the sky is darkening.\n\nNyx slows slightly.\n\n--------------------------------------------------\n\n[SHOT 8] — 12.70s–15.00s\nTHE LAST PIECE OF MOONLIGHT\n\nThe music becomes tense.\n\nCut toward the moon.\n\nIt is almost entirely dark.\n\nOnly a faint silver rim and weak crack lines remain.\n\nBack to Nyx.\n\nHis amber eyes dim slightly.\n\nHis wing/body movement becomes heavy.\n\nHe tries to continue.\n\nZara notices.\n\nZara:\n<d>[English] Nyx?!</d>\n\nNyx gives a tired exhale.\n\nThe moon shard begins slipping from his jaws.\n\nClose-up:\n\nsilver shard moves free.\n\nZara:\n<d>[English] KAEL!</d>\n\nThe fragment falls beside the bridge.\n\nUse a brief slow-motion feeling.\n\nIt spins through darkness.\n\nThe brilliant silver light illuminates:\nKael’s face,\nZara,\nNyx,\nthe ivory rails.\n\nKael reacts instinctively.\n\nHe does NOT jump from the bridge.\n\nHe lunges and slides safely across the broad walkway.\n\nOne hand grips / braces near the stable rib rail.\n\nHis other hand reaches.\n\nThen both hands close around the fragment.\n\nCATCH.\n\nSilver light bursts softly across his face.\n\nNo explosion.\n\nKael looks down at what he is holding.\n\nThen toward the towering Moon Shrine now close ahead.\n\nZara is already pointing toward the shrine.\n\nZara shouts:\n\n<d>[English] KAEL! GO!</d>\n\nKael rises with the fragment secured against his chest / in both hands.\n\nHis fear becomes determination.\n\nCamera:\nmoon fading\n→ Nyx weakening\n→ shard slipping\n→ slow follow of falling shard\n→ Kael safe lunge\n→ close-up catch\n→ dramatic low angle on Kael with Moon Shrine behind.\n\nHold final frame briefly:\n\nKael = carrying the last moonlight\nZara = pointing forward\nNyx = exhausted but safe\nBone Bridge = beneath them\nMoon Shrine = waiting\nMoon = nearly dark\n\nCUT TO BLACK.\n\nEND PART 3.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage boy voice.\n\nQualities:\n- youthful\n- clear\n- thoughtful\n- slightly anxious\n- increasingly brave\n- emotionally sincere\n\nPerformance progression:\n\n0–5 sec:\nfocused / questioning\n\n5–7 sec:\nfrustrated by canyon\n\n7–10 sec:\namazed and comedic\n\n10–13 sec:\nurgent\n\n13–15 sec:\nstartled → determined\n\nImportant comedy line:\n\n<d>[English] That is both amazing... and deeply uncomfortable.</d>\n\nDeliver genuinely, not as broad slapstick.\n\n--------------------------------------------------\n\n[ZARA VOICE]\n\nNatural teenage girl voice.\n\nQualities:\n- grounded\n- confident\n- quick\n- authoritative\n- dry humor\n- clear English\n\nKey dialogue:\n\n<d>[English] The shard has to reach that shrine.</d>\n\n<d>[English] Guardians. Bridge.</d>\n\n<d>[English] RUN!</d>\n\nFinal:\n\n<d>[English] KAEL! GO!</d>\n\nHer final line should be urgent and powerful but not screaming hysterically.\n\n--------------------------------------------------\n\n[NYX AUDIO]\n\nNo normal dialogue required.\n\nUse:\n- wing beats\n- deep breathing\n- soft throat resonance\n- annoyed exhale after failed flight\n- heavier breathing as moon fades\n- subtle weakened rumble near ending\n\nNo:\n- roaring\n- growling threats\n- fire\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN AUDIO]\n\nUse charming magical mechanical-bone sounds:\n\n- soft clacks\n- rounded wooden-like clicks\n- magical rune hum\n- light pieces aligning\n- tiny skull rotation sound\n- subtle comedic wave click\n\nBridge completion:\none warm resonant magical “THUMM.”\n\nNot a violent impact.\n\n--------------------------------------------------\n\n[MAGICAL SKY CURRENT AUDIO]\n\nUse:\n- rushing airy energy\n- glassy harmonic sweeps\n- low moon resonance\n\nDo NOT sound like:\n- tornado\n- explosion\n- violent storm\n\n--------------------------------------------------\n\n[MOON FRAGMENT AUDIO]\n\nSignature sound:\nclear silver crystalline hum.\n\nAs Nyx weakens:\nhum becomes more exposed.\n\nWhen shard slips:\nascending/falling glass tone.\n\nWhen Kael catches:\nsoft resonant silver “WHUM.”\n\nNo explosive impact.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nMaintain:\n- canyon wind\n- distant stone echo\n- mist ambience\n- character footsteps\n- bridge footsteps\n- bridge rune hum\n- distant Moon Shrine resonance\n\nDuring final shard fall:\nreduce environment temporarily so the fragment sound dominates.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL cinematic fantasy-adventure score.\n\nInstrumentation:\n- urgent strings\n- cello\n- low brass\n- frame drums\n- hand percussion\n- celesta\n- glockenspiel\n- glass harmonics\n- airy choir without lyrics\n- warm horn layer\n- low piano\n\nMUSIC ARC:\n\n0.00–1.70\nurgent mission setup.\n\n1.70–3.40\nhopeful lift attempt.\n\n3.40–4.90\nmagical turbulence.\n\n4.90–6.50\ndrop into uncertainty at canyon reveal.\n\n6.50–8.80\nmysterious magical build.\n\n8.80–10.60\nlarge uplifting Bone Bridge reveal.\n\n10.60–12.70\ndriving adventure rhythm.\n\n12.70–14.00\nmusic loses momentum as moon fades.\n\n14.00–15.00\ntense silver motif during shard fall →\nstrong hopeful chord on Kael’s catch →\nunresolved cut after “GO!”\n\nNo epic war music.\nNo horror score.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 3 should feel larger and more kinetic than Parts 1–2.\n\nCAMERA ARC:\n\n0–1.7 sec:\ndestination-focused wide composition.\n\n1.7–4.9 sec:\nupward and aerial Nyx coverage.\n\n4.9–6.5 sec:\nmassive canyon reveal.\n\n6.5–10.6 sec:\nmagical construction escalation.\n\n10.6–12.7 sec:\ndynamic tracking action.\n\n12.7–15 sec:\nslow time slightly for emotional/action climax.\n\nUse:\n- wide environmental reveals\n- low hero angle on Zara\n- upward dragon tracking\n- lateral bridge tracking\n- controlled parallax\n- close expression inserts\n- macro moon-fragment shot\n- dramatic final low angle\n\nAvoid:\n- shaky handheld\n- crash zoom\n- constant rapid cutting\n- 360 spins\n- extreme fisheye\n- Dutch-angle overload\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH PARTS 1–2 EXACTLY.\n\nUse:\n- dark expressive ink contours\n- watercolor/gouache-style painterly fills\n- cinematic moonlight\n- richly layered blue-black cliffs\n- beautiful cyan magical effects\n- expressive character faces\n- intricate dragon-scale textures\n- elegant skeleton design\n- atmospheric mist\n- soft silver bloom\n- subtle squash/stretch only where appropriate\n\nThe result should look like the uploaded storyboard artwork brought to cinematic life.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealism\n- live action\n- 3D CGI\n- anime\n- flat vector animation\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\nPART 3 opening:\n- deep navy\n- charcoal\n- cool silver moon\n- amber Nyx eyes\n- bright shard\n\nFlight-current sequence:\n- add pale cyan / silver energy ribbons\n\nCanyon:\n- darker blues\n- gray mist\n\nBone Bridge:\n- ivory architecture\n- beautiful pale cyan runic light\n\nRace:\n- cyan bridge glow\n- fading moonlight\n- silver shard\n\nEnding:\nthe moon is nearly dark.\n\nThe moon fragment becomes the DOMINANT illumination.\n\nKael’s face in the final shot should be lit strongly by:\n- silver-white light from below/front\n- faint cyan bridge bounce\n- little to no moon fill\n\nThe shrine remains dark behind him.\n\n--------------------------------------------------\n\n[MOTION RULES — KAEL]\n\nUse:\n- realistic stylized running\n- natural acceleration\n- controlled balance\n- expressive upper-body acting\n- hair and tunic responding to motion\n\nFor shard catch:\n\nKael MUST remain on the bridge.\n\nAction:\n\nfragment slips\n→ Kael sees it\n→ lunges across safe walkway\n→ slides on one knee / side\n→ braces against stable railing\n→ grabs fragment with both hands\n→ returns fully onto walkway\n\nNo jumping into canyon.\n\n--------------------------------------------------\n\n[MOTION RULES — ZARA]\n\nUse:\n- confident forward motion\n- strong pointing gestures\n- controlled hand command for skeletons\n- hair with substantial secondary motion\n- athletic natural running\n\nDuring bridge command:\nher movements should be minimal but decisive.\n\nShe does not need exaggerated spellcasting.\n\nThe guardians already recognize her.\n\n--------------------------------------------------\n\n[MOTION RULES — NYX]\n\nNyx must feel huge and weighty.\n\nFlight attempt:\n- strong takeoff\n- broad wing strokes\n- long body follow-through\n- magical current bends trajectory\n- Nyx corrects\n- safe landing\n\nRace:\n- lower controlled gliding\n- occasional foot/wing assistance if near bridge\n\nExhaustion:\n- slower wings\n- lowering head\n- reduced eye glow\n- heavier breathing\n\nNo physical collapse.\n\n--------------------------------------------------\n\n[MOTION RULES — SKELETON BRIDGE]\n\nTransformation:\n\nguardian runes illuminate\n→ components float apart cleanly\n→ pieces rotate\n→ long sections align\n→ pale energy connects\n→ walking surface forms\n→ arches rise\n→ final far section locks into place.\n\nMovement should resemble:\nbeautiful magical clockwork / puzzle assembly.\n\nNo violent snapping.\n\nNo painful reactions.\n\n--------------------------------------------------\n\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086322454679707867",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T05:23:29.849Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086322454679707867.mp4",
    "posterUrl": "/posters/x/2086322454679707867.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "nature",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Third Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 3 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, scale, costume, prop, lighting, camera, performance, and narrative reference.\n\nThis sequence continues DIRECTLY from PART 2.\n\nPART 2 ended with:\n- Kael learning that Nyx did NOT steal the moon fragment\n- Zara revealing:\n  <d>[English] He’s trying to return it.</d>\n- Nyx still carefully holding the ONLY glowing moon shard\n- the cracked moon losing more of its light\n- the skeleton guardians gathered around the group\n- the distant mountain and ancient Moon Shrine ahead\n- Zara warning:\n  <d>[English] And we’re running out of time.</d>\n\nGenerate a FINISHED cinematic animated sequence from the uploaded storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- storyboard panels\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- character reference drawings\n- environment diagrams\n- storyboard paper texture\n- title/header\n- footer graphics\n- color swatches\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact character identity\n2. Zara’s exact face, giant curly hair and blue rib-pattern shirt\n3. Nyx’s exact black midnight-dragon design\n4. Nyx’s amber-orange eyes\n5. Nyx’s luminous grin\n6. the single glowing moon fragment\n7. the ivory magical skeleton guardians\n8. the exact Living Bone Bridge design\n9. blue runic bridge energy\n10. the huge canyon geography\n11. the ancient Moon Shrine across the canyon\n12. the damaged moon state\n13. cold dark-fantasy color palette\n14. Nyx’s failed flight attempt\n15. Zara commanding the guardians\n16. bridge construction\n17. Kael/Zara race across bridge\n18. Nyx becoming magically exhausted\n19. the moon shard falling\n20. Kael catching the shard\n21. final “KAEL! GO!” cliffhanger\n\nTranslate the storyboard’s fifteen one-second panels into approximately **8 cinematic shots**.\n\nDO NOT create fifteen frantic one-second cuts.\n\nUse clear:\n\nmission\n→ failed solution\n→ impossible obstacle\n→ magical solution\n→ race\n→ moon failure\n→ falling shard\n→ heroic catch.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:30–0:45 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nfantasy adventure / magical teamwork\n\nDialogue:\nnative synchronized English dialogue\n\nAudio:\nnative dialogue + environmental sound + magic SFX + skeleton movement + original cinematic score\n\nEMOTIONAL ARC:\n\nurgent mission\n→ momentary hope\n→ failed shortcut\n→ impossible canyon\n→ Zara’s clever solution\n→ spectacular magical bridge\n→ comedic relief\n→ desperate race\n→ moon nearly dies\n→ shard falls\n→ Kael accepts responsibility\n\nCORE PART 3 PAYOFF:\n\nThe dragon carried the moonlight this far.\n\nThe Bonekeeper creates the path.\n\nNow the boy who wrongly accused them must become the one who carries the moonlight forward.\n\n--------------------------------------------------\n\n[PART 3 STORY LIMIT]\n\nPART 3 ONLY.\n\nSTART:\n\nZara points toward the ancient Moon Shrine across the valley.\n\nNyx still holds the moon shard.\n\nEND:\n\nKael has caught the moon shard safely with both hands.\n\nThe Moon Shrine towers immediately ahead.\n\nZara points toward it and shouts:\n\n<d>[English] KAEL! GO!</d>\n\nCUT TO BLACK.\n\nDO NOT show:\n- Kael climbing the shrine\n- Kael saying “Seriously?!”\n- Nyx lifting Kael toward the receptacle\n- shard being inserted\n- shrine activation\n- moon restoration\n- skeleton-to-firefly transformation\n- restored Noctara\n- final friendship dialogue\n\nThose belong to PART 4.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch Parts 1–2 and the uploaded PART 3 storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- pale-to-medium cool-toned skin\n- youthful angular stylized face\n- large expressive dark eyes\n- thick black eyebrows\n- swept-back tousled black hair\n- distinctive windswept pointed silhouette\n- lean narrow build\n- long thin arms and legs\n\nClothing:\n- simple charcoal-black tunic\n- dark trousers\n- practical dark boots\n- subtle belt\n- no armor\n- no cape\n- no weapon\n\nPART 3 PERFORMANCE ARC:\n\n0–3 sec:\nfocused on mission\n\n3–6 sec:\nhope → frustration\n\n6–10 sec:\nastonished by Zara’s solution\n\n10–12 sec:\nurgent sprint\n\n12–14 sec:\nfear when Nyx weakens\n\n14–15 sec:\ninstinctive courage and determination\n\nKael must NOT suddenly behave like a superhero.\n\nHis decisive act is simply:\n\nthe shard falls near the bridge\n→ Kael lunges/slides safely\n→ catches it with both hands.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — ZARA]\n\nMatch PART 2 and the uploaded storyboard EXACTLY.\n\nZARA — THE BONEKEEPER\n\nAppearance:\n- deep warm-brown skin\n- enormous dense curly black hair\n- cool-blue moonlight highlights through curls\n- large dark-brown eyes\n- strong expressive eyebrows\n- blue distressed top\n- pale stylized ribcage motif printed on the fabric\n- brown utility belt\n- practical fantasy trousers / waist wraps\n- forearm wraps\n- sturdy boots\n- compact athletic build\n\nPersonality:\n- fearless\n- intelligent\n- dry humor\n- decisive\n- experienced\n- protective\n\nPART 3 ROLE:\n\nZara is the tactical leader.\n\nShe knows:\n- the destination\n- the urgency\n- the valley\n- the skeleton guardians\n- how to create a safe route\n\nPerformance:\nurgent explanation\n→ skepticism at Nyx’s flight attempt\n→ confident idea\n→ commanding Bonekeeper\n→ sprinting leader\n→ alarm when shard falls\n→ decisive final “GO!”\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch Parts 1–2 EXACTLY.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- huge elongated black dragon\n- long serpentine neck\n- layered black-blue feather-like scales\n- wild backward-pointing spines\n- large pointed ears/horns\n- gigantic glowing amber-orange eyes\n- vertical pupils\n- luminous toothy grin\n- flexible long body\n- large dark wings\n\nAt the start:\nNyx still carries the ONE moon fragment carefully between his jaws.\n\nPART 3 PERFORMANCE:\n\ndetermined\n→ confident flight attempt\n→ surprised by unstable magic\n→ safely pushed back\n→ slightly annoyed\n→ races alongside friends\n→ becomes progressively magically exhausted\n→ loses strength when moon nearly dies\n→ accidentally loses the shard\n\nIMPORTANT:\n\nNyx is NOT physically injured.\n\nHe is experiencing magical exhaustion because his power is tied to the failing moon.\n\nNo fire breathing.\nNo attack.\nNo crash.\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN IDENTITY LOCK]\n\nMatch Part 2 exactly.\n\nAppearance:\n- clean ivory/cream bones\n- rounded stylized proportions\n- expressive skulls\n- large hollow eye sockets\n- pale blue magical eye/runic glow\n- family-friendly\n- individual personality\n\nNo:\n- gore\n- flesh\n- decay\n- horror\n- corpse realism\n\nPART 3 ROLE:\n\nThey voluntarily become magical architecture.\n\nTheir bodies may harmlessly separate and reconnect because they are magical constructs.\n\nThis is NOT physical dismemberment.\n\nVisual logic:\n\nrunes activate\n→ guardian pieces float\n→ components organize\n→ rib arches connect\n→ stable walkway forms\n→ Living Bone Bridge completed.\n\nThe skeleton guardians remain conscious and expressive.\n\nOne skull may wave at Kael.\n\nAnother may look mildly annoyed at becoming a railing.\n\n--------------------------------------------------\n\n[LIVING BONE BRIDGE LOCK]\n\nMatch the storyboard reference EXACTLY.\n\nThe bridge must look like elegant ancient magical architecture.\n\nDesign:\n- ivory structural arches\n- wide stable central walkway\n- curved rib-like side rails\n- guardian skull nodes\n- repeating skeleton-inspired architecture\n- pale cyan/blue magical runes\n- silver-blue energy flowing through joints\n- elegant curve across enormous canyon\n\nThe bridge must feel:\nsafe,\nsolid,\nbeautiful,\nstrange,\nslightly comedic.\n\nNOT:\na pile of bones.\n\nNOT:\na horror bridge.\n\n--------------------------------------------------\n\n[MOON SHRINE LOCK]\n\nUse the same shrine shown in the storyboard reference.\n\nLocation:\nhigh cliff on far side of canyon.\n\nDesign:\n- ancient dark stone\n- towering crescent/lunar arch\n- silver geometric carvings\n- inactive central altar\n- high fragment receptacle\n- ruined stairs\n- faint dormant runes\n\nIn PART 3:\n\nShrine remains completely INACTIVE.\n\nNo activation.\n\nNo silver energy wave.\n\nNo restored moon.\n\n--------------------------------------------------\n\n[MOON FRAGMENT LOCK]\n\nOnly ONE fragment exists.\n\nAppearance:\n- irregular crescent/crystalline silver shard\n- bright white-silver core\n- pale cyan translucent edges\n- soft floating silver particles\n- strong cool light\n\nTimeline:\n\n0–12+ sec:\nheld by Nyx\n\n12–14 sec:\nNyx weakens\n\naround 14 sec:\nshard slips free\n\nending:\nKael catches the SAME fragment with both hands\n\nDo NOT duplicate the shard.\n\nOnce Kael catches it:\nNyx no longer has it.\n\n--------------------------------------------------\n\n[MOON STATE CONTINUITY]\n\nStart:\nheavily cracked and dim.\n\nMiddle:\ncontinues losing brightness.\n\nNear 12–13 sec:\nalmost completely dark.\n\nUse:\n- faint silver rim\n- weak crack lines\n\nAs moon darkens:\nNyx’s strength decreases.\n\nThe moon shard becomes the strongest silver light source in the scene.\n\nDo NOT completely restore it.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK]\n\nContinue the Valley of Bones from PART 2.\n\nUse:\n- towering black-blue cliffs\n- massive canyon\n- pale moonlit mist\n- ruined paths\n- ancient arches\n- silver-gray vegetation\n- fossil-like formations\n- broken original stone bridge\n- distant Moon Shrine\n- deep shadowed valley\n- cold wind\n\nThe environment should feel enormous.\n\nScale is critical.\n\n--------------------------------------------------\n\n[STORY / CAUSE-AND-EFFECT LOGIC]\n\nZara shows destination\n→ group races toward shrine.\n\nNyx sees canyon\n→ tries flying directly.\n\nBroken moon disrupts sky magic\n→ magical current pushes Nyx safely back.\n\nThey discover destroyed bridge\n→ Kael sees no route.\n\nZara sees skeleton guardians\n→ gets solution.\n\nGuardians activate\n→ form Living Bone Bridge.\n\nBridge completed\n→ Kael and Zara sprint across.\n\nMoon dims\n→ Nyx loses magical strength.\n\nNyx weakens\n→ moon shard slips.\n\nShard falls\n→ Kael lunges safely.\n\nKael catches fragment\n→ responsibility transfers to him.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.70s\nTHE SHRINE ACROSS THE VALLEY\n\nContinue DIRECTLY from Part 2.\n\nWide cinematic shot.\n\nThe cracked moon hangs above a jagged mountain.\n\nZara points toward a distant ancient Moon Shrine across the huge valley.\n\nKael follows her gesture.\n\nNyx stands beside them holding the glowing shard.\n\nSkeleton guardians gather behind.\n\nZara speaks quickly:\n\n<d>[English] The shard has to reach that shrine.</d>\n\nKael sees how far away it is.\n\nZara:\n<d>[English] Before the moon goes dark.</d>\n\nThe moon flickers faintly.\n\nCamera:\ngroup medium-wide\n→ follow Zara’s pointing arm\n→ extreme wide reveal of distant shrine.\n\nSound:\n- cold canyon wind\n- faint moon hum\n- shard shimmer\n\nMusic:\nurgent low strings begin.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.70s–3.40s\nNYX TRIES THE OBVIOUS SOLUTION\n\nNyx looks from the shrine to the canyon.\n\nHis giant amber eyes narrow with confidence.\n\nHe gently adjusts the shard in his jaws.\n\nKael looks at him.\n\nKael:\n<d>[English] Can’t he just fly it there?</d>\n\nNyx opens his wings.\n\nZara:\n<d>[English] He can try.</d>\n\nNyx launches upward.\n\nCamera follows him dramatically.\n\nFor half a second:\nit appears to work.\n\nThen pale silver-blue magical currents begin spiraling across the sky.\n\nSound:\n- powerful wing beats\n- rising wind\n- magical air resonance\n\n--------------------------------------------------\n\n[SHOT 3] — 3.40s–4.90s\nTHE BROKEN MOON HAS TWISTED THE SKY\n\nThe silver-blue current wraps around Nyx.\n\nNOT an attack.\n\nIt behaves like an unstable magical jet stream.\n\nNyx struggles to move forward.\n\nHis long body bends with the current.\n\nHis wings compensate.\n\nThe shard remains safely secured.\n\nThen the current gently but firmly pushes him backward.\n\nNyx redirects downward and lands safely beside Kael and Zara.\n\nHis head spines are blown backward.\n\nHe looks up at the current with deeply offended disbelief.\n\nKael:\n<d>[English] ...So that’s a no.</d>\n\nNyx gives Kael an irritated side-eye.\n\nBrief comedy beat.\n\nSound:\n- rushing magical wind\n- wing strain\n- safe landing scrape\n- Nyx annoyed exhale\n\nNo crash.\n\n--------------------------------------------------\n\n[SHOT 4] — 4.90s–6.50s\nTHE BRIDGE IS GONE\n\nThe group reaches the canyon edge.\n\nCamera moves beyond Kael and dramatically pulls wide.\n\nReveal:\n\nan enormous chasm.\n\nThe original stone bridge is destroyed.\n\nOnly broken stone ends remain.\n\nAcross the void:\n\nthe Moon Shrine.\n\nKael steps close but safely behind the edge.\n\nHis face drops.\n\nDialogue:\n\nKael:\n<d>[English] How are we supposed to cross THAT?</d>\n\nNyx looks across.\n\nZara does not answer immediately.\n\nShe slowly turns toward the skeleton guardians behind them.\n\nA tiny confident smile appears.\n\nCamera:\nKael reaction\n→ huge canyon reveal\n→ Zara glance to guardians.\n\nMusic briefly suspends.\n\n--------------------------------------------------\n\n[SHOT 5] — 6.50s–8.80s\nTHE BONEKEEPER’S ANSWER\n\nHero medium-low shot of Zara.\n\nShe raises one hand.\n\nThe skeleton guardians straighten.\n\nTheir eye sockets glow pale cyan.\n\nBlue magical runes illuminate across their ivory forms.\n\nKael slowly turns.\n\nDialogue:\n\nKael:\n<d>[English] Zara...?</d>\n\nZara:\n<d>[English] Guardians.</d>\n\nShe sweeps her hand toward the canyon.\n\nZara:\n<d>[English] Bridge.</d>\n\nThen the magical construction begins.\n\nSkeleton guardian components float harmlessly apart like enchanted puzzle pieces.\n\nNo pain.\n\nNo fear.\n\nShow:\n- rib forms lifting\n- long ivory structural pieces rotating\n- skull guardian nodes floating into position\n- cyan runes linking pieces\n- stable walkway building outward\n\nOne floating skull looks toward Kael while being carried into position.\n\nIt gives him a cheerful wave.\n\nKael stares.\n\nSound:\n- soft bone clacks\n- runic hum\n- magical assembling tones\n- airy movements\n\nMusic:\nrising magical orchestral pattern.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.80s–10.60s\nTHE LIVING BONE BRIDGE\n\nPull VERY wide.\n\nThe final sections connect.\n\nA beautiful Living Bone Bridge now spans the entire canyon.\n\nIvory arches.\nPale blue rune energy.\nStable wide walkway.\nCurved guardian rails.\n\nThe Moon Shrine waits at the far side.\n\nKael is speechless.\n\nHe touches / looks at one guardian rail.\n\nA skeleton skull embedded nearby turns toward him.\n\nKael:\n<d>[English] That is both amazing...</d>\n\nTiny beat.\n\n<d>[English] ...and deeply uncomfortable.</d>\n\nZara is already running onto the bridge.\n\nShe looks back:\n\n<d>[English] RUN!</d>\n\nThe skeleton skull gestures forward as if agreeing with her.\n\nCamera:\ngiant bridge reveal\n→ Kael reaction\n→ Zara sprinting.\n\nMusic shifts directly into adventure momentum.\n\n--------------------------------------------------\n\n[SHOT 7] — 10.60s–12.70s\nRACE TOWARD THE SHRINE\n\nDynamic but smooth tracking shot.\n\nKael and Zara sprint across the Living Bone Bridge.\n\nZara leads.\n\nKael follows.\n\nNyx races/glides beside them through the lower stable air current, still holding the moon fragment.\n\nThe Moon Shrine grows larger.\n\nUse:\n- side tracking\n- bridge parallax\n- running close-ups\n- shrine destination ahead\n\nSkeleton guardian skulls along the rails react as they pass.\n\nOne appears to cheer them on silently.\n\nKael:\n<d>[English] Almost there!</d>\n\nThen the environment changes.\n\nMoonlight weakens.\n\nThe bridge’s cyan glow becomes more noticeable because the sky is darkening.\n\nNyx slows slightly.\n\n--------------------------------------------------\n\n[SHOT 8] — 12.70s–15.00s\nTHE LAST PIECE OF MOONLIGHT\n\nThe music becomes tense.\n\nCut toward the moon.\n\nIt is almost entirely dark.\n\nOnly a faint silver rim and weak crack lines remain.\n\nBack to Nyx.\n\nHis amber eyes dim slightly.\n\nHis wing/body movement becomes heavy.\n\nHe tries to continue.\n\nZara notices.\n\nZara:\n<d>[English] Nyx?!</d>\n\nNyx gives a tired exhale.\n\nThe moon shard begins slipping from his jaws.\n\nClose-up:\n\nsilver shard moves free.\n\nZara:\n<d>[English] KAEL!</d>\n\nThe fragment falls beside the bridge.\n\nUse a brief slow-motion feeling.\n\nIt spins through darkness.\n\nThe brilliant silver light illuminates:\nKael’s face,\nZara,\nNyx,\nthe ivory rails.\n\nKael reacts instinctively.\n\nHe does NOT jump from the bridge.\n\nHe lunges and slides safely across the broad walkway.\n\nOne hand grips / braces near the stable rib rail.\n\nHis other hand reaches.\n\nThen both hands close around the fragment.\n\nCATCH.\n\nSilver light bursts softly across his face.\n\nNo explosion.\n\nKael looks down at what he is holding.\n\nThen toward the towering Moon Shrine now close ahead.\n\nZara is already pointing toward the shrine.\n\nZara shouts:\n\n<d>[English] KAEL! GO!</d>\n\nKael rises with the fragment secured against his chest / in both hands.\n\nHis fear becomes determination.\n\nCamera:\nmoon fading\n→ Nyx weakening\n→ shard slipping\n→ slow follow of falling shard\n→ Kael safe lunge\n→ close-up catch\n→ dramatic low angle on Kael with Moon Shrine behind.\n\nHold final frame briefly:\n\nKael = carrying the last moonlight  \nZara = pointing forward  \nNyx = exhausted but safe  \nBone Bridge = beneath them  \nMoon Shrine = waiting  \nMoon = nearly dark\n\nCUT TO BLACK.\n\nEND PART 3.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage boy voice.\n\nQualities:\n- youthful\n- clear\n- thoughtful\n- slightly anxious\n- increasingly brave\n- emotionally sincere\n\nPerformance progression:\n\n0–5 sec:\nfocused / questioning\n\n5–7 sec:\nfrustrated by canyon\n\n7–10 sec:\namazed and comedic\n\n10–13 sec:\nurgent\n\n13–15 sec:\nstartled → determined\n\nImportant comedy line:\n\n<d>[English] That is both amazing... and deeply uncomfortable.</d>\n\nDeliver genuinely, not as broad slapstick.\n\n--------------------------------------------------\n\n[ZARA VOICE]\n\nNatural teenage girl voice.\n\nQualities:\n- grounded\n- confident\n- quick\n- authoritative\n- dry humor\n- clear English\n\nKey dialogue:\n\n<d>[English] The shard has to reach that shrine.</d>\n\n<d>[English] Guardians. Bridge.</d>\n\n<d>[English] RUN!</d>\n\nFinal:\n\n<d>[English] KAEL! GO!</d>\n\nHer final line should be urgent and powerful but not screaming hysterically.\n\n--------------------------------------------------\n\n[NYX AUDIO]\n\nNo normal dialogue required.\n\nUse:\n- wing beats\n- deep breathing\n- soft throat resonance\n- annoyed exhale after failed flight\n- heavier breathing as moon fades\n- subtle weakened rumble near ending\n\nNo:\n- roaring\n- growling threats\n- fire\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN AUDIO]\n\nUse charming magical mechanical-bone sounds:\n\n- soft clacks\n- rounded wooden-like clicks\n- magical rune hum\n- light pieces aligning\n- tiny skull rotation sound\n- subtle comedic wave click\n\nBridge completion:\none warm resonant magical “THUMM.”\n\nNot a violent impact.\n\n--------------------------------------------------\n\n[MAGICAL SKY CURRENT AUDIO]\n\nUse:\n- rushing airy energy\n- glassy harmonic sweeps\n- low moon resonance\n\nDo NOT sound like:\n- tornado\n- explosion\n- violent storm\n\n--------------------------------------------------\n\n[MOON FRAGMENT AUDIO]\n\nSignature sound:\nclear silver crystalline hum.\n\nAs Nyx weakens:\nhum becomes more exposed.\n\nWhen shard slips:\nascending/falling glass tone.\n\nWhen Kael catches:\nsoft resonant silver “WHUM.”\n\nNo explosive impact.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nMaintain:\n- canyon wind\n- distant stone echo\n- mist ambience\n- character footsteps\n- bridge footsteps\n- bridge rune hum\n- distant Moon Shrine resonance\n\nDuring final shard fall:\nreduce environment temporarily so the fragment sound dominates.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL cinematic fantasy-adventure score.\n\nInstrumentation:\n- urgent strings\n- cello\n- low brass\n- frame drums\n- hand percussion\n- celesta\n- glockenspiel\n- glass harmonics\n- airy choir without lyrics\n- warm horn layer\n- low piano\n\nMUSIC ARC:\n\n0.00–1.70\nurgent mission setup.\n\n1.70–3.40\nhopeful lift attempt.\n\n3.40–4.90\nmagical turbulence.\n\n4.90–6.50\ndrop into uncertainty at canyon reveal.\n\n6.50–8.80\nmysterious magical build.\n\n8.80–10.60\nlarge uplifting Bone Bridge reveal.\n\n10.60–12.70\ndriving adventure rhythm.\n\n12.70–14.00\nmusic loses momentum as moon fades.\n\n14.00–15.00\ntense silver motif during shard fall →\nstrong hopeful chord on Kael’s catch →\nunresolved cut after “GO!”\n\nNo epic war music.\nNo horror score.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 3 should feel larger and more kinetic than Parts 1–2.\n\nCAMERA ARC:\n\n0–1.7 sec:\ndestination-focused wide composition.\n\n1.7–4.9 sec:\nupward and aerial Nyx coverage.\n\n4.9–6.5 sec:\nmassive canyon reveal.\n\n6.5–10.6 sec:\nmagical construction escalation.\n\n10.6–12.7 sec:\ndynamic tracking action.\n\n12.7–15 sec:\nslow time slightly for emotional/action climax.\n\nUse:\n- wide environmental reveals\n- low hero angle on Zara\n- upward dragon tracking\n- lateral bridge tracking\n- controlled parallax\n- close expression inserts\n- macro moon-fragment shot\n- dramatic final low angle\n\nAvoid:\n- shaky handheld\n- crash zoom\n- constant rapid cutting\n- 360 spins\n- extreme fisheye\n- Dutch-angle overload\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH PARTS 1–2 EXACTLY.\n\nUse:\n- dark expressive ink contours\n- watercolor/gouache-style painterly fills\n- cinematic moonlight\n- richly layered blue-black cliffs\n- beautiful cyan magical effects\n- expressive character faces\n- intricate dragon-scale textures\n- elegant skeleton design\n- atmospheric mist\n- soft silver bloom\n- subtle squash/stretch only where appropriate\n\nThe result should look like the uploaded storyboard artwork brought to cinematic life.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealism\n- live action\n- 3D CGI\n- anime\n- flat vector animation\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\nPART 3 opening:\n- deep navy\n- charcoal\n- cool silver moon\n- amber Nyx eyes\n- bright shard\n\nFlight-current sequence:\n- add pale cyan / silver energy ribbons\n\nCanyon:\n- darker blues\n- gray mist\n\nBone Bridge:\n- ivory architecture\n- beautiful pale cyan runic light\n\nRace:\n- cyan bridge glow\n- fading moonlight\n- silver shard\n\nEnding:\nthe moon is nearly dark.\n\nThe moon fragment becomes the DOMINANT illumination.\n\nKael’s face in the final shot should be lit strongly by:\n- silver-white light from below/front\n- faint cyan bridge bounce\n- little to no moon fill\n\nThe shrine remains dark behind him.\n\n--------------------------------------------------\n\n[MOTION RULES — KAEL]\n\nUse:\n- realistic stylized running\n- natural acceleration\n- controlled balance\n- expressive upper-body acting\n- hair and tunic responding to motion\n\nFor shard catch:\n\nKael MUST remain on the bridge.\n\nAction:\n\nfragment slips\n→ Kael sees it\n→ lunges across safe walkway\n→ slides on one knee / side\n→ braces against stable railing\n→ grabs fragment with both hands\n→ returns fully onto walkway\n\nNo jumping into canyon.\n\n--------------------------------------------------\n\n[MOTION RULES — ZARA]\n\nUse:\n- confident forward motion\n- strong pointing gestures\n- controlled hand command for skeletons\n- hair with substantial secondary motion\n- athletic natural running\n\nDuring bridge command:\nher movements should be minimal but decisive.\n\nShe does not need exaggerated spellcasting.\n\nThe guardians already recognize her.\n\n--------------------------------------------------\n\n[MOTION RULES — NYX]\n\nNyx must feel huge and weighty.\n\nFlight attempt:\n- strong takeoff\n- broad wing strokes\n- long body follow-through\n- magical current bends trajectory\n- Nyx corrects\n- safe landing\n\nRace:\n- lower controlled gliding\n- occasional foot/wing assistance if near bridge\n\nExhaustion:\n- slower wings\n- lowering head\n- reduced eye glow\n- heavier breathing\n\nNo physical collapse.\n\n--------------------------------------------------\n\n[MOTION RULES — SKELETON BRIDGE]\n\nTransformation:\n\nguardian runes illuminate\n→ components float apart cleanly\n→ pieces rotate\n→ long sections align\n→ pale energy connects\n→ walking surface forms\n→ arches rise\n→ final far section locks into place.\n\nMovement should resemble:\nbeautiful magical clockwork / puzzle assembly.\n\nNo violent snapping.\n\nNo painful reactions.\n\n--------------------------------------------------\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086322454679707867"
  },
  {
    "id": "x-2086321924070834347",
    "title": "归还月亮的少年 · 第二幕",
    "titleEn": "The Boy Who Returned the Moon · Chapter 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《归还月亮的少年 · 第二幕》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Boy Who Returned the Moon · Chapter 2. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, costume, scale, prop, lighting, performance, and narrative reference.\n\nThis sequence continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n- Kael entering the forbidden Valley of Bones\n- Nyx disappearing deeper into the canyon\n- Nyx still carrying the ONLY glowing silver moon fragment\n- the giant moon remaining cracked and visibly dimmer\n- Kael believing Nyx stole the fragment\n- Kael deciding to follow him despite his fear\n\nGenerate a FINISHED cinematic animated sequence from this storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- storyboard panels\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- character reference drawings\n- environment reference images\n- storyboard title/header\n- off-white paper background\n- color swatches\n- footer graphics\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact face\n2. Kael’s swept-back black hairstyle\n3. Kael’s dark tunic and proportions\n4. Zara’s exact face and enormous curly hairstyle\n5. Zara’s blue skeletal-pattern top\n6. Nyx’s exact midnight-dragon design\n7. Nyx’s enormous amber eyes\n8. Nyx’s luminous toothy grin\n9. Nyx’s black feather-like scales\n10. the ONE silver moon fragment\n11. the stylized ivory skeleton guardians\n12. the Valley of Bones geography\n13. blue-gray moonlit mist\n14. the cracked dimming moon\n15. cold midnight-blue color palette\n16. Part 2 story progression\n17. Zara’s central truth reveal\n18. final “we’re running out of time” cliffhanger\n\nTranslate the storyboard’s 15 panels into approximately **8 cinematic shots**.\n\nDO NOT create fifteen rapid one-second cuts.\n\nUse clear progression:\n\nvalley mystery\n→ skeleton reveal\n→ Zara entrance\n→ comedy beat\n→ Nyx appears\n→ Kael accuses him\n→ Zara reveals the truth\n→ moon dims\n→ urgent alliance cliffhanger.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:15–0:30 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nfantasy mystery → adventure revelation\n\nDialogue:\nnative synchronized English dialogue\n\nAudio:\nnative dialogue + environmental sound + skeleton SFX + moon magic + original cinematic score\n\nEMOTIONAL ARC:\n\ncautious exploration\n→ eerie surprise\n→ overwhelmed\n→ mysterious hero entrance\n→ quirky comedy\n→ renewed tension\n→ confrontation\n→ major truth reveal\n→ cosmic urgency\n\nCORE PART 2 TWIST:\n\nKael believes Nyx stole the moon fragment.\n\nZara reveals:\n\n“He’s trying to return it.”\n\nThis line must land as the CENTRAL reversal of the entire sequence.\n\n--------------------------------------------------\n\n[PART 2 STORY LIMIT]\n\nPART 2 ONLY.\n\nSTART:\n\nKael enters deeper into the Valley of Bones alone.\n\nEND:\n\nKael, Zara, Nyx and the skeleton guardians look toward the distant mountain as the cracked moon dims again.\n\nZara says:\n\n<d>[English] And we’re running out of time.</d>\n\nCUT TO BLACK.\n\nDO NOT show:\n- detailed Moon Shrine explanation\n- Nyx trying to fly to shrine\n- magical wind pushing Nyx back\n- destroyed bridge\n- Bone Bridge construction\n- skeletons forming bridge\n- characters crossing canyon\n- fragment falling\n- Kael catching the fragment\n- shrine activation\n- moon restoration\n- skeleton-to-firefly transformation\n- final friendship ending\n\nThose belong to Parts 3 and 4.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch PART 1 and the uploaded PART 2 storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- youthful angular stylized face\n- pale-to-medium cool skin\n- large expressive dark eyes\n- thick dark eyebrows\n- swept-back messy black hair\n- recognizable pointed hair silhouette\n- lean narrow build\n- long arms and legs\n\nClothing:\n- simple charcoal-black tunic\n- dark trousers\n- dark practical boots\n- subtle belt\n- no armor\n- no cape\n- no weapon\n\nPersonality:\n- observant\n- courageous\n- impulsive\n- suspicious of Nyx\n- uncomfortable in the mysterious valley\n- willing to confront what he believes is wrong\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\ncareful, alert, nervous\n\n3–5 sec:\nstartled and overwhelmed\n\n5–8 sec:\nconfused by Zara and skeleton behavior\n\n8–11 sec:\nangry certainty returns when Nyx appears\n\n11–13 sec:\nshock as Zara contradicts him\n\n13–15 sec:\nbelief begins changing; urgency takes over\n\nIMPORTANT:\n\nKael is not hateful toward Nyx.\n\nHe genuinely believes he witnessed a theft.\n\nHis mistake must feel understandable.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — ZARA]\n\nMatch the uploaded storyboard EXACTLY.\n\nZARA — THE BONEKEEPER\n\nAppearance:\n- teenage girl\n- deep warm-brown skin\n- enormous dense black curly hair\n- huge rounded hair silhouette\n- subtle cool-blue moonlight rim on curls\n- large expressive brown eyes\n- strong dark eyebrows\n- blue short-sleeved distressed fantasy top\n- stylized pale rib/skeleton motif ON THE FABRIC ONLY\n- brown utility belt\n- dark/brown practical fantasy trousers\n- wrapped forearms\n- practical boots\n- compact athletic build\n\nIMPORTANT:\n\nThe rib design is clothing artwork.\n\nNo exposed anatomy.\n\nPersonality:\n- fearless\n- calm\n- clever\n- dry sense of humor\n- protective of the valley\n- emotionally grounded\n- already understands Nyx’s mission\n\nZARA PERFORMANCE:\n\nEntrance:\nconfident and controlled.\n\nComedy:\ndeadpan.\n\nTruth reveal:\ncompletely serious.\n\nEnding:\nurgent.\n\nShe should visually transform the skeleton-filled environment from “dangerous” into “strange but safe.”\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch PART 1 exactly.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- large elongated black dragon\n- long flexible serpentine neck\n- black-blue feather-like layered scales\n- wild swept-back head spines\n- large pointed ears/horns\n- massive glowing amber-orange eyes\n- vertical slit pupils\n- dramatic luminous teeth\n- elongated expressive snout\n- long dark body\n\nThe same moon shard remains with him.\n\nIMPORTANT:\n\nNyx is NOT eating the fragment.\n\nHold it carefully.\n\nHis body language should subtly reveal protectiveness:\n- careful head position\n- controlled jaw grip\n- fragment protected close to him\n- concern when moon weakens\n\nWhen Kael accuses him:\n\nNyx may blink,\ntilt his head,\nnarrow one eye,\nor look mildly offended.\n\nDo NOT make him attack.\n\nDo NOT make him roar.\n\nDo NOT use fire.\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN IDENTITY LOCK]\n\nMatch the storyboard.\n\nAncient magical skeleton guardians.\n\nAppearance:\n- smooth ivory / cream bones\n- stylized oversized skulls\n- large hollow expressive eye sockets\n- rounded family-friendly proportions\n- subtle pale-blue magical highlights if needed\n- small differences in height and personality\n\nNo:\n- flesh\n- gore\n- decay\n- blood\n- torn body remnants\n- horror texture\n\nThey are not zombies.\n\nThey are magical ancient guardians.\n\nPERFORMANCE ARC:\n\nfirst appearance:\nmysterious.\n\ngroup reveal:\nvisually intimidating only because there are many.\n\nafter Zara arrives:\ncurious and quirky.\n\nComedy examples:\n- one adjusts its own skull\n- one leans forward curiously\n- one scratches its head\n- one gives Kael a friendly wave\n- one looks embarrassed by the waving skeleton\n\nThey never attack Kael.\n\n--------------------------------------------------\n\n[MOON FRAGMENT IDENTITY LOCK]\n\nOnly ONE fragment exists.\n\nMatch Part 1.\n\nAppearance:\n- irregular silver-white crescent/crystal fragment\n- brilliant white core\n- translucent pale cyan edges\n- soft silver particle trail\n- cool moonlight aura\n\nIt remains with Nyx throughout Part 2.\n\nNo duplication.\n\nNo shape change.\n\nNo disappearance.\n\n--------------------------------------------------\n\n[MOON STATE CONTINUITY]\n\nThe moon begins Part 2:\n\n- cracked\n- missing one section\n- dimmer than at the beginning of Part 1\n\nDuring the ending:\n\na deep magical resonance occurs.\n\nMoon brightness decreases noticeably again.\n\nDo NOT:\n- break another giant piece away\n- destroy the moon\n- completely turn it black yet\n- restore it\n\nThe change should communicate:\n\nTIME IS RUNNING OUT.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — VALLEY OF BONES]\n\nUse the exact environment established by the uploaded storyboard.\n\nEnvironment:\n- huge black-blue canyon walls\n- tall jagged rock towers\n- cold moonlit path\n- pale blue ground mist\n- enormous fossil-like rib structures\n- ancient ruined stone features\n- sparse silver-gray vegetation\n- giant dark mountain in distance\n- scattered magical skeleton guardians\n- cold deep blue shadows\n- cracked moon overhead\n\nVisual feeling:\nancient,\nforbidden,\nquiet,\nmysterious,\nbeautiful.\n\nNOT horror.\n\nNo modern objects.\n\n--------------------------------------------------\n\n[WORLD / STORY LOGIC]\n\nClear cause-and-effect:\n\nKael enters valley\n→ hears rattling\n\nrattling\n→ skeleton guardians appear\n\nKael assumes danger\n→ Zara intervenes\n\nKael sees skeletons responding around Zara\n→ assumes she controls them\n\nZara says they act independently\n→ comedy breaks tension\n\nNyx arrives\n→ Kael’s suspicion immediately returns\n\nKael sees fragment\n→ accuses Nyx\n\nZara corrects him\n→ Kael’s understanding changes\n\nmoon dims again\n→ Zara reveals urgency\n\nNo unexplained magical action.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–2.00s\nTHE VALLEY IS TOO QUIET\n\nContinue directly from Part 1.\n\nWide rear shot.\n\nKael carefully walks deeper into the Valley of Bones.\n\nThe environment towers around him.\n\nHuge rib-like fossil structures curve from the ground.\n\nCold mist moves around his boots.\n\nThe cracked moon hangs high between canyon walls.\n\nFar behind him, Noctara’s warm castle glow is nearly gone.\n\nCamera slowly tracks backward in front of Kael.\n\nKael looks left.\n\nThen right.\n\nHis footsteps echo.\n\nHe whispers:\n\n<d>[English] Where did you go...?</d>\n\nNo answer.\n\nHe hears:\n\nCLACK...\n\nKael stops.\n\nCamera:\nwide valley\n→ tracking medium\n→ slow push toward Kael’s reaction.\n\nSound:\n- empty canyon wind\n- footsteps on stone\n- distant rock echoes\n- faint moon hum\n- one dry magical bone “clack”\n\nMusic:\nvery sparse low strings + distant celesta.\n\n--------------------------------------------------\n\n[SHOT 2] — 2.00s–3.80s\nONE SKELETON BECOMES AN AUDIENCE\n\nLow shot near Kael’s boots.\n\nA skeletal hand rises gently from behind a rock.\n\nKael freezes.\n\nAnother skull slowly peeks from behind a fossil arch.\n\nThen another guardian sits upright.\n\nKael turns.\n\nSeveral more skeletons rise from around the path.\n\nNo attack.\n\nThey simply stare.\n\nOne adjusts its skull so it sits straighter.\n\nKael slowly backs away.\n\nDialogue:\n\nKael:\n<d>[English] ...Okay.</d>\n\nHe looks behind him.\n\nMore skeletons.\n\nKael:\n<d>[English] That’s a lot of bones.</d>\n\nCamera:\nmacro skeletal hand\n→ Kael reaction\n→ widening circular composition revealing several guardians\n→ wider shot showing Kael surrounded.\n\nSound:\n- soft CLACK\n- gentle bone rattles\n- tiny skull adjustment\n- Kael’s nervous breath\n\nMusic adds light suspense pulses.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.80s–5.40s\nTHE BONEKEEPER\n\nKael continues backing away.\n\nSuddenly—\n\na figure drops safely from a low ruined ledge between Kael and the guardians.\n\nZARA lands in a crouch.\n\nSoft mist/dust puffs around her boots.\n\nShe rises.\n\nHer enormous curly hair catches a beautiful blue moonlight rim.\n\nAll the skeleton guardians stop moving.\n\nStrong cinematic reveal.\n\nKael stares at her.\n\nZara calmly looks at him.\n\nDialogue:\n\nZara:\n<d>[English] You’re loud.</d>\n\nKael:\n<d>[English] I’m surrounded by skeletons!</d>\n\nZara glances around casually.\n\nZara:\n<d>[English] They noticed.</d>\n\nCamera:\nKael surrounded\n→ quick downward motion reveal of Zara landing\n→ low-angle hero shot\n→ medium two-shot.\n\nMusic:\nbrief confident Zara motif.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.40s–7.30s\n“THEY CONTROL THEMSELVES.”\n\nKael points toward the skeleton guardians.\n\nDialogue:\n\nKael:\n<d>[English] You control them?!</d>\n\nZara raises one eyebrow.\n\nZara:\n<d>[English] No.</d>\n\nTiny pause.\n\nZara:\n<d>[English] They control themselves.</d>\n\nBehind her:\n\none skeleton gives Kael a cheerful little wave.\n\nKael slowly looks at it.\n\nThe skeleton waves again.\n\nAnother guardian looks at the waving skeleton with mild embarrassment.\n\nKael lowers his pointing hand slightly.\n\nDialogue:\n\nKael:\n<d>[English] ...Right.</d>\n\nCamera:\nmedium Kael/Zara\n→ rack focus past Zara to waving skeleton\n→ Kael reaction close-up.\n\nSound:\n- gentle finger-bone clack during wave\n- tiny comedic wooden percussion accent\n\nMusic briefly becomes lighter.\n\nThis is the major comedy relief beat.\n\n--------------------------------------------------\n\n[SHOT 5] — 7.30s–9.10s\nTHE DRAGON RETURNS\n\nA subtle silver light appears behind Zara.\n\nSkeleton guardians look toward it.\n\nZara does not panic.\n\nKael notices.\n\nHis expression changes immediately.\n\nCamera moves over Zara’s shoulder.\n\nFrom the darkness between enormous ruins emerges NYX.\n\nHis long black neck enters the moonlight.\n\nAmber eyes glow.\n\nThe silver moon fragment shines brilliantly between his jaws.\n\nKael’s relaxed confusion disappears.\n\nDialogue:\n\nKael:\n<d>[English] You!</d>\n\nNyx stops.\n\nHe looks at Kael.\n\nHe does not attack.\n\nThe glowing shard illuminates the bottom of his face.\n\nCamera:\nsilver light hint\n→ reveal Nyx silhouette\n→ push toward giant amber eye\n→ Kael reaction.\n\nMusic:\nmysterious low dragon motif returns.\n\n--------------------------------------------------\n\n[SHOT 6] — 9.10s–11.00s\n“HE STOLE THAT!”\n\nKael steps forward.\n\nHe points directly at Nyx.\n\nDialogue:\n\nKael:\n<d>[English] He stole that!</d>\n\nNyx’s enormous eye narrows.\n\nHe tilts his head slightly.\n\nExpression:\noffended confusion.\n\nZara turns slowly toward Kael.\n\nKael continues:\n\n<d>[English] I saw him take it from the moon!</d>\n\nNyx carefully adjusts the shard in his jaws without damaging it.\n\nZara looks from Kael to Nyx.\n\nCamera:\nKael foreground\n→ Nyx background\n→ Zara between them\n→ close Nyx reaction.\n\nNo combat stance.\n\nNo aggression.\n\n--------------------------------------------------\n\n[SHOT 7] — 11.00s–13.00s\nTHE STORY IS BACKWARDS\n\nCut closer to Zara.\n\nHer expression becomes completely serious.\n\nThe comedy disappears.\n\nDialogue:\n\nZara:\n<d>[English] No.</d>\n\nKael freezes.\n\nZara looks toward Nyx.\n\nNyx lowers his head slightly, protecting the shard.\n\nZara:\n\n<d>[English] He didn’t steal it.</d>\n\nKael:\n<d>[English] What?</d>\n\nZara points toward the glowing fragment.\n\nThen toward the distant mountain.\n\nZara:\n\n<d>[English] He’s trying to return it.</d>\n\nHOLD Kael’s reaction.\n\nHis pointing hand slowly lowers.\n\nHis eyes move:\n\nNyx\n→ shard\n→ Zara.\n\nHe realizes the story may be completely different from what he believed.\n\nCamera:\nZara serious close-up\n→ Nyx and shard\n→ Kael reaction close-up\n→ three-character composition.\n\nMusic:\nresolve previous suspicious motif into a surprising emotional chord.\n\nIMPORTANT:\n\nGive the line:\n\n“He’s trying to return it.”\n\nclear breathing space.\n\nIt is the MAIN Part 2 twist.\n\n--------------------------------------------------\n\n[SHOT 8] — 13.00s–15.00s\nTHE MOON LOSES MORE LIGHT\n\nBefore Kael can respond—\n\nA deep magical resonance passes through the valley.\n\nBOOOOM...\n\nNot an explosion.\n\nThe ground does NOT violently shake.\n\nEveryone looks upward.\n\nKael.\n\nZara.\n\nNyx.\n\nThe skeleton guardians.\n\nThe cracked moon flickers.\n\nIts silver glow decreases dramatically.\n\nThe moon fragment in Nyx’s jaws pulses brighter by contrast.\n\nNyx’s amber eyes widen with concern.\n\nZara immediately looks toward the distant mountain summit.\n\nFar through the canyon, barely visible:\n\na dark ancient shrine-like silhouette.\n\nKael sees it too.\n\nDialogue:\n\nKael, quieter:\n<d>[English] What’s happening?</d>\n\nZara:\n<d>[English] The moon is fading.</d>\n\nShe turns toward him.\n\nZara:\n<d>[English] And we’re running out of time.</d>\n\nFinal composition:\n\nKael = shocked but listening\nZara = urgent leader\nNyx = misunderstood guardian protecting shard\nSkeletons = alert behind them\nMoon = cracked and dimmer above\nDistant mountain = their next destination\n\nCamera:\ncharacters looking up\n→ huge damaged moon\n→ push toward Zara\n→ pull back into wide group + mountain composition.\n\nCUT TO BLACK.\n\nEND PART 2.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage boy voice.\n\nQualities:\n- youthful\n- intelligent\n- nervous early\n- emotionally expressive\n- clear English\n- sincere\n- not exaggerated\n\nPerformance arc:\n\n0–4 sec:\ncautious whisper.\n\n4–7 sec:\nnervous confusion.\n\n7–11 sec:\naccusatory certainty.\n\n11–13 sec:\nshocked disbelief.\n\n13–15 sec:\nconcern.\n\nImportant line:\n\n<d>[English] He stole that!</d>\n\nshould feel like Kael genuinely believes he is defending the moon.\n\nNot hatred.\n\n--------------------------------------------------\n\n[ZARA VOICE]\n\nNatural teenage girl voice.\n\nQualities:\n- confident\n- calm\n- dry humor\n- slightly low/grounded for her age\n- clear English\n- emotionally controlled\n\nPerformance:\n\nentrance:\ncool confidence.\n\nSkeleton comedy:\ndeadpan.\n\nTruth reveal:\nserious and deliberate.\n\nEnding:\nurgent.\n\nCritical delivery:\n\n<d>[English] He’s trying to return it.</d>\n\nSpeak clearly, calmly, with emphasis on “return.”\n\nThen final:\n\n<d>[English] And we’re running out of time.</d>\n\nshould be faster and more urgent.\n\n--------------------------------------------------\n\n[NYX AUDIO]\n\nNo spoken dialogue in PART 2.\n\nUse:\n- controlled deep breathing\n- soft throat resonance\n- scale/feather movement\n- small offended exhale when Kael accuses him\n- protective low hum around fragment\n\nNo:\n- roaring\n- evil growling\n- laughter\n- fire sounds\n\n--------------------------------------------------\n\n[SKELETON AUDIO]\n\nUse light stylized bone sounds:\n\n- soft clacks\n- gentle rattles\n- skull adjustment “tok”\n- finger-wave clicks\n- tiny curious head movement sounds\n\nNever make them sound gruesome.\n\nAvoid:\n- horror screams\n- zombie moans\n- shrieking\n- threatening rattling walls of sound\n\n--------------------------------------------------\n\n[VALLEY ENVIRONMENT AUDIO]\n\nMaintain:\n- cold canyon wind\n- stone echoes\n- subtle mist movement\n- occasional distant rock resonance\n- faint moon magic\n- Kael footsteps\n- sparse atmospheric ambience\n\nCompared with Noctara:\n\nmuch emptier,\ncolder,\nmore echoing.\n\n--------------------------------------------------\n\n[MOON DIMMING SOUND]\n\nAt final:\n\nuse a distant deep celestial resonance.\n\nStructure:\n\nsoft low pulse\n→ faint crystalline tremor\n→ moonlight hum weakens\n→ fragment briefly pulses\n\nNo explosion.\n\nNo thunder.\n\nNo destructive shockwave.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL fantasy mystery score.\n\nInstrumentation:\n- low strings\n- cello\n- bass clarinet\n- soft frame drum\n- celesta\n- glass harmonics\n- muted pizzicato\n- airy female/neutral choir texture without lyrics\n- subtle warm strings for truth reveal\n\nMUSIC ARC:\n\n0.00–2.00\nquiet valley mystery.\n\n2.00–3.80\ngentle eerie tension.\n\n3.80–5.40\nZara reveal motif.\n\n5.40–7.30\nbrief quirky skeleton-comedy rhythm.\n\n7.30–9.10\nNyx mystery motif returns.\n\n9.10–11.00\ntension.\n\n11.00–13.00\nmusic pulls back and resolves emotionally for truth reveal.\n\n13.00–15.00\nlow urgent pulse builds under fading moon.\n\nEnd unresolved.\n\nNo horror score.\n\nNo huge trailer percussion.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 2 camera progression:\n\n0–3.8 sec:\nslow exploratory movement.\n\n3.8–5.4 sec:\ndynamic but controlled Zara entrance.\n\n5.4–7.3 sec:\neye-level reaction comedy.\n\n7.3–11 sec:\nlarger scale tension with Nyx.\n\n11–13 sec:\nintimate truth reveal.\n\n13–15 sec:\nexpand back into cosmic/environmental stakes.\n\nUse:\n- slow tracking\n- low-angle reveal\n- rack focus\n- reaction close-ups\n- over-the-shoulder compositions\n- extreme scale framing with Nyx\n- atmospheric wide shots\n- slow final pull-back\n\nAvoid:\n- handheld shake\n- jump-scare camera\n- crash zoom\n- Dutch angles\n- 360-degree camera\n- excessive hard cuts\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n- delicate dark linework\n- watercolor/gouache painterly fills\n- cinematic blue moonlight\n- strong stylized silhouettes\n- expressive facial animation\n- atmospheric canyon backgrounds\n- intricate black-blue dragon scales\n- ivory skeleton guardians\n- subtle magical mist\n- warm skin rendering on Zara\n- rich eye highlights\n- cool silver shard glow\n\nThe animation should look like the supplied storyboard illustrations brought to cinematic life.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealism\n- live action\n- 3D CGI\n- anime\n- vector animation\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nPrimary PART 2 palette:\n\n- midnight navy\n- desaturated cobalt\n- charcoal blue\n- pale cyan\n- ivory\n- cool moon silver\n- warm brown skin\n- Zara blue\n- Nyx amber-orange eyes\n- silver-white moon shard\n\nPanels / first half:\ncold moonlit mystery.\n\nZara entrance:\nstrong blue rim light through curls.\n\nNyx reveal:\nuse contrast between:\nblack-blue scales,\namber eyes,\nsilver shard.\n\nTruth reveal:\nkeep lighting calm.\n\nFinal:\nmoonlight weakens noticeably.\n\nThe shard becomes relatively brighter.\n\nDo NOT darken scene so far that faces become unreadable.\n\n--------------------------------------------------\n\n[MOTION RULES — KAEL]\n\nUse believable teenage movement.\n\nValley:\ncareful steps,\nsmall head turns,\ntense shoulders.\n\nSkeleton reveal:\nquick startled movements but no exaggerated cartoon panic.\n\nAccusation:\none clear forward step and pointing gesture.\n\nTruth reveal:\nhand gradually lowers,\nbody relaxes slightly,\neyes shift between characters.\n\nNo fighting.\n\n--------------------------------------------------\n\n[MOTION RULES — ZARA]\n\nEntrance:\ncontrolled low jump from safe ledge.\n\nUse:\n- soft landing\n- hair follow-through\n- confident rise\n- minimal unnecessary movement\n\nConversation:\nvery composed.\n\nHer confidence contrasts with Kael.\n\nTruth reveal:\nalmost still.\n\nThat stillness gives her line authority.\n\n--------------------------------------------------\n\n[MOTION RULES — NYX]\n\nNyx should feel huge.\n\nUse:\n- slow serpentine head entry\n- delayed neck/body follow-through\n- small breathing movement\n- subtle feather-scale motion\n- massive eye blinks\n- gentle shard adjustment\n\nNo rapid attack movement.\n\n--------------------------------------------------\n\n[MOTION RULES — SKELETONS]\n\nMovement:\nslightly clacky but fluid.\n\nUse:\n- slow standing\n- head tilts\n- small shoulder movements\n- gentle waving\n- self-adjustment\n\nDo not make them move like aggressive zombies.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nMaintain exactly:\n\nONE Kael.\nONE Zara.\nONE Nyx.\nONE moon fragment.\nONE cracked moon.\n\nSame:\n- Kael clothes\n- Zara clothes\n- Zara hair\n- Nyx anatomy\n- skeleton style\n- valley architecture\n- moon damage\n- fragment shape\n- lighting direction\n\nZara appears for first time in Part 2.\n\nNyx appears after Zara.\n\nDo not show Nyx before his Shot 5 reveal except perhaps as an indistinct distant shadow if essential.\n\nOnce Nyx enters:\nhe remains with the group through ending.\n\nMoon fragment stays in his possession.\n\n--------------------------------------------------\n\n[EMOTIONAL PERFORMANCE PRIORITY]\n\nThe sequence has THREE emotional stages.\n\nSTAGE 1:\n\nKael thinks:\n\n“This place is dangerous.”\n\nThe skeletons visually support that assumption.\n\nSTAGE 2:\n\nZara and the waving skeleton reveal:\n\n“This place is strange, but not what Kael thought.”\n\nSTAGE 3:\n\nNyx appears and Kael again assumes:\n\n“There’s the villain.”\n\nZara reveals:\n\nKael is wrong AGAIN.\n\nThis reinforces the story theme:\n\nappearances can mislead.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard sheet.\nNo storyboard panels.\nNo captions.\nNo timestamps.\nNo dialogue bubbles.\nNo subtitles.\nNo production notes.\nNo title card.\nNo watermark.\nNo logos.\nNo brands.\n\nNo gore.\nNo blood.\nNo flesh on skeletons.\nNo corpses.\nNo zombies.\nNo horror.\nNo jump scares.\nNo skeleton attack.\nNo biting.\nNo fighting.\nNo weapons.\nNo injuries.\n\nNo angry Zara attack.\nNo Kael attacking Nyx.\nNo Nyx attacking Kael.\nNo dragon fire.\nNo dragon roar.\n\nNo duplicate skeleton characters caused by generation artifacts.\nNo duplicate Kael.\nNo duplicate Zara.\nNo duplicate Nyx.\nNo duplicate moon fragment.\n\nNo costume changes.\nNo hairstyle drift.\nNo Zara hair shrinking dramatically.\nNo Nyx becoming a different dragon type.\nNo moon restoration yet.\n\n...",
    "sourceUrl": "https://x.com/ManuAGI01/status/2086321924070834347",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T05:21:23.342Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086321924070834347.mp4",
    "posterUrl": "/posters/x/2086321924070834347.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "nature",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai\n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for “THE BOY WHO RETURNED THE MOON” as the PRIMARY visual, character, environment, costume, scale, prop, lighting, performance, and narrative reference.\n\nThis sequence continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n- Kael entering the forbidden Valley of Bones\n- Nyx disappearing deeper into the canyon\n- Nyx still carrying the ONLY glowing silver moon fragment\n- the giant moon remaining cracked and visibly dimmer\n- Kael believing Nyx stole the fragment\n- Kael deciding to follow him despite his fear\n\nGenerate a FINISHED cinematic animated sequence from this storyboard.\n\nDO NOT reproduce or animate the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- storyboard panels\n- panel numbers\n- timestamps\n- handwritten captions\n- speech bubbles\n- production notes\n- character reference drawings\n- environment reference images\n- storyboard title/header\n- off-white paper background\n- color swatches\n- footer graphics\n\nThe uploaded storyboard locks:\n\n1. Kael’s exact face\n2. Kael’s swept-back black hairstyle\n3. Kael’s dark tunic and proportions\n4. Zara’s exact face and enormous curly hairstyle\n5. Zara’s blue skeletal-pattern top\n6. Nyx’s exact midnight-dragon design\n7. Nyx’s enormous amber eyes\n8. Nyx’s luminous toothy grin\n9. Nyx’s black feather-like scales\n10. the ONE silver moon fragment\n11. the stylized ivory skeleton guardians\n12. the Valley of Bones geography\n13. blue-gray moonlit mist\n14. the cracked dimming moon\n15. cold midnight-blue color palette\n16. Part 2 story progression\n17. Zara’s central truth reveal\n18. final “we’re running out of time” cliffhanger\n\nTranslate the storyboard’s 15 panels into approximately **8 cinematic shots**.\n\nDO NOT create fifteen rapid one-second cuts.\n\nUse clear progression:\n\nvalley mystery\n→ skeleton reveal\n→ Zara entrance\n→ comedy beat\n→ Nyx appears\n→ Kael accuses him\n→ Zara reveals the truth\n→ moon dims\n→ urgent alliance cliffhanger.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory range:\n0:15–0:30 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium hand-painted 2D dark-fantasy animation\n\nGenre:\nfantasy mystery → adventure revelation\n\nDialogue:\nnative synchronized English dialogue\n\nAudio:\nnative dialogue + environmental sound + skeleton SFX + moon magic + original cinematic score\n\nEMOTIONAL ARC:\n\ncautious exploration\n→ eerie surprise\n→ overwhelmed\n→ mysterious hero entrance\n→ quirky comedy\n→ renewed tension\n→ confrontation\n→ major truth reveal\n→ cosmic urgency\n\nCORE PART 2 TWIST:\n\nKael believes Nyx stole the moon fragment.\n\nZara reveals:\n\n“He’s trying to return it.”\n\nThis line must land as the CENTRAL reversal of the entire sequence.\n\n--------------------------------------------------\n\n[PART 2 STORY LIMIT]\n\nPART 2 ONLY.\n\nSTART:\n\nKael enters deeper into the Valley of Bones alone.\n\nEND:\n\nKael, Zara, Nyx and the skeleton guardians look toward the distant mountain as the cracked moon dims again.\n\nZara says:\n\n<d>[English] And we’re running out of time.</d>\n\nCUT TO BLACK.\n\nDO NOT show:\n- detailed Moon Shrine explanation\n- Nyx trying to fly to shrine\n- magical wind pushing Nyx back\n- destroyed bridge\n- Bone Bridge construction\n- skeletons forming bridge\n- characters crossing canyon\n- fragment falling\n- Kael catching the fragment\n- shrine activation\n- moon restoration\n- skeleton-to-firefly transformation\n- final friendship ending\n\nThose belong to Parts 3 and 4.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — KAEL]\n\nMatch PART 1 and the uploaded PART 2 storyboard EXACTLY.\n\nKAEL:\n\nSlim teenage moon-watcher.\n\nAppearance:\n- youthful angular stylized face\n- pale-to-medium cool skin\n- large expressive dark eyes\n- thick dark eyebrows\n- swept-back messy black hair\n- recognizable pointed hair silhouette\n- lean narrow build\n- long arms and legs\n\nClothing:\n- simple charcoal-black tunic\n- dark trousers\n- dark practical boots\n- subtle belt\n- no armor\n- no cape\n- no weapon\n\nPersonality:\n- observant\n- courageous\n- impulsive\n- suspicious of Nyx\n- uncomfortable in the mysterious valley\n- willing to confront what he believes is wrong\n\nPART 2 PERFORMANCE ARC:\n\n0–3 sec:\ncareful, alert, nervous\n\n3–5 sec:\nstartled and overwhelmed\n\n5–8 sec:\nconfused by Zara and skeleton behavior\n\n8–11 sec:\nangry certainty returns when Nyx appears\n\n11–13 sec:\nshock as Zara contradicts him\n\n13–15 sec:\nbelief begins changing; urgency takes over\n\nIMPORTANT:\n\nKael is not hateful toward Nyx.\n\nHe genuinely believes he witnessed a theft.\n\nHis mistake must feel understandable.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — ZARA]\n\nMatch the uploaded storyboard EXACTLY.\n\nZARA — THE BONEKEEPER\n\nAppearance:\n- teenage girl\n- deep warm-brown skin\n- enormous dense black curly hair\n- huge rounded hair silhouette\n- subtle cool-blue moonlight rim on curls\n- large expressive brown eyes\n- strong dark eyebrows\n- blue short-sleeved distressed fantasy top\n- stylized pale rib/skeleton motif ON THE FABRIC ONLY\n- brown utility belt\n- dark/brown practical fantasy trousers\n- wrapped forearms\n- practical boots\n- compact athletic build\n\nIMPORTANT:\n\nThe rib design is clothing artwork.\n\nNo exposed anatomy.\n\nPersonality:\n- fearless\n- calm\n- clever\n- dry sense of humor\n- protective of the valley\n- emotionally grounded\n- already understands Nyx’s mission\n\nZARA PERFORMANCE:\n\nEntrance:\nconfident and controlled.\n\nComedy:\ndeadpan.\n\nTruth reveal:\ncompletely serious.\n\nEnding:\nurgent.\n\nShe should visually transform the skeleton-filled environment from “dangerous” into “strange but safe.”\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NYX]\n\nMatch PART 1 exactly.\n\nNYX — THE MIDNIGHT DRAGON\n\nAppearance:\n- large elongated black dragon\n- long flexible serpentine neck\n- black-blue feather-like layered scales\n- wild swept-back head spines\n- large pointed ears/horns\n- massive glowing amber-orange eyes\n- vertical slit pupils\n- dramatic luminous teeth\n- elongated expressive snout\n- long dark body\n\nThe same moon shard remains with him.\n\nIMPORTANT:\n\nNyx is NOT eating the fragment.\n\nHold it carefully.\n\nHis body language should subtly reveal protectiveness:\n- careful head position\n- controlled jaw grip\n- fragment protected close to him\n- concern when moon weakens\n\nWhen Kael accuses him:\n\nNyx may blink,\ntilt his head,\nnarrow one eye,\nor look mildly offended.\n\nDo NOT make him attack.\n\nDo NOT make him roar.\n\nDo NOT use fire.\n\n--------------------------------------------------\n\n[SKELETON GUARDIAN IDENTITY LOCK]\n\nMatch the storyboard.\n\nAncient magical skeleton guardians.\n\nAppearance:\n- smooth ivory / cream bones\n- stylized oversized skulls\n- large hollow expressive eye sockets\n- rounded family-friendly proportions\n- subtle pale-blue magical highlights if needed\n- small differences in height and personality\n\nNo:\n- flesh\n- gore\n- decay\n- blood\n- torn body remnants\n- horror texture\n\nThey are not zombies.\n\nThey are magical ancient guardians.\n\nPERFORMANCE ARC:\n\nfirst appearance:\nmysterious.\n\ngroup reveal:\nvisually intimidating only because there are many.\n\nafter Zara arrives:\ncurious and quirky.\n\nComedy examples:\n- one adjusts its own skull\n- one leans forward curiously\n- one scratches its head\n- one gives Kael a friendly wave\n- one looks embarrassed by the waving skeleton\n\nThey never attack Kael.\n\n--------------------------------------------------\n\n[MOON FRAGMENT IDENTITY LOCK]\n\nOnly ONE fragment exists.\n\nMatch Part 1.\n\nAppearance:\n- irregular silver-white crescent/crystal fragment\n- brilliant white core\n- translucent pale cyan edges\n- soft silver particle trail\n- cool moonlight aura\n\nIt remains with Nyx throughout Part 2.\n\nNo duplication.\n\nNo shape change.\n\nNo disappearance.\n\n--------------------------------------------------\n\n[MOON STATE CONTINUITY]\n\nThe moon begins Part 2:\n\n- cracked\n- missing one section\n- dimmer than at the beginning of Part 1\n\nDuring the ending:\n\na deep magical resonance occurs.\n\nMoon brightness decreases noticeably again.\n\nDo NOT:\n- break another giant piece away\n- destroy the moon\n- completely turn it black yet\n- restore it\n\nThe change should communicate:\n\nTIME IS RUNNING OUT.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — VALLEY OF BONES]\n\nUse the exact environment established by the uploaded storyboard.\n\nEnvironment:\n- huge black-blue canyon walls\n- tall jagged rock towers\n- cold moonlit path\n- pale blue ground mist\n- enormous fossil-like rib structures\n- ancient ruined stone features\n- sparse silver-gray vegetation\n- giant dark mountain in distance\n- scattered magical skeleton guardians\n- cold deep blue shadows\n- cracked moon overhead\n\nVisual feeling:\nancient,\nforbidden,\nquiet,\nmysterious,\nbeautiful.\n\nNOT horror.\n\nNo modern objects.\n\n--------------------------------------------------\n\n[WORLD / STORY LOGIC]\n\nClear cause-and-effect:\n\nKael enters valley\n→ hears rattling\n\nrattling\n→ skeleton guardians appear\n\nKael assumes danger\n→ Zara intervenes\n\nKael sees skeletons responding around Zara\n→ assumes she controls them\n\nZara says they act independently\n→ comedy breaks tension\n\nNyx arrives\n→ Kael’s suspicion immediately returns\n\nKael sees fragment\n→ accuses Nyx\n\nZara corrects him\n→ Kael’s understanding changes\n\nmoon dims again\n→ Zara reveals urgency\n\nNo unexplained magical action.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–2.00s\nTHE VALLEY IS TOO QUIET\n\nContinue directly from Part 1.\n\nWide rear shot.\n\nKael carefully walks deeper into the Valley of Bones.\n\nThe environment towers around him.\n\nHuge rib-like fossil structures curve from the ground.\n\nCold mist moves around his boots.\n\nThe cracked moon hangs high between canyon walls.\n\nFar behind him, Noctara’s warm castle glow is nearly gone.\n\nCamera slowly tracks backward in front of Kael.\n\nKael looks left.\n\nThen right.\n\nHis footsteps echo.\n\nHe whispers:\n\n<d>[English] Where did you go...?</d>\n\nNo answer.\n\nHe hears:\n\nCLACK...\n\nKael stops.\n\nCamera:\nwide valley\n→ tracking medium\n→ slow push toward Kael’s reaction.\n\nSound:\n- empty canyon wind\n- footsteps on stone\n- distant rock echoes\n- faint moon hum\n- one dry magical bone “clack”\n\nMusic:\nvery sparse low strings + distant celesta.\n\n--------------------------------------------------\n\n[SHOT 2] — 2.00s–3.80s\nONE SKELETON BECOMES AN AUDIENCE\n\nLow shot near Kael’s boots.\n\nA skeletal hand rises gently from behind a rock.\n\nKael freezes.\n\nAnother skull slowly peeks from behind a fossil arch.\n\nThen another guardian sits upright.\n\nKael turns.\n\nSeveral more skeletons rise from around the path.\n\nNo attack.\n\nThey simply stare.\n\nOne adjusts its skull so it sits straighter.\n\nKael slowly backs away.\n\nDialogue:\n\nKael:\n<d>[English] ...Okay.</d>\n\nHe looks behind him.\n\nMore skeletons.\n\nKael:\n<d>[English] That’s a lot of bones.</d>\n\nCamera:\nmacro skeletal hand\n→ Kael reaction\n→ widening circular composition revealing several guardians\n→ wider shot showing Kael surrounded.\n\nSound:\n- soft CLACK\n- gentle bone rattles\n- tiny skull adjustment\n- Kael’s nervous breath\n\nMusic adds light suspense pulses.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.80s–5.40s\nTHE BONEKEEPER\n\nKael continues backing away.\n\nSuddenly—\n\na figure drops safely from a low ruined ledge between Kael and the guardians.\n\nZARA lands in a crouch.\n\nSoft mist/dust puffs around her boots.\n\nShe rises.\n\nHer enormous curly hair catches a beautiful blue moonlight rim.\n\nAll the skeleton guardians stop moving.\n\nStrong cinematic reveal.\n\nKael stares at her.\n\nZara calmly looks at him.\n\nDialogue:\n\nZara:\n<d>[English] You’re loud.</d>\n\nKael:\n<d>[English] I’m surrounded by skeletons!</d>\n\nZara glances around casually.\n\nZara:\n<d>[English] They noticed.</d>\n\nCamera:\nKael surrounded\n→ quick downward motion reveal of Zara landing\n→ low-angle hero shot\n→ medium two-shot.\n\nMusic:\nbrief confident Zara motif.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.40s–7.30s\n“THEY CONTROL THEMSELVES.”\n\nKael points toward the skeleton guardians.\n\nDialogue:\n\nKael:\n<d>[English] You control them?!</d>\n\nZara raises one eyebrow.\n\nZara:\n<d>[English] No.</d>\n\nTiny pause.\n\nZara:\n<d>[English] They control themselves.</d>\n\nBehind her:\n\none skeleton gives Kael a cheerful little wave.\n\nKael slowly looks at it.\n\nThe skeleton waves again.\n\nAnother guardian looks at the waving skeleton with mild embarrassment.\n\nKael lowers his pointing hand slightly.\n\nDialogue:\n\nKael:\n<d>[English] ...Right.</d>\n\nCamera:\nmedium Kael/Zara\n→ rack focus past Zara to waving skeleton\n→ Kael reaction close-up.\n\nSound:\n- gentle finger-bone clack during wave\n- tiny comedic wooden percussion accent\n\nMusic briefly becomes lighter.\n\nThis is the major comedy relief beat.\n\n--------------------------------------------------\n\n[SHOT 5] — 7.30s–9.10s\nTHE DRAGON RETURNS\n\nA subtle silver light appears behind Zara.\n\nSkeleton guardians look toward it.\n\nZara does not panic.\n\nKael notices.\n\nHis expression changes immediately.\n\nCamera moves over Zara’s shoulder.\n\nFrom the darkness between enormous ruins emerges NYX.\n\nHis long black neck enters the moonlight.\n\nAmber eyes glow.\n\nThe silver moon fragment shines brilliantly between his jaws.\n\nKael’s relaxed confusion disappears.\n\nDialogue:\n\nKael:\n<d>[English] You!</d>\n\nNyx stops.\n\nHe looks at Kael.\n\nHe does not attack.\n\nThe glowing shard illuminates the bottom of his face.\n\nCamera:\nsilver light hint\n→ reveal Nyx silhouette\n→ push toward giant amber eye\n→ Kael reaction.\n\nMusic:\nmysterious low dragon motif returns.\n\n--------------------------------------------------\n\n[SHOT 6] — 9.10s–11.00s\n“HE STOLE THAT!”\n\nKael steps forward.\n\nHe points directly at Nyx.\n\nDialogue:\n\nKael:\n<d>[English] He stole that!</d>\n\nNyx’s enormous eye narrows.\n\nHe tilts his head slightly.\n\nExpression:\noffended confusion.\n\nZara turns slowly toward Kael.\n\nKael continues:\n\n<d>[English] I saw him take it from the moon!</d>\n\nNyx carefully adjusts the shard in his jaws without damaging it.\n\nZara looks from Kael to Nyx.\n\nCamera:\nKael foreground\n→ Nyx background\n→ Zara between them\n→ close Nyx reaction.\n\nNo combat stance.\n\nNo aggression.\n\n--------------------------------------------------\n\n[SHOT 7] — 11.00s–13.00s\nTHE STORY IS BACKWARDS\n\nCut closer to Zara.\n\nHer expression becomes completely serious.\n\nThe comedy disappears.\n\nDialogue:\n\nZara:\n<d>[English] No.</d>\n\nKael freezes.\n\nZara looks toward Nyx.\n\nNyx lowers his head slightly, protecting the shard.\n\nZara:\n\n<d>[English] He didn’t steal it.</d>\n\nKael:\n<d>[English] What?</d>\n\nZara points toward the glowing fragment.\n\nThen toward the distant mountain.\n\nZara:\n\n<d>[English] He’s trying to return it.</d>\n\nHOLD Kael’s reaction.\n\nHis pointing hand slowly lowers.\n\nHis eyes move:\n\nNyx\n→ shard\n→ Zara.\n\nHe realizes the story may be completely different from what he believed.\n\nCamera:\nZara serious close-up\n→ Nyx and shard\n→ Kael reaction close-up\n→ three-character composition.\n\nMusic:\nresolve previous suspicious motif into a surprising emotional chord.\n\nIMPORTANT:\n\nGive the line:\n\n“He’s trying to return it.”\n\nclear breathing space.\n\nIt is the MAIN Part 2 twist.\n\n--------------------------------------------------\n\n[SHOT 8] — 13.00s–15.00s\nTHE MOON LOSES MORE LIGHT\n\nBefore Kael can respond—\n\nA deep magical resonance passes through the valley.\n\nBOOOOM...\n\nNot an explosion.\n\nThe ground does NOT violently shake.\n\nEveryone looks upward.\n\nKael.\n\nZara.\n\nNyx.\n\nThe skeleton guardians.\n\nThe cracked moon flickers.\n\nIts silver glow decreases dramatically.\n\nThe moon fragment in Nyx’s jaws pulses brighter by contrast.\n\nNyx’s amber eyes widen with concern.\n\nZara immediately looks toward the distant mountain summit.\n\nFar through the canyon, barely visible:\n\na dark ancient shrine-like silhouette.\n\nKael sees it too.\n\nDialogue:\n\nKael, quieter:\n<d>[English] What’s happening?</d>\n\nZara:\n<d>[English] The moon is fading.</d>\n\nShe turns toward him.\n\nZara:\n<d>[English] And we’re running out of time.</d>\n\nFinal composition:\n\nKael = shocked but listening  \nZara = urgent leader  \nNyx = misunderstood guardian protecting shard  \nSkeletons = alert behind them  \nMoon = cracked and dimmer above  \nDistant mountain = their next destination\n\nCamera:\ncharacters looking up\n→ huge damaged moon\n→ push toward Zara\n→ pull back into wide group + mountain composition.\n\nCUT TO BLACK.\n\nEND PART 2.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[KAEL VOICE]\n\nNatural teenage boy voice.\n\nQualities:\n- youthful\n- intelligent\n- nervous early\n- emotionally expressive\n- clear English\n- sincere\n- not exaggerated\n\nPerformance arc:\n\n0–4 sec:\ncautious whisper.\n\n4–7 sec:\nnervous confusion.\n\n7–11 sec:\naccusatory certainty.\n\n11–13 sec:\nshocked disbelief.\n\n13–15 sec:\nconcern.\n\nImportant line:\n\n<d>[English] He stole that!</d>\n\nshould feel like Kael genuinely believes he is defending the moon.\n\nNot hatred.\n\n--------------------------------------------------\n\n[ZARA VOICE]\n\nNatural teenage girl voice.\n\nQualities:\n- confident\n- calm\n- dry humor\n- slightly low/grounded for her age\n- clear English\n- emotionally controlled\n\nPerformance:\n\nentrance:\ncool confidence.\n\nSkeleton comedy:\ndeadpan.\n\nTruth reveal:\nserious and deliberate.\n\nEnding:\nurgent.\n\nCritical delivery:\n\n<d>[English] He’s trying to return it.</d>\n\nSpeak clearly, calmly, with emphasis on “return.”\n\nThen final:\n\n<d>[English] And we’re running out of time.</d>\n\nshould be faster and more urgent.\n\n--------------------------------------------------\n\n[NYX AUDIO]\n\nNo spoken dialogue in PART 2.\n\nUse:\n- controlled deep breathing\n- soft throat resonance\n- scale/feather movement\n- small offended exhale when Kael accuses him\n- protective low hum around fragment\n\nNo:\n- roaring\n- evil growling\n- laughter\n- fire sounds\n\n--------------------------------------------------\n\n[SKELETON AUDIO]\n\nUse light stylized bone sounds:\n\n- soft clacks\n- gentle rattles\n- skull adjustment “tok”\n- finger-wave clicks\n- tiny curious head movement sounds\n\nNever make them sound gruesome.\n\nAvoid:\n- horror screams\n- zombie moans\n- shrieking\n- threatening rattling walls of sound\n\n--------------------------------------------------\n\n[VALLEY ENVIRONMENT AUDIO]\n\nMaintain:\n- cold canyon wind\n- stone echoes\n- subtle mist movement\n- occasional distant rock resonance\n- faint moon magic\n- Kael footsteps\n- sparse atmospheric ambience\n\nCompared with Noctara:\n\nmuch emptier,\ncolder,\nmore echoing.\n\n--------------------------------------------------\n\n[MOON DIMMING SOUND]\n\nAt final:\n\nuse a distant deep celestial resonance.\n\nStructure:\n\nsoft low pulse\n→ faint crystalline tremor\n→ moonlight hum weakens\n→ fragment briefly pulses\n\nNo explosion.\n\nNo thunder.\n\nNo destructive shockwave.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL fantasy mystery score.\n\nInstrumentation:\n- low strings\n- cello\n- bass clarinet\n- soft frame drum\n- celesta\n- glass harmonics\n- muted pizzicato\n- airy female/neutral choir texture without lyrics\n- subtle warm strings for truth reveal\n\nMUSIC ARC:\n\n0.00–2.00\nquiet valley mystery.\n\n2.00–3.80\ngentle eerie tension.\n\n3.80–5.40\nZara reveal motif.\n\n5.40–7.30\nbrief quirky skeleton-comedy rhythm.\n\n7.30–9.10\nNyx mystery motif returns.\n\n9.10–11.00\ntension.\n\n11.00–13.00\nmusic pulls back and resolves emotionally for truth reveal.\n\n13.00–15.00\nlow urgent pulse builds under fading moon.\n\nEnd unresolved.\n\nNo horror score.\n\nNo huge trailer percussion.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nPART 2 camera progression:\n\n0–3.8 sec:\nslow exploratory movement.\n\n3.8–5.4 sec:\ndynamic but controlled Zara entrance.\n\n5.4–7.3 sec:\neye-level reaction comedy.\n\n7.3–11 sec:\nlarger scale tension with Nyx.\n\n11–13 sec:\nintimate truth reveal.\n\n13–15 sec:\nexpand back into cosmic/environmental stakes.\n\nUse:\n- slow tracking\n- low-angle reveal\n- rack focus\n- reaction close-ups\n- over-the-shoulder compositions\n- extreme scale framing with Nyx\n- atmospheric wide shots\n- slow final pull-back\n\nAvoid:\n- handheld shake\n- jump-scare camera\n- crash zoom\n- Dutch angles\n- 360-degree camera\n- excessive hard cuts\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished hand-painted 2D dark-fantasy animation.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n- delicate dark linework\n- watercolor/gouache painterly fills\n- cinematic blue moonlight\n- strong stylized silhouettes\n- expressive facial animation\n- atmospheric canyon backgrounds\n- intricate black-blue dragon scales\n- ivory skeleton guardians\n- subtle magical mist\n- warm skin rendering on Zara\n- rich eye highlights\n- cool silver shard glow\n\nThe animation should look like the supplied storyboard illustrations brought to cinematic life.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealism\n- live action\n- 3D CGI\n- anime\n- vector animation\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nPrimary PART 2 palette:\n\n- midnight navy\n- desaturated cobalt\n- charcoal blue\n- pale cyan\n- ivory\n- cool moon silver\n- warm brown skin\n- Zara blue\n- Nyx amber-orange eyes\n- silver-white moon shard\n\nPanels / first half:\ncold moonlit mystery.\n\nZara entrance:\nstrong blue rim light through curls.\n\nNyx reveal:\nuse contrast between:\nblack-blue scales,\namber eyes,\nsilver shard.\n\nTruth reveal:\nkeep lighting calm.\n\nFinal:\nmoonlight weakens noticeably.\n\nThe shard becomes relatively brighter.\n\nDo NOT darken scene so far that faces become unreadable.\n\n--------------------------------------------------\n\n[MOTION RULES — KAEL]\n\nUse believable teenage movement.\n\nValley:\ncareful steps,\nsmall head turns,\ntense shoulders.\n\nSkeleton reveal:\nquick startled movements but no exaggerated cartoon panic.\n\nAccusation:\none clear forward step and pointing gesture.\n\nTruth reveal:\nhand gradually lowers,\nbody relaxes slightly,\neyes shift between characters.\n\nNo fighting.\n\n--------------------------------------------------\n\n[MOTION RULES — ZARA]\n\nEntrance:\ncontrolled low jump from safe ledge.\n\nUse:\n- soft landing\n- hair follow-through\n- confident rise\n- minimal unnecessary movement\n\nConversation:\nvery composed.\n\nHer confidence contrasts with Kael.\n\nTruth reveal:\nalmost still.\n\nThat stillness gives her line authority.\n\n--------------------------------------------------\n\n[MOTION RULES — NYX]\n\nNyx should feel huge.\n\nUse:\n- slow serpentine head entry\n- delayed neck/body follow-through\n- small breathing movement\n- subtle feather-scale motion\n- massive eye blinks\n- gentle shard adjustment\n\nNo rapid attack movement.\n\n--------------------------------------------------\n\n[MOTION RULES — SKELETONS]\n\nMovement:\nslightly clacky but fluid.\n\nUse:\n- slow standing\n- head tilts\n- small shoulder movements\n- gentle waving\n- self-adjustment\n\nDo not make them move like aggressive zombies.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nMaintain exactly:\n\nONE Kael.\nONE Zara.\nONE Nyx.\nONE moon fragment.\nONE cracked moon.\n\nSame:\n- Kael clothes\n- Zara clothes\n- Zara hair\n- Nyx anatomy\n- skeleton style\n- valley architecture\n- moon damage\n- fragment shape\n- lighting direction\n\nZara appears for first time in Part 2.\n\nNyx appears after Zara.\n\nDo not show Nyx before his Shot 5 reveal except perhaps as an indistinct distant shadow if essential.\n\nOnce Nyx enters:\nhe remains with the group through ending.\n\nMoon fragment stays in his possession.\n\n--------------------------------------------------\n\n[EMOTIONAL PERFORMANCE PRIORITY]\n\nThe sequence has THREE emotional stages.\n\nSTAGE 1:\n\nKael thinks:\n\n“This place is dangerous.”\n\nThe skeletons visually support that assumption.\n\nSTAGE 2:\n\nZara and the waving skeleton reveal:\n\n“This place is strange, but not what Kael thought.”\n\nSTAGE 3:\n\nNyx appears and Kael again assumes:\n\n“There’s the villain.”\n\nZara reveals:\n\nKael is wrong AGAIN.\n\nThis reinforces the story theme:\n\nappearances can mislead.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard sheet.\nNo storyboard panels.\nNo captions.\nNo timestamps.\nNo dialogue bubbles.\nNo subtitles.\nNo production notes.\nNo title card.\nNo watermark.\nNo logos.\nNo brands.\n\nNo gore.\nNo blood.\nNo flesh on skeletons.\nNo corpses.\nNo zombies.\nNo horror.\nNo jump scares.\nNo skeleton attack.\nNo biting.\nNo fighting.\nNo weapons.\nNo injuries.\n\nNo angry Zara attack.\nNo Kael attacking Nyx.\nNo Nyx attacking Kael.\nNo dragon fire.\nNo dragon roar.\n\nNo duplicate skeleton characters caused by generation artifacts.\nNo duplicate Kael.\nNo duplicate Zara.\nNo duplicate Nyx.\nNo duplicate moon fragment.\n\nNo costume changes.\nNo hairstyle drift.\nNo Zara hair shrinking dramatically.\nNo Nyx becoming a different dragon type.\nNo moon restoration yet.\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2086321924070834347"
  },
  {
    "id": "x-2086294491179495506",
    "title": "小小创作者与 3200 颗纸星星",
    "titleEn": "The Little Creator and 3,200 Paper Stars",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《小小创作者与 3200 颗纸星星》的 H3 成片。本页保留原始 X 来源链接。来源未公开完整 Prompt，本库不反推或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Little Creator and 3,200 Paper Stars. This page retains the original X source link. The source did not publish a complete Prompt, so this library does not infer or complete one. Hosted-video output: 15s · 1280×720.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2086294491179495506",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-09T03:32:22.831Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086294491179495506.mp4",
    "posterUrl": "/posters/x/2086294491179495506.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt :  \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for “The Little Creator and the 3,200 Paper Stars” as the PRIMARY visual, character, environment, scale, lighting, prop, performance, composition, and narrative reference.\n\nThis sequence continues IMMEDIATELY from PART 1.\n\nPART 1 ended with:\n- Luma still peacefully asleep above the rooftop\n- EXACTLY ONE glowing handmade paper star attached to her lantern surface\n- Milo beside Luma with his satchel still overflowing with glowing paper stars\n- Pip realizing Milo wants to use every remaining star\n- Milo whispering:\n  <d>[English] Maybe... all of them?</d>\n- Pip reacting in comic panic\n\nGenerate a FINISHED cinematic animated sequence from this storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten storyboard captions\n- production notes\n- reference-sheet drawings\n- scale diagrams\n- storyboard paper\n- footer graphics\n- storyboard header\n\nThe uploaded storyboard locks:\n\n1. Milo’s exact face and fluffy brown hairstyle\n2. Milo’s oversized dark navy-blue hoodie\n3. Milo’s brown star satchel\n4. Pip’s exact tiny yellow firefly design\n5. Luma’s exact giant cream-gold lantern design\n6. Luma’s gentle face and warm internal light\n7. Luma’s ribbons and tassels\n8. extreme tiny Milo / giant Luma scale\n9. handmade glowing folded-paper stars\n10. same nighttime rooftop\n11. same distant city skyline\n12. same indigo-blue + amber-gold palette\n13. progressive star decoration\n14. Luma’s gentle awakening\n15. Milo’s nervous “Surprise?”\n16. Luma understanding the gesture\n17. Luma carrying Milo safely higher\n18. paper stars lifting away\n19. exact glowing “3200” sky formation\n20. exact glowing “THANK YOU” sky formation\n21. final Milo + Pip + Luma celebration shot\n\nTranslate the storyboard’s 15 one-second panels into approximately 9 coherent cinematic shots.\n\nDO NOT create fifteen frantic one-second cuts.\n\nUse smooth visual progression:\n\ndecision\n→ decorating\n→ bigger decorating\n→ full reveal\n→ awakening\n→ nervous reaction\n→ understanding\n→ magical lift\n→ stars rise\n→ 3200\n→ THANK YOU\n→ emotional final image\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory time:\n0:15–0:30 of the full story\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D animated short film\n\nGenre:\nmagical milestone celebration / heartwarming gratitude fantasy\n\nDialogue:\nnatural synchronized English dialogue\n\nAudio:\nnative synchronized dialogue + rooftop ambience + magical paper-star effects + original cinematic music\n\nEMOTIONAL ARC:\n\nmischievous ambition\n→ playful decorating frenzy\n→ visual wonder\n→ “oh no, she woke up”\n→ nervous silence\n→ gentle acceptance\n→ magical surprise\n→ milestone reveal\n→ gratitude\n→ emotional celebration\n\nThe emotional meaning must be:\n\nMilo wanted to thank his 3,200 supporters.\n\nBut when the stars rise into the sky, he realizes that those 3,200 little lights are what made his own journey brighter.\n\nThe milestone belongs to the community.\n\n--------------------------------------------------\n\n[PART 2 STORY LIMIT]\n\nThis is the FINAL PART.\n\nSTART:\n\nMilo has decided to use all the paper stars.\n\nEND:\n\nMilo sits happily and safely on top of Luma with Pip beside him while glowing stars fill the sky above the city.\n\nThe milestone has been celebrated.\n\nNo cliffhanger.\n\nThe ending must feel complete, grateful, magical, and emotionally satisfying.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — MILO]\n\nMatch PART 1 and the uploaded storyboard EXACTLY.\n\nMILO:\n\nTiny young creator.\n\nAppearance:\n- warm light skin\n- large expressive dark-brown eyes\n- fluffy tousled brown hair\n- soft rounded youthful face\n- small appealing proportions\n\nClothing:\n- oversized dark navy-blue hoodie\n- dark trousers\n- little shoes\n- brown crossbody satchel\n\nPersonality:\n- grateful\n- creative\n- playful\n- ambitious\n- thoughtful\n- emotionally sincere\n\nPART 2 PERFORMANCE ARC:\n\n0–4 sec:\nexcited decorating energy\n\n4–6 sec:\nproud of his huge idea\n\n6–8 sec:\ncaught / nervous\n\n8–10 sec:\nsurprised by Luma’s kindness\n\n10–13 sec:\nastonished and emotional\n\n13–15 sec:\ndeep gratitude and joyful calm\n\nMilo must never appear arrogant.\n\nThe final emotion should be:\n\n“I can’t believe 3,200 people are here.”\n\nCRITICAL:\n\nMaintain exact:\n- face\n- brown hair\n- hoodie\n- satchel\n- proportions\n- age appearance\n- tiny scale\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — LUMA]\n\nMatch PART 1 and storyboard exactly.\n\nLUMA:\n\nGigantic living floating sky lantern.\n\nAppearance:\n- enormous round cream/parchment lantern body\n- warm amber-gold internal glow\n- subtle handmade paper texture\n- decorative swirls\n- large gentle expressive eyes\n- tiny peaceful mouth\n- subtle warm blush when awake\n- lower structural rim\n- long decorative ribbons/tassels\n\nScale:\ncolossal compared with Milo.\n\nPersonality:\n- gentle\n- warm\n- patient\n- sleepy\n- affectionate\n- quietly magical\n\nPERFORMANCE ARC:\n\n0–5 sec:\nsleeping\n\n5–6 sec:\neyelid twitch / awakening\n\n6–8 sec:\nsleepy curiosity\n\n8–9 sec:\nunderstands Milo’s intention\n\n9–15 sec:\nwarm, magical, supportive\n\nLuma must NEVER look:\n- angry\n- annoyed\n- frightening\n- threatening\n\nWhen she wakes, her reaction is:\n\n“What is this?”\n\nfollowed quickly by:\n\n“Oh... this is for your people.”\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — PIP]\n\nSame Pip from Part 1.\n\nAppearance:\n- tiny round yellow-gold firefly\n- big expressive eyes\n- tiny antennae\n- translucent wings\n- soft golden glow\n\nPersonality:\n- cautious\n- practical\n- loyal\n- comic\n- secretly proud of Milo\n\nPERFORMANCE:\n\npanic at beginning\n→ reluctant helper\n→ exhausted\n→ frozen when Luma wakes\n→ amazed by star magic\n→ emotionally happy\n→ proud final companion\n\n--------------------------------------------------\n\n[PAPER STAR IDENTITY LOCK]\n\nEvery paper star must maintain the same handcrafted design established in Part 1.\n\nClose-up appearance:\n- folded paper construction\n- visible triangular facets\n- cream / pale gold paper\n- warm golden inner illumination\n- delicate handmade imperfections\n- lightweight\n\nDo NOT turn them into:\n- generic five-point neon graphics\n- fireworks\n- laser particles\n- metallic stars\n- digital icons\n\nWhen thousands are visible at once, individual stars can become smaller glowing points at distance, but close and medium shots must preserve recognizable folded-paper-star forms.\n\n--------------------------------------------------\n\n[STAR BUILD CONTINUITY]\n\nStart:\n\nONE paper star already attached.\n\nThen progressively build:\n\n1 star\n→ 5–10 stars\n→ several clusters\n→ winding trails\n→ hundreds of visible glowing stars\n→ visually implied thousands covering much of Luma\n\nDo NOT instantly jump from one star to fully decorated Luma.\n\nThe first 4–5 seconds should clearly show Milo physically creating the decoration.\n\n--------------------------------------------------\n\n[EXACT “3200” FORMATION LOCK]\n\nThis is the most important visual information in the entire sequence.\n\nAt approximately 12 seconds into PART 2 / story time 0:27:\n\nHundreds/thousands of warm glowing paper-star lights must organize into the exact digits:\n\n3200\n\nRequirements:\n\n- exactly four digits\n- sequence must read LEFT TO RIGHT:\n  3 – 2 – 0 – 0\n- centered across the open night sky\n- huge and clearly legible\n- no comma\n- no decimal\n- no extra digit\n- do NOT render “3.2K”\n- do NOT render “320”\n- do NOT render “32000”\n- do NOT reverse digits\n- do NOT create malformed numerals\n\nThe number is made from many individual golden star points.\n\nIt should resemble an elegant constellation, not a font floating as digital text.\n\nGive the number sufficient screen time to read.\n\n--------------------------------------------------\n\n[EXACT “THANK YOU” FORMATION LOCK]\n\nAfter the “3200” formation:\n\nstars gently reorganize into:\n\nTHANK YOU\n\nRequirements:\n\n- exactly these two English words\n- uppercase\n- readable\n- centered\n- formed from golden paper-star lights\n- visually elegant\n- no extra message\n- no spelling errors\n\nDo NOT replace with:\n- THANKS\n- THANK U\n- THANKYOU\n- THANK YOU 3200\n- any other wording\n\nThe stars themselves are the letters.\n\nNo subtitle overlay.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — NIGHT ROOFTOP / CITY]\n\nMaintain exact continuity from Part 1.\n\nEnvironment:\n- same high rooftop\n- same nearby roof structures\n- same potted plants\n- same warm windows\n- same distant city\n- deep indigo night sky\n- same crescent moon if visible\n- subtle clouds\n- rooftop edge below\n- warm city-window lights\n\nAs Luma rises:\ncamera reveals more city depth.\n\nThe rooftop remains directly beneath them.\n\nNo transition to space.\n\nNo completely different location.\n\n--------------------------------------------------\n\n[SAFETY / SUPPORT RULE]\n\nWhen Luma rises higher:\n\nMilo must ALWAYS be safely supported.\n\nUse:\n- one thick ribbon loop functioning like a secure seat\nOR\n- Milo sitting securely on Luma’s structural rim\n\nNo dangling dangerously.\n\nNo falling.\n\nNo uncontrolled aerial motion.\n\nPip flies naturally beside him.\n\nLuma moves slowly and gently.\n\n--------------------------------------------------\n\n[MAGIC LOGIC]\n\nMilo attaches stars manually.\n\nLuma wakes and understands the gesture.\n\nLuma’s warm light strengthens.\n\nHer magic activates the stars.\n\nStars glow brighter\n→ loosen gently from lantern\n→ float upward\n→ orbit together\n→ form milestone number\n→ rearrange into gratitude message\n→ return to beautiful star field.\n\nNo explosion.\n\nNo magical shockwave.\n\nNo fireworks.\n\nMagic should feel:\ngraceful,\nwarm,\ncommunity-driven,\nbeautiful.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.80s\n“ALL OF THEM”\n\nContinue DIRECTLY from Part 1.\n\nMilo is beside Luma’s enormous paper surface.\n\nONE glowing paper star is already attached.\n\nHis satchel is open and overflowing with more stars.\n\nPip hovers beside him with huge worried eyes.\n\nMilo grins.\n\nDialogue:\n\nPip:\n<d>[English] You were serious?</d>\n\nMilo:\n<d>[English] Every single one.</d>\n\nPip:\n<d>[English] Oh boy.</d>\n\nMilo reaches into the satchel with both hands.\n\nHe pulls out two glowing stars.\n\nHe quickly but carefully attaches one.\n\nThen another.\n\nCamera:\nclose on single Part-1 star\n→ satchel overflowing\n→ Milo’s excited face\n→ close handmade-star placements.\n\nSound:\n- paper rustle\n- tiny attachment clicks\n- soft golden “ting” on each star\n- Pip wings\n- gentle lantern hum\n\nMusic:\nplayful pizzicato + marimba begins immediately.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.80s–3.70s\nTHE TINY THANK-YOU GETS BIGGER\n\nCreate a fluid miniature decorating montage.\n\nDo NOT use frantic hard cuts.\n\nShow Milo climbing along Luma’s hanging ribbons and lower structural rim.\n\nHe attaches glowing stars in curved patterns across her enormous lantern body.\n\nPip reluctantly assists by carrying one small star at a time.\n\nProgression should be visible:\n\nfew stars\n→ cluster\n→ glowing trail\n→ several clusters.\n\nMilo:\n<d>[English] Higher, Pip!</d>\n\nPip, struggling:\n<d>[English] I am a firefly, not a delivery service!</d>\n\nMilo laughs.\n\nLuma remains asleep.\n\nHer internal light slowly becomes slightly warmer as more stars are attached.\n\nCamera:\nsmooth lateral tracking across Luma\n→ close details\n→ wider scale shot showing growing patterns.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.70s–5.20s\nA LOT BRIGHTER\n\nContinue the montage as Milo finishes a sweeping line of stars.\n\nThe star coverage has become spectacular.\n\nUse a beautiful pull-back reveal.\n\nLuma now floats above the city with hundreds of glowing paper-star points wrapping gracefully around her giant lantern body.\n\nDo not completely obscure her cream paper surface.\n\nShe should still clearly look like Luma.\n\nMilo sits briefly on a ribbon knot catching his breath.\n\nPip hovers beside him completely exhausted.\n\nMilo looks up proudly.\n\nMilo:\n<d>[English] Okay...</d>\n\nTiny beat.\n\n<d>[English] Now that looks like a celebration.</d>\n\nPip:\n<d>[English] That looks visible from another country.</d>\n\nMilo giggles.\n\nCamera:\nmedium on Milo\n→ dramatic but gentle pull-back\n→ huge wide lantern-over-city reveal.\n\nSound:\n- hundreds of subtle star shimmers\n- nighttime city\n- ribbons\n- Pip exhausted buzz\n\nMusic:\nwarm celebratory flourish.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.20s–6.70s\nLUMA STARTS TO WAKE\n\nThe music suddenly becomes sparse.\n\nClose-up on Luma’s peaceful face.\n\nHer cheek glows.\n\nHer eyelid makes a tiny twitch.\n\nMilo freezes.\n\nPip freezes.\n\nAnother eyelid twitch.\n\nLuma’s giant eyes slowly open.\n\nHer gaze settles on tiny Milo.\n\nMilo is holding the FINAL remaining star in his hand.\n\nPip slowly hides behind Milo’s shoulder.\n\nMilo’s face changes:\n\nproud\n→ realization\n→ nervous.\n\nDialogue:\n\nPip, whispering:\n<d>[English] You did ask first... right?</d>\n\nMilo:\n<d>[English] ...No.</d>\n\nCamera:\nLuma eye close-up\n→ tiny Milo reflected in giant eye\n→ extreme scale shot.\n\nSound:\n- star shimmer reduces\n- soft paper movement\n- Luma awakening hum\n- tiny comedic silence\n\n--------------------------------------------------\n\n[SHOT 5] — 6.70s–8.20s\n“SURPRISE?”\n\nLuma looks at the hundreds/thousands of glowing decorations covering her.\n\nThen she looks back at Milo.\n\nMilo slowly raises the final paper star like an awkward peace offering.\n\nTiny nervous smile.\n\nMilo:\n<d>[English] ...Surprise?</d>\n\nHold a short silence.\n\nLuma looks at one glowing star.\n\nThen several.\n\nHer expression softens.\n\nHer eyes become warm.\n\nA faint rosy glow appears on her cheeks.\n\nLuma, gentle sleepy voice:\n<d>[English] Are these... thank-yous?</d>\n\nMilo nods.\n\nMilo:\n<d>[English] For everyone who stayed.</d>\n\nLuma smiles.\n\nLuma:\n<d>[English] Then they should see them.</d>\n\nCamera:\nthree-character scale composition\n→ close Milo\n→ close Luma’s eyes\n→ star detail\n→ Luma warm smile.\n\nMusic:\nemotional theme begins beneath the final line.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.20s–9.80s\nLUMA TAKES THE CELEBRATION HIGHER\n\nLuma’s internal amber light grows brighter.\n\nNot blinding.\n\nHer long ribbons gently move.\n\nOne ribbon curls beneath Milo and forms a secure supportive loop/seat.\n\nMilo settles safely into it.\n\nPip hovers beside him.\n\nLuma rises slowly above the rooftop.\n\nThe city expands beneath them.\n\nMilo grips the ribbon gently and stares outward.\n\nDialogue:\n\nMilo:\n<d>[English] Whoa...</d>\n\nPip:\n<d>[English] We are very high.</d>\n\nLuma, smiling:\n<d>[English] Just high enough.</d>\n\nCamera:\nmedium Milo supported in ribbon\n→ slow crane/pull-back\n→ city reveal below.\n\nSound:\n- ribbon movement\n- soft wind\n- rising lantern hum\n- distant city ambience\n\nMusic:\ngentle magical lift.\n\n--------------------------------------------------\n\n[SHOT 7] — 9.80s–11.40s\nTHE STARS BEGIN TO LEAVE\n\nClose-up.\n\nOne folded paper star on Luma’s surface trembles gently.\n\nIts edge lifts.\n\nThe star separates.\n\nIt floats upward.\n\nMilo notices.\n\nThen another star lifts.\n\nThen five.\n\nThen dozens.\n\nMilo reaches toward them.\n\nDialogue:\n\nMilo:\n<d>[English] Wait! The stars!</d>\n\nPip:\n<d>[English] They’re flying away!</d>\n\nLuma remains calm.\n\nLuma:\n<d>[English] Watch.</d>\n\nMore stars gently detach.\n\nThey do NOT explode.\n\nThey spiral gracefully upward in beautiful curved paths.\n\nCamera follows ONE star first.\n\nThen pull wider as dozens become hundreds.\n\nStars leave glowing trails against the deep indigo sky.\n\nSound:\n- soft folded-paper flutter\n- shimmering rise tones\n- wind\n- magical constellation ambience\n\n--------------------------------------------------\n\n[SHOT 8] — 11.40s–13.20s\n3,200 LITTLE LIGHTS\n\nThe stars rise above the city.\n\nHundreds become thousands of tiny warm points.\n\nThey swirl in elegant coordinated arcs.\n\nAt first:\nabstract spiral.\n\nThen:\none curved line settles.\n\nAnother.\n\nShapes become recognizable.\n\nCamera pulls VERY wide.\n\nThe glowing stars clearly form:\n\n3200\n\nacross the night sky.\n\nHOLD THE EXACT NUMBER.\n\nMake it large and unmistakably readable.\n\nMilo stares upward.\n\nHis eyes reflect “3200.”\n\nHis mouth opens slightly.\n\nHe becomes emotionally still.\n\nPip’s wings slow as he stares.\n\nLuma watches Milo rather than the number.\n\nMilo, barely above a whisper:\n<d>[English] Three thousand two hundred...</d>\n\nPip:\n<d>[English] That’s a lot of stars.</d>\n\nMilo:\n<d>[English] That’s a lot of people.</d>\n\nCamera:\nstar swarm\n→ formation transition\n→ enormous wide “3200”\n→ emotional Milo close-up with golden digits reflected in his eyes.\n\nMusic:\nreach the main emotional theme here.\n\nDo NOT rush this reveal.\n\n--------------------------------------------------\n\n[SHOT 9] — 13.20s–15.00s\nTHANK YOU — FINAL CELEBRATION\n\nThe “3200” constellation gently loosens.\n\nDo NOT dissolve randomly.\n\nStar trails elegantly sweep into new positions.\n\nThe lights reform into:\n\nTHANK YOU\n\nexactly.\n\nMilo places one hand over his heart.\n\nHis eyes become slightly glossy with emotion but he does NOT dramatically cry.\n\nMilo:\n<d>[English] Thank you... all of you.</d>\n\nLuma gently lifts Milo from the ribbon and allows him to sit safely on her upper structural rim / top surface.\n\nPip settles beside him.\n\nThe “THANK YOU” formation gradually relaxes back into hundreds of beautiful floating stars.\n\nA subtle heart-shaped star cluster may appear briefly.\n\nThe exact “3200” can remain faintly suggested as a distant constellation off to one side, but do not compete with the final characters.\n\nFINAL HERO COMPOSITION:\n\n- giant glowing Luma floating above the city\n- Milo sitting safely on top\n- Pip beside him\n- hundreds of warm paper stars floating around\n- distant rooftops and glowing windows below\n- deep indigo sky\n- crescent moon\n- soft golden reflections\n- tiny heart/star formations\n- peaceful magical atmosphere\n\nLuma softly asks:\n\n<d>[English] Ready for the next one?</d>\n\nMilo smiles.\n\n<d>[English] Always.</d>\n\nPip:\n<d>[English] After a nap.</d>\n\nMilo laughs softly.\n\nLuma smiles.\n\nHold the warm composition.\n\nFADE TO BLACK.\n\nEND.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[MILO VOICE]\n\nNatural youthful male/childlike creator voice.\n\nQualities:\n- warm\n- sincere\n- energetic\n- imaginative\n- emotionally expressive\n- clear English\n- not boastful\n\nPerformance:\n\n0–5 sec:\nexcited creative energy\n\n5–8 sec:\nawkward nervousness\n\n8–10 sec:\nwonder\n\n10–13 sec:\nsurprise turning emotional\n\n13–15 sec:\nsoft gratitude and confidence\n\nMost important emotional line:\n\n<d>[English] Thank you... all of you.</d>\n\nDeliver quietly and sincerely.\n\nNo shouting.\n\n--------------------------------------------------\n\n[LUMA VOICE]\n\nWarm feminine or gentle neutral giant-lantern voice.\n\nQualities:\n- soft\n- calm\n- slightly sleepy\n- warm\n- reassuring\n- subtly magical\n- never booming despite her giant size\n\nImportant lines:\n\n<d>[English] Are these... thank-yous?</d>\n\n<d>[English] Then they should see them.</d>\n\n<d>[English] Watch.</d>\n\n<d>[English] Ready for the next one?</d>\n\nKeep delivery gentle and affectionate.\n\n--------------------------------------------------\n\n[PIP VOICE]\n\nSmall cute firefly voice.\n\nQualities:\n- slightly high pitch\n- fast\n- expressive\n- dry comic timing\n- loyal\n- clear English\n\nComedy lines should be short so they do not overpower the milestone emotion.\n\nFinal:\n\n<d>[English] After a nap.</d>\n\nshould gently release the emotional tension with one small joke.\n\n--------------------------------------------------\n\n[ROOFTOP / CITY AUDIO]\n\nMaintain:\n- soft nighttime breeze\n- distant city atmosphere\n- faint rooftop ambience\n- gentle fabric ribbons\n- Pip wing buzz\n- Luma’s soft lantern resonance\n- paper-star rustles\n- distant quiet night life\n\nAs Luma rises:\nintroduce slightly stronger but still gentle wind.\n\nNo loud city traffic.\n\n--------------------------------------------------\n\n[PAPER STAR SOUND DESIGN]\n\nPhysical handling:\nsoft folded-paper rustle.\n\nAttachment:\ntiny “tap” / clip + magical warm “ting.”\n\nMass glowing:\nlayered delicate glockenspiel shimmer.\n\nDetachment:\npaper flutter + ascending bell tone.\n\nStar orbit:\nairy glass-like sparkle.\n\nNumber formation:\none soft harmonic resolution.\n\nTHANK YOU formation:\nwarmer emotional chime.\n\nAvoid:\n- laser sounds\n- sci-fi UI beeps\n- explosions\n- firework bangs\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL cinematic storybook score.\n\nInstrumentation:\n- soft piano\n- pizzicato strings\n- glockenspiel\n- light marimba\n- delicate harp\n- airy flute\n- warm sustained strings\n- subtle celesta\n- gentle acoustic plucks\n\nMUSIC ARC:\n\n0.00–3.70\nmischievous decorating rhythm.\n\n3.70–5.20\nbeautiful celebratory reveal.\n\n5.20–6.70\nmusic strips down for awakening.\n\n6.70–8.20\nsmall nervous-comedy motif resolving into warmth.\n\n8.20–9.80\nsoft magical rise.\n\n9.80–11.40\nwonder begins.\n\n11.40–13.20\nfull emotional milestone theme during “3200.”\n\n13.20–15.00\nheartwarming resolution during “THANK YOU” and final hero image.\n\nEnd on a warm resolved chord.\n\nNo epic trailer drums.\n\nNo EDM.\n\nNo sad melodrama.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nCAMERA ARC:\n\n0–4 sec:\nminiature creative coverage.\n\n4–5.2 sec:\nlarge-scale lantern reveal.\n\n5.2–8.2 sec:\nintimate waking / reaction coverage.\n\n8.2–9.8 sec:\ngraceful elevation.\n\n9.8–11.4 sec:\nfollow stars upward.\n\n11.4–13.2 sec:\nbiggest wide sky composition.\n\n13.2–15 sec:\nemotional close-up → final wide hero image.\n\nUse:\n- macro paper-star inserts\n- gentle tracking\n- slow dolly\n- scale-revealing pull-backs\n- upward tilt\n- controlled parallax\n- reflection in Milo’s eyes\n- very wide city skyline\n- final slow pull-back\n\nAvoid:\n- handheld shake\n- frantic cuts\n- crash zooms\n- Dutch angles\n- 360-degree spins\n- action-movie camera\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animation.\n\nMatch PART 1 exactly.\n\nUse:\n- delicate hand-drawn ink contours\n- watercolor/gouache-inspired fills\n- painterly indigo night\n- warm textured paper lantern\n- expressive rounded characters\n- subtle atmospheric depth\n- soft light bloom\n- detailed handmade paper stars\n- natural ribbon movement\n- cinematic light falloff\n- restrained squash and stretch\n\nThe final video must look like the uploaded storyboard illustrations came to life.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- 3D CGI\n- photorealism\n- anime\n- vector art\n- cel-shaded 3D\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\n0–4 sec:\ndeep indigo + small golden star points.\n\n4–5.2 sec:\nLuma becomes heavily decorated and warmer.\n\n5.2–8.2 sec:\nintimate cream / amber light around faces.\n\n8.2–11.4 sec:\nincrease sky depth and luminous star trails.\n\n11.4–13.2 sec:\npeak contrast:\n\nvery deep blue night\n+\nbrilliant warm golden “3200.”\n\n13.2–15 sec:\nsoft gold “THANK YOU”\n+\namber Luma\n+\ncool moonlit city\n+\ngentle gold star field.\n\nMilo’s face should receive warm star reflections.\n\nDo NOT overexpose the star formations.\n\nDigits and letters must retain crisp recognizable negative space.\n\n--------------------------------------------------\n\n[MOTION RULES — MILO]\n\nUse:\n- quick but believable star placement\n- careful ribbon movement\n- expressive head turns\n- hoodie secondary movement\n- natural tiny hand gestures\n- safe seating\n- emotional stillness during “3200”\n\nImportant:\n\nAt the number reveal, STOP unnecessary body motion.\n\nLet Milo simply stare.\n\nThis contrast makes the emotional beat stronger.\n\n--------------------------------------------------\n\n[MOTION RULES — LUMA]\n\nUse:\n- slow floating inertia\n- gentle vertical bob\n- paper-body softness\n- ribbon drag and lag\n- gradual eye opening\n- subtle cheek warmth\n- controlled upward movement\n\nNo sudden acceleration.\n\nNo jerking.\n\nNo rocket behavior.\n\n--------------------------------------------------\n\n[MOTION RULES — PIP]\n\nUse:\n- quick tiny hovering\n- natural wing blur\n- small body inertia\n- comic exhaustion\n- hiding behind Milo\n- awe-struck stillness during “3200”\n- gentle final landing beside Milo\n\n--------------------------------------------------\n\n[STAR MOTION RULES]\n\nDuring decorating:\nphysical objects stay attached.\n\nDuring activation:\n\none star lifts\n→ another\n→ small group\n→ dozens\n→ hundreds.\n\nNo instantaneous mass explosion.\n\nTrajectories:\n- arcs\n- spirals\n- gentle orbit paths\n- smooth coordinated curves\n\nWhen forming “3200”:\nslow down significantly.\n\nStars settle into fixed readable positions.\n\nWhen forming “THANK YOU”:\nstars smoothly travel from digit positions into letter positions.\n\nDo NOT morph digits themselves like digital typography.\n\nIndividual stars physically move to new locations.\n\n--------------------------------------------------\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2086053955315064990",
    "title": "雨中之刃剪纸片头",
    "titleEn": "Blade in the Rain Papercut Opening",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《雨中之刃剪纸片头》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1260×720。",
    "summaryEn": "An H3 output for Blade in the Rain Papercut Opening. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1260×720.",
    "prompt": "提示词灵感来源于 douyin 的【小彻的AI乱造】（国内的朋友可以关注下，真的把minimax 玩出花了）\n\n游戏宣传 PV，整体为纯二维中式剪纸版画动画与黑色电影平面合成。【美术风格】红黑视觉风格：深墨黑、朱砂红、暗猩红、骨白和少量旧金色构成高反差画面，人物使用硬边剪影、木刻纹理与套印错位效果，阴影只有一至两层。城市被高度平面化，只保留中式牌楼、红柱、木窗格、灯笼、屋檐、电线、潮湿石板路和高楼剪影，以粗糙木刻线条、剪纸层次、飞白墨痕与大面积负空间构成。融合旧式通缉海报、犯罪漫画封面、皮影戏、印章、雨线、纸屑和水墨笔触。画面像一张张被撕开、压印、重新拼贴的红黑犯罪海报，空间允许不合理地折叠、拉长、平移和错位。转场依靠人物剪影、短刀刀痕、纸张撕裂、红色印章、墨迹擦除和黑白反相完成。红色只用于灯笼、反派服饰、酒液、危险信号、印章与关键情绪爆点。\n【角色】女主为参考图中的黑发年轻女性，穿黑色长衣与长靴，手持一把短刀。保留她冷冽、克制、强势的气质；人物造型简洁平面，主要由黑色身体剪影、骨白色面部与刀刃高光、暗红色轮廓线组成。反派为提供参考图中的女性：黑色长发盘起，佩戴红色耳坠，身穿深红色华贵长袍，长袍有繁复的金色龙纹刺绣。她气质端庄、从容、优雅而危险，始终带着淡淡的审视感与掌控感；人物以深红剪纸轮廓、旧金色木刻纹样和黑色阴影表现，不要写实肖像，不要改变其红袍与金色刺绣的核心识别特征。\n【分镜】0–3 秒纯黑背景中落下斜向朱砂红雨线，女主的黑色长靴与短刀刀尖出现，只有人物下半身，没有完整街景。脚步落下时，骨白色木刻裂纹、红色印章残影和不规则纸片从地面向外弹开。短刀刀尖划过画面，留下一道骨白色的刀痕，切开黑暗。\n3–6 秒镜头沿短刀向上移动，女主身体由黑色剪影、骨白面部和暗红衣摆分层揭示。她从一张黑色通缉海报上的剪影瞬间翻转为木刻版画人物，再恢复为红黑剪纸动画形象。背景城市不是立体场景，而是由中式屋檐、窗棂、灯笼、电线和高楼轮廓逐层压印拼合。\n6–9 秒女主继续向前行走，人物保持侧面平移，背景像展开的长卷画一样向反方向滑动。巨大的暗红色印章圆形从她身后压下，黑色屋檐和骨白雨线穿过人物，将画面切成多个不规则纸片区域。女主冰冷的眼睛短暂出现在独立的横向木刻特写框中。\n9–12 秒一只骨白色纸剪蝴蝶掠过女主眼前，蝴蝶轮廓放大成为覆盖画面的红黑纸屑。女主握住短刀，刀刃横向扫过，但不进入写实战斗；刀光直接转化为白色书法笔锋、破碎的红色月亮和飞散的印刷颗粒。反派以红袍女性的皮影剪影出现：金色龙纹像木刻图案沿衣袖和披肩亮起，她站在红月中央，面容只显露一双冷静的眼睛与微微上扬的嘴角。\n12–15 秒进入快速红黑版画蒙太奇：女主侧脸、握刀手部、短刀轮廓、反派红袍上的金色龙纹、反派耳坠、红色满月、灯笼和雨夜屋檐不断重叠。人物动作以关键姿态、剪纸抽拉和海报定格表现，不制作连续写实打斗。每次鼓点触发黑白反相、单帧朱砂红闪和纸张撕裂。最后女主站在纯黑城市剪影中央，短刀垂在身侧；反派的红袍剪影在她身后如幕布般展开，猩红月亮收缩成一滴酒液，酒液扩散为原创游戏标题，出现英文文案：“BLADE IN THE RAIN”。\n【严格限制】全片保持二维剪纸、木刻版画、皮影与红黑犯罪海报美术不要抽象科技 MG，不要写实光影，不要立体 3D 城市，不要电影级真实透视，不要复杂连续打斗，不要大面积霓虹灯，不要赛博朋克写实街景。人物、建筑、房间和道具全部平面化、剪影化、纸张化、图形化，所有镜头都像一张动态的红黑犯罪版画海报。",
    "promptSourceUrl": "https://x.com/TanLuAI/status/2086053960700285194",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/TanLuAI/status/2086053955315064990",
    "sourceLabel": "X 原帖 · @TanLuAI",
    "author": "@TanLuAI",
    "publishedAt": "2026-08-08T11:36:34.612Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086053955315064990.mp4",
    "posterUrl": "/posters/x/2086053955315064990.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "有 Prompt",
      "完整 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Red-and-Black Papercut Game Opening",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2086051214425477564",
    "title": "小小花匠与沉睡的山岭巨人 · 下篇",
    "titleEn": "The Tiny Florist and the Sleeping Mountain Giant · Part 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《小小花匠与沉睡的山岭巨人 · 下篇》的 H3 成片。本页保留原始 X 来源链接。来源未公开完整 Prompt，本库不反推或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Tiny Florist and the Sleeping Mountain Giant · Part 2. This page retains the original X source link. The source did not publish a complete Prompt, so this library does not infer or complete one. Hosted-video output: 15s · 1280×720.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2086051214425477564",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-08T11:25:41.133Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086051214425477564.mp4",
    "posterUrl": "/posters/x/2086051214425477564.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "nature",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second Storyboard Image to video prompt for MiniMax H3 on @higgsfield \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for “The Tiny Florist and the Sleeping Mountain Giant” as the PRIMARY visual, character, environment, scale, prop, color, performance, and narrative reference.\n\nThis sequence continues DIRECTLY from PART 1.\n\nPART 1 ended with:\n- Brum still peacefully asleep beneath the ancient tree\n- exactly ONE small flower already tucked into his enormous beard\n- Mimi discovering that her woven flower basket is still completely full\n- Mimi getting an excited “bigger makeover” idea\n- Buzz realizing exactly where this is going and saying “NO.”\n\nGenerate a FINISHED cinematic animation from this storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- dialogue bubbles\n- storyboard headers\n- production notes\n- character reference drawings\n- scale charts\n- paper texture\n- footer decorations\n\nThe uploaded storyboard locks:\n\n1. Mimi’s exact identity and costume\n2. Buzz’s exact design and goggles\n3. Brum’s exact giant face, beard, vest and proportions\n4. extreme tiny-vs-giant scale\n5. existing single flower from Part 1\n6. progressive flower-beard makeover\n7. pink petal approaching Brum’s nose\n8. giant harmless sneeze\n9. Mimi safely holding one beard strand\n10. Brum waking gently\n11. Mimi’s tiny “Surprise?” reaction\n12. Brum lifting Mimi safely onto his palm\n13. pond reflection reveal\n14. Brum loving the makeover\n15. final Buzz “NO.” joke\n16. squirrel relaxing inside the finished beard\n17. final magical forest hero shot\n\nTranslate the 15 storyboard panels into approximately 9 fluid cinematic shots.\n\nDO NOT create fifteen rapid one-second cuts.\n\nPrioritize:\nclear cause → action → consequence → reaction → emotional payoff.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nEXACTLY 15 seconds\n\nStory time:\n0:15–0:30\n\nAspect Ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D animated short film\n\nGenre:\nwhimsical fantasy comedy + heartwarming friendship\n\nAudience:\nfamily-friendly\n\nDialogue:\nnatural synchronized English\n\nAudio:\nnative dialogue + SFX + forest ambience + original music\n\nEMOTIONAL ARC:\n\ncreative obsession\n→ escalating makeover comedy\n→ proud reveal\n→ suspicious nose twitch\n→ tiny petal suspense\n→ enormous harmless sneeze\n→ awkward “caught” moment\n→ giant discovers makeover\n→ unexpected emotional appreciation\n→ joyful friendship\n→ final comedy button\n\nThe finished 30-second story should feel COMPLETE at the end.\n\n--------------------------------------------------\n\n[PART 2 CORE STORY]\n\nMimi decides one flower is not enough.\n\nShe decorates Brum’s beard with:\n- daisies\n- pink blossoms\n- little yellow flowers\n- small vines\n- tiny mushrooms\n- one tiny sunflower\n\nThe enormous messy beard becomes a whimsical miniature beard garden.\n\nMimi proudly admires her masterpiece.\n\nBuzz notices Brum’s nose twitching.\n\nA loose pink petal drifts toward Brum’s nostril.\n\nThe petal tickles his nose.\n\nBrum inhales.\n\nMimi and Buzz realize what is about to happen.\n\nBrum sneezes.\n\nThe huge but harmless sneeze sends petals swirling through the air.\n\nMimi grabs one enormous beard strand safely.\n\nBuzz tumbles comedically.\n\nBrum wakes.\n\nMimi nervously says:\n\n<d>[English] Surprise?</d>\n\nInstead of getting angry, Brum discovers his decorated beard.\n\nHe sees himself reflected in a nearby pond.\n\nHe loves it.\n\nHe tells Mimi no one has decorated his beard in a hundred years.\n\nThen Brum asks Buzz if he helped.\n\nBuzz instantly replies:\n\n<d>[English] NO.</d>\n\nEnd with Brum, Mimi, Buzz and the squirrel happily together.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — MIMI]\n\nMatch PART 1 and the uploaded PART 2 storyboard EXACTLY.\n\nMIMI:\n\nTiny young florist approximately the height of Brum’s thumb.\n\nAppearance:\n- warm light skin\n- large expressive brown eyes\n- soft rounded face\n- fluffy chestnut-brown short bob\n- natural loose strands\n- same small childlike proportions\n\nClothing:\n- mint-green dress / work outfit\n- sunflower-yellow florist apron\n- small red-brown gardening boots\n- tiny florist/tool pouch\n- woven flower basket\n\nPersonality:\n- fearless\n- artistic\n- cheerful\n- mischievously confident\n- caring\n- extremely proud of her work\n\nPART 2 PERFORMANCE ARC:\n\n0–4 sec:\ninspired creative frenzy\n\n4–6 sec:\nextremely proud\n\n6–9 sec:\nconfusion → realization → panic\n\n9–11 sec:\nstartled but safe\n\n11–13 sec:\nnervous vulnerability\n\n13–15 sec:\nsurprise → relief → joy\n\nCRITICAL:\n\nKeep exact:\n- face\n- chestnut bob\n- outfit\n- apron\n- boots\n- proportions\n- tiny scale\n\nMimi must remain thumb-sized throughout.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — BRUM]\n\nMatch PART 1 EXACTLY.\n\nBRUM:\n\nEnormous gentle forest giant.\n\nAppearance:\n- warm tan skin\n- massive round face\n- enormous broad nose\n- thick brown eyebrows\n- huge textured chestnut/dark-brown beard\n- moss-green vest\n- large wooden buttons\n- mossy patches\n- enormous gentle hands\n\nPersonality:\n- peaceful\n- sleepy\n- warm\n- kind\n- teddy-bear gentle\n- slightly goofy\n\nPART 2 PERFORMANCE ARC:\n\nsleeping peacefully\n→ subtle nose twitch\n→ sneeze buildup\n→ enormous harmless sneeze\n→ groggy awakening\n→ confusion\n→ curiosity\n→ wonder\n→ heartfelt delight\n\nIMPORTANT:\n\nBrum must NEVER:\n- become angry\n- threaten Mimi\n- shout aggressively\n- grab Mimi\n- look evil\n\nThe first opened-eye reveal may feel enormous because of scale, but Brum himself is simply sleepy and confused.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — BUZZ]\n\nSame Buzz from PART 1.\n\nAppearance:\n- tiny round golden-yellow bee\n- dark stripes\n- large expressive eyes\n- tiny goggles\n- translucent wings\n- small antennae\n\nScale:\napproximately the size of Mimi’s head.\n\nPersonality:\n- cautious\n- dryly sarcastic\n- responsible\n- loyal\n- perpetually expecting Mimi’s plans to go wrong\n\nPART 2 PERFORMANCE ARC:\n\nhorrified by flower overload\n→ exhausted assistant\n→ notices danger first\n→ full sneeze panic\n→ tumbles safely\n→ hides\n→ relief\n→ denies responsibility\n\nThe FINAL “NO.” is Buzz’s biggest comedy beat.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — SQUIRREL]\n\nSame tiny squirrel from PART 1.\n\nAppearance:\n- warm brown fur\n- fluffy tail\n- large dark eyes\n- acorn\n\nThe squirrel is completely comfortable living inside Brum’s beard.\n\nIt should appear only subtly during the makeover or final shot.\n\nFinal:\nsquirrel sits happily among flowers inside the beard, eating its acorn.\n\n--------------------------------------------------\n\n[BEARD GARDEN CONTINUITY]\n\nAt the first frame:\n\nBrum’s beard still has EXACTLY the single flower from Part 1.\n\nBuild progressively.\n\nSTAGE 1:\nsingle existing flower\n\nSTAGE 2:\nMimi adds several daisies and pink blossoms\n\nSTAGE 3:\nsmall vines and additional flowers\n\nSTAGE 4:\ntiny mushrooms + small sunflower\n\nSTAGE 5:\nfinished beard garden\n\nFinished appearance:\n- brown beard remains dominant and visible\n- flowers woven naturally between curls\n- white daisies\n- pink blossoms\n- little yellow flowers\n- small lavender flowers\n- green vines\n- tiny red-and-white mushrooms\n- one small sunflower\n- bits of moss\n\nDo NOT cover Brum’s entire beard with flowers.\n\nIt must still clearly be a beard.\n\n--------------------------------------------------\n\n[FLOWER PHYSICS]\n\nFlowers are real lightweight flowers.\n\nThey should:\n- bend softly\n- move with Brum’s breathing\n- react naturally to breeze\n- shed only a few petals during sneeze\n\nNo glowing magical flowers.\n\nNo instant plant growth.\n\nMimi physically decorates the beard.\n\n--------------------------------------------------\n\n[KEY SNEEZE PETAL LOCK]\n\nUse ONE clearly readable soft pink petal.\n\nThe suspense sequence must clearly show:\n\nfinished beard\n→ Brum’s nose twitches\n→ pink petal loosens\n→ petal floats upward\n→ petal approaches nostril\n→ touches / tickles nose\n→ Brum inhales\n→ SNEEZE.\n\nDo not make the sneeze happen randomly.\n\nThe viewer must understand the cause.\n\n--------------------------------------------------\n\n[SNEEZE SAFETY RULE]\n\nThe sneeze is the largest comedy action of the film, but it is completely harmless.\n\nThe gust may cause:\n- flower petals to swirl\n- loose leaves to fly\n- Mimi’s hair to blow backward\n- Mimi’s apron to flutter\n- Buzz to spin through petals\n- butterflies to scatter\n- light flowers to shake\n\nMimi grabs one thick beard strand before being moved far.\n\nShe NEVER falls.\n\nBuzz NEVER crashes painfully.\n\nBrum’s beard remains intact.\n\nNo trees break.\n\nNo objects become dangerous projectiles.\n\nNo injuries.\n\nNo destruction.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK]\n\nUse the same enchanted forest clearing as Part 1.\n\nMaintain:\n- giant ancient tree\n- thick roots\n- golden sunlight\n- ferns\n- wildflowers\n- mushrooms\n- stones\n- leafy canopy\n- atmospheric forest depth\n- warm moss\n- nearby reflective pond\n- drifting petals\n- Brum’s same resting position\n\nThe pond should already exist spatially nearby so Brum can see his reflection naturally after waking.\n\nDo not teleport characters to another location.\n\n--------------------------------------------------\n\n[SCALE RULES]\n\nScale is CRITICAL.\n\nMimi = approximately Brum’s thumb height.\n\nBuzz = Mimi-head-sized.\n\nSquirrel = slightly larger than Mimi.\n\nBrum = colossal.\n\nVisual examples:\n- Mimi fits comfortably in Brum’s palm\n- Brum’s eye is larger than Mimi\n- Brum’s nose towers above Mimi\n- one beard strand is thick enough for Mimi to grip with both hands\n\nMaintain scale even during close-ups.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–2.20s\nONE FLOWER BECOMES A WHOLE PLAN\n\nContinue directly from Part 1.\n\nMimi stands inside Brum’s enormous beard beside the ONE flower she already placed.\n\nShe looks toward her overflowing basket.\n\nBuzz hovers beside her.\n\nBuzz:\n<d>[English] Mimi... no.</d>\n\nMimi gives him an innocent smile.\n\nMimi:\n<d>[English] Just a few more.</d>\n\nImmediately begin a fluid makeover montage.\n\nMimi grabs several daisies and pink blossoms.\n\nShe moves quickly but carefully across huge beard curls.\n\nShe tucks:\none flower,\nthen another,\nthen another.\n\nBuzz follows with increasing disbelief.\n\nDialogue:\n\nBuzz:\n<d>[English] That is already more than “a few.”</d>\n\nMimi:\n<d>[English] Art cannot be counted, Buzz.</d>\n\nCamera:\nclose-up of single flower\n→ pan toward full basket\n→ Mimi inspired expression\n→ tracking across beard as decorations begin.\n\nSound:\n- basket wicker\n- soft petals\n- beard fibers\n- Buzz wings\n- Brum snoring underneath everything\n\nMusic:\nfast light pizzicato makeover rhythm.\n\n--------------------------------------------------\n\n[SHOT 2] — 2.20s–4.20s\nTHE BEARD MAKEOVER ESCALATES\n\nContinue in one energetic montage.\n\nMimi:\n- loops a soft green vine around one beard curl\n- tucks white daisies into another\n- places a tiny mushroom between curls\n- adjusts several pink blossoms\n- carefully sets one small sunflower into the upper beard\n\nBuzz carries a tiny flower stem reluctantly.\n\nHe looks exhausted.\n\nA squirrel briefly peeks from the beard, watches them, takes one calm bite from its acorn, then disappears back inside.\n\nMimi finishes one last adjustment.\n\nDialogue:\n\nMimi:\n<d>[English] Daisy... vine... mushroom...</d>\n\nShe raises the tiny sunflower.\n\nMimi:\n<d>[English] And... masterpiece!</d>\n\nBuzz:\n<d>[English] We have different definitions of “small makeover.”</d>\n\nCamera:\nmacro flower placement\n→ beard-level lateral tracking\n→ squirrel reaction\n→ pull outward.\n\n--------------------------------------------------\n\n[SHOT 3] — 4.20s–5.70s\nTHE BEAUTIFUL BEARD GARDEN\n\nWide reveal.\n\nBrum remains completely asleep beneath the giant tree.\n\nHis massive brown beard is now transformed into a whimsical miniature flower garden.\n\nKeep the brown beard highly visible beneath:\n- flowers\n- vines\n- mushrooms\n- tiny sunflower\n\nA few butterflies naturally approach.\n\nMimi stands proudly on Brum’s chest / lower beard area with hands on hips.\n\nBuzz hovers beside her, exhausted.\n\nMimi:\n<d>[English] Perfect.</d>\n\nBuzz looks upward.\n\nHis eyes suddenly widen.\n\nBuzz:\n<d>[English] Mimi...</d>\n\nCamera:\nbeautiful wide reveal\n→ slow push toward Buzz’s worried face.\n\nMusic:\nmakeover melody resolves beautifully, then suddenly thins.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.70s–7.30s\nTHE NOSE TWITCH\n\nExtreme scale close-up.\n\nBrum’s enormous nose fills much of the frame.\n\nTWITCH.\n\nMimi looks upward.\n\nBuzz points urgently.\n\nBrum’s nose twitches again.\n\nTWITCH.\n\nOne soft pink flower petal loosens from the decorated beard.\n\nIt slowly rises on Brum’s breathing airflow.\n\nMimi watches it.\n\nBuzz watches it.\n\nCamera alternates focus:\n\nnose\n→ floating petal\n→ Mimi\n→ Buzz\n→ nostril.\n\nDialogue:\n\nMimi, whispering:\n<d>[English] What?</d>\n\nBuzz:\n<d>[English] Petal.</d>\n\nMimi:\n<d>[English] Where?</d>\n\nBuzz points.\n\nMimi sees it.\n\nMimi:\n<d>[English] Oh no.</d>\n\nSound:\n- nose twitch\n- subtle air inhale\n- tiny petal flutter\n- Buzz wings stop briefly\n\nMusic:\nplayful suspense plucks.\n\n--------------------------------------------------\n\n[SHOT 5] — 7.30s–8.60s\n“AAAAAH...”\n\nThe pink petal floats directly toward Brum’s nostril.\n\nTime feels slightly slower for comedy.\n\nThe petal gently touches the edge of his nose.\n\nBrum’s eyebrows squeeze.\n\nNostrils flare slightly.\n\nHis chest begins a giant inhale.\n\n<d>[English] Aaaaaah...</d>\n\nMimi freezes.\n\nBuzz’s eyes become enormous.\n\nMimi quickly grabs one thick beard strand with both arms.\n\nBuzz grabs onto a flower stem or hovers ready.\n\nMimi:\n<d>[English] Hold on!</d>\n\nBuzz:\n<d>[English] I KNEW IT!</d>\n\nCamera:\nmacro petal contact\n→ Brum pre-sneeze face\n→ tiny Mimi/Buzz reaction at beard scale\n→ wider shot showing huge size difference.\n\nSound:\nmassive but comedic inhale.\n\nMusic:\nascending comedic tension.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.60s–9.70s\nTHE GIANT SNEEZE\n\nBRUM:\n\n<d>[English] ACHOOOO!</d>\n\nA huge harmless gust bursts outward.\n\nShow a spectacular painterly swirl of:\n- pink petals\n- daisies\n- light leaves\n- flower pollen sparkles\n- butterflies\n\nMimi safely grips one thick beard hair.\n\nHer body stretches nearly horizontal from wind for a brief comedic beat.\n\nHer apron and hair flutter wildly.\n\nBuzz spins harmlessly through flower petals.\n\nHis goggles stay attached.\n\nThe squirrel ducks comfortably deeper into Brum’s beard while still holding its acorn.\n\nOnly a few loose flowers fly away.\n\nMost of the beard garden remains beautifully intact.\n\nCamera:\nmedium-wide side angle showing gust\n→ brief dynamic follow of swirling petals\n→ return toward Mimi gripping beard.\n\nSound:\nhuge warm cartoon sneeze\n- gust of air\n- petals\n- Mimi tiny surprised yell\n- Buzz spinning buzz\n\nMusic:\nlarge playful orchestral/comedy accent.\n\nNO violent impact.\n\n--------------------------------------------------\n\n[SHOT 7] — 9.70s–11.20s\nAND NOW THE GIANT IS AWAKE\n\nThe sneeze ends.\n\nEverything becomes suddenly quiet.\n\nPetals drift slowly downward.\n\nMimi hangs safely from one thick beard strand.\n\nHer basket hangs from one arm.\n\nBuzz hides behind the tiny sunflower.\n\nBrum’s giant eyelid slowly opens.\n\nHis enormous eye focuses.\n\nHe looks downward.\n\nMimi looks up.\n\nThe giant eye fills most of her world.\n\nPause.\n\nNo scary music.\n\nMimi slowly lifts one tiny hand.\n\nMimi:\n<d>[English] ...Surprise?</d>\n\nBuzz peeks from behind flower.\n\nBrum blinks.\n\nCamera:\nMimi hanging\n→ enormous eye opening\n→ extreme scale two-shot\n→ hold on awkward silence.\n\nSound:\n- one last petal falling\n- forest suddenly audible again\n- soft eye blink\n- Buzz nervous wing buzz\n\nMusic:\none tiny comedic pluck, then silence.\n\n--------------------------------------------------\n\n[SHOT 8] — 11.20s–13.20s\nBRUM DISCOVERS HIS MAKEOVER\n\nBrum does NOT get angry.\n\nHis expression is sleepy and confused.\n\nHe carefully places one enormous open palm below Mimi.\n\nMimi lets go of the beard and lands safely standing in the center of his palm.\n\nBuzz lands beside her.\n\nBrum uses his other hand to gently touch the flowers in his beard.\n\nHe notices:\ndaisies,\nvine,\nmushrooms,\nsunflower.\n\nBrum:\n<d>[English] Flowers?</d>\n\nMimi nervously nods.\n\nMimi:\n<d>[English] I thought your beard needed... a little help.</d>\n\nBrum raises one eyebrow.\n\nHe glances toward the nearby pond.\n\nCamera follows his gaze.\n\nShow Brum’s reflection in the still water.\n\nFor the first time he sees:\nhis enormous warm face\n+\nmagnificent flower-covered beard.\n\nHis confusion becomes surprise.\n\nThen the biggest gentle smile.\n\nDialogue:\n\nBrum, softly:\n<d>[English] Oh...</d>\n\nSound:\n- soft beard rustle\n- pond water\n- warm forest ambience\n\nMusic:\nheartwarming melody begins.\n\n--------------------------------------------------\n\n[SHOT 9] — 13.20s–15.00s\nTHE BIGGEST DAY BRIGHTENED\n\nBrum holds Mimi and Buzz safely on his open palm near his face.\n\nHis eyes are warm.\n\nBrum:\n<d>[English] No one has decorated my beard in a hundred years.</d>\n\nMimi’s nervous expression melts into a proud smile.\n\nBrum:\n<d>[English] I love it.</d>\n\nMimi:\n<d>[English] Really?</d>\n\nBrum nods.\n\nVery quick fluid transition into the FINAL hero composition:\n\nBrum now walks happily through the sunlit forest.\n\nHis flower beard is bright and beautiful.\n\nMimi sits proudly on his shoulder.\n\nBuzz flies beside them.\n\nThe squirrel sits comfortably inside the beard eating its acorn.\n\nButterflies gently follow the flowers.\n\nBrum glances toward Buzz.\n\nBrum:\n<d>[English] Did you help?</d>\n\nBuzz immediately shakes his entire body.\n\nBuzz:\n<d>[English] NO.</d>\n\nMimi bursts into a tiny laugh.\n\nBrum chuckles warmly.\n\nCamera:\nmedium emotional palm shot\n→ seamless pull/transition to final walking wide\n→ slow final pull-back through flowers and golden forest light.\n\nHold final composition briefly.\n\nFADE TO BLACK.\n\nEND.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[MIMI VOICE]\n\nNatural tiny young girl voice.\n\nQualities:\n- bright\n- playful\n- confident\n- warm\n- artistic\n- expressive\n- clear English\n- slightly mischievous\n\nPerformance progression:\n\n0–5 sec:\ncreative excitement.\n\n5–8 sec:\nconcerned whisper.\n\n8–10 sec:\ncomedic panic.\n\n10–12 sec:\ntiny nervous vulnerability.\n\n12–15 sec:\nrelief and joy.\n\nImportant line:\n\n<d>[English] ...Surprise?</d>\n\nDeliver this very softly and awkwardly.\n\nIt should be one of the funniest moments.\n\n--------------------------------------------------\n\n[BUZZ VOICE]\n\nSmall expressive bee voice.\n\nQualities:\n- slightly high\n- fast\n- skeptical\n- anxious\n- dry comedic timing\n- subtle buzzing quality\n\nImportant moments:\n\n<d>[English] Petal.</d>\n\n<d>[English] I KNEW IT!</d>\n\nFinal:\n\n<d>[English] NO.</d>\n\nThe final “NO.” must be immediate, loud relative to Buzz’s tiny voice, and perfectly timed for comedy.\n\n--------------------------------------------------\n\n[BRUM VOICE]\n\nLarge, deep but extremely gentle giant voice.\n\nQualities:\n- warm bass\n- sleepy\n- slow\n- kind\n- comforting\n- slightly rustic\n- no threatening resonance\n\nHis huge body size gives his voice depth, but the delivery remains soft.\n\nDialogue:\n\n<d>[English] Flowers?</d>\n\n<d>[English] No one has decorated my beard in a hundred years.</d>\n\n<d>[English] I love it.</d>\n\n<d>[English] Did you help?</d>\n\nNo shouting except the involuntary sneeze.\n\n--------------------------------------------------\n\n[SQUIRREL AUDIO]\n\nMinimal:\n- tiny acorn crunch\n- little squeak\n- soft beard rustle\n\n--------------------------------------------------\n\n[FOREST ENVIRONMENT AUDIO]\n\nUse:\n- birds\n- light breeze\n- leaves\n- insects\n- butterflies subtle wing flutter\n- Brum breathing\n- distant woodland atmosphere\n- pond ambience near reflection\n- soft footfalls during final walking shot\n\nKeep environment warm and peaceful.\n\n--------------------------------------------------\n\n[SOUND DESIGN — MAKEOVER]\n\nUse satisfying miniature sounds:\n\nflower placement:\nsoft “tup”\n\nvine movement:\nlight leaf rustle\n\nmushroom placement:\nsoft tiny “pop”\n\nbasket:\nwicker movement\n\nBuzz:\nwing vibration\n\nbeard:\nsoft dense hair fibers\n\nKeep all sounds delicate relative to Brum’s scale.\n\n--------------------------------------------------\n\n[SOUND DESIGN — SNEEZE]\n\nSneeze buildup:\n\nnose twitch\n→ subtle sniff\n→ longer inhale\n→ tension silence\n→ “ACHOOOO!”\n\nThe sneeze should have:\n- deep low-frequency body\n- warm cartoon quality\n- broad gust\n- fluttering petals\n\nDo NOT use:\n- explosive boom\n- shockwave\n- destructive impact\n- frightening sound design\n\nAfter sneeze:\nimmediate contrast into near silence.\n\nThis silence makes Brum opening his eye funnier.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nUse an ORIGINAL whimsical fantasy-animation score.\n\nInstrumentation:\n- pizzicato strings\n- flute\n- clarinet\n- bassoon\n- marimba\n- glockenspiel\n- light harp\n- warm strings\n- subtle acoustic plucks\n\nMUSIC ARC:\n\n0.00–2.20\nfast mischievous makeover rhythm.\n\n2.20–4.20\ncreative montage acceleration.\n\n4.20–5.70\nwarm “masterpiece” flourish.\n\n5.70–7.30\nmusic strips down to tiny suspense plucks.\n\n7.30–8.60\nascending comedic pre-sneeze tension.\n\n8.60–9.70\nbig playful sneeze accent.\n\n9.70–11.20\nnear silence / tiny comedy note.\n\n11.20–13.20\nsoft emotional warmth.\n\n13.20–15.00\njoyful friendship resolution.\n\nFinal Buzz “NO.” can receive one tiny comic musical punctuation.\n\nNo epic trailer music.\n\nNo scary music.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nUse cinematic storybook camera work.\n\nCAMERA ARC:\n\n0–4 sec:\nenergetic miniature makeover coverage.\n\n4–6 sec:\nbeautiful wide reveal.\n\n6–8.6 sec:\nincreasingly tight suspense coverage.\n\n8.6–9.7 sec:\nlargest dynamic movement during sneeze.\n\n9.7–11.2 sec:\nsudden stillness + extreme scale.\n\n11.2–13.2 sec:\nwarm character-focused coverage.\n\n13.2–15 sec:\nopen into a beautiful final wide shot.\n\nPreferred techniques:\n- macro close-ups\n- beard-level tracking\n- gentle dolly\n- rack focus\n- extreme scale compositions\n- slow-motion-feeling petal approach\n- controlled dynamic sneeze shot\n- pond reflection shot\n- final slow pull-back\n\nAvoid:\n- shaky handheld camera\n- extreme Dutch angles\n- action-movie spinning\n- excessive zooming\n- disorienting cuts\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animated short.\n\nMATCH PART 1 EXACTLY.\n\nUse:\n- hand-drawn ink contours\n- watercolor/gouache painted fills\n- warm storybook textures\n- rich but soft forest backgrounds\n- expressive facial animation\n- appealing rounded character forms\n- highly detailed beard texture\n- gentle atmospheric depth\n- natural flower movement\n- cinematic golden lighting\n- subtle squash and stretch\n\nThe finished video should look like the uploaded illustrations have come alive.\n\nNOT:\n- storyboard animatic\n- motion comic\n- slideshow\n- photorealism\n- 3D CGI\n- vector animation\n- anime\n- cel-shaded 3D\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\n0–5.7 sec:\n\nwarm greens + browns with rapidly increasing flower color:\n- pink\n- yellow\n- white\n- lavender\n- mint\n- sunflower gold\n\n5.7–8.6 sec:\n\nkeep environment warm but isolate the pink petal visually against Brum’s brown beard/nose.\n\n8.6–9.7 sec:\n\npetal swirl becomes brightest colorful action moment.\n\nUse:\n- pink petals\n- yellow flowers\n- pale lavender\n- leaf green\n- warm sunlight\n\n9.7–13.2 sec:\n\nreturn to softer natural forest palette.\n\n13.2–15 sec:\n\nrichest warm golden light.\n\nBrum’s beard flowers glow naturally in the sunlight.\n\nFinal frame:\nwarm gold,\nmoss green,\nbrown,\ncolorful floral accents.\n\nNo artificial neon glow.\n\n--------------------------------------------------\n\n[MOTION RULES — MIMI]\n\nUse:\n- tiny quick decorating movements\n- balancing across beard curls\n- reaching into basket\n- natural flower placement\n- proud hands-on-hips pose\n- sudden freeze when nose twitches\n- firm but safe beard grip during sneeze\n- natural nervous wave after Brum wakes\n- relaxed seated shoulder pose at final\n\nDuring sneeze:\nMimi may stretch with wind, but her hands stay securely on the beard strand.\n\nNo uncontrolled flight.\n\n--------------------------------------------------\n\n[MOTION RULES — BRUM]\n\nUntil sneeze:\n- slow breathing\n- small nose twitches only\n\nPre-sneeze:\n- chest expands\n- eyebrows tighten\n- nostrils flare\n- head moves only slightly\n\nSneeze:\none clear forward motion.\n\nAfter:\n- settles\n- eyes slowly open\n- gentle blinking\n- controlled careful hand movements\n\nWhen lifting Mimi:\nBrum places palm beneath her and lets her step/drop only a tiny safe distance.\n\nNo pinching.\n\nNo grabbing.\n\n--------------------------------------------------\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2086050663365333283",
    "title": "小小花匠与沉睡的山岭巨人 · 上篇",
    "titleEn": "The Tiny Florist and the Sleeping Mountain Giant · Part 1",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《小小花匠与沉睡的山岭巨人 · 上篇》的 H3 成片。本页保留原始 X 来源链接。来源未公开完整 Prompt，本库不反推或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Tiny Florist and the Sleeping Mountain Giant · Part 1. This page retains the original X source link. The source did not publish a complete Prompt, so this library does not infer or complete one. Hosted-video output: 15s · 1280×720.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2086050663365333283",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-08T11:23:29.750Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2086050663365333283.mp4",
    "posterUrl": "/posters/x/2086050663365333283.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "nature",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "First Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 1 storyboard image for “The Tiny Florist and the Sleeping Mountain Giant” as the PRIMARY visual, character, environment, scale, prop, performance, color, and narrative reference.\n\nGenerate a FINISHED animated sequence from the storyboard.\n\nDo NOT reproduce the storyboard sheet itself.\n\nThe final video must NOT contain:\n- storyboard borders\n- storyboard panels\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character reference drawings\n- scale reference diagrams\n- storyboard paper texture\n- footer decorations\n- title/header text\n\nThe uploaded storyboard locks:\n\n1. Mimi’s exact visual identity\n2. Brum’s exact giant design\n3. Buzz’s exact bee design\n4. squirrel design\n5. extreme tiny-vs-giant scale\n6. Brum’s enormous tangled beard\n7. Brum’s moss-green vest and giant buttons\n8. Mimi’s flower basket\n9. Mimi’s tiny turquoise comb\n10. twigs, leaves, moss and tiny nest inside beard\n11. same magical forest clearing\n12. enormous ancient tree\n13. warm golden forest lighting\n14. painterly storybook 2D animation style\n15. Part 1 story progression\n16. final basket-of-flowers cliffhanger\n\nAnimate the storyboard as one seamless cinematic short.\n\nDo NOT animate all 15 storyboard panels as fifteen one-second hard cuts.\n\nCombine the visual beats into approximately 8 coherent shots with natural cause → action → reaction storytelling.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration:\nexactly 15 seconds\n\nStory time:\n0:00–0:15\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nGenre:\nwhimsical fantasy comedy\n\nVisual medium:\npremium painterly 2D animated short film\n\nAudience:\nfamily-friendly\n\nDialogue language:\nEnglish\n\nAudio:\nnative synchronized dialogue + character sounds + environment + SFX + music\n\nEMOTIONAL ARC:\n\nmagical calm\n→ enormous sleepy reveal\n→ scale comedy\n→ Mimi notices a problem\n→ secret mission\n→ delicate grooming\n→ surprise beard resident\n→ silent teamwork\n→ first flower success\n→ dangerous BIG IDEA cliffhanger\n\nPart 1 should feel cute, clever and increasingly mischievous.\n\n--------------------------------------------------\n\n[PART 1 STORY LIMIT]\n\nPART 1 ONLY.\n\nSTART:\n\nBrum is peacefully asleep beneath the enormous ancient tree.\n\nEND:\n\nMimi has placed ONLY ONE flower in Brum’s beard.\n\nShe notices her still-full basket of flowers.\n\nHer face lights up with a much bigger decorating idea.\n\nBuzz realizes what she is thinking.\n\nCUT TO BLACK.\n\nDO NOT show:\n- multiple flowers being added\n- finished beard garden\n- vines wrapped through beard\n- mushrooms added to beard\n- tiny sunflower decoration\n- butterflies gathering around decorated beard\n- Brum’s nose twitch\n- floating petal near his nose\n- giant sneeze\n- Brum waking\n- Mimi hanging from beard hair\n- pond reflection\n- emotional approval\n- Brum walking through forest\n\nThose belong to PART 2.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — MIMI]\n\nMatch the uploaded storyboard exactly.\n\nMIMI is extremely tiny — approximately the height of one of Brum’s thumbs.\n\nAppearance:\n- tiny young florist\n- warm light skin\n- large expressive brown eyes\n- soft rounded face\n- fluffy chestnut-brown bob haircut\n- tousled short hair\n- natural loose strands around her face\n\nClothing:\n- mint-green dress / work outfit\n- sunflower-yellow florist apron\n- tiny red-brown gardening boots\n- miniature florist pouch\n- woven flower basket\n- same clothing colors and proportions throughout\n\nPersonality:\n- fearless\n- playful\n- artistic\n- curious\n- highly confident\n- caring\n- easily inspired\n\nPART 1 PERFORMANCE ARC:\n\n0–3 sec:\nhappy forest traveler → startled by giant snore\n\n3–5 sec:\nconcerned professional florist studying Brum’s beard\n\n5–9 sec:\ndetermined tiny adventurer\n\n9–12 sec:\nconfident groomer → startled by squirrel\n\n12–15 sec:\ndelighted by first flower → dangerous creative inspiration\n\nMimi must never look genuinely frightened of Brum.\n\nShe knows he is sleeping and believes she can improve his beard.\n\nCRITICAL IDENTITY:\n\nMaintain exact:\n- face\n- chestnut bob\n- clothing\n- yellow apron\n- green outfit\n- boots\n- scale\n- basket\n\nNo costume drift.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — BRUM]\n\nMatch the storyboard exactly.\n\nBRUM:\n\nan enormous gentle mountain/forest giant.\n\nAppearance:\n- warm tan skin\n- enormous rounded face\n- giant broad nose\n- thick dark-brown eyebrows\n- huge fluffy chestnut/dark-brown beard\n- very long tangled beard curls\n- mossy green vest\n- giant circular wooden buttons\n- slight moss / natural forest textures\n- enormous hands\n- peaceful closed eyes\n\nScale:\n\nBrum must feel truly enormous.\n\nMimi:\napproximately thumb-height.\n\nOne vest button:\nclose to Mimi’s torso height.\n\nOne thick beard curl:\nlarge enough for Mimi to stand beside and climb around.\n\nBrum’s face:\nmany times larger than Mimi’s entire body.\n\nPersonality:\n- gentle\n- peaceful\n- deep sleeper\n- warm teddy-bear energy\n\nDuring PART 1 Brum NEVER fully wakes.\n\nAllowed sleeping reactions:\n- breathing\n- enormous snore\n- small smile\n- eyebrow twitch\n- subtle beard movement from breath\n\nDo NOT open Brum’s eyes.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — BUZZ]\n\nMatch storyboard exactly.\n\nBUZZ:\n\ntiny round golden bee approximately the size of Mimi’s head.\n\nAppearance:\n- golden-yellow fuzzy round body\n- darker bee stripes\n- large expressive eyes\n- tiny blue-gray goggles\n- small translucent wings\n- little antennae\n- miniature tool carrier / tiny blue florist bucket when visible\n\nPersonality:\n- responsible\n- cautious\n- loyal\n- anxious\n- highly expressive\n- comedy warning character\n\nPART 1 PERFORMANCE:\n\nhappy companion\n→ blown by snore\n→ concerned about Mimi’s plan\n→ nervous during climb\n→ watches Brum carefully\n→ shocked by squirrel\n→ increasingly alarmed by flower idea\n\nBuzz should mostly communicate through:\n- wing buzzes\n- tiny gasps\n- squeaks\n- short dialogue\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — SQUIRREL]\n\nMatch the storyboard.\n\nA tiny warm-brown woodland squirrel.\n\nFeatures:\n- soft warm-brown fur\n- lighter belly\n- large dark eyes\n- fluffy curled tail\n- holding one acorn\n- comfortable living inside Brum’s beard\n\nPersonality:\n- surprisingly calm\n- protective of sleeping Brum\n- treats the beard like home\n\nThe squirrel appears ONLY after Mimi removes a twig.\n\nIts reveal should be surprising but adorable.\n\n--------------------------------------------------\n\n[BRUM'S BEARD — ENVIRONMENT LOCK]\n\nBrum’s beard is not merely facial hair.\n\nFrom Mimi’s scale, it behaves like a miniature landscape.\n\nUse:\n- enormous twisted beard curls\n- soft brown hair ridges\n- tunnels between curls\n- tangled twigs\n- green leaves\n- small moss patches\n- tiny nest\n- hidden squirrel space\n\nThe beard must still clearly look like natural hair.\n\nDo not turn it into literal soil or forest ground.\n\nHair physics:\nsoft,\nthick,\nspringy,\nheavy relative to Mimi.\n\nWhen Mimi brushes:\nindividual curls gently straighten and settle.\n\n--------------------------------------------------\n\n[KEY PROP LOCK — MIMI'S COMB]\n\nTiny turquoise/mint hand comb.\n\nSmall enough for Mimi.\n\nUsed to groom Brum’s beard carefully.\n\nDo not change it into a full-sized human comb.\n\n--------------------------------------------------\n\n[KEY PROP LOCK — FLOWER BASKET]\n\nMimi’s woven florist basket.\n\nContains many colorful fresh flowers:\n- pink\n- purple\n- yellow\n- white\n- small daisies\n- little wildflowers\n- leafy stems\n\nAt the beginning:\nbasket remains mostly full.\n\nAt the end:\nonly ONE flower has been removed.\n\nThe basket must still look obviously full.\n\nThis visual is essential to the cliffhanger.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK — MAGICAL FOREST]\n\nMaintain the exact storybook forest atmosphere from the uploaded storyboard.\n\nSetting:\n- enormous ancient tree\n- thick trunk\n- sprawling roots\n- mossy forest floor\n- ferns\n- mushrooms\n- wildflowers\n- scattered leaves\n- stones\n- soft grassy patches\n- distant forest depth\n- warm golden sunlight filtering through canopy\n- tiny drifting petals / leaves\n- atmospheric dust motes\n\nBrum sleeps at the base of the same tree.\n\nDo not relocate him.\n\nLighting direction must remain coherent throughout.\n\n--------------------------------------------------\n\n[SCALE RULES]\n\nExtreme scale contrast is one of the main visual hooks.\n\nAlways preserve:\n\nMimi = thumb-sized.\n\nBuzz = approximately Mimi-head-sized.\n\nSquirrel = comparable to / slightly larger than Mimi.\n\nBrum = gigantic.\n\nImportant visual scale examples:\n\nMimi climbs one vest button like a stair.\n\nMimi walks through beard curls like thick tree roots.\n\nOne beard hair bundle is large enough for Mimi to hold.\n\nBrum’s nose is enormous relative to Mimi.\n\nNever accidentally render Mimi adult-human-sized next to Brum.\n\n--------------------------------------------------\n\n[WORLD / PHYSICS RULES]\n\nAll actions follow understandable physical cause and effect.\n\nBrum snores\n→ air pushes leaves and petals\n→ Mimi and Buzz wobble backward.\n\nMimi sees messy beard\n→ decides to fix it.\n\nMimi climbs giant vest\n→ reaches beard.\n\nMimi combs beard\n→ twig becomes loose.\n\nMimi pulls twig\n→ squirrel hidden behind it appears.\n\nSquirrel warns them\n→ everyone becomes quiet.\n\nSquirrel leaves\n→ Mimi notices flower.\n\nMimi places ONE flower\n→ beard looks unexpectedly beautiful.\n\nMimi sees full basket\n→ inspiration escalates.\n\nNo unexplained magic.\n\nThe forest may feel magical, but no supernatural transformations occur in Part 1.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.80s\nTHE SLEEPING MOUNTAIN\n\nOpen with a beautiful cinematic wide shot of the enchanted forest clearing.\n\nGolden sunlight streams through the huge tree canopy.\n\nTiny particles float in the light.\n\nWildflowers sway.\n\nCamera slowly glides around one massive tree root.\n\nReveal BRUM asleep against the enormous ancient trunk.\n\nHis giant green-vested body occupies a large portion of the clearing.\n\nHis enormous brown beard spills across his chest like a tangled hillside.\n\nBrum inhales.\n\nThen—\n\n<d>[English] SNORRRR...</d>\n\nA harmless breath wave travels through nearby flowers.\n\nPetals and leaves flutter outward.\n\nCamera:\nwide forest establishing shot\n→ slow reveal of Brum\n→ gentle push toward sleeping face.\n\nSound:\n- forest birds\n- leaves\n- distant insects\n- low peaceful giant breathing\n- deep comedic snore\n- flowers fluttering\n\nMusic:\nsoft magical flute + pizzicato + warm strings.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.80s–3.30s\nTHE TINY FLORIST MEETS THE SNORE\n\nCut much lower to Mimi’s scale.\n\nMimi walks into the clearing carrying her flower basket.\n\nBuzz hovers beside her.\n\nAnother giant snore arrives.\n\nA visible but soft gust pushes:\n- Mimi’s short hair backward\n- her yellow apron outward\n- Buzz several centimeters sideways\n- loose petals past camera\n\nMimi plants her tiny boots and leans forward against the breeze.\n\nBuzz spins once and rights himself.\n\nMimi looks upward.\n\nHer eyes grow huge when she finally understands the “hill” is a sleeping giant.\n\nDialogue:\n\nMimi:\n<d>[English] Whoa...</d>\n\nBuzz:\n<d>[English] Very big napper.</d>\n\nMimi smiles rather than panics.\n\nCamera:\nlow character-level medium shot\n→ tilt upward along Brum’s enormous body\n→ emphasize extreme scale.\n\nSound:\n- snore gust\n- cloth flutter\n- Buzz wing buzz\n- falling leaves\n\n--------------------------------------------------\n\n[SHOT 3] — 3.30s–5.00s\nTHE REAL PROBLEM\n\nMimi studies Brum.\n\nHer attention moves from his peaceful sleeping face...\n\ndown into the enormous tangled beard.\n\nCamera follows her gaze.\n\nReveal close beard details:\n\n- tangled curls\n- twig\n- several leaves\n- moss\n- tiny nest tucked deep inside\n- messy knots\n\nMimi puts one tiny hand to her chin like a professional stylist inspecting a client.\n\nHer expression changes:\n\nsurprise\n→ concern\n→ determination.\n\nDialogue:\n\nMimi:\n<d>[English] Buzz...</d>\n\nBuzz:\n<d>[English] Yes?</d>\n\nMimi points dramatically toward beard.\n\n<d>[English] That beard needs help.</d>\n\nBuzz stares at the enormous beard.\n\nThen at sleeping Brum.\n\nBuzz:\n<d>[English] Of course it does.</d>\n\nCamera:\nMimi close-up\n→ POV into beard landscape\n→ macro details\n→ two-shot Mimi/Buzz.\n\nMusic:\nsmall mischievous plucked motif begins.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.00s–6.80s\nTHE SECRET BEARD MISSION\n\nMimi tightens her florist pouch and starts climbing Brum’s moss-green vest.\n\nShe uses Brum’s gigantic wooden vest buttons like climbing steps.\n\nShow scale clearly:\n\none button is nearly as large as Mimi’s torso.\n\nMimi climbs:\nbutton\n→ fold in vest\n→ next button.\n\nBuzz follows hovering close behind with her tiny supplies.\n\nBrum continues breathing deeply.\n\nHis huge chest rises gently beneath them.\n\nMimi pauses for one breath as the “ground” rises.\n\nThen continues.\n\nDialogue:\n\nMimi, whispering:\n<d>[English] Quiet feet.</d>\n\nBuzz:\n<d>[English] You are climbing a giant.</d>\n\nMimi:\n<d>[English] Very quiet feet.</d>\n\nCamera:\nside wide scale shot\n→ low view upward along giant vest\n→ close-up Mimi pulling herself onto final button\n→ reveal beard looming above.\n\nSound:\n- tiny boot taps\n- fabric creak\n- Buzz wing hum\n- Brum breathing\n- small button tap\n\n--------------------------------------------------\n\n[SHOT 5] — 6.80s–9.20s\nTHE BEARD LANDSCAPE\n\nMimi and Buzz reach Brum’s beard.\n\nFrom their scale, enormous brown curls surround them like thick twisting vines.\n\nMimi steps carefully onto a broad curl.\n\nShe removes her tiny turquoise comb.\n\nShe begins gently combing one small section.\n\nSCRITCH...\nSCRITCH...\n\nA leaf comes loose.\n\nThen another.\n\nOne beard curl slowly becomes neat and smooth.\n\nBrum’s sleeping mouth forms the tiniest peaceful smile.\n\nMimi notices.\n\nHer confidence grows.\n\nDialogue:\n\nMimi, whispering:\n<d>[English] See? He likes it.</d>\n\nBuzz watches Brum nervously.\n\nBuzz:\n<d>[English] He is unconscious.</d>\n\nMimi:\n<d>[English] Still counts.</d>\n\nCamera:\nwide beard-landscape reveal\n→ close-up tiny comb moving through giant hair\n→ leaf falling\n→ cut to Brum’s sleepy smile\n→ Mimi proud reaction.\n\nSound:\n- soft comb through hair\n- hair fibers\n- leaves falling\n- Brum sleepy exhale\n- Buzz wing flutter\n\nMusic:\ngentle playful grooming rhythm.\n\n--------------------------------------------------\n\n[SHOT 6] — 9.20s–11.40s\nTHE SECRET RESIDENT\n\nMimi notices one stubborn twig lodged between beard curls.\n\nShe sets down the comb.\n\nGrabs the twig with both tiny hands.\n\nShe pulls.\n\nOnce.\n\nNothing.\n\nShe braces one boot against a beard curl.\n\nPulls again.\n\nThe twig pops free.\n\nBehind it—\n\ntwo tiny eyes appear.\n\nMimi freezes.\n\nBuzz freezes.\n\nA warm-brown SQUIRREL suddenly pops out holding an acorn.\n\nAll three stare at each other.\n\nOne quiet beat.\n\nMimi:\n<d>[English] Oh!</d>\n\nBuzz:\n<d>[English] Oh no.</d>\n\nThe squirrel puts one finger/paw near its mouth.\n\n“Shhh.”\n\nThen points upward toward sleeping Brum.\n\nMimi instantly understands.\n\nShe slowly puts a finger over her own lips.\n\nDialogue, whispered:\n\nMimi:\n<d>[English] We won’t wake him.</d>\n\nThe squirrel gives a serious nod.\n\nCamera:\nclose-up twig\n→ tiny pull\n→ reveal squirrel\n→ reaction close-ups\n→ wider four-layer scale composition with sleeping Brum behind.\n\nSound:\n- twig creak\n- tiny “pop”\n- soft squirrel squeak\n- music briefly stops for reveal\n- Brum continuing to snore quietly\n\n--------------------------------------------------\n\n[SHOT 7] — 11.40s–13.30s\nONE LITTLE FLOWER\n\nThe squirrel quietly slips deeper through the beard and exits toward another curl, still carrying its acorn.\n\nMimi watches it go.\n\nThen she notices a single small purple-pink wildflower caught/growing near the beard edge.\n\nHer eyes brighten.\n\nBuzz immediately recognizes the expression.\n\nBuzz:\n<d>[English] Mimi...</d>\n\nMimi gently plucks/picks up the loose flower.\n\nDialogue:\n\nMimi:\n<d>[English] Just one.</d>\n\nBuzz narrows his eyes skeptically.\n\nMimi carefully tucks the flower between two neat beard curls.\n\nShe adjusts it once.\n\nSteps backward.\n\nThe tiny flower looks surprisingly beautiful against Brum’s huge brown beard.\n\nA tiny warm sparkle of sunlight catches its petals — natural light, not magical transformation.\n\nMimi beams.\n\nDialogue:\n\nMimi:\n<d>[English] Perfect.</d>\n\nCamera:\nflower insert\n→ Mimi’s idea expression\n→ gentle placement\n→ close beauty shot of single flower\n→ Mimi/Buzz reaction.\n\nSound:\n- tiny petal rustle\n- beard fibers\n- soft musical “ting”\n\n--------------------------------------------------\n\n[SHOT 8] — 13.30s–15.00s\nTHE MUCH BIGGER IDEA — CLIFFHANGER\n\nMimi admires the ONE flower.\n\nThen slowly turns.\n\nCamera follows her eyes toward her woven basket.\n\nThe basket is overflowing with:\n\n- pink flowers\n- purple flowers\n- white daisies\n- yellow blossoms\n- leafy stems\n\nMimi’s eyes widen.\n\nHer mouth slowly curls into an enormous creative grin.\n\nA metaphorical tiny “idea” sparkle may appear in her eye, but no literal floating lightbulb is necessary.\n\nBuzz looks at Mimi.\n\nThen the flower.\n\nThen the enormous basket.\n\nThen Brum’s gigantic beard.\n\nBuzz’s expression becomes pure dread.\n\nBuzz:\n<d>[English] Mimi... no.</d>\n\nMimi slowly reaches toward the basket.\n\nMimi, delighted whisper:\n<d>[English] Maybe... a few more?</d>\n\nBuzz:\n<d>[English] NO.</d>\n\nBrum snores peacefully in the background.\n\nThe single flower gently trembles from his breath.\n\nCamera:\nclose Mimi\n→ POV basket\n→ slow push into Mimi’s inspired grin\n→ Buzz reaction\n→ final wide shot showing tiny Mimi + full basket + enormous beard + sleeping Brum.\n\nCUT TO BLACK.\n\nEND PART 1.\n\nDo NOT begin decorating with additional flowers.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[MIMI VOICE]\n\nNatural young girl voice.\n\nQualities:\n- bright\n- confident\n- warm\n- creative\n- playful\n- slightly mischievous\n- clear English\n- not overly babyish\n\nHer delivery should often become a whisper because Brum is asleep.\n\nPerformance progression:\n\n0–3 sec:\nsurprised curiosity\n\n3–6 sec:\nprofessional determination\n\n6–11 sec:\nfocused whisper-comedy\n\n11–13 sec:\ngentle excitement\n\n13–15 sec:\ndangerous creative enthusiasm\n\n--------------------------------------------------\n\n[BUZZ VOICE]\n\nTiny expressive bee voice.\n\nQualities:\n- slightly higher pitch\n- quick\n- cautious\n- dry comic timing\n- understandable English\n- small buzzing undertone without harming intelligibility\n\nBuzz should sound like the sensible friend who already knows where this is going.\n\nFinal:\n\n<d>[English] NO.</d>\n\nshould land as the comedy button.\n\n--------------------------------------------------\n\n[BRUM AUDIO]\n\nBrum does NOT have normal dialogue in Part 1.\n\nUse only:\n- deep breathing\n- relaxed snoring\n- small satisfied sleepy exhale\n- perhaps a gentle sleepy hum while Mimi combs him\n\nSnoring should sound huge but cozy.\n\nNot monstrous.\n\n--------------------------------------------------\n\n[SQUIRREL AUDIO]\n\nNo full dialogue required.\n\nUse:\n- tiny squeak\n- quiet “shh” gesture sound\n- acorn movement\n- soft rustling\n\n--------------------------------------------------\n\n[FOREST ENVIRONMENT AUDIO]\n\nMaintain:\n- gentle birds\n- distant woodpecker/insects if subtle\n- leaves rustling\n- warm breeze\n- tiny petals falling\n- grass\n- mushrooms/forest floor movement\n- occasional distant woodland ambience\n\nUse strong scale contrast in audio:\n\nBrum’s breathing = huge low soft layer.\n\nMimi/Buzz = tiny close-detail sounds.\n\n--------------------------------------------------\n\n[SOUND EFFECT PRIORITIES]\n\nImportant synchronized effects:\n\nBrum snore:\ndeep warm “SNORRRR”\n\nsnore gust:\nsoft rush of air\n\nMimi climbing:\ntiny fabric/boot sounds\n\nBuzz:\nwing buzz changes with emotion\n\ncomb:\ndelicate “scritch-scritch”\n\ntwig:\ntiny creak then pop\n\nsquirrel reveal:\nsoft rustle + squeak\n\nflower placement:\nsmall petal rustle\n\nbasket:\nlight wicker movement\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nCreate an ORIGINAL whimsical fantasy-comedy score.\n\nInstrumentation:\n- pizzicato strings\n- soft flute\n- bassoon / clarinet for giant comedy\n- light marimba\n- glockenspiel\n- subtle acoustic plucks\n- warm strings\n- tiny bell accents\n\nMUSIC ARC:\n\n0.00–1.80\nmagical forest wonder.\n\n1.80–3.30\ngentle giant-scale comedy.\n\n3.30–5.00\ncurious “problem discovered” motif.\n\n5.00–6.80\nlight sneaking rhythm.\n\n6.80–9.20\nplayful grooming rhythm.\n\n9.20–11.40\ntiny suspense → cute reveal.\n\n11.40–13.30\nwarm charming flower motif.\n\n13.30–15.00\nmischievous “big idea” buildup ending with a short comedic sting after Buzz says “NO.”\n\nNo epic score.\nNo danger music.\nNo horror.\nNo dramatic action trailer music.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nThe visual hook depends heavily on scale.\n\nCAMERA PROGRESSION:\n\n0–3 sec:\nmajestic forest wides + tiny character scale reveal.\n\n3–6 sec:\nproblem-focused medium and close coverage.\n\n6–9 sec:\nminiature adventure perspective.\n\n9–12 sec:\nintimate beard-level comedy.\n\n12–15 sec:\nreaction-driven creative cliffhanger.\n\nPreferred camera techniques:\n- slow cinematic dolly\n- environmental reveal\n- tilt from Mimi to giant\n- macro beard details\n- shallow parallax through beard curls\n- over-the-shoulder views\n- low miniature POV\n- expressive reaction close-ups\n- slow final push-in\n\nAVOID:\n- handheld shaking\n- crash zooms\n- hyperfast cutting\n- Dutch angles\n- 360-degree spins\n- action-film camera behavior\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animation.\n\nMatch the storyboard artwork closely:\n\n- delicate ink-like outlines\n- rich hand-painted backgrounds\n- watercolor/gouache-style color texture\n- warm illustrated storybook aesthetic\n- cinematic dappled lighting\n- expressive faces\n- slightly rounded character designs\n- beautifully textured beard\n- lush forest foliage\n- clear silhouette staging\n- subtle squash and stretch\n- detailed but readable environment\n\nThe finished video should feel like the storyboard illustrations have come alive.\n\nNOT:\n- storyboard animation\n- motion comic\n- photorealistic film\n- 3D CGI\n- vector animation\n- anime\n- cel-shaded 3D\n- slideshow\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nPrimary palette:\n- forest green\n- moss green\n- warm brown\n- chestnut\n- golden yellow\n- mint\n- muted turquoise\n- flower pink\n- lavender\n- cream\n\nLighting:\nwarm late-morning / magical golden forest sunlight.\n\nUse:\n- leaf-shaped shadows\n- warm rim light\n- subtle volumetric rays\n- soft highlights on Brum’s beard\n- warm bounce on Mimi’s face\n- light catches on Buzz’s wings\n\nThe single flower at the end may receive a tiny natural golden highlight.\n\nNo artificial magical glow explosion.\n\n--------------------------------------------------\n\n[MOTION RULES — MIMI]\n\nMimi moves like a tiny energetic person in an enormous environment.\n\nUse:\n- small quick steps\n- careful climbing\n- expressive hands\n- hair bounce\n- apron flutter\n- natural effort pulling twig\n- gentle combing\n- subtle balance shifts on beard curls\n\nDo not:\n- make her weightless\n- make her fly\n- use superhero jumps\n\n--------------------------------------------------\n\n[MOTION RULES — BRUM]\n\nBrum remains mostly asleep and heavy.\n\nUse:\n- slow chest rise/fall\n- beard slightly moving with breath\n- nostrils subtly moving during snores\n- tiny sleepy smile during grooming\n- fingers completely relaxed\n\nNo major body movement.\n\nNo waking.\n\n--------------------------------------------------\n\n[MOTION RULES — BUZZ]\n\nUse:\n- natural bee hovering\n- rapid wings\n- small inertia when changing direction\n- blown sideways by snore\n- hovering at Mimi’s shoulder height\n- exaggerated eyebrow/eye reactions\n\n--------------------------------------------------\n\n[MOTION RULES — BEARD]\n\nThe beard must feel soft and dimensional.\n\nUse:\n- slight compression beneath Mimi\n- springy curl response\n- comb separating hair\n- gentle settling after brushing\n- individual loose hairs reacting to Brum’s breath\n\nDo not make beard behave like rigid branches.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nKeep exactly:\n- one Mimi\n- one Buzz\n- one Brum\n- one squirrel\n- same ancient tree\n- same forest clearing\n- same beard shape\n- same vest\n- same buttons\n- same basket\n- same comb\n- same outfit\n- same scale\n\nMimi begins on the forest floor.\n\nShe climbs Brum before reaching beard.\n\nDo not teleport her.\n\nThe squirrel appears only after the twig is removed.\n\nAfter squirrel exits, do not duplicate it elsewhere.\n\nOnly ONE flower is placed in Part 1.\n\nAt final frame:\nbasket remains full.\n\nBrum’s eyes remain closed.\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2085987791587135909",
    "title": "小园丁与出逃的雨云 · 第三幕",
    "titleEn": "The Little Gardener and the Runaway Rain Cloud · Chapter 3",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《小园丁与出逃的雨云 · 第三幕》的 H3 成片。本页保留原始 X 来源链接。来源未公开完整 Prompt，本库不反推或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Little Gardener and the Runaway Rain Cloud · Chapter 3. This page retains the original X source link. The source did not publish a complete Prompt, so this library does not infer or complete one. Hosted-video output: 15s · 1280×720.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2085987791587135909",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-08T07:13:39.950Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2085987791587135909.mp4",
    "posterUrl": "/posters/x/2085987791587135909.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "nature",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Third Storyboard Image to video using MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 3 storyboard image for “The Little Gardener and the Runaway Rain Cloud” as the PRIMARY visual, character, environment, emotional-performance, weather-effect, and narrative reference.\n\nThis sequence begins IMMEDIATELY after PART 2.\n\nPART 2 ended with:\n- Nimbus slightly larger and fluffier\n- Nimbus still wearing his flower crown\n- several gentle rain showers active around the greenhouse\n- Luma realizing the experiment may be going too far\n- Sprout soaked and panicking\n\nGenerate a FINISHED animated short sequence from the storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final animation must NOT contain:\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character reference drawings\n- camera notes\n- storyboard paper texture\n- footer decorations\n- title/header text\n\nThe uploaded storyboard locks:\n\n1. Luma’s exact Part 3 appearance\n2. Nimbus’s exact cloud identity\n3. Nimbus’s flower crown\n4. Sprout’s exact snail appearance\n5. greenhouse layout\n6. wet greenhouse continuity from Part 2\n7. shallow puddles\n8. multiple localized rain showers\n9. Sprout’s leaf-raft comedy\n10. Nimbus’s gradual emotional exhaustion\n11. Nimbus’s blue-white → soft blue-gray emotional color change\n12. the complete rain shutdown\n13. the quiet greenhouse beat\n14. Luma’s emotional realization\n15. final line: “You don’t have to rain for everyone.”\n\nAnimate the storyboard as one coherent cinematic sequence rather than 15 isolated panels.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration: exactly 15 seconds\n\nStory time:\n0:30–0:45 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual medium:\npremium painterly 2D animated film\n\nGenre:\nmagical garden comedy transitioning into heartfelt emotional storytelling\n\nDialogue:\nnative synchronized English dialogue\n\nAudio:\nnative synchronized voices + environmental sounds + rain + music\n\nEMOTIONAL ARC:\n\nrain comedy\n→ playful overload\n→ Nimbus tries even harder\n→ visual chaos peaks\n→ Luma notices Nimbus struggling\n→ Nimbus runs out of energy\n→ all rain stops\n→ emotional silence\n→ Nimbus withdraws\n→ Luma understands\n→ compassionate cliffhanger\n\nIMPORTANT:\n\nThe uploaded storyboard contains 15 one-second panels.\n\nDO NOT create 15 frantic hard cuts.\n\nTranslate them into approximately 8 coherent cinematic shots.\n\nAllow actions to flow naturally:\n\ncause\n→ action\n→ consequence\n→ reaction\n→ emotional change.\n\nThe first half should feel relatively quick and funny.\n\nThe second half must deliberately slow down.\n\n--------------------------------------------------\n\n[PART 3 STORY LIMIT]\n\nPART 3 ONLY.\n\nSTART:\n\nThe Super Cloud experiment is already working too well.\n\nEND:\n\nLuma sits beside a curled-up Nimbus and gently says:\n\n<d>[English] You don’t have to rain for everyone.</d>\n\nA SINGLE tiny raindrop begins forming beneath Nimbus.\n\nEND BEFORE THAT DROP FALLS.\n\nDO NOT show:\n- the final raindrop hitting the sunflower\n- sunflower fully blooming\n- Nimbus recovering emotionally\n- controlled one-plant watering\n- glowing mushrooms\n- magical greenhouse transformation\n- miniature rainbows\n- rainbow-drop charm\n- Nimbus sitting on Luma’s hat\n- final hero ending\n\nThose are PART 4.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — LUMA]\n\nMatch Parts 1–2 and the uploaded PART 3 storyboard exactly.\n\nLUMA:\n\napproximately 8-year-old little gardener.\n\nAppearance:\n- warm fair skin\n- soft rounded childlike face\n- large expressive dark-brown eyes\n- subtle freckles\n- fluffy auburn / reddish-brown hair\n- two messy buns\n- loose curls framing her cheeks\n- oversized yellow gardening hat\n\nClothing:\n- mint-green gardening overalls\n- coral-and-cream striped shirt\n- red gardening boots\n- small brown seed pouch attached at waist\n\nProportions:\n- child height\n- slightly oversized head\n- small expressive hands\n- appealing rounded silhouette\n\nPART 3 PERFORMANCE ARC:\n\n0–4 sec:\nbusy but still amused\n\n4–7 sec:\nexcited, encouraging Nimbus\n\n7–10 sec:\ncomedy turns into concern\n\n10–12 sec:\nworried realization\n\n12–15 sec:\nguilty, calm, compassionate\n\nLuma must not appear angry at Nimbus.\n\nShe gradually realizes that her encouragement caused him to push himself too hard.\n\nHer final emotion should communicate:\n\n“I finally understand.”\n\nCRITICAL:\n\nMaintain exact:\n- face\n- hairstyle\n- hair color\n- yellow hat\n- overalls\n- shirt\n- boots\n- pouch\n- height\n- age\n- proportions\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NIMBUS]\n\nNimbus must match Parts 1–2 exactly.\n\nSTARTING APPEARANCE:\n\n- fluffy blue-white living cloud\n- approximately 25–35% larger than his original Part 1 size\n- large expressive eyes\n- tiny eyebrows\n- cute small mouth\n- soft watercolor cloud edges\n- pale blue shading\n- tiny dangling raindrop-shaped feet\n- small handmade flower crown from Part 2\n\nFLOWER CROWN:\n- pink flowers\n- yellow flowers\n- tiny pale flowers\n- green leaves\n\nThe crown remains throughout PART 3.\n\nAs Nimbus becomes exhausted, it can tilt slightly.\n\nPERSONALITY:\n\n- sweet\n- helpful\n- eager to please\n- sensitive\n- determined to make Luma happy\n- incapable of saying when something is too much\n\nNimbus is NOT angry.\n\nEven when he “tries harder,” any determined expression should read as effort, NOT aggression.\n\n--------------------------------------------------\n\n[NIMBUS EMOTIONAL TRANSFORMATION]\n\nNimbus changes gradually through FOUR states.\n\nSTATE 1 — SUPER CLOUD\n\n0:00–0:05\n\n- bright blue-white\n- fluffy\n- flower crown upright\n- happy\n- energetic\n- producing multiple gentle showers\n\nSTATE 2 — TRYING HARDER\n\n0:05–0:08\n\n- slightly puffier\n- concentrating\n- subtle strain around eyes\n- rain remains active\n- still blue-white\n\nSTATE 3 — EXHAUSTED\n\n0:08–0:11\n\n- cloud edges begin drooping\n- eyes become heavy\n- flower crown slightly crooked\n- color becomes paler / slightly desaturated\n- rain becomes weaker\n- tiny rain feet hang lower\n- breathing/cloud pulses slower\n\nSTATE 4 — CURLED-UP NIMBUS\n\n0:11–0:15\n\n- slightly smaller\n- compact curled shape\n- muted soft blue-gray\n- tired sad eyes\n- crooked flower crown\n- no active rain\n- extremely gentle appearance\n\nIMPORTANT:\n\nThe gray-blue color represents EMOTIONAL EXHAUSTION ONLY.\n\nDO NOT make Nimbus:\n- black\n- dark charcoal\n- monstrous\n- evil\n- electrically charged\n- threatening\n- angry\n- storm-like\n\nNo lightning.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — SPROUT]\n\nSame Sprout from Parts 1–2.\n\nAppearance:\n- tiny green snail\n- tall expressive eye stalks\n- warm brown spiral shell\n- tiny yellow flower growing near/on shell\n- same miniature proportions\n\nPART 3 PERFORMANCE ARC:\n\npanic\n→ splash victim\n→ accidental leaf sailor\n→ dizzy / offended\n→ notices Nimbus’s exhaustion\n→ becomes quietly concerned\n\nSprout is the primary comedy element during the first half.\n\nDuring the emotional ending, Sprout becomes silent.\n\nNo normal conversational human dialogue needed.\n\nUse:\n- squeaks\n- gasps\n- little groans\n- startled “eep”\n- subtle worried chirp\n\n--------------------------------------------------\n\n[LEAF RAFT LOCK]\n\nSprout’s accidental raft:\n\n- broad fresh green garden leaf\n- curved slightly upward at edges\n- natural visible leaf veins\n- only large enough to support tiny Sprout\n- floats on a VERY shallow greenhouse puddle\n\nThis is visual comedy.\n\nThe puddle is only a few centimeters deep.\n\nNo dangerous water.\n\n--------------------------------------------------\n\n[GREENHOUSE CONTINUITY LOCK]\n\nSame rooftop greenhouse from Parts 1–2.\n\nMaintain:\n\n- glass ceiling\n- wooden support frame\n- planter boxes\n- flower pots\n- hanging vines\n- sunflower\n- plant benches\n- old metal watering can\n- gardening shelves\n- seed packets\n- tools\n- little ladder\n- distant rooftop city view\n- warm daylight direction\n\nPART 3 starts wet.\n\nShow continuity from Part 2:\n\n- droplets on leaves\n- wet wooden/stone floor\n- shallow reflective puddles\n- dripping pots\n- water beads on greenhouse glass\n- several localized rain zones\n\nDo NOT reset the greenhouse to dry.\n\nDo NOT deeply flood the greenhouse.\n\n--------------------------------------------------\n\n[RAIN EFFECT SCALE]\n\nRain should move through these levels:\n\nLEVEL 3 at beginning:\nseveral localized plant showers.\n\nLEVEL 4 around comedy peak:\nmany small plant-sized showers simultaneously.\n\nThen gradually decline:\n\nweaker showers\n→ scattered drops\n→ final droplets\n→ complete stop.\n\nThe rain must never become:\n- torrential\n- dangerous\n- destructive\n- storm-force\n\n--------------------------------------------------\n\n[PHYSICS / STORY CAUSE-AND-EFFECT]\n\nClear logic:\n\nLuma praised Nimbus\n→ Nimbus believes she wants more.\n\nNimbus sees plants responding positively\n→ he increases rain.\n\nMore rain\n→ puddles form.\n\nA large harmless splash lifts a broad leaf\n→ Sprout accidentally rides it.\n\nLuma continues giving encouraging signals\n→ Nimbus pushes himself harder.\n\nNimbus uses too much energy\n→ his rain weakens.\n\nLuma notices his body language\n→ her excitement disappears.\n\nNimbus attempts one final shower\n→ cannot sustain it.\n\nNimbus runs out of energy\n→ all rain stops.\n\nNimbus withdraws\n→ Luma realizes he was trying to please her.\n\nLuma stops asking for rain\n→ sits beside Nimbus instead.\n\nEmotional lesson begins.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.70s\nTHE EXPERIMENT IS WORKING TOO WELL\n\nContinue directly from Part 2.\n\nOpen on a wide shot of the sunny-but-rainy rooftop greenhouse.\n\nNimbus floats in the center of the greenhouse.\n\nHe is:\n- fluffy\n- blue-white\n- wearing his flower crown\n- still cute and eager\n\nSeveral localized rain showers fall beneath him.\n\nWater sparkles in the sunlight.\n\nSmall puddles reflect greenhouse beams.\n\nLuma runs below him holding TWO small gardening buckets.\n\nShe puts one bucket beneath a shower...\n\nthen sees another shower start beside her.\n\nShe quickly moves the second bucket.\n\nThen another rain zone activates behind her.\n\nHer expression is still playful.\n\nSprout moves quickly by snail standards, trying to avoid droplets.\n\nCamera:\n\nstart wide\n→ smooth lateral tracking with Luma\n→ small push toward Nimbus\n→ insert on buckets filling with rain.\n\nDialogue:\n\nLuma, breathlessly amused:\n<d>[English] Okay! That is definitely more rain!</d>\n\nNimbus happily:\n<d>[English] I’m helping!</d>\n\nLuma:\n<d>[English] You sure are!</d>\n\nSound:\n- multiple soft rain zones\n- bucket droplets\n- splashing\n- wet footsteps\n- plant leaves\n- Nimbus cloud hum\n\nMusic:\nenergetic playful pizzicato continuation from Part 2.\n\n--------------------------------------------------\n\n[SHOT 2] — 1.70s–3.40s\nSPROUT DISCOVERS WATER TRANSPORTATION\n\nCut low to Sprout.\n\nHe carefully crawls across the wet floor.\n\nA large but harmless raindrop falls beside him.\n\nSPLASH!\n\nThe splash slips underneath a broad nearby leaf.\n\nThe leaf lifts slightly.\n\nSprout accidentally slides onto it.\n\nThe shallow puddle begins carrying the leaf slowly across the greenhouse floor.\n\nSprout freezes.\n\nHis eye stalks stretch fully upward.\n\nHe looks left.\n\nThen right.\n\nThen directly toward camera.\n\nSprout:\n\n<d>[English] Eep?</d>\n\nCamera:\n\nmacro low-angle\n→ slow follow alongside leaf raft\n→ Sprout reaction close-up.\n\nSound:\n- single comic splash\n- tiny leaf sliding sound\n- water ripple\n- Sprout squeak\n\nDo NOT make the raft fast or dangerous.\n\n--------------------------------------------------\n\n[SHOT 3] — 3.40s–5.20s\nNIMBUS THINKS LUMA WANTS MORE\n\nLuma notices several flowers standing healthier beneath Nimbus’s rain.\n\nShe looks up at Nimbus.\n\nShe smiles and gives him a huge enthusiastic thumbs-up.\n\nDialogue:\n\nLuma:\n<d>[English] Look! They love it!</d>\n\nNimbus looks at the flowers.\n\nThen at Luma.\n\nHis expression brightens.\n\nNimbus:\n<d>[English] Really?</d>\n\nLuma:\n<d>[English] Really!</d>\n\nNimbus interprets this as:\n“keep going.”\n\nHe takes a tiny determined breath.\n\nHis cloud body expands subtly.\n\nThe flower crown wobbles.\n\nNimbus:\n<d>[English] I can do more!</d>\n\nCamera:\nplant detail\n→ Luma thumbs-up\n→ Nimbus reaction\n→ gentle push-in on Nimbus preparing.\n\nMusic:\nplayful rising motif.\n\n--------------------------------------------------\n\n[SHOT 4] — 5.20s–7.40s\nTHE GREENHOUSE GETS ITS OWN WEATHER\n\nNimbus compresses slightly...\n\nthen puffs outward.\n\nMore small showers appear.\n\nNOT one massive rain curtain.\n\nInstead:\nmultiple independent plant-sized showers.\n\nShow an energetic continuous camera move through the greenhouse:\n\n- flowers bouncing under droplets\n- hanging leaves dripping\n- vines swaying gently\n- water streaks on glass\n- a mushroom opens like a tiny umbrella\n- Luma runs through frame with one metal bucket upside-down over her yellow gardening hat like a silly umbrella\n\nLuma laughs.\n\nSprout continues floating on his leaf raft in the background.\n\nHis expression says he deeply regrets everything.\n\nDialogue:\n\nLuma, laughing:\n<d>[English] We made our own weather!</d>\n\nNimbus laughs softly.\n\nSprout:\n<d>[English] EEEP!</d>\n\nSound:\n- gentle rain\n- leaves\n- tiny puddle splashes\n- bucket tapping\n- cute mushroom pop\n- greenhouse ambience\n\nCamera:\nsmooth moving wide shot with layers of action.\n\nDo not make cuts too fast.\n\n--------------------------------------------------\n\n[SHOT 5] — 7.40s–8.70s\nSPROUT WOULD LIKE TO FILE A COMPLAINT\n\nSprout’s leaf raft gently reaches the edge of the puddle.\n\nIt softly bumps into a terracotta flower pot.\n\nBOOP.\n\nThe raft stops.\n\nSprout rocks backward slightly.\n\nTiny dizzy stars or subtle cartoon swirl may briefly appear around his eyes if consistent with painterly style.\n\nHe looks:\nwet,\ndizzy,\ndeeply offended.\n\nA final droplet lands directly on his flower.\n\nSprout slowly looks upward.\n\nSprout gives one exhausted squeak.\n\nCamera:\nlow close-up\n→ gentle bump\n→ reaction hold.\n\nKeep this comedy beat short.\n\n--------------------------------------------------\n\n[SHOT 6] — 8.70s–10.70s\nLUMA SEES NIMBUS STRUGGLING\n\nThe music begins reducing in energy.\n\nCut to Nimbus.\n\nSomething is different.\n\nHis cloud motion is slower.\n\nHis smile weakens.\n\nHis edges begin drooping.\n\nThe flower crown tilts slightly.\n\nHis eyes look heavy.\n\nRaindrops underneath him become less regular.\n\nInstead of smooth showers:\n\ndrip...\n\ndrip...\n\npause...\n\nNimbus tries to puff himself up again.\n\nHis body expands—\n\nbut immediately loses some volume.\n\nLuma stops running.\n\nHer smile disappears.\n\nShe lowers the bucket from above her head.\n\nDialogue:\n\nLuma:\n<d>[English] Nimbus?</d>\n\nNimbus forces a small smile.\n\nNimbus:\n<d>[English] I can keep going.</d>\n\nLuma watches more closely.\n\nNimbus tries again.\n\nOnly a few drops fall.\n\nLuma, now genuinely concerned:\n<d>[English] You look tired.</d>\n\nNimbus:\n<d>[English] I’m... helping.</d>\n\nCamera:\nclose-up Nimbus\n→ rack focus to Luma\n→ intimate two-character composition.\n\nSound:\nrain layers begin fading.\n\nMusic:\nplayful rhythm dissolves into a soft emotional pad/piano.\n\n--------------------------------------------------\n\n[SHOT 7] — 10.70s–12.20s\nTHE RAIN STOPS\n\nNimbus squeezes his eyes shut.\n\nHe tries one final time.\n\nHis cloud body puffs.\n\nFor one brief moment several droplets form.\n\nThen—\n\nthe fluff collapses gently inward.\n\nThe droplets stop.\n\nEvery localized rain shower throughout the greenhouse ends almost simultaneously.\n\nImportant audio transition:\n\nrain\n→ fewer drops\n→ one final drip\n→ SILENCE.\n\nOnly droplets already hanging from leaves continue falling naturally.\n\nPLINK.\n\nPLINK.\n\nNimbus becomes visibly smaller.\n\nHis color shifts subtly:\n\nbright blue-white\n→ pale blue\n→ muted soft blue-gray.\n\nLuma stands completely still.\n\nSprout becomes quiet.\n\nCamera:\n\nmedium Nimbus\n→ wide greenhouse shot as rain stops\n→ very slow push forward.\n\nDialogue:\n\nLuma, soft:\n<d>[English] Nimbus...</d>\n\nNo response.\n\nSound:\n- rain fading\n- final isolated droplets\n- room tone\n- soft distant rooftop wind\n\nThis should feel dramatically different from the first half.\n\n--------------------------------------------------\n\n[SHOT 8] — 12.20s–13.50s\nTHE QUIET GREENHOUSE\n\nWide still composition.\n\nNo active rain.\n\nMorning sunlight passes through wet greenhouse glass.\n\nThe floor contains small reflective puddles.\n\nLeaves sparkle with water.\n\nLuma stands quietly in the center.\n\nSprout has stepped off his leaf raft.\n\nBoth watch Nimbus.\n\nNimbus slowly drifts toward a quiet wooden shelf / planter corner surrounded by hanging vines.\n\nHis posture collapses.\n\nHe curls his cloud body inward.\n\nThe flower crown is crooked.\n\nHis eyes look down.\n\nHe settles into a compact sad little blue-gray cloud.\n\nNo dialogue for most of this shot.\n\nMusic:\nalmost nothing.\n\nPerhaps one soft piano note.\n\nCamera:\nvery slow tracking move following Nimbus\n→ stop at a respectful distance.\n\n--------------------------------------------------\n\n[SHOT 9] — 13.50s–15.00s\n“YOU DON’T HAVE TO RAIN FOR EVERYONE.”\n\nLuma gently removes her bucket and sets it aside.\n\nShe approaches Nimbus slowly.\n\nDo NOT have her immediately grab him.\n\nShe kneels/sits beside him at his eye level.\n\nSprout quietly approaches behind Luma.\n\nLuma looks at:\nNimbus,\nhis crooked flower crown,\nhis tired eyes.\n\nShe realizes what happened.\n\nHer face shows:\nguilt,\nunderstanding,\ncare.\n\nLuma speaks softly:\n\n<d>[English] I’m sorry, Nimbus.</d>\n\nNimbus slowly raises his eyes toward her.\n\nLuma:\n\n<d>[English] You don’t have to rain for everyone.</d>\n\nNimbus blinks.\n\nA SINGLE tiny raindrop begins forming beneath his cloud body.\n\nThis is NOT rain.\n\nIt is one tiny emotional magical drop.\n\nA subtle rainbow glint catches inside it.\n\nDO NOT let it fall.\n\nSprout watches quietly.\n\nCamera:\n\neye-level two-shot Luma + Nimbus\n→ subtle push-in\n→ close emotional view\n→ tiny raindrop forming at edge of frame.\n\nNo dialogue after Luma’s line.\n\nHold.\n\nCUT TO BLACK.\n\nEND PART 3.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[LUMA VOICE]\n\nNatural 8-year-old girl voice.\n\nVoice qualities:\n- warm\n- expressive\n- caring\n- playful early\n- sincere later\n- not overly cartoonish\n\nPERFORMANCE ARC:\n\n0–5 sec:\nbusy, excited, laughing.\n\n5–8 sec:\nhigh-energy garden comedy.\n\n8–11 sec:\nvoice becomes quieter and concerned.\n\n11–15 sec:\nsoft, compassionate, emotionally sincere.\n\nFINAL LINE IS VERY IMPORTANT:\n\n<d>[English] You don’t have to rain for everyone.</d>\n\nDeliver it gently.\n\nNo dramatic crying.\n\nNo melodramatic performance.\n\n--------------------------------------------------\n\n[NIMBUS VOICE]\n\nSoft youthful cloud voice.\n\nQualities:\n- shy\n- breathy\n- sweet\n- innocent\n- slightly high pitched\n\nBeginning:\nenergetic and eager.\n\nMiddle:\nincreasing effort in breathing.\n\nNear exhaustion:\nspeech becomes quieter,\nslower,\nless energetic.\n\nHis line:\n\n<d>[English] I’m... helping.</d>\n\nshould feel like Nimbus believes continuing is the right thing to do.\n\nDo not make him sound physically injured.\n\nHe is emotionally and magically exhausted.\n\n--------------------------------------------------\n\n[SPROUT AUDIO]\n\nEarly:\ncomic squeaks and “eep.”\n\nLate:\nquiet worried chirps only.\n\nAfter the rain stops:\nSprout should mostly remain silent.\n\n--------------------------------------------------\n\n[RAIN AUDIO PROGRESSION]\n\n0.00–5.20:\nseveral gentle rain zones.\n\n5.20–7.40:\npeak playful rain activity.\n\n7.40–8.70:\nstill busy.\n\n8.70–10.70:\nrain begins weakening.\n\n10.70–11.50:\nrain rapidly diminishes.\n\n11.50 onward:\nNO ACTIVE RAIN.\n\nOnly:\n- isolated leaf droplets\n- small puddle ripples\n- room tone\n\nThis sonic contrast is essential to the emotional transition.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nMaintain:\n- greenhouse ambience\n- distant rooftop birds\n- faint city atmosphere\n- water droplets\n- puddle splashes\n- leaves\n- glass\n- wooden planter sounds\n- Luma footsteps\n- bucket movements\n- Nimbus airy cloud motion\n- Sprout squeaks\n\nAfter rain stops:\n\ndramatically simplify environment audio.\n\nUse:\n- gentle wind outside\n- isolated water drip\n- faint leaf rustle\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nORIGINAL SCORE ONLY.\n\nInstrumentation:\n\nFIRST HALF:\n- pizzicato strings\n- marimba\n- light clarinet\n- flute\n- wooden percussion\n- tiny bell accents\n\nSECOND HALF:\n- soft piano\n- warm string pad\n- very gentle glockenspiel\n- subtle airy synth\n\nMUSIC ARC:\n\n0.00–3.40\ncontinuation of playful chaos.\n\n3.40–7.40\nfast whimsical garden-comedy rhythm.\n\n7.40–8.70\nbrief comic sting for Sprout.\n\n8.70–10.70\nrhythm gradually disappears.\n\n10.70–12.20\nmusic nearly stops.\n\n12.20–13.50\nvery sparse emotional piano.\n\n13.50–15.00\nsoft warm emotional harmony beneath Luma’s final line.\n\nDo NOT:\n- use sad melodramatic violin\n- use epic orchestration\n- make moment tragic\n- use horror music\n\nThe feeling is:\nunderstanding, not tragedy.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nCAMERA ARC MUST SUPPORT THE EMOTIONAL SHIFT.\n\nFIRST HALF:\nmore energetic.\n\nUse:\n- medium tracking\n- low-angle comedy\n- wider rain-zone shots\n- close reaction inserts\n- gentle lateral movement\n\nSECOND HALF:\nincreasing stillness.\n\nAfter 8.7 sec:\n- reduce camera speed\n- reduce cutting\n- favor close emotional compositions\n\nAfter rain stops:\ncamera should feel almost completely calm.\n\nFINAL:\nslow intimate push toward Luma and Nimbus.\n\nPreferred techniques:\n- smooth dolly\n- gentle pan\n- rack focus\n- macro water detail\n- low Sprout POV\n- character eye-level close-up\n- wide environmental contrast shot\n\nAvoid:\n- handheld shake\n- crash zoom\n- extreme Dutch angle\n- rapid 360 motion\n- excessive cuts\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animation.\n\nMATCH PARTS 1–2 EXACTLY.\n\nUse:\n- delicate ink contour lines\n- watercolor-like color fills\n- warm storybook texturing\n- hand-painted greenhouse backgrounds\n- expressive faces\n- rounded silhouettes\n- subtle squash/stretch\n- cinematic sunlight\n- soft atmospheric depth\n- beautiful transparent water effects\n\nThe finished video should feel like a high-budget animated short created directly from the storyboard artwork.\n\nNOT:\n- 3D CGI\n- photorealistic\n- anime\n- flat vector\n- slideshow\n- motion comic\n- storyboard animatic\n\n--------------------------------------------------\n\n[COLOR / LIGHTING ARC]\n\n0.00–7.40:\n\nbright and playful.\n\nUse:\n- golden sunlight\n- fresh green\n- sunflower yellow\n- sky blue\n- coral\n- mint\n- sparkling rain highlights\n\nNimbus:\nbright blue-white.\n\n7.40–10.70:\n\nenvironment remains colorful.\n\nONLY Nimbus gradually loses saturation.\n\nDo NOT darken the entire scene.\n\nNimbus transition:\n\nblue-white\n→ pale blue\n→ pale desaturated blue.\n\n10.70–13.50:\n\nNimbus becomes muted soft blue-gray.\n\nEnvironment:\nstill warm.\n\nThis contrast communicates:\n“The world is okay. Nimbus is exhausted.”\n\n13.50–15:\n\nwarm sunlight softly illuminates Luma’s face.\n\nNimbus remains cool blue-gray.\n\nFinal tiny raindrop catches:\none extremely subtle rainbow glint.\n\n--------------------------------------------------\n\n[MOTION RULES]\n\nLUMA:\n\nFirst half:\n- quick childlike steps\n- moving buckets\n- playful running\n- expressive reactions\n\nSecond half:\n- slows down\n- smaller movements\n- careful approach\n- natural sitting/kneeling motion\n\nHair and clothing:\nsubtle secondary movement.\n\nYellow hat:\nmust remain physically consistent.\n\nNIMBUS:\n\nBeginning:\n- buoyant\n- soft bobbing\n- cloud edges active\n- lively compression/expansion\n\nMiddle:\n- more deliberate puffing\n- visible effort\n\nExhaustion:\n- slower bobbing\n- drooping cloud edges\n- smaller breathing pulses\n- gradual descent\n- slight shrink\n\nFinal:\nalmost still.\n\nSPROUT:\n\n- slow natural snail motion\n- sudden eye-stalk reactions\n- leaf raft moves gently with water\n- no high-speed sliding\n\nRAIN:\n\nFollow gravity.\n\nUse believable:\n- droplet trajectories\n- splashes\n- ripples\n- leaf runoff\n- reflective puddles\n\n--------------------------------------------------\n\n[EMOTIONAL PERFORMANCE RULES]\n\nThe emotional turning point must NOT feel like Nimbus has been physically hurt.\n\nThis story is about over-helping and exhaustion.\n\nNimbus:\nsad because he thinks he failed.\n\nLuma:\nguilty because she realizes he was trying to please her.\n\nSprout:\nquiet because he understands the mood has changed.\n\nAvoid exaggerated tears.\n\nThe single forming raindrop at the ending provides enough emotion.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nKeep the same:\n- Luma\n- Nimbus\n- Sprout\n- flower crown\n- yellow gardening hat\n- mint overalls\n- coral shirt\n- red boots\n- seed pouch\n- Sprout shell\n- Sprout yellow flower\n- greenhouse\n- sunflower\n- watering can\n- plant locations\n- sunlight direction\n\nFlower crown remains attached to Nimbus.\n\nNimbus’s color shift must be gradual.\n\nPuddles remain shallow.\n\nRain accumulation must match Part 2 continuity.\n\nOnce rain stops:\ndo not restart it.\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2085986990919647644",
    "title": "小园丁与出逃的雨云 · 第二幕",
    "titleEn": "The Little Gardener and the Runaway Rain Cloud · Chapter 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《小园丁与出逃的雨云 · 第二幕》的 H3 成片。本页保留原始 X 来源链接。来源未公开完整 Prompt，本库不反推或补写。站内视频规格为 15 秒 · 2560×1440。",
    "summaryEn": "An H3 output for The Little Gardener and the Runaway Rain Cloud · Chapter 2. This page retains the original X source link. The source did not publish a complete Prompt, so this library does not infer or complete one. Hosted-video output: 15s · 2560×1440.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2085986990919647644",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-08-08T07:10:29.056Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2085986990919647644.mp4",
    "posterUrl": "/posters/x/2085986990919647644.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "nature",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second Storyboard Image to video using @Hailuo_AI's MiniMax H3 on @higgsfield @higgsfield_ai \n\nPrompt : \n\n[REFERENCE CONTROL]\n\nUse the uploaded PART 2 storyboard image for “The Little Gardener and the Runaway Rain Cloud” as the PRIMARY visual, character, environment, prop, color, composition, and narrative reference.\n\nThis sequence begins IMMEDIATELY after PART 1.\n\nPART 1 ended with Luma excitedly telling Nimbus:\n\n“You can help ALL my flowers!”\n\nGenerate a FINISHED animated sequence from the storyboard.\n\nDO NOT reproduce the storyboard sheet itself.\n\nThe final animation must NOT contain:\n- storyboard borders\n- panel numbers\n- timestamps\n- handwritten captions\n- production notes\n- character reference sheets\n- prop reference diagrams\n- paper texture\n- storyboard footer\n- storyboard title\n\nThe uploaded storyboard locks:\n\n1. Luma’s exact character design\n2. Nimbus’s exact tiny-cloud identity\n3. Sprout’s exact snail design\n4. Nimbus’s gradual Part 2 growth\n5. flower crown design\n6. hand fan\n7. wind chimes\n8. leaf cup of water\n9. greenhouse environment\n10. plant arrangements\n11. old watering can\n12. warm painterly 2D animation aesthetic\n13. playful rain effects\n14. exact emotional progression\n15. Part 2 cliffhanger\n\nAnimate the storyboard as one continuous, polished animated-film sequence.\n\n--------------------------------------------------\n\n[VIDEO GOAL]\n\nDuration: exactly 15 seconds\n\nStory time:\n0:15–0:30 of the complete 1-minute film\n\nAspect ratio:\n16:9 landscape\n\nFrame-rate feeling:\ncinematic 24 fps\n\nVisual style:\npremium painterly 2D animated film\n\nGenre:\nwhimsical magical garden comedy\n\nTone progression:\n\nBIG IDEA\n→ excited experiment\n→ silly cloud makeover\n→ Nimbus gets fluffier\n→ first successful rain\n→ Luma asks for more\n→ Nimbus becomes more powerful\n→ multiple showers begin\n→ cute greenhouse chaos\n→ “Uh-oh” cliffhanger\n\nThe sequence should feel increasingly energetic without ever becoming dangerous.\n\nIMPORTANT:\n\nThe storyboard contains fifteen one-second panels.\n\nDO NOT animate them as fifteen hard one-second cuts.\n\nConsolidate the story into approximately 8 fluid cinematic shots.\n\nAllow actions to develop naturally:\n\ncause\n→ motion\n→ consequence\n→ character reaction.\n\n--------------------------------------------------\n\n[PART 2 STORY LIMIT]\n\nThis sequence ONLY covers:\n\n“Operation Super Cloud.”\n\nStart:\nLuma excitedly explains her big idea.\n\nEnd:\nNimbus is slightly larger and raining over several plants simultaneously while Luma finally realizes this may be too much.\n\nDO NOT continue into Part 3.\n\nDo NOT show:\n- Nimbus becoming exhausted\n- Nimbus turning gray\n- Nimbus emotionally shutting down\n- complete rain shutdown\n- leaf-raft sequence as the major Part 3 event\n- Luma apologizing\n- “You don’t have to rain for everyone”\n- magical final greenhouse transformation\n- rainbow-drop charm\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — LUMA]\n\nMatch PART 1 and the uploaded PART 2 storyboard exactly.\n\nLUMA:\n\nApproximately 8 years old.\n\nAppearance:\n- warm fair skin\n- soft round childlike face\n- large dark brown expressive eyes\n- light freckles if visible\n- fluffy auburn / reddish-brown hair\n- two messy hair buns\n- loose curly strands framing face\n- oversized warm yellow gardening hat\n\nClothing:\n- mint-green gardening overalls\n- coral-and-cream striped shirt\n- red gardening boots\n- small brown seed/tool pouch at waist\n\nProportions:\n- small child stature\n- slightly oversized head\n- rounded appealing silhouette\n- small expressive hands\n\nPersonality:\n- kind\n- imaginative\n- impulsively enthusiastic\n- inventive\n- convinced every successful idea should immediately become a bigger idea\n\nPART 2 PERFORMANCE ARC:\n\nthrilled\n→ confident\n→ playful\n→ proud\n→ increasingly amazed\n→ briefly oblivious to warning signs\n→ surprised\n→ concerned at the final moment\n\nLuma must genuinely believe she is helping Nimbus.\n\nShe should NEVER appear selfish or cruel.\n\nKeep her lovable and innocent.\n\nCRITICAL IDENTITY:\n\nDo not alter:\n- face\n- hairstyle\n- hair color\n- hat\n- shirt\n- overalls\n- boots\n- seed pouch\n- age\n- proportions\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — NIMBUS]\n\nMatch PART 1 exactly.\n\nNimbus begins approximately pillow-sized.\n\nAppearance:\n- fluffy blue-white cloud body\n- soft watercolor edges\n- pale blue shadowing\n- large expressive eyes\n- tiny eyebrows\n- small mouth\n- rosy cheek tint\n- tiny dangling raindrop feet\n- buoyant soft silhouette\n\nPersonality:\n- shy\n- eager to please\n- gentle\n- trusting\n- curious\n- slightly nervous\n\nPART 2 GROWTH CONTINUITY:\n\nNimbus can gradually become approximately 20–35% larger and fluffier over the sequence.\n\nGrowth must be gradual.\n\nStage 1:\ntiny Nimbus from Part 1.\n\nStage 2:\nflower-crown Nimbus.\n\nStage 3:\nslightly fluffier after hand fan.\n\nStage 4:\nrounder after absorbing water.\n\nStage 5:\npuffier Nimbus capable of several gentle showers.\n\nDO NOT turn Nimbus into a giant cloud.\n\nDO NOT make him fill the greenhouse.\n\nDO NOT make him dark or gray.\n\nDO NOT make him angry.\n\nDO NOT add lightning.\n\nHis final Part 2 appearance remains:\nbright blue-white,\ncute,\nhappy/eager,\nslightly overloaded by his rain output.\n\n--------------------------------------------------\n\n[CHARACTER IDENTITY LOCK — SPROUT]\n\nSame Sprout from Part 1.\n\nDesign:\n- tiny green snail\n- tall expressive eye stalks\n- brown spiral shell\n- tiny yellow flower attached near/on shell\n- miniature scale\n\nPersonality:\n- skeptical\n- cautious\n- dramatic\n- comedic early-warning system\n\nPerformance progression:\n\nsuspicious\n→ worried\n→ alarmed\n→ splashed\n→ full dramatic panic\n\nSprout understands that “Operation Super Cloud” is becoming excessive long before Luma does.\n\nSprout does not need normal spoken dialogue.\n\nUse:\n- squeaks\n- gasps\n- tiny distressed sounds\n- eye-stalk reactions\n\n--------------------------------------------------\n\n[PART 2 PROP LOCKS]\n\nFLOWER CROWN\n\nSmall handmade flower crown.\n\nUse:\n- tiny pink flowers\n- yellow flowers\n- pale white flowers\n- small green leaves\n\nIt sits gently on Nimbus.\n\nOnce placed:\nkeep it on Nimbus for the remainder of PART 2.\n\nDo not randomly remove it.\n\n--------------------------------------------------\n\nHAND FAN\n\nSmall handmade garden hand fan.\n\nLight cream/tan material.\n\nDecorative floral pattern.\n\nLuma uses it ONLY to blow a gentle breeze at Nimbus.\n\nNo extreme wind.\n\n--------------------------------------------------\n\nWIND CHIMES\n\nSmall colorful handmade wind chimes.\n\nUse:\n- slim pastel hanging tubes\n- tiny central charm\n- delicate garden design\n\nThey hang beside Nimbus and create a soft musical breeze.\n\nDo not attach them permanently to Nimbus.\n\n--------------------------------------------------\n\nLEAF WATER CUP\n\nSmall green leaf folded naturally into a tiny cup.\n\nContains:\nclean sparkling water droplets.\n\nLuma offers it to Nimbus.\n\nNimbus absorbs/sips the water gently.\n\nNo bottle.\nNo branded container.\n\n--------------------------------------------------\n\n[ENVIRONMENT LOCK]\n\nUse the SAME rooftop greenhouse from Part 1.\n\nMaintain:\n- glass ceiling\n- warm wooden structures\n- plant benches\n- planter boxes\n- hanging vines\n- potted plants\n- sunflower\n- gardening shelves\n- old watering can\n- seed supplies\n- gardening tools\n- distant city rooftop skyline\n- sunlight direction\n\nAt the beginning:\nthe floor is mostly dry.\n\nBy the ending:\nthere are several shallow puddles and wet reflective areas.\n\nNever deeply flood the greenhouse.\n\n--------------------------------------------------\n\n[WEATHER / RAIN RULES]\n\nRain magic must remain:\n\ncute,\nlocalized,\ngentle,\nwhimsical,\nfamily-friendly.\n\nUse approximately four visual rain levels.\n\nLEVEL 1:\na few droplets.\n\nLEVEL 2:\none tiny plant-sized shower.\n\nLEVEL 3:\ntwo or three localized showers.\n\nLEVEL 4:\nmultiple gentle plant showers throughout part of the greenhouse.\n\nNever show:\n- violent rain\n- heavy storm sheets\n- thunder\n- lightning\n- dangerous wind\n- flooding\n- broken pots from impact\n- destroyed flowers\n\nEvery increase must be motivated by Nimbus trying harder.\n\n--------------------------------------------------\n\n[CAUSE-AND-EFFECT LOGIC]\n\nLuma believes Nimbus can help every flower.\n\nTherefore:\n\nLuma makes Nimbus feel special\n→ Nimbus becomes confident.\n\nLuma gently fans Nimbus\n→ cloud fluff expands.\n\nLuma adds pleasant wind chimes\n→ Nimbus relaxes.\n\nLuma gives Nimbus a tiny water refill\n→ Nimbus becomes slightly fuller.\n\nNimbus waters one plant\n→ plant responds positively.\n\nLuma gets excited\n→ asks him to try more.\n\nNimbus wants to make Luma happy\n→ increases rain output.\n\nMultiple plants receive rain\n→ puddles appear.\n\nNimbus increases output again\n→ greenhouse becomes busier.\n\nLuma finally sees:\nthis may be too much.\n\nThe escalating rain comes from Nimbus’s eagerness, NOT from malfunction or anger.\n\n--------------------------------------------------\n\nintegrated_multimodal_description:\n\n[SHOT 1] — 0.00s–1.80s\nLUMA’S BIG PLAN\n\nContinue directly from Part 1.\n\nWarm sunny greenhouse.\n\nLuma stands excitedly near Nimbus.\n\nNimbus floats above the old watering can at his original tiny size.\n\nSprout sits nearby.\n\nLuma spreads both arms wide toward the enormous greenhouse filled with flowers and plants.\n\nHer expression says she has just discovered the greatest gardening solution ever.\n\nNimbus looks from Luma...\n\nto all the plants...\n\nthen back to Luma.\n\nHis expression becomes uncertain.\n\nCamera:\n\nmedium three-character composition\n→ slight push toward Luma\n→ over-Luma shoulder showing dozens of plants\n→ reaction close-up Nimbus\n→ Sprout reaction.\n\nDialogue:\n\nLuma:\n<d>[English] Think about it, Nimbus!</d>\n\nShe gestures around.\n\nLuma:\n<d>[English] You can help ALL my flowers!</d>\n\nNimbus:\n<d>[English] All of them?</d>\n\nLuma:\n<d>[English] All of them!</d>\n\nSprout makes a suspicious:\n\n“Hmm?”\n\nMusic:\nplayful idea motif.\n\nSound:\n- soft greenhouse ambience\n- leaves\n- distant birds\n- Nimbus cloud puff\n- Sprout squeak\n\n--------------------------------------------------\n\n[SHOT 2] — 1.80s–3.60s\nOPERATION SUPER CLOUD\n\nLuma’s eyes brighten even more.\n\nShe raises one finger.\n\nDialogue:\n\nLuma:\n<d>[English] But first...</d>\n\nTiny dramatic beat.\n\nLuma:\n<d>[English] Operation Super Cloud!</d>\n\nNimbus:\n<d>[English] Super... Cloud?</d>\n\nLuma rapidly but safely gathers a few harmless gardening items.\n\nShow one fluid playful montage:\n\nLuma grabs:\n- tiny flower crown\n- hand fan\n- little wind chimes\n- leaf water cup\n\nShe returns with the supplies in a gardening basket.\n\nSprout watches the growing pile.\n\nHis eye stalks gradually rise higher.\n\nCamera:\nclose-up Luma's “idea face”\n→ whip-like but smooth pan toward supplies\n→ short tracking movement as she collects them\n→ Sprout suspicious reaction.\n\nSound:\n- basket movement\n- tiny tool clinks\n- leaves\n- playful pizzicato accents\n\n--------------------------------------------------\n\n[SHOT 3] — 3.60s–5.50s\nCLOUD STYLE + EXTRA FLUFF\n\nLuma kneels beside floating Nimbus.\n\nFirst she carefully places the tiny flower crown over the top of his fluffy body.\n\nNimbus blushes.\n\nHis cloud cheeks turn slightly rosy.\n\nLuma:\n\n<d>[English] Perfect!</d>\n\nNimbus gives a shy smile.\n\nThen Luma lifts the small floral hand fan.\n\nShe gently fans him.\n\nWHOOSH...\n\nNot strong.\n\nJust a pleasant garden breeze.\n\nNimbus’s cloud edges flutter.\n\nHe gradually expands slightly.\n\nHis silhouette becomes fluffier and rounder.\n\nHis little raindrop feet sway.\n\nNimbus looks down at himself in surprise.\n\nDialogue:\n\nLuma:\n<d>[English] Now... extra fluff!</d>\n\nNimbus:\n<d>[English] Oh!</d>\n\nLuma laughs softly.\n\nSprout looks from the fan to Nimbus.\n\nSprout’s expression:\n“This cannot possibly end well.”\n\nCamera:\nintimate medium\n→ close-up flower crown placement\n→ side angle showing fan breeze\n→ close-up Nimbus gently expanding\n→ Sprout deadpan reaction.\n\nSound:\n- soft flower rustle\n- fan flutter\n- gentle airy puff\n- tiny magical cloud shimmer\n\n--------------------------------------------------\n\n[SHOT 4] — 5.50s–7.30s\nPREMIUM WEATHER ATMOSPHERE + RAIN REFILL\n\nLuma hangs the little wind chimes beside Nimbus.\n\nThey gently sway in the greenhouse breeze.\n\nTING...\nTING...\n\nNimbus closes his eyes.\n\nHe relaxes and floats with the rhythm.\n\nHis cloud edges ripple softly.\n\nLuma then holds out the green leaf cup containing several sparkling droplets.\n\nDialogue:\n\nLuma:\n<d>[English] And one tiny refill.</d>\n\nNimbus opens his eyes.\n\nNimbus:\n<d>[English] For me?</d>\n\nLuma:\n<d>[English] For the Super Cloud.</d>\n\nNimbus shyly leans toward the leaf.\n\nHe absorbs/sips the water.\n\nThe water disappears gently into his cloud body.\n\nA subtle pale-blue ripple travels through his fluff.\n\nHe becomes slightly larger.\n\nOne or two droplets begin hanging beneath him.\n\nSprout notices.\n\nHis eyes widen.\n\nCamera:\nwind-chime close-up\n→ Nimbus relaxing\n→ leaf-cup insert\n→ close-up water absorbing into cloud\n→ slight push-out showing larger Nimbus\n→ Sprout reaction.\n\nSound:\n- delicate wind chimes\n- tiny water slurp/absorption\n- soft magical puff\n- Sprout nervous squeak\n\n--------------------------------------------------\n\n[SHOT 5] — 7.30s–9.30s\nTHE FIRST TEST\n\nLuma guides Nimbus gently toward one small potted plant.\n\nNimbus positions himself approximately 30–40 cm above the plant.\n\nHe looks nervous.\n\nLuma kneels nearby with hands clasped.\n\nSprout watches from behind the pot.\n\nLuma whispers:\n\n<d>[English] Just a little rain.</d>\n\nNimbus nods.\n\nHe closes his eyes.\n\nHis body compresses slightly.\n\nA tiny localized drizzle begins.\n\nApproximately six to ten small droplets fall onto the plant.\n\nThe soil darkens.\n\nLeaves lift gently.\n\nA tiny bud opens.\n\nNimbus opens his eyes.\n\nLuma gasps happily.\n\nDialogue:\n\nLuma:\n<d>[English] It worked!</d>\n\nNimbus:\n<d>[English] It did?</d>\n\nLuma:\n<d>[English] Look!</d>\n\nNimbus smiles proudly.\n\nCamera:\nmedium plant test\n→ close-up Nimbus concentrating\n→ macro rain droplets\n→ leaf response\n→ reaction two-shot Luma and Nimbus.\n\nMusic:\nwarm playful success flourish.\n\nSound:\n- gentle droplets\n- soft soil water sound\n- leaf movement\n- tiny magical chime\n\n--------------------------------------------------\n\n[SHOT 6] — 9.30s–11.20s\n“JUST A LITTLE MORE”\n\nLuma claps excitedly.\n\nShe turns and notices two nearby plants.\n\nShe points.\n\nDialogue:\n\nLuma:\n<d>[English] Can you do two?</d>\n\nNimbus looks at the plants.\n\nThen proudly nods.\n\nNimbus:\n<d>[English] I can try!</d>\n\nNimbus puffs himself slightly bigger.\n\nA second localized drizzle begins.\n\nThen another.\n\nNow two or three nearby plants receive gentle rain simultaneously.\n\nLuma moves between them, amazed.\n\nSprout slowly backs away.\n\nCamera:\nmedium Luma pointing\n→ close-up Nimbus determined face\n→ pull wider as multiple little showers appear\n→ Sprout slowly retreating.\n\nSound:\n- multiple gentle rain layers\n- plant leaves moving\n- growing whimsical score\n\nLuma:\n<d>[English] Nimbus, you’re amazing!</d>\n\nNimbus beams.\n\n--------------------------------------------------\n\n[SHOT 7] — 11.20s–13.20s\nTHE GARDEN GETS MORE RAIN THAN IT ORDERED\n\nNimbus hears Luma’s praise.\n\nHe wants to help even more.\n\nHis eyes brighten.\n\nNimbus:\n\n<d>[English] More?</d>\n\nBefore Luma fully answers, Nimbus takes a deeper breath.\n\nHis cloud body becomes puffier.\n\nSeveral gentle showers begin across different nearby pots.\n\nImportant:\nkeep showers localized, soft and visually charming.\n\nRain droplets:\n- bounce off leaves\n- sparkle in sunlight\n- create tiny shallow puddles\n- run down terracotta pots\n- form small rainbow reflections\n\nOne empty lightweight gardening pot near the floor becomes wet and slowly tips onto its side.\n\nNo plant is damaged.\n\nSprout gets splashed by one large harmless droplet.\n\nHis eyes become enormous.\n\nSprout:\n“EEP!”\n\nLuma initially laughs.\n\nDialogue:\n\nLuma:\n<d>[English] Wow!</d>\n\nShe looks around.\n\nHer smile slowly changes.\n\nLuma:\n<d>[English] Wait...</d>\n\nCamera:\nwide greenhouse reveal\n→ dynamic but smooth lateral camera move through several rain zones\n→ close splash on Sprout\n→ tipping empty pot\n→ Luma realization.\n\nMusic:\ntempo increases into cheerful escalating comedy.\n\nSound:\n- rain droplets\n- little splashes\n- leaves\n- pot gently rolling\n- Sprout squeal\n\n--------------------------------------------------\n\n[SHOT 8] — 13.20s–15.00s\nTHE SUPER CLOUD IDEA GETS VERY SUPER\n\nEnd with the strongest visual escalation.\n\nWide greenhouse shot.\n\nNimbus now floats higher in the center.\n\nHe is still cute and cheerful.\n\nHe is approximately 25–35% fluffier than at Part 1.\n\nHis flower crown remains visible.\n\nMultiple gentle plant-sized rain showers fall beneath and around him.\n\nShow:\n- wet leaves\n- tiny puddles\n- dripping planters\n- colorful flowers\n- glistening greenhouse floor\n- perhaps one small natural rainbow reflection through droplets\n\nLuma moves beneath Nimbus.\n\nAt first:\ndelighted.\n\nThen a shower begins beside her.\n\nAnother begins behind her.\n\nShe turns.\n\nAnother plant starts receiving rain.\n\nHer eyes widen.\n\nShe lifts both hands.\n\nDialogue:\n\nLuma:\n<d>[English] Nimbus?</d>\n\nNimbus happily:\n<d>[English] I’m helping!</d>\n\nLuma:\n<d>[English] Yes, but maybe—</d>\n\nA few more raindrops begin.\n\nLuma:\n<d>[English] ...that’s enough rain.</d>\n\nNimbus does not fully hear/understand yet.\n\nHe continues gently raining.\n\nSprout appears beside Luma.\n\nCompletely soaked.\n\nFlower on his shell drooping from water.\n\nHe looks directly toward Luma with maximum comic disbelief.\n\nLuma looks at Sprout.\n\nThen looks at Nimbus.\n\nFinal line:\n\nLuma:\n<d>[English] Uh-oh.</d>\n\nSprout gives a dramatic:\n\n<d>[English] EEP!</d>\n\nCamera:\nlarge wide composition\n→ slow push toward Luma\n→ Nimbus above\n→ Sprout reaction\n→ end on three-character composition.\n\nFINAL STORY EMOTIONS:\n\nNimbus:\nhappy and eager.\n\nLuma:\nfinally worried.\n\nSprout:\nfull dramatic panic.\n\nCUT TO BLACK.\n\n--------------------------------------------------\n\n[AUDIO DESCRIPTION]\n\nGenerate synchronized native stereo audio.\n\n--------------------------------------------------\n\n[LUMA VOICE]\n\nNatural young girl voice, approximately 8 years old.\n\nTone:\n- warm\n- bright\n- curious\n- playful\n- enthusiastic\n- emotionally readable\n\nDo NOT make her sound like an adult.\n\nDialogue should feel spontaneous rather than heavily performed.\n\nPerformance progression:\n\n0–5 sec:\nexcited inventor.\n\n5–9 sec:\ngentle coach.\n\n9–12 sec:\nthrilled encouragement.\n\n12–15 sec:\nsurprise turning to concern.\n\n--------------------------------------------------\n\n[NIMBUS VOICE]\n\nSmall gentle youthful cloud voice.\n\nTone:\n- soft\n- innocent\n- shy\n- breathy\n- cute\n- eager to please\n\nAs confidence increases:\nhis speech becomes slightly brighter.\n\nDo NOT change into:\n- deep voice\n- powerful storm voice\n- robotic voice\n- villain voice\n\n--------------------------------------------------\n\n[SPROUT AUDIO]\n\nNo full conversational speech.\n\nUse:\n- tiny suspicious hum\n- nervous squeak\n- startled gasp\n- splashed squeal\n- final dramatic “EEP!”\n\nKeep these subtle and funny.\n\n--------------------------------------------------\n\n[ENVIRONMENT AUDIO]\n\nMaintain:\n- soft rooftop breeze\n- birds\n- subtle city ambience\n- greenhouse wooden creaks\n- glass resonance\n- leaves\n- plant movement\n- gardening tools\n- light footsteps\n- Nimbus cloud puffs\n- water droplets\n- shallow splashes\n- wind chimes\n- Sprout squeaks\n\n--------------------------------------------------\n\n[RAIN SOUND DESIGN]\n\nRain intensity must evolve audibly.\n\n0:00–0:07:\nalmost no rain.\n\n0:07–0:09:\none gentle localized shower.\n\n0:09–0:11:\ntwo small showers.\n\n0:11–0:13:\nseveral soft rain zones.\n\n0:13–0:15:\nnoticeably busier but still gentle rainfall.\n\nNo thunder.\n\nNo huge storm roar.\n\nNo frightening bass.\n\n--------------------------------------------------\n\n[NON-DIEGETIC MUSIC]\n\nUse an original whimsical animation score.\n\nInstrumentation:\n- pizzicato strings\n- marimba\n- light wooden percussion\n- soft flute\n- clarinet\n- glockenspiel\n- gentle plucked instruments\n- tiny bell accents\n\nMusic progression:\n\n0.00–1.80\ncarry over Part 1’s playful BIG IDEA motif.\n\n1.80–5.50\nlight “experiment montage” rhythm.\n\n5.50–7.30\nsoft magical chime textures.\n\n7.30–9.30\nwarm success theme.\n\n9.30–11.20\nslightly faster playful rhythm.\n\n11.20–13.20\ncomedic escalation.\n\n13.20–15.00\nbusy playful motif followed by a small “uh-oh” ending sting.\n\nNo epic music.\n\nNo horror.\n\nNo dramatic danger music.\n\n--------------------------------------------------\n\n[CAMERA LANGUAGE]\n\nUse cinematic storybook camera movement.\n\nShot pacing progression:\n\n0–4 sec:\ncharacter-driven medium compositions.\n\n4–7 sec:\nclose-up prop and makeover details.\n\n7–11 sec:\nclear plant/rain demonstrations.\n\n11–13 sec:\nwider comedy compositions.\n\n13–15 sec:\nbig greenhouse reveal + reaction-driven ending.\n\nUse:\n- soft dolly\n- gentle push-ins\n- short lateral tracking\n- over-the-shoulder views\n- macro rain inserts\n- reaction close-ups\n- readable wide shots\n- subtle rack focus\n\nDo NOT use:\n- handheld shaking\n- crash zooms\n- extreme fisheye\n- Dutch-angle chaos\n- 360 spins\n- hyperfast cutting\n\n--------------------------------------------------\n\n[VISUAL STYLE]\n\nPremium finished painterly 2D animation.\n\nMatch Part 1 exactly:\n\n- delicate hand-drawn contour lines\n- watercolor-like fills\n- warm storybook textures\n- cinematic sunlight\n- expressive facial animation\n- rounded child-friendly shapes\n- natural background detail\n- subtle squash and stretch\n- rich but soft greenery\n- charming handmade props\n\nThe final animation should feel like the storyboard artwork has come alive.\n\nDO NOT render as:\n- storyboard animation\n- motion comic\n- 3D CGI\n- photorealism\n- anime\n- vector graphic\n- cel-shaded 3D\n- slideshow\n\n--------------------------------------------------\n\n[COLOR / LIGHTING]\n\nBeginning:\nwarm golden greenhouse sunlight.\n\nPrimary palette:\n- sunflower yellow\n- sky blue\n- leafy green\n- mint\n- coral\n- warm cream\n- terracotta\n- pale rainbow highlights\n\nNimbus:\nlight blue-white.\n\nAs Nimbus becomes fluffier:\nslightly increase cool blue shading within his cloud body.\n\nDo NOT darken him.\n\nAs rainfall increases:\nintroduce:\n- cool blue reflections\n- glossy wet leaves\n- small sparkling droplets\n- subtle rainbow refractions\n\nKeep environment sunny.\n\nThe rain should coexist with sunlight.\n\nThat combination produces the playful magical look.\n\n--------------------------------------------------\n\n[MOTION RULES]\n\nLUMA:\n\nUse natural child motion:\n- energetic gestures\n- crouching\n- excited pointing\n- quick little steps\n- expressive hands\n- head tilts\n- hair bouncing\n- hat movement\n- natural blinking\n\nNIMBUS:\n\nUse buoyant cloud physics:\n- slow floating\n- gentle bobbing\n- cloud compression before rain\n- subtle expansion after water\n- tiny jiggle after being fanned\n- edge fluff responding to wind chimes\n- raindrop feet swaying\n\nNimbus should gain fluff gradually.\n\nNever instant-transform.\n\nSPROUT:\n\nMovement remains mostly slow.\n\nUse sudden fast eye-stalk reactions for comedy.\n\nRAIN:\n\nDroplets should have:\n- convincing gravity\n- small splashes\n- reflections\n- natural surface interaction\n\nNo floating upward rain.\n\n--------------------------------------------------\n\n[CONTINUITY RULES]\n\nSame Luma throughout.\n\nSame Nimbus throughout.\n\nSame Sprout throughout.\n\nThe flower crown stays on Nimbus once applied.\n\nNimbus grows gradually, never shrinking again during Part 2.\n\nNimbus remains blue-white.\n\nLuma’s hat remains on.\n\nLuma’s seed pouch remains visible.\n\nSprout’s tiny yellow flower remains attached.\n\nThe greenhouse layout stays coherent.\n\nThe old watering can remains in the environment.\n\nThe original sunflower remains visible if framing allows.\n\nRain accumulation increases gradually.\n\nDo not reset the floor to dry between shots.\n\n--------------------------------------------------\n\n[NEGATIVE CONSTRAINTS]\n\nNo storyboard sheet.\nNo panel borders.\nNo panel numbers.\nNo captions.\nNo subtitles.\nNo timestamps.\nNo production notes.\nNo watermark.\nNo logos.\nNo brands.\n\nNo character identity drift.\nNo extra Lumas.\nNo extra Nimbus clouds.\nNo duplicate Sprout.\nNo outfit changes.\nNo hairstyle changes.\nNo missing yellow hat.\nNo missing flower crown after installation.\n\nNo giant Nimbus.\nNo black cloud.\nNo gray Nimbus yet.\nNo evil expressions.\nNo lightning.\nNo thunder.\nNo hurricane.\nNo dangerous wind.\nNo heavy flooding.\nNo destruction.\nNo smashed plants.\nNo broken greenhouse.\nNo injuries.\n\nNo Part 3 emotional collapse.\nNo exhausted Nimbus.\nNo Nimbus curled sadly in corner.\nNo Luma apology.\nNo “You don’t have to rain for everyone.”\nNo magical finale.\nNo rainbow charm.\n\nNo warped anatomy.\nNo malformed hands.\nNo duplicated fingers.\nNo melted cloud face.\nNo inconsistent character scale.\nNo flickering outfit details.\n\n--------------------------------------------------\n\n[FINAL 1.8-SECOND PRIORITY]\n\nThe final beat must clearly communicate:\n\n“The experiment worked TOO well.”\n\nWide sunny greenhouse.\n\nNimbus is slightly larger, fluffy, blue-white, wearing his flower crown.\n\nHe happily produces several gentle localized rain showers.\n\nPlants sparkle.\n\nTiny puddles have appeared.\n\nLuma stands below, looking around as multiple showers continue.\n\nShe realizes the problem.\n\n...",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2085929499653238799",
    "title": "柯克舰长的舰桥指令",
    "titleEn": "Captain Kirk Gives the Bridge Command",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《柯克舰长的舰桥指令》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Captain Kirk Gives the Bridge Command. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "\"Create a star trek scene on the bridge with the original captain Kirk, add ref image of Joe to the bridge with captain Kirk giving Joe and order to \"Engage\" and Joe respond \"Okie Dokie\"\n\n#aiart #hollywood #film #Advertising",
    "promptSourceUrl": "https://x.com/SolutionsJoeG/status/2085929499653238799",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/SolutionsJoeG/status/2085929499653238799",
    "sourceLabel": "X 原帖 · @SolutionsJoeG",
    "author": "@SolutionsJoeG",
    "publishedAt": "2026-08-08T03:22:02.070Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2085929499653238799.mp4",
    "posterUrl": "/posters/x/2085929499653238799.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "有 Prompt",
      "完整 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Model on Generative AI is getting really good .. I",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2085720283336622590",
    "title": "奥德修斯直面独眼巨人",
    "titleEn": "Odysseus Faces the Cyclops",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《奥德修斯直面独眼巨人》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 25 秒 · 1280×720。",
    "summaryEn": "An H3 output for Odysseus Faces the Cyclops. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 25s · 1280×720.",
    "prompt": "An ancient Greek warrior resembling Odysseus stands alone inside a colossal cave, illuminated only by shafts of golden light piercing through cracks in the ceiling. Behind him, an enormous Cyclops slowly emerges from the darkness, its single eye glowing like molten fire. Dust falls from the ceiling with every thunderous footstep. The warrior tightens his grip on a weathered bronze spear without looking back. The camera slowly circles before rapidly pulling away to reveal the overwhelming scale of the giant. Hyper-realistic Greek mythology, cinematic IMAX lighting, volumetric fog, dramatic shadows, epic scale, photorealistic, breathtaking.\n\nEpic ancient Greek mythology, grounded realism, sweeping IMAX cinematography, practical-looking costumes and armor, dramatic natural lighting, weathered bronze, marble architecture, vast Mediterranean landscapes, cinematic atmosphere, emotional scale, ultra-detailed, award-winning historical fantasy.",
    "promptSourceUrl": "https://x.com/apiframe/status/2085720292706619649",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/apiframe/status/2085720283336622590",
    "sourceLabel": "X 原帖 · @apiframe",
    "author": "@apiframe",
    "publishedAt": "2026-08-07T13:30:41.013Z",
    "addedAt": "2026-08-22T08:54:41.025Z",
    "mediaUrl": "/media/x-2085720283336622590.mp4",
    "posterUrl": "/posters/x/2085720283336622590.jpg",
    "duration": 25,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "有 Prompt",
      "完整 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Inspired by The Odyssey",
    "approvedAt": "2026-08-22T08:54:41.025Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083207597025354071",
    "title": "赛博战士对决远古斗士",
    "titleEn": "Cyber Warrior vs. Primordial Fighter",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《赛博战士对决远古斗士》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Cyber Warrior vs. Primordial Fighter. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create a 15-second cinematic showdown between the most technologically advanced human and the most primordial warrior.\n\nThe technological fighter wears adaptive nano-armor, holographic interfaces, energy weapons, and glowing cybernetic implants.\n\nThe primordial fighter is massive and feral, covered in ritual markings, bone armor, and animal hides, wielding a brutal stone weapon.\n\nSet the battle in a ruined landscape where a futuristic megacity merges with an ancient volcanic wilderness.\n\n00:00 — The fighters face each other as futuristic machinery hums and distant primal drums begin.\n\n00:02 — The technological fighter activates the nano-armor and says in a cold synthetic voice: “Evolution ends with me.”\n\n00:04 — The primordial warrior roars and replies in a deep, savage voice: “You have forgotten what strength is.”\n\n00:06 — They charge toward each other at full speed.\n\n00:08 — Their weapons collide, releasing a massive shockwave of blue plasma, fire, sparks, debris, lightning, and shattered holographic fragments.\n\n00:10 — The technological fighter launches drones and energy blades while the primordial warrior smashes the ground, creating volcanic cracks and a towering blast of stone and flame.\n\n00:12 — Both fighters unleash their final attacks simultaneously as the camera circles them in dramatic slow motion.\n\n00:14 — A blinding explosion consumes the battlefield.\n\n00:15 — The smoke begins to clear, revealing only a mysterious silhouette without showing who won.\n\nUse dynamic tracking shots, dramatic low angles, rapid orbiting camera movement, extreme close-ups, speed ramps, controlled handheld energy, and slow motion during key impacts.\n\nCinematic 4K, 16:9, ultra-realistic VFX, physically accurate destruction, volumetric smoke, plasma energy, fire, sparks, debris, lightning, synchronized impact sounds, futuristic weapon audio, primal roars, distinct English voices, and a cinematic orchestral score mixed with electronic pulses and tribal percussion.",
    "promptSourceUrl": "https://x.com/alex_bagnuoli89/status/2083207600141742136",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/alex_bagnuoli89/status/2083207597025354071",
    "sourceLabel": "X 原帖 · @alex_bagnuoli89",
    "author": "Alessandro Bagnuoli",
    "publishedAt": "2026-07-31T15:06:09.913Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083207597025354071.mp4",
    "posterUrl": "/posters/x/2083207597025354071.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Action & VFX",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The future has nano-armor, drones and energy weapons.\n\nThe past has rage, instinct and nothing to lose.\n\nWe created this 15-second battle with 𝗠𝗶𝗻𝗶𝗠𝗮𝘅 𝗛𝟯 𝗼𝗻 @openart_ai .\n\nOnly one silhouette survives. Who wins?\n\nFull prompt below. 👇\n\n#AIVideo #MiniMaxH3 #OpenArt",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083189719827878083",
    "title": "草莓饮料变形广告",
    "titleEn": "Strawberry Drink Transformation Commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《草莓饮料变形广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 7 秒 · 1280×720。",
    "summaryEn": "An H3 output for Strawberry Drink Transformation Commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 7s · 1280×720.",
    "prompt": "{   \"shots\": [     {       \"shot_id\": \"1\",       \"start_time\": \"00:00\",       \"end_time\": \"00:02\",       \"camera_angle\": \"Close-up\",       \"subject\": \"Strawberry\",       \"action\": \"A single, dull strawberry transforms into a vibrant, ripe strawberry, floating against a pink background.\",       \"visual_details\": \"Initial strawberry is muted, then becomes intensely red with visible seeds; pink background; subtle glow effect.\",       \"audio_cues\": \"Light, subtle, almost magical sound effect.\"     },     {       \"shot_id\": \"2\",       \"start_time\": \"00:02\",       \"end_time\": \"00:05\",       \"camera_angle\": \"Medium shot\",       \"subject\": \"Strawberry and beverage can\",       \"action\": \"The vibrant strawberry transforms into a white and pink beverage can with 'STRAWBERRY CREAM' text. Liquid pours into the can from above, and strawberries/water droplets float around it.\",       \"visual_details\": \"Can is sleek, covered in condensation; red liquid splashing into the can; numerous floating strawberries (whole and sliced) and water droplets; pink background.\",       \"audio_cues\": \"Sound of liquid pouring, bubbling, and light splashes; upbeat, commercial-style music.\"     },     {       \"shot_id\": \"3\",       \"start_time\": \"00:06\",       \"end_time\": \"00:10\",       \"camera_angle\": \"Close-up\",       \"subject\": \"Top of the beverage can\",       \"action\": \"Focus on the top of the can as the liquid continues to pour and splash, with droplets and strawberries in motion.\",       \"visual_details\": \"Intricate details of the can's opening and condensation; dynamic splashes of liquid; close-up of floating strawberry pieces.\",       \"audio_cues\": \"Enhanced sounds of liquid splashing and bubbling; continued music.\"     },   ] }",
    "promptSourceUrl": "https://x.com/GumVue/status/2083189719827878083",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/GumVue/status/2083189719827878083",
    "sourceLabel": "X 原帖 · @GumVue",
    "author": "Gumvue Studio",
    "publishedAt": "2026-07-31T13:55:07.657Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083189719827878083.mp4",
    "posterUrl": "/posters/x/2083189719827878083.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Bravo! @Hailuo_AI #MinimaxH3  prompt here {   \"shots\": [     {       \"shot_id\": \"1\",       \"start_time\": \"00:00\",       \"end_time\": \"00:02\",       \"camera_angle\": \"Close-up\",       \"subject\": \"Strawberry\",       \"action\": \"A single, dull strawberry transforms into a vibrant, ripe strawberry, floating against a pink background.\",       \"visual_details\": \"Initial strawberry is muted, then becomes intensely red with visible seeds; pink background; subtle glow effect.\",       \"audio_cues\": \"Light, subtle, almost magical sound effect.\"     },     {       \"shot_id\": \"2\",       \"start_time\": \"00:02\",       \"end_time\": \"00:05\",       \"camera_angle\": \"Medium shot\",       \"subject\": \"Strawberry and beverage can\",       \"action\": \"The vibrant strawberry transforms into a white and pink beverage can with 'STRAWBERRY CREAM' text. Liquid pours into the can from above, and strawberries/water droplets float around it.\",       \"visual_details\": \"Can is sleek, covered in condensation; red liquid splashing into the can; numerous floating strawberries (whole and sliced) and water droplets; pink background.\",       \"audio_cues\": \"Sound of liquid pouring, bubbling, and light splashes; upbeat, commercial-style music.\"     },     {       \"shot_id\": \"3\",       \"start_time\": \"00:06\",       \"end_time\": \"00:10\",       \"camera_angle\": \"Close-up\",       \"subject\": \"Top of the beverage can\",       \"action\": \"Focus on the top of the can as the liquid continues to pour and splash, with droplets and strawberries in motion.\",       \"visual_details\": \"Intricate details of the can's opening and condensation; dynamic splashes of liquid; close-up of floating strawberry pieces.\",       \"audio_cues\": \"Enhanced sounds of liquid splashing and bubbling; continued music.\"     },   ] }",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083180088775045626",
    "title": "冰霜枪手网页互动循环",
    "titleEn": "Ice Gunslinger Interactive Web Loop",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《冰霜枪手网页互动循环》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1278×720。",
    "summaryEn": "An H3 output for Ice Gunslinger Interactive Web Loop. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1278×720.",
    "prompt": "@Hailuo_AI Prompt #MiniMaxH3 in Chinese:\n\n无缝循环 | 网页首屏动画\n\n[核心概念] 她的每一个动作都对应一种网页交互行为，她不是站在网页前，而是在演示这个网页如何运作。加载、悬停、点击、滚动、拖拽、轮播、提交、刷新——八种交互依次发生，每一种都有对应的冰质界面反馈。\n\n[人物锁定 · @image1] 完全按照该参考图呈现，不进行任何重新设计。\n容貌：年轻女性，薄荷绿齐颏短发，发丝随动作轻扬；锐利的浅青碧色双眼，青色指甲，水晶坠耳饰。\n服装：白色剪裁西装——敞开的外套与长裤，领口环绕金色巴洛克卷草纹饰件，肩披缀满蓝色冰晶尖刺的深色皮草披领，腰间垂挂细金链，大腿绑黑色枪套束带。\n武器：两把华丽的银质左轮手枪，一手一把，枪管为冰蓝色水晶质，枪身雕花，转轮与击锤细节完整。\n场景：宏伟的宫殿大厅，高耸的拱形长窗，暖金色内饰，大理石地面有真实反射。窗外天光冷，室内金暖，冷暖对比贯穿全片。\n全片不得改动发型、面容、西装轮廓、金饰卷草纹、披领结构、枪械造型与比例。\n\n[解剖与武器锁定 · 最高优先级] 全片自始至终只有：两条手臂、两只手、每只手五指、恰好两把左轮——右手一把、左手一把，绝无第三把。拖拽段左枪明确入套，左手空出且五指清晰。所有转枪、滑步、射击及其运动模糊中，残影绝不可被渲染成额外肢体或额外枪械；手部每一帧结构完整、指节分明、扳机指位置明确。枪绝不悬空、绝不复制。\n\n[网页 UI/UX 系统 · 冰质界面] 画面按 12 栏网格构建，外边距 6%，栏间距 2%。安全区：第 1 至 5 栏为网页标题与按钮预留，人物与主冰晶结构全程不进入第 1 至 4 栏。\n所有界面元素由「冰」构成——发丝级霜线，半透明，具真实折射，如结在一层看不见的玻璃上。界面配色：霜线冰白 #EDF4F8，激活态冰蓝 #9FD8EC，确认态暖金 #D9A94C。\n骨架屏：加载段中，尚未「载入」的画面区域以磨砂半透明冰板占位，微微闪烁，内容载入时冰板融化、真实细节浮现。\n光标：一枚冰晶准星即是鼠标指针，全片跟随她当前的注意力移动，悬停时展开一圈细霜，点击时收缩并溅霜。\n悬停态：被准星悬停的物体表面泛起一层冰蓝微光并轻微上浮。\n涟漪反馈：每次「点击」命中处先扩出一圈材质涟漪，再生长冰晶。\n滚动条：右缘一根细冰柱随滚动段的横移同步充填。\n对齐参考线：拖拽段中，画面闪现发丝级的洋红对齐线 #E23D9B，元素吸附入位时消失。\n弹数指示：右下六格冰晶弹仓图标，随射击逐格碎裂。\n载入环：一圈环形霜环兼作页面载入条，1.2 秒起步，7.8 秒半圈，11.4 秒闭合爆亮，13.0 秒后逆向融化。\n缓动：所有界面与文字入场为标准缓出约 0.4 秒，退场缓入约 0.25 秒，吸附与回弹为轻微过冲后回位，绝不线性、绝不夸张弹跳，多元素错位约 0.06 秒。\n\n[文字系统 · 皆由场景物质生成] 共三处，字形统一为厚重镌刻衬线体——粗竖笔、发丝细横笔、凿刻楔形衬线、宽字距。每处均为三维实体，具厚度、倒角、折射与投影。\n以下引号内为画面实际渲染文字，须原样逐字呈现：\n1.8秒 \"GLACIA\"（G-L-A-C-I-A，全大写）由悬浮冰晶碎片在她身侧半空组装，每片旋转就位成为字母的一个刻面，冰体内部具次表面散射与气泡，边缘折射细彩虹，保持两拍后碎为粉雪。\n7.4秒 \"HAILUO AI\"（H-A-I-L-U-O 空格 A-I，全大写）以霜花在最大拱形长窗玻璃上蔓延书写——霜沿字形边缘呈羽毛状结晶生长，背后天光透过微亮，随后被继续蔓延的霜幕吞没。\n13.4秒 \"Cia0\"（大写 C、小写 i、小写 a、数字零；末位是数字零而非字母 O，须更窄更椭圆）以细雕刻纹刻于右枪银质枪身，与周围雕花同深度同色泽，仅在收枪入套、掠射高光扫过的一瞬显现，绝不发光、描边或放大。\n全片无任何其他文字、日文字符或字幕。\n\n[色彩渐变体系] 全片色彩以渐变构成，绝不平涂。宫殿内饰自受光面暖金 #C9A24A 渐变至阴影蜜褐 #8A6B35；大理石地面冷白 #E8E6E0 带灰纹，反射中带一层冰蓝。窗外天光自窗顶冷蓝 #A8C4D8 渐变至窗底近白。她的西装受光纯白 #F4F0E8 渐变至阴影暖灰 #C9C2B4。冰体自表层冰蓝 #9FD8EC 渐变至核心深青 #3D7A9E，最深处近墨蓝 #1E3A52，边缘折射完整细光谱。霜哑光白。粉雪在逆光中金边、在冷光中蓝边。冷色只属于冰、霜与窗光，暖色只属于宫殿与金饰，冷暖交界处始终是画面张力所在。\n\n[渲染品质] 最高保真。冰具真实次表面散射、内部气泡、生长纹与裂隙，边缘真实色散；霜以羽毛状晶枝逐帧生长，绝非淡入；粉雪具体积感受气流卷动。银枪抛光，雕花凹处压暗，转轮转动时反射流动。西装挺括、褶皱真实回弹；皮草逐撮渲染，冰刺在其间折射；发丝逐根渲染，甩动滞后。大理石反射她与冰晶的完整倒影。胶片式高光滚降，阴影提亮，最亮金处柔和光晕，细腻颗粒，绝不过曝。\n\n[特效 · 冰之弹道] 每发子弹是一颗冰种：弹道拖出瞬间结晶的霜线；命中处先扩涟漪、再以延时摄影速度生长冰花、冰枝、倒悬冰吊灯，逐帧可见生长；枪口喷冷雾而非火光；弹壳空中结霜、落地叮当结冰；她呼吸带白汽；足下薄霜外蔓；冰体折射窗光在金饰上投下游动冷斑\n\n@Hailuo_AI @Image1 ；变形镜条状光斑掠过每处高光；焦外粉雪呈椭圆光斑；高潮处大厅被冰晶架构半包裹，暖金自冰后透出。冰晶持续累积，绝不消失，直至终段统一升华。\n\n[动作编排 · 交互映射，每拍双手双枪明确]\n0.0-1.4 加载。无人物。窗玻璃霜结微距，四角霜框结晶延展，骨架屏冰板铺陈画面，载入环起步。\n1.4-3.0 载入完成。骨架冰板自左至右依次融化，大厅细节逐区浮现锐化，最后一块融化处是她——背对长窗，双枪在套，抬眼，一次白汽呼吸。准星光标出现。\"GLACIA\" 组装。\n3.0-4.6 悬停与点击按下。准星飘移至她的右侧枪套，枪套泛起冰蓝悬停微光并轻浮；她右手拔枪的瞬间即是按下——准星收缩溅霜。左手随后拔左枪，双枪各转一周停稳，动作利落克制。\n4.6-6.2 点击反馈。右枪射出第一发——弹道霜线飞越大厅，命中远柱：先一圈材质涟漪扩开，再炸开生长冰花。左枪抬起戒备。准星随弹道飞移并在命中点收缩。\n6.2-7.8 滚动。她沿大理石侧滑横移——整页随之「滚动」：前景柱、中景冰晶、远景长窗以不同速度视差掠过，右缘冰柱滚动条同步充填。滑行中双枪交替朝上射三发，各在头顶炸开一座倒悬冰吊灯，如同锚点依次钉入。霜同时在长窗上写出文字。\n7.8-9.6 拖拽与跳转，一镜到底。她滑停，左枪利落入套；左手抓住一块悬浮的冰晶面板向右横拖——洋红对齐参考线闪现，面板轻微过冲后吸附入位；吸附瞬间右枪朝面板中心射出一发，镜头贴弹穿过面板冰体——穿越时全画面被冰内折射充满，如一次页面跳转——穿出后甩回她，她已单膝滑跪停稳。\n9.6-11.4 轮播。八拍每拍 0.225 秒，如八张卡片依次滑入：一、白底冰蓝线稿，仅双枪与双手；二、冰蓝与暖金色块拼贴；三、二色剪影纯白配金；四、完整渲染一记真实双枪同射；五、拖影帧；六、整帧负片反转仅此一帧；七、线稿叠加冰晶折射；八、完整渲染短暂静止。每拍以水平滑切衔接如卡片轮播，冰晶持续累积。\n11.4-13.0 提交。两帧全静止。她起身，双枪缓缓合拢指向镜头——如按下主按钮——同时扣动扳机，两发弹道在镜头正前交汇，一朵冰花在镜面绽放生长为成功态；透过冰花看见她放下双枪、直视镜头的双眼。载入环闭合爆亮转暖金。\n13.0-15.0 刷新。全厅冰晶自下而上依次升华为粉雪与冷雾，如一次下拉刷新将页面推起。她双枪先右后左旋入枪套，掠射高光扫过右枪显现刻纹。首屏版式于第 1 至 5 栏短暂组装——霜线导航与三枚冰晶圆点、显示级 \"GLACIA\"、其下霜线圆角按钮内含 \"ENTER\"——保持 0.8 秒后融化。镜头推回窗玻璃霜结微距。最后一帧与第一帧完全一致。\n\n[摄影] 变形宽银幕，椭圆焦外，冷暖双色条状光斑，180 度快门。每镜头具明确动机与前中后三层纵深——粉雪、悬浮冰屑或披领冰刺始终在前景制造视差。人物始终位于第 6 至 12 栏。\n0.0-1.4 100mm 微距贴窗，霜枝逐帧生长掠过，随后离窗后拉。\n1.4-3.0 摇臂自窗顶缓降至她身后半身，逆窗光，骨架冰板在镜头前逐块融化。\n3.0-4.6 40mm 低位轻推，拔枪双臂划过前景，荷兰角三度。\n4.6-6.2 85mm 咬住枪口，射击瞬间甩镜追弹道，命中点急停看涟漪与冰花生长。\n6.2-7.8 24mm 低位横移与滑步同速同向，三层视差完整展开，三次射击各一次轻微震镜。\n7.8-9.6 一镜到底：随拖拽横移，吸附瞬间轻微过冲，随后贴弹穿板，穿越段升格至四分之一速，甩回实速。\n9.6-11.4 焦段随拍点跳切，85mm 与 24mm 交替，以水平滑切模拟轮播。\n11.4-13.0 50mm 正面缓推，冰花在镜面绽放时短暂定格。\n13.0-15.0 缓慢后拉环视升华的大厅，随后推回窗玻璃微距，与第一帧对位。\n转场一律以粉雪与冷雾溶解；仅轮播段硬切加静音门；提交前两帧全静止蓄力。\n\n[物理] 重量与利落优先。左轮真实后坐——腕部上弹、小臂吸收、肩部微沉，双枪同射全身后压半步。滑步真实借力，外套下摆与披领变向滞后回弹。拖拽面板时左臂可见牵引力，吸附瞬间手臂随之一顿。弹壳受重力落地弹跳。冰晶生长推开空气，粉雪受动作气流卷动。杜绝任何僵硬、停顿或不自然的关节角度。\n\n[音频] 每个音效卡在其画面拍点的准确帧上，混音首尾相接可循环。界面音与物理音融合：每种交互都有一枚清脆的冰质界面音——悬停一声轻霜，点击一声脆响，吸附一声实响，确认一声暖钟。底层：宏大室内安静混响，窗外极远风声，细微冰体应力吟。\n0.0-1.4 霜枝生长的细密结晶声，骨架冰板的低频闪烁嗡鸣，载入环起步的一声轻起。\n1.4-3.0 冰板逐块融化的水晶滴落音阶，她的呼吸与白汽，冷弦乐铺底进入，\"GLACIA\" 组装的逐片轻响。\n3.0-4.6 悬停轻霜音，拔枪的皮革与金属滑响、点击脆响，转枪机械咔哒，双击锤两声。\n4.6-6.2 第一声枪响带冰质混响尾音，涟漪扩开的一圈水纹音，冰花生长密集脆响，弹壳冰叮。\n6.2-7.8 滑步摩擦与滚动条充填的细流声，三声交替枪响各带回声，冰吊灯层叠生长，霜爬窗细密声。\n7.8-9.6 左枪入套一声，拖拽的冰面摩擦，对齐线两声轻嘀，吸附一声实响带轻微过冲回弹音；穿板段一切拉长降调，甩回实速，滑跪一声实响。\n9.6-11.4 八记打击卡在八拍——枪响、冰裂、击锤交替，水平滑切各带一声卡片掠",
    "promptSourceUrl": "https://x.com/Cia0_exe/status/2083180243687227676",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Cia0_exe/status/2083180088775045626",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-07-31T13:16:51.435Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083180088775045626.mp4",
    "posterUrl": "/posters/x/2083180088775045626.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1278×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Fashion",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "UI/UX for animation your apps \n\nMiniMax H3 on @Hailuo_AI  X midjourney V8.2 \n\nBuilt GLACIA a hero video where the character demonstrates the interface\n\n4,765 characters of Chinese prompt",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083172752790282615",
    "title": "波尔图法兰西斯克三明治喜剧食谱",
    "titleEn": "Porto Francesinha Comedy Recipe",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《波尔图法兰西斯克三明治喜剧食谱》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1698×720。",
    "summaryEn": "An H3 output for Porto Francesinha Comedy Recipe. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1698×720.",
    "prompt": "Style: <image_1> (don't reproduce the image, use it only as aesthetic reference)\n\nEdit: Comedy, fast paced, mixing close ups, medium shots and wide shots from uncanny and tilted angles. Super-imposed texts with doodles identify ingredients as they appear and add comic remarks.\n\nScene: In Porto, Portugal a crazy chef explains how to do a Francesinha in Portuguese from Portugal with Porto accent.",
    "promptSourceUrl": "https://x.com/imagineFERA/status/2083172755399233946",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/imagineFERA/status/2083172752790282615",
    "sourceLabel": "X 原帖 · @imagineFERA",
    "author": "FERA",
    "publishedAt": "2026-07-31T12:47:42.400Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083172752790282615.mp4",
    "posterUrl": "/posters/x/2083172752790282615.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1698×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Comedy",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Funny and useful one: How cultured a model is?\nIn this case @Hailuo_AI #MiniMaxH3.\n\nUse something that isn't widely known: a recipe, an accent, a location is something I always do in t2v (letting the model fill the blanks, I only provide general directions)\n\nIn this case a Portuguese dish from the North of Portugal (Prompt, Tips and honest Review👇)",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083143771508592938",
    "title": "金刚鹦鹉极慢速鸣叫",
    "titleEn": "Macaw Scream in Extreme Slow Motion",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《金刚鹦鹉极慢速鸣叫》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1698×720。",
    "summaryEn": "An H3 output for Macaw Scream in Extreme Slow Motion. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1698×720.",
    "prompt": "/gen prompt: Cinematic wildlife documentary, vertical framing, subject centered, National Geographic award-winning cinematography, anamorphic lens, rich natural color, natural sunlight, natural motion blur. States: the scream never stops, the camera always rises, every touched object reacts. Characters: a blue-and-gold macaw, cobalt wings, golden chest, green crown, white face patch, black beak; a young man, short dark curly hair, olive t-shirt. Israeli white stone apartment building, living room. 0-2s macro: silence, the macaw inhales, chest swelling, throat feathers rising, the man mid-sip. 2-5s the beak opens, the world slams into extreme slow motion: a ripple of distorted air expands from the beak, rings race across his coffee, the mug shattering into countable shards, coffee hanging in ropes, his cheeks rippling back. 5-9s time snaps back; the wave blows the shutters open, the camera rides it out, straight up the facade. 9-14s rising floor by floor: laundry lines whipping, a dog howling in one window; crawl on pigeons detonating off a ledge, feathers suspended, the ripple passing them; snap back. 14-17s the roof: solar heaters trembling, tank water rippling, car alarms flashing far below; bold 3D text \"105 DB\" slams centered mid-frame. 17-20s hard cut to the room: the macaw fluffs, tilts its head, one tiny chirp; the man frozen with the mug handle. Audio: silence, a heartbeat on the inhale; a sub-bass shockwave, one raw macaw screech Doppler shifting as the camera climbs, glass shatter, dogs and alarms join, biggest boom on the text, ringing silence, a tiny chirp. Photorealistic, no deformation, every feather barb visible, crisp letter edges, smooth fluid motion, 8k, extreme details. , resolution:720p, duration:20s, aspect_ratio:9:16",
    "promptSourceUrl": "https://x.com/yuvalav/status/2083143870922047519",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/yuvalav/status/2083143771508592938",
    "sourceLabel": "X 原帖 · @yuvalav",
    "author": "Yuval Avidani (יובל אבידני)",
    "publishedAt": "2026-07-31T10:52:32.724Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083143771508592938.mp4",
    "posterUrl": "/posters/x/2083143771508592938.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1698×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Fable 5 + The NEW MiniMax H3 Omni Model which is INSANE! 🤯\n\n-> Up to 15 seconds\n-> Supports Audio / Video / Images as input (multi-model, omni model)\n-> Up to 2K resolution!! \n-> Great in slow motion and texts!\n\n#MiniMaxH3 @Hailuo_AI \n\nPrompt below (in the comments) ⏬",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083136052638998714",
    "title": "黑莓香草汽水 UGC Vlog",
    "titleEn": "Blackberry Vanilla Soda UGC Vlog",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《黑莓香草汽水 UGC Vlog》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Blackberry Vanilla Soda UGC Vlog. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "CHARACTER & WORLD LOCK + FULL 15-SEC MASTER PROMPT\n\nA 15-second continuous UGC-style vlog in 9:16 vertical format. Jessy, a woman in her mid-20s — use her facial features exactly as shown in her reference photo, keep everything else about the scene freshly designed — wears an attractive pastel-colored dress (soft lavender, mint, or buttery yellow) that visually pops in contrast against the bubblegum-pink Olipop can. Location: a bright, clean modern kitchen/countertop with soft natural daylight. Product: a can of Olipop Blackberry Vanilla — pastel pink can, magenta blackberry illustration, white vanilla flower, dark green \"Blackberry Vanilla\" wordmark, \"Supports Digestive Health\" text near the top, \"A New Kind of Soda™\" tagline (use the uploaded product photo as the exact visual reference for the can). Style: authentic smartphone UGC vlog, natural handheld feel, realistic skin texture, no over-retouched look. She speaks casual, upbeat, conversational English directly to camera with natural lip-sync.\n\nThe video follows this exact beat sheet, with clean cuts between each moment:\n\n00:00–00:02 — Front-camera selfie, slight handheld sway, Jessy fans herself lightly in the kitchen, looking for something refreshing, smiles at camera: \"Okay I need something fun to drink right now.\"\n\n00:02–00:04 — Eye-level medium shot, she opens the fridge and pulls out the pink Olipop can, face lighting up with delight: \"Ooh, Blackberry Vanilla Olipop — yes please.\"\n\n00:04–00:06 — Close-up on her hands slowly turning the can, label details (\"Prebiotics, Botanics, Plant Fiber\") visible: \"Prebiotics AND it tastes like dessert? Sold.\"\n\n00:06–00:08 — Tight macro shot, her fingers pop the tab, condensation and a light fizz mist visible: \"Listen to that fizz.\"\n\n00:08–00:10 — Slow-motion macro shot, pink soda pouring over ice into a clear glass, bubbles rising: \"Look how pretty that color is.\"\n\n00:10–00:12 — Close-up on her face as she sips, eyes lighting up with genuine, pleasantly surprised reaction: \"Wait... that's actually so good.\"\n\n00:12–00:13 — Macro hero shot of the can alone on the counter, condensation dripping, soft bokeh kitchen background, slow subtle push-in, no dialog.\n\n00:13–00:14 — Medium shot, she smiles at camera, can raised slightly at chest height: \"Blackberry vanilla was the right call.\"\n\n00:14–00:15 — Close front-camera selfie, can held beside her face, warm bright smile into the lens: \"Go grab an Olipop, seriously.\"\n\nContinuous natural ambient kitchen sound, subtle upbeat pop background music underneath, her voice clear and casual throughout. Consistent face, outfit, and setting across every beat. No text overlays, no captions, no watermarks, no horizontal framing, no changes to the can's design or branding.",
    "promptSourceUrl": "https://x.com/AIwithJessica/status/2083136052638998714",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AIwithJessica/status/2083136052638998714",
    "sourceLabel": "X 原帖 · @AIwithJessica",
    "author": "Jessica Collins",
    "publishedAt": "2026-07-31T10:21:52.402Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083136052638998714.mp4",
    "posterUrl": "/posters/x/2083136052638998714.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I did NOT expect this level of realism from an AI-generated ad. 😳\n\nMade using MiniMax H3 by @Hailuo_AI \n\nH3 is a general-purpose multimodal video model that goes far beyond single-purpose tools. It understands text, images, video, and audio together, allowing it to better understand creative intent and generate natural, coherent audiovisual content in one end-to-end workflow.\n\nOne detail that really stood out to me was how well it handles on-screen text. The typography stays sharp, clear, and consistent throughout the video, with little to no distortion or deformation—something that's still a challenge for many AI video models.\n\n#MiniMaxH3\n\nPrompt:\nCHARACTER & WORLD LOCK + FULL 15-SEC MASTER PROMPT\n\nA 15-second continuous UGC-style vlog in 9:16 vertical format. Jessy, a woman in her mid-20s — use her facial features exactly as shown in her reference photo, keep everything else about the scene freshly designed — wears an attractive pastel-colored dress (soft lavender, mint, or buttery yellow) that visually pops in contrast against the bubblegum-pink Olipop can. Location: a bright, clean modern kitchen/countertop with soft natural daylight. Product: a can of Olipop Blackberry Vanilla — pastel pink can, magenta blackberry illustration, white vanilla flower, dark green \"Blackberry Vanilla\" wordmark, \"Supports Digestive Health\" text near the top, \"A New Kind of Soda™\" tagline (use the uploaded product photo as the exact visual reference for the can). Style: authentic smartphone UGC vlog, natural handheld feel, realistic skin texture, no over-retouched look. She speaks casual, upbeat, conversational English directly to camera with natural lip-sync.\n\nThe video follows this exact beat sheet, with clean cuts between each moment:\n\n00:00–00:02 — Front-camera selfie, slight handheld sway, Jessy fans herself lightly in the kitchen, looking for something refreshing, smiles at camera: \"Okay I need something fun to drink right now.\"\n\n00:02–00:04 — Eye-level medium shot, she opens the fridge and pulls out the pink Olipop can, face lighting up with delight: \"Ooh, Blackberry Vanilla Olipop — yes please.\"\n\n00:04–00:06 — Close-up on her hands slowly turning the can, label details (\"Prebiotics, Botanics, Plant Fiber\") visible: \"Prebiotics AND it tastes like dessert? Sold.\"\n\n00:06–00:08 — Tight macro shot, her fingers pop the tab, condensation and a light fizz mist visible: \"Listen to that fizz.\"\n\n00:08–00:10 — Slow-motion macro shot, pink soda pouring over ice into a clear glass, bubbles rising: \"Look how pretty that color is.\"\n\n00:10–00:12 — Close-up on her face as she sips, eyes lighting up with genuine, pleasantly surprised reaction: \"Wait... that's actually so good.\"\n\n00:12–00:13 — Macro hero shot of the can alone on the counter, condensation dripping, soft bokeh kitchen background, slow subtle push-in, no dialog.\n\n00:13–00:14 — Medium shot, she smiles at camera, can raised slightly at chest height: \"Blackberry vanilla was the right call.\"\n\n00:14–00:15 — Close front-camera selfie, can held beside her face, warm bright smile into the lens: \"Go grab an Olipop, seriously.\"\n\nContinuous natural ambient kitchen sound, subtle upbeat pop background music underneath, her voice clear and casual throughout. Consistent face, outfit, and setting across every beat. No text overlays, no captions, no watermarks, no horizontal framing, no changes to the can's design or branding.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083132770650571041",
    "title": "竹林武侠谜案",
    "titleEn": "Bamboo Forest Wuxia Mystery",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《竹林武侠谜案》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Bamboo Forest Wuxia Mystery. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "A 16:9 cinematic wuxia mystery set in a bamboo forest at night. Use a low-saturation palette of cold blue, ink green, charcoal, and gray. Thin mist fills the forest and fine snow drifts through the air. The mood is austere, lethal, and controlled, with the tension of a martial-arts sect investigating a case and exchanging secret intelligence.\n\nDeep in the forest, dense vertical bamboo fills the background while cold white mist-light glows in the distance. Soft, out-of-focus leaves partially obscure the foreground, creating the sense of watching from within the grove. A cool, soft front-side key lights the actors’ faces; backlight keeps the foreground dark and the distance luminous. Use shallow depth of field so leaves, snow, and bamboo dissolve into soft bokeh.\nPrioritize facial close-ups and measured shot/reverse-shot coverage. Keep the rhythm restrained but tense. Photoreal period-drama production value, cinematic lighting, and no modern elements. No subtitles, on-screen text, watermarks, modern clothing or architecture, animation styling, over-smoothed skin, bright daylight, or comic performance.",
    "promptSourceUrl": "https://x.com/AIwithAliya/status/2083132770650571041",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AIwithAliya/status/2083132770650571041",
    "sourceLabel": "X 原帖 · @AIwithAliya",
    "author": "Aaliya",
    "publishedAt": "2026-07-31T10:08:49.915Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083132770650571041.mp4",
    "posterUrl": "/posters/x/2083132770650571041.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tried MiniMax H3 on @Hailuo_AI second time.\n\nprompt\n\nA 16:9 cinematic wuxia mystery set in a bamboo forest at night. Use a low-saturation palette of cold blue, ink green, charcoal, and gray. Thin mist fills the forest and fine snow drifts through the air. The mood is austere, lethal, and controlled, with the tension of a martial-arts sect investigating a case and exchanging secret intelligence.\n\nDeep in the forest, dense vertical bamboo fills the background while cold white mist-light glows in the distance. Soft, out-of-focus leaves partially obscure the foreground, creating the sense of watching from within the grove. A cool, soft front-side key lights the actors’ faces; backlight keeps the foreground dark and the distance luminous. Use shallow depth of field so leaves, snow, and bamboo dissolve into soft bokeh.\nPrioritize facial close-ups and measured shot/reverse-shot coverage. Keep the rhythm restrained but tense. Photoreal period-drama production value, cinematic lighting, and no modern elements. No subtitles, on-screen text, watermarks, modern clothing or architecture, animation styling, over-smoothed skin, bright daylight, or comic performance.\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083090068378616089",
    "title": "黄金守卫网页首屏循环",
    "titleEn": "Golden Guardian Web Hero Loop",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《黄金守卫网页首屏循环》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 30 秒 · 1280×720。",
    "summaryEn": "An H3 output for Golden Guardian Web Hero Loop. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 30s · 1280×720.",
    "prompt": "@Hailuo_AI Prompt #MiniMaxH3 in Chinese:\n\n15秒 | 16:9 | 1440p | 24fps | 无缝循环 | 网页首屏动画\n\n[人物锁定 · @image1] 完全按照该参考图呈现，不进行任何重新设计。金色卷发、闭合的双眼、仰起的面容与安详神情；象牙色垂坠长袍；左右两枚金色玫瑰花章肩甲，含中心宝石与放射浮雕；金丝胸饰、十字垂饰、多层珠链；水晶巨剑，含金色巴洛克护手与剑柄头、缠绕剑柄，双臂高举过头横持，金色火柱自上贯入；白色大理石圆台；下方云海；深蓝天空；身后刻满金色符文的巨大光环与放射状水晶碎片及其硬边 RGB 衍射。全片不得改动发型、面部、肩甲浮雕、胸饰、长袍轮廓、护手纹样与剑身比例。\n她的双眼全片闭合，仅在 11.4 至 12.6 秒睁开，并在 15.0 秒前再次闭合。\n\n[字形锁定 · @image2] 本片所有拉丁文字必须完全采用 @image2 的字形——相同的衬线结构、凿刻感收笔、字宽比例与字距，不得发明或替换字体。材质取自 @image2：抛光铬银，每一笔画的左上缘带温暖琥珀色轮廓光，细微倒角，细腻颗粒。\n\n[解剖锁定 · 最高优先级] 全片自始至终只有两条手臂、两只手、每只手五指、一把剑。严禁多余的手臂、手掌、手指或第二把武器。快速旋转与运动模糊中，残影绝不可被渲染成额外肢体；手部每一帧结构完整、指节分明、无粘连。剑柄同时只被该拍指定的手握持；空出的手须明确可见且有清楚朝向。\n\n[网页 UI/UX 系统 · 全片骨架] 画面按 12 栏网格构建，左右外边距各占画幅宽度 6%，栏间距 2%。\n安全区：第 1 至 5 栏留空，供网页标题与行动按钮使用。人物、剑与光环全程不得进入第 1 至 4 栏。\n视窗框：铬色发丝级四角框线于 1.4 秒延展到位并保持，每次转场轻微重新吸附，绝不重置。\n载入指示：光环的进度弧兼作页面载入条——1.4 秒起步，7.8 秒约半圈，11.4 秒闭合爆亮，12.6 秒后逆向清空。\n副环：三层同心符文环于 3.2 秒依次向外展开，以不同速度反向旋转。\n准星：追踪剑身并随她运动持续修正，撞击时收紧。\n数据列：画面右缘细长字形条滚动，速度随配乐加快。\n计数器：持续攀升，9.6 秒跳动异常，11.4 秒爆红归零。\n频谱条：环外一圈径向条形随音频脉动。\n所有界面元素为铬色发丝线条与薄光，半透明，被水晶折射，统一笔画粗细，绝不越过面部，绝不进入第 1 至 4 栏。\n缓动：所有界面与文字元素以缓出曲线入场，约 0.4 秒；退场约 0.25 秒。绝不线性，绝不弹跳。多元素之间错位约 0.06 秒。\n\n[文字 · 稀少且为实体，绝非叠加] 全片只有四处文字，每一处都是场景中真实存在的物体，具备厚度、倒角、投影与光照响应，随镜头产生透视变形，被前景遮挡，被水晶折射。\n以下引号内为画面实际渲染的文字，须原样呈现，不得翻译、不得替换、不得增删：\n1.8秒 光环符文带上一段符文重组为 @image2 的字形 \"ASCENSION\"，呈深铬阴刻切入环体，随环旋转出画。\n4.6秒 \"SCROLL\" 作为小号铬色标签，下方带一枚细箭头，居于画面下缘第 6 至 7 栏，轻微脉动。\n12.8秒 首屏版式于第 1 至 5 栏组装：顶部一条铬色发丝导航横线与三个圆点；显示级 \"ASCENSION\"；其下一枚发丝圆角按钮内含小号 \"ENTER\"。保持 0.8 秒后整体溶解。\n14.2秒 \"CIA0\" 以深铬阴刻切入符文带，刻痕深度、笔画宽度、边缘磨损与氧化程度与周围符文完全一致。绝不更亮、绝不更大、无发光、无描边。它随环体弧面产生透视变形，被前方水晶碎片局部遮挡，仅在侧光扫过的一瞬显现，在正光下几乎重新隐没于金属之中。\n全片不出现任何日文字符、字幕或其他文字。\n\n[色彩] 天空自顶部深海军蓝 #0B2350 至宝石蓝 #12356B，地平线转暖白 #F0E4CE。火柱核心过曝白，向外为浅金 #FFE9B0、琥珀 #FFD98A、橙金 #E8A63C，最外缘半透明热扰动。剑身自剑尖冷青 #9FD8E0 渐变至护手暖金，中段穿插全光谱色散带沿剑身流动。长袍受光面象牙白 #F5EDDC 至暖灰米 #C9BBA4，褶皱深处冷蓝影 #8C93A6。金属高光 #FFF0C0、中调 #E0B454、暗部 #8A6220、最暗带绿调，绝不平涂。日冕自内向外为白、金、琥珀、品红、靛蓝、透明六段。云海暖白 #FFF6E6 至灰蓝 #7E96B4 至紫蓝 #4A5A85。铬色文字与界面 #E8ECEF，带暖琥珀 #D9A05B 轮廓光。全光谱仅以折射形式出现，绝不平涂。\n\n[风格与特效] 电影级手绘感 2.5D 动画，巴洛克式密度。胶片高光滚降，阴影提亮，最亮处柔和光晕，细腻颗粒，饱和浓郁不过曝。皮肤具次表面散射；金色为各向异性拉丝金属；水晶具真实色散；织物为多层布料解算，有惯性与回弹；头发逐丝渲染，逆光半透明。\n特效逐级升级：碎片投出硬边 RGB 衍射光带；金色高光上的星芒光斑；火柱以翻腾丝带状降下；上帝之光穿透云海；碎片剥落光之羽；棱镜焦散在肩甲与胸饰上爬行；圆台由内发光；可见的冲击波畸变环；日冕绽放时短暂过曝再回落；空气中始终悬浮尘埃与晶屑。\n\n[摄影] 变形宽银幕，椭圆焦外，蓝金双色条状光斑，180 度快门，轻微桶形畸变。每个镜头都有明确运动动机与前中后三层纵深——云絮、光之羽、晶屑或长袍一角始终位于前景制造视差。人物始终位于第 6 至 12 栏内。\n0.0-1.4\n\n@Hailuo_AI @Image1 微距置于水晶剑身内部，在晶体裂隙间穿行，光沿裂纹色散，随后自剑刃穿出进入开阔天空。\n1.4-3.2 摇臂沿光环内缘攀升，符文自近及远依次点亮，透视强烈压缩。\n3.2-4.6 35mm 弧形轨道绕至碎片之后，全光谱扫过镜头，她以逆光剪影浮现。\n4.6-6.0 40mm 低角度缓推，火柱自画面顶部入画与剑尖相接，构图收束为首屏比例。\n6.0-7.8 50mm 空中环绕，随她 180 度旋转同步绕行，斩击瞬间升格至四倍慢速再回落。\n7.8-9.6 18mm 一镜到底随她坠落穿过光环，穿越时全光谱填满画面，绝不切。\n9.6-11.4 焦段随拍点跳切，85mm 特写与 18mm 广角交替，中间插入一次甩镜。\n11.4-12.6 摇臂升降加滑动变焦，由低仰角升至面部齐平，背景压缩感随之改变。\n12.6-15.0 缓慢后拉，随后沿剑身推进再次进入水晶内部，与第一帧完全对位。\n开场与收束段一律使用带光晕的溶解与镜头位移，绝不硬切；仅 9.6 至 11.4 秒使用硬切并加静音门。\n\n[分镜]\n0.0-1.4 循环点。镜头在水晶剑身内部穿行，裂纹、气泡与色散缓缓掠过，随后自剑刃穿出，天空豁然展开。四角框线开始延展。\n1.4-3.2 点火。摇臂沿环内缘攀升，火沿环周逐个点亮符文奔跑一圈。1.8 秒处一段符文重组为深铬阴刻的 \"ASCENSION\" 并随环转出画。进度弧起步。\n3.2-4.6 浮现。轨道绕至碎片之后，全光谱扫过镜头，她以剪影浮现，双手同握剑柄将剑垂于身前。三层符文环展开，数据列开始滚动。\n4.6-6.0 举剑。双手将巨剑举过头顶横持，回到 @image1 姿态，下颌微仰，火柱降下与剑尖相接。\"SCROLL\" 落位于画面下缘。构图锁定为首屏比例。\n6.0-7.8 接光。她将剑举迎向降下的火柱——由剑身承接，双手始终紧握剑柄绝不松手——随即全身旋转 180 度带出一记横斩。棱镜光爆炸开，UI 全线爆红。双眼仍闭合。\n7.8-9.6 坠落。她背向下坠穿过光环中心，右手单手持剑于身侧缓慢旋转，左手向上伸展，长袍如伞面张开。一镜到底，穿越瞬间全光谱填满画面。\n9.6-11.4 蒙太奇，八拍每拍 0.225 秒：一、白底铬色线稿，仅剑与双臂轮廓；二、平涂光谱色块拼贴；三、二色剪影，象牙白配纯金；四、完整渲染，一记真实的空中回旋；五、拖影帧，拉出长条残像；六、整帧负片反转，仅此一帧；七、铬色线稿叠加彩虹衍射；八、完整渲染并短暂静止。光与云的扰动八拍之间持续累积，绝不重置。\n11.4-12.6 顶点。左手回握剑柄，双手同握将剑锋钉入虚空，冲击波向外扩散。她睁开双眼——青碧色，平静，直视镜头。睫毛先分离，眼睑上抬，瞳孔收缩后放大，泪膜捕捉火柱高光，眼角一道光。日冕绽放，进度弧闭合爆亮。\n12.6-15.0 归位。她上升落回圆台，双臂举剑横于头顶回到 @image1 姿态，各环反向折叠，日冕收敛为一道金线，双眼闭合。首屏版式于左区组装、保持、溶解。\"CIA0\" 随环转入画面，被侧光扫过而显现，随后转入正光重新隐没。镜头沿剑身推进，再次进入水晶内部。最后一帧与第一帧完全一致。\n\n[动态] 重量与优雅优先于摆姿势，动作连贯绝不生硬。长袍、卷发与垂带在每次转向时滞后，布料有真实惯性与回弹。巨剑沉重，挥动时可见承力与减速。一切略带高空浮力感。云层流动与人物动作互相影响。杜绝任何僵硬、停顿或不自然的关节角度。\n\n[音频] 每个音效卡在其画面拍点的准确帧上，混音首尾相接可循环。底层：高空空气感，远处风声，极淡持续人声。\n0.0-1.4 水晶内部的封闭共鸣，极轻的晶体应力声，穿出剑刃时空气骤然开阔。\n1.4-3.2 符文逐个点亮的清脆金属点火声，沿环奔跑形成加速连音，合唱自远处涌起，铬色文字落定时一记低音。\n3.2-4.6 全光谱扫过的高频泛音，布料舒展的柔和摩擦，副环展开的层叠环鸣。\n4.6-6.0 举剑的低频蓄势，火柱降临的持续嗡鸣，剑尖相接时一声清亮长音。\n6.0-7.8 撞击。管风琴与超低频同时落下，玻璃爆裂，反向镲片渐强，警报般明亮的界面爆红音。\n7.8-9.6 坠落气压呼啸，入环骤然滤波为水下感合唱，穿出时爆发大空间混响。\n9.6-11.4 八记打击声分别卡在八拍上，合唱断奏，每拍之间硬静音门，负片帧加一记反相金属噪音。\n11.4-12.6 顶点。完整合唱、管风琴、钟声与一记低沉铜管齐鸣，睁眼那一帧落下一声清亮的钟，混音随日冕展宽。\n12.6-15.0 收束为延长的合唱长音，各环以相反节奏收拢，最终回到与开场完全相同的水晶内部封闭共鸣。\n无对白。\n\n[限制] 画面中只有一个人物，两条手臂，两只手，一把剑。人物绝不进入第 1 至 4 栏。所有拉丁文字须完全匹配 @image2 的字形。\"CIA0\" 须准确拼写，绝不发光、描边、放大或比周围符文更亮。光谱仅以折射形式出现。界面层保持铬色发丝质感，绝不越过面部。特效持续累积直至归位段。双眼仅在 11.4 至 12.6 秒睁开。无日文字符、无品牌标识、无水印、无字幕。首尾帧完全一致，两帧中都不得残留任何浮动文字。.",
    "promptSourceUrl": "https://x.com/Cia0_exe/status/2083090227166658684",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Cia0_exe/status/2083090068378616089",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-07-31T07:19:08.900Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083090068378616089.mp4",
    "posterUrl": "/posters/x/2083090068378616089.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Fashion",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "UI/UX for animation your portfolio \n\nmade it with MiniMax H3 Ai @Hailuo_AI X midjourney V8.2 \n\nBuilt as a website hero, twelve column grid\n\nleft five columns kept clear for the headline.\n\nThe halo is the page loader.\n\n#MiniMaxH3 handles native audio & on-screen text + a character lock in ONE generation\n\nthat combo is where it pulls ahead of sd2 for me.\n\n4,367 characters of Chinese prompt",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083081579262591469",
    "title": "翡翠生物精华产品片",
    "titleEn": "Emerald Bio-Serum Product Film",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《翡翠生物精华产品片》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×1280。",
    "summaryEn": "An H3 output for Emerald Bio-Serum Product Film. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×1280.",
    "prompt": "Cinematic luxury skincare product commercial, vertical 9:16, 15 seconds. A translucent emerald green glass dropper bottle of INVENTIKO Bio-Serum 38% Renderfeel 20,000, labeled \"SERUM SKINCARE - SKIN REPAIR SKIN LIFTING\", sits on a sleek black and white pedestal against a dark teal gradient background with soft volumetric lighting and dramatic light rays. Smooth camera orbit and subtle zoom around the bottle, water droplets on glass, elegant reflections. Close-up macro shots of the bottle surface and text. Then a black dropper is slowly pulled out, filled with golden-yellow serum, and a perfect viscous drop forms and hangs at the tip in extreme slow-motion close-up against soft green bokeh. Quick transition to a soft-focus close-up of a young East Asian woman's glowing cheek and eye as a single golden serum droplet falls onto her skin, creating a delicate shimmering highlight. Final shot: the green bottle with dropper resting beside it on a white terrazzo surface against a green tiled wall with soft window light stripes. Ultra-realistic, high-end beauty advertising style, soft cinematic lighting, shallow depth of field, 4K quality, smooth camera movements.",
    "promptSourceUrl": "https://x.com/noorlewisx/status/2083081579262591469",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/noorlewisx/status/2083081579262591469",
    "sourceLabel": "X 原帖 · @noorlewisx",
    "author": "Noor",
    "publishedAt": "2026-07-31T06:45:24.937Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083081579262591469.mp4",
    "posterUrl": "/posters/x/2083081579262591469.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Creating a luxury skincare serum commercial with MiniMaxH3 on @Hailuo_AI was truly an incredible experience.\n\nThe cinematic lighting, refined product shots, and premium visual quality made every frame feel like a high-end beauty campaign.\n#MiniMaxH3\n\nPrompt:\n\nCinematic luxury skincare product commercial, vertical 9:16, 15 seconds. A translucent emerald green glass dropper bottle of INVENTIKO Bio-Serum 38% Renderfeel 20,000, labeled \"SERUM SKINCARE - SKIN REPAIR SKIN LIFTING\", sits on a sleek black and white pedestal against a dark teal gradient background with soft volumetric lighting and dramatic light rays. Smooth camera orbit and subtle zoom around the bottle, water droplets on glass, elegant reflections. Close-up macro shots of the bottle surface and text. Then a black dropper is slowly pulled out, filled with golden-yellow serum, and a perfect viscous drop forms and hangs at the tip in extreme slow-motion close-up against soft green bokeh. Quick transition to a soft-focus close-up of a young East Asian woman's glowing cheek and eye as a single golden serum droplet falls onto her skin, creating a delicate shimmering highlight. Final shot: the green bottle with dropper resting beside it on a white terrazzo surface against a green tiled wall with soft window light stripes. Ultra-realistic, high-end beauty advertising style, soft cinematic lighting, shallow depth of field, 4K quality, smooth camera movements.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083079304632520833",
    "title": "黑金奢华香水广告",
    "titleEn": "Black-and-Gold Perfume Commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《黑金奢华香水广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Black-and-Gold Perfume Commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create a 15-second ultra-premium luxury perfume commercial in cinematic 16:9.\n\nA crystal perfume bottle with a sleek black-and-gold design stands on a glossy reflective surface inside a dark luxury studio. The commercial opens with an extreme macro shot of the engraved cap, followed by slow cinematic rotations revealing every elegant detail. Dramatic golden light sweeps across the glass while soft volumetric light and subtle mist create a premium atmosphere.\n\nElegant water splashes, floating crystal shards, and shimmering golden particles move naturally around the bottle in ultra slow motion without covering the label. Every reflection is physically accurate with realistic caustics, premium glass refractions, and highly detailed liquid simulation.\n\nThe camera transitions through smooth macro close-ups, low-angle hero shots, orbiting movements, and dramatic push-ins using luxury advertising cinematography. Rich black marble, gold accents, and soft smoke enhance the high-end aesthetic.\n\nEnd with the bottle perfectly centered on a glossy pedestal as the lighting intensifies for a premium hero reveal. Display a fictional luxury brand logo \"NOIR ÉLITE\" with the tagline \"Crafted to Leave a Mark.\"\n\nPhotorealistic, Hollywood luxury TV commercial, ultra-detailed, flawless glass rendering, realistic water physics, cinematic color grading, HDR, ray-traced reflections, depth of field, premium product advertising, 8K quality.",
    "promptSourceUrl": "https://x.com/kingofdairyque/status/2083079304632520833",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/kingofdairyque/status/2083079304632520833",
    "sourceLabel": "X 原帖 · @kingofdairyque",
    "author": "Simply Ray",
    "publishedAt": "2026-07-31T06:36:22.623Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083079304632520833.mp4",
    "posterUrl": "/posters/x/2083079304632520833.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @Hailuo_AI continues to impress with premium product commercials.\nThis luxury perfume ad combines realistic character consistency, elegant camera movement, cinematic lighting, and stunning product close-ups. From the crystal bottle reflections to the final hero shot, every frame feels like a high-end beauty campaign.\nPrompt 👇\nCreate a 15-second ultra-premium luxury perfume commercial in cinematic 16:9.\n\nA crystal perfume bottle with a sleek black-and-gold design stands on a glossy reflective surface inside a dark luxury studio. The commercial opens with an extreme macro shot of the engraved cap, followed by slow cinematic rotations revealing every elegant detail. Dramatic golden light sweeps across the glass while soft volumetric light and subtle mist create a premium atmosphere.\n\nElegant water splashes, floating crystal shards, and shimmering golden particles move naturally around the bottle in ultra slow motion without covering the label. Every reflection is physically accurate with realistic caustics, premium glass refractions, and highly detailed liquid simulation.\n\nThe camera transitions through smooth macro close-ups, low-angle hero shots, orbiting movements, and dramatic push-ins using luxury advertising cinematography. Rich black marble, gold accents, and soft smoke enhance the high-end aesthetic.\n\nEnd with the bottle perfectly centered on a glossy pedestal as the lighting intensifies for a premium hero reveal. Display a fictional luxury brand logo \"NOIR ÉLITE\" with the tagline \"Crafted to Leave a Mark.\"\n\nPhotorealistic, Hollywood luxury TV commercial, ultra-detailed, flawless glass rendering, realistic water physics, cinematic color grading, HDR, ray-traced reflections, depth of field, premium product advertising, 8K quality.\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083075282186256521",
    "title": "晨间唇油 UGC 体验分享",
    "titleEn": "Morning Lip Oil UGC Testimonial",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《晨间唇油 UGC 体验分享》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Morning Lip Oil UGC Testimonial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use the uploaded reference image as the exact character reference. Preserve her facial identity, hairstyle, eye color, skin tone, body proportions, and facial consistency throughout every shot.\n\nCreate a 15-second ultra-realistic UGC-style beauty advertisement featuring a realistic young woman speaking directly to the smartphone camera in a bright, modern bedroom vanity area with warm morning sunlight, soft neutral décor, and an elegant illuminated mirror.\n\nShe smiles naturally while holding a premium luxury lip oil with a clear glass bottle and gold cap. In a friendly, conversational tone, she shares why it has become her favorite everyday beauty essential. She opens the product, applies the lip oil with the applicator in front of the mirror, then gently presses her lips together and smiles confidently, showing the glossy natural finish.\n\nCapture authentic handheld selfie shots, over-the-shoulder mirror angles, and cinematic close-ups highlighting the crystal-clear bottle, gold cap, soft reflections, applicator, glossy texture, and luxurious packaging. End with her holding the lip oil beside her face while smiling naturally into the camera.\n\nAuthentic UGC creator content, realistic facial expressions, natural hand gestures, healthy glowing skin, minimal makeup, luxury beauty aesthetic, smartphone-shot realism, subtle depth of field, premium color grading, vertical 9:16, 4K HDR, no text, subtitles, logos, or watermarks.",
    "promptSourceUrl": "https://x.com/SimplyAnnisa/status/2083075282186256521",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/SimplyAnnisa/status/2083075282186256521",
    "sourceLabel": "X 原帖 · @SimplyAnnisa",
    "author": "Anissa",
    "publishedAt": "2026-07-31T06:20:23.597Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083075282186256521.mp4",
    "posterUrl": "/posters/x/2083075282186256521.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "ugc"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @Hailuo_AI keeps raising the bar for realistic UGC content.\nThis beauty ad showcases natural facial expressions, lifelike skin textures, authentic smartphone-style camera movement, and convincing product interactions. The result feels polished, believable, and ready for real social media campaigns.\n\nPrompt:\nUse the uploaded reference image as the exact character reference. Preserve her facial identity, hairstyle, eye color, skin tone, body proportions, and facial consistency throughout every shot.\n\nCreate a 15-second ultra-realistic UGC-style beauty advertisement featuring a realistic young woman speaking directly to the smartphone camera in a bright, modern bedroom vanity area with warm morning sunlight, soft neutral décor, and an elegant illuminated mirror.\n\nShe smiles naturally while holding a premium luxury lip oil with a clear glass bottle and gold cap. In a friendly, conversational tone, she shares why it has become her favorite everyday beauty essential. She opens the product, applies the lip oil with the applicator in front of the mirror, then gently presses her lips together and smiles confidently, showing the glossy natural finish.\n\nCapture authentic handheld selfie shots, over-the-shoulder mirror angles, and cinematic close-ups highlighting the crystal-clear bottle, gold cap, soft reflections, applicator, glossy texture, and luxurious packaging. End with her holding the lip oil beside her face while smiling naturally into the camera.\n\nAuthentic UGC creator content, realistic facial expressions, natural hand gestures, healthy glowing skin, minimal makeup, luxury beauty aesthetic, smartphone-shot realism, subtle depth of field, premium color grading, vertical 9:16, 4K HDR, no text, subtitles, logos, or watermarks.\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083065653104107776",
    "title": "嘻哈角色对口型表演",
    "titleEn": "Hip-Hop Character Lip-Sync Performance",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《嘻哈角色对口型表演》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 108 秒 · 1280×720。",
    "summaryEn": "An H3 output for Hip-Hop Character Lip-Sync Performance. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 108s · 1280×720.",
    "prompt": "hip hop music video, make the character from @Video1 dance to the beat from music @Audio1 and he lip syncs in sync with @Audio1",
    "promptSourceUrl": "https://x.com/bennash/status/2083065655390031980",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/bennash/status/2083065653104107776",
    "sourceLabel": "X 原帖 · @bennash",
    "author": "Ben Nash",
    "publishedAt": "2026-07-31T05:42:07.845Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083065653104107776.mp4",
    "posterUrl": "/posters/x/2083065653104107776.jpg",
    "duration": 108,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Tomorrow Is A New Day\" - Let yesterday decay.\n\nI'm now honestly totally blown away with H3. It made my character dance to the beat and lip-sync as requested. I've never used a tool and model that could do this so easily at this level. H3 Prompt below. \n\nMade with Hailuo AI H3\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083064417798025721",
    "title": "现代战争 FPS 实机画面",
    "titleEn": "Modern warfare FPS gameplay",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《现代战争 FPS 实机画面》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Modern warfare FPS gameplay. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Camera: First-person perspective at eye level with authentic handheld player movement, as if recorded directly from a modern AAA military shooter. The player carries a highly detailed assault rifle with realistic animations, visible hands, tactical gloves, dynamic reload mechanics, and weapon sway.  **Opening Action:** The video immediately begins with the player already aiming down a roadway inside a modern military base. Multiple enemy soldiers are visible in the distance near sandbags, barricades, and military vehicles. The player carefully tracks one target, making small aim corrections while maintaining ADS (aim down sights). Fire several controlled bursts immediately at the visible enemies, producing realistic muzzle flashes, shell casings ejecting, smoke, recoil, hit reactions, and dust impacts around the targets. Continue firing in multiple short bursts while adjusting aim between enemies, simulating authentic FPS gameplay rather than scripted animation.  **Movement:** After the opening firefight, lower slightly from ADS and begin advancing cautiously along the road beside concrete barriers, Hesco walls, and parked military vehicles. Frequently check left and right corners, briefly stop to reacquire targets, then raise the weapon and fire additional controlled bursts whenever enemies appear ahead. Continue pushing forward with deliberate player-controlled movement, using cover naturally and maintaining believable tactical pacing.  **Environment:** Large modern military base with guard towers, armored vehicles, shipping containers, blast barriers, damaged buildings, smoke plumes, burning debris, scattered shell casings, dust clouds, and atmospheric battlefield haze. Cool natural daylight mixed with smoke and orange firelight creates a cinematic battlefield atmosphere.  **Camera Motion:** Authentic player-controlled movement with subtle head bob, weapon sway, natural mouse-look adjustments, small left-right corrections while aiming, realistic recoil, smooth tracking of moving targets, brief pauses before shooting, and fluid forward progression. Avoid cinematic camera moves—everything should feel like genuine live gameplay captured by a skilled player.  **Visual Quality:** Ultra-photorealistic, AAA game graphics with realistic PBR materials, detailed weapon models, physically accurate lighting, volumetric smoke, dynamic particle effects, crisp textures, realistic bullet impacts, muzzle flash illumination, motion blur only during rapid movement, and high-end military shooter presentation.  **Gameplay UI:** Display a realistic modern FPS HUD inspired by games like PUBG, Battlefield, or Call of Duty (without copying exact copyrighted assets). Include:  * Central dynamic crosshair or reticle * Ammo counter with magazine and reserve ammunition * Fire mode indicator * Compass at the top * Squad/team status panel * Mini-map in the upper corner * Health bar * Tactical equipment icons (grenades, medkit) * Hit markers when bullets connect * Directional damage indicators * Kill notification feed * Objective marker in the distance * Subtle interaction prompts and realistic HUD animations  The HUD should feel polished, modern, and fully integrated into the gameplay, enhancing the illusion of authentic recorded footage from a contemporary military FPS.",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2083064417798025721",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Just_sharon7/status/2083064417798025721",
    "sourceLabel": "X 原帖 · @Just_sharon7",
    "author": "Sharon Riley",
    "publishedAt": "2026-07-31T05:37:13.325Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083064417798025721.mp4",
    "posterUrl": "/posters/x/2083064417798025721.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Gaming",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "commercial"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @Hailuo_AI feels like a major step forward for AI video creation.  \n\nFrom cinematic storytelling and product commercials to UI mockups, game-inspired scenes, and branded content, it handles a wide range of creative workflows with impressive consistency.  \n\nPOV Modern Warfare FPS Gameplay Prompt  \n\nCamera: First-person perspective at eye level with authentic handheld player movement, as if recorded directly from a modern AAA military shooter. The player carries a highly detailed assault rifle with realistic animations, visible hands, tactical gloves, dynamic reload mechanics, and weapon sway.  **Opening Action:** The video immediately begins with the player already aiming down a roadway inside a modern military base. Multiple enemy soldiers are visible in the distance near sandbags, barricades, and military vehicles. The player carefully tracks one target, making small aim corrections while maintaining ADS (aim down sights). Fire several controlled bursts immediately at the visible enemies, producing realistic muzzle flashes, shell casings ejecting, smoke, recoil, hit reactions, and dust impacts around the targets. Continue firing in multiple short bursts while adjusting aim between enemies, simulating authentic FPS gameplay rather than scripted animation.  **Movement:** After the opening firefight, lower slightly from ADS and begin advancing cautiously along the road beside concrete barriers, Hesco walls, and parked military vehicles. Frequently check left and right corners, briefly stop to reacquire targets, then raise the weapon and fire additional controlled bursts whenever enemies appear ahead. Continue pushing forward with deliberate player-controlled movement, using cover naturally and maintaining believable tactical pacing.  **Environment:** Large modern military base with guard towers, armored vehicles, shipping containers, blast barriers, damaged buildings, smoke plumes, burning debris, scattered shell casings, dust clouds, and atmospheric battlefield haze. Cool natural daylight mixed with smoke and orange firelight creates a cinematic battlefield atmosphere.  **Camera Motion:** Authentic player-controlled movement with subtle head bob, weapon sway, natural mouse-look adjustments, small left-right corrections while aiming, realistic recoil, smooth tracking of moving targets, brief pauses before shooting, and fluid forward progression. Avoid cinematic camera moves—everything should feel like genuine live gameplay captured by a skilled player.  **Visual Quality:** Ultra-photorealistic, AAA game graphics with realistic PBR materials, detailed weapon models, physically accurate lighting, volumetric smoke, dynamic particle effects, crisp textures, realistic bullet impacts, muzzle flash illumination, motion blur only during rapid movement, and high-end military shooter presentation.  **Gameplay UI:** Display a realistic modern FPS HUD inspired by games like PUBG, Battlefield, or Call of Duty (without copying exact copyrighted assets). Include:  * Central dynamic crosshair or reticle * Ammo counter with magazine and reserve ammunition * Fire mode indicator * Compass at the top * Squad/team status panel * Mini-map in the upper corner * Health bar * Tactical equipment icons (grenades, medkit) * Hit markers when bullets connect * Directional damage indicators * Kill notification feed * Objective marker in the distance * Subtle interaction prompts and realistic HUD animations  The HUD should feel polished, modern, and fully integrated into the gameplay, enhancing the illusion of authentic recorded footage from a contemporary military FPS.\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083059583308751079",
    "title": "奢华香水商业片",
    "titleEn": "Luxury perfume commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《奢华香水商业片》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Luxury perfume commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Scene 1 (0–3s) – Luxury Reveal\nA luxury perfume bottle slowly emerges from darkness, standing on a glossy black pedestal. Soft golden light beams gradually reveal the crystal-clear glass, while subtle volumetric fog creates depth. Extreme macro shot, cinematic lighting, ultra-realistic reflections, premium luxury commercial aesthetic.\n\nScene 2 (3–6s) – Precision Details\nExtreme close-up of the perfume bottle as rich golden fragrance liquid gently swirls inside. The camera performs a smooth cinematic orbit, capturing intricate glass textures, flawless craftsmanship, and sparkling highlights. Ultra-detailed, photorealistic, premium commercial quality.\n\nScene 3 (6–9s) – AI Transformation\nThe perfume bottle seamlessly dissolves into glowing golden particles before instantly reforming into a futuristic luxury edition, maintaining perfect proportions and visual consistency. Dynamic orbit camera, seamless transformation, cinematic lighting, premium visual effects.\n\nScene 4 (9–12s) – Environment Transition\nWithout a single cut, the background transforms from a luxurious black studio into a futuristic neon metropolis. The perfume bottle remains perfectly consistent while the camera performs a dramatic dolly-in. Hyper-realistic, cinematic, high-end advertising style.\n\nScene 5 (12–15s) – Hero Finale\nSlow-motion hero shot of the luxury perfume bottle elegantly rotating as glowing golden particles float around it. Premium black background, flawless reflections, cinematic logo reveal, luxury commercial ending, 8K photorealistic, blockbuster advertising quality.",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2083059583308751079",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/CaliraVal/status/2083059583308751079",
    "sourceLabel": "X 原帖 · @CaliraVal",
    "author": "Calira",
    "publishedAt": "2026-07-31T05:18:00.693Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083059583308751079.mp4",
    "posterUrl": "/posters/x/2083059583308751079.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Creating a luxury perfume commercial with MiniMaxH3 on @Hailuo_AI was an amazing experience.\n\nThe cinematic lighting, elegant product shots, and premium visual quality made every scene feel like a high-end fragrance campaign.\n#MiniMaxH3\n\nPrompt:\n\nScene 1 (0–3s) – Luxury Reveal\nA luxury perfume bottle slowly emerges from darkness, standing on a glossy black pedestal. Soft golden light beams gradually reveal the crystal-clear glass, while subtle volumetric fog creates depth. Extreme macro shot, cinematic lighting, ultra-realistic reflections, premium luxury commercial aesthetic.\n\nScene 2 (3–6s) – Precision Details\nExtreme close-up of the perfume bottle as rich golden fragrance liquid gently swirls inside. The camera performs a smooth cinematic orbit, capturing intricate glass textures, flawless craftsmanship, and sparkling highlights. Ultra-detailed, photorealistic, premium commercial quality.\n\nScene 3 (6–9s) – AI Transformation\nThe perfume bottle seamlessly dissolves into glowing golden particles before instantly reforming into a futuristic luxury edition, maintaining perfect proportions and visual consistency. Dynamic orbit camera, seamless transformation, cinematic lighting, premium visual effects.\n\nScene 4 (9–12s) – Environment Transition\nWithout a single cut, the background transforms from a luxurious black studio into a futuristic neon metropolis. The perfume bottle remains perfectly consistent while the camera performs a dramatic dolly-in. Hyper-realistic, cinematic, high-end advertising style.\n\nScene 5 (12–15s) – Hero Finale\nSlow-motion hero shot of the luxury perfume bottle elegantly rotating as glowing golden particles float around it. Premium black background, flawless reflections, cinematic logo reveal, luxury commercial ending, 8K photorealistic, blockbuster advertising quality.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083056488122380671",
    "title": "圣家堂 FPV 穿越",
    "titleEn": "Sagrada Família FPV flight",
    "model": "MiniMax H3（来源标注）",
    "mode": "FL2VA",
    "summary": "《圣家堂 FPV 穿越》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×860。",
    "summaryEn": "An H3 output for Sagrada Família FPV flight. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×860.",
    "prompt": "Create a cinematic realistic high-speed drone video over Barcelona, Spain at golden hour, using the reference image as the exact location and flight-path guide. The drone starts very low above the Barcelona cityscape, almost skimming above the foreground rooftops and streets, facing La Sagrada Família. It accelerates quickly forward over the urban landscape, following the drawn path from the foreground toward the cathedral while maintaining a very low altitude. The drone then performs a dramatic climbing curve as La Sagrada Família fills the frame, executes a close cinematic orbit around the cathedral’s iconic spires, and after the orbit, pulls back while gaining altitude before tilting and opening toward the Barcelona skyline, Mediterranean Sea, Eixample grid, surrounding historic rooftops, golden-hour reflections, distant coastline, and glowing Barcelona atmosphere.\n\nThe red drawn line must not appear in the video; it is only a hidden flight-path reference. Add powerful emotional cinematic music: deep cello melody, dramatic percussion rhythm, orchestral strings, subtle choir texture, epic rising percussion, and a grand heroic crescendo as the drone completes its orbit around La Sagrada Família and reaches the final panoramic reveal. No text, no subtitles, no logos, no graphic overlays.@Image1",
    "promptSourceUrl": "https://x.com/Diplomeme/status/2083056488122380671",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Diplomeme/status/2083056488122380671",
    "sourceLabel": "X 原帖 · @Diplomeme",
    "author": "ᴍᴜʀᴘʜʏ",
    "publishedAt": "2026-07-31T05:05:42.743Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083056488122380671.mp4",
    "posterUrl": "/posters/x/2083056488122380671.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×860",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Travel",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This isn’t real drone footage. Or is it? \n\n15-second cinematic FPV flight over La sagrada Familia Barcelona. \n\nMade with GPT image 2 + Hailuo MiniMax H3 on @Hailuo_AI #MiniMaxH3 \n\nTry it with prompt : \n\nCreate a cinematic realistic high-speed drone video over Barcelona, Spain at golden hour, using the reference image as the exact location and flight-path guide. The drone starts very low above the Barcelona cityscape, almost skimming above the foreground rooftops and streets, facing La Sagrada Família. It accelerates quickly forward over the urban landscape, following the drawn path from the foreground toward the cathedral while maintaining a very low altitude. The drone then performs a dramatic climbing curve as La Sagrada Família fills the frame, executes a close cinematic orbit around the cathedral’s iconic spires, and after the orbit, pulls back while gaining altitude before tilting and opening toward the Barcelona skyline, Mediterranean Sea, Eixample grid, surrounding historic rooftops, golden-hour reflections, distant coastline, and glowing Barcelona atmosphere.\n\nThe red drawn line must not appear in the video; it is only a hidden flight-path reference. Add powerful emotional cinematic music: deep cello melody, dramatic percussion rhythm, orchestral strings, subtle choir texture, epic rising percussion, and a grand heroic crescendo as the drone completes its orbit around La Sagrada Família and reaches the final panoramic reveal. No text, no subtitles, no logos, no graphic overlays.@Image1",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083051609953903017",
    "title": "奢华腕表分镜广告",
    "titleEn": "Luxury Watch Storyboard Campaign",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《奢华腕表分镜广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Luxury Watch Storyboard Campaign. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Here`s the storyboard\n\nTITLE: Luxury Watch Commercial Storyboard\n\nCreate a premium single-page storyboard for a luxury watch commercial.\n\nEditorial presentation board.\n\n3:4 portrait.\n\nEight cinematic storyboard frames arranged in a clean luxury grid.\n\nMinimalist Swiss-inspired design.\n\nPremium typography.\n\nBlack, silver and champagne gold color palette.\n\nProfessional creative agency presentation.\n\nEach frame should look like a film storyboard while remaining highly realistic.\n\nFrame 1:\nLuxury mechanical watch inside an elegant leather presentation box.\n\nFrame 2:\nPresentation box opens revealing the watch with dramatic golden light.\n\nFrame 3:\nExtreme macro shot of the dial, hands, gears and sapphire crystal.\n\nFrame 4:\nElegant gentleman lifting the watch before a black-tie gala.\n\nFrame 5:\nWatch being fastened onto his wrist.\n\nFrame 6:\nClose-up of the sweeping second hand catching warm reflections.\n\nFrame 7:\nFashion editorial portrait with the watch as the hero.\n\nFrame 8:\nHero product shot on glossy black marble with floating metallic particles.\n\nLuxury presentation board.\n\nAgency-quality storyboard.\n\nMinimal annotations.\n\nWhite background.\n\n8K.\n\nUse storyboard as the complete creative direction, shot list, composition, pacing, and cinematic sequence.\n\nUse LuxuryWatch as the exact product reference. Preserve the watch's design, dial, case, leather strap, proportions, materials, and fine details consistently throughout every scene.\n\nCreate a world-class luxury watch commercial.",
    "promptSourceUrl": "https://x.com/I_amShiti/status/2083051633177686503",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/I_amShiti/status/2083051609953903017",
    "sourceLabel": "X 原帖 · @I_amShiti",
    "author": "Kuria | AI",
    "publishedAt": "2026-07-31T04:46:19.697Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083051609953903017.mp4",
    "posterUrl": "/posters/x/2083051609953903017.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I stopped thinking like a prompt writer and started thinking like a creative director.\n\nUsing Omni Reference, I built this luxury watch commercial with MiniMax H3, and seeing it bring the entire campaign together was the real test.\n\nCreated with @Hailuo_AI\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083048872566575568",
    "title": "NOVA X 手机 TVC",
    "titleEn": "NOVA X smartphone TVC",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《NOVA X 手机 TVC》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for NOVA X smartphone TVC. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create a 15-second premium cinematic branding TVC for a futuristic smartphone brand called NOVA X.\n\n0–5 seconds: Extreme close-up of a sleek futuristic smartphone resting on a glossy black surface. Dramatic studio lighting sweeps across the device, revealing its ultra-thin design and premium metallic edges. Cinematic camera push-in, macro product details, realistic reflections.\n\n5–10 seconds: The phone activates with a stunning futuristic interface. A young professional picks it up, and the environment transforms into a vibrant digital world with floating holographic visuals, smooth UI animations, and seamless interaction. Dynamic camera movement, premium commercial aesthetic.\n\n10–15 seconds: Hero product shot of the smartphone standing vertically against a minimalist futuristic background. The NOVA X logo appears elegantly with the tagline: “NOVA X — The Future in Your Hands.” Cinematic lighting, high-end TV commercial quality, photorealistic, ultra-detailed, smooth motion, polished advertising style, 4K.",
    "promptSourceUrl": "https://x.com/UrMeer289/status/2083048872566575568",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/UrMeer289/status/2083048872566575568",
    "sourceLabel": "X 原帖 · @UrMeer289",
    "author": "Ur Meer",
    "publishedAt": "2026-07-31T04:35:27.053Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083048872566575568.mp4",
    "posterUrl": "/posters/x/2083048872566575568.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created with MiniMax H3 on @Hailuo_AI \n\nPrompt: \nCreate a 15-second premium cinematic branding TVC for a futuristic smartphone brand called NOVA X.\n\n0–5 seconds: Extreme close-up of a sleek futuristic smartphone resting on a glossy black surface. Dramatic studio lighting sweeps across the device, revealing its ultra-thin design and premium metallic edges. Cinematic camera push-in, macro product details, realistic reflections.\n\n5–10 seconds: The phone activates with a stunning futuristic interface. A young professional picks it up, and the environment transforms into a vibrant digital world with floating holographic visuals, smooth UI animations, and seamless interaction. Dynamic camera movement, premium commercial aesthetic.\n\n10–15 seconds: Hero product shot of the smartphone standing vertically against a minimalist futuristic background. The NOVA X logo appears elegantly with the tagline: “NOVA X — The Future in Your Hands.” Cinematic lighting, high-end TV commercial quality, photorealistic, ultra-detailed, smooth motion, polished advertising style, 4K.\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083048782581858681",
    "title": "训练后健身房 Vlog",
    "titleEn": "Post-workout gym vlog",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《训练后健身房 Vlog》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Post-workout gym vlog. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "CAMERA: DV 16mm tape camcorder POV. CHASE films herself, occasionally propping the camera on a wall or gym bag. Natural handheld shake, imperfect framing, delayed focus, clumsy zooms, face occasionally cut off. Camcorder never visible.\nLOOK: Soft blurry tape aesthetic, faint tape noise, bloomed gym lights, flickering auto-exposure, muted contrast, realistic skin.\nSTYLE: Calm post-workout gym vlog. Relaxed, reflective, slightly tired but happy. Slow pacing, natural laughs.\nCHARACTER: CHASE, Korean idol (20s). Long black hair loosely tied, light post-workout glow, slim athletic build. Modest long-sleeve athletic top, loose joggers/leggings, socks or shoes off, towel around neck, no jewelry.\nSETTING: Quiet evening gym. Stretching mat, mirror wall, gym bag, protein shake, soft lighting, mostly empty.\n15s | 6 Cuts\nPropped medium: sits on mat, exhales. \"Okay, cool-down time.\"\nHandheld stretch: hamstring stretch, laughs. \"Yeah, that was a lot today.\"\nMacro: picks up protein shake, condensation visible. Ambient sound only.\nMedium: drinks shake, smiles. \"Okay, that's actually really good.\"\nClose: overhead stretch, rates workout. \"I'd give today like an eight out of ten.\"\nSelfie: grabs bag and shake, waves. \"Alright, I'm heading out—see you guys.\"",
    "promptSourceUrl": "https://x.com/doctorwasif/status/2083048782581858681",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/doctorwasif/status/2083048782581858681",
    "sourceLabel": "X 原帖 · @doctorwasif",
    "author": "WasifAI",
    "publishedAt": "2026-07-31T04:35:05.599Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083048782581858681.mp4",
    "posterUrl": "/posters/x/2083048782581858681.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Vlog",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "From natural dialogue to realistic handheld camera movement, H3 captures the small details that make a scene feel authentic. Really enjoyed creating this gym vlog with MiniMax H3.\n\n@Hailuo_AI\n\nprompt:\nCAMERA: DV 16mm tape camcorder POV. CHASE films herself, occasionally propping the camera on a wall or gym bag. Natural handheld shake, imperfect framing, delayed focus, clumsy zooms, face occasionally cut off. Camcorder never visible.\nLOOK: Soft blurry tape aesthetic, faint tape noise, bloomed gym lights, flickering auto-exposure, muted contrast, realistic skin.\nSTYLE: Calm post-workout gym vlog. Relaxed, reflective, slightly tired but happy. Slow pacing, natural laughs.\nCHARACTER: CHASE, Korean idol (20s). Long black hair loosely tied, light post-workout glow, slim athletic build. Modest long-sleeve athletic top, loose joggers/leggings, socks or shoes off, towel around neck, no jewelry.\nSETTING: Quiet evening gym. Stretching mat, mirror wall, gym bag, protein shake, soft lighting, mostly empty.\n15s | 6 Cuts\nPropped medium: sits on mat, exhales. \"Okay, cool-down time.\"\nHandheld stretch: hamstring stretch, laughs. \"Yeah, that was a lot today.\"\nMacro: picks up protein shake, condensation visible. Ambient sound only.\nMedium: drinks shake, smiles. \"Okay, that's actually really good.\"\nClose: overhead stretch, rates workout. \"I'd give today like an eight out of ten.\"\nSelfie: grabs bag and shake, waves. \"Alright, I'm heading out—see you guys.\" #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083045838377652641",
    "title": "爵士黑色电影动画片头",
    "titleEn": "Jazz-Noir Anime Title Sequence",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《爵士黑色电影动画片头》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Jazz-Noir Anime Title Sequence. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "@Arcane_Aii Using the attached reference image for the character and the attached audio track for timing, generate a jazz-noir anime title sequence in a stylized pop-art style. Every cut must land exactly on a beat of the audio — treat the track as the edit timeline.\n\nVISUAL STYLE — apply throughout:\nRender characters as flat, high-contrast black silhouettes against solid single-color backgrounds (saturated red, mustard yellow, deep blue). No naturalistic shading — each shot commits to one or two dominant colors only, like a screen-printed poster.\n\nCompose shots like comic-book panels and pulp-magazine covers: bold panel divisions splitting the frame, hard geometric shapes, dramatic off-center cropping, extreme close-ups intercut with full-body poses.\n\nUse freeze-frames as a core device: the character hits a dynamic pose (drawing a weapon, mid-stride, lighting a cigarette, turning toward camera) and the frame HOLDS completely still for one full beat before cutting. Do not smooth or interpolate through these holds — they must be dead stops.\n\nTitle typography in mid-century Saul Bass style: stark geometric letterforms, bold sans-serif, text that slides in as flat colored bars or snaps into frame on the beat.\n\nOverlay the entire sequence with heavy analog film grain, slight gate weave, and worn-print color texture — it should look like a scratched 35mm print of a 1960s jazz album cover, not clean digital animation.\n\nMOTION AND EDITING:\nAlternate between two modes — (1) fast kinetic bursts of motion (whip pans across silhouettes, rapid cut sequences, a silhouette sprinting across a solid color field) and (2) hard freeze-frame holds. The contrast between the two IS the rhythm. Cuts are hard cuts only: no dissolves, no fades, no morphing between shots. Each new shot snaps in on the downbeat with a fresh background color.\n\nEnd on the character in silhouette, frozen mid-pose against a solid red field, with the title card snapping in beside them on the final hit of the track.",
    "promptSourceUrl": "https://x.com/AIWarper/status/2083378754102202636",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AIWarper/status/2083045838377652641",
    "sourceLabel": "X 原帖 · @AIWarper",
    "author": "A.I.Warper",
    "publishedAt": "2026-07-31T04:23:23.646Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083045838377652641.mp4",
    "posterUrl": "/posters/x/2083045838377652641.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MINIMAX H3 \n\naudio generated from Suno + a single ref image with a prompt for a Cowboy Bepop style intro 🎺",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083043197937209852",
    "title": "韩国偶像 VHS 美妆 Vlog",
    "titleEn": "Korean Idol VHS Beauty Vlog",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《韩国偶像 VHS 美妆 Vlog》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Korean Idol VHS Beauty Vlog. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "DV 16mm camcorder vlog with soft VHS tape quality, handheld POV by CHASE and mirror-mounted vanity shots. CHASE, a Korean idol in her 20s with long black hair and dewy skin, casually picks up the fictional Lumina Velvet Lip Tint (frosted pink tube, gold cap) at a mall beauty store, saying, \"Okay, everyone's talking about this shade.\" She heads home, opens and swatches the lipstick, applies it smoothly in warm vanity lighting, admires the glossy finish, smiles at the mirror, then looks into the camera for a glowing close-up reveal and says, \"Okay wow... Lumina Velvet Lip Tint, you've got a new fan.\"",
    "promptSourceUrl": "https://x.com/saniaspeaks_/status/2083043197937209852",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/saniaspeaks_/status/2083043197937209852",
    "sourceLabel": "X 原帖 · @saniaspeaks_",
    "author": "𝗦𝗮𝗻𝗶𝗮",
    "publishedAt": "2026-07-31T04:12:54.116Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083043197937209852.mp4",
    "posterUrl": "/posters/x/2083043197937209852.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "commercial"
    ],
    "scenes": [
      "dialogue",
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I tested MiniMaxH3 by creating this entire beauty routine with @Hailuo_AI and I’m impressed by how well it understood the style, camera language, product flow, and performance.\n\nPrompt:\nDV 16mm camcorder vlog with soft VHS tape quality, handheld POV by CHASE and mirror-mounted vanity shots. CHASE, a Korean idol in her 20s with long black hair and dewy skin, casually picks up the fictional Lumina Velvet Lip Tint (frosted pink tube, gold cap) at a mall beauty store, saying, \"Okay, everyone's talking about this shade.\" She heads home, opens and swatches the lipstick, applies it smoothly in warm vanity lighting, admires the glossy finish, smiles at the mirror, then looks into the camera for a glowing close-up reveal and says, \"Okay wow... Lumina Velvet Lip Tint, you've got a new fan.\"\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083042578883383681",
    "title": "超碳酸海滩饮料广告",
    "titleEn": "SparkRush Ultra Fizz beach soft-drink commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《超碳酸海滩饮料广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×966。",
    "summaryEn": "An H3 output for SparkRush Ultra Fizz beach soft-drink commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×966.",
    "prompt": "（16:9 / 15秒 / 8カット / 日本語セリフ版）\n@ ImageA＝採用済みイメージボード。\n（以降、イメージボードと記述）\nイメージボードは 登場人物2人の雰囲気、水着、ビーチ環境、商品ボトルデザイン、色、爽快感、映像トーンの参考資料 としてのみ使用する。\n重要：イメージボードのレイアウトや構成そのものは絶対に映像内へ描写しない。\nイメージボード、8分割グリッド、コマ割り、番号、説明文、見出し、商品一覧、配色見本、ムードボード、ポスター、静止画コラージュ、絵コンテ風表示を作らない。\n完成物は1本の実写CM映像として出力し、画面内に文字・数字・ロゴ説明・CUT番号・パネル分割を表示しない。\n\n15秒、16:9、実写シネマティックCM、全8カット。\n真夏の明るい昼のビーチ。白い砂浜、透明感のある青い海、強い日差し、爽やかな風。\n登場人物は20代前半の成人男女2人。健康的でおしゃれ、自然な笑顔、友達同士の軽快で爽やかな空気感。\n商品は架空の超炭酸ペットボトル飲料 SPARKRUSH ULTRA FIZZ。\n透明ペットボトル、青いキャップ、濃いブルー主体のラベル、斜めライン、粒粒の炭酸バブル表現、勢いのある刺激的なデザイン。\n水やミネラルウォーターではなく、強い超炭酸飲料に見せること。\nラベルはエネルギッシュでおしゃれ、炭酸の刺激が伝わるグラフィック。\n実在ブランドのロゴや意匠は使わない。\n全カットで人物の顔、水着、商品ロゴ、ボトル形状、ラベル配色を統一する。\n字幕、テロップ、追加文字、別ロゴは入れない。\n\nCUT1 0:00-0:02\n\nワイドショット。真夏のビーチ。男女2人が浅瀬を走りながら海へ入っていく。水しぶきがきらめく。爽快で青春感のある始まり。\n男「うわー、最高！」\n\nCUT2 0:02-0:04\n\n2人がビーチボールで遊ぶ。明るく笑いながら投げ合う。空と海が鮮やかで、夏らしい開放感。\n女「ほら、ナイスキャッチ！」\n\nCUT3 0:04-0:06\n\n女性の上半身寄り。女性が SPARKRUSH ULTRA FIZZ のボトルをカメラに向けて掲げ、爽やかに笑う。ボトルの青いラインと粒粒の炭酸デザインがしっかり見える。\n女「これ、超炭酸だよ！」\n\nCUT4 0:06-0:08\n\n商品寄り。手でキャップをひねって開けた瞬間、プシュッという強い炭酸音。ボトル口から細かな泡と冷たいミストが弾ける。超炭酸の刺激を強く見せる。\nSE：炭酸の鋭い開栓音\n\nCUT5 0:08-0:10\n\n男性がボトルを飲む寄りショット。冷えたボトルの水滴、喉を通る爽快感、飲んだ瞬間の刺激的で気持ちいい表情。\n男「うおっ、キリッとくる！」\n\nCUT6 0:10-0:12\n\n2人のツーショット。笑い合いながらボトルを持つ。自然な距離感、眩しい日差し、夏が弾けるような空気感。\n女「すっきり爽快！」\n男「夏って感じ！」\n\nCUT7 0:12-0:14\n\n海辺の商品イメージカット。手前または岩場近くに SPARKRUSH ULTRA FIZZ のボトル。背景の海ではイルカの群れが数頭ジャンプしている。爽やかで開放的な奇跡のような演出。海しぶきと太陽光がきらめく。\nNA「超炭酸で、リフレッシュ！」\n\nCUT8 0:14-0:15\n\nエンドカット。海を背にして並ぶ2人、または海辺に座る2人の後ろ姿。画面の手前または横に商品ボトルを大きく配置。最後は商品が主役。\n2人「スパークラッシュ！」\n\n映像トーン：高級感のある夏の飲料CM、爽快、青春、開放感、清潔感、ファッション広告のような洗練。\nカメラ：ワイド、ミディアム、商品寄り、人物寄りをテンポよく切り替える。\n音：波音、軽快なサマーポップ、ビーチボールの軽い音、炭酸の開栓音、泡が弾ける音。\n口パクは日本語セリフに自然に合わせる。\n\n強い禁止事項\nイメージボードのムードボード自体を再現しない\n8分割パネル、コマ割り、静止画コラージュにしない\n番号、見出し、解説文、配色パレット、商品比較図を表示しない\nポスター、絵コンテ、広告ボード、資料画像のようにしない\n1枚絵の資料ではなく、自然な連続CM映像として出力する",
    "promptSourceUrl": "https://x.com/mi7_crypto/status/2083042584755376250",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/mi7_crypto/status/2083042578883383681",
    "sourceLabel": "X 原帖 · @mi7_crypto",
    "author": "M7［mi7］AI",
    "publishedAt": "2026-07-31T04:10:26.522Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083042578883383681.mp4",
    "posterUrl": "/posters/x/2083042578883383681.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×966",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "nature"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "架空の商品CM（ポン出し）\n『超炭酸飲料 SPARKRUSH ULTRA FIZZ』\nHailuoの新モデル【H3】アーリーアクセス（EA）検証\n@Hailuo_AI #MiniMaxH3\n商品プロダクトのCM制作にベストマッチなツール！\n5分位でこのCMは作られました！\n👇プロンプトはリプ欄で公開",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083039412506743096",
    "title": "沙漠越野车战斗追逐",
    "titleEn": "Desert Buggy Combat Chase",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《沙漠越野车战斗追逐》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 21 秒 · 1698×720。",
    "summaryEn": "An H3 output for Desert Buggy Combat Chase. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 21s · 1698×720.",
    "prompt": "Create a 15-second, 16:9 photoreal cinematic action sequence with native stereo audio. Treat the five images as coordinated multimodal references for identity, vehicle design, environment, performance, action, cinematography, and sound.\n\nUse Image 1 as the strict facial-identity reference for the female racer. Preserve her exact face, dark tied-back hair, amber goggles, dusty skin, black tactical suit, shoulder armor, gloves, and focused expression.\n\nUse Image 2 as the strict reference for her full costume, body proportions, armored black dune buggy, tire scale, exposed suspension, cyan headlights, roll cage, desert lighting, and industrial-outpost background.\n\nUse Image 3 as the action reference for the pursuing drones, vehicle speed, dust trails, fire, debris, chase intensity, and camera proximity.\n\nUse Image 4 as the location reference for Outpost 07, including its rusted towers, pipelines, bridges, rocky terrain, distant mountains, warm sunset atmosphere, and industrial scale.\n\nUse Image 5 as the cockpit and performance reference. Preserve the same steering wheel, open roll cage, goggles, gloves, facial identity, driving posture, and vehicle interior.\n\nSequence\n[0–2.5 seconds]\n\nOpen on Image 4 with an extreme-wide aerial establishing shot of Outpost 07 at sunset. The industrial complex fills the right side of frame while an empty desert route curves through the rocky foreground.\n\nThe camera dives rapidly toward the road as a small black dune buggy bursts from beneath an elevated pipeline, throwing a long dust plume behind it.\n\nKeep the vehicle moving consistently from left to right toward the outpost. No direction reversal.\n\nAudio begins with dry desert wind, distant industrial machinery, a low cinematic pulse, and the buggy engine approaching rapidly.\n\n[2.5–5 seconds]\n\nCut at peak engine sound to Image 5.\n\nTight frontal cockpit shot mounted just ahead of the driver. The vehicle shakes naturally over rough ground. Her hands hold the steering wheel firmly while she makes small, physically accurate corrections.\n\nHer amber goggles remain on top of her head. Loose strands of hair and fabric straps react to wind and vibration. Her eyes briefly check the left mirror, then return immediately to the road.\n\nA red warning light reflects across her face as a targeting alarm begins.\n\nDo not make her turn the steering wheel excessively. Her body, wheel movement, and vehicle direction must remain mechanically connected.\n\n[5–8 seconds]\n\nCut to a low front three-quarter tracking shot based on Image 3.\n\nThree pursuit drones descend behind the buggy in a triangular formation. Their rotors, stabilizers, and body movement respond realistically to speed and turbulence.\n\nThe lead drone fires into the sand beside the buggy. The impact creates a narrow eruption of dirt, sparks, and fragmented rock rather than an oversized fireball.\n\nThe racer steers sharply around the impact. The buggy’s front wheels turn first, the suspension compresses, the body leans, and the rear tires slide outward before regaining traction.\n\nThe camera tracks beside the vehicle without spinning or overtaking it.\n\n[8–11 seconds]\nContinue the same drift into a rear three-quarter shot.\n\nThe buggy races toward a narrowing passage between a rock wall and the outer structures of Outpost 07. The racer pulls a mechanical handbrake lever for one brief moment, rotating the vehicle through the opening.\n\nOne drone follows too closely and clips a rusted overhead pipe. Its wing breaks, sending the drone tumbling into the sand behind her.\n\nShow the collision in the background while keeping the buggy dominant and moving forward. No slow motion.\n\nAudio: tire scrape, suspension impact, metal tearing, drone rotors failing, engine rev rising.\n\n[11–13 seconds]\n\nCut to a wheel-level macro shot.\n\nThe right rear tire bites into loose sand. Stones fire backward while the suspension rebounds. The camera rises naturally along the buggy’s side and reveals the two remaining drones closing in.\n\nThe racer presses a guarded switch beside the steering wheel.\n\nA compact rear-mounted electromagnetic pulse discharges as a restrained blue-white distortion wave, briefly disrupting the drones’ lights and stabilizers.\n\nNo magical energy, lightning storm, or giant explosion.\n\n[13–15 seconds]\nCut to a low frontal hero shot as the buggy clears the outpost gate at full speed.\n\nThe two disabled drones fall into the dust behind it while the vehicle launches from a shallow ridge. Keep the jump low, heavy, and physically believable.\n\nDuring the brief airborne moment, cut to Image 1 for a tight close-up of the racer. Preserve her exact identity as warm firelight and cool dashboard light cross her face. Her expression remains controlled and determined.\n\nThe buggy lands hard beyond the gate. The suspension compresses, the engine roars, and the vehicle continues directly into the desert.\n\nEnd with a sharp cut to black on the landing impact.\n\nVisual Direction\nPremium live-action science-fiction action trailer with realistic CGI integration, warm orange sunset light, restrained steel-blue technology accents, dusty atmosphere, hard surface detail, cinematic contrast, subtle film grain, realistic motion blur, and 24 FPS movement.\n\nUse wide shots to establish scale, cockpit close-ups for tension, low tracking shots for speed, and mechanical macro shots for physical detail.\n\nKeep the editing fast but readable. Every shot must begin from the physical state established by the previous shot.\n\nAudio Direction\nNative stereo sound with:\nAggressive combustion engine\nTire friction over sand and rock\nSuspension rattles and chassis vibration\nDrone rotors and targeting alarms\nSand impacts, metal collisions, and falling debris\nLow percussion and rising electronic tension\nOne heavy bass impact on the final landing\n\nKeep music underneath the vehicle and environmental sounds. No dialogue or voice-over.\n\nRestrictions\nNo subtitles, titles, logos, watermarks, additional racers, pedestrians, creatures, motorcycles, futuristic city skyline, nighttime transition, costume changes, facial changes, vehicle redesign, additional wheels, floating vehicle parts, distorted hands, incorrect steering, reversed wheel rotation, teleportation, random explosions, oversized fireballs, weightless motion, impossible jumps, camera spins, circular camera moves, rapid zooms, fluid morphs, soft dissolves, or changes in travel direction.",
    "promptSourceUrl": "https://x.com/beginnersblog1/status/2083039427669230034",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/beginnersblog1/status/2083039412506743096",
    "sourceLabel": "X 原帖 · @beginnersblog1",
    "author": "Beginnersblog",
    "publishedAt": "2026-07-31T03:57:51.599Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083039412506743096.mp4",
    "posterUrl": "/posters/x/2083039412506743096.jpg",
    "duration": 21,
    "aspectRatio": "landscape",
    "resolution": "1698×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Horror",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark"
    ],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I tested Hailuo Minimax H3.\n\nI was using storyboards to guide MiniMax H3.\n\nThe results were inconsistent.\n\nSo I changed one thing:\nInstead of uploading the full storyboard, I turned every important frame into a separate reference image.\n\nMy workflow:\n- Create one clear 15-second concept\n- Plan the sequence as a storyboard\n- Generate separate images for the character, vehicle, location, action, and hero shot\n- Give each image one specific job\n- Write a time-coded prompt with camera movement, physics, sound, and continuity\n\nThe storyboard helped me think. The separate images helped the model understand.\n\nThat difference improved the entire sequence.\n\nthe prompt is in comments",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083033635364720815",
    "title": "机械公牛冲锋·野兽将至",
    "titleEn": "Mechanical bull charge The Beasts Are Coming",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《机械公牛冲锋·野兽将至》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Mechanical bull charge The Beasts Are Coming. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use the uploaded mechanical bull image as the primary reference for the bull, and use Rahul’s image as the character reference for the rider.\n\nCreate a 15-second ultra-realistic cinematic fantasy-sci-fi sequence set in The Harvester universe. The setting is a foggy ruined industrial lane at dawn or overcast light, with drifting smoke, hanging wires, broken stone and metal structures, wet ground, ash in the air, and distant furnace glow. The giant mechanical bull is ancient, battle-scarred, and alive with furnace energy inside its cracked armor.\n\n0–3s — Hook: Start with Rahul stumbling backward into frame, breathing hard, turning to face danger in the smoke behind him. The giant mechanical bull stands in front of him like a wall, half-silhouetted in steam, protecting him.\n\n3–6s: Close on Rahul placing one hand on the bull’s armored side, steadying himself. He looks at it with trust, not fear. The bull turns its head slightly toward Rahul. Its eye glows brighter, and its internal furnace begins to pulse.\n\n6–9s: A loud metallic roar echoes from somewhere in the fog ahead. The bull reacts instantly — it lowers its head, paws the ground, steam blasts from its vents, gears lock into place, and sparks spit from its chest cavity. Rahul pulls himself up and mounts the bull in one fast motion.\n\n9–12s: The bull rears slightly, then slams down hard. Rahul leans forward, gripping tight. Through the smoke ahead, we briefly see threatening silhouettes moving toward them, just enough to imply incoming enemies.\n\n12–15s — Payoff: The bull explodes forward into a full charge with Rahul riding it, smashing through smoke and embers. End on a powerful low-angle tracking shot as both disappear into the haze, charging straight toward the unseen threat.\n\nSound: deep metal groans, steam bursts, furnace rumble, distant enemy roar, hoof impact, sparks, low cinematic score.\n\nCinematography / Color Brief\n\nUse the same signature look: desaturated steel blue, charcoal grey, mist white, weathered bronze, and restrained furnace orange accents. Soft overcast light, smoky atmosphere, wet surfaces, and strong depth. Cinematography should feel premium and mythic: low angles, slow dramatic push-ins, tactile detail, subtle shake, sharp realism, and heavy scale.",
    "promptSourceUrl": "https://x.com/rahulnanda86/status/2083033635364720815",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/rahulnanda86/status/2083033635364720815",
    "sourceLabel": "X 原帖 · @rahulnanda86",
    "author": "Rahul Nanda",
    "publishedAt": "2026-07-31T03:34:54.221Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083033635364720815.mp4",
    "posterUrl": "/posters/x/2083033635364720815.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "fantasy"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Beasts Are Coming!!  \n\nMiniMax H3 Prompt/Made on @runwayml - \n\nUse the uploaded mechanical bull image as the primary reference for the bull, and use Rahul’s image as the character reference for the rider. \n\nCreate a 15-second ultra-realistic cinematic fantasy-sci-fi sequence set in The Harvester universe. The setting is a foggy ruined industrial lane at dawn or overcast light, with drifting smoke, hanging wires, broken stone and metal structures, wet ground, ash in the air, and distant furnace glow. The giant mechanical bull is ancient, battle-scarred, and alive with furnace energy inside its cracked armor.\n\n0–3s — Hook: Start with Rahul stumbling backward into frame, breathing hard, turning to face danger in the smoke behind him. The giant mechanical bull stands in front of him like a wall, half-silhouetted in steam, protecting him.\n\n3–6s: Close on Rahul placing one hand on the bull’s armored side, steadying himself. He looks at it with trust, not fear. The bull turns its head slightly toward Rahul. Its eye glows brighter, and its internal furnace begins to pulse.\n\n6–9s: A loud metallic roar echoes from somewhere in the fog ahead. The bull reacts instantly — it lowers its head, paws the ground, steam blasts from its vents, gears lock into place, and sparks spit from its chest cavity. Rahul pulls himself up and mounts the bull in one fast motion.\n\n9–12s: The bull rears slightly, then slams down hard. Rahul leans forward, gripping tight. Through the smoke ahead, we briefly see threatening silhouettes moving toward them, just enough to imply incoming enemies.\n\n12–15s — Payoff: The bull explodes forward into a full charge with Rahul riding it, smashing through smoke and embers. End on a powerful low-angle tracking shot as both disappear into the haze, charging straight toward the unseen threat.\n\nSound: deep metal groans, steam bursts, furnace rumble, distant enemy roar, hoof impact, sparks, low cinematic score.\n\nCinematography / Color Brief\n\nUse the same signature look: desaturated steel blue, charcoal grey, mist white, weathered bronze, and restrained furnace orange accents. Soft overcast light, smoky atmosphere, wet surfaces, and strong depth. Cinematography should feel premium and mythic: low angles, slow dramatic push-ins, tactile detail, subtle shake, sharp realism, and heavy scale.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083033353679221043",
    "title": "拳台边手机实拍击倒",
    "titleEn": "Ringside Phone-Footage Boxing Knockout",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《拳台边手机实拍击倒》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×1280。",
    "summaryEn": "An H3 output for Ringside Phone-Footage Boxing Knockout. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×1280.",
    "prompt": "@Hailuo_AI prompt：\n【风格】UGC拳赛现场实拍（Ringside Phone Footage），真实手机竖屏手持质感（9:16 Vertical, Handheld），现场生肉感：数码变焦、轻微失焦、运动模糊、闪光灯频闪，无滤镜无调色的直出画面\n【时长】15秒\n【场景】室内拳击馆比赛之夜：正规拳台、四周围绳、台上顶灯直打，台下观众黑压压一片举着手机，闪光灯此起彼伏，拍摄者站在台边第二排人群里手持手机拍摄\n【角色】主角1@图片1（黑袍蓝边战袍方，脱袍后黑色拳裤黑手套）、主角2@图片2（白袍绿边战袍方，脱袍后白色拳裤白手套）\n【拍摄机制】全片一镜到底的手机手持视角：画面始终带轻微手抖和呼吸晃动，关键时刻拍摄者会猛推数码变焦怼近、短暂失焦再对上，激烈处镜头被前排观众起身遮挡半秒又找回角度——所有\"不完美\"都是真实感的来源\n\n[00:00-00:03] 开局试探\n手机视角越过前排观众的后脑勺拍向拳台：铃声\"铛\"一响，两人从各自角落逼近，主角1压低重心左右晃动寻找角度，主角2双拳护脸小步游走，两人互相试探刺拳，拳套相碰发出闷响。\n【现场音】铃声、观众嘈杂人声、拍摄者身边有人喊\"打他！\"\n\n[00:03-00:07] 第一波对攻\n拍摄者猛推数码变焦怼近：主角1突进一记左刺拳接右摆拳，主角2侧头闪过、贴身回敬两记摆拳组合，两人拳套砰砰互击、汗珠在顶灯下甩出，画面因变焦过快糊了半秒又对上焦。\n【现场音】拳套击打的连续闷响、观众爆发的\"哦——！\"、拍摄者忍不住的一声\"卧槽\"。\n\n[00:07-00:10] 缠斗与遮挡\n两人进入贴身缠斗互相抱架，裁判上前分开——此时前排观众激动起身，手机画面被后脑勺挡住一瞬，拍摄者手忙脚乱举高手机越过人头继续拍：两人分开各退半步，主角2抹了把脸，主角1活动了下脖子。\n【现场音】裁判喊\"Break！\"，人群骚动声浪盖过一切。\n\n[00:10-00:13] 高潮重击\n举高俯角视角：主角2抢先出重拳，主角1一个下潜躲过，起身瞬间一记右上勾拳结结实实顶中主角2下巴——主角2整个人晃了两晃、踉跄后退撞上围绳滑坐到台面，全场瞬间炸裂，闪光灯连成一片。\n【现场音】重击一声爆响后观众声浪掀顶，拍摄者激动大吼，手机剧烈晃动。\n\n[00:13-00:15] 终场定格\n画面抖着拉回全景：裁判冲上前查看，主角1退到中立角高举双拳，台下无数手机屏幕的亮光连成星海，拍摄者边喊边把镜头怼向台上，画面在主角1振臂的糊焦瞬间定格。\n【现场音】欢呼、口哨、有人狂拍前座椅背，收束在鼎沸人声里。\n\n【备注】全片保持手机直出的粗粝感：允许过曝的顶灯、糊焦、拖影和构图歪斜，禁止任何电影感调色和稳定器丝滑运镜；击打动作写意图和结果（命中、踉跄、坐地），保持竞技体育范畴，无出血无伤情特写",
    "promptSourceUrl": "https://x.com/johnAGI168/status/2083033507278790722",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/johnAGI168/status/2083033353679221043",
    "sourceLabel": "X 原帖 · @johnAGI168",
    "author": "John",
    "publishedAt": "2026-07-31T03:33:47.062Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083033353679221043.mp4",
    "posterUrl": "/posters/x/2083033353679221043.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Action & VFX",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "ugc"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MiniMaxH3 EA 测试： Elon 一击即中🥊\n\n在我的这次测试里，相比 Seedance 2.0，H3 的名人角色参考成功跑通了。两张角色卡 + Omni Reference，直接生成擂台边手机实拍。\n\n手持抖动、观众遮挡、临场失焦、拳台氛围和击倒瞬间——真实感直接拉满。prompt 👇\n\n@Hailuo_AI  #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083032575908700581",
    "title": "醒来，我们去狩猎",
    "titleEn": "Wake up We Hunt mechanical bull mount",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《醒来，我们去狩猎》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Wake up We Hunt mechanical bull mount. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use the uploaded mechanical bull image as the main creature reference and Rahul’s image as the human character reference. Create a 15-second ultra-realistic cinematic sequence set in The Harvester universe. The scene takes place in a foggy dystopian industrial street full of smoke, ash, overhead wires, dim structures, and distant furnace glow. The giant mechanical bull is massive, battle-worn, and intimidating — a living machine-beast with heavy steel plating, exposed pistons, pipes, furnace light glowing inside cracked armor, huge curved horns, and hot steam venting from its body.\n\n0–3s — Hook: Start with a dramatic low-angle shot of the giant mechanical bull standing motionless in the street like a dead titan, steam drifting around it. Its orange core is dim. Rahul walks into frame, small beside it, cautious but familiar, like this is his companion.\n\n3–6s: Rahul places one hand firmly against the bull’s metal faceplate or neck armor. Close on the contact. He says in a calm, controlled tone: “Wake up. We hunt.” The bull’s eye glows brighter. Internal furnace light begins pulsing.\n\n6–9s: The bull violently powers on — steam blasts from vents, gears rotate, pistons lock, embers spit from its chest, and its massive head lifts. One hoof slams into the ground with weight. The whole frame shakes slightly.\n\n9–12s: Rahul pulls himself up and mounts the giant bull. The bull turns its head toward something unseen ahead in the smoke — distant danger, possibly droids or enemies approaching. We feel the next hunt is beginning.\n\n12–15s — Payoff: The bull charges forward with explosive force through smoke and sparks, Rahul riding it hard. End on a powerful forward rush into haze as the machine roars and the hunt begins.\n\nSound: deep metallic groans, steam bursts, furnace rumble, hoof impact, distant industrial ambience, low ominous score.",
    "promptSourceUrl": "https://x.com/rahulnanda86/status/2083032575908700581",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/rahulnanda86/status/2083032575908700581",
    "sourceLabel": "X 原帖 · @rahulnanda86",
    "author": "Rahul Nanda",
    "publishedAt": "2026-07-31T03:30:41.627Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083032575908700581.mp4",
    "posterUrl": "/posters/x/2083032575908700581.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "'Wake up, We Hunt'\n\nFirst test of MiniMax H3 on @runwayml \n\nMiniMax H3 Prompt - \n\nUse the uploaded mechanical bull image as the main creature reference and Rahul’s image as the human character reference. Create a 15-second ultra-realistic cinematic sequence set in The Harvester universe. The scene takes place in a foggy dystopian industrial street full of smoke, ash, overhead wires, dim structures, and distant furnace glow. The giant mechanical bull is massive, battle-worn, and intimidating — a living machine-beast with heavy steel plating, exposed pistons, pipes, furnace light glowing inside cracked armor, huge curved horns, and hot steam venting from its body.\n\n0–3s — Hook: Start with a dramatic low-angle shot of the giant mechanical bull standing motionless in the street like a dead titan, steam drifting around it. Its orange core is dim. Rahul walks into frame, small beside it, cautious but familiar, like this is his companion.\n\n3–6s: Rahul places one hand firmly against the bull’s metal faceplate or neck armor. Close on the contact. He says in a calm, controlled tone: “Wake up. We hunt.” The bull’s eye glows brighter. Internal furnace light begins pulsing.\n\n6–9s: The bull violently powers on — steam blasts from vents, gears rotate, pistons lock, embers spit from its chest, and its massive head lifts. One hoof slams into the ground with weight. The whole frame shakes slightly.\n\n9–12s: Rahul pulls himself up and mounts the giant bull. The bull turns its head toward something unseen ahead in the smoke — distant danger, possibly droids or enemies approaching. We feel the next hunt is beginning.\n\n12–15s — Payoff: The bull charges forward with explosive force through smoke and sparks, Rahul riding it hard. End on a powerful forward rush into haze as the machine roars and the hunt begins.\n\nSound: deep metallic groans, steam bursts, furnace rumble, hoof impact, distant industrial ambience, low ominous score.\n\n@MiniMax_AI @Hailuo_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083032443226030578",
    "title": "脏感嘻哈 MV 情绪",
    "titleEn": "Grimy hip-hop music film mood",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《脏感嘻哈 MV 情绪》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Grimy hip-hop music film mood. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "洗練されているが汚れた質感のヒップホップMV。地下雑誌、コピー機、アナログスキャン、攻撃的なビート感。",
    "promptSourceUrl": "https://x.com/tebasaki3D/status/2083032443226030578",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/tebasaki3D/status/2083032443226030578",
    "sourceLabel": "X 原帖 · @tebasaki3D",
    "author": "てばさき",
    "publishedAt": "2026-07-31T03:30:09.993Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083032443226030578.mp4",
    "posterUrl": "/posters/x/2083032443226030578.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "既にお気づきの方も多いかと思いますが…\n\nMiniMax H3、抽象・雰囲気プロンプトがかなり効きます🙌\n\nこの動画も、下記の超ざっくりしたプロンプトだけでテキストから生成しました😇\n\n洗練されているが汚れた質感のヒップホップMV。地下雑誌、コピー機、アナログスキャン、攻撃的なビート感。\n\nこの挙動を見ていると、なんだかSora2の再来というか、あの雰囲気を感じずにはいられないんですよね👀\n\nMV素材として生成して編集で仕上げる使い方とも相性が良さそう。\n\n何より、適当なプロンプトを投げて、今回はどんな映像が出てくるんだろう？ってガチャ感覚で遊べるのも楽しいです🤣\n\n動画生成難しい…と感じている人でも、かなり気軽に触りやすいモデルなんじゃないでしょうか。\n\nあとH3の強みは他にも色々あると思いますが、おっ◯いがすごく揺れます😇\n@Hailuo_AI #MinimaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083024913322103224",
    "title": "高端时装大片逐套亮相",
    "titleEn": "Premium cinematic fashion campaign outfit reveal",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《高端时装大片逐套亮相》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×1280。",
    "summaryEn": "An H3 output for Premium cinematic fashion campaign outfit reveal. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×1280.",
    "prompt": "Create a premium cinematic fashion campaign video using the five uploaded apparel images as strict product references.  The video should introduce each outfit one by one, while maintaining the exact original clothing design, silhouette, fabric texture, colors, patterns, accessories, stitching, and proportions from every reference image. Do not redesign, simplify, replace, or merge any garments.  Format: vertical 9:16 Duration: approximately 15 seconds Style: luxury fashion commercial, editorial, futuristic, minimal, sophisticated Visual quality: photorealistic, commercial-grade, high-end fashion film, sharp textile detail, realistic fabric movement  Sequence:  0–2 seconds: Begin with a dark minimal studio. A narrow beam of light slowly reveals the first outfit. Start with an extreme close-up of fabric texture, stitching, metallic details, and layered materials. Smooth cinematic camera movement.  2–5 seconds: Reveal the full first outfit on a stylish fashion model. The model makes one subtle pose while the camera performs a slow vertical tilt from shoes to face.  5–8 seconds: Transition through a fast match cut into the second and third outfits. Use elegant whip pans, fabric wipes, lighting flashes, and seamless motion transitions. Each outfit must remain visible long enough to understand its full design.  8–11 seconds: Show the fourth outfit in a wider architectural studio environment. The model walks slowly toward the camera while the fabric moves naturally. Add controlled wind, soft shadows, and reflective surfaces.  11–14 seconds: Reveal the fifth outfit as the hero look. Use a dramatic low-angle shot, slow orbiting camera movement, and stronger cinematic lighting. Emphasize the silhouette, construction, and key design details.  14–15 seconds: End with all five outfits appearing rapidly in rhythmic editorial cuts, followed by a clean final frame with the hero outfit centered against a minimal background.  Camera: Smooth dolly shots, macro textile close-ups, slow orbit, low-angle fashion shots, controlled handheld energy only during transitions.  Lighting: High-contrast luxury studio lighting, soft rim lights, precise highlights on fabric, metal, leather, and accessories. Avoid overexposure.  Motion: Natural model movement, restrained confident poses, realistic walking, subtle fabric physics. No exaggerated dancing.  Editing: Fast but elegant editorial pacing. Clean match cuts between outfits. Every transition should feel intentional and synchronized with the music.  Important: Preserve the exact identity and design of all five apparel references. Do not mix clothing elements between outfits. Do not change logos, prints, colors, garment length, footwear, accessories, or materials. Do not add random text, subtitles, watermarks, extra garments, duplicated limbs, distorted hands, warped fabric, or unstable faces.",
    "promptSourceUrl": "https://x.com/luxaios/status/2083024977151119578",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/luxaios/status/2083024913322103224",
    "sourceLabel": "X 原帖 · @luxaios",
    "author": "LUCY",
    "publishedAt": "2026-07-31T03:00:14.724Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083024913322103224.mp4",
    "posterUrl": "/posters/x/2083024913322103224.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "cinematic-realistic",
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "遅ればせながらHailuo H3使ってみました！\n5枚参照、プロンプトはリプ欄です。\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083021146341978345",
    "title": "赛博废土时装片与 VHS 故障",
    "titleEn": "Cyber-grunge fashion film with VHS glitch",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《赛博废土时装片与 VHS 故障》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 19 秒 · 1280×720。",
    "summaryEn": "An H3 output for Cyber-grunge fashion film with VHS glitch. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 19s · 1280×720.",
    "prompt": "Platinum-blonde subject, black patent-leather trench, orange firelight across the coat. Nighttime blaze, black smoke.\n\nVHS glitches, CCTV dropouts, 90s grain, scanlines, chromatic aberration, flash-to-white cuts.",
    "promptSourceUrl": "https://x.com/fal/status/2083021146341978345",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/fal/status/2083021146341978345",
    "sourceLabel": "X 原帖 · @fal",
    "author": "fal",
    "publishedAt": "2026-07-31T02:45:16.606Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083021146341978345.mp4",
    "posterUrl": "/posters/x/2083021146341978345.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "cinematic-realistic",
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cyber-grunge fashion film, made with MiniMax H3 on fal 🔥\n\nPlatinum-blonde subject, black patent-leather trench, orange firelight across the coat. Nighttime blaze, black smoke.\n\nVHS glitches, CCTV dropouts, 90s grain, scanlines, chromatic aberration, flash-to-white cuts.\n\n15s, one gen.\n\n@MiniMax_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083020024340693185",
    "title": "1980 年代开源家庭喜剧",
    "titleEn": "1980s open-source family comedy",
    "model": "MiniMax H3（来源标注）",
    "mode": "FL2VA",
    "summary": "《1980 年代开源家庭喜剧》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 12 秒 · 1280×720。",
    "summaryEn": "An H3 output for 1980s open-source family comedy. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 12s · 1280×720.",
    "prompt": "Use the supplied image as the exact opening frame. Create a hilarious, high-budget 1980s live-action family comedy movie scene, photographed on a real soundstage with practical robot costumes, animatronics, handmade props, vintage wardrobe, and authentic 35mm film color.\n\nThe entire group suddenly hears something above them. Everyone’s eyes dart upward at the same moment. The men, women, children, and robots dramatically tilt their heads back, point toward the ceiling, and erupt into chaotic celebration. They gasp, scream, laugh, jump, wave their arms, slap each other on the shoulders, and completely lose their minds with exaggerated 1980s comedy reactions. The children bounce excitedly. The adults stumble into one another. The robots flash their eyes, flap their mechanical arms, spin clumsily, and celebrate with funny practical animatronic movements.\n\nOne man shouts:\n\n“OPEN SOURCE?!”\n\nThe entire crowd answers together:\n\n“MINIMAX H3!”\n\nThey cheer wildly as one robot attempts a victory dance, loses its balance, and is caught by the shocked family at the final second.\n\nFast, energetic comic timing with believable ensemble choreography. Begin with a brief locked group shot, then a subtle handheld push-in as the chaos escalates. Keep every original person, robot, shirt design, and the large headline clearly recognizable. Preserve facial identity, wardrobe, composition, lighting, and 1980s production design.\n\nAuthentic optical softness, warm tungsten lighting, rich film grain, slight gate weave, practical effects, natural motion blur, expressive physical acting, polished studio-comedy sound mix, triumphant synthesizer sting, cheering, robot beeps, and comedic percussion.\n\nNo CGI, no digital-looking robots, no morphing, no extra people, no duplicated characters, no distorted faces, no rewritten text, no spelling changes, no modern clothing, no style shifts, and do not turn the scene into animation.",
    "promptSourceUrl": "https://x.com/BrentLynch/status/2083020024340693185",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/BrentLynch/status/2083020024340693185",
    "sourceLabel": "X 原帖 · @BrentLynch",
    "author": "Brent Lynch",
    "publishedAt": "2026-07-31T02:40:49.100Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083020024340693185.mp4",
    "posterUrl": "/posters/x/2083020024340693185.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Comedy",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "retro",
      "commercial"
    ],
    "scenes": [
      "dance",
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Open Source Minimax VIDEO\n\nThat's right!\n\nNot the LLM\n\nThe brand new \nMINIMAX \nH3 Video Model\n\n — 15-SECOND IMAGE-TO-VIDEO PROMPT  \n\nImage1\n\nUse the supplied image as the exact opening frame. Create a hilarious, high-budget 1980s live-action family comedy movie scene, photographed on a real soundstage with practical robot costumes, animatronics, handmade props, vintage wardrobe, and authentic 35mm film color.\n\nThe entire group suddenly hears something above them. Everyone’s eyes dart upward at the same moment. The men, women, children, and robots dramatically tilt their heads back, point toward the ceiling, and erupt into chaotic celebration. They gasp, scream, laugh, jump, wave their arms, slap each other on the shoulders, and completely lose their minds with exaggerated 1980s comedy reactions. The children bounce excitedly. The adults stumble into one another. The robots flash their eyes, flap their mechanical arms, spin clumsily, and celebrate with funny practical animatronic movements.\n\nOne man shouts:\n\n“OPEN SOURCE?!”\n\nThe entire crowd answers together:\n\n“MINIMAX H3!”\n\nThey cheer wildly as one robot attempts a victory dance, loses its balance, and is caught by the shocked family at the final second.\n\nFast, energetic comic timing with believable ensemble choreography. Begin with a brief locked group shot, then a subtle handheld push-in as the chaos escalates. Keep every original person, robot, shirt design, and the large headline clearly recognizable. Preserve facial identity, wardrobe, composition, lighting, and 1980s production design.\n\nAuthentic optical softness, warm tungsten lighting, rich film grain, slight gate weave, practical effects, natural motion blur, expressive physical acting, polished studio-comedy sound mix, triumphant synthesizer sting, cheering, robot beeps, and comedic percussion.\n\nNo CGI, no digital-looking robots, no morphing, no extra people, no duplicated characters, no distorted faces, no rewritten text, no spelling changes, no modern clothing, no style shifts, and do not turn the scene into animation.\n\n@Hailuo_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083013658230317082",
    "title": "RadiantGlo 护肤广告",
    "titleEn": "RadiantGlo skincare commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "Unknown",
    "summary": "《RadiantGlo 护肤广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for RadiantGlo skincare commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "The Brief: \"Jessy x RadiantGlo\"\n\n15 seconds. One face, one mask, one ritual — quiet luxury skincare.\n\nJessy stands at her mirror at night, cool blue light on her skin, studying her reflection before she says a word. She presses the remote, and the RadiantGlo mask blooms to life — red LED light rising from her jaw to her brow, glowing straight through the eye and mouth cutouts. No overlay, no CGI sheen, just real diffusion through real skin. A slow profile shot, cord resting naturally, breath audible in the silence. Macro push on her cheek — pores, peach fuzz, light catching under the surface. Then morning: she lifts the mask off, daylight replaces the red glow, and she walks through her sunlit apartment with nothing performed, just calm. Final beat: the mask alone on a matte surface, LEDs still glowing, logo catching the light — hero shot. Fade to black.\n\nCold nights fading into warm dawn color science throughout — think a skincare brand's flagship campaign, not an infomercial. Same face, same mask, same world, held together across every cut.",
    "promptSourceUrl": "https://x.com/AIwithJessica/status/2083013658230317082",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AIwithJessica/status/2083013658230317082",
    "sourceLabel": "X 原帖 · @AIwithJessica",
    "author": "Jessica Collins",
    "publishedAt": "2026-07-31T02:15:31.301Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083013658230317082.mp4",
    "posterUrl": "/posters/x/2083013658230317082.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "cinematic-realistic",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tried MiniMax H3 on my latest generation and the realism genuinely stopped me mid-scroll\nMade with @Hailuo_AI \n\nMiniMax H3 is a general-purpose multimodal video model that goes way beyond single-purpose tools. \n\nIt understands text, images, video, and audio together, reads creative intent, and outputs coherent, natural audiovisual content in one seamless pass. \n\n#hailuoai #hailuo #MiniMaxH3 #H3\n\nHere's what I fed it 👇\n\nThe Brief: \"Jessy x RadiantGlo\"\n\n15 seconds. One face, one mask, one ritual — quiet luxury skincare.\n\nJessy stands at her mirror at night, cool blue light on her skin, studying her reflection before she says a word. She presses the remote, and the RadiantGlo mask blooms to life — red LED light rising from her jaw to her brow, glowing straight through the eye and mouth cutouts. No overlay, no CGI sheen, just real diffusion through real skin. A slow profile shot, cord resting naturally, breath audible in the silence. Macro push on her cheek — pores, peach fuzz, light catching under the surface. Then morning: she lifts the mask off, daylight replaces the red glow, and she walks through her sunlit apartment with nothing performed, just calm. Final beat: the mask alone on a matte surface, LEDs still glowing, logo catching the light — hero shot. Fade to black.\n\nCold nights fading into warm dawn color science throughout — think a skincare brand's flagship campaign, not an infomercial. Same face, same mask, same world, held together across every cut.\n\nThat's it. That's the whole prompt. And [MODEL NAME] just... got it.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083011066800242986",
    "title": "浴室镜前护肤 UGC 广告",
    "titleEn": "Bathroom Mirror Skincare UGC Ad",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《浴室镜前护肤 UGC 广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Bathroom Mirror Skincare UGC Ad. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "UGC-style skincare video featuring a realistic young woman speaking directly to the camera in front of a bathroom mirror while holding a premium facial serum bottle. She smiles naturally, talks about her skincare routine, and demonstrates applying the serum to her face. Authentic, conversational presentation with genuine facial expressions and hand gestures, natural morning light, casual home bathroom setting, clean skincare aesthetic, healthy glowing skin, minimal makeup, smartphone-shot look, candid lifestyle beauty content, soft depth of field, luxury beauty brand feel, vertical 9:16, social media ad quality, highly realistic natural skin texture, not overly retouched.",
    "promptSourceUrl": "https://x.com/ZaraIrahh/status/2083011066800242986",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ZaraIrahh/status/2083011066800242986",
    "sourceLabel": "X 原帖 · @ZaraIrahh",
    "author": "Zara",
    "publishedAt": "2026-07-31T02:05:13.456Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083011066800242986.mp4",
    "posterUrl": "/posters/x/2083011066800242986.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "ugc"
    ],
    "scenes": [
      "product",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @Hailuo_AI continues to impress with realistic character generation and natural motion.\n\nThis UGC-style skincare ad captures authentic expressions, lifelike skin textures, and a genuine smartphone-shot feel.\n\nThe result feels polished, believable, and ready for social media campaigns.\n\nPrompt:\n\nUGC-style skincare video featuring a realistic young woman speaking directly to the camera in front of a bathroom mirror while holding a premium facial serum bottle. She smiles naturally, talks about her skincare routine, and demonstrates applying the serum to her face. Authentic, conversational presentation with genuine facial expressions and hand gestures, natural morning light, casual home bathroom setting, clean skincare aesthetic, healthy glowing skin, minimal makeup, smartphone-shot look, candid lifestyle beauty content, soft depth of field, luxury beauty brand feel, vertical 9:16, social media ad quality, highly realistic natural skin texture, not overly retouched.\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2083007144111325353",
    "title": "黑白八位机旧金山动画",
    "titleEn": "8-bit black and white San Francisco animation",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《黑白八位机旧金山动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 10 秒 · 1280×720。",
    "summaryEn": "An H3 output for 8-bit black and white San Francisco animation. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 10s · 1280×720.",
    "prompt": "𝟾-𝚋𝚒𝚝, 𝚋𝚕𝚊𝚌𝚔 𝚊𝚗𝚍 𝚠𝚑𝚒𝚝𝚎 𝚊𝚗𝚒𝚖𝚊𝚝𝚒𝚘𝚗 𝚘𝚏 𝚜𝚊𝚗 𝚏𝚛𝚊𝚗𝚌𝚒𝚜𝚌𝚘",
    "promptSourceUrl": "https://x.com/vercel_dev/status/2083007144111325353",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/vercel_dev/status/2083007144111325353",
    "sourceLabel": "X 原帖 · @vercel_dev",
    "author": "Vercel Developers",
    "publishedAt": "2026-07-31T01:49:38.214Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2083007144111325353.mp4",
    "posterUrl": "/posters/x/2083007144111325353.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is now on AI Gateway.\n\nTry it:\n𝚌𝚘𝚗𝚜𝚝 { 𝚟𝚒𝚍𝚎𝚘𝚜 } = 𝚊𝚠𝚊𝚒𝚝 𝚐𝚎𝚗𝚎𝚛𝚊𝚝𝚎𝚅𝚒𝚍𝚎𝚘({\n  𝚖𝚘𝚍𝚎𝚕: '𝚖𝚒𝚗𝚒𝚖𝚊𝚡/𝚖𝚒𝚗𝚒𝚖𝚊𝚡-𝚑𝟹',\n  𝚙𝚛𝚘𝚖𝚙𝚝: '𝟾-𝚋𝚒𝚝, 𝚋𝚕𝚊𝚌𝚔 𝚊𝚗𝚍 𝚠𝚑𝚒𝚝𝚎 𝚊𝚗𝚒𝚖𝚊𝚝𝚒𝚘𝚗 𝚘𝚏 𝚜𝚊𝚗 𝚏𝚛𝚊𝚗𝚌𝚒𝚜𝚌𝚘',\n});",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082999806948229386",
    "title": "豪华跑车落地页动效",
    "titleEn": "Luxury Sports Car Landing Page Motion",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《豪华跑车落地页动效》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 14 秒 · 1280×720。",
    "summaryEn": "An H3 output for Luxury Sports Car Landing Page Motion. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 14s · 1280×720.",
    "prompt": "A UI/UX motion showcase video for a high-performance luxury sports car brand landing page, created as an image-to-video project based on Product Image 1, with the car in the image remaining the core subject throughout.\n\nFirst screen – Hero Section:\n\nA premium black background, with the car positioned at the center of the page as the main visual focus. The camera slowly pushes in. A minimalist navigation bar appears at the top of the page:\n\n\"MODEL\"\n\"DESIGN\"\n\"PERFORMANCE\"\n\"TECHNOLOGY\"\n\"CONTACT\"\nOn the left side, a large brand headline appears:\n\"UNLEASH THE FUTURE\"\nBelow it, a subtitle reads:\n\"PURE PERFORMANCE. LIMITLESS SPEED.\"\n\nAt the bottom, two buttons appear:\n\n\"EXPLORE MODEL\"\n\"BOOK A TEST DRIVE\"\n\nScroll down into the vehicle design showcase section:\n\nMain headline:\n\n\"DESIGNED FOR SPEED\"\n\nShowcase text:\n\"AERODYNAMIC BODY\"\n\"CARBON FIBER STRUCTURE\"\n\"PRECISION ENGINEERING\"\n\nThese are paired with close-up shots of the car’s front fascia, headlights, wheels, and body lines.\n\nContinue scrolling into the performance section:\n\nHeadline:\n\n\"ENGINEERED TO DOMINATE\"\n\nDisplay performance specification cards:\n\"POWER\"\n\"HORSEPOWER\"\n\"TORQUE\"\n\"ACCELERATION\"\nUse dynamic, high-tech animated numeric displays for:\n\n\"0–100 KM/H\"\n\"3.2 SEC\"\n\nEnter the interior showcase section:\n\nHeadline:\n\n\"DRIVER-FOCUSED INTERIOR\"\n\nShowcase text:\n\n\"PREMIUM MATERIALS\"\n\"DIGITAL COCKPIT\"\n\"ADVANCED CONTROL SYSTEM\"\nFinal screen:\nA full view of the car is displayed, while light sweeps across the body in the background.\nCentered text:\n\n\"READY TO DRIVE\"\n\nButton:\n\"CONFIGURE YOUR CAR\"\n\nVisual style:\n\nA premium automotive brand website, similar to a supercar launch page. Dark metallic materials, carbon fiber textures, dynamic lighting, aerodynamic lines, and a cinematic luxury car advertisement aesthetic. The webpage uses large, bold, slightly slanted sans-serif typography, with fast yet smooth scrolling animation, parallax effects, mouse-hover zoom interactions, highlight scanning effects, floating information cards, and animated button feedback.",
    "promptSourceUrl": "https://x.com/BubbleBrain/status/2082999806948229386",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/BubbleBrain/status/2082999806948229386",
    "sourceLabel": "X 原帖 · @BubbleBrain",
    "author": "BubbleBrain",
    "publishedAt": "2026-07-31T01:20:28.898Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082999806948229386.mp4",
    "posterUrl": "/posters/x/2082999806948229386.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "My first try for MiniMax H3 from @Hailuo_AI . It is a good model.\n\nWelcome back! \n\nPrompt: https://youvibe.cn/s/ezDD4QBRlfkk \n\n------prompt---\nA UI/UX motion showcase video for a high-performance luxury sports car brand landing page, created as an image-to-video project based on Product Image 1, with the car in the image remaining the core subject throughout.\n\nFirst screen – Hero Section:\n\nA premium black background, with the car positioned at the center of the page as the main visual focus. The camera slowly pushes in. A minimalist navigation bar appears at the top of the page:\n\n\"MODEL\"\n\"DESIGN\"\n\"PERFORMANCE\"\n\"TECHNOLOGY\"\n\"CONTACT\"\nOn the left side, a large brand headline appears:\n\"UNLEASH THE FUTURE\"\nBelow it, a subtitle reads:\n\"PURE PERFORMANCE. LIMITLESS SPEED.\"\n\nAt the bottom, two buttons appear:\n\n\"EXPLORE MODEL\"\n\"BOOK A TEST DRIVE\"\n\nScroll down into the vehicle design showcase section:\n\nMain headline:\n\n\"DESIGNED FOR SPEED\"\n\nShowcase text:\n\"AERODYNAMIC BODY\"\n\"CARBON FIBER STRUCTURE\"\n\"PRECISION ENGINEERING\"\n\nThese are paired with close-up shots of the car’s front fascia, headlights, wheels, and body lines.\n\nContinue scrolling into the performance section:\n\nHeadline:\n\n\"ENGINEERED TO DOMINATE\"\n\nDisplay performance specification cards:\n\"POWER\"\n\"HORSEPOWER\"\n\"TORQUE\"\n\"ACCELERATION\"\nUse dynamic, high-tech animated numeric displays for:\n\n\"0–100 KM/H\"\n\"3.2 SEC\"\n\nEnter the interior showcase section:\n\nHeadline:\n\n\"DRIVER-FOCUSED INTERIOR\"\n\nShowcase text:\n\n\"PREMIUM MATERIALS\"\n\"DIGITAL COCKPIT\"\n\"ADVANCED CONTROL SYSTEM\"\nFinal screen:\nA full view of the car is displayed, while light sweeps across the body in the background.\nCentered text:\n\n\"READY TO DRIVE\"\n\nButton:\n\"CONFIGURE YOUR CAR\"\n\nVisual style:\n\nA premium automotive brand website, similar to a supercar launch page. Dark metallic materials, carbon fiber textures, dynamic lighting, aerodynamic lines, and a cinematic luxury car advertisement aesthetic. The webpage uses large, bold, slightly slanted sans-serif typography, with fast yet smooth scrolling animation, parallax effects, mouse-hover zoom interactions, highlight scanning effects, floating information cards, and animated button feedback.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082970376745726278",
    "title": "火龙与焰骑士掠过焦土世界",
    "titleEn": "Fire dragon and rider over scorched world",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《火龙与焰骑士掠过焦土世界》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Fire dragon and rider over scorched world. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "fire dragon and his rider of living flame,\nrising over a scorched world, cutting through smoke and sunset.",
    "promptSourceUrl": "https://x.com/kulufly/status/2082970376745726278",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/kulufly/status/2082970376745726278",
    "sourceLabel": "X 原帖 · @kulufly",
    "author": "Kulufly",
    "publishedAt": "2026-07-30T23:23:32.191Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082970376745726278.mp4",
    "posterUrl": "/posters/x/2082970376745726278.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "fire dragon and his rider of living flame, \nrising over a scorched world, cutting through smoke and sunset.\n\nCreated in @Hailuo_AI \n\n#hailuoai #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082964095410512241",
    "title": "手工剪纸定格短片",
    "titleEn": "Handcrafted paper-cut stop-motion short",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《手工剪纸定格短片》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 960×720。",
    "summaryEn": "An H3 output for Handcrafted paper-cut stop-motion short. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 960×720.",
    "prompt": "15-second cinematic stop motion animation in handcrafted paper cut style. Everything is physically assembled by invisible hands, one element at a time. Every object appears with authentic stop motion timing, tiny frame imperfections, subtle paper wobble, handmade placement, slight camera vibration, natural easing, tactile textures, soft shadows, and realistic depth. Nothing fades or morphs. Every piece is manually placed into the scene exactly like a professional stop motion production.\n\nScene begins with an empty dark navy paper background only.\n\n0-2s\nLarge crumpled blue paper sheets slide in from every edge and are pressed into place one after another, forming the underwater cave walls.\n\n2-4s\nPaper coral pieces appear individually from the bottom. Red coral, green seaweed, beige coral, pink coral, shells, pebbles and tiny bubbles are carefully placed one by one with visible handcrafted motion.\n\n4-6s\nThe glowing jellyfish is lowered from above. The bell lands first, then each tentacle is attached individually, gently bouncing. The jellyfish softly floats with slow natural stop motion movement.\n\n6-8s\nThe paper astronaut child slides into frame from the left. The body arrives first, then the helmet is placed on top with a tiny bounce. One arm rotates slightly as if reaching forward.\n\n8-10s\nThe white cat appears from below. The body is placed first, then the transparent helmet is carefully attached afterward. The tail makes a tiny handmade wiggle.\n\n10-12s\nSmall paper bubbles are added individually around both characters. Each bubble pops into place with slight stop motion jitter. The child and cat gently drift upward a few millimeters.\n\n12-15s\nThe complete underwater world comes alive. The jellyfish slowly pulses. Tentacles sway naturally. Sea plants lightly oscillate. Tiny bubbles continue rising. The astronaut child and cat subtly float together while holding position. Warm glow from the jellyfish softly illuminates nearby paper textures. Camera performs a slow cinematic push-in with realistic macro depth of field.\n\nPremium handcrafted stop motion, museum-quality paper art, tactile cardstock textures, realistic layered paper thickness, authentic frame-by-frame animation, tiny alignment imperfections, soft studio lighting, cinematic composition, macro photography, volumetric lighting, ultra detailed, emotionally warm, professional stop motion film, 15 seconds, no text, no watermark.",
    "promptSourceUrl": "https://x.com/AiMeowing/status/2082964098925355430",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AiMeowing/status/2082964095410512241",
    "sourceLabel": "X 原帖 · @AiMeowing",
    "author": "Nisho 👦🏻🙀",
    "publishedAt": "2026-07-30T22:58:34.604Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082964095410512241.mp4",
    "posterUrl": "/posters/x/2082964095410512241.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "960×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I challenged H3 to do Stop motion video!\n\nlet's watch .✦˚\n\nPic + Prompt in comment to test by yourself \n\nᓚᘏᗢ ─ #MiniMaxH3 from @Hailuo_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082945363431080299",
    "title": "Lilia Astra 预告片头",
    "titleEn": "Lilia Astra title sequence",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《Lilia Astra 预告片头》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Lilia Astra title sequence. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "高品質アニメ映像。\nExpression progression: eyes -> brows -> eyelids -> mouth -> gaze.\n【参照画像とキャラクター固定】\nSame girl, same person, same identity in every shot. 画像1を、この少女の顔・目・髪型・髪色・衣装・装飾・体型・装備を含む唯一の最優先参照として使用する。\n全カットで、画像1と同じ一人の少女を厳密に維持する。\n\n画像1内の文字、UI、レイアウト、背景、ポーズ見本、表情一覧、および小さく描かれた顔の差異は参照しない。\n画像1に複数の表情差分や小さな顔見本が含まれていても、それらを別人として扱わず、同一人物のバリエーション情報としてのみ扱う。\n顔の平均化、別人化、衣装交換、複製、クローン、追加人物を禁止。\n許可する変化は、表情、視線、口元、ポーズ、呼吸、自然な髪・布の動きのみ。\n表情変化は必ず 「目 -> 眉 -> まぶた -> 口 -> 視線」 の順序で見せる。\n顔は画像1の\n小さく柔らかな卵型、繊細な顎、紫から青へ層状に変化する大きな虹彩、マゼンタとシアンの反射、多層のキャッチライト、細い上まつ毛、淡い肌、長い桃色のツインテール、斜め前髪、頭の紫花と濃紫リボン\nを厳密に維持する。\n\n衣装・装飾・体型・装備は画像1の\n深い紫のショートケープ、古金の縁取りと星花刺繍、白いフリルブラウス、黒紫のコルセット、ダークブラウンの革ベルト、短い重ねスカート、白いサイハイ、紫黒と古金のブーツ、紫水晶の胸飾り\nを維持する。\n 画像1で明確に確認できる形状・配色・材質を最優先し、不明瞭な箇所のみ同一デザインとして自然に補完する。\n小道具は、金の透かし彫りと紫水晶を持つダークブラウンの革装丁の魔導書、腰の紫色のガラス小瓶、紫水晶のチャーム。\n魔導書は一冊だけで、開始時は閉じており、常に左手で支え、右手だけで表紙を開き紙のページを一枚めくる。\nガラス小瓶とチャームは腰に固定したまま使用しない。\n\n【固定レンダリング：3D toon/cel-look】\n細く制御された色付き線画。線色は黒ではなく深い紫褐色。\n陰影は2〜3層のトゥーン影と柔らかな中間グラデーション。\n配色は桃色、深紫、白、古金、宝石のマゼンタを主軸にする。\n瞳は多層、髪は真珠質、金属と水晶は局所的な鋭いハイライト。\n肌はマット、髪は絹、ケープは織布、ブラウスは薄い白布、ベルトと魔導書は革、装飾は古金、宝石とガラス小瓶は透明結晶として材質を分離する。\n顔、まつ毛、髪束、フリル、金装飾、革、魔法図形、濡れた石床、真鍮扉、アーチ窓は高密度に描写する。\n太い輪郭線、単層だけの平坦なセル塗り、汎用的な3Dアニメ顔、滑らかなプラスチックCG素材、低密度背景、実写、半実写、鈍い色、画風混在を禁止。\n\n【世界とトーン】\n真夜中の王立魔導文書庫。\n濡れた黒い石床、背の高いアーチ窓、真鍮の星形錠、奥まで続く暗い書架。\n黒い余白、アイボリー、深紫、マゼンタ、古金の幾何学色面を使った映画ティザー。\n比率はサスペンス60%、ノワールジャズ40%。\n神秘的、都会的、冷静、俊敏。\nホラー、重い悲劇、陽気なジャズMVにはしない。\n物語の核は「魔導書の紋章から星が一つ盗まれた」。\n戦闘ではなく、少女が異変を発見し、残された魔法痕跡を追う瞬間を見せる。\n\n【カメラと編集設計】\nCamera state: position=three-quarter and over-shoulder; height=eye level then low angle; direction=tracks then push-in; subject scale=close-up to medium; vanishing axis=archive depth axis; purpose=to show the clue chain and reveal the missing star. Narrow split panels open into a vast archive space, creating a clear scale change.\nカメラは各動作の因果を見せるために動く。\n瞳の異変は極端な寄り、手掛かりの進行は左右非対称の分割画面、扉の解錠は低い追従、真相への接近は奥行き方向の押し込み、最後の決意は胸上の3/4正面で停止する。\n動きの最中も被写体を画面中央へ固定し続けず、斜め構図、前景遮蔽、奥行き差を使う。\nソフトディゾルブと液体モーフを使わず、すべて硬い切替、マスク、ワイプ、タイル再配置でつなぐ。\n黒画面にアイボリーの細い線が一本だけ引かれ、その線が画像1の右目の輪郭へ変わる。\n極端な顔の寄り。\n虹彩に映る魔導書の星花紋章から、星が一つだけ音もなく消える。\n\n少女は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に、ごく小さな微笑みから警戒した静かな表情へ変わり、視線を画面右下の魔導書へ落とす。\n虹彩の円形反射が古い天文レコード盤のように回転し、硬い円形ワイプとなる。\n画面が左右非対称の二分割へスナップする。\n左の細いパネルには閉じた魔導書の金具の接写。\n右の大きなパネルには濡れた石床を進む彼女のブーツと長いケープの裾。\n深紫とマゼンタの色面、細い古金フレームが一片ずつ組み上がる。\n\n少女が右手の指先を魔導書の紫水晶の数センチ手前で止めると、水晶が一度だけ自発的に点灯し、その点灯を起点に細いマゼンタの魔法痕跡線が水晶から扉へ伸びる。\n扉は痕跡線が到達する前まで閉じたまま。\n痕跡線の先端が扉中央の星形錠の中心と正確に重なると、真鍮の錠が一度だけ回転して扉が垂直に開く。\n扉の縦線がそのまま硬い縦カットになる。\n開いた暗がりから、人物ではない平面的な長い影が石床を横切り、全画面を一度だけワイプする。\nこの影に顔、身体、立体感を与えない。\n影が抜けると、少女は左手で支えた一冊の魔導書を右手で開き、紙のページを一枚だけめくる。\n紙のページ端が次の画面を硬く切る。\n\nページには幾何学的な星図だけが浮かび、読める文字は出さない。\n三つの細いフレームがタイル状にスナップして、瞳の反射、星形錠、途切れたマゼンタの痕跡を同時に示す。\n少女は一歩だけ進み、低い背面カメラが肩越しまで上昇して、奥の巨大なアーチ窓と一点だけ欠けた星座を見せる。\n\n巨大な英字「A」の輪郭が画面外から滑り込み、その内部だけが次の映像を見せる硬い文字マスクになる。\n少女が右手の掌を上へ向けると、その掌の上に星花型の魔法陣が一つだけ展開する。\n魔法陣は古金の外周、マゼンタの細線、紫水晶色の中心で構成され、一点の星だけ黒く欠けている。\n\n欠けた一点から赤紫の直線が横へ走り、胸上の構図へ硬く切る。\nカメラが魔法陣から顔へ短く鋭く振る。\n少女は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に異変を理解し、怯えず、静かな決意へ変わる。\n魔法陣を消し、右手で魔導書を閉じ、閉じた表紙へ右手を置く。\n魔導書の紫水晶と胸飾りだけが一度ずつ呼応して点灯する。\n最後は黒、深紫、アイボリーの矩形タイルが硬くスナップして一枚の画面を組み上げる。\n終端タイプはポーズロック（pose lock）。\n\n左に少女の胸上3/4正面、右に十分な黒い余白。\n右側に英字タイトル 「LILIA ASTRA: THE STOLEN STAR」 を一度だけ、完全に正しい綴りで表示する。\nその下に 「COMING SOON」 を一度だけ表示する。\nほかの文字、役職、名前、漢字、ひらがな、カタカナ、疑似文字、読めない記号は一切表示しない。\n少女は一冊の閉じた魔導書を左手で支え、右手を表紙に置いたまま静止する。\n平面的な影は画面右外縁で止まり、新しい要素は入らない。\n\nタイトルを二秒間完全に保持する。\n緊張した和音と一発のドラムヒットの余韻が終わる正確なフレームでカットする。\n【音響】\n完全オリジナルの15秒タイトル音楽。既存曲や既存メロディを模倣しない。\n\n低いドローン、緊張した弦のピチカート、冷たいシンセパルス、低いキック、疎らなブラシ、短いウォーキングベース断片、短いバリトンサックス、クリップしたブラスアクセント。\n冒頭は低域とハイハット。\n低いキックは3秒付近。\nジャズベースの動きは6秒付近。\n短いサックスとブラスのリフは10秒付近。\n13秒で緊張した和音と一発のドラムヒットを置き、その余韻を15秒まで保持する。\n\n台詞、ナレーション、歌唱なし。\n効果音は見える動作に同期し、その瞬間だけ鳴る。\n革装丁の魔導書が開閉する瞬間だけ、短い革のきしみ。\n紙のページ一枚がめくれる瞬間だけ、乾いた紙の音。\n真鍮の星形錠が回転する瞬間だけ、短く乾いた真鍮の鳴り。\n\nブーツが濡れた石床へ着地する瞬間だけ、湿った石床の足音。\nケープと髪が動く瞬間だけ、軽い布の擦れ。\n紫水晶が点灯する瞬間だけ、細い結晶共鳴音。\n材質を混同しない。\n【禁止】\n別人化、顔崩れ、目色変更、髪型変更、衣装簡略化、衣装交換、アクセサリー欠落、装備の増殖、魔導書の複製、手指破綻、余分な腕、追加人物、群衆、クローン、顔平均化、戦闘、爆発、流血、恐怖顔、悪霊、ホラー照明、実写化、半実写化、画風変化、太い輪郭線、低詳細背景、プラスチック質感、ソフトディゾルブ、液体モーフ、過剰な被写界深度、手ぶれ、読めない文字、綴り間違い、重複タイトル、字幕、ロゴ、透かし。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2082945363431080299",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/haruuraeadss/status/2082945363431080299",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-07-30T21:44:08.552Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082945363431080299.mp4",
    "posterUrl": "/posters/x/2082945363431080299.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "anime"
    ],
    "scenes": [
      "nature",
      "sci-fi"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "映画トレーラー風\n\n@Hailuo_AI　\n#MiniMaxH3　\nプロンプト👇\n                                                                                                                                                                                                                                  \n高品質アニメ映像。\nExpression progression: eyes -> brows -> eyelids -> mouth -> gaze.\n【参照画像とキャラクター固定】\nSame girl, same person, same identity in every shot. 画像1を、この少女の顔・目・髪型・髪色・衣装・装飾・体型・装備を含む唯一の最優先参照として使用する。\n全カットで、画像1と同じ一人の少女を厳密に維持する。\n\n画像1内の文字、UI、レイアウト、背景、ポーズ見本、表情一覧、および小さく描かれた顔の差異は参照しない。\n画像1に複数の表情差分や小さな顔見本が含まれていても、それらを別人として扱わず、同一人物のバリエーション情報としてのみ扱う。\n顔の平均化、別人化、衣装交換、複製、クローン、追加人物を禁止。\n許可する変化は、表情、視線、口元、ポーズ、呼吸、自然な髪・布の動きのみ。\n表情変化は必ず 「目 -> 眉 -> まぶた -> 口 -> 視線」 の順序で見せる。\n顔は画像1の\n小さく柔らかな卵型、繊細な顎、紫から青へ層状に変化する大きな虹彩、マゼンタとシアンの反射、多層のキャッチライト、細い上まつ毛、淡い肌、長い桃色のツインテール、斜め前髪、頭の紫花と濃紫リボン\nを厳密に維持する。\n\n衣装・装飾・体型・装備は画像1の\n深い紫のショートケープ、古金の縁取りと星花刺繍、白いフリルブラウス、黒紫のコルセット、ダークブラウンの革ベルト、短い重ねスカート、白いサイハイ、紫黒と古金のブーツ、紫水晶の胸飾り\nを維持する。\n 画像1で明確に確認できる形状・配色・材質を最優先し、不明瞭な箇所のみ同一デザインとして自然に補完する。\n小道具は、金の透かし彫りと紫水晶を持つダークブラウンの革装丁の魔導書、腰の紫色のガラス小瓶、紫水晶のチャーム。\n魔導書は一冊だけで、開始時は閉じており、常に左手で支え、右手だけで表紙を開き紙のページを一枚めくる。\nガラス小瓶とチャームは腰に固定したまま使用しない。\n\n【固定レンダリング：3D toon/cel-look】\n細く制御された色付き線画。線色は黒ではなく深い紫褐色。\n陰影は2〜3層のトゥーン影と柔らかな中間グラデーション。\n配色は桃色、深紫、白、古金、宝石のマゼンタを主軸にする。\n瞳は多層、髪は真珠質、金属と水晶は局所的な鋭いハイライト。\n肌はマット、髪は絹、ケープは織布、ブラウスは薄い白布、ベルトと魔導書は革、装飾は古金、宝石とガラス小瓶は透明結晶として材質を分離する。\n顔、まつ毛、髪束、フリル、金装飾、革、魔法図形、濡れた石床、真鍮扉、アーチ窓は高密度に描写する。\n太い輪郭線、単層だけの平坦なセル塗り、汎用的な3Dアニメ顔、滑らかなプラスチックCG素材、低密度背景、実写、半実写、鈍い色、画風混在を禁止。\n\n【世界とトーン】\n真夜中の王立魔導文書庫。\n濡れた黒い石床、背の高いアーチ窓、真鍮の星形錠、奥まで続く暗い書架。\n黒い余白、アイボリー、深紫、マゼンタ、古金の幾何学色面を使った映画ティザー。\n比率はサスペンス60%、ノワールジャズ40%。\n神秘的、都会的、冷静、俊敏。\nホラー、重い悲劇、陽気なジャズMVにはしない。\n物語の核は「魔導書の紋章から星が一つ盗まれた」。\n戦闘ではなく、少女が異変を発見し、残された魔法痕跡を追う瞬間を見せる。\n\n【カメラと編集設計】\nCamera state: position=three-quarter and over-shoulder; height=eye level then low angle; direction=tracks then push-in; subject scale=close-up to medium; vanishing axis=archive depth axis; purpose=to show the clue chain and reveal the missing star. Narrow split panels open into a vast archive space, creating a clear scale change.\nカメラは各動作の因果を見せるために動く。\n瞳の異変は極端な寄り、手掛かりの進行は左右非対称の分割画面、扉の解錠は低い追従、真相への接近は奥行き方向の押し込み、最後の決意は胸上の3/4正面で停止する。\n動きの最中も被写体を画面中央へ固定し続けず、斜め構図、前景遮蔽、奥行き差を使う。\nソフトディゾルブと液体モーフを使わず、すべて硬い切替、マスク、ワイプ、タイル再配置でつなぐ。\n黒画面にアイボリーの細い線が一本だけ引かれ、その線が画像1の右目の輪郭へ変わる。\n極端な顔の寄り。\n虹彩に映る魔導書の星花紋章から、星が一つだけ音もなく消える。\n\n少女は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に、ごく小さな微笑みから警戒した静かな表情へ変わり、視線を画面右下の魔導書へ落とす。\n虹彩の円形反射が古い天文レコード盤のように回転し、硬い円形ワイプとなる。\n画面が左右非対称の二分割へスナップする。\n左の細いパネルには閉じた魔導書の金具の接写。\n右の大きなパネルには濡れた石床を進む彼女のブーツと長いケープの裾。\n深紫とマゼンタの色面、細い古金フレームが一片ずつ組み上がる。\n\n少女が右手の指先を魔導書の紫水晶の数センチ手前で止めると、水晶が一度だけ自発的に点灯し、その点灯を起点に細いマゼンタの魔法痕跡線が水晶から扉へ伸びる。\n扉は痕跡線が到達する前まで閉じたまま。\n痕跡線の先端が扉中央の星形錠の中心と正確に重なると、真鍮の錠が一度だけ回転して扉が垂直に開く。\n扉の縦線がそのまま硬い縦カットになる。\n開いた暗がりから、人物ではない平面的な長い影が石床を横切り、全画面を一度だけワイプする。\nこの影に顔、身体、立体感を与えない。\n影が抜けると、少女は左手で支えた一冊の魔導書を右手で開き、紙のページを一枚だけめくる。\n紙のページ端が次の画面を硬く切る。\n\nページには幾何学的な星図だけが浮かび、読める文字は出さない。\n三つの細いフレームがタイル状にスナップして、瞳の反射、星形錠、途切れたマゼンタの痕跡を同時に示す。\n少女は一歩だけ進み、低い背面カメラが肩越しまで上昇して、奥の巨大なアーチ窓と一点だけ欠けた星座を見せる。\n\n巨大な英字「A」の輪郭が画面外から滑り込み、その内部だけが次の映像を見せる硬い文字マスクになる。\n少女が右手の掌を上へ向けると、その掌の上に星花型の魔法陣が一つだけ展開する。\n魔法陣は古金の外周、マゼンタの細線、紫水晶色の中心で構成され、一点の星だけ黒く欠けている。\n\n欠けた一点から赤紫の直線が横へ走り、胸上の構図へ硬く切る。\nカメラが魔法陣から顔へ短く鋭く振る。\n少女は「目 -> 眉 -> まぶた -> 口 -> 視線」の順に異変を理解し、怯えず、静かな決意へ変わる。\n魔法陣を消し、右手で魔導書を閉じ、閉じた表紙へ右手を置く。\n魔導書の紫水晶と胸飾りだけが一度ずつ呼応して点灯する。\n最後は黒、深紫、アイボリーの矩形タイルが硬くスナップして一枚の画面を組み上げる。\n終端タイプはポーズロック（pose lock）。\n\n左に少女の胸上3/4正面、右に十分な黒い余白。\n右側に英字タイトル 「LILIA ASTRA: THE STOLEN STAR」 を一度だけ、完全に正しい綴りで表示する。\nその下に 「COMING SOON」 を一度だけ表示する。\nほかの文字、役職、名前、漢字、ひらがな、カタカナ、疑似文字、読めない記号は一切表示しない。\n少女は一冊の閉じた魔導書を左手で支え、右手を表紙に置いたまま静止する。\n平面的な影は画面右外縁で止まり、新しい要素は入らない。\n\nタイトルを二秒間完全に保持する。\n緊張した和音と一発のドラムヒットの余韻が終わる正確なフレームでカットする。\n【音響】\n完全オリジナルの15秒タイトル音楽。既存曲や既存メロディを模倣しない。\n\n低いドローン、緊張した弦のピチカート、冷たいシンセパルス、低いキック、疎らなブラシ、短いウォーキングベース断片、短いバリトンサックス、クリップしたブラスアクセント。\n冒頭は低域とハイハット。\n低いキックは3秒付近。\nジャズベースの動きは6秒付近。\n短いサックスとブラスのリフは10秒付近。\n13秒で緊張した和音と一発のドラムヒットを置き、その余韻を15秒まで保持する。\n\n台詞、ナレーション、歌唱なし。\n効果音は見える動作に同期し、その瞬間だけ鳴る。\n革装丁の魔導書が開閉する瞬間だけ、短い革のきしみ。\n紙のページ一枚がめくれる瞬間だけ、乾いた紙の音。\n真鍮の星形錠が回転する瞬間だけ、短く乾いた真鍮の鳴り。\n\nブーツが濡れた石床へ着地する瞬間だけ、湿った石床の足音。\nケープと髪が動く瞬間だけ、軽い布の擦れ。\n紫水晶が点灯する瞬間だけ、細い結晶共鳴音。\n材質を混同しない。\n【禁止】\n別人化、顔崩れ、目色変更、髪型変更、衣装簡略化、衣装交換、アクセサリー欠落、装備の増殖、魔導書の複製、手指破綻、余分な腕、追加人物、群衆、クローン、顔平均化、戦闘、爆発、流血、恐怖顔、悪霊、ホラー照明、実写化、半実写化、画風変化、太い輪郭線、低詳細背景、プラスチック質感、ソフトディゾルブ、液体モーフ、過剰な被写界深度、手ぶれ、読めない文字、綴り間違い、重複タイトル、字幕、ロゴ、透かし。",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082937573559722460",
    "title": "漂浮生态城中的探险者与猫",
    "titleEn": "Explorer and Cat in a Floating Eco-City",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《漂浮生态城中的探险者与猫》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 43 秒 · 1280×720。",
    "summaryEn": "An H3 output for Explorer and Cat in a Floating Eco-City. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 43s · 1280×720.",
    "prompt": "Ultra cinematic AAA adventure game trailer, one continuous seamless camera shot, Unreal Engine 5 quality, hyper realistic, 60 FPS, cinematic lighting, premium animation, realistic physics, smooth transitions, no text, no logo.  The camera begins behind a young silver-haired explorer and his fluffy orange cat standing together on a giant ancient tree branch overlooking a breathtaking futuristic nature city in the year 2100. Giant waterfalls flow through floating islands, enormous glowing trees surround eco-friendly futuristic towers, flying drones move naturally between the trees, distant floating platforms connect the landscape, soft volumetric sunlight fills the valley, leaves gently drift through the air, birds fly across the sky, mist slowly rolls through the forest.  Far away, a colossal ancient guardian peacefully walks through the valley between giant trees, partially hidden by fog, creating mystery and wonder.  The explorer and his cat quietly observe the distant guardian.  After two seconds, the explorer slowly raises his left hand.  He gently touches the glowing wrist device on his arm.  A soft circular pulse of blue light expands through the air.  Elegant transparent holographic interface panels smoothly appear around him with premium glass animation.  The interface feels futuristic, clean and luxurious.  Panels gracefully float into position using smooth spring motion.  Explorer Inventory.  Adventure Gear.  Explorer Suit.  Accessories.  Energy Tools.  Skills.  Artifacts.  Companion Gear.  Quest Journal.  Player Statistics.  Beautiful holographic cards slowly rotate in space.  Crystal accessories sparkle.  Energy gadgets softly glow.  Explorer suit pieces rotate in three dimensions.  Energy bow slowly turns with animated light effects.  The cursor appears naturally with smooth realistic movement.  It selects the glowing Energy Bow.  The selected card enlarges slightly.  The bow transforms into blue light particles and gracefully flies onto the explorer's back.  A quiver of glowing energy arrows materializes beside it.  Next the explorer suit activates.  Shoulder pieces.  Chest protection.  Explorer gloves.  Travel boots.  Utility belt.  Crystal rings.  Energy necklace.  Each item gently assembles onto the explorer with elegant magnetic energy effects.  Blue energy flows through every piece as it activates.  The cat watches curiously.  The cursor opens Companion Gear.  A lightweight explorer harness, glowing collar, utility backpack and protective accessories gently assemble onto the cat.  The cat happily stretches, shakes its body and looks proudly toward the valley.  The holographic interface slowly dissolves into glowing particles.  The environment becomes quiet again.  The explorer reaches for the glowing Energy Bow.  The cat lowers into an excited ready position.  They exchange a quick glance.  Without hesitation they begin running together across the giant tree branch.  The camera follows closely from behind with cinematic motion.  At the edge of the branch they leap together into the breathtaking futuristic valley.  The explorer's cape flows dramatically in the wind.  The cat glides gracefully beside him.  Leaves swirl around them.  Flying drones pass overhead.  The waterfalls rush below.  The camera follows their descent through the incredible landscape.  In the distance, the colossal guardian slowly turns its head after sensing their arrival.  The adventure has only just begun.  Fade to black.  Premium PlayStation first-party game trailer quality.  Natural facial animation.  Realistic cloth simulation.  Dynamic hair physics.  Cinematic camera movement.  AAA environmental storytelling.  Award-winning game cinematic.  Seamless single take.  No combat.  No violence.  Focus on exploration, discovery and adventure.",
    "promptSourceUrl": "https://x.com/AiMeowing/status/2082937578513186893",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AiMeowing/status/2082937573559722460",
    "sourceLabel": "X 原帖 · @AiMeowing",
    "author": "Nisho 👦🏻🙀",
    "publishedAt": "2026-07-30T21:13:11.302Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082937573559722460.mp4",
    "posterUrl": "/posters/x/2082937573559722460.jpg",
    "duration": 43,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Gaming",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I challenged #MiniMaxH3 from @Hailuo_AI to do AAA game promo with UI of tools, can it be delivered well. \n\nWatch this long promo and let me know!\n(I did some cuts for hook by After effect)\n\nshort version 15 sec in comment with prompt",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082934406910513162",
    "title": "无缝循环武士刀刃重构",
    "titleEn": "Seamless-loop cinematic warrior blade reconstruction",
    "model": "MiniMax H3（来源标注）",
    "mode": "FL2VA",
    "summary": "《无缝循环武士刀刃重构》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Seamless-loop cinematic warrior blade reconstruction. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "@Image1  15s | 16:9 | 1440p | 24fps | PERFECT SEAMLESS LOOP  [REFERENCE LOCK] Render exactly as @image1. Preserve the character exactly as shown. Do not redesign, reinterpret, or modify any aspect of her appearance.  Do NOT alter her hairstyle, bangs, facial proportions, eye color, expression, clothing, coat silhouette, sash placement, body proportions, sword guard, hilt ornament, blade length, or overall composition.  The character must always remain holding the sword throughout the entire sequence. The sword never leaves her hand.  ---  [LOOP DESIGN]  Create a mathematically perfect infinite loop.  Frame 1 and Frame 360 must match perfectly.  Maintain identical: • Pose • Camera angle • Lens • Lighting • Sword position • Character expression • Particle placement • Composition  The sword continuously shatters and reconstructs while remaining in the character's grip, creating an endless cycle with no visible transition.  ---  [COLOR PALETTE]  Background: Smooth vertical gradient  Top: Luminous Cyan-Teal (#2AA6B4)  Bottom: Deep Petrol (#0F4A55)  No environment. No horizon. No props. Only a clean cinematic graphic backdrop.  Character  Coat: Bright Coral (#FF9273)  Inner Garment: Deep Ocean (#14454E)  Hair: Natural black with subtle blue-white rim highlights.  Metal & Glass: Bright silver-white.  Accent colors appear ONLY through light refraction and chromatic dispersion.  Gold (#FFD27A)  Electric Cyan (#6FE3F5)  Soft Magenta (#FF9ED8)  The main palette must always dominate.  ---  [VISUAL STYLE]  Premium 2.5D Anime High-end cinematic illustration Ultra-clean rendering Feature-film quality  Bright high-key lighting  Elegant bloom  Sharp specular highlights  Volumetric glow  HDR dynamic range  Soft filmic contrast  Luminous skin with subtle subsurface scattering.  Heavy cotton twill coat with realistic cloth simulation.  Hair moves naturally in layered strands with soft flyaways.  Glass has authentic thickness, edge refraction, internal fracture lines, microscopic scratches, and realistic prism reflections.  ---  [EFFECTS]  Progressively increase visual intensity.  • Time Echo Motion Trails  • Liquid Mercury Blade Reconstruction  • Kaleidoscope Mirror Fractures  • Shockwave Refraction  • One-frame Negative Flash  • Glass Crack Frame Wipes  • Reverse Gravity Debris  • Chromatic Dispersion  • Dynamic Caustic Lighting  • Floating Crystal Dust  • High-speed Light Trails  • Micro Lens Distortion during impacts  Effects should evolve naturally and never become repetitive.  ---  [CAMERA]  Premium anamorphic cinema language.  85mm intimate close-ups  35mm cinematic medium shots  24mm aggressive dynamic movement  Oval bokeh  Blue anamorphic streak flares  Subtle barrel distortion  180-degree shutter  Depth of field shifts emotionally.  Shallow during hero moments.  Deep during action moments.  Rack focus:  Blade → Eyes at 2.4s  Blade → Eyes at 14.2s  Smooth cinematic easing with invisible acceleration.  ---  [LIGHTING]  Hard cinematic sunlight originates from screen-right behind the character.  Strong rim lighting illuminates  Hair  Jawline  Shoulders  Sword edge  Soft cyan bounce light fills from the background.  Floating shards cast constantly changing colored kick lights.  Shadows remain soft, open and cinematic.  Brightness peaks around 10.4 seconds before gradually returning to the exact opening lighting by 15 seconds.  ---  [SWORD ENGRAVING]  A finely engraved  CRYPTOSAM  appears on the ricasso directly above the sword guard.  Small letter-spaced capital letters carved directly into polished steel.  This engraving is part of the blade itself.  Never an overlay.  Visible in Frame One.  Visible throughout the animation.  Visible after every blade reconstruction.  The engraving catches the sunlight whenever the camera glides across the blade.  ---  [GLASS SHARDS]  Every shard reflects a different scene.  Examples include  Blue sky  Soft white clouds  Golden wheat fields  Warm sunset  Lens flares  The character's own reflection  Moments that have not yet happened  No empty glass.  Every fragment contributes to the visual storytelling.  ---  [KINETIC TYPOGRAPHY]  Typography exists behind the character on the gradient background.  Use only premium cinematic condensed sans-serif fonts.  Animated with  Impact slams  Scroll reveals  Prism sweeps  Glass shattering  Shockwave distortion  Perspective warping  Never cover the character's face.  Timeline  0.8s  \"REFLECTION\"  Coral → Gold  2.2s  \"A BLADE THAT BREAKS\"  \"AND REFORGES FOREVER\"  Pale Cyan → White  5.0s  \"ONE IMAGE\"  \"FIFTEEN SECONDS\"  White → Coral  7.2s  \"EVERY SHARD REMEMBERS\"  Cyan → Magenta → Gold  10.6s  \"NO CUT\"  Gold → White  Warp under expanding shockwave.  13.2s  \"MIRRORFALL\"  Silver → Pale Gold  Soft coral glow.  Typography completely dissolves before 15 seconds.  ---  [SHOT 01]  0.0–1.6s  LOOP ANCHOR  The character firmly grips the sword with both hands.  An ultra-fast macro camera races along the polished blade toward the hilt.  Cloud reflections glide across the steel.  An anamorphic flare slides along the cutting edge.  The engraved word  CRYPTOSAM  briefly catches the sunlight.  Without releasing the sword, the blade suddenly fractures into thousands of crystalline shards while the hilt remains firmly in her hands.  A glass-crack transition wipes into the next shot.  ---  [SHOT 02]  1.6–3.2s  BULLET-TIME ORBIT  Time freezes.  The character remains perfectly still while tightly holding the sword hilt.  The shattered blade floats in front of her hand.  The camera performs a smooth 180-degree orbit through the suspended crystal field.  Each shard reflects a different reality.  Reverse-gravity debris hangs motionless.  As the orbit finishes, time resumes and the floating shards accelerate back toward the hilt, beginning to reconstruct the blade directly in her grip.",
    "promptSourceUrl": "https://x.com/ogbenniasamuel2/status/2082934524019912805",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ogbenniasamuel2/status/2082934406910513162",
    "sourceLabel": "X 原帖 · @ogbenniasamuel2",
    "author": "𝗰𝗿𝘆𝗽𝘁𝗼__𝘀𝗮𝗺 📑✍️🌹",
    "publishedAt": "2026-07-30T21:00:36.314Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082934406910513162.mp4",
    "posterUrl": "/posters/x/2082934406910513162.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "At this point, I can confidently say @Hailuo_AI #MiniMaxH3 is absolutely GOATED. 🔥\n\nJust look at this.\nPrompt below. 👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082932693613043716",
    "title": "节拍同步 K-Pop 多模态 MV",
    "titleEn": "Beat-Synced K-Pop Multimodal MV",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《节拍同步 K-Pop 多模态 MV》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Beat-Synced K-Pop Multimodal MV. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Make an MV from Audio1 . Show these Kpop members from Image1 singing and dancing to the music. Add coarse grain, glitch effects, and grunge effects. Keep the edit fast and use hard cuts only—no fades or soft transitions. Cuts should occur within 3 seconds. Make the cuts based on the beat of the song. Use the typographic reference from Image2",
    "promptSourceUrl": "https://x.com/aisearchio/status/2082932695479517354",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/aisearchio/status/2082932693613043716",
    "sourceLabel": "X 原帖 · @aisearchio",
    "author": "⚡AI Search⚡",
    "publishedAt": "2026-07-30T20:53:47.832Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082932693613043716.mp4",
    "posterUrl": "/posters/x/2082932693613043716.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is pretty good at making MVs.\n\nBecause it's multimodal, it can take in audio, images, and video. Here, I uploaded a song made from ACE Step and images from Nano Banana. \n\nIt's able to sync the video to the audio quite well.\n\nPrompt:",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082919486399943073",
    "title": "KALDR 极冷香水多模态广告",
    "titleEn": "KALDR cold perfume commercial multi-reference",
    "model": "MiniMax H3（来源标注）",
    "mode": "Unknown",
    "summary": "《KALDR 极冷香水多模态广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1700×720。",
    "summaryEn": "An H3 output for KALDR cold perfume commercial multi-reference. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1700×720.",
    "prompt": "REFERENCE USAGE:\nImage 1 — the KALDR bottle. Preserve the smoked charcoal glass, bevelled edges, pale blue liquid, brushed gunmetal cap and the etched silver KALDR mark exactly.\nImage 2 — the closing brand lockup. Preserve the letterforms, tracking, silver finish and the COLD KEEPS line exactly. It appears only in the final shot, alone on black.\nImage 3 — the location, palette, cloud, surf and film texture for the whole film.\nImage 4 — a three panel character sheet of the woman. Use it only to match her face, hair, frost detail and black greatcoat. Do not reproduce the sheet's panel layout, its grey backdrop or its flat studio lighting. She must appear as a single person inside the Image 3 environment, lit by that location.\nAudio 1 — the voice for the woman's spoken line. Match its timbre, pitch, accent, breathiness and pacing exactly. Use it only for her voice, not as background audio or ambience.\n\nPremium luxury fragrance brand film. Photoreal 35mm anamorphic, high contrast, fine grain, cold specular highlights, deep crushed blacks. Palette strictly black sand, gunmetal, glacial blue and silver, with no warm tones anywhere. Setting: a black volcanic beach on a northern coast at blue hour under heavy storm cloud. Tone: severe, violent, expensive. This is a fast, aggressive fashion edit, not a calm product film. Never slow, never ceremonial, never a slideshow of pretty holds.\n\n[0-1.5s] Cold open, no establishing shot. Camera at water level, lens half submerged, [Static shot]. A sheet of freezing surf races straight at the lens. Her boots stride hard into frame through the water, coat hem whipping. HARD CUT.\n\n[1.5-2.6s] Extreme macro insert. Frost crystals race across black glass and flood the etched KALDR mark with rime, filling the letters. HARD CUT.\n\n[2.6-4.2s] Behind her, low and close, [Tracking shot] pushing fast. She walks straight into an oncoming wave that towers over her, coat streaming sideways, never breaking stride. HARD CUT.\n\n[4.2-5.4s] Reverse, low angle from the water. The wave collapses over her. Camera [Shake] on the impact. Cut on the peak of the white spray. HARD CUT.\n\n[5.4-9s] SLOW MOTION. The wave freezes solid around her mid collapse. She stands inside a shell of blue black ice, the spray suspended in the air as a thousand hanging shards, her coat frozen mid snap, her face calm and unbothered. Camera [Tracking shot] orbits right through the suspended shards, holding her at centre. This is the hero image of the film.\n\n[9-9.8s] Snap back to real time. The ice shell detonates outward. Camera [Pull out] fast and hard through the exploding shards. HARD CUT.\n\n[9.8-12.6s] Tight close up, [Push in]. She walks out of the collapsing ice straight at the lens, ice water running off her coat, looking directly into camera. She says quietly: \"I was never meant to melt.\" She holds the look for a beat after the line. HARD CUT.\n\n[12.6-13.8s] The bottle alone, upright on wet black sand, shot low and close on a macro lens. Frost blooms outward from its base toward the lens. No text of any kind in this shot. HARD CUT.\n\n[13.8-15s] Pure black frame. The brand lockup from Image 2 appears alone, centred, exactly as in the reference. The wordmark KALDR emerges from darkness at low opacity, letter spacing opening over 0.4 seconds into sharp focus with a faint edge glow, and the line COLD KEEPS sits smaller beneath it. Nothing else is in this frame. Critical: the bottle, the woman, the beach and the wordmark must never share a frame. The lockup must never overlap or be occluded by the bottle, and the etched mark on the glass must never appear in the same shot as the lockup. Hold the lockup to the end.\n\nDialogue and lip sync: only one line is spoken, in the shot beginning at 9.8 seconds: \"I was never meant to melt.\" Match the voice in Audio 1 exactly, including its timbre, breathiness, pacing and the way the sentence lands without rising at the end. Do not substitute a different voice, do not add reverb or processing that changes its character. Her lip movement must match the words frame accurately and stop when the line ends. Her breath fogs on the words. Do not repeat the line and do not add any other spoken words anywhere in the film.\n\nPacing: hold every cut until its stated timecode and do not shorten the last two shots. The first four shots are fast and aggressive. The slow motion shot is the only long hold in the film. Everything after it is fast again.\n\nAudio: hard coastal wind, heavy surf, a low sub bass drone. At 5.8 seconds, one deep impact hit, a splintering glass freeze crackle and a reversed swell. Then cut the wind, the surf and all low end to complete silence from 6 seconds to 9 seconds, leaving only isolated ice ticks and faint crystal ring in that window. This silence is mandatory and must be obvious. At 9 seconds the full mix slams back in with a hard low kick and the crash of shattering ice into water. As she speaks, duck the wind, surf and low end so her voice sits forward and dry. The final second holds one sustained low string chord decaying into wind. No voiceover, no narrator, no lyrics, no recognisable melody.\n\nDo not add other people, boats, buildings, vehicles, modern objects, subtitles, on screen captions, watermarks, or any text other than specified. Do not warp or distort the bottle or the wordmark. No cartoon or plastic CGI water or ice. Keep skin texture real, no over smoothing. Never let the ice or spray form a face, eyes, or a creature.",
    "promptSourceUrl": "https://x.com/abulu8/status/2082919493060514051",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/abulu8/status/2082919486399943073",
    "sourceLabel": "X 原帖 · @abulu8",
    "author": "Syed Abuthahir ∞",
    "publishedAt": "2026-07-30T20:01:18.987Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082919486399943073.mp4",
    "posterUrl": "/posters/x/2082919486399943073.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1700×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "KALDR. Cold, sharp, unforgiving. 🧊\n\nTested a perfume commercial using multiple references, 4 Image references + 1 voice clip. \n\nReally impressed with how #MiniMaxH3 handled the reference consistency and brought everything together. @Hailuo_AI\n\nPrompt below 👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082916440403742779",
    "title": "极致电影感场景挑战",
    "titleEn": "Most cinematic scenes challenge",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《极致电影感场景挑战》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1698×720。",
    "summaryEn": "An H3 output for Most cinematic scenes challenge. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1698×720.",
    "prompt": "The most cinematic scenes you can make.",
    "promptSourceUrl": "https://x.com/JamianGerard/status/2082916440403742779",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/JamianGerard/status/2082916440403742779",
    "sourceLabel": "X 原帖 · @JamianGerard",
    "author": "Jamian Gerard",
    "publishedAt": "2026-07-30T19:49:12.765Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082916440403742779.mp4",
    "posterUrl": "/posters/x/2082916440403742779.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1698×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Got early access to @Hailuo_AI’s new video model, #MiniMaxH3, so I tested it with one simple prompt:\n\n“The most cinematic scenes you can make.”\n\nSometimes your prompts do not have to be long. This is what it created.\n\nAny other CPPs who want to join the challenge, feel free 🎬",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082909383424446745",
    "title": "奇幻卡组构筑战斗界面",
    "titleEn": "Fantasy Deck-Builder Battle UI",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《奇幻卡组构筑战斗界面》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Fantasy Deck-Builder Battle UI. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use Image 1 for the board, art direction, and lighting. Use Image 2 for the exact UI style — preserve its card shape, filigree, colors, and iconography.\n[0–2s] Slow isometric drift across the courtyard. The card fan deals in from the bottom edge, five cards arcing into place one at a time with a slight overshoot and settle. The energy orb fills to 3.\n[2–5s] Cursor lifts the second card; it scales up, tilts toward camera, and glows. Dragging it over the left enemy paints three hex tiles red with a soft pulse. Release — the card burns from the bottom edge upward and dissolves into embers.\n[5–8s] Camera snaps to a low three-quarter angle for the resolution: a shockwave crosses the three red hexes, damage numbers arc up and fade, one skeleton shatters into bone fragments that settle physically on the tiles.\n[8–11s] Return to isometric. Energy drops to 1. The remaining cards slide left to close the gap. Cursor hovers END TURN; the button pulses gold.\n[11–15s] Enemy turn: a red banner wipes across the top of frame, the surviving skeletons step two hexes forward with staggered timing, and a threat marker appears over the player unit. The banner wipes out. Cards deal in again for the next turn. Keep all UI perfectly screen-locked and crisp regardless of camera movement. Text must be legible and spelled correctly.\nAudio: parchment card slides, a wooden clack on card placement, a low bone-crunch impact, a soft chime on energy spend, muted lute drone underneath.\nNegative: no UI drift or warping, no invented card names, no misspellings, no gore, no blood, no camera shake during UI interaction.",
    "promptSourceUrl": "https://x.com/AllaAisling/status/2082909383424446745",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AllaAisling/status/2082909383424446745",
    "sourceLabel": "X 原帖 · @AllaAisling",
    "author": "Alexandra Aisling",
    "publishedAt": "2026-07-30T19:21:10.250Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082909383424446745.mp4",
    "posterUrl": "/posters/x/2082909383424446745.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Gaming",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Trying another direction with @Hailuo_AI : Gaming visuals \"Deck-Builder Turn\" \nIsometric tactics roguelike. All the interest is in UI choreography, not action.   \n\nPrompt for video: Use Image 1 for the board, art direction, and lighting. Use Image 2 for the exact UI style — preserve its card shape, filigree, colors, and iconography. \n[0–2s] Slow isometric drift across the courtyard. The card fan deals in from the bottom edge, five cards arcing into place one at a time with a slight overshoot and settle. The energy orb fills to 3. \n[2–5s] Cursor lifts the second card; it scales up, tilts toward camera, and glows. Dragging it over the left enemy paints three hex tiles red with a soft pulse. Release — the card burns from the bottom edge upward and dissolves into embers. \n[5–8s] Camera snaps to a low three-quarter angle for the resolution: a shockwave crosses the three red hexes, damage numbers arc up and fade, one skeleton shatters into bone fragments that settle physically on the tiles. \n[8–11s] Return to isometric. Energy drops to 1. The remaining cards slide left to close the gap. Cursor hovers END TURN; the button pulses gold. \n[11–15s] Enemy turn: a red banner wipes across the top of frame, the surviving skeletons step two hexes forward with staggered timing, and a threat marker appears over the player unit. The banner wipes out. Cards deal in again for the next turn. Keep all UI perfectly screen-locked and crisp regardless of camera movement. Text must be legible and spelled correctly. \nAudio: parchment card slides, a wooden clack on card placement, a low bone-crunch impact, a soft chime on energy spend, muted lute drone underneath. \nNegative: no UI drift or warping, no invented card names, no misspellings, no gore, no blood, no camera shake during UI interaction.\n #MiniMaxH3\nPrompt for the reference images 👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082897758084907132",
    "title": "沙漠秀场金属礼服时尚广告",
    "titleEn": "Luxury metallic gown desert runway fashion ad",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《沙漠秀场金属礼服时尚广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 12 秒 · 1280×720。",
    "summaryEn": "An H3 output for Luxury metallic gown desert runway fashion ad. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 12s · 1280×720.",
    "prompt": "12s cinematic luxury fashion ad, 4K, 24fps, ultra-photorealistic. A high-fashion model in a flowing iridescent metallic gown walks along a minimalist desert runway at golden hour. Cinematic aerial, tracking, orbit, close-up, and hero shots with realistic cloth physics, flowing fabric, shimmering reflections, soft wind, golden sun flare, HDR lighting, Vogue-style editorial, seamless continuity, consistent face/outfit, no flicker, no warping, no extra limbs, fade to black.",
    "promptSourceUrl": "https://x.com/noorwithwifi/status/2082897758084907132",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/noorwithwifi/status/2082897758084907132",
    "sourceLabel": "X 原帖 · @noorwithwifi",
    "author": "NoorAI",
    "publishedAt": "2026-07-30T18:34:58.553Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082897758084907132.mp4",
    "posterUrl": "/posters/x/2082897758084907132.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 handles high-fashion fabric physics and lighting transitions incredibly well.\n\nGenerated with @Hailuo_AI #MiniMaxH3\n\nPrompt:\n12s cinematic luxury fashion ad, 4K, 24fps, ultra-photorealistic. A high-fashion model in a flowing iridescent metallic gown walks along a minimalist desert runway at golden hour. Cinematic aerial, tracking, orbit, close-up, and hero shots with realistic cloth physics, flowing fabric, shimmering reflections, soft wind, golden sun flare, HDR lighting, Vogue-style editorial, seamless continuity, consistent face/outfit, no flicker, no warping, no extra limbs, fade to black.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082893990995619967",
    "title": "彩色颜料武器选择界面",
    "titleEn": "Colorful Paint-Weapon Selection Screen",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《彩色颜料武器选择界面》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Colorful Paint-Weapon Selection Screen. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use Image 1 for the character, Image 2 for the UI style, and use the uploaded weapon selection image as the reference for all paint weapons. The selectable weapons should closely match the colorful ink guns shown in the reference image, including compact paint shooters, dual paint pistols, heavy paint cannons, bucket/slosher weapons, large paint brushes, paint rollers, sniper-style paint rifles, and other stylized colorful ink weapons. Match their playful proportions, vibrant colors, chunky plastic construction, and game-like appearance while avoiding exact logos or branding.\n\n[0–2 seconds]\nHigh-angle overhead shot. The character stands on a vivid, highly saturated purple floor, looking up toward the camera, matching Image 1 exactly. A colorful game menu matching Image 2 appears on the right: START NEW GAME, CONTINUE (highlighted), SETTINGS, EXIT GAME. Player profile MINIMAX appears in the top left. The cursor selects CONTINUE.\n\n[2–4 seconds]\nThe camera pushes toward her right hand. A PAINT WEAPON SELECT panel slides in, using the uploaded weapon selection screen as its layout and visual reference. Multiple colorful paint weapons are displayed exactly like the reference image in a grid with rarity and level indicators. The cursor moves between different weapons. Every time a weapon is highlighted, the paint gun in her hands instantly transforms into the selected model with colorful ink particles, playful mechanical animation, spinning parts, and glossy paint effects. She remains standing throughout.\n\n[4–7 seconds]\nThe camera smoothly arcs around her left side. The selector rapidly cycles through many weapons from the reference image: compact paint shooters, oversized paint brushes, dual paint pistols, heavy paint cannons, bucket-style splash weapons, sniper paint rifles, giant paint rollers, and experimental ink launchers. With each selection, the weapon in her hands changes immediately to match the highlighted weapon while she naturally adjusts her grip. Every swap is accompanied by bursts of colorful ink, animated UI effects, splashing paint particles, and satisfying transition animations.\n\n[7–8.5 seconds]\nPull back to a medium shot. CONFIRM LOADOUT flashes. The cursor clicks it. All UI panels collapse and disappear. She confidently spins the selected paint weapon once before resting it on her shoulder, remaining in a relaxed standing pose.\n\n[8.5–10 seconds]\nA colorful LOADING bar fills from 0% to 100%. The purple testing room dissolves beneath spreading ink splashes that transition into the game world.\n\n[10–15 seconds]\nThe world loads into a bright, colorful ink-covered city filled with graffiti, paint-coated streets, animated billboards, moving NPCs, and playful architecture. The camera settles into a third-person view behind the character. HUD elements fade in: minimap, health, paint gauge, current weapon icon matching the selected weapon, ammo counter, and objective marker. She runs forward while firing the selected colorful paint weapon, coating walls, streets, obstacles, and scenery with thick, glossy paint that matches the weapon's color.",
    "promptSourceUrl": "https://x.com/craftian_keskin/status/2082893990995619967",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/craftian_keskin/status/2082893990995619967",
    "sourceLabel": "X 原帖 · @craftian_keskin",
    "author": "Keskin",
    "publishedAt": "2026-07-30T18:20:00.409Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082893990995619967.mp4",
    "posterUrl": "/posters/x/2082893990995619967.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Gaming",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "So I've been playing Splatoon Raiders, and here is the weapon selection screen I did with Hailuo H3. \n\nHailuo H3 Prompt; \nUse Image 1 for the character, Image 2 for the UI style, and use the uploaded weapon selection image as the reference for all paint weapons. The selectable weapons should closely match the colorful ink guns shown in the reference image, including compact paint shooters, dual paint pistols, heavy paint cannons, bucket/slosher weapons, large paint brushes, paint rollers, sniper-style paint rifles, and other stylized colorful ink weapons. Match their playful proportions, vibrant colors, chunky plastic construction, and game-like appearance while avoiding exact logos or branding.\n\n[0–2 seconds]\nHigh-angle overhead shot. The character stands on a vivid, highly saturated purple floor, looking up toward the camera, matching Image 1 exactly. A colorful game menu matching Image 2 appears on the right: START NEW GAME, CONTINUE (highlighted), SETTINGS, EXIT GAME. Player profile MINIMAX appears in the top left. The cursor selects CONTINUE.\n\n[2–4 seconds]\nThe camera pushes toward her right hand. A PAINT WEAPON SELECT panel slides in, using the uploaded weapon selection screen as its layout and visual reference. Multiple colorful paint weapons are displayed exactly like the reference image in a grid with rarity and level indicators. The cursor moves between different weapons. Every time a weapon is highlighted, the paint gun in her hands instantly transforms into the selected model with colorful ink particles, playful mechanical animation, spinning parts, and glossy paint effects. She remains standing throughout.\n\n[4–7 seconds]\nThe camera smoothly arcs around her left side. The selector rapidly cycles through many weapons from the reference image: compact paint shooters, oversized paint brushes, dual paint pistols, heavy paint cannons, bucket-style splash weapons, sniper paint rifles, giant paint rollers, and experimental ink launchers. With each selection, the weapon in her hands changes immediately to match the highlighted weapon while she naturally adjusts her grip. Every swap is accompanied by bursts of colorful ink, animated UI effects, splashing paint particles, and satisfying transition animations.\n\n[7–8.5 seconds]\nPull back to a medium shot. CONFIRM LOADOUT flashes. The cursor clicks it. All UI panels collapse and disappear. She confidently spins the selected paint weapon once before resting it on her shoulder, remaining in a relaxed standing pose.\n\n[8.5–10 seconds]\nA colorful LOADING bar fills from 0% to 100%. The purple testing room dissolves beneath spreading ink splashes that transition into the game world.\n\n[10–15 seconds]\nThe world loads into a bright, colorful ink-covered city filled with graffiti, paint-coated streets, animated billboards, moving NPCs, and playful architecture. The camera settles into a third-person view behind the character. HUD elements fade in: minimap, health, paint gauge, current weapon icon matching the selected weapon, ammo counter, and objective marker. She runs forward while firing the selected colorful paint weapon, coating walls, streets, obstacles, and scenery with thick, glossy paint that matches the weapon's color.\n\n@Hailuo_AI, #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082892924694712509",
    "title": "都市街头超能力重拳动作",
    "titleEn": "Urban street superpowered punch action sequence",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《都市街头超能力重拳动作》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 28 秒 · 1280×720。",
    "summaryEn": "An H3 output for Urban street superpowered punch action sequence. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 28s · 1280×720.",
    "prompt": "PART 1 (0–15 SECONDS)\n\nLOCATION DETAIL: Gritty urban street at golden hour dusk. Wet concrete ground reflecting golden-orange light. Industrial concrete pillars and brick walls frame both sides. Metal railings visible. Golden-orange sky in background. Blue shadows cast across street creating high-contrast cinematic look. Volumetric dust particles floating in air catching light.\n\nCAMERA ANGLES & CINEMATOGRAPHY:\n\n0-1s: WIDE ESTABLISHING SHOT, camera positioned at street level, tilted slightly Dutch angle to show desperation. Slow dolly-in toward the fighter\n1-3s: LOW-ANGLE CLOSE-UP on beaten fighter's face as goons surround him. Camera slightly beneath eye-level looking upward at aggressive goons (makes them look menacing, him vulnerable)\n3-5s: OVERHEAD SHOT as golden-blue energy beam descends. Camera positioned directly above, looking down at the beam engulfing him\n5-7s: FAST 360-DEGREE ORBITAL CAMERA ROTATION around fighter during energy explosion. Energy aura visible radiating outward. Multiple angles showing transformation from all sides\n7-8s: EXTREME CLOSE-UP on fighter's face as eyes glow. Camera inches from face, eyes filling 60% of frame. Electricity crackling visible around eyes\n8-9s: LOW-ANGLE UPWARD SHOT as fighter stands fully transformed. Camera positioned beneath him looking up he now TOWERS over everyone. Shows dominance shift\n9-12s: DYNAMIC WHIP-PAN following first goon's charge. Camera pans rapidly to track movement, then HARD CUT to:\n12-14s: SLOW-MOTION SIDE PROFILE SHOT of devastating punch. Camera positioned at exact side angle to catch punch trajectory. Slow-motion impact  shockwave visible in air, goon's body reacting to force\n14-15s: TRACKING SHOT following goon flying backward. Camera moves with the goon's body trajectory, then IMPACT SHOT from goon's POV as he crashes into wall (camera shakes violently on impact)\n\nSTORY BEATS:\n\nA beaten-down fighter in torn shirt, bruises covering his face, blood dripping, is VICIOUSLY surrounded by three aggressive goons on a wet industrial street at golden hour. He's backed against a brick wall, barely defending himself as punches rain down. He's on the verge of collapse.\n\nSUDDENLY  a massive golden-blue energy beam EXPLODES down from the sky with CATACLYSMIC force. The beam engulfs him in blinding light. Massive ENERGY EXPLOSION  the entire street SHAKES, dust clouds ERUPT, volumetric fog fills the air.\n\nWhen the light clears, he stands COMPLETELY TRANSFORMED — eyes BLAZING with godly golden-blue power, energy aura RAGING violently around his entire body with visible electricity tendrils crackling, muscles VISIBLY ENHANCED and defined, face twisted in COLD RAGE and predatory intensity.\n\nHe breathes heavily, stares at the goons with KILLING INTENT. The goons freeze in ABSOLUTE TERROR.\n\nFirst goon CHARGES desperately forward. The fighter EXPLODES into motion with IMPOSSIBLE SUPERHUMAN VELOCITY. DEVASTATING PUNCH  fist CRACKS through air with SONIC intensity. SLOW-MOTION IMPACT  shockwave visible, goon's face CONTORTS in agony, blood SPRAYS, body LAUNCHES like ragdoll 20 feet backward, CRASHES through brick wall with BONE-CRUSHING force, debris ERUPTS everywhere.\n\nPART 2 -\n\nLOCATION DETAIL (Escalated): Same urban street, now a DEMOLITION ZONE. Concrete cracked and breaking. Brick wall partially destroyed from first impact. Metal railings BENT and snapping. Wooden crates and boxes SPLINTERED. Pillars showing cracks spreading. Golden hour light now more dramatic with darker blue shadows. Heavy dust and debris particles in air. Total DEVASTATION visible.\n\nCAMERA ANGLES & CINEMATOGRAPHY:\n\n15-17s: WIDE SHOT showing street arena. Camera pulls back to show scale of destruction and two remaining enemies. High-angle shot showing ground cracks\n17-19s: RAPID MULTI-ANGLE FIGHT EXCHANGE alternating extreme close-ups and medium shots. Fast cutting every 0.3-0.5 seconds matching fight tempo:\nClose-up on second goon's face as he launches kicks (intensity capture)\nLow-angle shot of fighter's parries\nHigh-angle overhead shot of exchange\nSide profile close-up on impact moments\n19-21s: MID-AIR GRAB SHOT from side angle as fighter lifts goon. Camera positioned at exact moment of lift showing superhuman strength\n21-22s: VIOLENT SLAM SHOT  camera positioned ground-level looking upward as goon is SLAMMED onto concrete. IMPACT slow-motion, cracks SPREAD like spiderwebs across ground, dust ERUPTS upward\n22-24s: THROW SEQUENCE  camera tracks throw trajectory. Goon FLIES through wooden crates, structures BEND and SHATTER, camera follows impact (tracking shot from side)\n24-26s: INTENSE FINAL FIGHT EXCHANGE with scarred leader. Rapid multi-angle cuts showing:\nLow-angle shot of fighter's punch\nHigh-angle overhead of counter-attack\nClose-up on scarred goon's face showing desperation\nMetal railings BENDING in background from force\n26-27s: EXTREME CLOSE-UP on fighter's fist as it CHARGES energy. Golden-blue glow INTENSIFIES, electricity CRACKLING violently around fist. Camera macro-level, glow filling entire frame\n27-28s: SLOW-MOTION FINAL PUNCH setup. Camera positioned LOW-ANGLE BEHIND punch trajectory. Fist PULLS BACK with godly force visible (air distorting around fist)\n28-29s: CATACLYSMIC IMPACT SLOW-MOTION. Camera positioned to capture scarred goon's face as punch connects. Eyes go WIDE in SHOCK, body CONTORTS from impact force, entire body LAUNCHED backward with EARTH-SHATTERING velocity\n29-30s: PILLAR DESTRUCTION SHOT camera positioned wide to show goon CRASHES through concrete pillar like PAPER, pillar EXPLODES into massive chunks, structure COLLAPSES, debris MUSHROOMS into air, dust cloud engulfs entire area. Final frame: TIGHT CLOSE-UP on fighter's face from straight-on angle. Eyes still BLAZING with power. Breathing HEAVILY, chest HEAVING. Energy aura SLOWLY FADES. Cold MERCILESS confidence in stare. Fade to black.\n\nSTORY BEATS:\n\nSecond goon attempts RAPID KICK COMBO  fighter PARRIES with BLINDING speed. Each block CRASHES with SONIC booms. GRAB  he SLAMS goon down VIOLENTLY onto concrete, CRACKS spread like spiderwebs. THROW goon FLIES through wooden boxes, metal structures BEND and SHATTER from impact force.\n\nThird goon  scarred leader, last hope  steps forward DESPERATELY. This is BRUTAL WARFARE. EXPLOSIVE hand-to-hand combat: PUNCH blocked, COUNTERATTACK devastating, SPINNING KICK with BONE-BREAKING force, GRAPPLE with SUPERHUMAN pressure. Ground SHATTERS beneath them. Metal railings BEND and SNAP. Every collision sends SHOCKWAVES across street.\n\nFINAL MOMENT  fighter CHARGES energy into fist. Fist GLOWS BLINDING GOLD-BLUE. Pulls back with GODLY FORCE. Unleashes CATACLYSMIC POWERED PUNCH.\n\nSLOW-MOTION IMPACT  scarred goon's body CONTORTS, eyes WIDE in SHOCK, LAUNCHED backward with EARTH-SHATTERING velocity, CRASHES THROUGH concrete pillar like PAPER, pillar EXPLODES into massive chunks, entire structure COLLAPSES, dust MUSHROOMS into air.\n\nFighter stands alone in TOTAL DEVASTATION. Breathing HEAVILY, chest HEAVING, eyes still BLAZING. Energy aura SLOWLY FADES. Stares DIRECTLY at camera with COLD MERCILESS confidence.",
    "promptSourceUrl": "https://x.com/itsshara_ai/status/2082892936963031405",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/itsshara_ai/status/2082892924694712509",
    "sourceLabel": "X 原帖 · @itsshara_ai",
    "author": "Shara I Ai Video Creator",
    "publishedAt": "2026-07-30T18:15:46.183Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082892924694712509.mp4",
    "posterUrl": "/posters/x/2082892924694712509.jpg",
    "duration": 28,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "No stunt team. No rig. No second take.\n\nMiniMax H3.\n\nPrompt below:\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082873433478582726",
    "title": "皮克斯风小鼠冒险动画",
    "titleEn": "Pixar-style mouse adventure 3D animation",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《皮克斯风小鼠冒险动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Pixar-style mouse adventure 3D animation. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "{       \"clip_id\": \"animated_adventure\",       \"genre\": \"3D Animation / Pixar Style\",       \"total_duration\": \"15s\",       \"aspect_ratio\": \"16:9\",       \"style_keywords\": \"Pixar-style 3D animation, vibrant colors, expressive characters, cinematic lighting, whimsical, high detail textures\",       \"shots\": [         {           \"shot_num\": 1,           \"duration\": \"4s\",           \"prompt\": \"Wide shot of a tiny adventurous mouse wearing a leather aviator cap standing on the edge of a giant mushroom in an enchanted forest, giant flowers towering overhead, Pixar-style 3D animation, vibrant colors, soft cinematic lighting\",           \"camera\": \"low angle push in\",           \"transition\": \"hard cut\"         },         {           \"shot_num\": 2,           \"duration\": \"4s\",           \"prompt\": \"Medium shot of the mouse launching into the air on a dandelion seed parachute, floating through sunbeams, pollen particles sparkling around, joyful expression, 3D animated film style, warm golden light\",           \"camera\": \"tracking shot from below\",           \"transition\": \"match cut on motion\"         },         {           \"shot_num\": 3,           \"duration\": \"4s\",           \"prompt\": \"Dynamic action shot of the mouse swooping through a hollow log, fireflies lighting the way, exaggerated squash-and-stretch animation, motion blur on background, Pixar-style adventure sequence\",           \"camera\": \"POV following through tunnel\",           \"transition\": \"whip pan\"         },         {           \"shot_num\": 4,           \"duration\": \"3s\",           \"prompt\": \"Wide shot of the mouse landing triumphantly on a lily pad in a moonlit pond, ripples spreading outward, bioluminescent plants glowing, 3D animation, magical atmosphere, satisfied smile\",           \"camera\": \"crane shot rising up\",           \"transition\": \"end\"         }       ]     },",
    "promptSourceUrl": "https://x.com/sebatheepan/status/2082873433478582726",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/sebatheepan/status/2082873433478582726",
    "sourceLabel": "X 原帖 · @sebatheepan",
    "author": "Pan",
    "publishedAt": "2026-07-30T16:58:19.115Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082873433478582726.mp4",
    "posterUrl": "/posters/x/2082873433478582726.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuio MiniMax H3 - 3D animation \n\nPrompt ⤵️\n\n{       \"clip_id\": \"animated_adventure\",       \"genre\": \"3D Animation / Pixar Style\",       \"total_duration\": \"15s\",       \"aspect_ratio\": \"16:9\",       \"style_keywords\": \"Pixar-style 3D animation, vibrant colors, expressive characters, cinematic lighting, whimsical, high detail textures\",       \"shots\": [         {           \"shot_num\": 1,           \"duration\": \"4s\",           \"prompt\": \"Wide shot of a tiny adventurous mouse wearing a leather aviator cap standing on the edge of a giant mushroom in an enchanted forest, giant flowers towering overhead, Pixar-style 3D animation, vibrant colors, soft cinematic lighting\",           \"camera\": \"low angle push in\",           \"transition\": \"hard cut\"         },         {           \"shot_num\": 2,           \"duration\": \"4s\",           \"prompt\": \"Medium shot of the mouse launching into the air on a dandelion seed parachute, floating through sunbeams, pollen particles sparkling around, joyful expression, 3D animated film style, warm golden light\",           \"camera\": \"tracking shot from below\",           \"transition\": \"match cut on motion\"         },         {           \"shot_num\": 3,           \"duration\": \"4s\",           \"prompt\": \"Dynamic action shot of the mouse swooping through a hollow log, fireflies lighting the way, exaggerated squash-and-stretch animation, motion blur on background, Pixar-style adventure sequence\",           \"camera\": \"POV following through tunnel\",           \"transition\": \"whip pan\"         },         {           \"shot_num\": 4,           \"duration\": \"3s\",           \"prompt\": \"Wide shot of the mouse landing triumphantly on a lily pad in a moonlit pond, ripples spreading outward, bioluminescent plants glowing, 3D animation, magical atmosphere, satisfied smile\",           \"camera\": \"crane shot rising up\",           \"transition\": \"end\"         }       ]     },\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082857595824451623",
    "title": "格子纹跑车夜到黎明广告",
    "titleEn": "Luxury checkered sports-coupe night-to-dawn commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "Unknown",
    "summary": "《格子纹跑车夜到黎明广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Luxury checkered sports-coupe night-to-dawn commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "High-end luxury auto commercial, cinematic night-to-dawn lighting, wet reflective surfaces, anamorphic lens flares, rich emerald green and deep black color grade, heavy film grain.\nUse  @Audio1 for background music\n\nUse the provided 4x4 grid as the only visual reference @Image1 . You must show ALL 16 panels from the grid in sequence. Lock the exact same character (blonde swept-back hair, stern heavy-browed face, green-and-black checkered haori) and the exact same green-and-black checkered sports coupe in every shot. Do not invent new shots or change anything.\n\n[TIMELINE - ALL 16 PANELS]\n0.0-0.9s: Panel 1 – Extreme wide rain-soaked city road with distant headlights and skyline\n0.9-1.8s: Panel 2 – Character standing arms crossed under streetlight\n\n1.8-2.7s: Panel 3 – Extreme close-up of character’s angry scowling face\n2.7-3.6s: Panel 4 – Character standing next to the checkered car under city lights\n3.6-4.5s: Panel 5 – Character walking toward the car from behind in the rain\n4.5-5.4s: Panel 6 – Close-up of hand gripping the door handle\n5.4-6.3s: Panel 7 – Character sitting in the driver’s seat, intense expression\n6.3-7.2s: Panel 8 – Dashboard and steering wheel close-up with green lights\n7.2-8.1s: Panel 9 – Low rear three-quarter shot of the car accelerating in the rain\n8.1-9.0s: Panel 10 – Side tracking shot of character driving the checkered car\n9.0-9.9s: Panel 11 – Car driving on mountain road at sunset\n9.9-10.8s: Panel 12 – Aerial/coastal highway shot of the car at sunset\n10.8-11.7s: Panel 13 – Character standing next to the car on rooftop with city skyline\n11.7-12.6s: Panel 14 – Character leaning on the car, arms crossed, hero shot\n12.6-13.5s: Panel 15 – Extreme close-up of the car’s front headlights and grille\n13.5-15.0s: Panel 16 – Character and car silhouetted against the rising sun\ncoherent character and vehicle identity across every cut 📷Image1 , cinematic color grade, heavy film grain, no artifacts, movie-level continuity, fast-paced 15-second commercial.",
    "promptSourceUrl": "https://x.com/shirawiggles/status/2082857598802415992",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/shirawiggles/status/2082857595824451623",
    "sourceLabel": "X 原帖 · @shirawiggles",
    "author": "Shira Wiggles",
    "publishedAt": "2026-07-30T15:55:23.124Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082857595824451623.mp4",
    "posterUrl": "/posters/x/2082857595824451623.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I tested my new “Grid + Video Master Prompt” with MiniMax H3 in early access, and I also added the background music I wanted as a reference.\n\nHere you have the results👇\n\nCheck the prompt + screenshots + link for the master prompt in the replies 💋\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082854898907931081",
    "title": "暗黑流行三人组 MV 与标题",
    "titleEn": "Dark-pop trio music-film performance with on-screen titles",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《暗黑流行三人组 MV 与标题》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×808。",
    "summaryEn": "An H3 output for Dark-pop trio music-film performance with on-screen titles. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×808.",
    "prompt": "15 seconds, 16:9 dark-pop music video performance. Use @Image 1 as the strict identity reference for the three women — faces, hair and wardrobe unchanged in every shot. Use @Image 2 as the reference for the two titles only: ultra-condensed heavy caps, bone white, photocopied print surface.  The three women perform and sing to camera throughout. Follow the beats below shot by shot with continuous natural camera movement and real physical motion inside every frame — this is live performance footage, never a slideshow, never a sequence of held poses.  VOCAL: a single female lead voice, gritty and low, close-mic'd with tape saturation. The lyric lines are: \"Dead air on the line\" — sung at 2.7 seconds \"Run it till it burns\" — sung at 6.0 seconds \"I don't slow down\" — sung at 10.0 seconds Lip-sync locks precisely to these lines on whichever woman is singing. Between the lines all three mouth along, breathe hard and move.  [0.0–2.7s] The three stand in near-blackness, sand-beige studio crushed dark. Handheld camera drifting slowly forward, breathing. The platinum-crop woman rolls her shoulders and steps toward lens. Her hair lifts. The auburn woman in mirrored shades tilts her head up. Real movement, low energy, building. Title over the darkness: \"DEAD AIR\".  [2.7–6.0s] The auburn woman in the mirrored shades sings the first line straight into the lens, medium-close, mouth clearly forming the words. Camera whip-pans off her to the platinum-crop woman mid-stride, then crash-zooms back. Hard cuts every 0.3 seconds. Bodies in constant motion — hair whipping, chains swinging, hips driving, hands reaching for camera. Single-frame white flashes between cuts.  [6.0–10.0s] All three together, performing hard. The chestnut-haired woman takes the second line, singing directly to lens while the other two move behind her, bodies loose and driving. Camera orbits the trio fast, then whip-pans, then handheld snap-zooms. Cuts every 0.2 seconds. Hair, fur and fishnet all moving. Faces alive, mouths working, sweat catching the light.  [10.0–12.7s] Slow down hard, two long shots. The chestnut-haired woman alone, medium-close, singing the third line slowly and holding eye contact with the lens all the way through. Camera almost still, only breathing. Her wet hair moves across her face. She finishes the line and her mouth stays slightly open.  [12.7–15.0s] Everything detonates. Flash cutting, single-frame plates, all three thrashing and singing at once, faces close to lens, black frames and inverted negatives between cuts. Camera whipping violently. Final title hammers in: \"AFTERBURN\". Cut hard to black with the music.  MUSIC: driving industrial dark-pop, distorted 808 kick landing at 2.7 seconds, full arrangement through 6.0 to 10.0 seconds with clipped brass stabs and heavy tape saturation, dropping to near silence and one breath at 12.0 seconds for a full second, then the final two seconds at full weight with one metallic crash on \"AFTERBURN\". The lead vocal sits forward in the mix. Do not imitate an existing melody.  Warm sand-beige studio sweep, hard raking light from camera-left, Kodak Portra 400 grain, tape dropout, scanline tear, chromatic fringing, light leaks. Hard cuts only, full speed throughout, every shot carrying real motion. Titles spelled exactly as written, English only.",
    "promptSourceUrl": "https://x.com/ivanka_humeniuk/status/2082854905476251967",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ivanka_humeniuk/status/2082854898907931081",
    "sourceLabel": "X 原帖 · @ivanka_humeniuk",
    "author": "Ivanna | AI Art & Prompts",
    "publishedAt": "2026-07-30T15:44:40.129Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082854898907931081.mp4",
    "posterUrl": "/posters/x/2082854898907931081.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×808",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Three girls. One studio. Fifteen seconds.\n\nEverything you see and hear was generated in a single pass — the performance, the lip-sync, the cuts, the typography, the track. No edit, no sound design, no post.\n\nMade with MiniMax H3 on @Hailuo_AI  🎬\n\nPrompts in the thread 🧵\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082853567543615837",
    "title": "世界上最倒霉的超级英雄",
    "titleEn": "The World's Unluckiest Superhero",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《世界上最倒霉的超级英雄》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 30 秒 · 1280×720。",
    "summaryEn": "An H3 output for The World's Unluckiest Superhero. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 30s · 1280×720.",
    "prompt": "A documentary about a superhero who has extremely bad luck and ends up saving people by accident through the destruction caused by his own misfortune. Dialogue in English. Scene direction with unique composition. Every cut, every camera angle, and every movement is of exceptionally high quality; the composition is guided by an experienced film director. The comedy is genuinely interesting, and throughout the 15 seconds everything unfolds in a perfectly crafted way, with a comedic payoff that can make anyone laugh.",
    "promptSourceUrl": "https://x.com/NACHOS2D_/status/2082853571222028586",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/NACHOS2D_/status/2082853567543615837",
    "sourceLabel": "X 原帖 · @NACHOS2D_",
    "author": "nachos2d",
    "publishedAt": "2026-07-30T15:39:22.707Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082853567543615837.mp4",
    "posterUrl": "/posters/x/2082853567543615837.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Comedy",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The most impressive thing about Hailuo 3.0 might be the price.\n\nSeeing a model generate videos in 2K with this level of quality is genuinely impressive. @Hailuo_AI really cooked with this release.#MiniMaxH3\n\nI created a comedy scene about an incredibly unlucky superhero who ends up capturing villains completely by accident 😂\n\nPrompt below.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082850740603867462",
    "title": "黄昏厨房里的发光生物",
    "titleEn": "Luminous Creature in a Dusk Kitchen",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《黄昏厨房里的发光生物》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Luminous Creature in a Dusk Kitchen. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "15 seconds, 16:9 landscape. Blend live-action footage of a small kitchen at dusk with hand-drawn luminous animation. The last sunset light lingers at the window. The lived-in kitchen contains an old wooden table, a half-washed mug, a lightly fogged glass bottle, and a hanging dish towel.\n- Shoot as if someone is filming one-handed on a phone: subtle hand tremor, hesitant close-focus pulls, backlit exposure breathing, and slightly coarse noise in the shadows. It should feel like an astonishing event captured in a rush at home, not a carefully dressed commercial.\n- Do not show giant eyes, split mouths, fangs, threatening behavior, lunges, sudden black frames, or jump scares. Use only room tone, cloth friction, a soft mug clink, faucet drips, the camera operator’s footsteps and quiet breathing, plus gentle electronic tones and tiny vocalizations from the drawn creatures.",
    "promptSourceUrl": "https://x.com/underwoodxie96/status/2082850740603867462",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/underwoodxie96/status/2082850740603867462",
    "sourceLabel": "X 原帖 · @underwoodxie96",
    "author": "underwood",
    "publishedAt": "2026-07-30T15:28:08.712Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082850740603867462.mp4",
    "posterUrl": "/posters/x/2082850740603867462.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Horror",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "commercial"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I think I just caught tiny creatures living in my kitchen.\nNo jump scares, no threatening monsters, and none of that overly polished commercial look—just dusk, a handheld camera, and luminous hand-drawn creatures quietly appearing in a real space.\n\nGenerated with @Hailuo_AI’s #MiniMaxH3. The blend of live action and animation turned out surprisingly natural.\n\nprompt：\n- 15 seconds, 16:9 landscape. Blend live-action footage of a small kitchen at dusk with hand-drawn luminous animation. The last sunset light lingers at the window. The lived-in kitchen contains an old wooden table, a half-washed mug, a lightly fogged glass bottle, and a hanging dish towel.\n- Shoot as if someone is filming one-handed on a phone: subtle hand tremor, hesitant close-focus pulls, backlit exposure breathing, and slightly coarse noise in the shadows. It should feel like an astonishing event captured in a rush at home, not a carefully dressed commercial.\n- Do not show giant eyes, split mouths, fangs, threatening behavior, lunges, sudden black frames, or jump scares. Use only room tone, cloth friction, a soft mug clink, faucet drips, the camera operator’s footsteps and quiet breathing, plus gentle electronic tones and tiny vocalizations from the drawn creatures.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082847686185672833",
    "title": "九宫格女说唱歌手同屏表演",
    "titleEn": "3x3 contact sheet female rapper grid performance",
    "model": "MiniMax H3（来源标注）",
    "mode": "FL2VA",
    "summary": "《九宫格女说唱歌手同屏表演》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 12 秒 · 1280×720。",
    "summaryEn": "An H3 output for 3x3 contact sheet female rapper grid performance. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 12s · 1280×720.",
    "prompt": "Use @Image 1 as the exact opening frame. It is a 3x3 contact sheet of nine separate portraits of the same female rapper, each in a different location. Treat every panel as its own live video cell with its own camera, lighting and performance. 12 seconds, 16:9.  Keep the grid absolutely fixed for the entire shot: identical panel positions, identical panel sizes, identical white gutters, no zoom, no pan, no rotation, no drift of the grid itself. Only the content inside each panel may move. Preserve each panel's identity exactly: same face, same hair, same wardrobe, same jewellery, same background as in @Image 1. Panel 1 stage with mic, panel 2 graffiti alley, panel 3 studio console with bandana, panel 4 rooftop with pink car, panel 5 roller shutter with bucket hat and chains, panel 6 purple club, panel 7 sunset rooftop with braids, panel 8 neon alley, panel 9 black backdrop close-up.  [0.0-1.2s] All nine panels sit almost still like breathing photographs, only micro-movement: hair drift, blinking, smoke, light flicker. A low sub-bass swell rises. [1.2-3.0s] The beat drops. Panel 1 raps to camera: \"MiniMax.\" Panel 5 answers: \"H3.\" The other seven nod on the beat. [3.0-5.0s] Panels 2, 3 and 7 trade the line one after another: \"New drop.\" Each one takes the beat, the rest bounce in rhythm. [5.0-7.5s] Panels 4, 6, 8 and 9 answer: \"Go and try it.\" Hands move, chains swing, ad-libs. [7.5-10.0s] All nine perform in unison, tightly lip-synced toH3, go and try it.\" [10.0-11.3s] The beat stops dead. In one synchronized move all nine raise a hand into a finger-gun aimed straight at the camera, chin down, hard contact. [11.3-12.0s] They fire the imaginary shot on a single low hit, a tiny recoil ripples through all nine panels at once, then freeze.                Audio: one continuous hip-hop beat across all panels, roughly 92 BPM, deep 808 sub, crisp trap hats, sparse hand claps. Female voice, confident, slightly playful, not aggressive. Layer nine slightly detunedn section for a choir effect. End on one low hit plus a shorttail of room tone.  Keep the lip-sync accurate to the words in every speaking panel. No subtitles, no on-screen text, no captions, no logos, no watermarks. No panel may drift, crossfade, resize or swap places. Do not add people. Dreen single shot at any point.",
    "promptSourceUrl": "https://x.com/lukasersil/status/2082847686185672833",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/lukasersil/status/2082847686185672833",
    "sourceLabel": "X 原帖 · @lukasersil",
    "author": "Lukáš Eršil",
    "publishedAt": "2026-07-30T15:16:00.482Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082847686185672833.mp4",
    "posterUrl": "/posters/x/2082847686185672833.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Prompt pro #MinimaxH3:\n\nUse @Image 1 as the exact opening frame. It is a 3x3 contact sheet of nine separate portraits of the same female rapper, each in a different location. Treat every panel as its own live video cell with its own camera, lighting and performance. 12 seconds, 16:9.  Keep the grid absolutely fixed for the entire shot: identical panel positions, identical panel sizes, identical white gutters, no zoom, no pan, no rotation, no drift of the grid itself. Only the content inside each panel may move. Preserve each panel's identity exactly: same face, same hair, same wardrobe, same jewellery, same background as in @Image 1. Panel 1 stage with mic, panel 2 graffiti alley, panel 3 studio console with bandana, panel 4 rooftop with pink car, panel 5 roller shutter with bucket hat and chains, panel 6 purple club, panel 7 sunset rooftop with braids, panel 8 neon alley, panel 9 black backdrop close-up.  [0.0-1.2s] All nine panels sit almost still like breathing photographs, only micro-movement: hair drift, blinking, smoke, light flicker. A low sub-bass swell rises. [1.2-3.0s] The beat drops. Panel 1 raps to camera: \"MiniMax.\" Panel 5 answers: \"H3.\" The other seven nod on the beat. [3.0-5.0s] Panels 2, 3 and 7 trade the line one after another: \"New drop.\" Each one takes the beat, the rest bounce in rhythm. [5.0-7.5s] Panels 4, 6, 8 and 9 answer: \"Go and try it.\" Hands move, chains swing, ad-libs. [7.5-10.0s] All nine perform in unison, tightly lip-synced toH3, go and try it.\" [10.0-11.3s] The beat stops dead. In one synchronized move all nine raise a hand into a finger-gun aimed straight at the camera, chin down, hard contact. [11.3-12.0s] They fire the imaginary shot on a single low hit, a tiny recoil ripples through all nine panels at once, then freeze.                Audio: one continuous hip-hop beat across all panels, roughly 92 BPM, deep 808 sub, crisp trap hats, sparse hand claps. Female voice, confident, slightly playful, not aggressive. Layer nine slightly detunedn section for a choir effect. End on one low hit plus a shorttail of room tone.  Keep the lip-sync accurate to the words in every speaking panel. No subtitles, no on-screen text, no captions, no logos, no watermarks. No panel may drift, crossfade, resize or swap places. Do not add people. Dreen single shot at any point.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082845923152150697",
    "title": "大理石楼梯唇彩广告",
    "titleEn": "Marble Staircase Lip Gloss Commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《大理石楼梯唇彩广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Marble Staircase Lip Gloss Commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use the uploaded reference image as the exact character reference. Preserve her facial identity, hairstyle, eye color, makeup, skin tone, body proportions, pearl necklace, rings, and facial consistency throughout every shot.\n\nCreate a 15-second ultra-realistic luxury lip gloss commercial. The video opens with a close-up of the woman looking directly into the camera while confidently holding the lip gloss beside her face. She then gracefully walks down an elegant marble staircase as the camera smoothly tracks backward. Cut to her sitting on a luxurious blush-pink velvet sofa, smiling softly while placing the lip gloss on a marble coffee table. Cinematic macro shots showcase the applicator being removed, the glossy texture, and the premium packaging. She applies the lip gloss while looking into a vanity mirror, followed by extreme close-ups of her glossy lips catching soft studio light. Finish with a confident beauty pose holding the lip gloss beside her face, then transition to a premium floating hero shot of the product rotating above a glossy marble surface with warm rose-gold lighting.\n\nUltra-realistic beauty cinematography, luxury cosmetic campaign, smooth gimbal movement, realistic skin textures, macro product photography, shallow depth of field, creamy bokeh, premium color grading, 4K HDR, 16:9, no text, subtitles, logos, or watermarks.",
    "promptSourceUrl": "https://x.com/AIwithSynthia/status/2082845923152150697",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AIwithSynthia/status/2082845923152150697",
    "sourceLabel": "X 原帖 · @AIwithSynthia",
    "author": "Synthia",
    "publishedAt": "2026-07-30T15:09:00.142Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082845923152150697.mp4",
    "posterUrl": "/posters/x/2082845923152150697.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tested MiniMax H3 on @Hailuo_AI with Omni Reference to create a luxury lip gloss campaign, combining elegant beauty cinematography, macro product shots, and premium cosmetic visuals.\n\nThe consistent character, realistic lighting, and commercial-quality production make this feel remarkably close to a professional beauty advertisement.\n\n#MiniMaxH3 \n\nPrompt 👇\n\nUse the uploaded reference image as the exact character reference. Preserve her facial identity, hairstyle, eye color, makeup, skin tone, body proportions, pearl necklace, rings, and facial consistency throughout every shot.\n\nCreate a 15-second ultra-realistic luxury lip gloss commercial. The video opens with a close-up of the woman looking directly into the camera while confidently holding the lip gloss beside her face. She then gracefully walks down an elegant marble staircase as the camera smoothly tracks backward. Cut to her sitting on a luxurious blush-pink velvet sofa, smiling softly while placing the lip gloss on a marble coffee table. Cinematic macro shots showcase the applicator being removed, the glossy texture, and the premium packaging. She applies the lip gloss while looking into a vanity mirror, followed by extreme close-ups of her glossy lips catching soft studio light. Finish with a confident beauty pose holding the lip gloss beside her face, then transition to a premium floating hero shot of the product rotating above a glossy marble surface with warm rose-gold lighting.\n\nUltra-realistic beauty cinematography, luxury cosmetic campaign, smooth gimbal movement, realistic skin textures, macro product photography, shallow depth of field, creamy bokeh, premium color grading, 4K HDR, 16:9, no text, subtitles, logos, or watermarks.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082840410268057697",
    "title": "厨房巨蛛喜剧短片",
    "titleEn": "Giant Kitchen Spider Comedy Short",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《厨房巨蛛喜剧短片》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1698×720。",
    "summaryEn": "An H3 output for Giant Kitchen Spider Comedy Short. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1698×720.",
    "prompt": "3D Pixar-style animated comedy short film, ultra-premium feature film quality, highly expressive facial animation, cinematic storytelling, cozy modern kitchen, warm golden lighting, realistic global illumination, shallow depth of field, ultra-detailed textures, realistic hair simulation, emotional acting, cinematic camera movement, smooth multi-shot sequence, fast-paced visual comedy, no text, no subtitles, no watermark.\n\nA bright, cozy modern kitchen during the afternoon. Warm sunlight streams through the windows. A young woman with long messy dark hair, oversized expressive brown eyes, freckles, wearing an oversized black T-shirt and house slippers, walks into the kitchen carrying a mug.\n\nShe suddenly freezes.\n\nOn the white marble kitchen island sits an enormous cartoon-style spider about the size of a football. It has fuzzy black legs, oversized expressive eyes, and an intimidating stare.\n\nShot 1: Wide cinematic shot. The woman and giant spider lock eyes from opposite sides of the kitchen island. Everything becomes still.\n\nShot 2: Extreme close-up of the woman's face. Her pupils shrink. Sweat forms on her forehead. Her breathing becomes rapid. Tiny facial micro-expressions show pure panic.\n\nShot 3: Extreme close-up of the spider. It slowly raises its front legs, narrows its eyes, and curls its fangs into a confident grin as if challenging her.\n\nShot 4: Medium shot. The woman slowly reaches toward a nearby frying pan without taking her eyes off the spider. Her hand trembles violently.\n\nShot 5: Close-up of her gripping the frying pan with both hands. Her knuckles turn white from fear.\n\nShot 6: Wide shot. She nervously points the frying pan toward the spider like a shield while cautiously stepping backward.\n\nThe spider suddenly takes one slow step forward.\n\nThe woman gasps.\n\nAnother step.\n\nShe screams.\n\nShot 7: Low-angle shot from behind the spider, making it appear huge while the frightened woman backs away across the kitchen.\n\nThe spider tilts its head and smirks mischievously.\n\nThe woman gathers all her courage, raises the frying pan high above her head, and charges toward the spider with a terrified battle cry.\n\nJust before she swings—\n\nThe spider suddenly spreads its legs, launches into the air, and leaps directly toward the camera.\n\nThe woman instantly drops into a crouch, covers her head with both hands, and screams in terror.\n\nFreeze on the spider flying through the air toward the viewer.\n\nEnd of first 15 seconds.\n\nUltra-detailed Pixar-quality animation, expressive facial acting, dynamic cinematic camera work, alternating wide shots and extreme close-ups every 1–2 seconds, emotional comedy timing, smooth character animation, premium lighting, blockbuster animated short-film quality, perfect character consistency, viral social-media storytelling.\n\nThis ending creates a strong cliffhanger for the extension, where an unexpected hero can arrive and flip the situation into comedy.",
    "promptSourceUrl": "https://x.com/Ciri_ai/status/2082840410268057697",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Ciri_ai/status/2082840410268057697",
    "sourceLabel": "X 原帖 · @Ciri_ai",
    "author": "Ciri",
    "publishedAt": "2026-07-30T14:47:05.768Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082840410268057697.mp4",
    "posterUrl": "/posters/x/2082840410268057697.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1698×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "food",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This doesn’t feel like just another AI tool anymore it feels like a full creative engine.\n\nI’ve been really impressed by how MiniMax H3 brings visuals, motion, and audio together in one seamless workflow.\n\nBuilt by @Hailuo_AI, it understands text, images, video, and audio simultaneously, making it incredibly easy to create cinematic, Pixar style content through a single creative pipeline.\n\n#MiniMaxH3 \n\nPrompt: Prompt: 3D Pixar-style animated comedy short film, ultra-premium feature film quality, highly expressive facial animation, cinematic storytelling, cozy modern kitchen, warm golden lighting, realistic global illumination, shallow depth of field, ultra-detailed textures, realistic hair simulation, emotional acting, cinematic camera movement, smooth multi-shot sequence, fast-paced visual comedy, no text, no subtitles, no watermark.\n\nA bright, cozy modern kitchen during the afternoon. Warm sunlight streams through the windows. A young woman with long messy dark hair, oversized expressive brown eyes, freckles, wearing an oversized black T-shirt and house slippers, walks into the kitchen carrying a mug.\n\nShe suddenly freezes.\n\nOn the white marble kitchen island sits an enormous cartoon-style spider about the size of a football. It has fuzzy black legs, oversized expressive eyes, and an intimidating stare.\n\nShot 1: Wide cinematic shot. The woman and giant spider lock eyes from opposite sides of the kitchen island. Everything becomes still.\n\nShot 2: Extreme close-up of the woman's face. Her pupils shrink. Sweat forms on her forehead. Her breathing becomes rapid. Tiny facial micro-expressions show pure panic.\n\nShot 3: Extreme close-up of the spider. It slowly raises its front legs, narrows its eyes, and curls its fangs into a confident grin as if challenging her.\n\nShot 4: Medium shot. The woman slowly reaches toward a nearby frying pan without taking her eyes off the spider. Her hand trembles violently.\n\nShot 5: Close-up of her gripping the frying pan with both hands. Her knuckles turn white from fear.\n\nShot 6: Wide shot. She nervously points the frying pan toward the spider like a shield while cautiously stepping backward.\n\nThe spider suddenly takes one slow step forward.\n\nThe woman gasps.\n\nAnother step.\n\nShe screams.\n\nShot 7: Low-angle shot from behind the spider, making it appear huge while the frightened woman backs away across the kitchen.\n\nThe spider tilts its head and smirks mischievously.\n\nThe woman gathers all her courage, raises the frying pan high above her head, and charges toward the spider with a terrified battle cry.\n\nJust before she swings—\n\nThe spider suddenly spreads its legs, launches into the air, and leaps directly toward the camera.\n\nThe woman instantly drops into a crouch, covers her head with both hands, and screams in terror.\n\nFreeze on the spider flying through the air toward the viewer.\n\nEnd of first 15 seconds.\n\nUltra-detailed Pixar-quality animation, expressive facial acting, dynamic cinematic camera work, alternating wide shots and extreme close-ups every 1–2 seconds, emotional comedy timing, smooth character animation, premium lighting, blockbuster animated short-film quality, perfect character consistency, viral social-media storytelling.\n\nThis ending creates a strong cliffhanger for the extension, where an unexpected hero can arrive and flip the situation into comedy.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082837405812613617",
    "title": "寿司之夜手机 UGC Vlog",
    "titleEn": "Sushi Night Smartphone UGC Vlog",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《寿司之夜手机 UGC Vlog》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1698×720。",
    "summaryEn": "An H3 output for Sushi Night Smartphone UGC Vlog. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1698×720.",
    "prompt": "Duration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, slight motion blur, TikTok/Reels energy — NOT cinematic, NOT overly polished. Feels like a real creator filmed this on their phone.\nProduct Reference: Use the uploaded assorted sushi platter image as the only product reference. Keep the sushi arrangement, fish cuts, rice texture, garnish placement, colors, and proportions identical in every shot—do not redesign or alter the food.\nCamera: iPhone 15 Pro front + back camera switching, handheld with natural wobble, slight autofocus breathing, vertical-style framing cropped to 16:9, occasional finger near the edge of the lens, realistic phone HDR, warm apartment lighting, subtle reflections on the takeaway container.\nCharacter Description\nName (for reference): Aoi\nA young Japanese woman in her early 20s with a naturally beautiful, approachable appearance. Straight shoulder-length black hair with soft bangs, light natural makeup, warm expressive eyes, wearing an oversized beige knit sweater. Friendly, energetic personality like a genuine lifestyle creator. Sitting in a cozy modern Tokyo apartment with warm ambient lighting.\nShot Breakdown\nSHOT 1 (0–2s) — The Reveal\nSelfie angle, laughing while holding the sushi takeaway box toward the camera. Quick jump cut.\nDialogue: \"Okay... sushi night it is.\"\nSHOT 2 (2–4s) — The Open\nOverhead handheld shot as she opens the container. Fresh sushi is revealed with slight camera shake as she leans closer.\nSHOT 3 (4–6s) — The Close-Up\nQuick punch-in on the sushi. Autofocus locks naturally onto the salmon and tuna while the background softly blurs. Ambient apartment sounds only.\nSHOT 4 (6–8s) — The Pick-Up\nLow candid angle as chopsticks lift a salmon nigiri. Natural movement, slight handheld shake like a friend filming across the table.\nSHOT 5 (8–10s) — The Reaction\nBack to selfie cam. She takes a bite, smiles with widened eyes.\nDialogue: \"Wow... that's amazing.\"\nSHOT 6 (10–12s) — The Cutaway\nClose handheld shot of the sushi tray on the table while her hand reaches in for another piece just outside the frame. Casual lifestyle b-roll feel.\nSHOT 7 (12–14s) — The Wrap-Up\nSelfie angle. She raises a piece of sushi with chopsticks toward the camera.\nDialogue: \"Best decision tonight.\"\nSHOT 8 (14–15s) — End Tag\nQuick freeze on the sushi tray with natural handheld wobble. Soft handwritten-style overlay: \"sushi night = perfect 🍣\" Cut to black.\nLook & Feel\nWarm indoor lighting, slightly grainy smartphone sensor look, imperfect framing, authentic reactions, minimal dialogue (3 short lines), hard jump cuts, casual pacing, genuine creator content.\nNegative Prompt\ncinematic color grading, overly smooth gimbal movement, studio lighting, professional voiceover, staged acting, CGI food, unrealistic fish texture, distorted sushi, extra fingers, warped hands, text glitches, logo distortion, overly polished commercial feel.",
    "promptSourceUrl": "https://x.com/oggii_0/status/2082837405812613617",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/oggii_0/status/2082837405812613617",
    "sourceLabel": "X 原帖 · @oggii_0",
    "author": "Oogie",
    "publishedAt": "2026-07-30T14:35:09.450Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082837405812613617.mp4",
    "posterUrl": "/posters/x/2082837405812613617.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1698×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "ugc"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This no longer feels like just another AI tool it feels like a complete creative engine.\n\nI've been impressed by how MiniMax H3 seamlessly brings together visuals, motion, and audio into one unified workflow.\n\nBuilt by @Hailuo_AI, it can understand text, images, video, and audio all at once, making it possible to create cinematic content through a single creative process.\n\n#MiniMaxH3\n\nPrompt: Duration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, slight motion blur, TikTok/Reels energy — NOT cinematic, NOT overly polished. Feels like a real creator filmed this on their phone.\nProduct Reference: Use the uploaded assorted sushi platter image as the only product reference. Keep the sushi arrangement, fish cuts, rice texture, garnish placement, colors, and proportions identical in every shot—do not redesign or alter the food.\nCamera: iPhone 15 Pro front + back camera switching, handheld with natural wobble, slight autofocus breathing, vertical-style framing cropped to 16:9, occasional finger near the edge of the lens, realistic phone HDR, warm apartment lighting, subtle reflections on the takeaway container.\nCharacter Description\nName (for reference): Aoi\nA young Japanese woman in her early 20s with a naturally beautiful, approachable appearance. Straight shoulder-length black hair with soft bangs, light natural makeup, warm expressive eyes, wearing an oversized beige knit sweater. Friendly, energetic personality like a genuine lifestyle creator. Sitting in a cozy modern Tokyo apartment with warm ambient lighting.\nShot Breakdown\nSHOT 1 (0–2s) — The Reveal\nSelfie angle, laughing while holding the sushi takeaway box toward the camera. Quick jump cut.\nDialogue: \"Okay... sushi night it is.\"\nSHOT 2 (2–4s) — The Open\nOverhead handheld shot as she opens the container. Fresh sushi is revealed with slight camera shake as she leans closer.\nSHOT 3 (4–6s) — The Close-Up\nQuick punch-in on the sushi. Autofocus locks naturally onto the salmon and tuna while the background softly blurs. Ambient apartment sounds only.\nSHOT 4 (6–8s) — The Pick-Up\nLow candid angle as chopsticks lift a salmon nigiri. Natural movement, slight handheld shake like a friend filming across the table.\nSHOT 5 (8–10s) — The Reaction\nBack to selfie cam. She takes a bite, smiles with widened eyes.\nDialogue: \"Wow... that's amazing.\"\nSHOT 6 (10–12s) — The Cutaway\nClose handheld shot of the sushi tray on the table while her hand reaches in for another piece just outside the frame. Casual lifestyle b-roll feel.\nSHOT 7 (12–14s) — The Wrap-Up\nSelfie angle. She raises a piece of sushi with chopsticks toward the camera.\nDialogue: \"Best decision tonight.\"\nSHOT 8 (14–15s) — End Tag\nQuick freeze on the sushi tray with natural handheld wobble. Soft handwritten-style overlay: \"sushi night = perfect 🍣\" Cut to black.\nLook & Feel\nWarm indoor lighting, slightly grainy smartphone sensor look, imperfect framing, authentic reactions, minimal dialogue (3 short lines), hard jump cuts, casual pacing, genuine creator content.\nNegative Prompt\ncinematic color grading, overly smooth gimbal movement, studio lighting, professional voiceover, staged acting, CGI food, unrealistic fish texture, distorted sushi, extra fingers, warped hands, text glitches, logo distortion, overly polished commercial feel.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082830892209221921",
    "title": "赛博朋克冰蓝角色觉醒",
    "titleEn": "Cyberpunk Ice-Blue Character Awakening",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《赛博朋克冰蓝角色觉醒》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Cyberpunk Ice-Blue Character Awakening. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use the uploaded image as the exact source image and first frame. Preserve the original character design, composition, face, pale skin, glowing icy blue eyes, braided black hair, cybernetic head implants, exposed mechanical neck and chest structure, white and cobalt-blue jacket, blue serpentine mechanical creatures, labels, symbols, and the clean white background. Keep the result extremely faithful to the original illustration. The animation should feel like the original artwork has come alive, not been redesigned.  Create a premium, stylish, high-end anime animation with a refined cyberpunk aesthetic. The overall impression should be elegant, cold, hypnotic, dangerous, and visually luxurious. Emphasize precise Japanese anime craftsmanship: crisp linework, beautiful cel shading, highly detailed mechanical rendering, controlled highlights, graceful motion, and strong temporal consistency. The animation should look polished, expensive, and intentional.  The central character remains calm, dominant, and nearly still. She faces forward and maintains an unwavering, piercing gaze directly into the camera. Her expression is cool, emotionless, intelligent, and slightly predatory. Avoid exaggerated acting. Only subtle micro-movements are allowed in her eyelids, breathing, neck posture, and facial tension. Her presence should feel powerful even in stillness.  The main animated elements are the multiple blue mechanical serpents surrounding her. They move with smooth, layered, snake-like elegance. Each serpent has slightly different timing, path, and rhythm, so the movement feels organic and sophisticated rather than synchronized. Their bodies undulate in natural S-curves with believable weight, inertia, and continuous motion. Some serpents slowly raise their heads, some glide across her shoulders and around her face, some tighten or loosen their coils, and some subtly cross behind or in front of the composition to reshape the frame. Their movement should feel alive, hypnotic, and controlled, never frantic. Preserve the segmentation, scale texture, metallic fittings, joints, and cable-like details. Small articulated components should shift with delicate mechanical precision.  Her long green tongue moves like a real serpent tongue but remains elegant and stylized. It slowly extends, retracts, and sways with controlled, organic softness. The forked tip performs subtle tasting-the-air flicks. The tongue should feel eerie, sensual, and mesmerizing, never comedic. The tongue motion should complement the serpents’ motion, acting like a focal accent that draws attention toward her mouth and face.  Add subtle secondary motion throughout the scene. Her shoulders and upper torso rise and fall gently with slow breathing. Tiny motions should ripple through the hood, jacket folds, loose blue fabric, fine cables, and a few hair details. The exposed mechanical chest and neck assembly should feel alive through micro-movements: slight cable tension shifts, minute piston adjustments, tiny joint rotations, and restrained vibration. These details must remain subtle and tasteful, never distracting from the face and serpents.  Lighting should remain clean and striking. Preserve the white background and the strong graphic contrast between the white, cobalt blue, black mechanical detailing, and neon green tongue. Let reflections glide softly across the blue serpents’ scales and metallic joints. Metallic parts in the neck and chest may shimmer subtly. The green tongue should remain a vivid accent with slight wet gloss or faint luminous emphasis. Maintain a polished illustration-like finish while introducing dimensionality through moving highlights and soft shadow transitions.  The camera should perform an extremely slow and smooth frontal dolly-in over the full 15 seconds. The framing begins as a stable medium close-up and gradually becomes slightly tighter and more intense. The camera should never move abruptly. No shake, no tilt, no sudden zoom, no spinning, and no dramatic reframing. The face, eyes, and tongue should become increasingly dominant as the sequence progresses, while the surrounding serpents continue to frame the composition. The movement of the camera should feel cinematic and hypnotic, reinforcing tension through slow approach.  Structure the motion over the 15-second duration with deliberate pacing:  Seconds 0–3: begin in near stillness. Only the faintest breathing, tiny hood and cable motion, and a barely perceptible serpent twitch. Establish tension and the character’s unwavering blue-eyed stare.  Seconds 3–6: the serpents begin to awaken. One or two serpents slowly lift their heads and start gentle undulating motion. Reflections start traveling across their blue surfaces. The tongue begins a slight side-to-side sway.  Seconds 6–9: more serpents join the motion. The body language of the serpents becomes richer and more layered, circling her shoulders and face with elegant S-curve motion. The tongue extends a little farther and retracts slowly, with subtle fork-tip flicks. The camera continues its slow push inward.  Seconds 9–12: reach the most hypnotic phase. The serpents are now fully active but still graceful and controlled, framing her face dynamically. Their movement should feel luxurious and predatory. The tongue performs its most visually striking serpent-like motion here, synchronized with the heightened tension. Highlights glide across scales and metal details. Her gaze remains locked and emotionally unreadable.  Seconds 12–15: sustain and slightly resolve the intensity. The serpents continue moving but begin to feel more settled, as if entering a poised equilibrium. The camera reaches its closest framing, emphasizing her eyes and green tongue while still preserving enough serpents in frame for visual impact. End on a powerful, composed, hypnotic final moment that feels elegant and complete.  The emotional tone should be futuristic, eerie, sensual, intelligent, and luxurious. The final result should resemble a high-budget anime opening shot or a premium animated key visual. Prioritize elegance, polish, and controlled visual storytelling over excessive motion. Every motion should feel deliberate and aesthetically composed.  Maintain strict model fidelity and visual stability. Do not alter the character’s face shape, eye design, hairstyle, skin tone, clothing, mechanical anatomy, serpent design, or color palette. Do not add new objects, extra serpents, extra tongues, extra limbs, or background elements. Do not morph the face or let the eyes drift. Do not let the serpents fuse together unnaturally or intersect in broken ways. Do not allow clipping through the body or clothing. Do not change the jacket symbols or graphic details. Do not shift into 3D, live action, or a different illustration style. Avoid flicker, line wobble, perspective drift, detail melting, unstable anatomy, scale inconsistency, random deformation, abrupt motion, or composition changes.",
    "promptSourceUrl": "https://x.com/ainextastro/status/2082830892209221921",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ainextastro/status/2082830892209221921",
    "sourceLabel": "X 原帖 · @ainextastro",
    "author": "astro",
    "publishedAt": "2026-07-30T14:09:16.486Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082830892209221921.mp4",
    "posterUrl": "/posters/x/2082830892209221921.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Horror",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "illustration"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "過去作を動かしてみた\nprompt\n\nUse the uploaded image as the exact source image and first frame. Preserve the original character design, composition, face, pale skin, glowing icy blue eyes, braided black hair, cybernetic head implants, exposed mechanical neck and chest structure, white and cobalt-blue jacket, blue serpentine mechanical creatures, labels, symbols, and the clean white background. Keep the result extremely faithful to the original illustration. The animation should feel like the original artwork has come alive, not been redesigned.  Create a premium, stylish, high-end anime animation with a refined cyberpunk aesthetic. The overall impression should be elegant, cold, hypnotic, dangerous, and visually luxurious. Emphasize precise Japanese anime craftsmanship: crisp linework, beautiful cel shading, highly detailed mechanical rendering, controlled highlights, graceful motion, and strong temporal consistency. The animation should look polished, expensive, and intentional.  The central character remains calm, dominant, and nearly still. She faces forward and maintains an unwavering, piercing gaze directly into the camera. Her expression is cool, emotionless, intelligent, and slightly predatory. Avoid exaggerated acting. Only subtle micro-movements are allowed in her eyelids, breathing, neck posture, and facial tension. Her presence should feel powerful even in stillness.  The main animated elements are the multiple blue mechanical serpents surrounding her. They move with smooth, layered, snake-like elegance. Each serpent has slightly different timing, path, and rhythm, so the movement feels organic and sophisticated rather than synchronized. Their bodies undulate in natural S-curves with believable weight, inertia, and continuous motion. Some serpents slowly raise their heads, some glide across her shoulders and around her face, some tighten or loosen their coils, and some subtly cross behind or in front of the composition to reshape the frame. Their movement should feel alive, hypnotic, and controlled, never frantic. Preserve the segmentation, scale texture, metallic fittings, joints, and cable-like details. Small articulated components should shift with delicate mechanical precision.  Her long green tongue moves like a real serpent tongue but remains elegant and stylized. It slowly extends, retracts, and sways with controlled, organic softness. The forked tip performs subtle tasting-the-air flicks. The tongue should feel eerie, sensual, and mesmerizing, never comedic. The tongue motion should complement the serpents’ motion, acting like a focal accent that draws attention toward her mouth and face.  Add subtle secondary motion throughout the scene. Her shoulders and upper torso rise and fall gently with slow breathing. Tiny motions should ripple through the hood, jacket folds, loose blue fabric, fine cables, and a few hair details. The exposed mechanical chest and neck assembly should feel alive through micro-movements: slight cable tension shifts, minute piston adjustments, tiny joint rotations, and restrained vibration. These details must remain subtle and tasteful, never distracting from the face and serpents.  Lighting should remain clean and striking. Preserve the white background and the strong graphic contrast between the white, cobalt blue, black mechanical detailing, and neon green tongue. Let reflections glide softly across the blue serpents’ scales and metallic joints. Metallic parts in the neck and chest may shimmer subtly. The green tongue should remain a vivid accent with slight wet gloss or faint luminous emphasis. Maintain a polished illustration-like finish while introducing dimensionality through moving highlights and soft shadow transitions.  The camera should perform an extremely slow and smooth frontal dolly-in over the full 15 seconds. The framing begins as a stable medium close-up and gradually becomes slightly tighter and more intense. The camera should never move abruptly. No shake, no tilt, no sudden zoom, no spinning, and no dramatic reframing. The face, eyes, and tongue should become increasingly dominant as the sequence progresses, while the surrounding serpents continue to frame the composition. The movement of the camera should feel cinematic and hypnotic, reinforcing tension through slow approach.  Structure the motion over the 15-second duration with deliberate pacing:  Seconds 0–3: begin in near stillness. Only the faintest breathing, tiny hood and cable motion, and a barely perceptible serpent twitch. Establish tension and the character’s unwavering blue-eyed stare.  Seconds 3–6: the serpents begin to awaken. One or two serpents slowly lift their heads and start gentle undulating motion. Reflections start traveling across their blue surfaces. The tongue begins a slight side-to-side sway.  Seconds 6–9: more serpents join the motion. The body language of the serpents becomes richer and more layered, circling her shoulders and face with elegant S-curve motion. The tongue extends a little farther and retracts slowly, with subtle fork-tip flicks. The camera continues its slow push inward.  Seconds 9–12: reach the most hypnotic phase. The serpents are now fully active but still graceful and controlled, framing her face dynamically. Their movement should feel luxurious and predatory. The tongue performs its most visually striking serpent-like motion here, synchronized with the heightened tension. Highlights glide across scales and metal details. Her gaze remains locked and emotionally unreadable.  Seconds 12–15: sustain and slightly resolve the intensity. The serpents continue moving but begin to feel more settled, as if entering a poised equilibrium. The camera reaches its closest framing, emphasizing her eyes and green tongue while still preserving enough serpents in frame for visual impact. End on a powerful, composed, hypnotic final moment that feels elegant and complete.  The emotional tone should be futuristic, eerie, sensual, intelligent, and luxurious. The final result should resemble a high-budget anime opening shot or a premium animated key visual. Prioritize elegance, polish, and controlled visual storytelling over excessive motion. Every motion should feel deliberate and aesthetically composed.  Maintain strict model fidelity and visual stability. Do not alter the character’s face shape, eye design, hairstyle, skin tone, clothing, mechanical anatomy, serpent design, or color palette. Do not add new objects, extra serpents, extra tongues, extra limbs, or background elements. Do not morph the face or let the eyes drift. Do not let the serpents fuse together unnaturally or intersect in broken ways. Do not allow clipping through the body or clothing. Do not change the jacket symbols or graphic details. Do not shift into 3D, live action, or a different illustration style. Avoid flicker, line wobble, perspective drift, detail melting, unstable anatomy, scale inconsistency, random deformation, abrupt motion, or composition changes.\n\nMidjourney #MiniMaxH3 @Hailuo_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082828444484960451",
    "title": "手绘姜烧猪肉料理动画",
    "titleEn": "Hand-Drawn Ginger Pork Cooking Anime",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《手绘姜烧猪肉料理动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Hand-Drawn Ginger Pork Cooking Anime. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "手描きの日本の2Dアニメ、温かみのあるセルシェーディング、居心地のいい町の定食屋の厨房。15秒、ゆっくり見せる6カット、各約2.5秒。寄りのクローズアップ中心で、食欲をそそる濃厚な演出：脂、照り、湯気、たっぷりの量。\n1 まな板を真上から：包丁が厚切りの豚ロースを切る、白い脂の霜降り。\n2 斜め45度：熱々の鉄のフライパンに肉を並べ入れる、油が弾け、縁が反り返る。\n3 マクロ：脂が溶け出し、深い焼き色の焦げ目が広がる。\n4 寄り：濃い生姜醤油ダレを回し入れる、ぐつぐつと煮立って濃厚な照りのタレになり、肉に絡みつく。\n5 横から：フライパンを揺すり、すべての肉に鏡のような照りがまとわり、湯気が立ち上る。\n6 ラスト：湯気の立つ白ご飯の丼に生姜焼きを山盛りにのせ、タレが米粒の間へ垂れていく。\n料理と手元のみ。顔なし、食べるシーンなし、文字なし、BGMなし、ナレーションなし。実写風なし、3Dなし。",
    "promptSourceUrl": "https://x.com/ozuozuai99/status/2082828444484960451",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ozuozuai99/status/2082828444484960451",
    "sourceLabel": "X 原帖 · @ozuozuai99",
    "author": "オズ",
    "publishedAt": "2026-07-30T13:59:32.903Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082828444484960451.mp4",
    "posterUrl": "/posters/x/2082828444484960451.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "food",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "新モデル Hailuo H3が登場✨\n\n先行体験で触らせていただきました！\n最近はAI動画生成の費用が高騰しどうなるのやらと思っていましたが安くてこのクオリティは結構いいですね！\n\nプロンプトも載せとくのでご参考に👇\n\n手描きの日本の2Dアニメ、温かみのあるセルシェーディング、居心地のいい町の定食屋の厨房。15秒、ゆっくり見せる6カット、各約2.5秒。寄りのクローズアップ中心で、食欲をそそる濃厚な演出：脂、照り、湯気、たっぷりの量。\n1 まな板を真上から：包丁が厚切りの豚ロースを切る、白い脂の霜降り。\n2 斜め45度：熱々の鉄のフライパンに肉を並べ入れる、油が弾け、縁が反り返る。\n3 マクロ：脂が溶け出し、深い焼き色の焦げ目が広がる。\n4 寄り：濃い生姜醤油ダレを回し入れる、ぐつぐつと煮立って濃厚な照りのタレになり、肉に絡みつく。\n5 横から：フライパンを揺すり、すべての肉に鏡のような照りがまとわり、湯気が立ち上る。\n6 ラスト：湯気の立つ白ご飯の丼に生姜焼きを山盛りにのせ、タレが米粒の間へ垂れていく。\n料理と手元のみ。顔なし、食べるシーンなし、文字なし、BGMなし、ナレーションなし。実写風なし、3Dなし。\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082827829172117943",
    "title": "草莓四季匹配剪辑食品广告",
    "titleEn": "Strawberry seasonal match-cut food commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《草莓四季匹配剪辑食品广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1698×720。",
    "summaryEn": "An H3 output for Strawberry seasonal match-cut food commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1698×720.",
    "prompt": "Premium food commercial. Extreme macro shot of a single ripe strawberry falling toward a white ceramic bowl. Before it lands, the environment changes through four seamless seasonal match cuts: a sunlit spring greenhouse, a hot summer picnic, an autumn kitchen and a candlelit winter dining room. The strawberry stays in exactly the same screen position, size, orientation and downward motion through every transition. It lands in fresh cream during the winter scene, sending one graceful crown-shaped splash upward. Photorealistic food texture, high-speed product cinematography, clean commercial lighting. Each season has distinct environmental sound, but one continuous musical phrase connects the entire sequence. No hands, labels or additional fruit.",
    "promptSourceUrl": "https://x.com/HBCoop_/status/2082827841390198845",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/HBCoop_/status/2082827829172117943",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-07-30T13:57:06.201Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082827829172117943.mp4",
    "posterUrl": "/posters/x/2082827829172117943.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1698×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Text to video in the new MiniMax H3 from @Hailuo_AI\n\nPrompt below:",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082820121882665060",
    "title": "白色无影棚奢华眼镜广告",
    "titleEn": "White-Cyclorama Luxury Eyewear Campaign",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《白色无影棚奢华眼镜广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×1280。",
    "summaryEn": "An H3 output for White-Cyclorama Luxury Eyewear Campaign. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×1280.",
    "prompt": "Create a premium 9:16 fashion-eyewear commercial. Match the reference video’s shot rhythm, edit speed, white-cyclorama look, and severe fashion attitude. Use a seamless minimal white studio with clean, bold, avant-garde art direction worthy of a global luxury campaign.\n- Use Image 1 for the key visual: two full-body female models, one Black and one white, preserving their elevated wardrobe, body language, studio lighting, runway presence, and cool attitude. Use Image 2 for facial details. Both models wear futuristic luxury eyewear based on Image 3: wraparound curved lenses, a sharp cat-eye/goggle hybrid silhouette, mirrored reflections, streamlined temples, and the finish of a premium fashion accessory.",
    "promptSourceUrl": "https://x.com/hey_leomartin/status/2082820121882665060",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/hey_leomartin/status/2082820121882665060",
    "sourceLabel": "X 原帖 · @hey_leomartin",
    "author": "Leo Martin",
    "publishedAt": "2026-07-30T13:26:28.640Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082820121882665060.mp4",
    "posterUrl": "/posters/x/2082820121882665060.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is really impressive for commercial creation.\n\nCreating ads with AI is becoming a completely different experience. With multimodal references, you can combine images, videos, audio, and text to build more consistent and controllable visuals.\n\nUp to 12 references + native 2K generation makes H3 a powerful tool for bringing advertising ideas to life.\n\nprompt：\n- Create a premium 9:16 fashion-eyewear commercial. Match the reference video’s shot rhythm, edit speed, white-cyclorama look, and severe fashion attitude. Use a seamless minimal white studio with clean, bold, avant-garde art direction worthy of a global luxury campaign.\n- Use Image 1 for the key visual: two full-body female models, one Black and one white, preserving their elevated wardrobe, body language, studio lighting, runway presence, and cool attitude. Use Image 2 for facial details. Both models wear futuristic luxury eyewear based on Image 3: wraparound curved lenses, a sharp cat-eye/goggle hybrid silhouette, mirrored reflections, streamlined temples, and the finish of a premium fashion accessory.\n\n#hailuoai #hailuo #MiniMaxH3 #H3\n@Hailuo_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082818054598066367",
    "title": "暗黑说唱时尚杂志音乐成片",
    "titleEn": "Dark Rap Fashion Zine Music film",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《暗黑说唱时尚杂志音乐成片》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Dark Rap Fashion Zine Music film. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Style: dark-rap -rap music video with photoreal high-fashion polish and the texture of a scanned film magazine—high contrast without looking cheap. film scans, underground-music posters, and zine collage. Add coarse grain, subtle gate weave, halftone dots, rough print edges, and slight scan misregistration. Keep the edit fast and use hard cuts only—no fades or soft transitions. Match the typographic treatment and surface texture of the reference images.",
    "promptSourceUrl": "https://x.com/MO_IAI/status/2082818054598066367",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/MO_IAI/status/2082818054598066367",
    "sourceLabel": "X 原帖 · @MO_IAI",
    "author": "MO",
    "publishedAt": "2026-07-30T13:18:15.761Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082818054598066367.mp4",
    "posterUrl": "/posters/x/2082818054598066367.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Now Music Videos are easy with H3\nMore testing on the way.\nPrompt:\nStyle: dark-rap -rap music video with photoreal high-fashion polish and the texture of a scanned film magazine—high contrast without looking cheap. film scans, underground-music posters, and zine collage. Add coarse grain, subtle gate weave, halftone dots, rough print edges, and slight scan misregistration. Keep the edit fast and use hard cuts only—no fades or soft transitions. Match the typographic treatment and surface texture of the reference images.\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082804771220804089",
    "title": "The Brief: \"Jessica x The Bag\"",
    "titleEn": "The Brief: \"Jessica x The Bag\"",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《The Brief: \"Jessica x The Bag\"》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for The Brief: \"Jessica x The Bag\". This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "The Brief: \"Jessica x The Bag\"\n\n15 seconds. One face, one bag, one mood — effortless luxury.\n\nJessica walks into frame on a sun-warmed city street, the cognac leather bag with its hand-scribbled blue design swinging from her hand. She's in a fitted black dress with tan accents that pick up the bag's exact shade — every detail intentional. The camera pushes in as she lifts the bag, gold hardware flashing in the light, her expression shifting from composed to a knowing half-smile. A slow turn through a marble-lit interior, dress catching the glow, bag cradled at her arm. Final beat: she faces the lens dead-on, bag held forward, hero shot, soft rim light framing her like a campaign still. Fade.\n\nWarm, glossy, early-2000s fashion-film color science throughout — think The Devil Wears Prada meets a runway ad. Same face, same bag, same world, held together across every cut.",
    "promptSourceUrl": "https://x.com/AIwithJessica/status/2082804771220804089",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AIwithJessica/status/2082804771220804089",
    "sourceLabel": "X 原帖 · @AIwithJessica",
    "author": "Jessica Collins",
    "publishedAt": "2026-07-30T12:25:28.757Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082804771220804089.mp4",
    "posterUrl": "/posters/x/2082804771220804089.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Fashion",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tried the new MiniMax H3 model and the quality genuinely stopped me mid-scroll 🤯\n\nMade with @Hailuo_AI\n\nH3 is a general-purpose multimodal video model that goes way beyond single-purpose tools, it understands text, images, video, and audio together, reads creative intent, and outputs coherent, natural audiovisual content in one seamless pass.\n\n#MiniMaxH3\n\nHere's what I fed it 👇\n\nThe Brief: \"Jessica x The Bag\"\n\n15 seconds. One face, one bag, one mood — effortless luxury.\n\nJessica walks into frame on a sun-warmed city street, the cognac leather bag with its hand-scribbled blue design swinging from her hand. She's in a fitted black dress with tan accents that pick up the bag's exact shade — every detail intentional. The camera pushes in as she lifts the bag, gold hardware flashing in the light, her expression shifting from composed to a knowing half-smile. A slow turn through a marble-lit interior, dress catching the glow, bag cradled at her arm. Final beat: she faces the lens dead-on, bag held forward, hero shot, soft rim light framing her like a campaign still. Fade.\n\nWarm, glossy, early-2000s fashion-film color science throughout — think The Devil Wears Prada meets a runway ad. Same face, same bag, same world, held together across every cut.\n\nThat's it. That's the whole prompt. And H3 just... got it.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082799981426037151",
    "title": "风暴悬崖高尔夫物理挑战",
    "titleEn": "Storm-Cliff Golf Physics Challenge",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《风暴悬崖高尔夫物理挑战》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Storm-Cliff Golf Physics Challenge. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Minimax H3 Prompt\n\nA dramatic sports sequence begins on a lush, windswept coastal cliff during a violent rainstorm. A golfer in bright yellow rain gear stands on the tee box, gripping a driver. His stance is planted firmly, eyes locked on a distant green across a churning ocean cove. As he swings, the clubhead connects with a resounding crack, launching the dimpled white ball into the stormy sky. The camera immediately accelerates and locks onto the golf ball mid-flight. The camera tracks directly behind the spinning sphere, maintaining perfect alignment as it cuts through the driving rain. The ocean waves and dark clouds become a streaking motion blur while the ball remains sharp and centered. The howling wind roars as the projectile travels straight and true. The manicured green grows rapidly larger. In the final seconds, the camera closes in tightly as the ball strikes the turf. End on an extreme close-up of the ball ripping through the wet grass, throwing up a spray of dirt before dropping perfectly into the cup.",
    "promptSourceUrl": "https://x.com/Dheepanratnam/status/2082799984055906687",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Dheepanratnam/status/2082799981426037151",
    "sourceLabel": "X 原帖 · @Dheepanratnam",
    "author": "Dheepan Ratnam",
    "publishedAt": "2026-07-30T12:06:26.781Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082799981426037151.mp4",
    "posterUrl": "/posters/x/2082799981426037151.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Action & VFX",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I just tested the new Hailuo Minimax H3 model with an old prompt ⤵️\n\nother models struggle with this\n\nthe physics are actually insane. violent rainstorm, high-speed ball tracking, and perfect water spray.\n\nits the only model that could do it other than \nseedance 2.0 @Hailuo_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082799917140197798",
    "title": "披萨之夜 UGC 开箱 Vlog",
    "titleEn": "Pizza night UGC Domino’s vlog",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《披萨之夜 UGC 开箱 Vlog》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Pizza night UGC Domino’s vlog. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "VIDEO PROMPT — \"Pizza Night Vlog\" (UGC iPhone Style)\n\nDuration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, slight motion blur, TikTok/Reels energy — NOT cinematic, NOT overly polished. Feels like a real creator filmed this on their phone.\nProduct Reference: Use the uploaded Domino's Pepperoni Pizza image as the only product reference. Keep crust thickness, cheese texture, bake color, pepperoni placement, and proportions identical in every cut — no redesigning the pizza.\nCamera: iPhone 15 Pro front + back camera switching, handheld, natural wobble, autofocus hunting slightly (realistic), vertical-style framing cropped to 16:9, occasional finger near lens edge, natural room lighting + phone flash reflections on the pizza box.\n\nCharacter Description\n\nName (for reference): Mia\n\nAwoman in her mid-20s, naturally attractive and beautiful with an approachable, girl-next-door charm — not overly done up. Wavy sandy-blonde hair pulled back loosely, light natural makeup, wearing a cozy oversized cream sweater. Warm, genuine smile, expressive eyes, casual energetic personality like a real lifestyle vlogger. Sits in a softly lit modern kitchen/living room.\n\nShot Breakdown\nSHOT 1 (0–2s) — The Grab\n\nSelfie-angle, she's mid-laugh holding up the Domino's box to camera. Quick jump cut. Dialogue: \"Okay so it's officially pizza night—\"\n\nSHOT 2 (2–4s) — The Open\nCut to overhead handheld shot, box flips open, steam rising off the pizza, slight camera shake as she leans in.\n\nSHOT 3 (4–6s) — The Zoom\nQuick zoom-punch into the pizza, phone camera autofocus adjusts naturally, cheese and pepperoni in focus, ambient kitchen sounds.\n\nSHOT 4 (6–8s) — The Pull\nCut to her hands lifting a slice, natural cheese pull, filmed from a slightly low candid angle like a friend filming across the table.\n\nSHOT 5 (8–10s) — The Reaction\nCut back to selfie-cam, she takes a bite, eyes widen, quick genuine reaction. Dialogue: \"Oh my god, that's so good.\"\n\nSHOT 6 (10–12s) — The Candid Cutaway\nJump cut to a close, slightly shaky shot of the pizza box on the counter, her hand grabbing another slice off-frame, casual b-roll energy.\n\nSHOT 7 (12–14s) — The Wrap-Up\nBack to selfie angle, she grins at camera, holding slice up like a toast. Dialogue: \"Dominos, y'all know what to do.\"\n\nSHOT 8 (14–15s) — End Tag\nQuick freeze/cut to the box logo close-up, natural handheld wobble, soft text overlay in casual font: \"pizza night = solved 🍕\" — cut to black.\n\nLook & Feel\nWarm indoor lighting, slightly grainy natural phone sensor look, imperfect framing, real reactions, minimal dialogue (3 short lines total), authentic pacing with hard jump cuts instead of smooth transitions.\n\nNegative Prompt\n\ncinematic grade, overly smooth camera moves, studio lighting, professional voiceover, staged acting, CGI look, plastic cheese, distorted pepperoni, extra fingers, warped hands, text glitches, logo distortion, overly polished commercial feel.",
    "promptSourceUrl": "https://x.com/ShamiWeb3/status/2082799917140197798",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ShamiWeb3/status/2082799917140197798",
    "sourceLabel": "X 原帖 · @ShamiWeb3",
    "author": "Shami",
    "publishedAt": "2026-07-30T12:06:11.454Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082799917140197798.mp4",
    "posterUrl": "/posters/x/2082799917140197798.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Vlog",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc",
      "commercial"
    ],
    "scenes": [
      "dialogue",
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This doesn't feel like a specialized AI tool anymore; it feels like a complete creative engine.\n\nMiniMax H3 impressed me with how naturally it combines visuals, motion, and sound into a single workflow.\n\nPowered by @Hailuo_AI, the model understands text, images, video, and audio simultaneously, making it possible to generate cinematic content from one unified creative process.\n\n#MiniMaxH3\n\nPrompt:\nVIDEO PROMPT — \"Pizza Night Vlog\" (UGC iPhone Style)\n\nDuration: 15 seconds | Aspect Ratio: 16:9 | Style: Authentic UGC / iPhone selfie-vlog, handheld, natural light, slight motion blur, TikTok/Reels energy — NOT cinematic, NOT overly polished. Feels like a real creator filmed this on their phone.\nProduct Reference: Use the uploaded Domino's Pepperoni Pizza image as the only product reference. Keep crust thickness, cheese texture, bake color, pepperoni placement, and proportions identical in every cut — no redesigning the pizza.\nCamera: iPhone 15 Pro front + back camera switching, handheld, natural wobble, autofocus hunting slightly (realistic), vertical-style framing cropped to 16:9, occasional finger near lens edge, natural room lighting + phone flash reflections on the pizza box.\n\nCharacter Description\n\nName (for reference): Mia\n\nAwoman in her mid-20s, naturally attractive and beautiful with an approachable, girl-next-door charm — not overly done up. Wavy sandy-blonde hair pulled back loosely, light natural makeup, wearing a cozy oversized cream sweater. Warm, genuine smile, expressive eyes, casual energetic personality like a real lifestyle vlogger. Sits in a softly lit modern kitchen/living room.\n\nShot Breakdown\nSHOT 1 (0–2s) — The Grab\n\nSelfie-angle, she's mid-laugh holding up the Domino's box to camera. Quick jump cut. Dialogue: \"Okay so it's officially pizza night—\"\n\nSHOT 2 (2–4s) — The Open\nCut to overhead handheld shot, box flips open, steam rising off the pizza, slight camera shake as she leans in.\n\nSHOT 3 (4–6s) — The Zoom\nQuick zoom-punch into the pizza, phone camera autofocus adjusts naturally, cheese and pepperoni in focus, ambient kitchen sounds.\n\nSHOT 4 (6–8s) — The Pull\nCut to her hands lifting a slice, natural cheese pull, filmed from a slightly low candid angle like a friend filming across the table.\n\nSHOT 5 (8–10s) — The Reaction\nCut back to selfie-cam, she takes a bite, eyes widen, quick genuine reaction. Dialogue: \"Oh my god, that's so good.\"\n\nSHOT 6 (10–12s) — The Candid Cutaway\nJump cut to a close, slightly shaky shot of the pizza box on the counter, her hand grabbing another slice off-frame, casual b-roll energy.\n\nSHOT 7 (12–14s) — The Wrap-Up\nBack to selfie angle, she grins at camera, holding slice up like a toast. Dialogue: \"Dominos, y'all know what to do.\"\n\nSHOT 8 (14–15s) — End Tag\nQuick freeze/cut to the box logo close-up, natural handheld wobble, soft text overlay in casual font: \"pizza night = solved 🍕\" — cut to black.\n\nLook & Feel\nWarm indoor lighting, slightly grainy natural phone sensor look, imperfect framing, real reactions, minimal dialogue (3 short lines total), authentic pacing with hard jump cuts instead of smooth transitions.\n\nNegative Prompt\n\ncinematic grade, overly smooth camera moves, studio lighting, professional voiceover, staged acting, CGI look, plastic cheese, distorted pepperoni, extra fingers, warped hands, text glitches, logo distortion, overly polished commercial feel.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082799648528335119",
    "title": "水彩动画感官蒙太奇快切",
    "titleEn": "Watercolor anime fetish montage rapid cuts",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《水彩动画感官蒙太奇快切》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×1280。",
    "summaryEn": "An H3 output for Watercolor anime fetish montage rapid cuts. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×1280.",
    "prompt": "style:\n  visual: \"日本の水彩画風フルカラーアニメ\"\n  editing: \"0.5秒おきの高速カット割り\"\n  animation: \"作画枚数多め、滑らかな24fps\"\n  tone: \"官能的だが直接的にはせず、温度・湿度・心拍を映像化する\"\n\nproject:\n  id: \"fetish_montage_v3_with_broll\"\n  format: \"vertical_9x16\"\n  fps: 24\n  shot_duration: \"0.5s\"\n  total_shots: 28\n  total_duration: \"14s\"\n  ratio: \"fetish 22 : b-roll 6（約3.5:1）\"\n\nbroll_philosophy:\n  role: \"呼吸・象徴・感情の外部化\"\n  placement: \"フェチ的なカットが連続した直後、または感情の転換点\"\n  rule: \"直接的な性的表現は避け、心拍・温度・湿度・緊張を伝える\"\n\nbroll_categories:\n  nature_micro:\n    - \"雨粒が窓を一筋伝う\"\n    - \"蝋燭の炎が揺れる。息で消えそうな瞬間\"\n    - \"氷が割れる音と共に、ひびが走る\"\n    - \"花弁が一枚、テーブルへ落ちる\"\n    - \"蜘蛛の巣の水滴が揺れる\"\n    - \"煙草の灰が静かに落ちる\"\n    - \"湯気が立ち上り、消える\"\n    - \"夜の街灯の周りを虫が一匹飛ぶ\"\n\n  object_symbol:\n    - \"壁時計の秒針のアップ。止まりそうに見える\"\n    - \"コーヒーの黒い表面に天井が映る\"\n    - \"床に脱ぎ捨てられた靴。片方だけ\"\n    - \"誰もいないベッドの皺\"\n    - \"鳴り続ける電話。誰も出ない\"\n    - \"灰皿の吸い殻が二本。片方には口紅\"\n    - \"ドアの隙間から漏れる光\"\n    - \"鏡に映る空の椅子\"\n\n  texture_landscape:\n    - \"夜の濡れた路面。ヘッドライトが滲む\"\n    - \"ネオンサインのフリッカー\"\n    - \"シーシャの炭が赤く脈打つ\"\n    - \"金属表面に映る歪んだ顔\"\n    - \"革張りソファに沈む手の跡\"\n    - \"湯船の水面が揺れる\"\n\n  body_abstract:\n    - \"心臓の鼓動に合わせ、布団が微かに上下する\"\n    - \"人物の影だけが壁を伝う\"\n    - \"ガラス越しの人物の輪郭。ピントがずれる\"\n    - \"水滴の中に顔が映り込む\"\n\nsequence:\n  - id: 01\n    type: \"fetish\"\n    action: \"髪を持ち上げ、うなじを晒す\"\n    scale: \"macro\"\n\n  - id: 02\n    type: \"fetish\"\n    action: \"下唇を噛み、ゆっくり離す\"\n    scale: \"extreme_macro\"\n\n  - id: 03\n    type: \"b-roll\"\n    category: \"object_symbol\"\n    action: \"壁時計の秒針が、カチッと一秒進む\"\n    intent: \"時間の伸縮を意識させる\"\n\n  - id: 04\n    type: \"fetish\"\n    action: \"袖口のボタンを外し、袖を一度まくる\"\n    scale: \"macro\"\n\n  - id: 05\n    type: \"fetish\"\n    action: \"襟がわずかにずれ、鎖骨に影が落ちる\"\n    scale: \"macro\"\n\n  - id: 06\n    type: \"fetish\"\n    action: \"グラスの縁を指先でなぞる。結露が一滴落ちる\"\n    scale: \"extreme_macro\"\n\n  - id: 07\n    type: \"b-roll\"\n    category: \"nature_micro\"\n    action: \"蝋燭の炎が、一度だけ大きく揺れる\"\n    intent: \"呼吸の比喩。緊張のピーク\"\n\n  - id: 08\n    type: \"fetish\"\n    action: \"イヤリングを外す。直後に髪が肩へ落ちる\"\n    scale: \"macro\"\n\n  - id: 09\n    type: \"fetish\"\n    action: \"ヒールから足が抜ける\"\n    scale: \"macro\"\n\n  - id: 10\n    type: \"fetish\"\n    action: \"喉が小さく動く。嚥下\"\n    scale: \"extreme_macro\"\n\n  - id: 11\n    type: \"b-roll\"\n    category: \"texture_landscape\"\n    action: \"シーシャの炭が、脈打つように赤く光る\"\n    intent: \"心拍の外部化\"\n\n  - id: 12\n    type: \"fetish\"\n    action: \"口元から煙をゆっくり吐く\"\n    scale: \"macro_side\"\n\n  - id: 13\n    type: \"fetish\"\n    action: \"薄手の布越しに、肩甲骨が静かに動く\"\n    scale: \"macro_back\"\n\n  - id: 14\n    type: \"fetish\"\n    action: \"ペンのキャップを歯で咥え、ゆっくり引き抜く\"\n    scale: \"extreme_macro\"\n\n  - id: 15\n    type: \"b-roll\"\n    category: \"body_abstract\"\n    action: \"布団が呼吸に合わせて微かに上下する\"\n    intent: \"緊張の小休止。まだ続くという予告\"",
    "promptSourceUrl": "https://x.com/yachimat_manga/status/2082799651237843119",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/yachimat_manga/status/2082799648528335119",
    "sourceLabel": "X 原帖 · @yachimat_manga",
    "author": "yachimat - AI Short Anime",
    "publishedAt": "2026-07-30T12:05:07.412Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082799648528335119.mp4",
    "posterUrl": "/posters/x/2082799648528335119.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3フェティッシュモンタージュ\n質感結構良くない？ (プロンプトはリプ欄）\n\n@Hailuo_AI #MinimaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082798969499832514",
    "title": "四种泳姿美学演示",
    "titleEn": "Swimming strokes aesthetic collection",
    "model": "MiniMax H3（来源标注）",
    "mode": "FL2VA",
    "summary": "《四种泳姿美学演示》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Swimming strokes aesthetic collection. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "【风格】游泳姿态美学展示（Swimming Strokes Showcase），电影感写真质感（16:9 Cinematic, Photorealistic），高速摄影升格（Super Slow-mo），阳光泳池通透水蓝影调，长发水中飘动的唯美运动风\n【时长】15秒\n【场景】一条极长的泳道，向远处延伸看不到尽头，阳光透过玻璃顶棚斜射入水，在池底投下晃动的光网，水质通透碧蓝，画面里没有出发台、记分屏等比赛元素\n【角色】主角@图片1\n【结构机制】四种泳姿依次展示，切换全部在水中完成：靠身体滚转（Roll）或下潜滑行（Underwater Glide）自然过渡，不触壁、不转身，一条泳道游到底；每种泳姿约3秒，配一个专属机位和标志性画面\n\n[00:00-00:02.5] 开场：入水滑行（The Glide）\n侧面水下机位：她从池边轻盈跃入，身体呈流线型在水下滑行，长发在身后完全舒展开像一片黑色的绸缎，气泡串从指尖向后飞，阳光光柱扫过她的身体。\n【音效】入水一声轻\"唰\"，水下低频的宁静嗡响。\n\n[00:02.5-00:05.5] 第一式：蛙泳（Breaststroke）\n正面低水位机位：她浮出水面开始蛙泳，双臂对称外划、双腿蹬夹，身体一收一展从容优雅，每次抬头换气露出全脸，湿发贴在颊边，身前推出一圈圈弧形波纹缓缓荡向镜头。\n【音效】\"呼——蓬\"的舒缓划水节奏，换气声轻而清晰。\n\n[00:05.5-00:08.5] 第二式：滚转变仰泳（Roll → Backstroke）\n她在水中侧身一滚翻成仰面——切头顶俯拍机位：她脸朝上漂游，双臂轮流向后划出风车弧线，长发在头后的水里散成一片扇形随波摆动，眼睛望着顶棚光影，嘴角带笑，水珠从抬起的指尖甩落。\n【音效】滚转一声水响，之后是左右交替的轻快划水声。\n\n[00:08.5-00:11.5] 第三式：滚转变自由泳（Roll → Freestyle）\n她再侧滚翻回俯身——切侧面平行跟移机位：双臂交替抡水向前，侧脸贴水换气，打腿踢出细碎白沫，速度感起来，发丝随每次转头在水面甩出弧线，身后拉出一条长长的浪花尾迹伸向泳道深处。\n【音效】划水声变得连贯有力，节奏渐快。\n\n[00:11.5-00:14] 第四式：蝶泳压轴（Butterfly, Super Slow-mo）\n正面水面机位升格：她双臂同时抡起越过头顶，胸背波浪起伏，每次划水在身前炸开一对对称的水花翅膀，水珠悬在阳光里发亮，抬头换气瞬间水从下颌成帘滑落，长发向后甩出一道水弧。\n【音效】升格段\"哗——哗——\"的划水鼓点被拉长，水珠洒落如碎铃。\n\n[00:14-00:15] 收尾：下潜定格（The Descent）\n她顺势下潜，切水下机位：身体在光网里舒展成一条线滑行，长发向后飘散，她回头望向镜头露出笑容，气泡从唇边升起，画面定格。\n【音效】水声归于宁静的低频嗡响，定格配一声上扬收束音。\n\n【备注】四种泳姿姿态必须分明：蛙泳对称蹬夹、仰泳面部朝上、自由泳交替划臂、蝶泳双臂同步波浪推进；泳姿切换只用水中滚转或下潜滑行完成，全程不出现池壁；长发是贯穿全片的美学元素，每个镜头都要有发丝在水中或水面的动态",
    "promptSourceUrl": "https://x.com/johnAGI168/status/2082798969499832514",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/johnAGI168/status/2082798969499832514",
    "sourceLabel": "X 原帖 · @johnAGI168",
    "author": "John",
    "publishedAt": "2026-07-30T12:02:25.519Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082798969499832514.mp4",
    "posterUrl": "/posters/x/2082798969499832514.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3 早期访问测试⚙️\n\nSeedance 终于是迎来了对手，已经是非常接近了！\n\nprompt 👇\n\n【风格】游泳姿态美学展示（Swimming Strokes Showcase），电影感写真质感（16:9 Cinematic, Photorealistic），高速摄影升格（Super Slow-mo），阳光泳池通透水蓝影调，长发水中飘动的唯美运动风\n【时长】15秒\n【场景】一条极长的泳道，向远处延伸看不到尽头，阳光透过玻璃顶棚斜射入水，在池底投下晃动的光网，水质通透碧蓝，画面里没有出发台、记分屏等比赛元素\n【角色】主角@图片1\n【结构机制】四种泳姿依次展示，切换全部在水中完成：靠身体滚转（Roll）或下潜滑行（Underwater Glide）自然过渡，不触壁、不转身，一条泳道游到底；每种泳姿约3秒，配一个专属机位和标志性画面\n\n[00:00-00:02.5] 开场：入水滑行（The Glide）\n侧面水下机位：她从池边轻盈跃入，身体呈流线型在水下滑行，长发在身后完全舒展开像一片黑色的绸缎，气泡串从指尖向后飞，阳光光柱扫过她的身体。\n【音效】入水一声轻\"唰\"，水下低频的宁静嗡响。\n\n[00:02.5-00:05.5] 第一式：蛙泳（Breaststroke）\n正面低水位机位：她浮出水面开始蛙泳，双臂对称外划、双腿蹬夹，身体一收一展从容优雅，每次抬头换气露出全脸，湿发贴在颊边，身前推出一圈圈弧形波纹缓缓荡向镜头。\n【音效】\"呼——蓬\"的舒缓划水节奏，换气声轻而清晰。\n\n[00:05.5-00:08.5] 第二式：滚转变仰泳（Roll → Backstroke）\n她在水中侧身一滚翻成仰面——切头顶俯拍机位：她脸朝上漂游，双臂轮流向后划出风车弧线，长发在头后的水里散成一片扇形随波摆动，眼睛望着顶棚光影，嘴角带笑，水珠从抬起的指尖甩落。\n【音效】滚转一声水响，之后是左右交替的轻快划水声。\n\n[00:08.5-00:11.5] 第三式：滚转变自由泳（Roll → Freestyle）\n她再侧滚翻回俯身——切侧面平行跟移机位：双臂交替抡水向前，侧脸贴水换气，打腿踢出细碎白沫，速度感起来，发丝随每次转头在水面甩出弧线，身后拉出一条长长的浪花尾迹伸向泳道深处。\n【音效】划水声变得连贯有力，节奏渐快。\n\n[00:11.5-00:14] 第四式：蝶泳压轴（Butterfly, Super Slow-mo）\n正面水面机位升格：她双臂同时抡起越过头顶，胸背波浪起伏，每次划水在身前炸开一对对称的水花翅膀，水珠悬在阳光里发亮，抬头换气瞬间水从下颌成帘滑落，长发向后甩出一道水弧。\n【音效】升格段\"哗——哗——\"的划水鼓点被拉长，水珠洒落如碎铃。\n\n[00:14-00:15] 收尾：下潜定格（The Descent）\n她顺势下潜，切水下机位：身体在光网里舒展成一条线滑行，长发向后飘散，她回头望向镜头露出笑容，气泡从唇边升起，画面定格。\n【音效】水声归于宁静的低频嗡响，定格配一声上扬收束音。\n\n【备注】四种泳姿姿态必须分明：蛙泳对称蹬夹、仰泳面部朝上、自由泳交替划臂、蝶泳双臂同步波浪推进；泳姿切换只用水中滚转或下潜滑行完成，全程不出现池壁；长发是贯穿全片的美学元素，每个镜头都要有发丝在水中或水面的动态",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082798959014064531",
    "title": "温室红茶异世界动画",
    "titleEn": "Greenhouse tea isekai anime",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《温室红茶异世界动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Greenhouse tea isekai anime. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "高品質アニメ映像。\n\n作品トーンと世界観は、透明感のある夏のガラス温室から、紅茶の渦の内側に存在するオリジナルの小さな不思議の国へ連続する、上品で夢幻的な叙情ファンタジー。澄んだ白、水色、琥珀色、淡い金色を主役にし、怖さや混沌ではなく、好奇心、浮遊感、静かな高揚を描く。今回は1枚のソース参照画像image1のみを使用し、image1をキャラクター、衣装、2D手描き画風、配色、現実側のガラス温室背景の唯一の参照として扱う。\n\n【同一人物固定】\nimage1と完全に同じ一人の少女を全編で維持する。小さく繊細な同じ顔立ち、大きな青緑色の宝石のような瞳、同じ目の形、多層の虹彩と白いキャッチライト、淡い頬、細いまつ毛、真珠のような白銀髪、柔らかな長いウェーブ、編み込みを含むまとめ髪、淡い金色のリボン、青い雫形の髪飾りを固定する。表情変化でも目、眉、まぶた、口、視線の順を守る。白い長袖のハイネックドレス、淡い水色の刺繍と縁取り、淡い金色の腰リボン、華奢で上品な体格、白・水色・淡い金色の人物固有配色を変えない。現実の温室と不思議の国に同時に複数の少女を存在させず、光学トランジションの前後で同じ一人の少女として連続させる。変えてよいのは表情、視線、呼吸、歩行、手の動き、髪と袖とスカート裾の自然な揺れだけ。別人化、顔の平均化、衣装変更、装飾欠落、分身、余分な人物、既存作品を想起させる有名キャラクターを出さない。\n【2D hand-drawn／2D手描き画風固定】\nimage1の細く淡い青灰色と暖灰色の線、透明感のある二層以上の色影、柔らかなグラデーション、白・ミルキーブルー・澄んだ空色・琥珀色・淡い金色の明るい配色を全編で維持する。白銀髪の真珠質ハイライト、瞳の多層反射、肌の淡い発光感、ドレスの皺と刺繍、花、葉、ガラス、白い家具、床タイル、紅茶まで高密度の手描き筆致で描く。布は柔らかく、ガラスは透明で硬質、紅茶は澄んだ琥珀色、異世界の道と建築は半透明の白いガラス質として素材を描き分ける。太い輪郭にしない。簡略TVアニメにしない。平坦な単層セル影にしない。低密度背景にしない。滑らかなCG・3Dにしない。半写実にしない。実写にしない。くすんだ色や画風混合にしない。\n\n【舞台と小物固定】\n現実側はimage1と同じ明るいガラス温室。白い窓枠と屋根骨格、青空と白い雲、外の緑と白い花、白い丸テーブル、白い金属椅子、淡い青灰色の床タイルを維持する。開始時はimage1の少女の顔、目、髪、衣装、温室背景、白・水色・淡い金色の配色、明るさ、柔らかなコントラストを維持し、追加される小物はティーポットだけ。小物の所有者はimage1の少女一人だけ。透明なガラスのティーポット一つ、透明なティーカップ一つ、透明なソーサー一つだけを使う。開始状態ではティーカップはソーサー中央に置かれ、底に浅い琥珀色の紅茶が入っている。少女は右手でティーポットの取っ手を持ち、左手はソーサーの横へ軽く添える。ティーポット、カップ、ソーサーを増殖、交換、変形させない。不思議の国ではこれらを人物や巨大な小道具として複製しない。\n少女の表情は、目→眉→まぶた→口→視線の順に変化させる。まず左右の目の中にある小さなキャッチライトがわずかに明るくなり、次に眉が少し上がり、その後まぶたが柔らかく開き、口元に小さな驚きと微笑みが生まれ、最後に視線が渦の中心へ下がる。反射線、窓枠、光点を顔中央へ重ねず、顔と両目を常に明瞭に保つ。\n\n最初のカメラ高は白い丸テーブルと同じ低い位置。少女の斜め前から、手前にティーカップとソーサーを大きく、奥に少女の胸上と顔を明瞭に置く。背景消失軸は床タイルと右側の窓枠が温室奥へ斜めに伸びる方向へ揃える。この構図の目的は、少女、ポット先端、カップ中央の着水点を最初の一画で同時に読ませること。\n少女が右手首だけをゆっくり傾けると、ティーポットの先端から一筋だけの琥珀色の紅茶がカップ中央へ落ちる。ポット先端、液体の筋、着水点を一画で読み取れるようにし、紅茶はテーブル外へこぼれない。この着水点を変化の起点とする。着水点から一つの時計回りの渦が生まれ、青空の反射と淡い金色の日差しを巻き込みながら、琥珀、青、白の三色の螺旋へ育つ。\n\nカメラはテーブルと同じ低い位置からカップの斜め上へ滑らかに上がり、人物サイズを胸上の中景から背景の顔が読める近景へ保ったまま、紅茶面へ近づく。背景消失軸は温室奥から円形のカップ中央へ収束させる。この移動の目的は、現実の着水点を異世界への案内線へ変えること。\n\n螺旋の半分がまだ琥珀色の紅茶、もう半分が細い光の道になっている形成中の中間状態を短く明瞭に見せる。続いて螺旋が画面全体を満たした瞬間、その同じ回転方向、曲率、色の並びを保ったまま、一本の光の道へ完成する。暗転、瞬間移動、破裂、液体の中を溺れる表現にせず、紅茶面から異世界の俯瞰へ連続する光学的な形状一致として接続する。\n\n変化後の完成状態である不思議の国は、傾いた白いガラスの庭道、空中でゆっくり反転する白い花壇、輪のように連なる温室アーチ、上下が穏やかに入れ替わる空色の庭で構成する。文字盤、数字、看板、動物、人型住民は出さない。\n\nカメラ高は少女の胸より少し低い位置へ移り、同じ少女の斜め前を後退しながら、胸から膝までの中景で追従する。背景消失軸は琥珀と青の光の道が円形アーチの奥へ伸びる方向へ揃える。この移動の目的は、紅茶の渦だった曲線が少女を導く冒険路へ変わったことを読ませること。\n\n少女は光の道に沿って軽やかに三歩進み、第一歩で傾いた庭道へ乗り、第二歩で浮かぶ白い床片を渡り、第三歩で円形のガラスアーチをくぐる。各足裏を順に接地させ、髪、袖、スカート裾は歩みより少し遅れて同じ風向きへ揺れる。滑走、飛行、急回転に見せない。\n\n少女がアーチをくぐると、周囲の白い花が外側から順に淡い水色へ染まり、頭上のガラス骨格がゆっくり四分の一回転して上下の庭をつなぐ。少女自身は直立を保ち、重力方向を急変させない。\n\nカメラは胸より少し低い位置を保って少女の肩越しへ短く回り込み、人物サイズを胸上に寄せる。背景消失軸は前方の光の道が作る一つの円へ収束させる。この移動の目的は、少女の冒険の先に帰還路が生まれることを示すこと。光の道は再び一つの時計回りの螺旋となり、その円周が現実のティーカップの縁と同じ形、大きさ、角度へ整列する。\n\nカメラは螺旋の中心を通り、琥珀色の紅茶面の極近景から現実の温室へ滑らかに引く。カメラ高はテーブルと同じ低い位置へ戻り、人物サイズを奥の胸上へ戻す。背景消失軸は円形のカップ中央から床タイルと窓枠が温室奥へ伸びる開始方向へ戻す。この移動の目的は、異世界の冒険を一杯の紅茶の中へ回収し、少女の感情へ報酬を返すこと。同じ少女が帰還した連続性を保つ。\n\n終端では少女が右手のティーポットを垂直へ戻し、紅茶の筋をきれいに止め、ティーポットを胸より低い高さで保持する。左手はソーサー横に添えたまま。カップ内の最後の渦がゆっくり縮み、中心に青と淡い金の小さな反射一つだけを残して静止する。\n\nカメラはテーブル高さの斜め前で、手前の静かな紅茶面、奥の少女の三分の四顔と穏やかな微笑みを同時に読ませる。紅茶面の反射が完全に止まり、少女の両目、ティーポット、カップ、渦の名残が一画で明瞭になった瞬間に切る。\n\n液体を黒く濁らせない。渦を黒い穴、荒れた水面、別形状の物体へ変えない。顔を歪ませない。手指を増やさない。余分な人物や余分な食器を出さない。ストロボ、激しい明滅、強い水平フレア、白飛びなし。文字、字幕、ロゴ、透かしなし。BGMなし、音楽なし、環境音なし、フォーリーなし、音響生成なし。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2082798959014064531",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/haruuraeadss/status/2082798959014064531",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-07-30T12:02:23.019Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082798959014064531.mp4",
    "posterUrl": "/posters/x/2082798959014064531.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "不思議な紅茶\n@Hailuo_AI\n#MiniMaxH3　\n\nプロンプト👇\n                                                                                                                                                                                                                                            \n高品質アニメ映像。\n\n作品トーンと世界観は、透明感のある夏のガラス温室から、紅茶の渦の内側に存在するオリジナルの小さな不思議の国へ連続する、上品で夢幻的な叙情ファンタジー。澄んだ白、水色、琥珀色、淡い金色を主役にし、怖さや混沌ではなく、好奇心、浮遊感、静かな高揚を描く。今回は1枚のソース参照画像image1のみを使用し、image1をキャラクター、衣装、2D手描き画風、配色、現実側のガラス温室背景の唯一の参照として扱う。\n\n【同一人物固定】\nimage1と完全に同じ一人の少女を全編で維持する。小さく繊細な同じ顔立ち、大きな青緑色の宝石のような瞳、同じ目の形、多層の虹彩と白いキャッチライト、淡い頬、細いまつ毛、真珠のような白銀髪、柔らかな長いウェーブ、編み込みを含むまとめ髪、淡い金色のリボン、青い雫形の髪飾りを固定する。表情変化でも目、眉、まぶた、口、視線の順を守る。白い長袖のハイネックドレス、淡い水色の刺繍と縁取り、淡い金色の腰リボン、華奢で上品な体格、白・水色・淡い金色の人物固有配色を変えない。現実の温室と不思議の国に同時に複数の少女を存在させず、光学トランジションの前後で同じ一人の少女として連続させる。変えてよいのは表情、視線、呼吸、歩行、手の動き、髪と袖とスカート裾の自然な揺れだけ。別人化、顔の平均化、衣装変更、装飾欠落、分身、余分な人物、既存作品を想起させる有名キャラクターを出さない。\n【2D hand-drawn／2D手描き画風固定】\nimage1の細く淡い青灰色と暖灰色の線、透明感のある二層以上の色影、柔らかなグラデーション、白・ミルキーブルー・澄んだ空色・琥珀色・淡い金色の明るい配色を全編で維持する。白銀髪の真珠質ハイライト、瞳の多層反射、肌の淡い発光感、ドレスの皺と刺繍、花、葉、ガラス、白い家具、床タイル、紅茶まで高密度の手描き筆致で描く。布は柔らかく、ガラスは透明で硬質、紅茶は澄んだ琥珀色、異世界の道と建築は半透明の白いガラス質として素材を描き分ける。太い輪郭にしない。簡略TVアニメにしない。平坦な単層セル影にしない。低密度背景にしない。滑らかなCG・3Dにしない。半写実にしない。実写にしない。くすんだ色や画風混合にしない。\n\n【舞台と小物固定】\n現実側はimage1と同じ明るいガラス温室。白い窓枠と屋根骨格、青空と白い雲、外の緑と白い花、白い丸テーブル、白い金属椅子、淡い青灰色の床タイルを維持する。開始時はimage1の少女の顔、目、髪、衣装、温室背景、白・水色・淡い金色の配色、明るさ、柔らかなコントラストを維持し、追加される小物はティーポットだけ。小物の所有者はimage1の少女一人だけ。透明なガラスのティーポット一つ、透明なティーカップ一つ、透明なソーサー一つだけを使う。開始状態ではティーカップはソーサー中央に置かれ、底に浅い琥珀色の紅茶が入っている。少女は右手でティーポットの取っ手を持ち、左手はソーサーの横へ軽く添える。ティーポット、カップ、ソーサーを増殖、交換、変形させない。不思議の国ではこれらを人物や巨大な小道具として複製しない。\n少女の表情は、目→眉→まぶた→口→視線の順に変化させる。まず左右の目の中にある小さなキャッチライトがわずかに明るくなり、次に眉が少し上がり、その後まぶたが柔らかく開き、口元に小さな驚きと微笑みが生まれ、最後に視線が渦の中心へ下がる。反射線、窓枠、光点を顔中央へ重ねず、顔と両目を常に明瞭に保つ。\n\n最初のカメラ高は白い丸テーブルと同じ低い位置。少女の斜め前から、手前にティーカップとソーサーを大きく、奥に少女の胸上と顔を明瞭に置く。背景消失軸は床タイルと右側の窓枠が温室奥へ斜めに伸びる方向へ揃える。この構図の目的は、少女、ポット先端、カップ中央の着水点を最初の一画で同時に読ませること。\n少女が右手首だけをゆっくり傾けると、ティーポットの先端から一筋だけの琥珀色の紅茶がカップ中央へ落ちる。ポット先端、液体の筋、着水点を一画で読み取れるようにし、紅茶はテーブル外へこぼれない。この着水点を変化の起点とする。着水点から一つの時計回りの渦が生まれ、青空の反射と淡い金色の日差しを巻き込みながら、琥珀、青、白の三色の螺旋へ育つ。\n\nカメラはテーブルと同じ低い位置からカップの斜め上へ滑らかに上がり、人物サイズを胸上の中景から背景の顔が読める近景へ保ったまま、紅茶面へ近づく。背景消失軸は温室奥から円形のカップ中央へ収束させる。この移動の目的は、現実の着水点を異世界への案内線へ変えること。\n\n螺旋の半分がまだ琥珀色の紅茶、もう半分が細い光の道になっている形成中の中間状態を短く明瞭に見せる。続いて螺旋が画面全体を満たした瞬間、その同じ回転方向、曲率、色の並びを保ったまま、一本の光の道へ完成する。暗転、瞬間移動、破裂、液体の中を溺れる表現にせず、紅茶面から異世界の俯瞰へ連続する光学的な形状一致として接続する。\n\n変化後の完成状態である不思議の国は、傾いた白いガラスの庭道、空中でゆっくり反転する白い花壇、輪のように連なる温室アーチ、上下が穏やかに入れ替わる空色の庭で構成する。文字盤、数字、看板、動物、人型住民は出さない。\n\nカメラ高は少女の胸より少し低い位置へ移り、同じ少女の斜め前を後退しながら、胸から膝までの中景で追従する。背景消失軸は琥珀と青の光の道が円形アーチの奥へ伸びる方向へ揃える。この移動の目的は、紅茶の渦だった曲線が少女を導く冒険路へ変わったことを読ませること。\n\n少女は光の道に沿って軽やかに三歩進み、第一歩で傾いた庭道へ乗り、第二歩で浮かぶ白い床片を渡り、第三歩で円形のガラスアーチをくぐる。各足裏を順に接地させ、髪、袖、スカート裾は歩みより少し遅れて同じ風向きへ揺れる。滑走、飛行、急回転に見せない。\n\n少女がアーチをくぐると、周囲の白い花が外側から順に淡い水色へ染まり、頭上のガラス骨格がゆっくり四分の一回転して上下の庭をつなぐ。少女自身は直立を保ち、重力方向を急変させない。\n\nカメラは胸より少し低い位置を保って少女の肩越しへ短く回り込み、人物サイズを胸上に寄せる。背景消失軸は前方の光の道が作る一つの円へ収束させる。この移動の目的は、少女の冒険の先に帰還路が生まれることを示すこと。光の道は再び一つの時計回りの螺旋となり、その円周が現実のティーカップの縁と同じ形、大きさ、角度へ整列する。\n\nカメラは螺旋の中心を通り、琥珀色の紅茶面の極近景から現実の温室へ滑らかに引く。カメラ高はテーブルと同じ低い位置へ戻り、人物サイズを奥の胸上へ戻す。背景消失軸は円形のカップ中央から床タイルと窓枠が温室奥へ伸びる開始方向へ戻す。この移動の目的は、異世界の冒険を一杯の紅茶の中へ回収し、少女の感情へ報酬を返すこと。同じ少女が帰還した連続性を保つ。\n\n終端では少女が右手のティーポットを垂直へ戻し、紅茶の筋をきれいに止め、ティーポットを胸より低い高さで保持する。左手はソーサー横に添えたまま。カップ内の最後の渦がゆっくり縮み、中心に青と淡い金の小さな反射一つだけを残して静止する。\n\nカメラはテーブル高さの斜め前で、手前の静かな紅茶面、奥の少女の三分の四顔と穏やかな微笑みを同時に読ませる。紅茶面の反射が完全に止まり、少女の両目、ティーポット、カップ、渦の名残が一画で明瞭になった瞬間に切る。\n\n液体を黒く濁らせない。渦を黒い穴、荒れた水面、別形状の物体へ変えない。顔を歪ませない。手指を増やさない。余分な人物や余分な食器を出さない。ストロボ、激しい明滅、強い水平フレア、白飛びなし。文字、字幕、ロゴ、透かしなし。BGMなし、音楽なし、環境音なし、フォーリーなし、音響生成なし。",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082798559901143224",
    "title": "Image Model: GPT Image 2",
    "titleEn": "Image Model: GPT Image 2",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《Image Model: GPT Image 2》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Image Model: GPT Image 2. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Claude\n\nInteractive Game Equipment UI (15s)\n\nUse Image 1 as a STRICT CHARACTER REFERENCE. Preserve the character exactly: oversized yellow hooded jacket, glossy black faceless helmet, black armored tactical torso, tattooed left arm, glossy red tactical cargo pants with hanging straps and pouches, black combat boots, and the massive crimson-black biomechanical right arm with organic tendrils and clawed fingers. Do not alter colors, proportions, outfit, or mechanical design.\n0–2 Seconds\n\nHigh-angle overhead cinematic shot. The character sits on a vivid, highly saturated purple floor with one knee raised and the enormous crimson biomechanical arm resting beside him. He slowly tilts his head upward toward the camera. A premium AAA futuristic game interface fades in.\n\nTop left:\n Player Profile\n REAPER\n Level 42\nRight side menu:\nSTART NEW GAME\nCONTINUE (highlighted)\n\nLOADOUT\n\nSETTINGS\nEXIT GAME\nA sleek white cursor smoothly clicks CONTINUE.\n2–4 Seconds\nCamera pushes in toward the biomechanical right arm.\n\nA translucent RIGHT ARM EQUIPMENT panel slides in.\n\nEquipment list:\n\nPHANTOM GRIP\nABYSS CLAW\nHELLFORGE TALON\n\nCRIMSON REAVER\nThe selector moves to ABYSS CLAW.\nThe right arm begins transforming.\nRed-black armor plates separate.\n\nOrganic crimson tendrils weave together.\n\nMechanical pistons rotate.\n\nClaw fingers extend into longer razor blades.\n\nInternal red energy pulses through exposed cables.\n\nSmall particles and sparks emit as the transformation locks into place.\n\n4–7 Seconds\n\nCamera smoothly arcs around the character's right side.\n\nA large ARMAMENT CUSTOMIZATION interface appears.\n\nComponent grid:\n\nShoulder\n\nUpper Arm\n\nElbow\n\nForearm\nWrist\nClaw System\nCore Reactor\nTendril Matrix\nSelections cycle rapidly.\nThe biomechanical arm partially disassembles.\nArmor segments detach.\n\nInternal skeletal frame becomes visible.\n\nHydraulic pistons rotate.\n\nMechanical tendrils retract and regrow.\n\nNew crimson armor slides into place.\n\nClaw geometry reshapes into a more aggressive configuration before locking with a satisfying mechanical snap.\n\n7–8.5 Seconds\n\nCamera pulls back to a medium shot.\n\nA large glowing CONFIRM CONFIGURATION button appears.\n\nCursor clicks.\n\nEvery interface panel folds inward using holographic animations and disappears.\n\nThe character rises from the seated position.\n\nHe slowly flexes the upgraded biomechanical arm.\n\nThe massive claws open and close once.\n\nCrimson energy briefly surges through the entire limb.\n\n8.5–10 Seconds\n\nA cinematic LOADING bar appears across the bottom.\n\nProgress accelerates from 0% → 100%.\n\nThe purple environment gradually fades into darkness.\n\nGolden light leaks through the scene.\n\nSubtle volumetric fog forms.\n\nAmbient audio transitions from UI sounds to a living city.\n\n10–15 Seconds\n\nThe world fully loads.\n\nThe character now stands inside a dense cyberpunk megacity.\n\nRain-soaked streets.\n\nNeon holographic advertisements.\n\nSteam rising from vents.\n\nMoving crowds.\n\nHover drones patrol overhead.\n\nMotorcycles race past.\n\nTowering futuristic skyscrapers loom in the distance.\n\nCamera settles into a polished third-person gameplay perspective behind the character.\n\nHUD fades in:\n\nMinimap (top right)\n\nHealth & Energy (bottom left)\n\nAbility Icons\n\nWeapon Indicator\nObjective Marker\nThe character activates the crimson biomechanical arm, flexes the claws once, then confidently walks into the neon-lit street as the mission begins.\nStyle: AAA Unreal Engine 5 cinematic, photorealistic PBR materials, production-quality game UI, premium sci-fi HUD, seamless mechanical transformation, cinematic camera motion, ray-traced lighting, volumetric fog, rain reflections, 4K, 60fps, no glitches, no UI clipping, no extra limbs, perfect character consistency, movie-grade color grading.",
    "promptSourceUrl": "https://x.com/ShamsAmin56/status/2082798564032282747",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ShamsAmin56/status/2082798559901143224",
    "sourceLabel": "X 原帖 · @ShamsAmin56",
    "author": "Shams",
    "publishedAt": "2026-07-30T12:00:47.863Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082798559901143224.mp4",
    "posterUrl": "/posters/x/2082798559901143224.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Got early access to MiniMax H3 from\n@Hailuo_AI, Designed a prototype for a game UI \"Reaper\"\n\nThis is ultimately game changing \n\nVideo Model: MiniMax H3\nImage Model: GPT Image 2\n#MiniMaxH3\n\nReference Media and Prompt in replies",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082794108998078815",
    "title": "喷气背包俯冲山峡",
    "titleEn": "Jetpack Dive Through a Mountain Gorge",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《喷气背包俯冲山峡》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Jetpack Dive Through a Mountain Gorge. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Ultra-fast paced cinematic action sequence, a man in a sleek black tactical suit with advanced winged jetpack on his back, dramatic high-altitude jump from a towering cliff edge, immediately diving headfirst at extreme speed down a deep rocky mountain gorge with jagged cliffs blurring past, intense free-fall with wind and dust whipping around him, at the very last second he suddenly activates powerful rocket launcher boosters, bright orange flames and thrust exploding from the jetpack, sharp high-speed swap and barrel roll to the side avoiding the rocky wall by inches, then rocketing forward aggressively straight towards an enemy vehicle ahead, eyes focused and intense, dynamic motion blur, dramatic golden hour lighting, highly detailed, epic wide cinematic angle, over-the-top action style, intense energy, 8k",
    "promptSourceUrl": "https://x.com/Ankit_patel211/status/2082794111384580244",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Ankit_patel211/status/2082794108998078815",
    "sourceLabel": "X 原帖 · @Ankit_patel211",
    "author": "ANKIT PATEL 🇮🇳 | AI",
    "publishedAt": "2026-07-30T11:43:06.685Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082794108998078815.mp4",
    "posterUrl": "/posters/x/2082794108998078815.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Action & VFX",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "My first test with early access to @Hailuo_AI MiniMax H3. \n\nPrompt shared below 👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082790356983447606",
    "title": "雨夜韩国黑色电影重逢",
    "titleEn": "Rain-Soaked Korean Noir Reunion",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《雨夜韩国黑色电影重逢》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 19 秒 · 1280×720。",
    "summaryEn": "An H3 output for Rain-Soaked Korean Noir Reunion. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 19s · 1280×720.",
    "prompt": "16:9, 15s, hyper-realistic Korean noir crime teaser.\n\nA rain-soaked Korean woman (late 20s, trench coat) enters an abandoned underground nightclub searching for her missing sister. A scarred crime boss (40s), presumed dead for years, emerges from the shadows with a lit cigarette. Atmosphere: 90s Korean noir, practical neon, rain, smoke, film grain, tactile realism, no CGI gloss.\n\nShots:\n\n1. She enters: \"언니... 여기 있어?\"\n2. Lighter ignites, revealing his scar. \"오랜만이네.\"\n3. She freezes: \"당신... 죽은 줄 알았는데.\"\n4. Slow push-in. He smiles: \"네 언니가 날 죽였다고 생각했겠지.\"\n5. They lock eyes. Thunder. Cut to black.\n\nAudio: Rain, thunder, jazz crackle, lighter click, intimate silence during dialogue.",
    "promptSourceUrl": "https://x.com/doctorwasif/status/2082790356983447606",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/doctorwasif/status/2082790356983447606",
    "sourceLabel": "X 原帖 · @doctorwasif",
    "author": "WasifAI",
    "publishedAt": "2026-07-30T11:28:12.135Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082790356983447606.mp4",
    "posterUrl": "/posters/x/2082790356983447606.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Built this Korean crime-thriller teaser with MiniMax H3 @Hailuo_AI and the level of control is on another level.\n\nFrom cinematic noir to nuanced dialogue, H3 understands complex creative intent across text, visuals, motion, and audio. \n\nPrompt:\n16:9, 15s, hyper-realistic Korean noir crime teaser.\n\nA rain-soaked Korean woman (late 20s, trench coat) enters an abandoned underground nightclub searching for her missing sister. A scarred crime boss (40s), presumed dead for years, emerges from the shadows with a lit cigarette. Atmosphere: 90s Korean noir, practical neon, rain, smoke, film grain, tactile realism, no CGI gloss.\n\nShots:\n\n1. She enters: \"언니... 여기 있어?\"\n2. Lighter ignites, revealing his scar. \"오랜만이네.\"\n3. She freezes: \"당신... 죽은 줄 알았는데.\"\n4. Slow push-in. He smiles: \"네 언니가 날 죽였다고 생각했겠지.\"\n5. They lock eyes. Thunder. Cut to black.\n\nAudio: Rain, thunder, jazz crackle, lighter click, intimate silence during dialogue. #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082783319075291312",
    "title": "奢华无线耳机产品演示",
    "titleEn": "Luxury wireless headphones product collection",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《奢华无线耳机产品演示》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Luxury wireless headphones product collection. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create a 15-second luxury cinematic product showcase for premium wireless over-ear headphones.\n\n0–4s: Begin with an extreme macro tracking shot moving across the soft memory-foam ear cushion, fine fabric texture, brushed-metal hinge and precision-machined controls. A narrow light band travels across the surface, revealing realistic materials against a deep black studio background.\n\n4–8s: Pull back into a three-quarter hero view. The headphones rotate slowly above a glossy reflective pedestal. The ear cups pivot naturally while the adjustable headband extends slightly, demonstrating flexible construction and comfort. Maintain exact symmetry, stable geometry and consistent proportions.\n\n8–12s: Transition into an elegant exploded-view reveal. The ear cushion, acoustic driver, internal sound chamber, control ring and outer shell separate smoothly in perfect alignment. Subtle luminous sound waves pulse outward from the driver while the camera performs a restrained side orbit.\n\n12–15s: Every component reconnects seamlessly. The headphones settle into a centered front-facing hero composition as soft rim lighting defines the silhouette. Complete a gentle dolly-in toward the ear cups. Premium technology-commercial finish, controlled reflections, realistic shadows, shallow depth of field, crisp surface detail, stable product shape, no hands, no distortion, no onscreen text.",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2082783319075291312",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/LudovicCreator/status/2082783319075291312",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-07-30T11:00:14.167Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082783319075291312.mp4",
    "posterUrl": "/posters/x/2082783319075291312.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cinematic-realistic",
      "commercial"
    ],
    "scenes": [
      "music-performance",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "New test using Minimax H3 by Hailuo AI : Product Showcase\n\nHere an exemple text to video that you can adapt your own products\n\nPrompt : \n\nCreate a 15-second luxury cinematic product showcase for premium wireless over-ear headphones.\n\n0–4s: Begin with an extreme macro tracking shot moving across the soft memory-foam ear cushion, fine fabric texture, brushed-metal hinge and precision-machined controls. A narrow light band travels across the surface, revealing realistic materials against a deep black studio background.\n\n4–8s: Pull back into a three-quarter hero view. The headphones rotate slowly above a glossy reflective pedestal. The ear cups pivot naturally while the adjustable headband extends slightly, demonstrating flexible construction and comfort. Maintain exact symmetry, stable geometry and consistent proportions.\n\n8–12s: Transition into an elegant exploded-view reveal. The ear cushion, acoustic driver, internal sound chamber, control ring and outer shell separate smoothly in perfect alignment. Subtle luminous sound waves pulse outward from the driver while the camera performs a restrained side orbit.\n\n12–15s: Every component reconnects seamlessly. The headphones settle into a centered front-facing hero composition as soft rim lighting defines the silhouette. Complete a gentle dolly-in toward the ear cups. Premium technology-commercial finish, controlled reflections, realistic shadows, shallow depth of field, crisp surface detail, stable product shape, no hands, no distortion, no onscreen text.\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082783123830378938",
    "title": "竖屏改装车街头文化短片",
    "titleEn": "Vertical car transformation street-culture film",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《竖屏改装车街头文化短片》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Vertical car transformation street-culture film. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "FORMAT:\n15-second cinematic vertical car transformation film\n9:16 aspect ratio\nUltra-realistic automotive commercial aesthetic\nJapanese street culture + luxury cinematic realism\nHigh-energy transformation pacing\n\nCONCEPT:\n“FROM STOCK TO STREET WEAPON”\n\nA normal Maruti Suzuki Swift slowly transforms into a fully modified street machine through cinematic garage sequences, mechanical close-ups, aggressive lighting, and dramatic reveal shots.\n\nTIMELINE:\n\n0:00–0:02\nOld stock white Maruti Suzuki Swift parked inside a dim industrial garage.\nDust particles in sunlight.\nStatic realistic atmosphere.\nSoft engine ticking sounds.\n\n0:02–0:04\nQuick cinematic close-ups:\ntools dropping on metal table,\ngarage lights turning on,\nmechanic gloves tightening,\ncar jack lifting vehicle.\nFast-paced editing begins.\n\n0:04–0:06\nTransformation montage:\nnew alloy wheels installed,\nwide tires mounted,\nlowered suspension,\ncustom front bumper attached,\nblack spoiler fitted.\nMechanical sparks and cinematic motion blur.\n\n0:06–0:08\nPaint transformation sequence.\nOriginal white paint peels away into glossy metallic midnight black.\nWater droplets and reflections across body panels.\nNeon garage lighting reflects on paint.\n\n0:08–0:10\nInterior upgrade shots:\ndigital dashboard glowing,\nsport steering wheel,\nbucket seats,\nambient red lighting,\npush-start ignition.\nEngine rev echoes.\n\n0:10–0:12\nAggressive cinematic reveal:\nmodified Swift exits garage slowly.\nHeadlights turn on through smoke.\nLow-angle tracking shots.\nWet asphalt reflections.\nTurbo flutter sound.\n\n0:12–0:15\nFinal hero sequence:\ncar speeding through empty city streets at night,\nrolling shots,\ndrifting corner,\nrain reflections,\nmotion blur,\ncamera orbit around vehicle.\nEnd with parked hero shot under neon lights.\n\nFINAL TEXT:\n“SWIFT — REBORN.”\n\nVISUAL STYLE:\nFast & Furious-inspired realism\nMoody cinematic lighting\nJapanese underground street-racing atmosphere\nWet asphalt reflections\nPractical garage lighting\nHigh-detail metallic textures\nSubtle film grain\nDeep contrast shadows\nRealistic smoke and sparks\n\nCAMERA:\nARRI Alexa Mini LF aesthetic\nAnamorphic lens look\nLow-angle tracking shots\nWheel close-ups\nSlow-motion mechanical shots\nDrone-style orbit reveal\nHandheld garage realism\n\nSOUND DESIGN:\nMetal tool impacts\nHydraulic jack sounds\nEngine revs\nTurbo flutter\nDeep bass transitions\nTire screeches\nCinematic trailer hits\nAggressive phonk-inspired background energy\n\nQUALITY:\nMasterpiece-level automotive realism\nLuxury commercial production quality\nPinterest-saveable cinematic car edit\nUltra-detailed reflections and textures\nFeels like a premium car modification advertisement",
    "promptSourceUrl": "https://x.com/fromjtown/status/2082783476466450861",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/fromjtown/status/2082783123830378938",
    "sourceLabel": "X 原帖 · @fromjtown",
    "author": "JTOWN 🕷️",
    "publishedAt": "2026-07-30T10:59:27.617Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082783123830378938.mp4",
    "posterUrl": "/posters/x/2082783123830378938.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Got an Early access to test the new @Hailuo_AI #MiniMaxH3 it’s indeed amazing!\n\nPrompt below 👇🏾",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082782535252410508",
    "title": "科幻探险者逼近黑暗之门",
    "titleEn": "Sci-fi explorer approaches dark gateway mystery",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《科幻探险者逼近黑暗之门》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Sci-fi explorer approaches dark gateway mystery. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create a 15-second, 16:9 photoreal cinematic sci-fi mystery sequence with native stereo audio.\n\nUse Image 1 as the strict anchor reference for the explorer’s identity, face, short dark hair, charcoal hooded coat, black tactical clothing, boots, gateway design, wet obsidian landscape, jagged rock formations, lighting, materials, and cold blue/crimson color palette. Use Images 2–4 as sequential visual keyframes and preserve complete character, costume, environment, scale, and screen-direction continuity.\n\n[0–3 seconds] Begin with Image 2. Ultra-wide symmetrical establishing shot from behind the explorer. She walks slowly toward the colossal circular gateway, remaining very small against its immense scale. The camera performs a restrained forward dolly at ground level, following the crimson reflection across the wet surface. Her boots create small splashes. Mist drifts laterally around the distant rocks. The glowing ring pulses faintly, but its center remains completely black.\n\n[3–6 seconds] Cut on her next footstep to a low rear three-quarter tracking shot. Move beside her at walking speed while keeping the gateway fixed ahead. Her coat responds naturally to the wind. The crimson illumination travels across the wet fabric and reflective ground. Do not change her walking direction or position relative to the gateway.\n\n[6–9 seconds] Cut to Image 3. Tight three-quarter facial close-up. Slowly push toward her eyes as she stops walking and looks upward. Preserve her exact facial identity. Her expression shifts subtly from focused caution to disturbed recognition. Fine strands of hair move across her face. The gateway remains softly recognizable behind her. A dim crimson reflection appears in her eyes.\n\n[9–12 seconds] Cut to a close shot of her gloved hand rising cautiously into frame. Track forward with the hand as she approaches the threshold. The air near the black center distorts slightly, pulling mist and loose droplets toward it. Keep the effect restrained and physically coherent, with no lightning, explosions, or fantasy energy.\n\n[12–15 seconds] Transition into Image 4. Medium-wide side view near the gateway. She extends her hand toward the pitch-black center but stops just before contact. The crimson ring suddenly dims for half a second. All ambient wind and rain sound disappear. From inside the void, a quiet version of her own breathing answers one beat later. She freezes. End on a slow push toward the narrow space between her fingertips and the darkness, cutting to black before contact.\n\nVisual style: premium live-action science-fiction film, realistic skin and fabric, cold blue ambient light, restrained crimson rim light, deep blacks, wet reflections, fine atmospheric mist, subtle film grain, controlled contrast, 24 FPS, natural motion blur, cinematic depth of field.\n\nAudio: distant low-frequency gateway resonance, cold wind, light rain, boots splashing through shallow water, fabric movement, restrained breathing, and a deep sub-bass pulse that grows gradually. Remove all environmental sound during the final silence, leaving only the delayed breathing from inside the void.\n\nNo dialogue, subtitles, title cards, logos, watermarks, extra characters, vehicles, weapons, creatures, lightning, explosions, camera spins, random zooms, rapid montage, costume changes, facial changes, distorted hands, warped gateway geometry, daylight, or bright saturated colors.",
    "promptSourceUrl": "https://x.com/beginnersblog1/status/2082782535252410508",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/beginnersblog1/status/2082782535252410508",
    "sourceLabel": "X 原帖 · @beginnersblog1",
    "author": "Beginnersblog",
    "publishedAt": "2026-07-30T10:57:07.289Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082782535252410508.mp4",
    "posterUrl": "/posters/x/2082782535252410508.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "motion-graphics"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tried Hailuo Minimax H3\n\nUse this prompt \"Create a 15-second, 16:9 photoreal cinematic sci-fi mystery sequence with native stereo audio.\n\nUse Image 1 as the strict anchor reference for the explorer’s identity, face, short dark hair, charcoal hooded coat, black tactical clothing, boots, gateway design, wet obsidian landscape, jagged rock formations, lighting, materials, and cold blue/crimson color palette. Use Images 2–4 as sequential visual keyframes and preserve complete character, costume, environment, scale, and screen-direction continuity.\n\n[0–3 seconds] Begin with Image 2. Ultra-wide symmetrical establishing shot from behind the explorer. She walks slowly toward the colossal circular gateway, remaining very small against its immense scale. The camera performs a restrained forward dolly at ground level, following the crimson reflection across the wet surface. Her boots create small splashes. Mist drifts laterally around the distant rocks. The glowing ring pulses faintly, but its center remains completely black.\n\n[3–6 seconds] Cut on her next footstep to a low rear three-quarter tracking shot. Move beside her at walking speed while keeping the gateway fixed ahead. Her coat responds naturally to the wind. The crimson illumination travels across the wet fabric and reflective ground. Do not change her walking direction or position relative to the gateway.\n\n[6–9 seconds] Cut to Image 3. Tight three-quarter facial close-up. Slowly push toward her eyes as she stops walking and looks upward. Preserve her exact facial identity. Her expression shifts subtly from focused caution to disturbed recognition. Fine strands of hair move across her face. The gateway remains softly recognizable behind her. A dim crimson reflection appears in her eyes.\n\n[9–12 seconds] Cut to a close shot of her gloved hand rising cautiously into frame. Track forward with the hand as she approaches the threshold. The air near the black center distorts slightly, pulling mist and loose droplets toward it. Keep the effect restrained and physically coherent, with no lightning, explosions, or fantasy energy.\n\n[12–15 seconds] Transition into Image 4. Medium-wide side view near the gateway. She extends her hand toward the pitch-black center but stops just before contact. The crimson ring suddenly dims for half a second. All ambient wind and rain sound disappear. From inside the void, a quiet version of her own breathing answers one beat later. She freezes. End on a slow push toward the narrow space between her fingertips and the darkness, cutting to black before contact.\n\nVisual style: premium live-action science-fiction film, realistic skin and fabric, cold blue ambient light, restrained crimson rim light, deep blacks, wet reflections, fine atmospheric mist, subtle film grain, controlled contrast, 24 FPS, natural motion blur, cinematic depth of field.\n\nAudio: distant low-frequency gateway resonance, cold wind, light rain, boots splashing through shallow water, fabric movement, restrained breathing, and a deep sub-bass pulse that grows gradually. Remove all environmental sound during the final silence, leaving only the delayed breathing from inside the void.\n\nNo dialogue, subtitles, title cards, logos, watermarks, extra characters, vehicles, weapons, creatures, lightning, explosions, camera spins, random zooms, rapid montage, costume changes, facial changes, distorted hands, warped gateway geometry, daylight, or bright saturated colors.\"\n\n@Hailuo_AI  #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082777543824789559",
    "title": "高速运动产品落地页",
    "titleEn": "High-Speed Sports Product Landing Page",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《高速运动产品落地页》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for High-Speed Sports Product Landing Page. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create a dynamic product-landing-page UI/UX demo inspired by Nike's digital language, built around the product in Image 1. Use oversized, bold, italicized sans-serif typography and backgrounds that combine speed-driven light streaks with dark carbon fiber or breathable performance-mesh textures. Show a smooth, fast, powerful scroll through the page, plus high-impact hover interactions with scale-ups and color inversion.",
    "promptSourceUrl": "https://x.com/Voxyz_ai/status/2082777543824789559",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Voxyz_ai/status/2082777543824789559",
    "sourceLabel": "X 原帖 · @Voxyz_ai",
    "author": "Vox",
    "publishedAt": "2026-07-30T10:37:17.240Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082777543824789559.mp4",
    "posterUrl": "/posters/x/2082777543824789559.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This product page demo is a single generation, 15 seconds long. Scroll, hover states, product cards.\n\nMade entirely with MiniMax H3 from @Hailuo_AI. Omni Reference takes up to 12 references at once across video, text, image and audio, which is what keeps the look consistent.\n\nPrompt below👇\n\nCreate a dynamic product-landing-page UI/UX demo inspired by Nike's digital language, built around the product in Image 1. Use oversized, bold, italicized sans-serif typography and backgrounds that combine speed-driven light streaks with dark carbon fiber or breathable performance-mesh textures. Show a smooth, fast, powerful scroll through the page, plus high-impact hover interactions with scale-ups and color inversion.\n\n#hailuoai #hailuo #MiniMaxH3 #H3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082776032042783085",
    "title": "超高端香水电影级广告",
    "titleEn": "Ultra-premium cinematic luxury perfume commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《超高端香水电影级广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Ultra-premium cinematic luxury perfume commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create a 15-second ultra-premium cinematic luxury perfume commercial designed to fully showcase MiniMax H3's multimodal capabilities.\n\nOutput Settings:\n• Duration: 15 seconds\n• Resolution: 2K (1440p)\n• Frame Rate: 24 FPS\n• Aspect Ratio: 9:16 Vertical\n• Native Stereo Audio\n• Commercial-grade quality\n\nScene 1 (0–3s)\nExtreme macro shot of a crystal luxury perfume bottle resting on polished black marble. Golden sunrise light passes through the glass, creating realistic caustics and sparkling reflections. Tiny perfume particles float naturally in the air. Slow cinematic push-in with perfect depth of field.\n\nScene 2 (3–6s)\nA luxury fashion model wearing an elegant black dress walks confidently through a futuristic golden hallway. The camera smoothly circles around her while she gracefully picks up the perfume bottle. Hair moves naturally, realistic fabric simulation, premium fashion campaign quality.\n\nScene 3 (6–10s)\nShe sprays the perfume. The mist transforms into flowing golden particles that evolve into blooming flowers, liquid gold ribbons, shimmering butterflies, floating crystal fragments, and elegant light trails surrounding her. Every movement is physically realistic with seamless cinematic transitions.\n\nScene 4 (10–13s)\nHero beauty sequence. Slow-motion close-ups of the model's face, glowing skin, elegant eyes, glossy lips, luxury jewelry, and the perfume bottle. Beautiful bokeh, premium studio lighting, realistic reflections, cinematic lens movement.\n\nScene 5 (13–15s)\nEpic ending. The perfume bottle rotates slowly on a glossy black pedestal surrounded by floating golden particles and soft luxury lighting. The model stands elegantly in the background. Premium cinematic composition worthy of a global luxury brand campaign.\n\nVisual Style:\nLuxury brand commercial, Dior Beauty, Chanel, Tom Ford, Apple-level production quality, Hollywood cinematography, ultra-photorealistic, ray tracing, global illumination, volumetric lighting, realistic reflections, premium color grading, cinematic depth of field, physically accurate materials, ultra-detailed textures, smooth camera movement, flawless facial animation, natural body motion, production-ready advertising quality.\n\nNegative Prompt:\nLow quality, blurry, noise, flickering, unstable motion, distorted anatomy, bad hands, bad face, duplicate objects, watermark, logo, text overlay, compression artifacts, oversaturated colors, unrealistic physics, low-detail textures.",
    "promptSourceUrl": "https://x.com/SadiaMalik182/status/2082776032042783085",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/SadiaMalik182/status/2082776032042783085",
    "sourceLabel": "X 原帖 · @SadiaMalik182",
    "author": "Sadia",
    "publishedAt": "2026-07-30T10:31:16.803Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082776032042783085.mp4",
    "posterUrl": "/posters/x/2082776032042783085.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Luxury fragrance, cinematic visuals, and premium storytelling\n\nall generated with a single prompt\nCreated using MiniMax H3 on @Hailuo_AI. \n\nThe quality, lighting, and realism are seriously impressive.🔥\n\n#MiniMaxH3 @Hailuo_AI \n\nPrompt 👇 \nCreate a 15-second ultra-premium cinematic luxury perfume commercial designed to fully showcase MiniMax H3's multimodal capabilities.\n\nOutput Settings:\n• Duration: 15 seconds\n• Resolution: 2K (1440p)\n• Frame Rate: 24 FPS\n• Aspect Ratio: 9:16 Vertical\n• Native Stereo Audio\n• Commercial-grade quality\n\nScene 1 (0–3s)\nExtreme macro shot of a crystal luxury perfume bottle resting on polished black marble. Golden sunrise light passes through the glass, creating realistic caustics and sparkling reflections. Tiny perfume particles float naturally in the air. Slow cinematic push-in with perfect depth of field.\n\nScene 2 (3–6s)\nA luxury fashion model wearing an elegant black dress walks confidently through a futuristic golden hallway. The camera smoothly circles around her while she gracefully picks up the perfume bottle. Hair moves naturally, realistic fabric simulation, premium fashion campaign quality.\n\nScene 3 (6–10s)\nShe sprays the perfume. The mist transforms into flowing golden particles that evolve into blooming flowers, liquid gold ribbons, shimmering butterflies, floating crystal fragments, and elegant light trails surrounding her. Every movement is physically realistic with seamless cinematic transitions.\n\nScene 4 (10–13s)\nHero beauty sequence. Slow-motion close-ups of the model's face, glowing skin, elegant eyes, glossy lips, luxury jewelry, and the perfume bottle. Beautiful bokeh, premium studio lighting, realistic reflections, cinematic lens movement.\n\nScene 5 (13–15s)\nEpic ending. The perfume bottle rotates slowly on a glossy black pedestal surrounded by floating golden particles and soft luxury lighting. The model stands elegantly in the background. Premium cinematic composition worthy of a global luxury brand campaign.\n\nVisual Style:\nLuxury brand commercial, Dior Beauty, Chanel, Tom Ford, Apple-level production quality, Hollywood cinematography, ultra-photorealistic, ray tracing, global illumination, volumetric lighting, realistic reflections, premium color grading, cinematic depth of field, physically accurate materials, ultra-detailed textures, smooth camera movement, flawless facial animation, natural body motion, production-ready advertising quality.\n\nNegative Prompt:\nLow quality, blurry, noise, flickering, unstable motion, distorted anatomy, bad hands, bad face, duplicate objects, watermark, logo, text overlay, compression artifacts, oversaturated colors, unrealistic physics, low-detail textures.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082774662342377881",
    "title": "手持镜头双人争吵",
    "titleEn": "Handheld Two-Person Argument",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《手持镜头双人争吵》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Handheld Two-Person Argument. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Two characters arguing in English. Make the emotion feel very realistic  angry, tense, the friction between them should be palpable. Shaky handheld camera. No music.",
    "promptSourceUrl": "https://x.com/heydin_ai/status/2082774665123246479",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/heydin_ai/status/2082774662342377881",
    "sourceLabel": "X 原帖 · @heydin_ai",
    "author": "Dinda Prasetyo",
    "publishedAt": "2026-07-30T10:25:50.241Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082774662342377881.mp4",
    "posterUrl": "/posters/x/2082774662342377881.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "ugc"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Early access to @Hailuo_AI  H3 and I had to test it immediately 👀\n\nKept the first test simple  2 characters, one straightforward prompt. I just wanted to see how the model reasons through emotion and tension between characters on its own.\n\nResult: genuinely impressive. The eye contact, the anger, the mic-tension  it read the scene better than expected.\n\nPrompt's in the comments 👇\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082774526098874724",
    "title": "武士剑影无缝循环",
    "titleEn": "Seamless Warrior Sword Reflection Loop",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《武士剑影无缝循环》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 30 秒 · 1280×720。",
    "summaryEn": "An H3 output for Seamless Warrior Sword Reflection Loop. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 30s · 1280×720.",
    "prompt": "Prompt #MiniMaxH3 in Chinese:\n\n@image1\n\n15s | 16:9 | 1440p | 24fps | SEAMLESS LOOP\n\n[LOCK] Render exactly as @image1. 完全按照该参考图呈现，不进行任何重新设计。Do not alter hair, bangs, eye color, coat silhouette, sash, hilt ornament, or blade proportion.\n\n[LOOP] A perfect cycle. The final frame must match frame one exactly — same composition, focal length, lighting, blade position and pose — so playback runs endlessly with no seam. The story is a blade breaking and unbreaking forever.\n\n[PALETTE] Background: smooth vertical gradient, luminous cyan-teal #2AA6B4 top into deep petrol #0F4A55 bottom — a graphic plane, no environment, no horizon, no props. Coat bright coral #FF9273. Inner garment #14454E. Hair true black with blue-white highlights. Metal and glass bright silver-white.\nAccents as refracted light ONLY, never flat fill: gold #FFD27A, cyan #6FE3F5, magenta #FF9ED8. Core palette always dominates.\n\n[STYLE] 2.5D anime, bright cinematic. High-key, generous bloom, crisp speculars, luminous not grim. Skin luminous with subsurface warmth at ears and fingertips. Coat as heavy matte cotton twill, deep folds. Hair in clumped strands with flyaways catching the rim. Glass with real thickness — edge refraction, internal fracture lines, micro-scratches.\n\n[EFFECTS] Escalating: time-echo ghosting trailing every fast move, each afterimage a different accent; liquid mercury flow as the blade sets; kaleidoscope fracture splitting the frame into mirror symmetry; shockwave refraction warping the background and bending the type; one-frame negative invert on hard impacts; glass-crack frame wipes; reverse-gravity debris; prismatic dispersion off every shard edge casting moving caustics across her face and coat.\n\n[CAMERA + LIGHT] Anamorphic — oval bokeh, blue streak flares, slight barrel distortion wide. 85mm close, 35mm wide, 24mm on moves. 180-degree shutter. Shallow focus on hero beats, deep on montage frames. Rack focus blade-to-eyes at 2.4s and 14.2s.\nHard sun key from screen right, high and behind — bright rim on jaw, shoulder and blade edge. Broad cyan bounce fill off the background plane. Kicker thrown by the shards, hue shifting each beat. Shadows open, never crushed. Brightest at 10.4, back to the opening state by 15.0.\n\n[SIGNATURE] A fine engraved CIA0 sits on the blade's ricasso above the hilt, small letter-spaced capitals cut into steel, catching the key light as the camera races past in the opening shot. Part of the blade, not an overlay — present in frame one, in the final frame, and on the reforged blade at 14.4s.\n\n[SHARDS] Vary reflected content shard to shard: blue sky with cumulus, sunlit golden farmland, a low sun with lens flare, a fragment of her own face, pure white cloud. A few reflect a beat that has not happened yet. Never blank glass.\n\n[TYPE] Behind her on the gradient plane. Condensed sans for Latin, brush-weight for Japanese. Gradient fill, soft glow, hard offset shadow. Kinetic — slams, scrolls, shatters, warps under shockwaves. Never covers her face. Japanese runs VERTICAL in a side third; Latin STACKED, two lines max, cropped by frame edge for scale.\n0.8s 「鏡」 coral into gold\n2.2s \"A BLADE THAT BREAKS\" / \"AND UNBREAKS FOREVER\" pale cyan into white\n5.0s \"ONE IMAGE\" / \"FIFTEEN SECONDS\" white into coral\n7.2s 「反射」 diagonal prism sweep, cyan-white-magenta-gold\n10.6s \"NO CUT\" gold into white, warping under the shockwave\n13.2s title 「鏡落」 above \"MIRRORFALL\", silver into pale gold, coral glow — must fully dissolve before 15.0\n\n[SHOTS]\n0.0-1.6 LOOP POINT. Extreme close-up racing along the blade flat toward the hilt, clouds streaking on steel, flare sliding down the edge, the engraved mark passing through frame. Blade detonates. Glass-crack wipe.\n1.6-3.2 BULLET-TIME ORBIT. Time freezes on the detonation; camera arcs 180 degrees through the suspended shard field, threading between fragments, debris hanging in reverse gravity. Motion resumes as the arc completes.\n\n@Image1 3.2-4.8 TOP-DOWN CRANE. Bird's eye directly above; shards lock edge-on into a ring around her, then the camera drops fast down the vertical axis into a low hero angle as the ring fires outward.\n4.8-6.6 MIRROR PORTAL, one take, no cut. Camera pushes into the reflection inside a single shard, passes through, emerges behind her at a new angle. Time-echo ghosts trail her lunge.\n6.6-8.4 SPIKE. Kaleidoscope fracture splits the frame into sixfold symmetry, resolves, and duplicates of her peel out of the shard surfaces, each trailing a different accent. Max two of her in frame. Extreme dutch angle.\n8.4-10.4 SAKUGA MONTAGE. One exchange per frame: line-art on flat cyan, two-color coral-teal silhouette, heavy smear frame, one-frame negative invert, full render. Debris accumulates, never resets.\n10.4-12.4 ESCALATION. Camera barrel-rolls a full 360 pushing in low. A shard catches an incoming strike, splits it into spectrum, fires it back doubled — the shockwave warps the background and bends the type.\n12.4-15.0 REVERSAL. Literal time reversal in slow motion. Debris rises, cracks close, light retreats, shards spiral back along her arm and flow together as liquid mercury before setting into one unbroken blade, the engraved mark resolving last. Camera eases into the exact opening framing. Final frame matches frame one.\n\n[MOTION] Momentum over posing. Cloth, hair and coat lag every direction change. Wrist and grip rotate with the blade, weight carries through footwork. Speed ramp each beat — slow wind-up, fast snap, settle. 杜绝任何僵硬、停顿或不自然的关节角度。\n\n[AUDIO] Every cue on the exact frame of its beat. The mix loops as cleanly as the picture. Bed: open air, bright wind, thin sustained glass resonance.\n0.0-1.6 rising metallic ring, crystalline detonation on the break frame.\n1.6-3.2 filtered vacuum for the frozen orbit, faint chimes and a low drone panning with the arc, rush as motion resumes.\n3.2-4.8 hard glass snap on the lock frame, deep whoosh on the drop, wide fan of air as the ring fires.\n4.8-6.6 pressure drop entering the shard, muffled interior tone, sharp release on the exit.\n6.6-8.4 stuttering echo as the kaleidoscope splits, sub-bass swell, one chime per duplicate.\n8.4-10.4 staccato impacts locked frame-for-frame to the cuts, glass cracking, blade screech, hard gate on the invert frame.\n10.4-12.4 doppler roll following the barrel turn, spectral riser on the split, doubled bright impact out.\n12.4-15.0 the mix runs in reverse — reversed chimes, glass reassembling, liquid metallic pour, resolving into the opening ring.\nMusic: glass bells and plucks in at 3.2, strings rise at 6.6, choir swell peaks at 10.4, reversing from 12.4 back into its first note. No dialogue.\n\n[LIMITS] Max two figures. Accents as refracted light only. Effects accumulate until the reversal. All type carries its gradient. First and last frame identical, no floating type in either. Engraved mark reads CIA0, spelled exactly. No logos, watermarks, or extra subtitles. No environment. Never let the frame go dark or desaturated.",
    "promptSourceUrl": "https://x.com/Cia0_exe/status/2082774668562538636",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Cia0_exe/status/2082774526098874724",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-07-30T10:25:17.758Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082774526098874724.mp4",
    "posterUrl": "/posters/x/2082774526098874724.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Fashion",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "UI/UX for animation your website. \n\nmade it with MiniMax H3 Ai @Hailuo_AI X midjourney V8.2 \n\nZero setting by design, type and colour carry the entire cinematic load.\n\n#MiniMaxH3 available for character work, promos, and brand films \n\n6,789 characters of prompt",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082770042630943156",
    "title": "无线电员守住撤离桥",
    "titleEn": "Radio operator evacuation bridge",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《无线电员守住撤离桥》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Radio operator evacuation bridge. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "FORMAT\n15 seconds | 16:9 | photoreal live-action war thriller\nFictional Sahelian city at blue-hour dawn.\nUrgent, human, suspenseful, non-graphic.\n\nREFERENCE CONTROL\nImage 1 = locked AMINA identity and wardrobe.\nPreserve face, close-cropped hair, deep-brown skin, navy field jacket, silver earpiece, canvas satchel, analogue radio, boots and determined expression.\n\nImage 2 = locked environment.\nPreserve concrete bridge, dry riverbed, dense low-rise city, rooftop antennae, dusty atmosphere, cool-blue shadows and restrained amber dawn.\n\nNo real countries, flags, organizations or landmarks.\n\nSTORY\nAmina is the last radio operator keeping an evacuation bridge open. Communications fail during a distant attack. She must activate the backup relay and transmit clearance before automatic controls close the bridge.\n\n00–02s — CHANNEL OFFLINE\nExtreme close-up inside a dark concrete communications room.\n\nAmina grips the analogue radio as red emergency lights pulse. Dust trembles, cables quiver, broken static fills the room.\n\nAutomated voice:\n“Emergency channel offline.”\n\nShe looks up and immediately moves.\n\n02–04s — BRIDGE CLOSING\nWide aerial-oblique rooftop view.\n\nCivilians cross the bridge through windblown dust.\n\nCheckpoint signal flickers green → red as shutdown begins.\n\nA distant shockwave moves through city haze. No visible casualties.\n\n04–06.5s — RUN\nLow handheld tracking behind Amina running through a damaged service corridor.\n\nEmergency lights flash, dust falls, cables briefly spark, boots hit wet concrete.\n\nShe pushes through a heavy metal door into blowing dust.\n\n06.5–09s — FLASHBACK\nTwo minutes earlier.\n\nMedium side-profile on rooftop.\n\nAmina sees the unstable bridge signal, manually rotates the antenna and connects her radio to backup power.\n\nSlow camera push. Wind moves her satchel strap as first amber sunlight reaches her jacket.\n\n09–12.5s — TRANSMISSION\nRapid cross-cut montage:\n\nfrequency dial locks →\nantenna aligns →\nbridge cables vibrate →\nred signal flashes →\nAmina speaks into radio:\n\n“Channel Seven, keep the bridge open.”\n\nOverhead shot: civilians continue crossing.\n\nSharp hard cuts, brief black frames, natural motion blur.\n\n12.5–15s — SIGNAL RESTORED\nSound suddenly drops.\n\nAmina listens on the rooftop.\n\nClear radio tone.\n\nCheckpoint signal changes red → steady green.\n\nEvacuation continues.\n\nAmina exhales and watches the bridge as dawn reaches the city.\n\nSlow push toward her face.\nCut to black.\n\nCAMERA\nLarge-format live-action realism.\nPrecise nonlinear cross-cutting.\nHandheld tracking, aerial-oblique coverage, restrained push-ins and close details.\nNatural motion blur, fine film grain, subtle gate weave.\nNo impossible camera movement.\n\nENVIRONMENT + PHYSICS\nCool-blue shadows against muted amber sunrise.\nDeep blacks, dusty atmosphere, weathered concrete.\n\nWind realistically affects dust, clothing, cables and antenna equipment.\nAll movement and interactions remain physically believable.\n\nAUDIO\nStereo radio static, electrical hum, wind, boots, antenna creaks, distant low artillery and helicopter rotors.\n\nRestrained suspenseful sub-bass builds through the montage.\n\nVISUAL\nPhotoreal cinematic war-thriller footage.\nImmense scale contrasted with one determined operator.\nGrounded, tense and human.\n\nNo subtitles, text, logos, watermarks, political messaging, flags or military insignia.\n\nCONTINUITY\nSame Amina identity, wardrobe, radio and satchel throughout.\nSame bridge, city, atmosphere and dawn lighting.\nMaintain geographic and temporal continuity across all shots.",
    "promptSourceUrl": "https://x.com/Diplomeme/status/2082770042630943156",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Diplomeme/status/2082770042630943156",
    "sourceLabel": "X 原帖 · @Diplomeme",
    "author": "ᴍᴜʀᴘʜʏ",
    "publishedAt": "2026-07-30T10:07:28.816Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082770042630943156.mp4",
    "posterUrl": "/posters/x/2082770042630943156.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tested the MiniMax H3 model, and the quality genuinely impressed me!\n\nCreated using @Hailuo_AI\n\nH3 is a general-purpose multimodal video model designed to handle more than specialized generation tasks.\n\nIt brings text, image, video, and audio understanding into a unified system, translating creative ideas into coherent, natural audiovisual content through a single end-to-end workflow. #MiniMaxH3\n\nPrompt ↓ \n\nFORMAT\n15 seconds | 16:9 | photoreal live-action war thriller\nFictional Sahelian city at blue-hour dawn.\nUrgent, human, suspenseful, non-graphic.\n\nREFERENCE CONTROL\nImage 1 = locked AMINA identity and wardrobe.\nPreserve face, close-cropped hair, deep-brown skin, navy field jacket, silver earpiece, canvas satchel, analogue radio, boots and determined expression.\n\nImage 2 = locked environment.\nPreserve concrete bridge, dry riverbed, dense low-rise city, rooftop antennae, dusty atmosphere, cool-blue shadows and restrained amber dawn.\n\nNo real countries, flags, organizations or landmarks.\n\nSTORY\nAmina is the last radio operator keeping an evacuation bridge open. Communications fail during a distant attack. She must activate the backup relay and transmit clearance before automatic controls close the bridge.\n\n00–02s — CHANNEL OFFLINE\nExtreme close-up inside a dark concrete communications room.\n\nAmina grips the analogue radio as red emergency lights pulse. Dust trembles, cables quiver, broken static fills the room.\n\nAutomated voice:\n“Emergency channel offline.”\n\nShe looks up and immediately moves.\n\n02–04s — BRIDGE CLOSING\nWide aerial-oblique rooftop view.\n\nCivilians cross the bridge through windblown dust.\n\nCheckpoint signal flickers green → red as shutdown begins.\n\nA distant shockwave moves through city haze. No visible casualties.\n\n04–06.5s — RUN\nLow handheld tracking behind Amina running through a damaged service corridor.\n\nEmergency lights flash, dust falls, cables briefly spark, boots hit wet concrete.\n\nShe pushes through a heavy metal door into blowing dust.\n\n06.5–09s — FLASHBACK\nTwo minutes earlier.\n\nMedium side-profile on rooftop.\n\nAmina sees the unstable bridge signal, manually rotates the antenna and connects her radio to backup power.\n\nSlow camera push. Wind moves her satchel strap as first amber sunlight reaches her jacket.\n\n09–12.5s — TRANSMISSION\nRapid cross-cut montage:\n\nfrequency dial locks →\nantenna aligns →\nbridge cables vibrate →\nred signal flashes →\nAmina speaks into radio:\n\n“Channel Seven, keep the bridge open.”\n\nOverhead shot: civilians continue crossing.\n\nSharp hard cuts, brief black frames, natural motion blur.\n\n12.5–15s — SIGNAL RESTORED\nSound suddenly drops.\n\nAmina listens on the rooftop.\n\nClear radio tone.\n\nCheckpoint signal changes red → steady green.\n\nEvacuation continues.\n\nAmina exhales and watches the bridge as dawn reaches the city.\n\nSlow push toward her face.\nCut to black.\n\nCAMERA\nLarge-format live-action realism.\nPrecise nonlinear cross-cutting.\nHandheld tracking, aerial-oblique coverage, restrained push-ins and close details.\nNatural motion blur, fine film grain, subtle gate weave.\nNo impossible camera movement.\n\nENVIRONMENT + PHYSICS\nCool-blue shadows against muted amber sunrise.\nDeep blacks, dusty atmosphere, weathered concrete.\n\nWind realistically affects dust, clothing, cables and antenna equipment.\nAll movement and interactions remain physically believable.\n\nAUDIO\nStereo radio static, electrical hum, wind, boots, antenna creaks, distant low artillery and helicopter rotors.\n\nRestrained suspenseful sub-bass builds through the montage.\n\nVISUAL \nPhotoreal cinematic war-thriller footage.\nImmense scale contrasted with one determined operator.\nGrounded, tense and human.\n\nNo subtitles, text, logos, watermarks, political messaging, flags or military insignia.\n\nCONTINUITY\nSame Amina identity, wardrobe, radio and satchel throughout.\nSame bridge, city, atmosphere and dawn lighting.\nMaintain geographic and temporal continuity across all shots.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082768164413428159",
    "title": "海外零食 ASMR 多镜开箱动画",
    "titleEn": "ASMR multi-cut overseas snack unboxing anime",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《海外零食 ASMR 多镜开箱动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for ASMR multi-cut overseas snack unboxing anime. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "日本のフルカラーアニメ映画風、シネマティックな高品質映像。BGMなし、セリフなし、字幕なし、文字なし、環境音のみ。\n\n# 映像スタイル\n添付イラストのキャラクター本人感を最優先。\n過度な写実化・厚塗り・複雑な陰影は禁止。\nキャラクターの顔立ち、髪型、体型、衣装、アクセは変えない。\n\n# 制約\n全カットで画角とアングルを変えて単調にしない。\n文字・ロゴは入れない。自然な生活感のある仕草にする。各カットは完全な固定カメラ（カメラワーク・カメラ移動なし）。フェードやモーフィングは禁止し、明確なハードカットで瞬時に切り替える。所作は極めてゆっくり、静かに、最小限に。「キャラクターの顔」と「手元の細かい作業」は絶対に同じカットに入れない。\n\n# 登場人物\n## 女性\n添付イラストの成人女性。\n\n# 情景\n夜のリビングのローテーブル。ランプの光の輪の中に段ボールや包みが待っている。\n\n# シーン\n女性が海外のお菓子詰め合わせを開封する。テープ、緩衝材、包み紙。開ける音の快感を最大限に立てながら、中身が見えるまでの期待感で引っ張る。開封の工程と仕草を交互に見せる。\nこの1本だけの見せ場として、「カラフルな包み紙がテーブルに広がっていく俯瞰」を後半のカットに必ず入れる。\n\ncut1:\n超クローズアップ。ランプの光の輪の中、箱のテープにカッターの刃が当たる。\n環境音：刃を出すカチカチという音だけ。\n\ncut2:\n超クローズアップ、真上俯瞰。テープがすーっと切れていく。\n環境音：テープの切れる音。\n\ncut3:\nクローズアップ、正面。瞳がきらきらして、蓋に手をかける。\n環境音：夜の静けさ。\n\ncut4:\nクローズアップ、箱の内側視点。蓋が開き、覗き込む顔が現れる。\n環境音：段ボールの軋み。\n\ncut5:\n超クローズアップ、手元。緩衝材のプチプチを一枚めくる。指がひとつだけぷちんと潰す。\n環境音：プチプチの音を主役に。\n\ncut6:\nクローズアップ、横顔。息を止めて中を見つめる。\n環境音：静けさ。\n\ncut7:\n超クローズアップ。薄紙を一枚ずつぺりぺりと剥がしていく。\n環境音：薄紙の繊細な音。\n\ncut8:\n超クローズアップ。海外のお菓子詰め合わせの一部がちらりと見える。まだ全体は見せない。\n環境音：包装の最後の音。\n\ncut9:\nミディアム、真後ろ。ランプの光の輪の中で開封を続ける後ろ姿。\n環境音：包みの音、夜の静けさ。\n\ncut10:\nクローズアップ、俯瞰。海外のお菓子詰め合わせが姿を現す。指先でそっと持ち上げる。\n環境音：包装が落ちる音。\n\ncut11:\nクローズアップ、横顔。海外のお菓子詰め合わせをランプの光にかざしてうっとり見つめる。\n環境音：静けさ。\n\ncut12:\n超クローズアップ、手元。付属品や説明書をぱらりとめくって確認する。\n環境音：紙のめくれる音。\n\ncut13:\nクローズアップ、斜め上。海外のお菓子詰め合わせを部屋の定位置にそっと置き、角度を整える。\n環境音：置く小さな音。\n\ncut14:\n超クローズアップ。空箱をばりばりと畳む。開封の儀の後片付け。\n環境音：段ボールの音。\n\ncut15:\nワイド、テーブル全景。開け終わった包みと戦利品が光の輪に残る。\n環境音：夜の静けさ。\n\ncut16:\nワイド。一番かわいい缶を棚に飾る。ランプの光の中、満足の空気が満ちる。\n環境音：静寂へフェード。",
    "promptSourceUrl": "https://x.com/aiehon_aya/status/2082768164413428159",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/aiehon_aya/status/2082768164413428159",
    "sourceLabel": "X 原帖 · @aiehon_aya",
    "author": "妖精アーヤさん",
    "publishedAt": "2026-07-30T10:00:01.014Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082768164413428159.mp4",
    "posterUrl": "/posters/x/2082768164413428159.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo 新モデル MiniMaxH3で作る\nASMRマルチカット\n─────────────\nMiniMaxH3：15秒 150クレジット\n（Seedance2.0の半額以下！）\n─────────────\nプロンプト：\n日本のフルカラーアニメ映画風、シネマティックな高品質映像。BGMなし、セリフなし、字幕なし、文字なし、環境音のみ。\n\n# 映像スタイル\n添付イラストのキャラクター本人感を最優先。\n過度な写実化・厚塗り・複雑な陰影は禁止。\nキャラクターの顔立ち、髪型、体型、衣装、アクセは変えない。\n\n# 制約\n全カットで画角とアングルを変えて単調にしない。\n文字・ロゴは入れない。自然な生活感のある仕草にする。各カットは完全な固定カメラ（カメラワーク・カメラ移動なし）。フェードやモーフィングは禁止し、明確なハードカットで瞬時に切り替える。所作は極めてゆっくり、静かに、最小限に。「キャラクターの顔」と「手元の細かい作業」は絶対に同じカットに入れない。\n\n# 登場人物\n## 女性\n添付イラストの成人女性。\n\n# 情景\n夜のリビングのローテーブル。ランプの光の輪の中に段ボールや包みが待っている。\n\n# シーン\n女性が海外のお菓子詰め合わせを開封する。テープ、緩衝材、包み紙。開ける音の快感を最大限に立てながら、中身が見えるまでの期待感で引っ張る。開封の工程と仕草を交互に見せる。\nこの1本だけの見せ場として、「カラフルな包み紙がテーブルに広がっていく俯瞰」を後半のカットに必ず入れる。\n\ncut1:\n超クローズアップ。ランプの光の輪の中、箱のテープにカッターの刃が当たる。\n環境音：刃を出すカチカチという音だけ。\n\ncut2:\n超クローズアップ、真上俯瞰。テープがすーっと切れていく。\n環境音：テープの切れる音。\n\ncut3:\nクローズアップ、正面。瞳がきらきらして、蓋に手をかける。\n環境音：夜の静けさ。\n\ncut4:\nクローズアップ、箱の内側視点。蓋が開き、覗き込む顔が現れる。\n環境音：段ボールの軋み。\n\ncut5:\n超クローズアップ、手元。緩衝材のプチプチを一枚めくる。指がひとつだけぷちんと潰す。\n環境音：プチプチの音を主役に。\n\ncut6:\nクローズアップ、横顔。息を止めて中を見つめる。\n環境音：静けさ。\n\ncut7:\n超クローズアップ。薄紙を一枚ずつぺりぺりと剥がしていく。\n環境音：薄紙の繊細な音。\n\ncut8:\n超クローズアップ。海外のお菓子詰め合わせの一部がちらりと見える。まだ全体は見せない。\n環境音：包装の最後の音。\n\ncut9:\nミディアム、真後ろ。ランプの光の輪の中で開封を続ける後ろ姿。\n環境音：包みの音、夜の静けさ。\n\ncut10:\nクローズアップ、俯瞰。海外のお菓子詰め合わせが姿を現す。指先でそっと持ち上げる。\n環境音：包装が落ちる音。\n\ncut11:\nクローズアップ、横顔。海外のお菓子詰め合わせをランプの光にかざしてうっとり見つめる。\n環境音：静けさ。\n\ncut12:\n超クローズアップ、手元。付属品や説明書をぱらりとめくって確認する。\n環境音：紙のめくれる音。\n\ncut13:\nクローズアップ、斜め上。海外のお菓子詰め合わせを部屋の定位置にそっと置き、角度を整える。\n環境音：置く小さな音。\n\ncut14:\n超クローズアップ。空箱をばりばりと畳む。開封の儀の後片付け。\n環境音：段ボールの音。\n\ncut15:\nワイド、テーブル全景。開け終わった包みと戦利品が光の輪に残る。\n環境音：夜の静けさ。\n\ncut16:\nワイド。一番かわいい缶を棚に飾る。ランプの光の中、満足の空気が満ちる。\n環境音：静寂へフェード。\n─────────────\n現在はアーリーアクセスでお試し中です。近日一般公開されるとのこと。楽しみにしててね🧚\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082762739697815758",
    "title": "神雕侠侣趣味讲单词 dream",
    "titleEn": "Condor Heroes characters teach English word dream",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《神雕侠侣趣味讲单词 dream》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Condor Heroes characters teach English word dream. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "神雕侠侣主角趣味讲单词 dream 教程",
    "promptSourceUrl": "https://x.com/nicekate8888/status/2082762739697815758",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/nicekate8888/status/2082762739697815758",
    "sourceLabel": "X 原帖 · @nicekate8888",
    "author": "nicekate",
    "publishedAt": "2026-07-30T09:38:27.661Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082762739697815758.mp4",
    "posterUrl": "/posters/x/2082762739697815758.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Comedy",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "提示词：神雕侠侣主角趣味讲单词 dream 教程\n\n视频由 MiniMax H3 生成，怎么感觉男主有些油腻，还有点眼熟，不过声音和嘴型对得很好\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082754694250717186",
    "title": "夜店霹雳舞环绕镜头",
    "titleEn": "Nightclub Breakdance Camera Orbit",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《夜店霹雳舞环绕镜头》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Nightclub Breakdance Camera Orbit. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use @ image1 as the visual bible. The board is a 16-image 4x4 storyboard shot\nburst, not a first frame and not a production-board layout. Create one cohesive\n15-second 16:9 multi-shot music-video b-roll clip as real cinematic nightclub\nfootage only.\n\nTranslate the 16-frame progression into one continuous acrobatic male\nbreakdance phrase: 0-3s establish the crowd circle, low fisheye master, top\nrock entry, and the start of a clockwise circular dolly orbit; 3-6s show crowd\nopening, drop to floor, six-step / three-step, and readable hand/foot floor\ncontact; 6-9s show CC / kick-out, sweep, power-move setup, and windmill or\nbackspin-like power motion; 9-12s hold the rotation through the second power\nbeat and catch a baby/chair freeze; 12-15s complete the orbit to the front,\nrelease into final footwork, land a decisive freeze/tableau, and show crowd\neruption.\n\nPreserve the board's central athletic male breaker, adult nightclub crowd\ncircle, glossy wet black floor, red backlight, acid green lasers, white strobes,\nblue haze, water mist, fisheye lens, and circular camera geography. The camera\nmay rotate around the dancer, but the movement must stay motivated and readable:\nlow circular dolly, Steadicam crowd-circle wrap, low tracking inserts, and\nfront-angle freeze completion. Avoid random flying camera, chaotic spinning,\nteleporting perspective jumps, losing the dancer, impossible acrobatics,\nduplicate limbs, foreground blockers, visible camera gear, text, logos, singer,\nor children.",
    "promptSourceUrl": "https://x.com/luisnomad/status/2082754696754663842",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/luisnomad/status/2082754694250717186",
    "sourceLabel": "X 原帖 · @luisnomad",
    "author": "Luis",
    "publishedAt": "2026-07-30T09:06:29.477Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082754694250717186.mp4",
    "posterUrl": "/posters/x/2082754694250717186.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing @Hailuo_AI new model H3, and the results are quite impressive! Here's a comparison of the same prompt using H3 (left) vs Seedance 2 (right). Prompt and reference image in first comment #H3 #MinimaxH3 @Hailuo_AI",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082747871103942674",
    "title": "孟买季风 FPV 航拍穿越",
    "titleEn": "Mumbai monsoon FPV drone flight",
    "model": "MiniMax H3（来源标注）",
    "mode": "FL2VA",
    "summary": "《孟买季风 FPV 航拍穿越》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×960。",
    "summaryEn": "An H3 output for Mumbai monsoon FPV drone flight. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×960.",
    "prompt": "Reference: Use the attached @ Image1  1 as the exact first frame and environment reference. The green drawn line is only the camera flight-path guide. Do not show the green line in the final video.\n\nCreate a 15-second hyper-real cinematic FPV drone video over Mumbai in heavy monsoon weather, using the same skyline, interchange, sea edge, road geometry, dense buildings, rainy morning atmosphere, grey clouds, and wet surfaces from the reference image. Keep everything realistic and grounded.  The drone must follow the green guideline path as a smooth continuous move: begin near the left side of the interchange, rise and sweep across the middle highway corridor, continue through the right-side city edge, then climb higher into the skyline, perform a tight spiral/orbit near the upper tower cluster, descend back toward the interchange, curve around the large circular loop, and finally glide out toward the lower-right water edge. The move should feel fast, elegant, immersive, and controlled — like a premium FPV drone shot, not a random flyover.\n\nTIMELINE\n\n 0:00–0:03 Start low near the left feeder ramp above the wet interchange. Heavy rain falls across the lens. The drone accelerates forward and slightly upward, revealing slick roads, tiny moving traffic, dark sea barriers, and the circular loop ahead.\n\n0:03–0:06 Follow the path upward across the mid-frame highway corridor. Keep the skyline growing larger. Wet roads shimmer, rain haze softens the distance, and the sea remains visible to the right. Motion is quick but readable.\n\n0:06–0:09 Sweep along the right-side city edge and climb higher. Dense towers, mid-rise blocks, and dark green patches below feel soaked and monsoon-heavy. The camera begins rising toward the upper skyline.\n\n0:09–0:11 Reach the upper skyline cluster and perform a tight spiral/orbit around the central high-rises. Storm clouds loom above, rain streaks rush past, and the city beneath feels deep, layered, and cinematic.\n\n0:11–0:13 Exit the spiral and descend in a smooth arc back toward the interchange. The circular loop becomes dominant again. Maintain realistic drone inertia and readable geography.\n\n0:13–0:15 Curve around the large circular flyover loop and finish with a fast glide toward the lower-right over the water, ending on a dramatic monsoon reveal of wet infrastructure, dark sea, and rain-washed Mumbai.\n\nCAMERA LANGUAGE  FPV drone POV, smooth stabilized flight, realistic banking turns, slight motion blur from speed and rain, believable acceleration and inertia, one continuous shot, no teleporting, no abrupt cuts, no impossible passes through buildings.\n\nVISUAL STYLE  Hyper-real cinematic realism, heavy monsoon rain, grey storm clouds, wet roads, reflective flyovers, rain haze, cool blue-grey morning light, dark sea, atmospheric depth, realistic Mumbai scale, premium documentary-film look.\n\nAUDIO  No voiceover. No dialogue. Use heavy rainfall, rushing wind, distant thunder, soft drone motor whirr, faint traffic hum, sea movement, and subtle cinematic tension music building with the motion.\n\nIMPORTANT  Do not show the green route line. No text, captions, logos, or watermark. Keep Mumbai realistic, monsoon-heavy, and geographically believable.",
    "promptSourceUrl": "https://x.com/CurieuxExplorer/status/2082747871103942674",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/CurieuxExplorer/status/2082747871103942674",
    "sourceLabel": "X 原帖 · @CurieuxExplorer",
    "author": "Dev Khanna",
    "publishedAt": "2026-07-30T08:39:22.712Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082747871103942674.mp4",
    "posterUrl": "/posters/x/2082747871103942674.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×960",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Travel",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Quite flight Mumbai 👀\n\nMiniMax H3 from @Hailuo_AI [early access]\n11 concepts 🧵👇 #MumbaiRains  \n\nPrompt:\n\nReference: Use the attached @ Image1  1 as the exact first frame and environment reference. The green drawn line is only the camera flight-path guide. Do not show the green line in the final video.  \n\nCreate a 15-second hyper-real cinematic FPV drone video over Mumbai in heavy monsoon weather, using the same skyline, interchange, sea edge, road geometry, dense buildings, rainy morning atmosphere, grey clouds, and wet surfaces from the reference image. Keep everything realistic and grounded.  The drone must follow the green guideline path as a smooth continuous move: begin near the left side of the interchange, rise and sweep across the middle highway corridor, continue through the right-side city edge, then climb higher into the skyline, perform a tight spiral/orbit near the upper tower cluster, descend back toward the interchange, curve around the large circular loop, and finally glide out toward the lower-right water edge. The move should feel fast, elegant, immersive, and controlled — like a premium FPV drone shot, not a random flyover.  \n\nTIMELINE \n\n 0:00–0:03 Start low near the left feeder ramp above the wet interchange. Heavy rain falls across the lens. The drone accelerates forward and slightly upward, revealing slick roads, tiny moving traffic, dark sea barriers, and the circular loop ahead.  \n\n0:03–0:06 Follow the path upward across the mid-frame highway corridor. Keep the skyline growing larger. Wet roads shimmer, rain haze softens the distance, and the sea remains visible to the right. Motion is quick but readable.  \n\n0:06–0:09 Sweep along the right-side city edge and climb higher. Dense towers, mid-rise blocks, and dark green patches below feel soaked and monsoon-heavy. The camera begins rising toward the upper skyline.  \n\n0:09–0:11 Reach the upper skyline cluster and perform a tight spiral/orbit around the central high-rises. Storm clouds loom above, rain streaks rush past, and the city beneath feels deep, layered, and cinematic.  \n\n0:11–0:13 Exit the spiral and descend in a smooth arc back toward the interchange. The circular loop becomes dominant again. Maintain realistic drone inertia and readable geography.  \n\n0:13–0:15 Curve around the large circular flyover loop and finish with a fast glide toward the lower-right over the water, ending on a dramatic monsoon reveal of wet infrastructure, dark sea, and rain-washed Mumbai.  \n\nCAMERA LANGUAGE  FPV drone POV, smooth stabilized flight, realistic banking turns, slight motion blur from speed and rain, believable acceleration and inertia, one continuous shot, no teleporting, no abrupt cuts, no impossible passes through buildings.  \n\nVISUAL STYLE  Hyper-real cinematic realism, heavy monsoon rain, grey storm clouds, wet roads, reflective flyovers, rain haze, cool blue-grey morning light, dark sea, atmospheric depth, realistic Mumbai scale, premium documentary-film look.  \n\nAUDIO  No voiceover. No dialogue. Use heavy rainfall, rushing wind, distant thunder, soft drone motor whirr, faint traffic hum, sea movement, and subtle cinematic tension music building with the motion.  \n\nIMPORTANT  Do not show the green route line. No text, captions, logos, or watermark. Keep Mumbai realistic, monsoon-heavy, and geographically believable.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082747838782386563",
    "title": "公园巨型锦鲤意外",
    "titleEn": "Giant koi park incident",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《公园巨型锦鲤意外》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Giant koi park incident. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "15-second, 16:9 vertical, continuous single-take video that looks like authentic smartphone footage accidentally captured by a passerby in a city park. Overcast natural daylight, subtle handheld shake, limited phone stabilization, occasional autofocus adjustment, and realistic smartphone compression. The absurd event is filmed with a completely serious, unscripted documentary feeling.  0–3s: [Handheld medium shot] A middle-aged man wearing a dark business suit and tie crouches beside the stone edge of a pond. With a completely serious expression, he slowly scatters pieces of bread from a paper bag to several ordinary koi. Small ripples spread across the water as the fish gather in front of him.  3–7s: [Camera instinctively moves closer] An abnormally huge orange-and-white koi suddenly surges out of the murky water, briefly lifting its upper body above the surface and biting down on the entire bread bag in the man’s left hand. He freezes for half a second, then grips the bag with both hands and leans backward. Startled, the person filming steps back. The image briefly loses focus before locking onto the man and the giant fish again.  7–11s: [Close handheld action shot] The giant koi pulls violently toward the deeper part of the pond. The wet paper bag stretches, the man’s arms tense, and his leather shoes slide repeatedly across the wet stone. His knee strikes the edge of the pond. He tries to brace himself with his right foot, but the sole loses traction and his center of gravity moves past the edge. Water, pieces of bread, and fallen leaves scatter from the force as the camera operator hurriedly moves sideways.  11–15s: [Impact and final hold] The paper bag suddenly tears. The man loses all support and pitches forward into the pond, creating one heavy, realistic splash. The camera quickly tilts downward while keeping the center of the pond visible. The man resurfaces with duckweed covering his head and his wet tie stuck across his face. The giant koi calmly swims past him with the remains of the bread bag still in its mouth. The camera holds on the man’s stunned expression while the koi casually swims away beside him.  Keep the man’s face, dark suit, tie, and paper bag visually consistent throughout. The koi must retain the same orange-and-white markings and enormous size. The pulling, sliding, loss of balance, and fall must show believable weight, inertia, traction, and water displacement. Natural park ambience and a realistic splash only. No dialogue, no subtitles, no music. Avoid cuts, character teleportation, changes in the fish’s size, extra limbs, and cartoonish acting.",
    "promptSourceUrl": "https://x.com/underwoodxie96/status/2082747838782386563",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/underwoodxie96/status/2082747838782386563",
    "sourceLabel": "X 原帖 · @underwoodxie96",
    "author": "underwood",
    "publishedAt": "2026-07-30T08:39:15.006Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082747838782386563.mp4",
    "posterUrl": "/posters/x/2082747838782386563.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Community Showcase",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 @Hailuo_AI just dropped, and I tested it with this 15-second prompt. The results were more interesting than I expected.\n\nThe character movement, physical interactions, and shot continuity are all impressive. At first glance, it already feels very close to Seedance 2.0.\n\nWatch the final video and find the full prompt below 👇\n\n15-second, 16:9 vertical, continuous single-take video that looks like authentic smartphone footage accidentally captured by a passerby in a city park. Overcast natural daylight, subtle handheld shake, limited phone stabilization, occasional autofocus adjustment, and realistic smartphone compression. The absurd event is filmed with a completely serious, unscripted documentary feeling.  0–3s: [Handheld medium shot] A middle-aged man wearing a dark business suit and tie crouches beside the stone edge of a pond. With a completely serious expression, he slowly scatters pieces of bread from a paper bag to several ordinary koi. Small ripples spread across the water as the fish gather in front of him.  3–7s: [Camera instinctively moves closer] An abnormally huge orange-and-white koi suddenly surges out of the murky water, briefly lifting its upper body above the surface and biting down on the entire bread bag in the man’s left hand. He freezes for half a second, then grips the bag with both hands and leans backward. Startled, the person filming steps back. The image briefly loses focus before locking onto the man and the giant fish again.  7–11s: [Close handheld action shot] The giant koi pulls violently toward the deeper part of the pond. The wet paper bag stretches, the man’s arms tense, and his leather shoes slide repeatedly across the wet stone. His knee strikes the edge of the pond. He tries to brace himself with his right foot, but the sole loses traction and his center of gravity moves past the edge. Water, pieces of bread, and fallen leaves scatter from the force as the camera operator hurriedly moves sideways.  11–15s: [Impact and final hold] The paper bag suddenly tears. The man loses all support and pitches forward into the pond, creating one heavy, realistic splash. The camera quickly tilts downward while keeping the center of the pond visible. The man resurfaces with duckweed covering his head and his wet tie stuck across his face. The giant koi calmly swims past him with the remains of the bread bag still in its mouth. The camera holds on the man’s stunned expression while the koi casually swims away beside him.  Keep the man’s face, dark suit, tie, and paper bag visually consistent throughout. The koi must retain the same orange-and-white markings and enormous size. The pulling, sliding, loss of balance, and fall must show believable weight, inertia, traction, and water displacement. Natural park ambience and a realistic splash only. No dialogue, no subtitles, no music. Avoid cuts, character teleportation, changes in the fish’s size, extra limbs, and cartoonish acting.\n\n#MiniMaxH3 #AIVideo",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082747126497300930",
    "title": "一九九八首尔雨夜公交站家用录像",
    "titleEn": "1998 Seoul rainy bus stop archival home film",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《一九九八首尔雨夜公交站家用录像》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for 1998 Seoul rainy bus stop archival home film. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Create an authentic 15-second archival home video that appears to have been recorded in Seoul, South Korea, during the summer of 1998 using a consumer Hi8 or VHS-C camcorder. The camcorder is never visible. The audience only sees the footage captured through the camera, exactly as if watching a forgotten family videotape decades later. This is not a vlog, not a selfie, and not a cinematic film. The unseen camera operator casually documents an ordinary moment from everyday life.\n\nThe footage should faithfully recreate genuine late-1990s analogue camcorder characteristics: soft 480i resolution, subtle VHS tape grain, faint interlacing, mild colour fading, occasional autofocus breathing, slight exposure shifts, gentle handheld shake, imperfect framing, tiny zoom corrections, realistic motion blur, analogue colour bleed, and faint tape hiss. No HDR, no digital sharpening, no beauty filters, no stabilisation, no cinematic lighting, no slow motion, and no modern camera effects.\n\nA light summer rain has just begun in a quiet Seoul neighborhood. A young Korean woman in her early twenties waits at a simple city bus stop beneath a clear plastic umbrella. She wears a pale blue oversized T-shirt tucked into relaxed-fit blue jeans, classic white sneakers, and carries a canvas shoulder bag, perfectly matching authentic Korean street fashion from 1998. Around her, commuters quietly wait while late-1990s Korean buses, sedans, and bicycles pass by on the wet street. Handwritten Hangul shop signs, overhead utility wires, convenience stores, and low-rise apartment buildings create an unmistakably authentic Seoul atmosphere.\n\nAs the rain becomes slightly heavier, a small orange tabby cat runs from beneath a parked car toward the bus stop. The woman notices it, crouches down with a warm smile, and gently shelters the cat under her umbrella. The cat calmly sits beside her while several nearby commuters smile at the unexpected moment. A city bus approaches in the background, its headlights reflecting across the rain-soaked road. The woman looks toward the arriving bus, softly laughs, gives the cat one gentle pat, and steps aboard as the doors close. The camera remains outside, lingering for a few seconds on the rainy street as the bus slowly drives away before the recording ends abruptly, exactly as if someone simply pressed the stop button on the camcorder.\n\nGeneration Requirements: Produce one continuous, uninterrupted 15-second recording with no cuts or transitions. Every detail must accurately recreate Seoul in 1998, including clothing, hairstyles, architecture, vehicles, signage, weather, and everyday behavior. Human movement must be completely natural with consistent facial identity, realistic anatomy, authentic hand movements, believable crowd interactions, and subtle environmental details such as rain, passing traffic, distant conversations, bus engine sounds, birds, and ambient city noise. Avoid AI artifacts, duplicated pedestrians, warped limbs, floating objects, unnatural expressions, perfect symmetry, or anything suggesting the footage was AI-generated. The finished result should be indistinguishable from genuine late-1990s home camcorder footage discovered in a family archive.",
    "promptSourceUrl": "https://x.com/itxabdullaa/status/2082747126497300930",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/itxabdullaa/status/2082747126497300930",
    "sourceLabel": "X 原帖 · @itxabdullaa",
    "author": "ÀBDŪLLÂH",
    "publishedAt": "2026-07-30T08:36:25.184Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082747126497300930.mp4",
    "posterUrl": "/posters/x/2082747126497300930.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "ugc"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax H3 on @Hailuo_AI #MiniMaxH3\n\nConcept: Seoul Summer, 1998 – Rainy Bus Stop\n\nPrompt:\n\nCreate an authentic 15-second archival home video that appears to have been recorded in Seoul, South Korea, during the summer of 1998 using a consumer Hi8 or VHS-C camcorder. The camcorder is never visible. The audience only sees the footage captured through the camera, exactly as if watching a forgotten family videotape decades later. This is not a vlog, not a selfie, and not a cinematic film. The unseen camera operator casually documents an ordinary moment from everyday life.\n\nThe footage should faithfully recreate genuine late-1990s analogue camcorder characteristics: soft 480i resolution, subtle VHS tape grain, faint interlacing, mild colour fading, occasional autofocus breathing, slight exposure shifts, gentle handheld shake, imperfect framing, tiny zoom corrections, realistic motion blur, analogue colour bleed, and faint tape hiss. No HDR, no digital sharpening, no beauty filters, no stabilisation, no cinematic lighting, no slow motion, and no modern camera effects.\n\nA light summer rain has just begun in a quiet Seoul neighborhood. A young Korean woman in her early twenties waits at a simple city bus stop beneath a clear plastic umbrella. She wears a pale blue oversized T-shirt tucked into relaxed-fit blue jeans, classic white sneakers, and carries a canvas shoulder bag, perfectly matching authentic Korean street fashion from 1998. Around her, commuters quietly wait while late-1990s Korean buses, sedans, and bicycles pass by on the wet street. Handwritten Hangul shop signs, overhead utility wires, convenience stores, and low-rise apartment buildings create an unmistakably authentic Seoul atmosphere.\n\nAs the rain becomes slightly heavier, a small orange tabby cat runs from beneath a parked car toward the bus stop. The woman notices it, crouches down with a warm smile, and gently shelters the cat under her umbrella. The cat calmly sits beside her while several nearby commuters smile at the unexpected moment. A city bus approaches in the background, its headlights reflecting across the rain-soaked road. The woman looks toward the arriving bus, softly laughs, gives the cat one gentle pat, and steps aboard as the doors close. The camera remains outside, lingering for a few seconds on the rainy street as the bus slowly drives away before the recording ends abruptly, exactly as if someone simply pressed the stop button on the camcorder.\n\nGeneration Requirements: Produce one continuous, uninterrupted 15-second recording with no cuts or transitions. Every detail must accurately recreate Seoul in 1998, including clothing, hairstyles, architecture, vehicles, signage, weather, and everyday behavior. Human movement must be completely natural with consistent facial identity, realistic anatomy, authentic hand movements, believable crowd interactions, and subtle environmental details such as rain, passing traffic, distant conversations, bus engine sounds, birds, and ambient city noise. Avoid AI artifacts, duplicated pedestrians, warped limbs, floating objects, unnatural expressions, perfect symmetry, or anything suggesting the footage was AI-generated. The finished result should be indistinguishable from genuine late-1990s home camcorder footage discovered in a family archive.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082746108493848857",
    "title": "水墨虾跃入未来都市",
    "titleEn": "Ink-wash shrimp leap into futuristic city",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《水墨虾跃入未来都市》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Ink-wash shrimp leap into futuristic city. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "水墨虾跃入未来城市——一张宣纸在夜色摩天楼之间展开，八只黑白水墨虾从纸面跃出，身体仍保留飞白与墨晕，虾须化成发光城市轨道，墨滴变成霓虹雨；画面有巨大尺度反差、强烈纵深、电影级光影和一眼停留的奇观感，不要文字、不要水印。",
    "promptSourceUrl": "https://x.com/nicekate8888/status/2082746108493848857",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/nicekate8888/status/2082746108493848857",
    "sourceLabel": "X 原帖 · @nicekate8888",
    "author": "nicekate",
    "publishedAt": "2026-07-30T08:32:22.473Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082746108493848857.mp4",
    "posterUrl": "/posters/x/2082746108493848857.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "来一个搞笑视频《水墨虾跃入未来城市》\n\n视频1 MiniMax H3 文本生成视频，水墨虾的虾须很酷\n\n视频2-3 分别是 Grok Imagine、Gemini Omni，虾的数量搞错了\n\n三款模型在指令遵循上都有些问题，不过这轮对比下来，MiniMax H3 的完成度和指令遵循最好\n\n提示词：\n水墨虾跃入未来城市——一张宣纸在夜色摩天楼之间展开，八只黑白水墨虾从纸面跃出，身体仍保留飞白与墨晕，虾须化成发光城市轨道，墨滴变成霓虹雨；画面有巨大尺度反差、强烈纵深、电影级光影和一眼停留的奇观感，不要文字、不要水印。\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082745513573831106",
    "title": "破碎宫殿中的喷火巨龙",
    "titleEn": "Fire-breathing dragon in broken palace",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《破碎宫殿中的喷火巨龙》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Fire-breathing dragon in broken palace. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Fire-Breathing Dragon in a Broken Palace**  A colossal fire-breathing dragon with obsidian-black scales streaked by glowing molten cracks stalks through the ruins of a shattered royal palace during a violent thunderstorm. Rain crashes through the collapsed marble ceilings and broken palace domes, lightning flashes behind towering fractured columns and crumbling statues of forgotten kings, while thick smoke and mist drift across the flooded stone floors.  The dragon exhales a massive stream of blazing fire, turning the rain into clouds of steam and illuminating the ruined palace with fiery orange light. Every thunderous roar sends shockwaves through the ancient structure—massive pillars split apart, cracked walls collapse, shattered stained-glass windows explode outward, and glowing magma-like fissures spread beneath its claws across the palace floor.  The camera tracks low beside the dragon, capturing ultra-detailed wet scales, razor-sharp claws crushing marble tiles, smoke pouring from its nostrils, glowing molten eyes, and fire reflecting across the rain-soaked stone. Cinematic camera movement with dynamic close-ups, slow-motion embers, volumetric smoke, and dramatic lightning flashes emphasize the creature's immense scale.  **Final moment:** The dragon spreads its colossal wings atop the broken royal throne as a gigantic bolt of lightning strikes the ruined palace behind it. It unleashes an earth-shaking roar while breathing an enormous column of fire into the stormy sky, casting its gigantic shadow across the shattered palace walls.  **Mythic creature realism, cinematic fantasy, ultra-detailed dragon anatomy, broken palace environment, heavy rain, lightning storm, volumetric fog, fire and smoke simulation, dynamic destruction, epic scale, dramatic lighting, photorealistic CGI, cinematic 4K.**",
    "promptSourceUrl": "https://x.com/dreamydigiarts/status/2082745513573831106",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/dreamydigiarts/status/2082745513573831106",
    "sourceLabel": "X 原帖 · @dreamydigiarts",
    "author": "dreamy digital arts",
    "publishedAt": "2026-07-30T08:30:00.633Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082745513573831106.mp4",
    "posterUrl": "/posters/x/2082745513573831106.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing new model MiniMax H3 from @Hailuo_AI \n\n#MiniMaxH3\n\nPrompt 👇\n\nFire-Breathing Dragon in a Broken Palace**  A colossal fire-breathing dragon with obsidian-black scales streaked by glowing molten cracks stalks through the ruins of a shattered royal palace during a violent thunderstorm. Rain crashes through the collapsed marble ceilings and broken palace domes, lightning flashes behind towering fractured columns and crumbling statues of forgotten kings, while thick smoke and mist drift across the flooded stone floors.  The dragon exhales a massive stream of blazing fire, turning the rain into clouds of steam and illuminating the ruined palace with fiery orange light. Every thunderous roar sends shockwaves through the ancient structure—massive pillars split apart, cracked walls collapse, shattered stained-glass windows explode outward, and glowing magma-like fissures spread beneath its claws across the palace floor.  The camera tracks low beside the dragon, capturing ultra-detailed wet scales, razor-sharp claws crushing marble tiles, smoke pouring from its nostrils, glowing molten eyes, and fire reflecting across the rain-soaked stone. Cinematic camera movement with dynamic close-ups, slow-motion embers, volumetric smoke, and dramatic lightning flashes emphasize the creature's immense scale.  **Final moment:** The dragon spreads its colossal wings atop the broken royal throne as a gigantic bolt of lightning strikes the ruined palace behind it. It unleashes an earth-shaking roar while breathing an enormous column of fire into the stormy sky, casting its gigantic shadow across the shattered palace walls.  **Mythic creature realism, cinematic fantasy, ultra-detailed dragon anatomy, broken palace environment, heavy rain, lightning storm, volumetric fog, fire and smoke simulation, dynamic destruction, epic scale, dramatic lighting, photorealistic CGI, cinematic 4K.**",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082731549707927857",
    "title": "日出峡谷翼装飞行",
    "titleEn": "Sunrise Wingsuit Canyon Run",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《日出峡谷翼装飞行》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Sunrise Wingsuit Canyon Run. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "CANYON WINGSUIT THREAD WORLD BUILD A wingsuit pilot threading an impossible canyon route through ancient stone monuments during sunrise wind. Fear, precision, airspeed. LOOK BIBLE Ultra-wide cliff geography, warm edge light, cold shadows in canyon depth, fast ground blur, rock texture crisp near passes. HERO LOCK A female wingsuit pilot in matte black-and-copper suit, mirrored visor, compact oxygen rig. SOUND BED Wind shear, cloth tension, breathing through mask, stone flyby roar, pebble cascades from near misses. BEAT GRID 0–3s The camera drops with her off a sandstone arch into open canyon air, her body flattening into glide posture. 3–6s It locks just behind her shoulder as she skims a rock wall close enough to brush dust from it. The canyon splits ahead. 6–10s She angles through the narrower gap between two suspended stone bridges, sunrise blasting through the opening and turning the walls into streaks. 10–15s She flares hard above a hidden oasis basin and shoots through the center of a ruined archway, exiting into full gold morning light over still water. OUTPUT LOCKS No parachute deployment. Keep it pure wingsuit glide. Speed always readable.",
    "promptSourceUrl": "https://x.com/sebatheepan/status/2082731549707927857",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/sebatheepan/status/2082731549707927857",
    "sourceLabel": "X 原帖 · @sebatheepan",
    "author": "Pan",
    "publishedAt": "2026-07-30T07:34:31.388Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082731549707927857.mp4",
    "posterUrl": "/posters/x/2082731549707927857.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Action & VFX",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3 @Hailuo_AI \n\nPrompt ⤵️\n\nCANYON WINGSUIT THREAD WORLD BUILD A wingsuit pilot threading an impossible canyon route through ancient stone monuments during sunrise wind. Fear, precision, airspeed. LOOK BIBLE Ultra-wide cliff geography, warm edge light, cold shadows in canyon depth, fast ground blur, rock texture crisp near passes. HERO LOCK A female wingsuit pilot in matte black-and-copper suit, mirrored visor, compact oxygen rig. SOUND BED Wind shear, cloth tension, breathing through mask, stone flyby roar, pebble cascades from near misses. BEAT GRID 0–3s The camera drops with her off a sandstone arch into open canyon air, her body flattening into glide posture. 3–6s It locks just behind her shoulder as she skims a rock wall close enough to brush dust from it. The canyon splits ahead. 6–10s She angles through the narrower gap between two suspended stone bridges, sunrise blasting through the opening and turning the walls into streaks. 10–15s She flares hard above a hidden oasis basin and shoots through the center of a ruined archway, exiting into full gold morning light over still water. OUTPUT LOCKS No parachute deployment. Keep it pure wingsuit glide. Speed always readable.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082727329390854152",
    "title": "空西装定格组装",
    "titleEn": "Empty Suit Stop-Motion Assembly",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《空西装定格组装》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 12 秒 · 720×864。",
    "summaryEn": "An H3 output for Empty Suit Stop-Motion Assembly. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 12s · 720×864.",
    "prompt": "Static locked-off camera, no zoom, no pan, no handheld drift. The frame never moves.    Open on an empty landscape: flat azure sky, six soft cumulus clouds drifting slowly, a   band of ochre and green eroded hills along the bottom. Nothing else. Hold this for a beat.    Then the composition assembles itself, piece by piece, in surreal stop-motion cadence:    First, a torso rises into frame from the bottom edge: a man in a flecked brown tweed   jacket, white collar, dark tie. It settles and locks.    Above the collar, a vertical column starts growing upward from nothing: dark earth,   roots, dried vegetation and moss, building like soil stacking against gravity.    Chunks fly in from the frame edges and snap into the column with weight: coral and   fungal growths in salmon pink, chartreuse and rust from above; a large blue elephant   head with a visible eye and pale tusks sliding in from the right; branching red vascular   coral whipping in from the right edge.    The face parts land last and land hard: a huge glossy red fleshy mouth with white teeth   locks into the lower half, its trunk-like red tongue curling into a tight loop; a second   inverted red mouth snaps into the upper half.    Final details settle: a brown and orange moth wing unfolds at the upper right, small   pink petal-like slices freeze mid-flight on the left with fine particles suspended   around them, a small blue elephant figure tucks into the crown.    The clouds keep drifting behind and through the structure the entire time, unbothered.    Everything comes to rest in a still, balanced composition. Only the clouds move.    Photoreal 3D render, hyperreal wet organic surfaces, hard midday light, deep saturation,   crisp shadows. Surrealist portrait, Magritte staging. 12 seconds, 5:6 vertical.",
    "promptSourceUrl": "https://x.com/lukasersil/status/2082727332637290602",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/lukasersil/status/2082727329390854152",
    "sourceLabel": "X 原帖 · @lukasersil",
    "author": "Lukáš Eršil",
    "publishedAt": "2026-07-30T07:17:45.186Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082727329390854152.mp4",
    "posterUrl": "/posters/x/2082727329390854152.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "720×864",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Locked-off camera. Twelve seconds. The frame never moves, everything inside it does.😍\n\nAn empty suit on the horizon. Then the head builds itself out of soil, moss, coral, lips, an elephant. One continuous shot, no cuts.\n\nThe hard part of AI video was never a nice frame. It's holding the camera nailed down while the subject does all the work. Most models start drifting the second you ask them to stay still.\n\nMade in #MiniMaxH3 early access in @Hailuo_AI. Audio is native, straight out of the model.\n\nPrompt in comment👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082718662482411584",
    "title": "少林寺中的雷神降临",
    "titleEn": "Lightning God at a Shaolin Monastery",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《少林寺中的雷神降临》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 5 秒 · 1280×720。",
    "summaryEn": "An H3 output for Lightning God at a Shaolin Monastery. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 5s · 1280×720.",
    "prompt": "Buddhist monastery courtyard at dawn, mist drifting between stone pillars. Wide shot, shallow depth of field: a row of Shaolin priests kneeling in prayer, hands pressed together, saffron and grey robes, heads bowed. Soft ambient temple bells. Background monastery architecture slightly defocused, foreground priest sharply in focus.  Sudden thunderclap. A bolt of lightning crashes down center-frame — the Lightning God lands in a crouch, white kimono and white wide-brimmed hat, faint electricity crackling off the fabric. He rises slowly. Close-up: his eyes ignite, glowing white-blue lightning, no pupils, just arcing current.  He explodes into motion. Fast tracking shot as he closes distance on the first priest, palm strike trailing a lightning arc — priest is zapped, muscles seizing, thrown backward in a shower of sparks.  Quick cut: side angle, second priest lunges with a staff; the god sidesteps in a blur, lightning crackling along his arm, backhands the staff aside and zaps the priest's chest — priest crumples, smoke rising off his robes.  Quick cut: low angle, third priest sweeps a leg; god leaps straight up, hovers a beat with lightning rippling off his kimono hem, then drops down zapping the priest's shoulder.  Fast intercut: downed priests groaning, hands pressing into the ground, pushing themselves back up one by one, robes scorched but eyes determined — they re-engage together, converging on the god from three sides.  Wide shot, all four in frame: priests strike in unison, god pivots between them in a continuous whirl, every block and counter sparking with lightning, his white hat never falling. Camera whip-pans following the fastest exchanges, strobing light flashes matching each hit.  Final beat: god stands center, lightning arcing off him in all directions, breathing storm-light, priests circling back to their feet around him, ready to strike again — freeze on his glowing eyes, thunder rumbling.  Camera: mix of fixed depth-of-field opening shot, fast tracki",
    "promptSourceUrl": "https://x.com/seirdotmk/status/2082718792069566546",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/seirdotmk/status/2082718662482411584",
    "sourceLabel": "X 原帖 · @seirdotmk",
    "author": "Myron AI artist",
    "publishedAt": "2026-07-30T06:43:18.834Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082718662482411584.mp4",
    "posterUrl": "/posters/x/2082718662482411584.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Master is back! @Hailuo_AI just came outwith their new #MiniMaxH3, and I got early access!\nIt has Precision Editing & Control: Strict instruction-following and micro editing across entities, environments, audio.\nThis is what enabled me to create this fantastic fight scene in a single prompt, prompt is below:",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082712692159344891",
    "title": "飞空艇穿越浮空王国",
    "titleEn": "Skyship Through a Floating Kingdom",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《飞空艇穿越浮空王国》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Skyship Through a Floating Kingdom. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Skyship flight across a floating kingdom (single continuous shot) From above a sea of endless clouds, the camera dives toward a majestic skyship soaring between gigantic floating islands. Lock-on: the airship accelerating into a violent thunderstorm suspended in the sky. The camera slingshots beneath the hull, whips across the towering sails, then drops tight behind the glowing engines: lightning tears across the clouds, rain lashes the deck, shattered rock fragments tumble through the air. A collapsing floating island sends massive boulders crashing into the storm; the captain steers beneath a breaking stone arch before threading between waterfalls pouring into the sky, hanging bridges and drifting islands in one uninterrupted line. The camera mirrors every impossible maneuver before bursting through the storm into brilliant sunlight, revealing an endless kingdom of floating continents stretching beyond the horizon above a sea of clouds.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2082712692159344891",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Strength04_X/status/2082712692159344891",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-07-30T06:19:35.398Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082712692159344891.mp4",
    "posterUrl": "/posters/x/2082712692159344891.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Skyship Through the Storm \n\nMade with MiniMax H3 on @Hailuo_AI #MiniMaxH3 \n\nPrompt - Skyship flight across a floating kingdom (single continuous shot) From above a sea of endless clouds, the camera dives toward a majestic skyship soaring between gigantic floating islands. Lock-on: the airship accelerating into a violent thunderstorm suspended in the sky. The camera slingshots beneath the hull, whips across the towering sails, then drops tight behind the glowing engines: lightning tears across the clouds, rain lashes the deck, shattered rock fragments tumble through the air. A collapsing floating island sends massive boulders crashing into the storm; the captain steers beneath a breaking stone arch before threading between waterfalls pouring into the sky, hanging bridges and drifting islands in one uninterrupted line. The camera mirrors every impossible maneuver before bursting through the storm into brilliant sunlight, revealing an endless kingdom of floating continents stretching beyond the horizon above a sea of clouds.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082711476347998615",
    "title": "暗黑流行赛博废土说唱 MV 风格",
    "titleEn": "Dark-pop cyber-grunge rap music film style",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《暗黑流行赛博废土说唱 MV 风格》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Dark-pop cyber-grunge rap music film style. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Style: dark-pop / cyber-grunge / rap music video with photoreal high-fashion polish and the texture of a scanned film magazine—high contrast without looking cheap. Reference late-1990s to early-2000s indie magazines, photocopies, film scans, underground-music posters, and zine collage. Add coarse grain, subtle gate weave, halftone dots, rough print edges, and slight scan misregistration. Keep the edit fast and use hard cuts only—no fades or soft transitions. Match the typographic treatment and surface texture of the reference images.",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2082711476347998615",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Just_sharon7/status/2082711476347998615",
    "sourceLabel": "X 原帖 · @Just_sharon7",
    "author": "Sharon Riley",
    "publishedAt": "2026-07-30T06:14:45.526Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082711476347998615.mp4",
    "posterUrl": "/posters/x/2082711476347998615.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "photoreal"
    ],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tried MiniMax H3 model and I am seriously blown away by the quality!\n\nMade by using @Hailuo_AI \n\nH3 is a general-purpose multimodal video model that goes beyond specialized tools. \n\nIt unifies understanding of text, images, video, and audio to interpret creative intent and generate coherent, natural audiovisual content in one end-to-end workflow.\n\nPrompt\n\n(Partial prompt — expand as needed)\nStyle: dark-pop / cyber-grunge / rap music video with photoreal high-fashion polish and the texture of a scanned film magazine—high contrast without looking cheap. Reference late-1990s to early-2000s indie magazines, photocopies, film scans, underground-music posters, and zine collage. Add coarse grain, subtle gate weave, halftone dots, rough print edges, and slight scan misregistration. Keep the edit fast and use hard cuts only—no fades or soft transitions. Match the typographic treatment and surface texture of the reference images. #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082707037256056852",
    "title": "梦幻夏日温室动画",
    "titleEn": "Dreamlike Summer Greenhouse Anime",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《梦幻夏日温室动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Dreamlike Summer Greenhouse Anime. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "高品質アニメ映像。\n\n【作品トーン】\n透明感のある夏のガラス温室を舞台にした、上品で夢のような午後の叙情アニメ。映像全体は高密度な2D手描きアニメとして統一し、静かな呼吸、光の移動、素材ごとの反射と透過、繊細な表情変化を中心に描く。\n【参照画像の役割】\n画像1のみを使用する。画像1は登場する一人のキャラクターの参照として扱い、顔立ち、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、全身シルエット、雰囲気、キャラクター固有色だけを参照する。\n\n画像1そのものの背景、部屋、家具、建築、ポーズ、手の位置、画角、カメラ距離、フレーミング、構図、照明状態は再現しない。画像1の背景要素を温室内へ持ち込まない。\n画像1と同じ人物を、全編で同じ一人のキャラクターとして厳密に維持する。顔の輪郭、目の形、左右の瞳の特徴、虹彩色、前髪、髪の長さ、髪型、髪飾り、衣装構造、装飾位置、体格、年齢感、キャラクター固有色を変えない。表情、視線、口元、呼吸、肩の力、手の動き、自然な髪と衣装の揺れだけ変化してよい。特徴の混合、顔の平均化、別人化、衣装交換、装飾消失、髪型変更、クローン化、分身、余計な人物の追加は禁止する。\n\n【絵柄固定】\nキャラクター同一性とは別条件として、映像全体を繊細で高密度な2D手描きアニメに固定する。細く淡い寒色系の線、透明感のある多層の色影、柔らかなグラデーション、肌の淡い発光感、髪の繊細なハイライト、瞳の多層反射を維持する。花、ガラス、紅茶、空、雲、白い家具、床タイル、温室建築も同じ密度の手描き筆致で描く。太い輪郭、簡略化されたTVアニメ調、平坦な単層セル影、低密度背景、滑らかなCG、3D化、半写実、実写化、画風混合にしない。\n\n【キャラクターに合わせて変化させる要素】\n場所と出来事は固定し、小物の意匠、アクセント色、光の印象、紅茶の色味、卓上の小さな添え物だけを画像1のキャラクターに合わせて調整する。\n\n温室の白と淡い青を基調とした明るさは維持しながら、画像1の髪色、瞳色、衣装色、装飾色から一色だけを補助アクセントとして抽出する。アクセント色はカップの縁、ソーサーの細い装飾、卓上の花、屈折光のわずかな色味に限定し、背景全体をキャラクター色で塗り替えない。\nティーカップとソーサーの形状は、画像1の衣装や装飾の印象に合わせて、上品、可憐、端正、幻想的などの範囲でわずかに意匠を調整してよい。ただし透明なガラス製であること、カップ一つ、取っ手一つ、ソーサー一枚であることは固定する。\n\n紅茶は透明感のある温かい色を保ち、画像1のキャラクターカラーに合わせて、琥珀色、淡い蜂蜜色、薄い薔薇色、淡い青緑を含む幻想的な茶色などへ穏やかに調整してよい。濁った液体、不透明な液体、蛍光色にはしない。\n卓上の添え物は一つだけ置く。画像1の雰囲気に合わせ、柑橘の一切れ、小さな砂糖菓子、一輪の花、薄い果実片などから最も自然なものを一つ選ぶ。添え物は途中で増殖、移動、変形させない。\n\n【固定舞台】\n舞台は明るい夏のガラス温室。白い窓枠と屋根骨格、澄んだ青空、白い雲、外の緑と花、白い丸テーブル、白い金属椅子、花鉢、淡い青灰色の床タイルを維持する。白と淡い青を基調とし、日差しが作る淡い金色と、画像1から抽出した補助アクセント色だけを使用する。\n\n【小物固定】\n開始状態では、透明なガラスのティーカップ一つに紅茶が入り、誰の手にも保持されず、透明なガラスのソーサー一枚の中央に置かれている。卓上の添え物一つは白い丸テーブルの上に置いたままにする。\n\n後半でキャラクター本人だけが右手でカップの取っ手を持つ。カップ、取っ手、紅茶、ソーサー、添え物を増殖、交換、消失、変形させない。\n\n【開始カメラ】\n最初のカメラは白い丸テーブルの斜め前、テーブル面と同じ低い高さからキャラクターの斜め前へ向く。前景にティーカップとソーサーを大きく置き、奥にキャラクターの胸上と顔を明瞭に見せる。背景消失軸は床タイルと右側の窓枠が温室奥へ伸びる方向に揃える。\n\n紅茶とキャラクターを最初から同じ画面で紹介する。キャラクターは穏やかに呼吸し、髪先、袖、衣装の軽い装飾だけが温室の微風でわずかに揺れる。\n【光の発生】\n屋外の薄い雲が横へ流れて日差しが開き、温室の窓ガラスを通った白い陽光が紅茶を透過する。紅茶と透明ガラスの屈折によって生まれた柔らかな光模様が、ソーサーの縁から白いテーブルへ現れる。\n\n光模様の基本色は淡い金色とし、画像1のキャラクター固有色がごく薄く混じる。キャラクターは新しい光に気づき、肩の力を少し抜き、上半身を光の方向へごくわずかに向ける。\n【光を追うカメラ移動】\n光模様が雲の移動に合わせて連続して流れ始め、カメラは光を案内線として追う。カメラ位置はテーブル斜め前から床近くへ滑らかに下がり、光の進行方向へ向いたまま、被写体サイズをカップの近景から衣装の中景へ変える。背景消失軸はテーブル奥から床タイルの斜め奥へ移す。\n\n透明ガラスでは細い透明な反射、紅茶では色のある透過光、白い布では柔らかな拡散光、金属装飾では短い鋭い反射、床タイルでは淡い鏡面光として、素材ごとに描き分ける。\n\n続けて光は白いテーブルから床タイル、衣装の裾や下部装飾、腰周辺の装飾へ連続して移る。カメラは床近くからキャラクターの斜め前へ向いたまま胸の高さまで上がり、被写体サイズを衣装の中景から胸上へ戻す。背景消失軸は床タイルから右上へ伸びる窓枠と屋根骨格へ受け渡す。\n画像1の衣装に裾飾りや腰飾りがない場合は、存在しない装飾を追加せず、実際に存在する衣装の縁、布の重なり、胸元、袖口、髪飾りの順に光を移す。背景の窓枠、花、椅子は緩やかなパララックスを作り、温室の奥行きを保つ。\n\n【表情変化】\n光は顔中央へ入らず、肩、髪の外周、衣装の輪郭だけを淡く縁取る。\n\n表情変化は、最初に左右の瞳の内側に元からあるキャッチライトだけが同時にわずかに明るくなり、次に眉が少し上がり、その後まぶたが柔らかく緩み、続いて口角が自然に上がり、最後に視線を紅茶へ落とす順序で見せる。\nカメラは斜め前の胸上を保ち、画像1の瞳、顔立ち、髪型、髪飾り、衣装の主要装飾を読みやすく見せる。\n\n【カップを持ち上げる動作】\nキャラクターは右手の指でカップの取っ手だけを丁寧につまみ、透明なティーカップをソーサーから静かに持ち上げる。左手は白いテーブルの縁へ軽く添えたままにする。\n\nカップを胸元より少し低い高さで保持すると、紅茶に小さな波紋が生まれ、温室の白い屋根骨格、青空、流れる雲が紅茶の表面へ自然に映り込む。\n【終端カメラ】\n終端のカメラはキャラクターの斜め前、胸の高さから紅茶面と顔へ向く。キャラクターを胸上の大きさで保ち、手前に紅茶面、奥に実像の三分の四顔を置く。背景消失軸は白い屋根骨格が右上へ抜ける方向に揃える。\n\nカメラをわずかに下向きへ傾け、キャラクターが透明なティーカップを胸元より少し低い位置で静かに保持する。手前の紅茶面に温室の青空と、同じキャラクターの三分の四顔が一度だけ明瞭に映る。\n\n紅茶面の顔は光学的な反射像であり、別人、分身、第二の人物ではない。奥の実像と同じ顔立ち、髪型、表情、視線方向を保つ。奥の実像は穏やかな微笑みを維持し、紅茶の波紋が収まり、実像と反射像が同時に澄んだ瞬間に切る。\n\n【禁止事項】\n光点、水平フレア、ガラス反射を顔中央、鼻筋、眉間、瞳の表面へ重ねない。顔を光で隠さない。背景を画像1の背景へ置き換えない。温室以外の場所へ変更しない。文字、字幕、ロゴ、透かしなし。BGMなし、音楽なし、環境音なし、フォーリーなし、音響生成なし。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2082707037256056852",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/haruuraeadss/status/2082707037256056852",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-07-30T05:57:07.164Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082707037256056852.mp4",
    "posterUrl": "/posters/x/2082707037256056852.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でMV寄りの映像も試してみました。\n\n雰囲気めちゃくちゃよくて、アニメMVのような映像ともかなり相性が良い印象です。\n2K・15秒でも150クレジットなので、気軽にいろいろ試せるのも良いですね🎬\n\n@Hailuo_AI　#MiniMaxH3　\nプロンプト👇\n                                                              \n高品質アニメ映像。\n\n【作品トーン】\n透明感のある夏のガラス温室を舞台にした、上品で夢のような午後の叙情アニメ。映像全体は高密度な2D手描きアニメとして統一し、静かな呼吸、光の移動、素材ごとの反射と透過、繊細な表情変化を中心に描く。\n【参照画像の役割】\n画像1のみを使用する。画像1は登場する一人のキャラクターの参照として扱い、顔立ち、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、全身シルエット、雰囲気、キャラクター固有色だけを参照する。\n\n画像1そのものの背景、部屋、家具、建築、ポーズ、手の位置、画角、カメラ距離、フレーミング、構図、照明状態は再現しない。画像1の背景要素を温室内へ持ち込まない。\n画像1と同じ人物を、全編で同じ一人のキャラクターとして厳密に維持する。顔の輪郭、目の形、左右の瞳の特徴、虹彩色、前髪、髪の長さ、髪型、髪飾り、衣装構造、装飾位置、体格、年齢感、キャラクター固有色を変えない。表情、視線、口元、呼吸、肩の力、手の動き、自然な髪と衣装の揺れだけ変化してよい。特徴の混合、顔の平均化、別人化、衣装交換、装飾消失、髪型変更、クローン化、分身、余計な人物の追加は禁止する。\n\n【絵柄固定】\nキャラクター同一性とは別条件として、映像全体を繊細で高密度な2D手描きアニメに固定する。細く淡い寒色系の線、透明感のある多層の色影、柔らかなグラデーション、肌の淡い発光感、髪の繊細なハイライト、瞳の多層反射を維持する。花、ガラス、紅茶、空、雲、白い家具、床タイル、温室建築も同じ密度の手描き筆致で描く。太い輪郭、簡略化されたTVアニメ調、平坦な単層セル影、低密度背景、滑らかなCG、3D化、半写実、実写化、画風混合にしない。\n\n【キャラクターに合わせて変化させる要素】\n場所と出来事は固定し、小物の意匠、アクセント色、光の印象、紅茶の色味、卓上の小さな添え物だけを画像1のキャラクターに合わせて調整する。\n\n温室の白と淡い青を基調とした明るさは維持しながら、画像1の髪色、瞳色、衣装色、装飾色から一色だけを補助アクセントとして抽出する。アクセント色はカップの縁、ソーサーの細い装飾、卓上の花、屈折光のわずかな色味に限定し、背景全体をキャラクター色で塗り替えない。\nティーカップとソーサーの形状は、画像1の衣装や装飾の印象に合わせて、上品、可憐、端正、幻想的などの範囲でわずかに意匠を調整してよい。ただし透明なガラス製であること、カップ一つ、取っ手一つ、ソーサー一枚であることは固定する。\n\n紅茶は透明感のある温かい色を保ち、画像1のキャラクターカラーに合わせて、琥珀色、淡い蜂蜜色、薄い薔薇色、淡い青緑を含む幻想的な茶色などへ穏やかに調整してよい。濁った液体、不透明な液体、蛍光色にはしない。\n卓上の添え物は一つだけ置く。画像1の雰囲気に合わせ、柑橘の一切れ、小さな砂糖菓子、一輪の花、薄い果実片などから最も自然なものを一つ選ぶ。添え物は途中で増殖、移動、変形させない。\n\n【固定舞台】\n舞台は明るい夏のガラス温室。白い窓枠と屋根骨格、澄んだ青空、白い雲、外の緑と花、白い丸テーブル、白い金属椅子、花鉢、淡い青灰色の床タイルを維持する。白と淡い青を基調とし、日差しが作る淡い金色と、画像1から抽出した補助アクセント色だけを使用する。\n\n【小物固定】\n開始状態では、透明なガラスのティーカップ一つに紅茶が入り、誰の手にも保持されず、透明なガラスのソーサー一枚の中央に置かれている。卓上の添え物一つは白い丸テーブルの上に置いたままにする。\n\n後半でキャラクター本人だけが右手でカップの取っ手を持つ。カップ、取っ手、紅茶、ソーサー、添え物を増殖、交換、消失、変形させない。\n\n【開始カメラ】\n最初のカメラは白い丸テーブルの斜め前、テーブル面と同じ低い高さからキャラクターの斜め前へ向く。前景にティーカップとソーサーを大きく置き、奥にキャラクターの胸上と顔を明瞭に見せる。背景消失軸は床タイルと右側の窓枠が温室奥へ伸びる方向に揃える。\n\n紅茶とキャラクターを最初から同じ画面で紹介する。キャラクターは穏やかに呼吸し、髪先、袖、衣装の軽い装飾だけが温室の微風でわずかに揺れる。\n【光の発生】\n屋外の薄い雲が横へ流れて日差しが開き、温室の窓ガラスを通った白い陽光が紅茶を透過する。紅茶と透明ガラスの屈折によって生まれた柔らかな光模様が、ソーサーの縁から白いテーブルへ現れる。\n\n光模様の基本色は淡い金色とし、画像1のキャラクター固有色がごく薄く混じる。キャラクターは新しい光に気づき、肩の力を少し抜き、上半身を光の方向へごくわずかに向ける。\n【光を追うカメラ移動】\n光模様が雲の移動に合わせて連続して流れ始め、カメラは光を案内線として追う。カメラ位置はテーブル斜め前から床近くへ滑らかに下がり、光の進行方向へ向いたまま、被写体サイズをカップの近景から衣装の中景へ変える。背景消失軸はテーブル奥から床タイルの斜め奥へ移す。\n\n透明ガラスでは細い透明な反射、紅茶では色のある透過光、白い布では柔らかな拡散光、金属装飾では短い鋭い反射、床タイルでは淡い鏡面光として、素材ごとに描き分ける。\n\n続けて光は白いテーブルから床タイル、衣装の裾や下部装飾、腰周辺の装飾へ連続して移る。カメラは床近くからキャラクターの斜め前へ向いたまま胸の高さまで上がり、被写体サイズを衣装の中景から胸上へ戻す。背景消失軸は床タイルから右上へ伸びる窓枠と屋根骨格へ受け渡す。\n画像1の衣装に裾飾りや腰飾りがない場合は、存在しない装飾を追加せず、実際に存在する衣装の縁、布の重なり、胸元、袖口、髪飾りの順に光を移す。背景の窓枠、花、椅子は緩やかなパララックスを作り、温室の奥行きを保つ。\n\n【表情変化】\n光は顔中央へ入らず、肩、髪の外周、衣装の輪郭だけを淡く縁取る。\n\n表情変化は、最初に左右の瞳の内側に元からあるキャッチライトだけが同時にわずかに明るくなり、次に眉が少し上がり、その後まぶたが柔らかく緩み、続いて口角が自然に上がり、最後に視線を紅茶へ落とす順序で見せる。\nカメラは斜め前の胸上を保ち、画像1の瞳、顔立ち、髪型、髪飾り、衣装の主要装飾を読みやすく見せる。\n\n【カップを持ち上げる動作】\nキャラクターは右手の指でカップの取っ手だけを丁寧につまみ、透明なティーカップをソーサーから静かに持ち上げる。左手は白いテーブルの縁へ軽く添えたままにする。\n\nカップを胸元より少し低い高さで保持すると、紅茶に小さな波紋が生まれ、温室の白い屋根骨格、青空、流れる雲が紅茶の表面へ自然に映り込む。\n【終端カメラ】\n終端のカメラはキャラクターの斜め前、胸の高さから紅茶面と顔へ向く。キャラクターを胸上の大きさで保ち、手前に紅茶面、奥に実像の三分の四顔を置く。背景消失軸は白い屋根骨格が右上へ抜ける方向に揃える。\n\nカメラをわずかに下向きへ傾け、キャラクターが透明なティーカップを胸元より少し低い位置で静かに保持する。手前の紅茶面に温室の青空と、同じキャラクターの三分の四顔が一度だけ明瞭に映る。\n\n紅茶面の顔は光学的な反射像であり、別人、分身、第二の人物ではない。奥の実像と同じ顔立ち、髪型、表情、視線方向を保つ。奥の実像は穏やかな微笑みを維持し、紅茶の波紋が収まり、実像と反射像が同時に澄んだ瞬間に切る。\n\n【禁止事項】\n光点、水平フレア、ガラス反射を顔中央、鼻筋、眉間、瞳の表面へ重ねない。顔を光で隠さない。背景を画像1の背景へ置き換えない。温室以外の場所へ変更しない。文字、字幕、ロゴ、透かしなし。BGMなし、音楽なし、環境音なし、フォーリーなし、音響生成なし。",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082704405619679353",
    "title": "天然材质巨龙定格动画",
    "titleEn": "Natural-Material Dragon Stop Motion",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《天然材质巨龙定格动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 12 秒 · 960×720。",
    "summaryEn": "An H3 output for Natural-Material Dragon Stop Motion. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 12s · 960×720.",
    "prompt": "A top-down stop-motion style video of four hands assembling a detailed dragon head sculpture on a clean white surface using only natural materials.\nStart with four hands holding and positioning bare brown twigs and branches, carefully arranging them into the basic outline of a dragon’s head, neck, and curved spine with small protruding sticks for spikes.\nNext, hands place large pieces of rough textured tree bark to form the dragon’s face, snout, and jaw, carefully fitting the pieces so the bark grain creates natural scales and ridges. Yellow lichen spots appear on the forehead and cheeks. Small sharp wooden shards are inserted as teeth.\nThen hands add layers of thin bark strips and light-colored wood pieces along the neck to create overlapping scale armor. Fresh green moss is packed into the upper neck and crest area. Dried brown and golden autumn leaves are carefully layered in a flowing mane along the side of the neck.\nHands continue refining: placing two curved wooden sticks as horns, adding small pinecones as decorative spikes, inserting a realistic glass eye into the socket, and tucking pine needles into the mouth and around the head like whiskers and eyebrows.\nFinal shot: the completed dragon head sculpture sits alone on the white background — a fierce yet elegant nature-made dragon with bark scales, moss crest, leaf mane, wooden horns, sharp teeth, and a single glowing eye. Soft even lighting, clean white background, hyper-detailed textures, realistic stop-motion assembly feel, 9-second duration.",
    "promptSourceUrl": "https://x.com/Maercihh/status/2082704405619679353",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Maercihh/status/2082704405619679353",
    "sourceLabel": "X 原帖 · @Maercihh",
    "author": "Maercih",
    "publishedAt": "2026-07-30T05:46:39.733Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082704405619679353.mp4",
    "posterUrl": "/posters/x/2082704405619679353.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "960×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo AI just dropped Minimax H3. 👀\n\nHere's one of my first generations with it.\n\nPrompt:\n\nA top-down stop-motion style video of four hands assembling a detailed dragon head sculpture on a clean white surface using only natural materials.\nStart with four hands holding and positioning bare brown twigs and branches, carefully arranging them into the basic outline of a dragon’s head, neck, and curved spine with small protruding sticks for spikes.\nNext, hands place large pieces of rough textured tree bark to form the dragon’s face, snout, and jaw, carefully fitting the pieces so the bark grain creates natural scales and ridges. Yellow lichen spots appear on the forehead and cheeks. Small sharp wooden shards are inserted as teeth.\nThen hands add layers of thin bark strips and light-colored wood pieces along the neck to create overlapping scale armor. Fresh green moss is packed into the upper neck and crest area. Dried brown and golden autumn leaves are carefully layered in a flowing mane along the side of the neck.\nHands continue refining: placing two curved wooden sticks as horns, adding small pinecones as decorative spikes, inserting a realistic glass eye into the socket, and tucking pine needles into the mouth and around the head like whiskers and eyebrows.\nFinal shot: the completed dragon head sculpture sits alone on the white background — a fierce yet elegant nature-made dragon with bark scales, moss crest, leaf mane, wooden horns, sharp teeth, and a single glowing eye. Soft even lighting, clean white background, hyper-detailed textures, realistic stop-motion assembly feel, 9-second duration.\n\n @Hailuo_AI  #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082700637444452380",
    "title": "悬崖下行至海边",
    "titleEn": "Cliffside Descent to the Ocean",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《悬崖下行至海边》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Cliffside Descent to the Ocean. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "The shot opens tight on a beautiful woman face, wind tearing at her coat as she stands on a jagged cliff. The camera eases over her shoulder and tilts down, following her line of sight. We descend past the crumbling lip into the void, the frame widening to reveal cliffs dropping away. The ocean roars into view, violent surf detonating against rock. The move continues into a high overhead, locking into an aerial top-down that frames the lone figure at the precipice as waves explode below.",
    "promptSourceUrl": "https://x.com/umesh_ai/status/2082700637444452380",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/umesh_ai/status/2082700637444452380",
    "sourceLabel": "X 原帖 · @umesh_ai",
    "author": "Umesh",
    "publishedAt": "2026-07-30T05:31:41.330Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082700637444452380.mp4",
    "posterUrl": "/posters/x/2082700637444452380.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing the same prompt across newer models gives us a better idea of how they are evolving.\n\nThis is one of my favourite prompts!\n\nGenerated using @Hailuo_AI’s newer model with early access.\n\n#MiniMaxH3\n\nPrompt : The shot opens tight on a beautiful woman face, wind tearing at her coat as she stands on a jagged cliff. The camera eases over her shoulder and tilts down, following her line of sight. We descend past the crumbling lip into the void, the frame widening to reveal cliffs dropping away. The ocean roars into view, violent surf detonating against rock. The move continues into a high overhead, locking into an aerial top-down that frames the lone figure at the precipice as waves explode below.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082693906224009447",
    "title": "金铠女剑士岩台动作戏",
    "titleEn": "Armored swordswoman rock-plateau action sequence",
    "model": "MiniMax H3（来源标注）",
    "mode": "FL2VA",
    "summary": "《金铠女剑士岩台动作戏》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Armored swordswoman rock-plateau action sequence. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "@image1 完全按照参考图呈现,不重新设计角色。面部五官比例、金色浮雕铠甲纹样、服装剪裁与巨剑造型,全片保持严格一致,不因动作幅度、镜头角度或光线变化产生偏差。\n\n武器锁定:全片只使用这把巨剑,不出现任何其他武器或道具。\n\n场景:高处开阔的灰白岩石台地,四周散布着大小不一的巨岩,最远处矗立一块屋子大小的黑色巨岩。天空是清透的灰蓝色,浮着几片白云。侧向强光在岩石上投下长影。空气中悬浮着细小浮尘,被光线照出颗粒感;岩缝间偶有暗红余烬般的光点隐约明灭。\n\n剑的机制:剑身表面是焦黑的炭质外壳,壳下裂缝中流动着橙红熔岩,流速与亮度随她的发力实时联动——蓄力时如心跳般规律搏动,全力挥砍前的瞬间由暗转亮。剑尖始终有熔融的橙红液滴缓慢垂落,落地后凝成一小块暗红结晶。每完成一次切割,炭壳裂缝会略微扩大、露出更多内部熔光,全片末段剑刃发光面积明显多于开场,形成清晰的能量攀升弧线。\n\n切割效果铁律:她的剑不是劈开岩石,而是熔穿它。每一道切口必须呈现完整过程——刃口经过处岩面先烧成暗红并向内塌软,断面呈现熔融的橙红发光切面并挂着流动的熔滴,岩块随后才顺着切面滑开分离,分离时切面之间拉出细长的熔融丝线。切割瞬间偶有细小岩屑因高温爆裂点状飞溅,带着橙色光点划出短促弧线。所有发光切口在全片持续发亮不熄灭,断面的橙红光晕实时照亮她的铠甲表面与周边岩壁,形成随光源移动的动态反光;岩石落地时溅出的熔渣与火星留在地面累积。\n\n视觉效果:挥剑时刃后拖出扭曲空气与橙红热痕;高温使剑刃周围空气持续升腾扭曲;熔滴落地激起细小火星与青烟,余烬如萤火虫般在空中缓慢飘升;岩石碎块被斩飞时带着燃烧的边缘旋转;切割瞬间自切口喷出一道细长白炽气流,剑刃全亮时刃线周围溢出轻微白色光晕;她的发丝与衣摆被上升热气不均匀地掀动,呈现不对称的飘动。命中瞬间插入纯白冲击帧,全片共两次(4.5-6秒与10.5-12.5秒)。高速动作段落(1.5-3秒、3-4.5秒、7.5-9秒、10.5-12.5秒)叠加笔刷质感速度线与三重残影,缓慢或静态镜头不套用此效果。\n\n动作原则:视线先到、剑刃后到。挥剑前肩膀先转、腰胯发力、手臂最后到位,收势时手腕再补一记轻甩为刃部增速。每次落剑后剑尖因巨剑重量继续下沉半拍;大力劈砍后她的躯干会顺着剑势多转一点角度才收住,呈现真实惯性与延迟,而非机械式一刀归位。落地时膝盖沉降缓冲、脚尖抓地。她全程双手与单手交替持握,换手时剑柄在掌中转半圈。呼吸与发力同步——蓄力时短促吸气,发力瞬间从齿间压出短促吐息,大招前肩背明显起伏。每一击节奏不完全相同:部分快、脆、一闪而过,部分沉、重、带延迟惯性,避免全片动作节奏单一化。\n\n风格:2.5D剧场版动漫,赛璐璐上色配厚涂光影,线条锐利。冷色调环境(灰白岩石、灰蓝天空)与炽热橙红剑光形成互补色对比。模拟35mm变形宽银幕镜头:横向椭圆散景、浅景深。16:9,4K,24帧,15秒。镜头每1.5秒换一个角度,不重复。\n\n声音:熔岩流动的低沉咕噜声、炭壳开裂的细碎脆响、巨剑破空的沉重呼啸、岩石熔穿时的滋滋声、岩块滑开与落地的闷响、熔滴滴落的轻响、冲刺时靴底碾过碎石的摩擦声、她的呼吸与发力吐息。没有音乐。\n\n0-1.5秒 微距特写,浅景深:镜头贴着剑身缓慢滑过,焦黑炭壳裂缝里橙红熔岩如心跳般规律搏动,一颗熔滴垂落拉长后坠地溅起火星。焦点后拉,她双手握住剑柄,指节逐一收紧,黑色指甲压进缠绳。她垂眼看向第一块巨岩,瞳孔微微收紧锁定目标。\n\n1.5-3秒 侧面平移追踪,与她同速:右脚踏前瞬间脚跟先着地抓稳,腰胯先转带动双臂甩出,巨剑横向扫过第一块巨岩。刃口经过处岩面瞬间烧成暗红塌软,断面亮起熔融橙光,细小岩屑带火星四溅。岩石停半拍后顺切面滑开两半,切面间拉出数条熔丝,橙光映上她的铠甲侧面。她收剑时剑尖因重量继续下沉半拍。\n\n3-4.5秒 低角度仰拍:两块巨岩立于她两侧。她脚尖为轴原地半旋,双手举剑过顶,先是一记快、脆的斩击劈向左侧,再借剑势惯性沉重地回身劈向右侧,一快一重形成节奏对比。两道切口几乎同时亮起,四块岩体外滑砸地,溅起熔渣青烟,几缕余烬缓缓升起。她落定时双膝微屈,发丝被热气不均匀地掀向一侧。\n\n4.5-6秒 极近垂直特写:她双臂完全伸展将剑举过头顶,肩背明显收紧蓄力,剑身裂缝里熔岩流速骤然加快、亮度提升。随即整个躯干下沉发力垂直劈落,牙关咬紧、短促吐息喷出——剑刃自高大立岩顶端一路熔穿至底部,岩体沿切面倒向两侧,切口喷出细长白炽气流。插入纯白冲击帧,镜头随冲击轻微震动。\n\n6-7.5秒 侧向缓慢横移,体积光穿过热气:她剑尖点地稍作停顿,肩膀剧烈起伏,呼出一口白气,重心在两脚间轻轻晃动找回平衡。镜头横移扫过身后——三处切口仍明亮燃烧,熔滴缓缓垂落,地面已积起暗红结晶与青烟,橙光在她脸颊与铠甲上跳动反射。剑柄在掌中转半圈换到另一只手。\n\n7.5-9秒 高速环绕加速:数块小岩石被她一脚踢向空中。她双脚蹬地纵身跃起,膝盖微收加快转速,空中旋身一圈连续挥剑四次,每一击手臂完全伸展并在末端补一记手腕轻甩——每块岩石被熔穿的瞬间亮起橙色断面,燃烧的边缘",
    "promptSourceUrl": "https://x.com/Cia0_exe/status/2082694269329195217",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Cia0_exe/status/2082693906224009447",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-07-30T05:04:56.482Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082693906224009447.mp4",
    "posterUrl": "/posters/x/2082693906224009447.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hi, @Hailuo_AI love your new model!\n\nThanks for the early access to MiniMax H3.\n\ncombined with reference images from Midjourney V8.2 for testing AI action film scenes\n\nit delivers great cinematic camera angles and smooth movement.\n\nusing Chinese prompts in comments\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082684613475082714",
    "title": "混凝土广场尖翻落地",
    "titleEn": "Concrete-Plaza Kickflip Drop",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《混凝土广场尖翻落地》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 10 秒 · 1280×720。",
    "summaryEn": "An H3 output for Concrete-Plaza Kickflip Drop. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 10s · 1280×720.",
    "prompt": "プロンプト\n\nSCENE CONTEXT Late afternoon, empty two-level concrete plaza. A young woman skateboarder rolls along the raised upper deck to its edge and launches off the TOP of a 10-step stair set with a kickflip, clearing the whole drop in one jump and landing on the lower plaza below. She travels downward from takeoff to landing.  ACTIVE REFERENCES @image1 — young woman skateboarder, long straight black hair with one bright cobalt-blue streak falling on the left side of her face, blunt fringe across the eyebrows, thin black winged eyeliner, freckles over both cheeks. 100% matches the reference.  LOCATION MAP Two levels. She rides the raised UPPER deck: pale polished concrete with black urethane scuff arcs. The upper deck ends at a hard edge, and that edge is the top step of a 10-step stair set descending away from her, chipped steel nosing, matte black handrail on the left. The LOWER plaza sits one human height below, flat open concrete, and is the landing. Beyond it: shuttered glass frontage receding into depth, haze 15% at 30 meters. Low sun at the far left, 15 degrees up, shadows raking left to right. Sunlit concrete warm bone-grey, stair treads cool blue-grey in shadow; the cobalt streak and the handrail carry the only saturated notes.  FIRST FRAME / BLOCKING Frame one is already moving: she is mid-push on the upper deck, right foot swinging back to the tail, board rolling camera-left to camera-right, camera beside her. Past her the deck edge cuts across frame and the lower plaza opens below it, so the ground visibly drops away in the direction she is heading.  FORMAT MODE Timed multishot, five beats. Cuts land only at 2.4s, 3.8s, 7.0s and 9.0s and the camera does not cut on its own.  OPTICS CUT 1 — LENS LOCK 180 degrees fisheye, barrel distortion, WS at wheel height. CUT 2 — LENS LOCK 84 degrees wide rectilinear, WS angled down over the edge. CUT 3 — LENS LOCK 47 degrees neutral, WS full profile of the descent. CUT 4 — LENS LOCK 63 degrees observational, MS to WS low on the landing. CUT 5 — LENS LOCK 18 degrees portrait compression, MCU. LENS CHECK at each cut: FOV as stated, no drift mid-segment.  CAMERA Operator on her right flank, the shadow side, half a beat behind her. CUT 1 — upper deck, glides at 22 km/h level with the trucks, 20 cm ride height, 1 to 2 cm handheld tremor. CUT 2 — upper deck behind her right shoulder, chest height, tilted DOWN 25 degrees so the stair descent and the landing fill the bottom half of frame. CUT 3 — right of the stairs at top-step height, tracking the descent at 3 km/h lateral, focus on the board. CUT 4 — lower plaza, 30 cm off the ground, tilted UP 20 degrees at the stairs above, rising 40 cm as she rolls out. CUT 5 — eye level, slow push from chest to face at 2 km/h.  ACTION 0.0s to 2.4s — CUT 1. On the upper deck she pushes twice, back foot slapping the deck and returning to the tail, rolling at 22 km/h toward the edge. Wheels rumble over the seams. The lower plaza is visible dropping away ahead of her. 2.4s HARD CUT 2.4s to 3.8s — CUT 2. From behind her, the 10-step drop and the landing lie in the lower half of frame. She sinks into a deep crouch, eyes down on the landing, slides her front foot to the bolts and snaps the tail down. The board pops and she leaves the top edge into open air above the steps, the ground now far beneath her feet. 3.8s HARD CUT 3.8s to 7.0s — CUT 3. Full profile of the descent, slow motion at 40 percent speed. She enters at the upper left of frame at the top step and travels down and to the right on a falling arc, the 10 steps passing beneath her, the landing waiting in the bottom right. Her front foot flicks off the nose edge and the board rotates one complete 360-degree roll along its long axis as she falls, knees tucked, arms wide. The griptape comes back under her feet and she catches it over the bolts, wheels aimed down. 7.0s HARD CUT 7.0s to 9.0s — CUT 4. Real time. She drops in from the top of frame, out of the air, and all four wheels bite the lower plaza at once with a hard slap. Her knees compress deep and absorb the drop, and she rolls away camera-right at 20 km/h, the stair set standing above and behind her. 9.0s HARD CUT 9.0s to 10.0s — CUT 5. Real time. Rolling out, she turns her head back over her left shoulder and looks UP at the stairs she came off, holds one beat, faces forward.  PERFORMANCE Pore-level skin realism, living eyes with a hard catch-light from the low sun. On the run-up her jaw is set, eyes down on the landing far below. On landing the impact travels through her shoulders, she blinks once, then exhales and her mouth lifts barely at one corner.  PHYSICS Real board mass: the deck flexes and rebounds on landing, wheels squash against the concrete. From the pop onward she only loses altitude, falling on a clean gravity parabola while forward inertia carries her out over the steps, touching down one human height lower and several meters further out than her takeoff point, gaining speed as she falls. The kickflip rotates at a constant rate. Contact shadows stay welded to the wheels on both levels and detach in the air, hair and fabric lagging above her as she drops.  LIGHTING Low sun as key from the far left at 15 degrees elevation, raking across the upper deck and rimming her left edge, the cobalt streak glowing where light passes through it. The stair treads sit in their own soft shadow while the lower plaza stays open and sunlit, so the two levels read as clearly separated heights. White balance 5600K, fixed across all beats.  WARDROBE Oversized faded black cotton tee flapping upward as she falls, baggy light-wash denim jeans with frayed hems over the heels, black canvas low-top skate shoes with the suede toe worn through from griptape.  AUDIO Lofi hiphop bed at 72 BPM throughout: dusty vinyl crackle, soft boom-bap kick and brushed snare, mellow jazz piano loop, warm bass. On top: wheels rumbling over seams, the crack of the tail on the top step, the board fluttering as it flips, a quiet beat of air as she falls, a hard slap as the wheels hit the lower plaza.  STYLE Photoreal live-action skate promo video, professional street skating footage, fine film grain, real motion blur on the wheels and the flipping board.  OUTPUT SETTINGS 8K, anamorphic 2.39:1, 24 fps. CUTS 1, 2, 4 and 5 real time. CUT 3 slow motion at 40 percent, constant for the whole beat.  POSITIVE LOCKS She starts on the upper deck and finishes on the lower plaza: takeoff at the top of the stairs, landing at the bottom, one human height below. Her motion is downward through the whole flight, entering each airborne frame high and leaving it low. The stair set stays exactly 10 steps and descends away from her direction of travel. The board completes one full 360-degree kickflip rotation and she lands with both feet over the bolts, rolling away upright past the bottom step. Screen direction stays camera-left to camera-right. Same rider, same face, same cobalt-blue streak, same wardrobe and same board in all five beats.",
    "promptSourceUrl": "https://x.com/eijo_AIart/status/2082684616155271248",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/eijo_AIart/status/2082684613475082714",
    "sourceLabel": "X 原帖 · @eijo_AIart",
    "author": "eijo(SAT577)",
    "publishedAt": "2026-07-30T04:28:00.918Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082684613475082714.mp4",
    "posterUrl": "/posters/x/2082684613475082714.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Action & VFX",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailueの最新モデル「Minimaxi H3」でキックフリップをキメる！\n\n今はアーリーアクセスで使わせてもらってますが、本リリースが楽しみですね☺️\n\nプロンプトは👇\n\n@Hailuo_AI, #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082669221222207488",
    "title": "1970 年代涩谷都市传说恐怖片",
    "titleEn": "1970s Shibuya Urban-Legend Horror",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《1970 年代涩谷都市传说恐怖片》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for 1970s Shibuya Urban-Legend Horror. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "😱👇\n\n【風格】\n1970年代日本都市傳說恐怖電影（Japanese Urban Legend Horror Movie），經典日系恐怖片風格，昭和時代澀谷街景，高密度群眾演出，電影級分鏡，強烈戲劇張力，歡樂復古流行音樂逐漸扭曲變調為詭異恐怖配樂，緊湊快節奏剪輯，陰森都市怪談氛圍，9:16直式畫面。\n\n【音樂】\n開場：歡樂復古1970年代日本流行樂。\n中段：音樂逐漸失真、降速、出現不和諧音。\n結尾：低沉詭異弦樂、女性怪笑聲、電子故障音（Glitch）。\n\n【總時長】15秒，中文對話\n\n【場景】1970年代東京澀谷區熱鬧商店街 → 暗巷 → 都市傳說恐怖事件\n\n【鏡頭1（00:00 - 00:03）】\n熱鬧的1970年代澀谷街頭。  大量打扮前衛新潮的年輕男女逛街聊天。\n地面散落報紙、飲料罐與垃圾。  突然一名穿著女僕咖啡廳制服的粉紅色長髮可愛女店員從狹窄暗巷衝出來。  她身穿低胸黑色與白色相間的女僕裝，搭配白色長手套、黑色網襪與高跟鞋，臉色慘白、滿臉恐懼。  一邊奔跑一邊尖叫：\n「救命啊！有鬼！」下一秒她重重跌倒在地，當場昏迷。  周圍路人紛紛停下腳步圍觀。\n【鏡頭2（00:03 - 00:06）】\n一名醉醺醺的平頭流氓推開圍觀群眾。  肩膀露出龍紋刺青。  他粗暴地大喊：\n「滾開！」人群驚慌後退。  流氓蹲到女店員旁邊。  不耐煩地用力搖晃她肩膀：\n「小姐！妳醒醒！」\n【鏡頭3（00:06 - 00:08）】\n女店員虛弱睜開眼睛。  顫抖地說：\n「有鬼……」她艱難抬起左手。  指向旁邊漆黑暗巷。  接著再次失去意識。  流氓順勢轉頭望向暗巷。  背景音樂突然停止。\n【鏡頭4（00:08 - 00:10）】\n暗巷深處。  垃圾堆旁飄浮著一名白衣長黑髮女鬼。  長髮遮住半張臉。  蒼白雙眼死死瞪著鏡頭。  周圍空氣微微扭曲。  遠方傳來低沉詭異笑聲。\n【鏡頭5（00:10 - 00:12）】\n流氓臉部超特寫。  他震驚得不自覺張開嘴巴。  頭緩緩歪向右側。  醉意瞬間消失。  額頭冒出冷汗。\n【鏡頭6（00:12 - 00:14）】\n女鬼也做出完全相同的動作。  嘴巴慢慢張開。  頭歪向右側。  接著發出尖銳而詭異的笑聲：\n「呵呵呵呵呵……」下一秒。  女鬼突然高速朝鏡頭飛撲而來。  長髮與白衣瞬間佔滿整個畫面。\n【鏡頭7（00:14 - 00:15）】\n畫面突然發生強烈電子故障。  Glitch！  黑白噪訊瘋狂閃爍。  尖銳雜音爆發。  畫面瞬間切成全黑。  只剩女鬼最後一聲詭異笑聲在黑暗中迴盪。",
    "promptSourceUrl": "https://x.com/drjoetw/status/2082669221222207488",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/drjoetw/status/2082669221222207488",
    "sourceLabel": "X 原帖 · @drjoetw",
    "author": "The Anxious Mind",
    "publishedAt": "2026-07-30T03:26:51.119Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082669221222207488.mp4",
    "posterUrl": "/posters/x/2082669221222207488.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Horror",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "He was ready to save her... but no one warned him about what was waiting. The rescue mission took an unexpected turn. Would you stay or run? 😨🔥\n(created using @Hailuo_AI early access to #MiniMaxH3) \nPrompt below 😱👇\n\n【風格】\n1970年代日本都市傳說恐怖電影（Japanese Urban Legend Horror Movie），經典日系恐怖片風格，昭和時代澀谷街景，高密度群眾演出，電影級分鏡，強烈戲劇張力，歡樂復古流行音樂逐漸扭曲變調為詭異恐怖配樂，緊湊快節奏剪輯，陰森都市怪談氛圍，9:16直式畫面。\n\n【音樂】\n開場：歡樂復古1970年代日本流行樂。\n中段：音樂逐漸失真、降速、出現不和諧音。\n結尾：低沉詭異弦樂、女性怪笑聲、電子故障音（Glitch）。\n\n【總時長】15秒，中文對話\n\n【場景】1970年代東京澀谷區熱鬧商店街 → 暗巷 → 都市傳說恐怖事件\n\n【鏡頭1（00:00 - 00:03）】\n熱鬧的1970年代澀谷街頭。  大量打扮前衛新潮的年輕男女逛街聊天。\n地面散落報紙、飲料罐與垃圾。  突然一名穿著女僕咖啡廳制服的粉紅色長髮可愛女店員從狹窄暗巷衝出來。  她身穿低胸黑色與白色相間的女僕裝，搭配白色長手套、黑色網襪與高跟鞋，臉色慘白、滿臉恐懼。  一邊奔跑一邊尖叫：\n「救命啊！有鬼！」下一秒她重重跌倒在地，當場昏迷。  周圍路人紛紛停下腳步圍觀。\n【鏡頭2（00:03 - 00:06）】\n一名醉醺醺的平頭流氓推開圍觀群眾。  肩膀露出龍紋刺青。  他粗暴地大喊：\n「滾開！」人群驚慌後退。  流氓蹲到女店員旁邊。  不耐煩地用力搖晃她肩膀：\n「小姐！妳醒醒！」\n【鏡頭3（00:06 - 00:08）】\n女店員虛弱睜開眼睛。  顫抖地說：\n「有鬼……」她艱難抬起左手。  指向旁邊漆黑暗巷。  接著再次失去意識。  流氓順勢轉頭望向暗巷。  背景音樂突然停止。\n【鏡頭4（00:08 - 00:10）】\n暗巷深處。  垃圾堆旁飄浮著一名白衣長黑髮女鬼。  長髮遮住半張臉。  蒼白雙眼死死瞪著鏡頭。  周圍空氣微微扭曲。  遠方傳來低沉詭異笑聲。\n【鏡頭5（00:10 - 00:12）】\n流氓臉部超特寫。  他震驚得不自覺張開嘴巴。  頭緩緩歪向右側。  醉意瞬間消失。  額頭冒出冷汗。\n【鏡頭6（00:12 - 00:14）】\n女鬼也做出完全相同的動作。  嘴巴慢慢張開。  頭歪向右側。  接著發出尖銳而詭異的笑聲：\n「呵呵呵呵呵……」下一秒。  女鬼突然高速朝鏡頭飛撲而來。  長髮與白衣瞬間佔滿整個畫面。\n【鏡頭7（00:14 - 00:15）】\n畫面突然發生強烈電子故障。  Glitch！  黑白噪訊瘋狂閃爍。  尖銳雜音爆發。  畫面瞬間切成全黑。  只剩女鬼最後一聲詭異笑聲在黑暗中迴盪。\n\n#AIDrama #AIFilm #AIMovie #AIVideo #Microdrama",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082667261718872509",
    "title": "奢华翡翠项链广告分镜",
    "titleEn": "Luxury Emerald Necklace Storyboard",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《奢华翡翠项链广告分镜》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×960。",
    "summaryEn": "An H3 output for Luxury Emerald Necklace Storyboard. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×960.",
    "prompt": "TITLE: Luxury Emerald Necklace Commercial Storyboard\n\nFORMAT:\n• Single-page premium storyboard\n• 3:4 portrait ratio\n• Editorial presentation board\n• 8 cinematic frames arranged in a clean luxury grid\n• Ultra-high-resolution, 8K\n\nHEADER:\n• Elegant serif typography\n• White, emerald green, champagne gold, and black luxury palette\n• Premium information cards\n• Luxury jewelry icons\n• Section: \"Why This Style Works\"\n\nSTORYBOARD SCENES:\n1. A handcrafted emerald necklace resting inside an elegant velvet jewelry box on polished black marble under a dramatic spotlight.\n2. The jewelry box slowly opening as warm golden light reveals the brilliant emerald necklace.\n3. Extreme macro close-up of the emerald gemstone and diamond accents with breathtaking clarity, crystal reflections, and luxurious sparkle.\n4. A graceful woman preparing for an exclusive gala, delicately lifting the necklace from the box.\n5. The emerald necklace being elegantly fastened around her neck, emphasizing craftsmanship and sophistication.\n6. Emotional cinematic close-up highlighting the necklace glowing against flawless skin with soft luxury lighting.\n7. High-fashion editorial portrait featuring the model in an elegant couture gown, with the emerald necklace as the unmistakable focal point.\n8. Hero product shot of the emerald necklace displayed on a glossy black reflective pedestal surrounded by floating emerald particles, subtle golden dust, dramatic rim lighting, and premium luxury branding composition.\n\nSTYLE:\nInspired by world-class luxury jewelry campaigns, premium fashion editorial, soft cinematic lighting, refined elegance, crystal-clear gemstone reflections, ultra-realistic textures, shallow depth of field, award-winning commercial photography, luxury branding aesthetic, 8K masterpiece.",
    "promptSourceUrl": "https://x.com/itxabdullaa/status/2082667261718872509",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/itxabdullaa/status/2082667261718872509",
    "sourceLabel": "X 原帖 · @itxabdullaa",
    "author": "ÀBDŪLLÂH",
    "publishedAt": "2026-07-30T03:19:03.937Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082667261718872509.mp4",
    "posterUrl": "/posters/x/2082667261718872509.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×960",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made this luxury emerald jewelry commercial using GPT Image 2 × MiniMax H3 with Omni Reference on @Hailuo_AI \n\nFrom elegant macro gemstone shots to a high-end fashion editorial aesthetic, every frame was crafted to feel like a luxury brand campaign. #MiniMaxH3\n\nWhat do you think—does this feel like a real luxury jewelry commercial?\n\nPrompt below 👇🏻\n\n TITLE: Luxury Emerald Necklace Commercial Storyboard\n\nFORMAT:\n• Single-page premium storyboard\n• 3:4 portrait ratio\n• Editorial presentation board\n• 8 cinematic frames arranged in a clean luxury grid\n• Ultra-high-resolution, 8K\n\nHEADER:\n• Elegant serif typography\n• White, emerald green, champagne gold, and black luxury palette\n• Premium information cards\n• Luxury jewelry icons\n• Section: \"Why This Style Works\"\n\nSTORYBOARD SCENES:\n1. A handcrafted emerald necklace resting inside an elegant velvet jewelry box on polished black marble under a dramatic spotlight.\n2. The jewelry box slowly opening as warm golden light reveals the brilliant emerald necklace.\n3. Extreme macro close-up of the emerald gemstone and diamond accents with breathtaking clarity, crystal reflections, and luxurious sparkle.\n4. A graceful woman preparing for an exclusive gala, delicately lifting the necklace from the box.\n5. The emerald necklace being elegantly fastened around her neck, emphasizing craftsmanship and sophistication.\n6. Emotional cinematic close-up highlighting the necklace glowing against flawless skin with soft luxury lighting.\n7. High-fashion editorial portrait featuring the model in an elegant couture gown, with the emerald necklace as the unmistakable focal point.\n8. Hero product shot of the emerald necklace displayed on a glossy black reflective pedestal surrounded by floating emerald particles, subtle golden dust, dramatic rim lighting, and premium luxury branding composition.\n\nSTYLE:\nInspired by world-class luxury jewelry campaigns, premium fashion editorial, soft cinematic lighting, refined elegance, crystal-clear gemstone reflections, ultra-realistic textures, shallow depth of field, award-winning commercial photography, luxury branding aesthetic, 8K masterpiece.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082659300090994748",
    "title": "无菌实验室收容失效",
    "titleEn": "Containment Breach in a Sterile Lab",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《无菌实验室收容失效》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1678×720。",
    "summaryEn": "An H3 output for Containment Breach in a Sterile Lab. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1678×720.",
    "prompt": "Slow-motion handheld close-up, 16:9. Shaky organic handheld shot shoving in close on a researcher in a pressurized white hazmat suit, faceplate fogged and cracking, stumbling backward across a gleaming sterile lab floor as a glistening mutant creature — too many jointed limbs, translucent skin pulsing with visible organs, a maw that splits wrong-way-wide — lunges from an overturned containment tank, wet strands of tissue trailing through the air behind it. Around him the team scatters in chaos: a woman in a matching suit shoves past his shoulder dragging a fallen colleague by the arm, two men wrestle a heavy steel door against something pushing back from the other side, someone is dragged backward across a workstation beside him, knocking a spray of shattered glassware and spilled reagent vials into the air, more suited figures stumble and scramble through the background past racks of specimen jars cracking under sudden impact. In slow motion glass, fluid, and torn suit fabric explode across the whole frame, droplets and shards suspended everywhere, catching the cold light. Dynamic jostling handheld movement at chest height in the thick of the chaos, camera nearly clipped by a whipping limb. Sterile unreal palette — clinical white, cold surgical blue, sickly bio-luminescent green glow bleeding from the creature's flesh, harsh alarm-red strobing at the edges of frame. All faces anatomically correct behind fogged visors, stable, realistic, genuine terror, no warping. Heavy coarse analog 35mm film grain baked into every frame, thick and consistent, grain crawling over highlights and shadows, like a scanned celluloid film print, never clean, never digital-looking. Flat overhead cleanroom lighting cut through by pulsing red alarm strobes, dramatic slow-motion intensity, cinematic photorealistic, ultra-detailed.",
    "promptSourceUrl": "https://x.com/plasm0/status/2082659449747939485",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/plasm0/status/2082659300090994748",
    "sourceLabel": "X 原帖 · @plasm0",
    "author": "ρŁ𝐀𝔰Ｍʘ",
    "publishedAt": "2026-07-30T02:47:25.737Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082659300090994748.mp4",
    "posterUrl": "/posters/x/2082659300090994748.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1678×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Horror",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.0 vs. MiniMax H3\nsame prompt (full prompt in comments)\n\nThis one is Seedance 2.0.  \nBoth look great, but I see more details and better audio in SD2 👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082658222247137433",
    "title": "暗黑奇幻酒馆斗殴",
    "titleEn": "Dark-Fantasy Tavern Fight",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《暗黑奇幻酒馆斗殴》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Dark-Fantasy Tavern Fight. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Style: Hyper-realistic dark fantasy tavern cinematic, grounded physical combat, realistic body momentum, medieval atmosphere, gritty lighting, practical effects, handheld cinematic camera, physically accurate movement, ultra-detailed textures, realistic collisions and environmental interaction.\n\nDuration: 15 seconds\n\nAspect Ratio: 16:9\n\nIMPORTANT:\n\nKeep the SAME characters fully consistent with the reference: 📷Image1\n\n— Nyssa: athletic woman, dark curly hair tied back, bronze leather armor, cloth wraps, sword on hip\n\n— Grok: massive muscular green orc, scarred skin, fur armor, heavy axes, large physical weight\n\nCombat MUST obey realistic physics:\n\n— no floating\n\n— no anime flips\n\n— realistic inertia and recovery\n\n— believable weight transfer\n\n— realistic impacts and exhaustion\n\n— grounded footwork\n\n— environment reacts physically to impacts\n\n[00:00-00:02]\n\nWide cinematic shot inside a crowded medieval tavern filled with drunk mercenaries, candles, smoke, spilled ale, and loud cheering. Nyssa walks through the tavern cautiously while patrons stare. Grok sits at a massive wooden table drinking heavily. Deep bass-heavy tavern music and crowd ambience.\n\n[00:02-00:04]\n\nClose-up tension sequence:\n\n— Nyssa locks eyes with Grok\n\n— Grok slowly stands, towering over everyone\n\n— benches scrape across the floor\n\n— mugs shake from his weight\n\n— crowd forms a fighting circle cheering loudly\n\nHandheld camera movement feels realistic and grounded.\n\n[00:04-00:07]\n\nThe fight erupts violently. Grok swings a heavy punch with believable momentum. Nyssa narrowly dodges while stumbling realistically into a wooden table. Wood cracks and splinters physically on impact. She counters with fast grounded strikes to Grok’s ribs and legs. Crowd roars and throws coins and mugs into the air.\n\n[00:07-00:10]\n\nCombat intensifies through the tavern:\n\n— Grok grabs Nyssa and throws her across a table\n\n— table explodes realistically beneath her weight\n\n— Nyssa rolls across the floor recovering naturally\n\n— Grok charges heavily, smashing furniture\n\n— patrons jump away realistically\n\nNo exaggerated movement, only physically believable combat.\n\n[00:10-00:12]\n\nNyssa uses speed and positioning intelligently. She dodges another heavy attack causing Grok to crash into a support beam. Dust and debris fall from the ceiling. She climbs briefly onto the bar counter and leaps down with controlled realistic momentum, locking Grok’s arm and using leverage to throw him off balance.\n\n[00:12-00:15]\n\nFinal cinematic climax. Nyssa tackles Grok to the ground hard onto broken wooden debris. The floor shakes from the impact. She pins him down on top of him while holding a blade near his throat. Grok struggles realistically beneath her massive weight difference and exhaustion. Tavern crowd erupts cheering wildly, slamming mugs against tables. Camera slowly pushes inward on Nyssa breathing heavily while sweat, dirt, and candlelight flicker across both fighters.\n\nAudio:\n\nHeavy cinematic tavern battle soundtrack with deep drums, loud cheering crowds, breaking wood, metal impacts, realistic body hits, roaring fire ambience, mugs crashing, heavy footsteps, and gritty medieval atmosphere.\n\nNegative prompts:\n\nNo floating, no anime combat, no superhero physics, no unrealistic flips, no weightless motion, no blurry characters, no low detail, no cartoon style, no slow reactions, no clipping through objects.",
    "promptSourceUrl": "https://x.com/craftian_keskin/status/2082658222247137433",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/craftian_keskin/status/2082658222247137433",
    "sourceLabel": "X 原帖 · @craftian_keskin",
    "author": "Keskin",
    "publishedAt": "2026-07-30T02:43:08.759Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082658222247137433.mp4",
    "posterUrl": "/posters/x/2082658222247137433.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3 Video Model is here! \n\nHere is the Prompt for my Orc Warrior vs Human Rogue: \n\nStyle: Hyper-realistic dark fantasy tavern cinematic, grounded physical combat, realistic body momentum, medieval atmosphere, gritty lighting, practical effects, handheld cinematic camera, physically accurate movement, ultra-detailed textures, realistic collisions and environmental interaction.\n\nDuration: 15 seconds\n\nAspect Ratio: 16:9\n\nIMPORTANT:\n\nKeep the SAME characters fully consistent with the reference: 📷Image1\n\n— Nyssa: athletic woman, dark curly hair tied back, bronze leather armor, cloth wraps, sword on hip\n\n— Grok: massive muscular green orc, scarred skin, fur armor, heavy axes, large physical weight\n\nCombat MUST obey realistic physics:\n\n— no floating\n\n— no anime flips\n\n— realistic inertia and recovery\n\n— believable weight transfer\n\n— realistic impacts and exhaustion\n\n— grounded footwork\n\n— environment reacts physically to impacts\n\n[00:00-00:02]\n\nWide cinematic shot inside a crowded medieval tavern filled with drunk mercenaries, candles, smoke, spilled ale, and loud cheering. Nyssa walks through the tavern cautiously while patrons stare. Grok sits at a massive wooden table drinking heavily. Deep bass-heavy tavern music and crowd ambience.\n\n[00:02-00:04]\n\nClose-up tension sequence:\n\n— Nyssa locks eyes with Grok\n\n— Grok slowly stands, towering over everyone\n\n— benches scrape across the floor\n\n— mugs shake from his weight\n\n— crowd forms a fighting circle cheering loudly\n\nHandheld camera movement feels realistic and grounded.\n\n[00:04-00:07]\n\nThe fight erupts violently. Grok swings a heavy punch with believable momentum. Nyssa narrowly dodges while stumbling realistically into a wooden table. Wood cracks and splinters physically on impact. She counters with fast grounded strikes to Grok’s ribs and legs. Crowd roars and throws coins and mugs into the air.\n\n[00:07-00:10]\n\nCombat intensifies through the tavern:\n\n— Grok grabs Nyssa and throws her across a table\n\n— table explodes realistically beneath her weight\n\n— Nyssa rolls across the floor recovering naturally\n\n— Grok charges heavily, smashing furniture\n\n— patrons jump away realistically\n\nNo exaggerated movement, only physically believable combat.\n\n[00:10-00:12]\n\nNyssa uses speed and positioning intelligently. She dodges another heavy attack causing Grok to crash into a support beam. Dust and debris fall from the ceiling. She climbs briefly onto the bar counter and leaps down with controlled realistic momentum, locking Grok’s arm and using leverage to throw him off balance.\n\n[00:12-00:15]\n\nFinal cinematic climax. Nyssa tackles Grok to the ground hard onto broken wooden debris. The floor shakes from the impact. She pins him down on top of him while holding a blade near his throat. Grok struggles realistically beneath her massive weight difference and exhaustion. Tavern crowd erupts cheering wildly, slamming mugs against tables. Camera slowly pushes inward on Nyssa breathing heavily while sweat, dirt, and candlelight flicker across both fighters.\n\nAudio:\n\nHeavy cinematic tavern battle soundtrack with deep drums, loud cheering crowds, breaking wood, metal impacts, realistic body hits, roaring fire ambience, mugs crashing, heavy footsteps, and gritty medieval atmosphere.\n\nNegative prompts:\n\nNo floating, no anime combat, no superhero physics, no unrealistic flips, no weightless motion, no blurry characters, no low detail, no cartoon style, no slow reactions, no clipping through objects.\n\n@Hailuo_AI  #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082652041709773075",
    "title": "奇幻学院分享面包场景",
    "titleEn": "Fantasy Academy Bread-Sharing Scene",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《奇幻学院分享面包场景》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Fantasy Academy Bread-Sharing Scene. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "高品質アニメ映像。世界観は、上品な現代ファンタジー学園を舞台にした2D手描きの日常会話アニメ。表情設計は必ず、目 → 眉 → まぶた → 口 → 視線、の順で行う。\n\n【参照画像と人物固定】\n画像1は人物Aの参照画像。画像1に描かれたキャラクターを、全編で同じ人物として維持する。画像1は、人物Aの顔、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、シルエット、雰囲気、キャラクターカラーの参照にのみ使用する。\n画像2は人物Bの参照画像。画像2に描かれたキャラクターを、全編で同じ人物として維持する。画像2は、人物Bの顔、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、シルエット、雰囲気、キャラクターカラーの参照にのみ使用する。\n\n参照画像そのものの背景、部屋、家具、ポーズ、画角、フレーミング、光源配置は再現しない。変化してよいのは、表情、視線、口元、姿勢、呼吸、自然な髪と衣装の揺れだけ。\n\n人物Aと人物Bの特徴を混ぜない。顔の平均化、衣装交換、髪型交換、別人化、クローン化、分身、余計な人物の追加は禁止。\n\n画像2に武器や手持ち品が含まれていても、人物Bの両手には何も持たせない。画像1に非戦闘用の装飾や小型バッグがある場合のみ、人物Aに自然に残してよい。ただしバッグは身体側に固定し、手で持つ、開ける、動作に使うことは禁止。武器、大型装備、発光する戦闘装備は持たせない。\n\n【小物・色味・雰囲気の調整】\n場所、座る位置、会話内容、パンを分ける動作、カメラ構成は固定する。そのうえで、人物Aと人物Bの参照画像に合わせて、髪飾り、ブローチ、リボン、バッグなどの装飾、小物、配色、光の印象、背景の差し色を自然に調整する。\n\n全体の線色、影色、背景の差し色、午後光の色温度は、人物Aと人物Bのキャラクターカラーと雰囲気に合わせて調整する。ただし、明るく上品で清潔な現代ファンタジー学園の日常感は維持し、暗い戦闘空間、ホラー、退廃、極端な逆光にはしない。\n【絵柄固定】\n細く繊細な色付き線画、透明感のある二層から三層の色影、柔らかなグラデーション。彩度は低くしすぎず、参照人物の固有色を髪、瞳、衣装、反射光、背景の差し色へ自然に反映する。\n\n瞳は宝石のような虹彩の奥行きと複数の小さなキャッチライト、髪は真珠のような多層ハイライト。髪、まつ毛、レース、刺繍、髪飾り、装飾、衣服の皺を高密度に描く。布は上質なマット素材、レース、金属、宝石、リボンはそれぞれ異なる繊細な質感で描き分ける。\n\n背景も同じ2D手描き筆致で、クリーム色の壁、木製ベンチ、植栽、遠景の校舎を十分な密度で描く。太線、厚輪郭、簡略TVアニメ化、平坦な単層セル塗り、低密度背景、滑らかな3D素材、半写実、実写質感、強いくすみ、生成AI平均顔、画風混合は禁止。\n【固定場所と人物配置】\n放課後の学園中庭にある木製ベンチ。人物Bは画面左、人物Aは画面右。二人は隣り合って座る。\n\n人物Aだけが、薄い紙包みに入った丸いあんパン一個を右手で下側から持つ。開始時のパンは完全な丸形で、歯形も欠けもない。指は紙包みの下側に置き、食べる位置から離す。パン、紙包み、手指を増やさない。\n\n【Cut 1】\nカメラ位置はベンチ正面中央。カメラ高は座った二人の目線の高さ。正面から二人を胸上で同じ大きさに収め、背景の消失軸は二人の間の奥へ抜ける。目的は、二人の関係とパンの持ち主が人物Aであることを示すこと。\n\n人物Bはパンを見たあと人物Aへ視線を戻し、両手を胸元で合わせ、身体ごと少し近づく。瞳を輝かせ、眉を上げ、まぶたを期待で開き、甘える口元を作り、人物Aへ視線を合わせる。\n人物Bだけが明るく甘い声で言う。\n「そのアンパン、一口ちょうだい。」\n\n短い止めを入れ、さらに甘えて言う。\n「お願い、一生のお願い！」\n\n人物Aは話さず、あきれながらも優しい目で人物Bを見る。\n【Cut 2】\nカメラ位置は人物Aの斜め前。カメラ高は二人の目線の高さ。人物Aの肩越しから人物Bへ向き、二人の顔とパンを中近景で見せる。背景の消失軸は右奥へ変える。目的は、返答、差し出す手、一口だけかじる動作を一続きで見せること。\n\n人物Aは瞳を柔らかくし、眉を少し下げ、まぶたを穏やかに細め、困った笑顔の口元を作り、人物Bへ視線を合わせる。\n\n小さく息を吐き、短い止めを入れ、人物Aだけが落ち着いた優しい声で言う。\n「もう、しょうがないなー。ひとくちだけだよ」\n言い終わりながら、人物Aは右手の一個のあんパンを人物Bの口元へ差し出す。持ち主は人物Aのままで、手渡さない。\n\n接触前：パンは完全な丸形。人物Bの歯とパンの間に小さな隙間がある。\n接触：人物Bが顔を近づけ、歯がパンの上側の端へ接触する瞬間を見せ、一口だけかじる。\n\n接触後：パンにだけ小さな半月形の一口分が欠け、人物Bは顔を離す。人物Aは同じパンを右手で胸元へ戻す。手や指を噛まない。パン、紙包み、手指を複製しない。\n\n【Cut 3】\nカメラ位置は人物Bの斜め前。カメラ高は二人の目線の高さ。人物B側から人物A側へ向き、人物Bを胸上でやや大きく見せたあと、短く左へ移動して二人の胸上ツーショットにする。背景の消失軸は左奥へ変える。目的は、人物Bの満足顔と人物Aが返す微笑みを同じ終端画へ置くこと。\n人物Bは口を閉じて小さく二回だけ噛み、飲み込む。短い止めのあと、瞳を明るくし、眉を柔らかな弧にし、まぶたを幸せそうに細め、満足した笑顔の口元を作り、人物Aへ視線を戻す。\n\n人物Bだけが弾む優しい声で言う。\n「んー、おいしい！」\n\n人物Aは一口分だけ欠けた同じパンを右手で胸元に持ち、無言で温かな微笑みを返す。二人の顔とパンを同じ画面に保ち、人物Bが「おいしい」の最後の母音を言い終え、二人の笑顔が完成した瞬間に切る。\n【音声】\n人物の外見、年齢感、雰囲気に合った日本語の二人の声を生成する。人物Bは明るく甘い声、人物Aは落ち着いた優しい声。話者を交換せず、三つの台詞を一字一句そのまま、重ねず、口の形と発声を同期させる。\n\n台詞以外の会話、ナレーション、字幕、画面内文字は入れない。BGMなし、音楽なし。環境音とフォーリーのみ。弱い風、遠い鳥、布擦れ、紙包みの紙音、柔らかなパン生地を一口かじった瞬間の小さな圧縮音を、見えている動作へ同期させる。誇張した咀嚼音は入れない。\n\n【重要な禁止】\n話者交換、同時発話、口パクずれ、パンの複製、勝手な手渡し、二口以上食べる、食べる前の歯形、手指の変形、人物の追加、顔を隠す前景、顔中央の光点、字幕、ロゴを出さない。",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2082652041709773075",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/haruuraeadss/status/2082652041709773075",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-07-30T02:18:35.204Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082652041709773075.mp4",
    "posterUrl": "/posters/x/2082652041709773075.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3、日本語のセリフが想像以上に自然でびっくりしました。\n\nアニメらしい映像の雰囲気とも相性が良く、真剣にアニメを作ろうか迷うくらい、日本語のセリフの自然さが良かったです。\n\n2K・15秒でも消費は150クレジット。\n音声付き動画をかなり気軽に試せます🎬\n@Hailuo_AI\n#MiniMaxH3\nプロンプト👇\n\n高品質アニメ映像。世界観は、上品な現代ファンタジー学園を舞台にした2D手描きの日常会話アニメ。表情設計は必ず、目 → 眉 → まぶた → 口 → 視線、の順で行う。\n\n【参照画像と人物固定】\n画像1は人物Aの参照画像。画像1に描かれたキャラクターを、全編で同じ人物として維持する。画像1は、人物Aの顔、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、シルエット、雰囲気、キャラクターカラーの参照にのみ使用する。\n画像2は人物Bの参照画像。画像2に描かれたキャラクターを、全編で同じ人物として維持する。画像2は、人物Bの顔、目の形、虹彩色、髪型、髪色、衣装、装飾、体格、シルエット、雰囲気、キャラクターカラーの参照にのみ使用する。\n\n参照画像そのものの背景、部屋、家具、ポーズ、画角、フレーミング、光源配置は再現しない。変化してよいのは、表情、視線、口元、姿勢、呼吸、自然な髪と衣装の揺れだけ。\n\n人物Aと人物Bの特徴を混ぜない。顔の平均化、衣装交換、髪型交換、別人化、クローン化、分身、余計な人物の追加は禁止。\n\n画像2に武器や手持ち品が含まれていても、人物Bの両手には何も持たせない。画像1に非戦闘用の装飾や小型バッグがある場合のみ、人物Aに自然に残してよい。ただしバッグは身体側に固定し、手で持つ、開ける、動作に使うことは禁止。武器、大型装備、発光する戦闘装備は持たせない。\n\n【小物・色味・雰囲気の調整】\n場所、座る位置、会話内容、パンを分ける動作、カメラ構成は固定する。そのうえで、人物Aと人物Bの参照画像に合わせて、髪飾り、ブローチ、リボン、バッグなどの装飾、小物、配色、光の印象、背景の差し色を自然に調整する。\n\n全体の線色、影色、背景の差し色、午後光の色温度は、人物Aと人物Bのキャラクターカラーと雰囲気に合わせて調整する。ただし、明るく上品で清潔な現代ファンタジー学園の日常感は維持し、暗い戦闘空間、ホラー、退廃、極端な逆光にはしない。\n【絵柄固定】\n細く繊細な色付き線画、透明感のある二層から三層の色影、柔らかなグラデーション。彩度は低くしすぎず、参照人物の固有色を髪、瞳、衣装、反射光、背景の差し色へ自然に反映する。\n\n瞳は宝石のような虹彩の奥行きと複数の小さなキャッチライト、髪は真珠のような多層ハイライト。髪、まつ毛、レース、刺繍、髪飾り、装飾、衣服の皺を高密度に描く。布は上質なマット素材、レース、金属、宝石、リボンはそれぞれ異なる繊細な質感で描き分ける。\n\n背景も同じ2D手描き筆致で、クリーム色の壁、木製ベンチ、植栽、遠景の校舎を十分な密度で描く。太線、厚輪郭、簡略TVアニメ化、平坦な単層セル塗り、低密度背景、滑らかな3D素材、半写実、実写質感、強いくすみ、生成AI平均顔、画風混合は禁止。\n【固定場所と人物配置】\n放課後の学園中庭にある木製ベンチ。人物Bは画面左、人物Aは画面右。二人は隣り合って座る。\n\n人物Aだけが、薄い紙包みに入った丸いあんパン一個を右手で下側から持つ。開始時のパンは完全な丸形で、歯形も欠けもない。指は紙包みの下側に置き、食べる位置から離す。パン、紙包み、手指を増やさない。\n\n【Cut 1】\nカメラ位置はベンチ正面中央。カメラ高は座った二人の目線の高さ。正面から二人を胸上で同じ大きさに収め、背景の消失軸は二人の間の奥へ抜ける。目的は、二人の関係とパンの持ち主が人物Aであることを示すこと。\n\n人物Bはパンを見たあと人物Aへ視線を戻し、両手を胸元で合わせ、身体ごと少し近づく。瞳を輝かせ、眉を上げ、まぶたを期待で開き、甘える口元を作り、人物Aへ視線を合わせる。\n人物Bだけが明るく甘い声で言う。\n「そのアンパン、一口ちょうだい。」\n\n短い止めを入れ、さらに甘えて言う。\n「お願い、一生のお願い！」\n\n人物Aは話さず、あきれながらも優しい目で人物Bを見る。\n【Cut 2】\nカメラ位置は人物Aの斜め前。カメラ高は二人の目線の高さ。人物Aの肩越しから人物Bへ向き、二人の顔とパンを中近景で見せる。背景の消失軸は右奥へ変える。目的は、返答、差し出す手、一口だけかじる動作を一続きで見せること。\n\n人物Aは瞳を柔らかくし、眉を少し下げ、まぶたを穏やかに細め、困った笑顔の口元を作り、人物Bへ視線を合わせる。\n\n小さく息を吐き、短い止めを入れ、人物Aだけが落ち着いた優しい声で言う。\n「もう、しょうがないなー。ひとくちだけだよ」\n言い終わりながら、人物Aは右手の一個のあんパンを人物Bの口元へ差し出す。持ち主は人物Aのままで、手渡さない。\n\n接触前：パンは完全な丸形。人物Bの歯とパンの間に小さな隙間がある。\n接触：人物Bが顔を近づけ、歯がパンの上側の端へ接触する瞬間を見せ、一口だけかじる。\n\n接触後：パンにだけ小さな半月形の一口分が欠け、人物Bは顔を離す。人物Aは同じパンを右手で胸元へ戻す。手や指を噛まない。パン、紙包み、手指を複製しない。\n\n【Cut 3】\nカメラ位置は人物Bの斜め前。カメラ高は二人の目線の高さ。人物B側から人物A側へ向き、人物Bを胸上でやや大きく見せたあと、短く左へ移動して二人の胸上ツーショットにする。背景の消失軸は左奥へ変える。目的は、人物Bの満足顔と人物Aが返す微笑みを同じ終端画へ置くこと。\n人物Bは口を閉じて小さく二回だけ噛み、飲み込む。短い止めのあと、瞳を明るくし、眉を柔らかな弧にし、まぶたを幸せそうに細め、満足した笑顔の口元を作り、人物Aへ視線を戻す。\n\n人物Bだけが弾む優しい声で言う。\n「んー、おいしい！」\n\n人物Aは一口分だけ欠けた同じパンを右手で胸元に持ち、無言で温かな微笑みを返す。二人の顔とパンを同じ画面に保ち、人物Bが「おいしい」の最後の母音を言い終え、二人の笑顔が完成した瞬間に切る。\n【音声】\n人物の外見、年齢感、雰囲気に合った日本語の二人の声を生成する。人物Bは明るく甘い声、人物Aは落ち着いた優しい声。話者を交換せず、三つの台詞を一字一句そのまま、重ねず、口の形と発声を同期させる。\n\n台詞以外の会話、ナレーション、字幕、画面内文字は入れない。BGMなし、音楽なし。環境音とフォーリーのみ。弱い風、遠い鳥、布擦れ、紙包みの紙音、柔らかなパン生地を一口かじった瞬間の小さな圧縮音を、見えている動作へ同期させる。誇張した咀嚼音は入れない。\n\n【重要な禁止】\n話者交換、同時発話、口パクずれ、パンの複製、勝手な手渡し、二口以上食べる、食べる前の歯形、手指の変形、人物の追加、顔を隠す前景、顔中央の光点、字幕、ロゴを出さない。",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082650526362333627",
    "title": "电影叙事作品 333627",
    "titleEn": "Cinematic Story Study 333627",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《电影叙事作品 333627》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×1280。",
    "summaryEn": "An H3 output for Cinematic Story Study 333627. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×1280.",
    "prompt": "電影感，現代奢華風場景：高端企業總裁辦公室，夜景落地窗，昏暗奢華燈光。\n\n角色設定：\n霸總 (參考 📷圖片1 )：35歲，劍眉星目，深邃冷峻，穿黑色襯衫領口敞開，深灰色合身西裝褲。\n\n秘書 (參考 📷圖片2 )：28歲，瓜子臉，長直黑髮，白色修身襯衫＋黑色窄裙。\n\n清潔婦 (參考 📷圖片3 )：60歲，白髮盤髻，細框老花眼鏡，深藍色清潔制服，乾淨溫和。\n\n音樂風格：戲劇化的BGM，融合短劇的懸疑與浪漫風格。\n\n分鏡腳本：0-15秒\n\n0-2s：清潔婦驚慌轉身想往門口逃去。\n\n2-5s：霸總大步上前，一把將清潔婦壁咚壓在門板上。\n5-8s：霸總低下頭，嘴唇幾乎碰觸清潔婦的耳垂，低聲挑逗：「想躲？門都被我堵死了。」\n8-10s：清潔婦背靠門板，呼吸急促，眼鏡微微歪斜：「總裁別這樣……」\n\n10s-13s：沒等清潔婦話說完，霸總突然用力將清潔婦的肩膀壓在門板上，開始熱情地親吻清潔婦，劇烈喘息。清潔婦的表情也從抗拒逐漸變成服從。\n\n13-15s：門外的走廊上，秘書悄悄靠近門邊，貼耳偷聽，臉色越來越驚訝。畫面結尾 Fade into black.\n\n#AIDrama #AIFilm #AIMovie #AIVideo #霸總 #Microdrama",
    "promptSourceUrl": "https://x.com/drjoetw/status/2082650526362333627",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/drjoetw/status/2082650526362333627",
    "sourceLabel": "X 原帖 · @drjoetw",
    "author": "The Anxious Mind",
    "publishedAt": "2026-07-30T02:12:33.917Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082650526362333627.mp4",
    "posterUrl": "/posters/x/2082650526362333627.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The CEO's pursuit was simply too irresistible, even for the sophisticated older woman. Age may separate them, but chemistry brings them together. ✨👔💖\n(created using @Hailuo_AI, 2K NEW model #MiniMaxH3)\nprompt below 👇\n\n電影感，現代奢華風場景：高端企業總裁辦公室，夜景落地窗，昏暗奢華燈光。\n\n角色設定：\n霸總 (參考 📷圖片1 )：35歲，劍眉星目，深邃冷峻，穿黑色襯衫領口敞開，深灰色合身西裝褲。\n\n秘書 (參考 📷圖片2 )：28歲，瓜子臉，長直黑髮，白色修身襯衫＋黑色窄裙。\n\n清潔婦 (參考 📷圖片3 )：60歲，白髮盤髻，細框老花眼鏡，深藍色清潔制服，乾淨溫和。\n\n音樂風格：戲劇化的BGM，融合短劇的懸疑與浪漫風格。\n\n分鏡腳本：0-15秒\n\n0-2s：清潔婦驚慌轉身想往門口逃去。\n\n2-5s：霸總大步上前，一把將清潔婦壁咚壓在門板上。\n5-8s：霸總低下頭，嘴唇幾乎碰觸清潔婦的耳垂，低聲挑逗：「想躲？門都被我堵死了。」\n8-10s：清潔婦背靠門板，呼吸急促，眼鏡微微歪斜：「總裁別這樣……」\n\n10s-13s：沒等清潔婦話說完，霸總突然用力將清潔婦的肩膀壓在門板上，開始熱情地親吻清潔婦，劇烈喘息。清潔婦的表情也從抗拒逐漸變成服從。\n\n13-15s：門外的走廊上，秘書悄悄靠近門邊，貼耳偷聽，臉色越來越驚訝。畫面結尾 Fade into black.\n\n#AIDrama #AIFilm #AIMovie #AIVideo #霸總 #Microdrama",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082646007444300101",
    "title": "小偷尴尬的产品借口",
    "titleEn": "The Thief's Awkward Product Excuse",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《小偷尴尬的产品借口》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×828。",
    "summaryEn": "An H3 output for The Thief's Awkward Product Excuse. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×828.",
    "prompt": "日本語での会話シーン。ピンクシャツを着た泥棒の男が、おばあさんに見つかり、こっぴどく叱られて立ちすくんでいる。男はそばにいる少女に向かって言い訳を始める。「別に盗むつもりじゃなかったんだ、ちょっと借りただけだったんだよ」と弁解していると、おばあさんがふと思い出したように話を遮り、「そういえば、はいるお スリーが出たんだって」と告知するように明るく言う。少女と男は一瞬きょとんとした表情を浮かべ、泥棒は「そそ、そうだよ！はいるおスリーが出たんだよ！そのためにパンをちょっと借りたんだよ！」と気まずい空気が笑いに変わる。少女は画面に向かって軽くアピールするような仕草を見せ、まるでコマーシャルの一言のように場の空気を切り替える。カメラは9種類の異なるアングルで切り替わり、男が言い訳する肩越しショット、少女が告知する瞬間のクローズアップ、おばあさんと男が驚く表情のアップ、三人の位置関係が分かる引きの全体ショットでしめくくる。コミカルな緊張感と告知の明るさを両立させるように構成される。",
    "promptSourceUrl": "https://x.com/ponzponz15/status/2082646007444300101",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/ponzponz15/status/2082646007444300101",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-07-30T01:54:36.523Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082646007444300101.mp4",
    "posterUrl": "/posters/x/2082646007444300101.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×828",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo 3出たので日本語チェック・・前回のパン泥棒のエクストラカット作ってみた🍞\n\nprompt:日本語での会話シーン。ピンクシャツを着た泥棒の男が、おばあさんに見つかり、こっぴどく叱られて立ちすくんでいる。男はそばにいる少女に向かって言い訳を始める。「別に盗むつもりじゃなかったんだ、ちょっと借りただけだったんだよ」と弁解していると、おばあさんがふと思い出したように話を遮り、「そういえば、はいるお スリーが出たんだって」と告知するように明るく言う。少女と男は一瞬きょとんとした表情を浮かべ、泥棒は「そそ、そうだよ！はいるおスリーが出たんだよ！そのためにパンをちょっと借りたんだよ！」と気まずい空気が笑いに変わる。少女は画面に向かって軽くアピールするような仕草を見せ、まるでコマーシャルの一言のように場の空気を切り替える。カメラは9種類の異なるアングルで切り替わり、男が言い訳する肩越しショット、少女が告知する瞬間のクローズアップ、おばあさんと男が驚く表情のアップ、三人の位置関係が分かる引きの全体ショットでしめくくる。コミカルな緊張感と告知の明るさを両立させるように構成される。\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082612279866986858",
    "title": "要不要一起跳舞",
    "titleEn": "Would you like to dance",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《要不要一起跳舞》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Would you like to dance. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Would you like to dance?",
    "promptSourceUrl": "https://x.com/OrctonAI/status/2082612279866986858",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/OrctonAI/status/2082612279866986858",
    "sourceLabel": "X 原帖 · @OrctonAI",
    "author": "OrctonAI",
    "publishedAt": "2026-07-29T23:40:35.242Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082612279866986858.mp4",
    "posterUrl": "/posters/x/2082612279866986858.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Been playing around with  @Hailuo_AI  new H3 video model…\n“Would you like to dance?”\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082611421225845158",
    "title": "硬核科幻沙漠九宫格动画",
    "titleEn": "Hard sci-fi desert 3x3 grid animation sequence",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《硬核科幻沙漠九宫格动画》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×810。",
    "summaryEn": "An H3 output for Hard sci-fi desert 3x3 grid animation sequence. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×810.",
    "prompt": "Epic desert sci-fi, harsh golden-hour light, volumetric sand haze, anamorphic lens flares, subtle handheld drift, heavy atmospheric particles, 35mm film texture.\n\nUse the provided 3x3 grid as the only visual reference. Lock the exact same woman (face, scars, green eyes, headscarf, desert armor, cape) and environments from every panel. Do not invent new outfits or faces.\n\n[TIMELINE] 0-1.7s: [Wide establishing] Lone figure walking across endless dunes toward twin massive suns. Heat shimmer, long shadow stretching behind her.\n1.7-3.4s: [Extreme close-up] Her scarred face fills the frame. Eyes locked on camera, wind tugging the dirty headscarf, grit on skin, unblinking intensity.\n\n3.4-5.1s: [Tight detail insert] Gloved hand grips the rusted multi-tool weapon. Metal catches the orange light, cable dangling, fingers adjusting the grip with quiet precision.\n\n5.1-6.8s: [Low wide silhouette] The towering industrial structure rises against the giant sun. Dust storms swirl at its base, scale feeling ancient and oppressive.\n\n6.8-8.5s: [Medium tracking] She advances through thick sandstorm, lower face wrapped, cape and scarf whipping hard. Eyes visible above the cloth, determined stride.\n\n8.5-10.2s: [Hero wide] She stands alone on the rocky outcrop, cape flowing, surveying the empty desert below. Wind howls, silhouette sharp against the hazy sky.\n\n10.2-11.9s: [Dynamic tracking + low angle] She charges down the steep dune face, boots kicking up thick clouds of sand, body leaning forward with full momentum.\n\n11.9-13.6s: [Macro atmospheric] Golden sand particles explode and float in extreme slow motion, backlit by the blazing sun, pure texture and light.\n\n13.6-15s: [Intimate close-up] Inside the dark cave, shafts of light cut through. She sits against the rock, armor dusted, breathing controlled, eyes sharp and watching the entrance.\n\nPhotorealistic, ultra-detailed skin and fabric textures, perfect sand physics, coherent character identity across every cut, cinematic color grade, heavy film grain, no artifacts, movie-level continuity.",
    "promptSourceUrl": "https://x.com/techhalla/status/2082611424824623135",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/techhalla/status/2082611421225845158",
    "sourceLabel": "X 原帖 · @techhalla",
    "author": "TechHalla",
    "publishedAt": "2026-07-29T23:37:10.526Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082611421225845158.mp4",
    "posterUrl": "/posters/x/2082611421225845158.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×810",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "nature",
      "sci-fi"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 + grid 3x3 prompt below 👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082605878679609528",
    "title": "高能情感表演口型同步",
    "titleEn": "Emotional performance lip-sync to reference audio",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《高能情感表演口型同步》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 720×1280。",
    "summaryEn": "An H3 output for Emotional performance lip-sync to reference audio. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 720×1280.",
    "prompt": "The man sings the song with great emotion and energy. His movements are rapid and in sync with the music. He looks straight ahead. He is full of drive and energy, and his arm movements complement the motion of his body. Handheld shooting.\nHe sings passionately and emotionally with perfect lip sync the lyrics: “My chest goes quick My pulse on fire One little smile And I climb higher I don’t stand still I don’t stay calm You turn my whole Body on.”",
    "promptSourceUrl": "https://x.com/Kiber_Alla/status/2082605880776741109",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/Kiber_Alla/status/2082605878679609528",
    "sourceLabel": "X 原帖 · @Kiber_Alla",
    "author": "Kiber Alla",
    "publishedAt": "2026-07-29T23:15:09.080Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082605878679609528.mp4",
    "posterUrl": "/posters/x/2082605878679609528.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Music Video",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just tested the new MiniMax H3 @Hailuo_AI \nOne reference image + audio.\nOmni Reference ON.\nNo audio changes after generation.\n\nPrompt below 👇\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082563241062875568",
    "title": "五段电影对白重场戏",
    "titleEn": "Five cinematic dialogue set-piece scenes",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《五段电影对白重场戏》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 161 秒 · 1286×720。",
    "summaryEn": "An H3 output for Five cinematic dialogue set-piece scenes. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 161s · 1286×720.",
    "prompt": "SCENE CONTEXT A middle-aged man stands in the middle of a dirt road in open desert and holds a pistol level at the person filming him. He gives an instruction, is answered by name, and warns them not to say it again. Nothing is fired.\n\nTIMELINE 0.0–3.0s — he is already aimed and steady; his first line lands. 3.0–6.0s — the off-frame reply uses his name; his jaw sets and the muzzle does not drop. 6.0–9.0s — his second line; he takes half a step forward and re-sights. 9.0–11.0s — the off-frame voice starts his name again and stops. 11.0–15.0s — his last line; wind lifts dust across the road behind him; he holds the aim. Hold to black-free end.\n\nACTIVE REFERENCES No image or video references are active in this shot. Build the man from description only. Man, fifties, white, average build, standing, thinning brown hair, a brown moustache, thin-rimmed prescription glasses; light green long-sleeved button-down shirt with the sleeves buttoned at the wrist, dark work trousers; stern, brow furrowed, sweat starting at the hairline; frightened underneath and controlling it; speaking voice flat, plain, mid-western American, no theatricality. Second speaker: the unseen person the pistol is aimed at, occupying the camera position, never entering frame and never seen. Voice only — male, forties, close to the lens, low and careful.\n\nLOCATION MAP Camera on the dirt road 3.5 m from him at his chest height, facing him square-on so the muzzle is aligned with the lens axis. FG: rutted dirt and loose gravel across the lower frame. MG: the man alone, centre frame. BG: the dirt road receding to a vanishing point at x 50%, y 44%, low scrub and dry brush from x 0% to x 30% and x 70% to x 100%, a distant fence line at x 62%, y 48%, flat desert horizon at y 46%, overcast sky filling the upper half with one pale blue break low at x 24%, y 36%. He occupies x 50%, y 50%, body frontal, both feet planted, right arm extended toward camera with the silver pistol at x 50%, y 56%, held at the very front of the depth plane. His eyeline is straight down the lens. The first visible frame already has him aimed and in position — no empty establishing frame, no walk-in, no delayed reveal, and he remains the only figure in frame for the entire take.\n\nFORMAT MODE Single continuous take, no internal cuts, no transitions, no fades. Real-time motion. His lines are on camera; the second voice is an intentional off-frame speaker at the camera position. No other voices, no narration, no subtitles, no score.\n\nCAMERA 47° diagonal field of view, camera 3.5 m from him at chest height, natural human-eye perspective, no distortion, natural face and body proportions, the road and horizon readable behind him, comfortable depth of field with the pistol slightly forward of his focal plane. Handheld, shoulder-mounted mass, operator breath and small involuntary settling — the camera is being held by the person he is aiming at, so it flinches minutely rather than gliding. Total travel under 20 cm, no push-in, no zoom, no focus rack. Focus holds his eyes behind the glasses; the pistol stays just soft.\n\nACTION TIMING He is already sighted along the barrel as the shot opens and speaks without lowering it. The off-frame voice answers with his name; his jaw sets and the muzzle does not waver. He speaks his second line and takes half a step forward on the dirt, re-sighting as his weight lands. The off-frame voice begins his name again and stops. He delivers his last line, and wind lifts a curtain of dust across the road behind him. He holds the aim to the end without firing and without lowering the weapon.\n\nPHYSICS The pistol carries real weight — his wrist and forearm show the load and micro-correct continuously; the muzzle drifts a few millimetres and is pulled back on line. His half step lands heel-first in soft dirt with visible weight transfer and a small dust puff that drifts and thins. Shirt fabric pulls across the extended shoulder and creases at the elbow. Wind moves the loose dust and the dry scrub in the same direction and at the same time. Thinning hair lifts and settles with the gust. Sweat beads at his temple and does not run.\n\nLIGHTING Primary source is the full overcast sky acting as one enormous soft toplight, giving even illumination with minimal harsh shadow — a soft shadow under the brow and the nose only, no hard edges anywhere. Secondary is dry ground bounce filling under the chin and lifting the shadow side of the pistol. Both lenses of his glasses carry a broad soft reflection of the sky, and his eyes stay readable behind them. Exposure is set for his face; the sky sits bright but not clipped. No sun shaft, no lens flare, no added key, no light change during the take.\n\nAUDIO Dry wind across open ground, loose grit moving on the road, dry brush ticking, one distant bird, his own breathing close and unsteady. Ambient ducks under the voices. Exactly this exchange, flat and plain, no shouting — 25 words total: HIM: \"Put it on the ground. Slowly.\" OFF-FRAME: \"You're not going to shoot me, Ray.\" HIM: \"I've been wrong about myself before.\" OFF-FRAME: \"Ray—\" HIM: \"Don't say my name again.\" His first line opens the shot. The off-frame reply lands as his jaw sets. His second line comes as he takes the half step. The off-frame \"Ray—\" is cut off by his last line, which is quieter than everything before it. His lips move only for his own lines; there is no dialogue in the pauses and no muttering between beats.\n\nPOSITIVE LOCKS Exactly one visible person is in frame for the whole take — nobody enters at any edge, no second figure appears on the road, at the fence line or in the reflections on his glasses, and the off-frame speaker is never seen. The pistol stays in his right hand, stays aimed at the lens, and is never fired, never lowered and never re-holstered; no muzzle flash, no recoil, no gunshot. He stays on the road and never turns away from camera. Naturalistic muted palette of dry greens and browns against pale overcast grey. Live-action photoreal footage shot on ARRI Alexa 35, tense neutral mood, natural film grain.\n\nSCENE CONTEXT In a frosted country-house conservatory on a grey morning, an older woman tells a younger one that the letter will not be answered. The younger woman holds her composure and loses it only in her hands.\n\nFORMAT MODE Controlled three-segment sequence, one HARD CUT and one INSERT CUT. Real-time motion.\n\nLOCATION MAP Camera inside a glazed conservatory, 3 m from the two women, facing them with the steamed glass wall behind them. FG: potted fern fronds at the frame edge. MG: older woman screen-left, younger woman screen-right, 80 cm apart. BG: glass panes, frost-white winter garden, a cast-iron bench, one low oil lamp. Soft light comes from the whole glass wall behind and above them.\n\nFIRST FRAME AND SPATIAL BLOCKING Both women are in frame from the first visible frame. Older woman at x 32%, y 46%, three-quarter profile facing screen-right, gloved hands folded at her waist. Younger woman at x 68%, y 44%, torso squared to her, gaze locked on her face, a folded letter held in both hands at x 62%, y 76%. No empty establishing frame.\n\nOPTICS LENS LOCK SEGMENT 1 = 47°, camera 3 m out, natural proportions, both faces and the conservatory geography readable. LENS LOCK SEGMENT 2 = 18°, camera 6.5 m back down the conservatory aisle, strong compression, glass panes and ferns stacking behind the younger woman, razor-thin focus on her eyes. LENS LOCK SEGMENT 3 = 29°, camera 4 m out and lower, close on her hands by lens reach, the room dissolving to creamy bokeh behind.\n\nCAMERA Segment 1 locked-off at eye height. Segment 2 long lens with the faintest human drift. Segment 3 handheld, settling downward toward the letter.\n\nACTION TIMING BEAT 1 — The older woman speaks her line, chin level, hands unmoving. The younger woman does not react in her face. HARD CUT BEAT 2 — Long-lens hold on the younger woman. A long beat. Her eyelids redden; she blinks once and keeps her eyes forward, refusing to look down. INSERT CUT BEAT 3 — Tight on her hands. Her thumbs press and slowly crush the folded letter's edge until the paper buckles, then stop. Hold on the buckled paper.\n\nPHYSICS Paper buckles progressively under thumb pressure and holds its new shape. Glove leather stretches over her knuckles. Condensation runs down glass panes in slow irregular tracks. Silk and wool hold real weight; her skirts do not float. Breath fogs faintly in the cold air.\n\nLIGHTING Primary source is vast soft north daylight through the steamed glass wall behind and above them, motivated, wrapping both faces. Secondary: cold blue bounce off the frost outside, and one warm low oil lamp deep in frame. Soft shadow rolloff with clearly readable eyes and catchlights — restrained low contrast, not silhouette.\n\nAUDIO Rain ticking on glass, a distant door, the faint hiss of the oil lamp, a bird outside. Ambient ducks under the voice. The older woman speaks exactly this line as the shot opens: \"There will be no reply, and you will not ask again.\" No other dialogue, no score. The younger woman's lips stay still.\n\nPOSITIVE LOCKS Exactly two women appear and no servant or third figure enters. Both hold their positions across all segments. Costume, gloves, the folded letter and the frost state remain identical across the cuts. Cinematic photorealistic footage shot on ARRI Alexa 35, pearl-grey and soft-green grade, low contrast, natural film grain.\n\nSCENE CONTEXT A weathered man rests against a timber rail in blistering noon heat while something is brought up the street toward him. A voice across the street offers him a way out and he declines it.\n\nTIMELINE 0.0–3.0s — the off-frame line; a gust lifts dust across frame. 3.0–6.0s — his answer; he turns his face out of the dust and works his jaw once. 6.0–9.5s — the second off-frame line; his eyes track slowly screen-left to screen-right. 9.5–12.5s — his last line, flat and final. 12.5–15.0s — he pushes off the rail with one shoulder and straightens out of the top of frame; dust closes behind him.\n\nACTIVE REFERENCES  📷Image1  is the reference and the exact first visible frame — source of truth for his face, hat, wardrobe and the light. Man, fifties, leaning against a timber rail, pale wide-brim hat pushed back with a visible pale forehead line, dust-caked collarless shirt torn at one shoulder, deep sun creases, broken capillaries, grit in the stubble; long past being surprised; speaking voice dry, laconic, low volume. 100% matches the reference. Second speaker: an unseen man across the street, off-frame screen-left at the same level, never entering frame. Voice only — male, middle-aged, flat, carrying across distance.\n\nLOCATION MAP Camera on the street 4.5 m from him at his eye height. FG: nothing between camera and subject except airborne dust. MG: the man against grey weathered timber spanning x 0% to x 55%. BG: a blurred wagon at x 76%, y 30% and two standing figures at x 84%, y 34% and x 92%, y 36%, dissolved in heat shimmer, pale sky at the top edge. He occupies x 34%, y 50%, chest-up, torso angled to screen-right, eyes to screen-left. The first visible frame already contains him in exactly this position. He is the only sharp figure and remains so.\n\nFORMAT MODE Single continuous take, no internal cuts, no transitions. Real-time motion. His lines are on camera; the second voice is an intentional off-frame reply. No other voices, no narration, no subtitles, no music.\nCAMERA 29° diagonal field of view, camera 4.5 m out at his eye height, close framing achieved through lens reach, his face razor-sharp, the wagon and the two figures compressed flat behind him and melting into bokeh, visible heat shimmer between camera and background. Locked off with the faintest human drift; no push-in, no focus rack.\n\nACTION TIMING He squints harder as a gust lifts dust across the frame, and turns his face a few degrees out of it as he answers. He works his jaw once. His eyes track slowly from screen-left to screen-right, following something crossing behind camera. A long beat. He delivers his last line, then pushes off the rail with one shoulder and straightens, rising out of the top edge of frame as the dust closes behind him.\n\nPHYSICS Airborne dust crosses in real gusts, catches the light and settles on his shoulders. The torn shirt fabric lifts at the shoulder and falls. The timber rail does not move when he pushes off it, and his weight transfers audibly into his boots. Sweat and grit sit on the skin without running.\n\nLIGHTING Inherit  📷Image1  exactly and do not relight. Primary: hard vertical noon sun. Secondary: brilliant bounce off pale caliche dust filling the shadows from below. Deep shadow under the hat brim across the bridge of his nose. Highlights on the timber and sky sit near clipping. No fill added, no light change during the take.\n\nAUDIO Dry wind against timber, distant wagon wheels and harness, a fly, boot grit. Ambient ducks under the voices. Exactly this exchange, laconic and flat, spoken across distance without echo effects — 22 words total: OFF-FRAME MAN: \"They're bringing him up now.\" HIM: \"I can see the dust.\" OFF-FRAME MAN: \"You want to be somewhere else?\" HIM: \"I've been somewhere else. Didn't help.\" The off-frame line opens the shot. He answers into the gust. The second off-frame line comes while his eyes track across. His last line lands as he pushes off the rail. His lips move only for his own lines.\n\nPOSITIVE LOCKS Exactly one sharp figure is in frame; the two background figures stay unresolved in the heat haze and never become characters or come into focus, and no one crosses between camera and subject. Hat, torn shirt, dust layer and the timber rail stay exactly as in  📷Image1 . Cinematic photorealistic footage shot on ARRI Alexa 35, natural film grain.\n\nSCENE CONTEXT On a narrow Naples balcony at blue hour, a woman tells a man she has already taken the job. He does not argue, which is worse.\n\nFORMAT MODE Controlled three-segment sequence, two HARD CUTs. Real-time motion. Restrained camera throughout.\n\nLOCATION MAP Camera on the balcony, 2 m from the couple, facing them with the lit kitchen doorway behind them screen-centre-left. FG: iron rail and two espresso cups. MG: woman and man standing 60 cm apart. BG: alley canyon, drying laundry, a lit window opposite, string lights. Warm light comes from behind them through the doorway; cool dusk fills from camera side.\n\nFIRST FRAME AND SPATIAL BLOCKING Both are in frame from the first visible frame. Woman at x 34%, y 46%, torso facing screen-right, one hand on the iron rail. Man at x 68%, y 44%, three-quarter profile facing her, hands at his sides. Their eyelines meet at x 50%. The two espresso cups sit on the rail at x 52%, y 74%.\n\nOPTICS LENS LOCK SEGMENT 1 = 47°, camera 2.2 m out, natural proportions, both faces and the alley depth readable. LENS LOCK SEGMENT 2 = 29°, camera 4.5 m back along the balcony, close on the woman by lens reach, alley lights compressing into soft round bokeh behind her. LENS LOCK SEGMENT 3 = 84°, camera 1 m from the man, foreground rail presence, the whole balcony and doorway visible to the edges.\n\nCAMERA Segment 1 handheld with almost no travel, operator breath only. Segment 2 locked-off. Segment 3 handheld, a slow 20 cm settle downward as he lowers his head. Camera stays on the dusk side of both faces.\n\nACTION TIMING BEAT 1 — She speaks her line, then looks at her own hand on the rail rather than at him. He does not move. HARD CUT BEAT 2 — Held on her. A long beat. Her mouth begins a word and abandons it; she swallows and lifts her eyes back to him. HARD CUT BEAT 3 — On him. He nods once, small, and picks up one of the espresso cups without drinking from it. The shot holds on his hand around the cup as the string lights flicker once.\n\nPHYSICS Cool air moves the drying laundry in the background on real cloth delay. Her knit sleeve compresses where her forearm presses the rail. The cup has weight; the saucer chatters faintly when lifted. Loose hair moves independently of her head turn.\n\nLIGHTING Primary source is the warm kitchen bulb behind them through the doorway, motivated, giving both a warm hair and shoulder rim. Fill is cool ambient dusk from camera side. Soft shadow rolloff, not crushed silhouettes — eyes stay readable with visible catchlights. Practical string lights and one lit window across the alley sit in the background.\n\nAUDIO Distant scooter, a television two floors down, pigeons, faint street voices. Ambient ducks under the voice. She speaks exactly this line as the shot opens: \"I signed it on Tuesday.\" No other dialogue, no score. His lips stay still.\n\nPOSITIVE LOCKS Exactly two people are on the balcony and no one appears in the doorway behind them. Their positions, wardrobe, the two cups and the laundry arrangement remain identical across the cuts. Cinematic photorealistic footage shot on ARRI Alexa 35, warm-amber against slate-blue grade, low contrast, natural film grain.\n\nSCENE CONTEXT In a shelled Normandy farmhouse at night, a colonel orders a captain to take the lane at first light. The captain accepts the order knowing what it costs.\n\nFORMAT MODE Controlled three-segment sequence, one HARD CUT and one REVERSE CUT. Real-time motion.\n\nLOCATION MAP Camera inside a roofless farmhouse room, 2.5 m from a map table. FG: table edge, tin mug, field telephone. MG: colonel standing screen-left of the table, captain standing screen-right. BG: pinned maps on plaster, splintered beams, a hole in the roof. Primary light is a hurricane lamp on the table between them.\n\nFIRST FRAME AND SPATIAL BLOCKING Both officers are already at the table in the first visible frame. Colonel at x 30%, y 44%, three-quarter profile facing screen-right, index finger planted on the map at x 46%, y 76%. Captain at x 70%, y 42%, torso squared to the colonel, gaze down on the finger, not the face. Lamp at x 50%, y 66%. No empty establishing frame.\n\nOPTICS LENS LOCK SEGMENT 1 = 47°, camera 2.5 m from the table, natural proportions, both officers and the map geography readable. LENS LOCK SEGMENT 2 = 29°, camera 4 m back, close on the captain by lens reach, the pinned map wall compressed soft behind him. LENS LOCK SEGMENT 3 = 84°, camera 1.1 m above the map surface angled up past the lamp, foreground lamp glass looming, both men and the broken roof visible to the edges.\n\nCAMERA Segment 1 locked-off at chest height. Segment 2 handheld with operator breath, camera on the lamp's shadow side. Segment 3 handheld low, slow rise past the lamp flame.\n\nACTION TIMING BEAT 1 — The colonel speaks his line, finger pressing hard enough to crease the map. Plaster dust drifts through the lamp beam. REVERSE CUT BEAT 2 — On the captain. A distant artillery flash strobes blue across his face. A muscle jumps in his cheek; he nods once and looks up. HARD CUT BEAT 3 — Low wide past the lamp. The captain squares his cap, turns and walks out of frame screen-right. The colonel stays, hand flattening the creased map, and the lamp flame leans in the draught of the door.\n\nPHYSICS Paper creases and holds the crease under finger pressure. The lamp flame bends and recovers when air moves; its light level dips accordingly. Plaster dust falls with real settling drift. Boots find rubble with uneven weight transfer and grit sound. Wool uniform holds damp weight and does not billow.\n\nLIGHTING Primary source is the hurricane lamp on the table, motivated, hard and warm, underlighting both faces and leaving eye sockets in shadow. Secondary: cold moonlight through the roof hole rimming shoulders, plus intermittent distant artillery flash on the back wall. Exposure is set for the lamp; shadows are allowed to crush. No added frontal fill.\n\nAUDIO Distant shellfire, a field telephone hum, wind through the broken roof, drip of water. Ambient ducks under the voice. The colonel speaks exactly this line as the shot opens: \"You take the lane at first light, with what you have.\" No other dialogue, no score. The captain's lips stay still.\n\nPOSITIVE LOCKS Exactly two officers are in the room and no other figures enter. Both stay on their own side of the table until the captain exits in the final beat. Uniform state, mud, plaster dust and the lamp position remain identical across the cuts. Cinematic photorealistic footage shot on ARRI Alexa 35, desaturated olive-and-amber grade, natural film grain.",
    "promptSourceUrl": "https://x.com/maxescu/status/2082563245504856385",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/maxescu/status/2082563241062875568",
    "sourceLabel": "X 原帖 · @maxescu",
    "author": "Alex Patrascu",
    "publishedAt": "2026-07-29T20:25:43.480Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082563241062875568.mp4",
    "posterUrl": "/posters/x/2082563241062875568.jpg",
    "duration": 161,
    "aspectRatio": "landscape",
    "resolution": "1286×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax's new model, @Hailuo_AI H3, is finally here! \n\nSpecs:\n- 15-sec duration\n- 2560 x 1440 native resolution\n- Omni model with 12 references (image, video, audio)\n\nHere are a few cinematic scenes I tried.\nPrompts below 👇\n\n#MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082549709240160587",
    "title": "霓虹噩梦汉堡广告",
    "titleEn": "Neon Nightmare Burger Commercial",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《霓虹噩梦汉堡广告》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Neon Nightmare Burger Commercial. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Cinematic 15-second commercial shot, fast-paced editing, dark humor, creepy comedy style. Ultra-realistic, 8k resolution. A surreal hamburger restaurant with moody neon lighting (red and green). Close-ups of distorted, wide-eyed hungry people smiling creepily while eating massive juicy cheeseburgers. Fast cuts: sizzling bacon in flames, towering stacks of burger trays, a mysterious retro waiter with a unnerving wide smile. Hyper-detailed melted cheese, glossy sauce drips, chaotic and fun atmosphere, 90s thriller aesthetic crossed with a fast-food ad.",
    "promptSourceUrl": "https://x.com/xavier_mitjana/status/2082549709240160587",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/xavier_mitjana/status/2082549709240160587",
    "sourceLabel": "X 原帖 · @xavier_mitjana",
    "author": "Xavier Mitjana",
    "publishedAt": "2026-07-29T19:31:57.242Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082549709240160587.mp4",
    "posterUrl": "/posters/x/2082549709240160587.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "UGC & Advertising",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "El panorama del vídeo con IA vuelve a respirar.\n\nAhí va mi primer test con el nuevo Hailuo H3, la verdad es que el modelo pinta muy bien.\n\nPrompt: Cinematic 15-second commercial shot, fast-paced editing, dark humor, creepy comedy style. Ultra-realistic, 8k resolution. A surreal hamburger restaurant with moody neon lighting (red and green). Close-ups of distorted, wide-eyed hungry people smiling creepily while eating massive juicy cheeseburgers. Fast cuts: sizzling bacon in flames, towering stacks of burger trays, a mysterious retro waiter with a unnerving wide smile. Hyper-detailed melted cheese, glossy sauce drips, chaotic and fun atmosphere, 90s thriller aesthetic crossed with a fast-food ad.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082548512286224793",
    "title": "实拍关系告白对峙戏",
    "titleEn": "Live-action relationship confession drama scene",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《实拍关系告白对峙戏》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Live-action relationship confession drama scene. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Grounded photorealistic live-action relationship drama, natural 24fps motion.\nFORMAT\n\nTwo connected 15-second clips, 16:9, grounded photorealistic live-action relationship drama, natural 24fps motion.\n\nSHARED STYLE\n\nA small living room at night. One warm floor lamp mixes with cold blue streetlight through the curtains. A framed wedding photograph hangs behind the couple. Restrained handheld camera, realistic skin and fabric, tense silence, no melodramatic shouting.\n\nCONTINUITY\n\nThe husband remains frame left beside the coffee table, holding his wife’s unlocked phone containing the evidence. The wife remains frame right near the window. Preserve their faces, clothes, positions and screen direction across every cut.\n\nCLIP 1 — THE CONFESSION\n\n0.0–5.0\n\nCamera: Slow 35mm push toward both characters.\n\nAction: The husband stares at the phone, then raises his eyes toward her.\n\nHUSBAND: “How long has this been going on?”\n\n5.0–8.0\n\nCamera: Cut to a restrained 50mm close-up of the wife.\n\nAction: She cannot meet his eyes. After a short pause:\n\nWIFE: “Six months.”\n\n8.0–15.0\n\nCamera: Close on the husband. His grip tightens around the phone, but he controls his voice.\n\nHUSBAND: “Six months? I gave you twelve years. Why?”\n\nFinal frame: The wife finally looks directly at him, preparing to answer.\n\nCLIP 2 — THE REASON\n\nBegin from the exact positions and expressions inherited from Clip 1.\n\n0.0–6.5\n\nCamera: Slow push toward the wife as she folds her arms defensively.\n\nWIFE: “Because I’m tired of being broke. Every dream with you becomes ‘maybe one day.’”\n\n6.5–10.0\n\nCamera: Cut to the husband. His anger gives way to disbelief.\n\nHUSBAND: “So you chose him for his money?”\n\n10.0–15.0\n\nCamera: Return to the wife. She holds his gaze without apologizing.\n\nWIFE: “I chose someone who can give me the life I want.”\n\nFinal frame: Hold on the husband as the meaning lands. His hand lowers, and the phone slips onto the sofa.",
    "promptSourceUrl": "https://x.com/NEXUS_TO_NOVA/status/2082548515587358962",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/NEXUS_TO_NOVA/status/2082548512286224793",
    "sourceLabel": "X 原帖 · @NEXUS_TO_NOVA",
    "author": "NΞXUS",
    "publishedAt": "2026-07-29T19:27:11.866Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082548512286224793.mp4",
    "posterUrl": "/posters/x/2082548512286224793.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Got early access to the new @Hailuo_AI video model MiniMax H3.\n\nI used it to direct a live-action confession scene, with a focus on keeping the emotional line and physical continuity locked. #MiniMaxH3\n\nPrompt below👇",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082522979339288619",
    "title": "小机器人赛博朋克逃亡",
    "titleEn": "Little Robot Cyberpunk Escape",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《小机器人赛博朋克逃亡》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Little Robot Cyberpunk Escape. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "A stunning action packed scene following a little robot running away from soldiers, drones and other things that are trying to get him. Chase cam, dystopian futuristic cyberpunk environment, cinematic lighting. The camera follows the robot from behind, chase cam style. Every now and then the robot looks behind him and the entities chasing it are shown.",
    "promptSourceUrl": "https://x.com/AIandDesign/status/2082522979339288619",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/AIandDesign/status/2082522979339288619",
    "sourceLabel": "X 原帖 · @AIandDesign",
    "author": "⭕ AI & Design (Marco)",
    "publishedAt": "2026-07-29T17:45:44.337Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082522979339288619.mp4",
    "posterUrl": "/posters/x/2082522979339288619.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I am testing @Hailuo_AI #MiniMaxH3 and it packs a punch!\n\nPrompt:\n\nA stunning action packed scene following a little robot running away from soldiers, drones and other things that are trying to get him. Chase cam, dystopian futuristic cyberpunk environment, cinematic lighting. The camera follows the robot from behind, chase cam style. Every now and then the robot looks behind him and the entities chasing it are shown.",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082519983842595100",
    "title": "斗篷复仇者对决赛博处刑者",
    "titleEn": "Cloaked Avenger vs. Cyborg Executioner",
    "model": "MiniMax H3（来源标注）",
    "mode": "Ref2VA",
    "summary": "《斗篷复仇者对决赛博处刑者》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 15 秒 · 1280×720。",
    "summaryEn": "An H3 output for Cloaked Avenger vs. Cyborg Executioner. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 15s · 1280×720.",
    "prompt": "Use @Image1 as the cloaked avenger identity lock and wardrobe reference. Use @Image2 as the environment, lighting, and street layout reference. Use @Image3 as the cyborg executioner identity lock, armor design, and weapon reference.\n\nCreate a 15-second ultra-cinematic confrontation in the ruined streets of Sector 7 at night. Heavy ashfall, wet pavement, orange emergency lights, smoke columns, and reflective industrial architecture strictly follow @Image2. The cloaked avenger from @Image1 faces the red cyborg executioner from @Image3 in the center of the empty street. Maintain facial features, wardrobe, armor geometry, and environment consistency across all shots. No onscreen text, no floating letters, clean visuals.\n\n0–4s: Wide establishing shot. The cloaked avenger from @Image1 walks slowly through the rain-dark street toward the cyborg from @Image3. Camera performs a slow push-in. Coat edges move in the wind, boots splash through shallow water, ash drifts through orange backlight.\n\n4–8s: Medium close-up on @Image1. He stops, lifts his chin slightly, and says clearly in English: “Long time I was looking for you.” Deliver slowly, low, and intense, with precise lip-sync. Camera holds a subtle handheld drift. Wet hood, tired eyes, and grim determination remain consistent with @Image1.\n\n8–11s: Medium close-up on @Image3. The cyborg tilts its head, red visor brightening, and replies in a cold metallic voice: “finally”. Deliver with a short pause before the word and precise mouth-panel synchronization. Servo motion, red eye glow, and armored posture strictly follow @Image3.\n\n11–15s: Sudden action payoff. Whip pan into a dynamic wide shot as @Image3 ignites the red energy blade and lunges forward. @Image1 instantly draws his blade and rushes to meet the attack. They collide at center frame in a shower of sparks, rain spray, and reflected orange-red light. End on a dramatic frozen impact moment with crossed blades, smoke and ash swirling around them, camera orbiting slightly into a powerful hero frame. Gritty cinematic realism, strong contrast, volumetric haze, realistic body momentum, normal proportions without stretch.",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2082519987688722576",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/LudovicCreator/status/2082519983842595100",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-07-29T17:33:50.155Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082519983842595100.mp4",
    "posterUrl": "/posters/x/2082519983842595100.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Narrative Film",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just started to test Minimax H3 by Hailuo AI..\n\nFirst test using 3 Character sheet in Omni Reference , find the sheet and prompt first comment..\n\nMore tests to come next days ...\n\n@Hailuo_AI #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2082501605803597837",
    "title": "写实角色转面设定图",
    "titleEn": "Photoreal Character Turnaround Sheet",
    "model": "MiniMax H3（来源标注）",
    "mode": "T2VA",
    "summary": "《写实角色转面设定图》的 H3 成片。本页保留原始 X 来源链接。公开 Prompt 按原文保留，未翻译、改写或补写。站内视频规格为 7 秒 · 1280×720。",
    "summaryEn": "An H3 output for Photoreal Character Turnaround Sheet. This page retains the original X source link. The public Prompt is preserved verbatim without translation, rewriting, or completion. Hosted-video output: 7s · 1280×720.",
    "prompt": "※実写ver。アニメキャラの場合はリプ参照（↓）\n──────────────────\n【スタイル指定】\n- 添付画像のデザイン（衣装・髪型・体格・配色）を完全に維持したまま、実在の人物が衣装を着用して撮影したようなフォトリアルな実写に変換する\n- 生地・肌・髪・金属などの質感は写真として自然なリアリティで再現する\n- デザインそのものの変更（衣装や小物の追加・削除）はしない\n- スタジオ写真撮影のような自然光・均一な照明にする\n\n【共通ルール】\n- SHOT区切りのみで進行する（秒数指定なし）\n- 各SHOTの最後は静止カットで終える。滑らかな変形ではなく静止→カットで切り替える\n- 各SHOTの画面には、そのSHOTで指定された要素だけを映す。それ以外の角度・別バージョンの絵を混ぜない\n- 背景は無地の単色スタジオ背景（白またはライトグレー）に統一する\n- 字幕・テキスト・ロゴ・透かしは表示しない\n\nSHOT 1：正面の立ち姿と、その顔の実写クローズアップを並べて表示する。\n\nSHOT 2：横向き（右）の立ち姿と、その横顔の実写クローズアップを並べて表示する。\n\nSHOT 3：横向き（左）の立ち姿と、その横顔の実写クローズアップを並べて表示する。\n\nSHOT 4：後ろ向きの立ち姿と、その後頭部の実写クローズアップを並べて表示する。\n\nSHOT 5：正面の立ち姿と、表情差分6パターン（普通・笑顔・怒り・驚き・悲しみ・ウインク）の実写クローズアップを並べて表示する。\n\nSHOT 6：正面の立ち姿と、身に着けているもの（衣装・小物・アクセサリー）の素材ディテール実写マクロ接写を並べて表示する。\n\nSHOT 7：正面の立ち姿・横向き（右）の立ち姿・後ろ向きの立ち姿と、正面の顔の実写クローズアップをまとめて1枚に並べる\n──────────────────",
    "promptSourceUrl": "https://x.com/aiehon_aya/status/2082501605803597837",
    "promptCompleteness": "complete",
    "sourceUrl": "https://x.com/aiehon_aya/status/2082501605803597837",
    "sourceLabel": "X 原帖 · @aiehon_aya",
    "author": "妖精アーヤさん",
    "publishedAt": "2026-07-29T16:20:48.489Z",
    "addedAt": "2026-08-22T09:15:40.742Z",
    "mediaUrl": "/media/x-2082501605803597837.mp4",
    "posterUrl": "/posters/x/2082501605803597837.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "完整 Prompt",
      "Anime",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo 新モデル MiniMaxH3 いいじゃん！\nキャラクターシートもGptimg2より綺麗！\n──────────────────\nプロンプト：\n※実写ver。アニメキャラの場合はリプ参照（↓）\n──────────────────\n【スタイル指定】\n- 添付画像のデザイン（衣装・髪型・体格・配色）を完全に維持したまま、実在の人物が衣装を着用して撮影したようなフォトリアルな実写に変換する\n- 生地・肌・髪・金属などの質感は写真として自然なリアリティで再現する\n- デザインそのものの変更（衣装や小物の追加・削除）はしない\n- スタジオ写真撮影のような自然光・均一な照明にする\n\n【共通ルール】\n- SHOT区切りのみで進行する（秒数指定なし）\n- 各SHOTの最後は静止カットで終える。滑らかな変形ではなく静止→カットで切り替える\n- 各SHOTの画面には、そのSHOTで指定された要素だけを映す。それ以外の角度・別バージョンの絵を混ぜない\n- 背景は無地の単色スタジオ背景（白またはライトグレー）に統一する\n- 字幕・テキスト・ロゴ・透かしは表示しない\n\nSHOT 1：正面の立ち姿と、その顔の実写クローズアップを並べて表示する。\n\nSHOT 2：横向き（右）の立ち姿と、その横顔の実写クローズアップを並べて表示する。\n\nSHOT 3：横向き（左）の立ち姿と、その横顔の実写クローズアップを並べて表示する。\n\nSHOT 4：後ろ向きの立ち姿と、その後頭部の実写クローズアップを並べて表示する。\n\nSHOT 5：正面の立ち姿と、表情差分6パターン（普通・笑顔・怒り・驚き・悲しみ・ウインク）の実写クローズアップを並べて表示する。\n\nSHOT 6：正面の立ち姿と、身に着けているもの（衣装・小物・アクセサリー）の素材ディテール実写マクロ接写を並べて表示する。\n\nSHOT 7：正面の立ち姿・横向き（右）の立ち姿・後ろ向きの立ち姿と、正面の顔の実写クローズアップをまとめて1枚に並べる\n──────────────────\n@Hailuo_AI  #MiniMaxH3",
    "approvedAt": "2026-08-22T09:15:40.742Z",
    "editorialBasis": "The case retains the original creator attribution and X status URL; the visible video supports the editorial title and summary."
  },
  {
    "id": "x-2091392123497279974",
    "title": "雷华扇舞角色预告",
    "titleEn": "Thunder Fan-Dance Character Trailer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "古装角色先在暗色舞台亮相，随后以中英标题卡、雷电和发光文字完成预告片式揭幕。 原生视频规格为 30 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A costumed character enters on a dark stage before bilingual title cards, lightning, and glowing typography complete a trailer-style reveal. Native-video output: 30s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cia0_exe/status/2091392123497279974",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-08-23T05:08:33.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091392123497279974.mp4",
    "posterUrl": "/posters/x/2091392123497279974.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "dark"
    ],
    "scenes": [
      "combat",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n\nmake a trailer skill character with thunder elementals by #MiniMaxH3 complete wth typography art\n\n7,438 chinese prompt",
    "engagement": {
      "replies": 22,
      "reposts": 2,
      "likes": 40,
      "views": 12592,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091610649914962079",
    "title": "邻里归家的普通一天",
    "titleEn": "An Ordinary Walk Home in Suburbia",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名女子从街边走到门廊，再进入家门；克制的日常动作和连续空间构成写实短片。 原生视频规格为 28 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman walks from a suburban street to her porch and enters the house, forming a restrained, realistic everyday sequence. Native-video output: 28s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2091610649914962079",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-23T19:36:53.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091610649914962079.mp4",
    "posterUrl": "/posters/x/2091610649914962079.jpg",
    "duration": 28,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This was AI generated. \n\nBeen mostly trying all kinds of weird experiments to push the MiniMax H3 model, so this time I tried a completely normcore video where nothing spectacular happens, but shot in 720p. Can you tell it's AI? If so, where?",
    "engagement": {
      "replies": 16,
      "reposts": 0,
      "likes": 44,
      "views": 4606,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091681626480873718",
    "title": "粉色女仆的室内招呼",
    "titleEn": "A Pink-Clad Maid Greets the Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉色女仆角色从白底展示转入明亮室内，以挥手、坐姿和表情变化完成角色短片。 原生视频规格为 10 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-clad maid moves from a clean character showcase into a bright room, greeting the viewer through gestures, poses, and expressions. Native-video output: 10s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tototoP210/status/2091681626480873718",
    "sourceLabel": "X 原帖 · @tototoP210",
    "author": "永久恋愛ととと",
    "publishedAt": "2026-08-24T00:18:55.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091681626480873718.mp4",
    "posterUrl": "/posters/x/2091681626480873718.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今日のMiniMaxH3のテストのハルちゃん\n音声は差し替え\nアップスケールを使おうとしたけど、最初から画質夜雨にしてゆっくり作った方が良いものができることが判明してしまった\nやはり時間はかけなきゃダメか　ケチケチせずに",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 96,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091679892442149010",
    "title": "冰原企鹅乐队进行曲",
    "titleEn": "An Ice-Cave Penguin Band Anthem",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴皇冠的企鹅吉他手与礼帽鼓手在水晶冰洞中轮番演奏，最终组成完整乐队。 原生视频规格为 240 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A crowned penguin guitarist and top-hatted drummer perform inside a crystal ice cave before assembling as a full band. Native-video output: 240s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IthacaNFT/status/2091679892442149010",
    "sourceLabel": "X 原帖 · @IthacaNFT",
    "author": "Ithaca Pengu 🐧✳️",
    "publishedAt": "2026-08-24T00:12:02.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091679892442149010.mp4",
    "posterUrl": "/posters/x/2091679892442149010.jpg",
    "duration": 240,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Had fun making this @pudgypenguins anthem with MiniJam; the music video was made using Maestro w MiniMax H3… \n\nall 100% local using zero data centers, thanks to @cocktailpeanut’s Pinokio 🐧🫡",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 9,
      "views": 850,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091679311623548944",
    "title": "猫耳少女从夜晚醒到清晨",
    "titleEn": "A Cat-Eared Girl Wakes Into Morning",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "猫耳少女在夜色卧室中醒来，镜头经过床边特写，最后走入被夕阳照亮的街道。 原生视频规格为 65 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared girl wakes in a dark bedroom, passes through bedside close-ups, and finally walks into a sunset-lit street. Native-video output: 65s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/alfredplpl/status/2091679311623548944",
    "sourceLabel": "X 原帖 · @alfredplpl",
    "author": "あるふ",
    "publishedAt": "2026-08-24T00:09:44.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091679311623548944.mp4",
    "posterUrl": "/posters/x/2091679311623548944.jpg",
    "duration": 65,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "クリエイティブエージェントによる曲と参照画像からの全自動MV生成。#MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 690,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091678907917308060",
    "title": "纸艺地标从金字塔切换到马德里",
    "titleEn": "Paper-Theatre Landmarks Shift from Pyramids to Madrid",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "折纸舞台先升起金字塔，再以刷痕标题切换巴塞罗那与马德里的城市地标。 原生视频规格为 17 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A folded-paper stage raises pyramids, then uses brush-stroke titles to transition between landmarks in Barcelona and Madrid. Native-video output: 17s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wakeupgfx/status/2091678907917308060",
    "sourceLabel": "X 原帖 · @wakeupgfx",
    "author": "Rubén Rivas",
    "publishedAt": "2026-08-24T00:08:07.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091678907917308060.mp4",
    "posterUrl": "/posters/x/2091678907917308060.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🧪 AI Test\n\nRan @techhalla's paper-theatre H3 prompt across 4 capitals. Text held clean in every frame — until an object crossed it mid-cut. Fixed by staging, not rewriting.\n\nLandmarks are harder: \"oval stadium\" = Colosseum until you feed it a real photo.\n\n#MinimaxH3 #AIWorkflow",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 18,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091678685984329849",
    "title": "社保事务所信息动态图形",
    "titleEn": "Social-Insurance Office Motion Explainer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "米白和橙色信息图以文档、日历与办公人物为核心，组合成一支日语事务所说明动画。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beige-and-orange graphics combine documents, calendars, and office characters into a Japanese social-insurance explainer. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2091678685984329849",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-08-24T00:07:14.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091678685984329849.mp4",
    "posterUrl": "/posters/x/2091678685984329849.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI × モーショングラフィックス\n\nMiniMax H3\n\nRunway(@runwayml ) unlimitedで制作\n\nprompt:",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 21,
      "views": 1487,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091676198485328169",
    "title": "牙刷日的日语信息动画",
    "titleEn": "A Japanese Toothbrush-Day Motion Card",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝色迷你角色与巨大日期数字依次翻动，最后落在牙刷日口号和刷牙图标上。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny blue character and oversized date numerals animate into a final Toothbrush Day message and brushing icon. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/x_nonotan/status/2091676198485328169",
    "sourceLabel": "X 原帖 · @x_nonotan",
    "author": "ののたん",
    "publishedAt": "2026-08-23T23:57:21.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091676198485328169.mp4",
    "posterUrl": "/posters/x/2091676198485328169.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "8/24は歯ブラシの日ということで🪥\n\n自分でやってみると改めてminimaxH3によるリリックモーションの凄さがわかりますね😆\n\n日本語も結構な精度でちゃんと出るし、ほんとあとは演出力と言語化能力ですね🤔\n\n#AIイラスト #AIアニメ \ncreated by Polloai",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 58,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091675224844718539",
    "title": "未来舱室的物件与人物变换",
    "titleEn": "Object and Character Transitions in a Futuristic Chamber",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "未来感舱室里，手指点向托盘物件，微缩人物与远处光门随镜头切换出现。 原生视频规格为 21 秒 · 1024×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Inside a futuristic chamber, a hand points to objects on a tray as miniature figures and a distant light portal appear across cuts. Native-video output: 21s · 1024×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pepe_program/status/2091675224844718539",
    "sourceLabel": "X 原帖 · @pepe_program",
    "author": "ペペチ児",
    "publishedAt": "2026-08-23T23:53:29.000Z",
    "addedAt": "2026-08-24T01:35:51.925Z",
    "mediaUrl": "/media/x-2091675224844718539.mp4",
    "posterUrl": "/posters/x/2091675224844718539.jpg",
    "duration": 21,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMaxH3\nわかってきたー",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 9,
      "snapshotAt": "2026-08-24T01:35:51.925Z"
    },
    "approvedAt": "2026-08-24T01:35:51.925Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091674280849551824",
    "title": "泳池边的人像镜头转换",
    "titleEn": "Poolside Portrait Shot Transitions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "同一位粉色泳装人物在泳池边由全身镜头切换到面部近景，再回到环境人像。 原生视频规格为 28 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same pink-swimsuit subject moves from a full-body poolside view to facial close-ups and back to an environmental portrait. Native-video output: 28s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AiCreatorS1881/status/2091674280849551824",
    "sourceLabel": "X 原帖 · @AiCreatorS1881",
    "author": "こたじろう",
    "publishedAt": "2026-08-23T23:49:44.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091674280849551824.mp4",
    "posterUrl": "/posters/x/2091674280849551824.jpg",
    "duration": 28,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "少し大人になったYayoiさんのIV\n\nK2R_yayoi_S1_1k(LoRA)を使用した静止画をMiniMax H3で動画化。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 489,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091673049884176816",
    "title": "自拍女性穿行长廊的连续镜头",
    "titleEn": "A Selfie Walk Through Connected Hallways",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "女性保持自拍构图穿过室内长廊与公共空间，画面持续保留参考图和人物一致性标记。 原生视频规格为 65 秒 · 960×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman keeps a selfie composition while moving through hallways and public interiors, retaining the reference-image identity across the long sequence. Native-video output: 65s · 960×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DeepForest_Inc/status/2091673049884176816",
    "sourceLabel": "X 原帖 · @DeepForest_Inc",
    "author": "🟢DeepForest_Studio_AI🟢",
    "publishedAt": "2026-08-23T23:44:51.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091673049884176816.mp4",
    "posterUrl": "/posters/x/2091673049884176816.jpg",
    "duration": 65,
    "aspectRatio": "landscape",
    "resolution": "960×576",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 + ComfyUI-H3-Multishot を使用して引用元の65秒無言版から、今度は英語セリフ付きに挑戦！🙌\n\n3枚のリファレンス画像を使って65秒の長時間生成✨\n人物IDと背景の連続性を維持しながら、英語セリフもかなり安定。長尺でも最後までシームレスに生成出来た😎\n\n#MiniMaxH3 #ComfyUI\n\n補足😊\n\n日本語のセリフも生成できますが、長尺で試していると、指定したセリフ以外の意味不明な発話が入ることが何度かありました。\n\nそこで今回は、安定性を重視してセリフを英語に変更。指定した内容をかなり安定して話してくれました。少なくとも今回のテストでは、長尺の音声生成は日本語より英語の方が安定している印象です。\n\nまた今回は3枚のリファレンス画像を使用。動画内左側に表示している3枚が実際に使用したReference Imagesです。65秒を通して人物のIDはかなりよく維持され、廊下から図書館へ移動しても背景の連続性はほぼ保たれました。\n\n一方で、長尺になるにつれて後半では口の動きと音声にほんの少しズレが出る場面もありました。また、人物のIDは維持されているものの、後半になるにつれて画質は少しずつ低下していきます。\n\nそれでも、人物・背景・音声を維持しながら65秒まで連続生成できたのはかなり面白い結果でした。まだまだ色々試せそうです😊",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 23,
      "views": 1370,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091672323611095231",
    "title": "霓虹舞台上的周一角色秀",
    "titleEn": "Monday Character Show on a Neon Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "彩色角色在竖屏霓虹舞台中旋转起舞，最后以醒目的日语周一标题收束。 原生视频规格为 5 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A colorful character spins and dances on a vertical neon stage before a bold Japanese Monday title closes the clip. Native-video output: 5s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nanohana_hanana/status/2091672323611095231",
    "sourceLabel": "X 原帖 · @nanohana_hanana",
    "author": "Hanana",
    "publishedAt": "2026-08-23T23:41:57.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091672323611095231.mp4",
    "posterUrl": "/posters/x/2091672323611095231.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "全力月曜日！\n\nMiniMaxH3\nタイトルロゴはアマモリさんのプロンプトを使わせていただきました！かわいい。\n今週も頑張りましょう。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 4,
      "views": 133,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091667671771140454",
    "title": "手工皂从静物到产品广告",
    "titleEn": "Handmade Soap Turned into a Product Spot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "花朵与手工皂静物经过微距质感镜头和英文卖点字幕，组成完整产品广告。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Floral handmade-soap still lifes become a complete product spot through macro texture shots and English selling-point captions. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AllaAisling/status/2091667671771140454",
    "sourceLabel": "X 原帖 · @AllaAisling",
    "author": "Alexandra Aisling",
    "publishedAt": "2026-08-23T23:23:28.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091667671771140454.mp4",
    "posterUrl": "/posters/x/2091667671771140454.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Using MiniMax Design in @Hailuo_AI  and agentic workflows, I built this ad, starting from a handmade soaps image ⤵️\n\n #MiniMaxH3 on #MiniMaxDesign",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 7,
      "views": 396,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091663130749440463",
    "title": "骑士查理王小猎犬的夜店宣传片",
    "titleEn": "A Cavalier Spaniel's Nightclub Promo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿粉色礼服的小猎犬在华丽夜店中登场，配合日语宣传文字完成幽默角色广告。 原生视频规格为 15 秒 · 1152×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A Cavalier Spaniel in a pink dress poses inside a lavish nightclub as Japanese promotional copy completes a playful character ad. Native-video output: 15s · 1152×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kakurega_films/status/2091663130749440463",
    "sourceLabel": "X 原帖 · @kakurega_films",
    "author": "苔こっこ｜AI映像・音楽制作",
    "publishedAt": "2026-08-23T23:05:26.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091663130749440463.mp4",
    "posterUrl": "/posters/x/2091663130749440463.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "当店人気No.1キャバ嬢、リアちゃん✨\n\n……キャバリア🐶だけに。笑\n\nMiniMax H3で、15秒の「嬢紹介プロモ」っぽく作ってみました。\n\nこういうくだらないダジャレを、\n無駄に高品質な映像にできるのも生成AIの面白いところ😂",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 92,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091662948381122658",
    "title": "梵高画作的厚涂文字变形",
    "titleEn": "Van Gogh Paintings Become Sculpted Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "厚重颜料纹理从画布表面展开，经过《星夜》式图像，最终凝成金色 VAN GOGH 立体字。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Thick paint textures spread across a canvas, pass through a Starry Night-inspired image, and resolve into sculpted gold VAN GOGH lettering. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/vkuoo/status/2091662948381122658",
    "sourceLabel": "X 原帖 · @vkuoo",
    "author": "vkuoo",
    "publishedAt": "2026-08-23T23:04:42.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091662948381122658.mp4",
    "posterUrl": "/posters/x/2091662948381122658.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created text animations with Minimax H3 on @krea_ai . ✨",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 258,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091661911469310184",
    "title": "金色英雄的漫画风音乐片段",
    "titleEn": "A Golden Hero in a Graphic Music Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "金色英雄在黑白漫画拼贴与强烈英文排版之间跃动，形成节奏鲜明的音乐视觉段落。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A golden hero moves through monochrome comic collages and bold English typography in a tightly paced music-video sequence. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MarkBenjaminAI/status/2091661911469310184",
    "sourceLabel": "X 原帖 · @MarkBenjaminAI",
    "author": "Mark Benjamin",
    "publishedAt": "2026-08-23T23:00:35.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091661911469310184.mp4",
    "posterUrl": "/posters/x/2091661911469310184.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Making MV has never been easier! \n\nI made this MV with a character reference, song from Suno only! \n\nMiniMax Design handled the rest! And what made that so easy is using MiniMaxH3, because it is truly great at displaying Text in videos, it doesn't show glitched Text like Seedance 2!\n\nHonestly, I really like MiniMax Design, and I think I'm gonna make my own skills there to make my workflows easier.\n\nIncredibly thankful to @Hailuo_AI for giving me the chance to try it!\n\n#MiniMaxDesign #MiniMaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 6,
      "views": 165,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091659975231988159",
    "title": "20项运动的高速图形巡礼",
    "titleEn": "A Fast Graphic Tour of 20 Sports",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "第一视角运动镜头与桌球等项目标题高速切换，最后汇聚为彩色 20 SPORTS 总览。 原生视频规格为 42 秒 · 1672×941。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "First-person athletic shots and title cards such as table tennis cut rapidly before converging into a colorful 20 SPORTS finale. Native-video output: 42s · 1672×941. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_animer/status/2091659975231988159",
    "sourceLabel": "X 原帖 · @ai_animer",
    "author": "雨雲(Amagumo)806 ｜ AI Animer",
    "publishedAt": "2026-08-23T22:52:53.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091659975231988159.mp4",
    "posterUrl": "/posters/x/2091659975231988159.jpg",
    "duration": 42,
    "aspectRatio": "landscape",
    "resolution": "1672×941",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "20 SPORTS — ONE MOTION [PR]\n\nMinMax H3のT2V（参照物なし・プロンプトのみ）で\n20スポーツの紹介動画を作ってみました。\n\nPOVを多用した2Dアニメで、迫力とスピード感を表現してました🔥\n\nワークフロー & プロンプトはリプ欄からTapTV👇\n\nAI & Tools\nワークフロー：TapNow\n動画：MiniMax H3\n音楽：SUNO\n編集：Filmora",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 27,
      "views": 550,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091657957981261970",
    "title": "黑白房间里的持枪对峙",
    "titleEn": "A Gunpoint Standoff in a Monochrome Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白室内场景中，持枪女子与头巾男子近距离对峙，越肩镜头强化紧张感。 原生视频规格为 14 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a monochrome room, an armed woman confronts a turbaned man at close range, with over-the-shoulder angles heightening the tension. Native-video output: 14s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nonameoasis/status/2091657957981261970",
    "sourceLabel": "X 原帖 · @nonameoasis",
    "author": "nonameoasis",
    "publishedAt": "2026-08-23T22:44:52.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091657957981261970.mp4",
    "posterUrl": "/posters/x/2091657957981261970.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Original Prompt from @CharaspowerAI \ni made this scene with Jennai in #MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 3,
      "views": 41,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091656223523209336",
    "title": "冰链破碎后的奇幻女战士",
    "titleEn": "A Fantasy Warrior Emerges from Shattering Ice Chains",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "透明冰链在特写中崩裂，碎片飞散后显露站在荒原中的蓝衣奇幻女战士。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Transparent ice chains fracture in close-up, scattering shards that reveal a blue-clad fantasy warrior standing in a barren landscape. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nonameoasis/status/2091656223523209336",
    "sourceLabel": "X 原帖 · @nonameoasis",
    "author": "nonameoasis",
    "publishedAt": "2026-08-23T22:37:59.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091656223523209336.mp4",
    "posterUrl": "/posters/x/2091656223523209336.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Well Jennai has gone out to a starring role, this time with #MiniMaxH3 in @Hailuo_AI  that looks good Goddess that you think? ⤵️ You can guess what Goddess is about?",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 48,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091654707236499474",
    "title": "夕阳下的古装人物对话",
    "titleEn": "A Period-Drama Conversation at Sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "古装男女在夕阳山景前近距离交谈，交替特写突出人物表情与情绪张力。 原生视频规格为 31 秒 · 2576×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A period-costumed man and woman converse against a sunset mountain backdrop, with alternating close-ups emphasizing facial emotion. Native-video output: 31s · 2576×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PixelAigc/status/2091654707236499474",
    "sourceLabel": "X 原帖 · @PixelAigc",
    "author": "PixelAIGC",
    "publishedAt": "2026-08-23T22:31:57.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091654707236499474.mp4",
    "posterUrl": "/posters/x/2091654707236499474.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "2576×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "我把那个电影对话做成skill\n生成了一些电影对话场景提示词\n在minimax h3 本地部署里跑出30秒视频\n效果还不错\n看看这个大话西游的场景复刻\n怎么样？\n\n提示词如下：⬇️",
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 10,
      "views": 1071,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091646670392164706",
    "title": "红黑剪影的歌词动画",
    "titleEn": "Red-and-Black Lyric Animation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红黑人物剪影、棋盘格与大号英文歌词在横屏画面中连续重组。 原生视频规格为 25 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Red-and-black character silhouettes, checkerboard patterns, and oversized English lyrics continuously recompose across a widescreen frame. Native-video output: 25s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiehon_aya/status/2091646670392164706",
    "sourceLabel": "X 原帖 · @aiehon_aya",
    "author": "妖精アーヤさん",
    "publishedAt": "2026-08-23T22:00:01.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091646670392164706.mp4",
    "posterUrl": "/posters/x/2091646670392164706.jpg",
    "duration": 25,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMAX H3 × モーショングラフィック\nリリックMV\n──────\nモーグラは2Dベタ塗と相性いい。\n2Dベタ塗の画像生成はSeedream 5 proが得意。\n──────\nSousaku AI ならどっちも使えるし、なんならH3が9/30まで50%オフで使えてお得よ https://sousaku.ai/signin?share_code=WP6H4K  #SousakuAI [PR]",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 15,
      "views": 458,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091645327874932893",
    "title": "暗黑王冠徽标揭幕",
    "titleEn": "A Dark Crown Logo Reveal",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红色能量花在黑底中绽放，粒子凝聚成王冠与 THE BURIED KING 标志。 原生视频规格为 5 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red energy bloom opens against black before particles form a crown and THE BURIED KING emblem. Native-video output: 5s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NigelWatsonTFTC/status/2091645327874932893",
    "sourceLabel": "X 原帖 · @NigelWatsonTFTC",
    "author": "Nigel Watson",
    "publishedAt": "2026-08-23T21:54:41.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091645327874932893.mp4",
    "posterUrl": "/posters/x/2091645327874932893.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3 is really good for logo reveals",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 5,
      "views": 332,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091644432433234415",
    "title": "彩色几何节拍中的舞者剪影",
    "titleEn": "A Dancer Silhouette in Geometric Beat Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色舞者剪影在黄、青、粉色几何背景间切换，MOVE 与 BEAT 字样跟随节拍出现。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black dancer silhouette switches between yellow, cyan, and pink geometric backgrounds as MOVE and BEAT typography lands on the rhythm. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2091644432433234415",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-08-23T21:51:08.000Z",
    "addedAt": "2026-08-24T01:35:51.926Z",
    "mediaUrl": "/media/x-2091644432433234415.mp4",
    "posterUrl": "/posters/x/2091644432433234415.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "dance",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「MiniMax H3 x ダンス」改良版\nダンスもモーショングラフィックも変えてみた\n\nMiniMax H3のt2vで生成\n#MiniMaxDesign @Hailuo_AI \n\nプロンプトはこちら↓\n「MiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィー動画を制作する。\n\n主人公は、全身が黒一色で描かれたシンプルな人間のシルエット。人物を画面中央に配置し、頭から足先まで全身が見える構図を基本とする。顔、髪、服装などの細部は描かず、洗練された美しい人体のシルエットとして表現する。背景には明るく鮮やかな色を使用し、黒い人物との明確なコントラストを維持する。\n\n主人公は、プロのストリートダンサーが披露するような、クールでスタイリッシュなダンスを踊る。ヒップホップをベースに、ポッピング、ロッキング、グライディング、ボディアイソレーション、ヒールトゥ、細かく高速なフットワーク、腕を大きく使ったムーブを滑らかに組み合わせる。単に激しく動くのではなく、全身から余裕と自信が感じられる、音楽性の高い洗練されたパフォーマンスにする。足のステップ、身体のヒット、胸や肩の動き、腕の停止と解放を、キック、スネア、ベースなどのビートに正確にシンクロさせる。\n\nダンサーの動きに合わせて、背景の色、図形、線、円、グリッド、光、パターンがビートごとに次々と変化する。長方形、円、三角形、太いラインなどのシェイプが、画面の上下左右や奥行き方向から勢いよくスライドインし、次のビートでスライドアウトする。複数のシェイプが異なる方向から重なり合い、分割、回転、拡大、縮小を繰り返しながら、リズミカルに画面構成を作り替えていく。ただし、シェイプでダンサーの全身を隠さず、人物の動きは常に明確に見えるようにする。\n\n腕を鋭く振ると背景にラインが走り、足を踏み込むと円形のシェイプが広がり、身体をヒットさせると背景の色面が瞬時に切り替わる。ダンサーが身体を静止させる瞬間には、周囲のシェイプも一瞬動きを止め、次のビートで別方向へ一気にスライドする。人物のダンスと背景のモーショングラフィックが互いに反応し、一体となって動いているように見せる。\n\n背景には「DANCE」「MOVE」「BEAT」という単語を、正確なスペルの大文字でキネティック・タイポグラフィーとして登場させる。文字は一度表示されて終わるのではなく、登場後も音楽のビートに合わせて脈打つように拡大・縮小を繰り返し、前後左右への3D回転、奥行き方向への移動、軽い伸縮、色の切り替えを続ける。3D回転した文字は再び正面を向き、各単語が明確に読める瞬間を必ず作る。文字の欠落、重複、誤字、別の文字への変化は起こさず、「DANCE」「MOVE」「BEAT」以外の文字は表示しない。タイポグラフィーはダンサーの背後や周囲で動かし、黒いシルエットの視認性を保つ。\n\nカメラは基本的にダンサーの全身を正面から捉えながら、音楽の強いビートに合わせてアングルを大胆に切り替える。正面からの構図に加えて、人物を斜め上から見下ろすダイナミックな俯瞰ショットと、斜め下から見上げる力強いローアングルショットを入れる。俯瞰ショットでは床面を流れるシェイプやタイポグラフィーとダンサーのフットワークを印象的に見せ、ローアングルショットではダンサーの腕や上半身の動きを大きく、存在感のあるシルエットとして捉える。\n\nさらに、短いズームイン、ズームアウト、左右へのスライド、わずかなカメラの傾きや回り込みを加える。カットが切り替わってもダンスの動きとリズムを途切れさせず、同じ人物による一続きの振り付けとして自然につなげる。どのアングルでも全身をできるだけ画面内に収め、人物が中央から大きく外れないようにする。背景、シェイプ、タイポグラフィーの目まぐるしい動きと大胆なアングル変化によって、映像に立体感、スピード感、ファッション映像のような洗練された印象を与える。\n\n人物の形状、体格、手足の本数を15秒間一貫させ、手足の変形、身体の分裂、不自然な残像を発生させない。アングルが変わっても、同一人物の黒いシルエットを維持する。\n\n大胆でありながら統一感のある配色、シャープなグラフィック、強いコントラスト、正確なビートシンクによって、とにかくカッコよく、スタイリッシュで、ファッション性の高い映像に仕上げる。終盤に向けてダンス、シェイプ、タイポグラフィーの動きを徐々に高密度にし、最後は大げさな決めポーズを取らず、ビートの余韻に合わせて身体の力を抜き、片足に体重を預けた自然で余裕のある姿勢でクールに締めくくる。SNSで拡散されそうな、センスが良く、中毒性があり、何度も見返したくなる高品質な映像にする。」",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 19,
      "views": 1039,
      "snapshotAt": "2026-08-24T01:35:51.926Z"
    },
    "approvedAt": "2026-08-24T01:35:51.926Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091749172361978149",
    "title": "从洗衣到晾晒的连续家务",
    "titleEn": "A Continuous Laundry Day from Sorting to Folding",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "一名女子从分类衣物、装入洗衣机，到阳台晾晒和桌边折叠，连续完成一整套日常家务。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman moves continuously from sorting and loading laundry to balcony drying and table-side folding. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A young Japanese woman carries a full laundry basket from her bedroom to a small laundry area, kneels down, and actively sorts white and dark clothes into separate piles. She then picks up the laundry, walks several steps to the washing machine, loads the clothes one by one, closes the door, and starts the machine. The scene transitions to her walking onto the balcony with freshly washed laundry, placing the basket down, shaking out a wet white shirt, putting it on a hanger, and hanging it on the drying rack as a gentle breeze moves the clothes naturally. After the clothes dry, she walks back inside carrying a stack of laundry, places it on a table, and actively folds a T-shirt by shaking it open, folding the sleeves inward, folding it in half, and placing it onto the stack before starting the next garment. Photorealistic Japanese lifestyle video, natural handheld camera, realistic household physics, authentic walking and body movement, natural fabric behavior, realistic washing-machine motion, subtle wind, continuous action, neutral morning daylight, consistent character and environment, seamless transitions, no static posing, no teleportation, no floating objects, no distorted hands, no CGI appearance, no text overlays, logos, or watermarks.",
    "promptSourceUrl": "https://x.com/aiwithaly/status/2091749172361978149",
    "sourceUrl": "https://x.com/aiwithaly/status/2091749172361978149",
    "sourceLabel": "X 原帖 · @aiwithaly",
    "author": "Ali",
    "publishedAt": "2026-08-24T04:47:20.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2091749172361978149.mp4",
    "posterUrl": "/posters/x/2091749172361978149.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "ugc"
    ],
    "scenes": [
      "combat",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A simple laundry day, brought to life with natural movement and everyday realism. \n\nFrom sorting and washing to hanging and folding, every step feels authentic and continuous.\n\nCreated with MiniMax H3.\n\nPrompt:\n\nA young Japanese woman carries a full laundry basket from her bedroom to a small laundry area, kneels down, and actively sorts white and dark clothes into separate piles. She then picks up the laundry, walks several steps to the washing machine, loads the clothes one by one, closes the door, and starts the machine. The scene transitions to her walking onto the balcony with freshly washed laundry, placing the basket down, shaking out a wet white shirt, putting it on a hanger, and hanging it on the drying rack as a gentle breeze moves the clothes naturally. After the clothes dry, she walks back inside carrying a stack of laundry, places it on a table, and actively folds a T-shirt by shaking it open, folding the sleeves inward, folding it in half, and placing it onto the stack before starting the next garment. Photorealistic Japanese lifestyle video, natural handheld camera, realistic household physics, authentic walking and body movement, natural fabric behavior, realistic washing-machine motion, subtle wind, continuous action, neutral morning daylight, consistent character and environment, seamless transitions, no static posing, no teleportation, no floating objects, no distorted hands, no CGI appearance, no text overlays, logos, or watermarks.",
    "engagement": {
      "replies": 49,
      "reposts": 5,
      "likes": 217,
      "views": 9905,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091775336849789065",
    "title": "喷漆唤醒墙上巨龙",
    "titleEn": "Graffiti Paint Awakens a Wall-Bound Dragon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝发涂鸦师按下喷漆罐，金色笔触穿过标题卡并唤醒墙上的彩色巨龙。 原生视频规格为 29 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired graffiti artist sprays a golden stroke through a title card and awakens a colorful dragon in the wall. Native-video output: 29s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cia0_exe/status/2091775336849789065",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-08-24T06:31:18.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2091775336849789065.mp4",
    "posterUrl": "/posters/x/2091775336849789065.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n\nGraffiti soul - Painting brings dragon to life\n\n4,567 Chinese prompt",
    "engagement": {
      "replies": 24,
      "reposts": 4,
      "likes": 36,
      "views": 6785,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091893350748025010",
    "title": "一根金线织出奇幻世界与徽标",
    "titleEn": "One Golden Thread Weaves a Fantasy World and Emblem",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一根发光金线在黑色织物上分叉，编织出山脉、城邦和道路，最终汇成浮雕徽标。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A glowing golden thread branches across dark fabric, weaving mountains, kingdoms, and roads before converging into an embossed emblem. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create an epic fantasy title sequence built around a vast living tapestry. The sequence must feel like an entire world is being born from a single thread.\n\nBegin in near darkness. Only the faint texture of woven fabric is visible. At the very center, one lone thread awakens. It trembles, pulls tight, glows faintly, and begins moving across the surface like a living force. As it races forward, it splits into more threads. Those threads split again, multiplying rapidly into branching strands of gold, crimson, ivory, midnight blue, and ash-gray. Everything in the sequence must originate from this first thread and its descendants. The world should feel as if it is being generated from one living source.\n\nThe camera moves dynamically and continuously through the tapestry, never static. It races low across the woven surface, whips past rising embroidered forms, glides along rivers made of flowing thread, dives through deep folds and valleys of cloth, and sweeps between giant raised textile structures. The camera must always remain on the exposed upper surface of the tapestry. It may skim across folds and pass through openings between raised embroidered forms, but it must never travel beneath the fabric.\n\nThe world should not simply appear. It must actively weave itself into existence in real time. At every moment, visible threads must remain in motion, constantly splitting, branching, knotting, braiding, tightening, and constructing new terrain, landmarks, symbols, and regions ahead of the camera. The world must never feel complete until the final logo reveal. The strongest visual rule is this: if something appears, it should be because threads are physically making it.\n\nCities, kingdoms, and landmarks must all be built through textile transformation. Threads knot together into towers, fortresses, ceremonial structures, bridges, and gates. Layers of cloth fold upward into cliffs, walls, and ridges. Roads are sewn rapidly across the land. Banners unfurl from the weave. Forests spread outward stitch by stitch as fibers become roots, trunks, branches, and canopies. Rivers are pulled into place by shimmering strands. Coastlines tighten into shape like stretched cloth under tension. Nothing should materialize passively. Everything must be woven, stitched, tied, pulled, or braided into existence by moving thread.\n\nEach major region should feel like its own construction event. In one kingdom, royal motifs and heraldic embroidery bloom outward in ornate patterns. In another, scorched threads blacken and curl as if marked by war. In a frozen region, pale threads crystallize into icy stitched structures. In a desert realm, sand-colored silk folds into dunes and radiant sun motifs. In a battlefield region, red threads race violently across the surface like spreading conflict. In every case, the region should be visibly built by thread activity, not simply revealed.\n\nUse aggressive but elegant camera choreography: fast push-ins, sweeping arcs, sharp climbs, dramatic pull-backs, overhead reveals, and low-angle passes through giant stitched landmarks and dense moving fibers. Let the camera chase the leading edge of construction. Often follow one strand as it races into empty fabric and begins creating the next formation. The camera should feel as if it is surfing the growth of the world itself.\n\nThe tapestry should feel handcrafted, ancient, ceremonial, and mythic. Show rich textile detail everywhere: layered cloth, frayed edges, gold thread, dense embroidery, woven symbols, stitched heraldry, seams, folds, braided cords, hanging fibers, and visible tension in the fabric. If desired, briefly glimpse looms, spools, or unseen weaving forces at the edges of the frame, but keep the main focus on the living tapestry and its self-weaving world.\n\nThe pacing should escalate constantly. Start with the mystery of one awakened thread. Then let that thread multiply into many. Then let those many threads begin forming entire regions. Then let multiple regions awaken at once as long strands travel across the tapestry and connect distant kingdoms, roads, battles, and symbols. The sequence should feel as if history, bloodlines, war, and power are all being woven from one original source.\n\nLighting should be dramatic and cinematic, with warm highlights skimming across thread textures, deep shadows in folds of cloth, soft glow passing through thin fabric, and floating dust in the air. The tone should feel majestic, ancient, noble, immersive, and alive.\n\nToward the climax, reveal that all kingdoms, roads, battles, and symbols are interconnected through one vast woven design. Long threads travel across the entire tapestry and begin drawing everything toward a central point. Regional patterns, sigils, and story elements align into one unified composition.\n\nIf a logo reference image is supplied, use that logo as the exact design authority for the ending. Preserve its silhouette, proportions, and core graphic structure. Do not simply reveal, overlay, or fade in the logo. Instead, the final logo must be physically built in real time from the same thread lineage that created the world. The first awakened thread and all the threads that descended from it should ultimately converge and construct the final emblem. Kingdoms, roads, rivers, banners, heraldic motifs, and regional patterns must unravel back into fibers and feed the final construction. Braided cords define the main silhouette. Dense embroidery fills the interior forms. Fine metallic threads complete the small details. The final logo should feel as if the whole world has transformed into it.\n\nIf no logo reference image is supplied, end with an original fantasy crest or title emblem emerging through the same thread-construction logic.\n\nEnd with the completed logo fully stitched into the tapestry as a monumental embroidered insignia. The surrounding lands continue to ripple subtly with life as a few final strands glide into place, tighten, and lock the design.\n\nTone: majestic, ancient, dramatic, elegant, immersive, ceremonial, and alive.\nStyle keywords: living tapestry, single-origin thread, embroidered fantasy world, woven kingdoms, dynamic textile transformation, cinematic title sequence, mythic handcrafted world-building, epic camera motion.\nFormat: 16:9, highly detailed, no modern elements, no direct copy of any existing franchise.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2091893704101367936",
    "sourceUrl": "https://x.com/aimikoda/status/2091893350748025010",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-24T14:20:14.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2091893350748025010.mp4",
    "posterUrl": "/posters/x/2091893350748025010.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 - Tapestry\n\nOne thread turns into a whole world.\n\nIt splits, travels across the tapestry and keeps building everything in its path, all the way to the final logo. \n\nJust give your logo as a reference. Prompt in the replies.",
    "engagement": {
      "replies": 13,
      "reposts": 10,
      "likes": 134,
      "views": 5149,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091936838944715142",
    "title": "后室空间的一镜长途穿行",
    "titleEn": "A One-Shot Journey Through the Backrooms",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从空旷后室走廊持续前进，穿入地铁车厢并来到站台，完成近一分钟的空间续接实验。 原生视频规格为 56 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera advances through empty backrooms corridors, enters a subway car, and reaches a platform in a nearly one-minute continuation experiment. Native-video output: 56s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2091936838944715142",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-24T17:13:03.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2091936838944715142.mp4",
    "posterUrl": "/posters/x/2091936838944715142.jpg",
    "duration": 56,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Long video generation experiment -- Generated the entire 1 minute Backrooms short film with MiniMax H3, on my PC.\n\nDidn't have to stitch short clips or anything, generated in one shot. Yes there are some quirks but this was the first attempt, it will only get better.",
    "engagement": {
      "replies": 5,
      "reposts": 6,
      "likes": 66,
      "views": 3569,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092053560989270489",
    "title": "女学生穿越高密度动漫动态图形",
    "titleEn": "A Schoolgirl Races Through Dense Anime Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "动漫女学生骑车、奔跑并穿过眼睛特写和物件转场，在高密度图形层之间保持连续动作。 原生视频规格为 17 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An anime schoolgirl cycles and runs through eye close-ups and object-driven transitions across dense motion-graphic layers. Native-video output: 17s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create an extremely dense, highly sophisticated 18-second Japanese anime motion-graphics video.\n\nUse the supplied reference video as a visual and editing reference.\n\nThe goal is NOT to reproduce the reference literally.\n\nInstead, create a more complex evolution of its visual language:\nanime character animation + everyday objects + manga graphics + cinematic camera movement + layered 2.5D compositing + rapid transformations + strong easing and acceleration.\n\nThe result should look like an exceptionally detailed Adobe After Effects composition made from many independently animated layers.\n\nIMPORTANT:\nDo NOT use the visual technique of large rectangles sliding across the screen as transitions.\n\nDo NOT make scenes transition by moving rectangular panels.\n\nDo NOT create the impression of PowerPoint slides or template motion graphics.\n\nTransitions must be caused by actual objects, character movement, camera movement, depth, perspective, particles, paper, hair, environmental objects or rapidly changing compositions.\n\nMAIN CHARACTER:\n\nA short Japanese middle-school girl, approximately 13–14 years old, anime style.\n\nShoulder-length dark brown hair.\nExpressive brown eyes.\nJapanese middle-school sailor-style school uniform.\nWhite blouse.\nNavy skirt.\n\nKeep her appearance consistent throughout the entire video.\n\nShe is the central character.\n\nHer entire body should remain anatomically correct whenever she is shown in a full-body shot.\n\nAnimate:\nhair,\nskirt,\narms,\nhands,\nlegs,\nfeet,\nfacial expression,\neyes,\nbody weight,\nand secondary motion.\n\nThe girl must never appear frozen while only the camera moves.\n\n==================================================\nCORE MOTION PRINCIPLE\n=====================\n\nUse extremely strong Adobe After Effects-style easing.\n\nMotion must NOT have constant velocity.\n\nUse:\n\nslow anticipation,\nrapid acceleration,\nstrong motion blur,\nrapid deceleration,\novershoot,\nsettling,\nbrief holds,\nthen sudden acceleration again.\n\nUse different easing curves for different objects.\n\nSome objects should accelerate immediately.\n\nSome objects should move slowly and then suddenly accelerate.\n\nSome objects should overshoot their destination slightly and settle back.\n\nSome objects should stop abruptly.\n\nCreate rhythmic contrast between extremely fast movement and brief moments of stillness.\n\nThe video should breathe.\n\nDo NOT make everything move at the same speed.\n\nDo NOT make every movement smooth.\n\nUse deliberate timing similar to professional motion-graphics animation.\n\n==================================================\nINFORMATION DENSITY\n===================\n\nDouble the visual information density compared with a normal 18-second AI video.\n\nThis does NOT mean simply adding random objects.\n\nAt almost every moment, multiple independent events should occur simultaneously.\n\nFor example:\n\nthe girl is running,\nher hair is moving,\npapers are flying,\na bicycle wheel is rotating,\nbackground buildings are moving with parallax,\nsmall graphic lines are animating,\nthe camera is orbiting,\nand a foreground object is crossing the lens.\n\nAll movements must have different timing and different speeds.\n\nCreate a layered hierarchy:\n\nFOREGROUND:\nlarge objects extremely close to the camera.\n\nMIDGROUND:\ngirl and important objects.\n\nBACKGROUND:\nenvironment, buildings, sky, graphic elements.\n\nEach layer should move at a different speed.\n\n==================================================\n0–2 SECONDS\n===========\n\nBegin almost completely still.\n\nThe girl is standing in a quiet street.\n\nVery subtle movement:\nhair moving slightly,\na paper sheet on the ground,\nclouds slowly moving,\na bicycle wheel turning very slowly.\n\nHold this calm composition briefly.\n\nThen suddenly:\n\nthe girl's eyes look toward the camera.\n\nA paper airplane rapidly enters the frame.\n\nIt accelerates extremely quickly.\n\nThe camera instantly follows the paper airplane with a strong whip-pan.\n\nStrong motion blur.\n\nThe paper airplane passes extremely close to the lens and completely fills the frame naturally.\n\nThis is the transition.\n\nNO RECTANGULAR TRANSITION.\n\n==================================================\n2–4 SECONDS\n===========\n\nThe camera emerges from behind the paper airplane into a dense anime city.\n\nThe girl is running.\n\nThe camera tracks sideways.\n\nShe runs from right to left.\n\nBackground buildings move in strong parallax.\n\nForeground:\ntelephone pole,\ntree branches,\npaper,\nbicycle,\nschool bag.\n\nSome pass extremely close to the lens.\n\nThe girl suddenly jumps.\n\nUse a strong anticipation movement before the jump.\n\nShe crouches slightly.\n\nThen accelerates upward rapidly.\n\nAt the peak of the jump:\n\nTHE CAMERA SUDDENLY STOPS.\n\nThe girl continues moving.\n\nHer hair and skirt continue with secondary motion.\n\nThen the camera rapidly rotates approximately 90 degrees.\n\nThe entire environment transforms into a monochrome manga world.\n\nThe girl remains in full color.\n\n==================================================\n4–6 SECONDS\n===========\n\nThe manga environment becomes extremely dense.\n\nMultiple recognizable objects appear simultaneously:\n\nschool bag,\nnotebook,\npencil,\nsneakers,\nbicycle,\npaper airplane,\nschool bell,\napple,\nglass of water.\n\nDo not arrange them as static icons.\n\nEvery object has independent movement.\n\nThe bicycle wheel rotates.\n\nThe paper airplane flies.\n\nThe pencil rolls.\n\nThe notebook pages flutter.\n\nThe school bell swings.\n\nThe apple rotates slowly.\n\nThe water inside the glass moves naturally.\n\nThe camera travels diagonally between the objects.\n\nThe objects move at different depths.\n\nSuddenly the pencil accelerates toward the camera.\n\nIt passes directly in front of the lens.\n\nUse the pencil itself as the transition.\n\nNO RECTANGLES.\n\n==================================================\n6–8 SECONDS\n===========\n\nThe camera emerges from behind the pencil into a colorful anime music-video environment.\n\nThe girl is standing in the center.\n\nBehind her:\nfictional school-band performers,\nguitar,\nbass,\ndrums,\nmicrophone,\nspeakers,\nstage lights.\n\nEverything moves.\n\nThe drummer's arms move.\n\nThe guitar player moves.\n\nThe speaker cones vibrate.\n\nStage lights sweep across the scene.\n\nThe girl turns her head.\n\nHer hair follows with delayed secondary motion.\n\nThen she suddenly turns her entire body.\n\nThe camera performs a fast orbit around her.\n\nDuring the orbit, the background changes from the band stage into a summer landscape.\n\nUse a strong acceleration at the beginning of the orbit.\n\nSlow down dramatically near the end.\n\nBriefly hold the final position.\n\n==================================================\n8–10 SECONDS\n============\n\nExtreme close-up of the girl's face.\n\nVery calm.\n\nShe blinks.\n\nHer expression changes subtly.\n\nThe camera slowly moves sideways across her eye.\n\nThe iris contains a reflection of a completely different scene.\n\nThe camera suddenly accelerates toward the reflection.\n\nThe eye reflection expands naturally.\n\nEnter the reflected scene.\n\nNO SIMPLE ZOOM TRANSITION.\n\n==================================================\n10–12 SECONDS\n=============\n\nInside the reflection:\n\na table filled with recognizable everyday objects.\n\nGlass of water.\nPlate with cake.\nApple.\nNotebook.\nPencil.\nAlarm clock.\nSchool bag.\nSmall fan.\nPaper airplane.\n\nAll objects are alive with subtle independent animation.\n\nThe fan rotates.\n\nThe water moves.\n\nThe clock hands move.\n\nThe paper flutters.\n\nThe apple slowly rotates.\n\nThe cake remains physically stable.\n\nThe girl's hair moves from the airflow of the fan.\n\nShe reaches toward the glass.\n\nHer hand moves slowly.\n\nThen suddenly accelerates.\n\nShe touches the glass.\n\nThe water ripples.\n\nAt the exact same moment, the camera performs a rapid lateral movement.\n\nThe glass passes directly in front of the lens.\n\nUse the glass as the transition.\n\n==================================================\n12–14 SECONDS\n=============\n\nThe camera emerges from behind the glass.\n\nThe girl is running outdoors again.\n\nLow-angle camera.\n\nVery fast lateral tracking.\n\nTrees, houses, utility poles, clouds and bicycles move at different speeds.\n\nA bicycle suddenly crosses the foreground.\n\nThe camera follows the bicycle wheel.\n\nThe wheel rotates rapidly.\n\nThen the wheel slows dramatically.\n\nThe camera slows with it.\n\nEverything almost freezes.\n\nBrief moment of silence and visual stillness.\n\nThen:\n\nTHE WHEEL SUDDENLY ACCELERATES.\n\nThe camera is pulled into the spinning wheel.\n\n==================================================\n14–16 SECONDS\n=============\n\nThe spinning wheel transforms into a large technical illustration of the bicycle and motorcycle.\n\nBut do NOT use static technical diagrams.\n\nThe mechanical components move independently.\n\nWheel rotates.\n\nHandlebar turns.\n\nBrake lever moves.\n\nSuspension compresses.\n\nChain moves.\n\nSmall graphic lines animate around the machine.\n\nUse only meaningful readable labels:\n\n\"WHEEL\"\n\"BRAKE\"\n\"ENGINE\"\n\nThe girl walks through the technical layers.\n\nThe camera travels diagonally through the machine.\n\nShe reaches toward the camera.\n\nHer hand becomes extremely large in perspective.\n\nShe suddenly snaps her fingers.\n\nEverything freezes.\n\n==================================================\n16–18 SECONDS\n=============\n\nAbsolute stillness for a brief moment.\n\nThe girl looks directly at the camera.\n\nThen:\n\nALL ELEMENTS EXPLODE INTO MOTION.\n\nPaper airplanes.\nNotebook pages.\nComic lines.\nLeaves.\nGraphic circles.\nSmall photographs.\nPencils.\nMusical notes.\nBicycle parts.\nClouds.\nColored particles.\n\nAll move outward with different acceleration curves.\n\nThe camera rapidly orbits around the girl approximately 180 degrees.\n\nStrong motion blur.\n\nThe girl spins once.\n\nHer hair and skirt follow naturally.\n\nThe camera decelerates sharply.\n\nEverything suddenly becomes clean white.\n\nThe girl remains alone.\n\nShe looks toward the camera.\n\nSmall smile.\n\nReadable Japanese text:\n\n「行こう！」\n\nThe text appears cleanly.\n\nThe girl suddenly runs sideways out of frame.\n\nThe camera follows her for a fraction of a second.\n\nThen abrupt white flash.\n\nEND.\n\n==================================================\nCAMERA LANGUAGE\n===============\n\nCamera movement is a major part of the animation.\n\nUse:\n\nfast lateral tracking,\nslow lateral tracking,\ndiagonal tracking,\ncamera orbit,\n90-degree rotation,\n180-degree orbit,\nlow-angle tracking,\noverhead movement,\nwhip-pan,\nrapid parallax,\nforeground occlusion,\nlens-level object passes,\nsudden stops,\nsudden acceleration.\n\nDo NOT repeatedly move only forward and backward.\n\nDo NOT make the camera continuously fly through tunnels.\n\nCamera movement must have rhythm.\n\nUse:\n\nslow → fast → stop → slow → explosive acceleration → deceleration → hold → fast.\n\n==================================================\nSECONDARY MOTION\n================\n\nEverything should have secondary motion.\n\nHair follows the head with slight delay.\n\nSkirt follows body movement.\n\nPaper bends and rotates.\n\nWater ripples.\n\nClock hands move.\n\nWheels rotate.\n\nClouds drift.\n\nLeaves react to wind.\n\nBicycle components move.\n\nMusicians perform.\n\nSpeakers vibrate.\n\nGraphic elements react to nearby movement.\n\nDo not freeze background objects.\n\n==================================================\nTEXT\n====\n\nUse very little text.\n\nOnly use clearly readable real words.\n\nAllowed examples:\n\n\"START\"\n\"READY?\"\n\"GO!\"\n\"AFTER SCHOOL\"\n\"WHEEL\"\n\"BRAKE\"\n\"ENGINE\"\n\nJapanese:\n\n「放課後」\n「行こう！」\n\nNever generate meaningless text.\n\nNever generate fake Japanese.\n\nNever generate random alphabet strings.\n\nNever generate distorted pseudo-typography.\n\n==================================================\nVISUAL QUALITY\n==============\n\nHigh-quality Japanese anime.\n\nClean line art.\n\nDetailed character animation.\n\nProfessional motion-graphics composition.\n\nDense layered composition.\n\n2.5D depth.\n\nStrong parallax.\n\nCinematic lighting.\n\nControlled motion blur.\n\nSophisticated easing.\n\nStrong anticipation and follow-through.\n\nClear visual hierarchy.\n\nEven though the information density is extremely high, the main character must always remain visually understandable.\n\nThe video should feel like an ambitious professional Adobe After Effects music-video sequence rather than a collection of unrelated AI shots.\n\nNO large sliding rectangles.\n\nNO PowerPoint-like transitions.\n\nNO static object arrangements.\n\nNO meaningless text.\n\nNO random symbols.\n\nNO gibberish.\n\nNO repetitive camera movement.\n\nNO constant velocity.\n\nNO frozen characters.\n\nNO missing limbs.\n\nNO missing lower body.\n\nNO duplicated characters.\n\nNO deformed hands.\n\nPrioritize complex simultaneous motion, strong easing, camera choreography and physical object-based transitions.",
    "promptSourceUrl": "https://x.com/goldwing_zoome/status/2092053560989270489",
    "sourceUrl": "https://x.com/goldwing_zoome/status/2092053560989270489",
    "sourceLabel": "X 原帖 · @goldwing_zoome",
    "author": "goldwing",
    "publishedAt": "2026-08-25T00:56:52.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092053560989270489.mp4",
    "posterUrl": "/posters/x/2092053560989270489.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こんにちは動画。\n#MiniMaxH3\n\nプロンプトは以下です。\nCreate an extremely dense, highly sophisticated 18-second Japanese anime motion-graphics video.\n\nUse the supplied reference video as a visual and editing reference.\n\nThe goal is NOT to reproduce the reference literally.\n\nInstead, create a more complex evolution of its visual language:\nanime character animation + everyday objects + manga graphics + cinematic camera movement + layered 2.5D compositing + rapid transformations + strong easing and acceleration.\n\nThe result should look like an exceptionally detailed Adobe After Effects composition made from many independently animated layers.\n\nIMPORTANT:\nDo NOT use the visual technique of large rectangles sliding across the screen as transitions.\n\nDo NOT make scenes transition by moving rectangular panels.\n\nDo NOT create the impression of PowerPoint slides or template motion graphics.\n\nTransitions must be caused by actual objects, character movement, camera movement, depth, perspective, particles, paper, hair, environmental objects or rapidly changing compositions.\n\nMAIN CHARACTER:\n\nA short Japanese middle-school girl, approximately 13–14 years old, anime style.\n\nShoulder-length dark brown hair.\nExpressive brown eyes.\nJapanese middle-school sailor-style school uniform.\nWhite blouse.\nNavy skirt.\n\nKeep her appearance consistent throughout the entire video.\n\nShe is the central character.\n\nHer entire body should remain anatomically correct whenever she is shown in a full-body shot.\n\nAnimate:\nhair,\nskirt,\narms,\nhands,\nlegs,\nfeet,\nfacial expression,\neyes,\nbody weight,\nand secondary motion.\n\nThe girl must never appear frozen while only the camera moves.\n\n==================================================\nCORE MOTION PRINCIPLE\n=====================\n\nUse extremely strong Adobe After Effects-style easing.\n\nMotion must NOT have constant velocity.\n\nUse:\n\nslow anticipation,\nrapid acceleration,\nstrong motion blur,\nrapid deceleration,\novershoot,\nsettling,\nbrief holds,\nthen sudden acceleration again.\n\nUse different easing curves for different objects.\n\nSome objects should accelerate immediately.\n\nSome objects should move slowly and then suddenly accelerate.\n\nSome objects should overshoot their destination slightly and settle back.\n\nSome objects should stop abruptly.\n\nCreate rhythmic contrast between extremely fast movement and brief moments of stillness.\n\nThe video should breathe.\n\nDo NOT make everything move at the same speed.\n\nDo NOT make every movement smooth.\n\nUse deliberate timing similar to professional motion-graphics animation.\n\n==================================================\nINFORMATION DENSITY\n===================\n\nDouble the visual information density compared with a normal 18-second AI video.\n\nThis does NOT mean simply adding random objects.\n\nAt almost every moment, multiple independent events should occur simultaneously.\n\nFor example:\n\nthe girl is running,\nher hair is moving,\npapers are flying,\na bicycle wheel is rotating,\nbackground buildings are moving with parallax,\nsmall graphic lines are animating,\nthe camera is orbiting,\nand a foreground object is crossing the lens.\n\nAll movements must have different timing and different speeds.\n\nCreate a layered hierarchy:\n\nFOREGROUND:\nlarge objects extremely close to the camera.\n\nMIDGROUND:\ngirl and important objects.\n\nBACKGROUND:\nenvironment, buildings, sky, graphic elements.\n\nEach layer should move at a different speed.\n\n==================================================\n0–2 SECONDS\n===========\n\nBegin almost completely still.\n\nThe girl is standing in a quiet street.\n\nVery subtle movement:\nhair moving slightly,\na paper sheet on the ground,\nclouds slowly moving,\na bicycle wheel turning very slowly.\n\nHold this calm composition briefly.\n\nThen suddenly:\n\nthe girl's eyes look toward the camera.\n\nA paper airplane rapidly enters the frame.\n\nIt accelerates extremely quickly.\n\nThe camera instantly follows the paper airplane with a strong whip-pan.\n\nStrong motion blur.\n\nThe paper airplane passes extremely close to the lens and completely fills the frame naturally.\n\nThis is the transition.\n\nNO RECTANGULAR TRANSITION.\n\n==================================================\n2–4 SECONDS\n===========\n\nThe camera emerges from behind the paper airplane into a dense anime city.\n\nThe girl is running.\n\nThe camera tracks sideways.\n\nShe runs from right to left.\n\nBackground buildings move in strong parallax.\n\nForeground:\ntelephone pole,\ntree branches,\npaper,\nbicycle,\nschool bag.\n\nSome pass extremely close to the lens.\n\nThe girl suddenly jumps.\n\nUse a strong anticipation movement before the jump.\n\nShe crouches slightly.\n\nThen accelerates upward rapidly.\n\nAt the peak of the jump:\n\nTHE CAMERA SUDDENLY STOPS.\n\nThe girl continues moving.\n\nHer hair and skirt continue with secondary motion.\n\nThen the camera rapidly rotates approximately 90 degrees.\n\nThe entire environment transforms into a monochrome manga world.\n\nThe girl remains in full color.\n\n==================================================\n4–6 SECONDS\n===========\n\nThe manga environment becomes extremely dense.\n\nMultiple recognizable objects appear simultaneously:\n\nschool bag,\nnotebook,\npencil,\nsneakers,\nbicycle,\npaper airplane,\nschool bell,\napple,\nglass of water.\n\nDo not arrange them as static icons.\n\nEvery object has independent movement.\n\nThe bicycle wheel rotates.\n\nThe paper airplane flies.\n\nThe pencil rolls.\n\nThe notebook pages flutter.\n\nThe school bell swings.\n\nThe apple rotates slowly.\n\nThe water inside the glass moves naturally.\n\nThe camera travels diagonally between the objects.\n\nThe objects move at different depths.\n\nSuddenly the pencil accelerates toward the camera.\n\nIt passes directly in front of the lens.\n\nUse the pencil itself as the transition.\n\nNO RECTANGLES.\n\n==================================================\n6–8 SECONDS\n===========\n\nThe camera emerges from behind the pencil into a colorful anime music-video environment.\n\nThe girl is standing in the center.\n\nBehind her:\nfictional school-band performers,\nguitar,\nbass,\ndrums,\nmicrophone,\nspeakers,\nstage lights.\n\nEverything moves.\n\nThe drummer's arms move.\n\nThe guitar player moves.\n\nThe speaker cones vibrate.\n\nStage lights sweep across the scene.\n\nThe girl turns her head.\n\nHer hair follows with delayed secondary motion.\n\nThen she suddenly turns her entire body.\n\nThe camera performs a fast orbit around her.\n\nDuring the orbit, the background changes from the band stage into a summer landscape.\n\nUse a strong acceleration at the beginning of the orbit.\n\nSlow down dramatically near the end.\n\nBriefly hold the final position.\n\n==================================================\n8–10 SECONDS\n============\n\nExtreme close-up of the girl's face.\n\nVery calm.\n\nShe blinks.\n\nHer expression changes subtly.\n\nThe camera slowly moves sideways across her eye.\n\nThe iris contains a reflection of a completely different scene.\n\nThe camera suddenly accelerates toward the reflection.\n\nThe eye reflection expands naturally.\n\nEnter the reflected scene.\n\nNO SIMPLE ZOOM TRANSITION.\n\n==================================================\n10–12 SECONDS\n=============\n\nInside the reflection:\n\na table filled with recognizable everyday objects.\n\nGlass of water.\nPlate with cake.\nApple.\nNotebook.\nPencil.\nAlarm clock.\nSchool bag.\nSmall fan.\nPaper airplane.\n\nAll objects are alive with subtle independent animation.\n\nThe fan rotates.\n\nThe water moves.\n\nThe clock hands move.\n\nThe paper flutters.\n\nThe apple slowly rotates.\n\nThe cake remains physically stable.\n\nThe girl's hair moves from the airflow of the fan.\n\nShe reaches toward the glass.\n\nHer hand moves slowly.\n\nThen suddenly accelerates.\n\nShe touches the glass.\n\nThe water ripples.\n\nAt the exact same moment, the camera performs a rapid lateral movement.\n\nThe glass passes directly in front of the lens.\n\nUse the glass as the transition.\n\n==================================================\n12–14 SECONDS\n=============\n\nThe camera emerges from behind the glass.\n\nThe girl is running outdoors again.\n\nLow-angle camera.\n\nVery fast lateral tracking.\n\nTrees, houses, utility poles, clouds and bicycles move at different speeds.\n\nA bicycle suddenly crosses the foreground.\n\nThe camera follows the bicycle wheel.\n\nThe wheel rotates rapidly.\n\nThen the wheel slows dramatically.\n\nThe camera slows with it.\n\nEverything almost freezes.\n\nBrief moment of silence and visual stillness.\n\nThen:\n\nTHE WHEEL SUDDENLY ACCELERATES.\n\nThe camera is pulled into the spinning wheel.\n\n==================================================\n14–16 SECONDS\n=============\n\nThe spinning wheel transforms into a large technical illustration of the bicycle and motorcycle.\n\nBut do NOT use static technical diagrams.\n\nThe mechanical components move independently.\n\nWheel rotates.\n\nHandlebar turns.\n\nBrake lever moves.\n\nSuspension compresses.\n\nChain moves.\n\nSmall graphic lines animate around the machine.\n\nUse only meaningful readable labels:\n\n\"WHEEL\"\n\"BRAKE\"\n\"ENGINE\"\n\nThe girl walks through the technical layers.\n\nThe camera travels diagonally through the machine.\n\nShe reaches toward the camera.\n\nHer hand becomes extremely large in perspective.\n\nShe suddenly snaps her fingers.\n\nEverything freezes.\n\n==================================================\n16–18 SECONDS\n=============\n\nAbsolute stillness for a brief moment.\n\nThe girl looks directly at the camera.\n\nThen:\n\nALL ELEMENTS EXPLODE INTO MOTION.\n\nPaper airplanes.\nNotebook pages.\nComic lines.\nLeaves.\nGraphic circles.\nSmall photographs.\nPencils.\nMusical notes.\nBicycle parts.\nClouds.\nColored particles.\n\nAll move outward with different acceleration curves.\n\nThe camera rapidly orbits around the girl approximately 180 degrees.\n\nStrong motion blur.\n\nThe girl spins once.\n\nHer hair and skirt follow naturally.\n\nThe camera decelerates sharply.\n\nEverything suddenly becomes clean white.\n\nThe girl remains alone.\n\nShe looks toward the camera.\n\nSmall smile.\n\nReadable Japanese text:\n\n「行こう！」\n\nThe text appears cleanly.\n\nThe girl suddenly runs sideways out of frame.\n\nThe camera follows her for a fraction of a second.\n\nThen abrupt white flash.\n\nEND.\n\n==================================================\nCAMERA LANGUAGE\n===============\n\nCamera movement is a major part of the animation.\n\nUse:\n\nfast lateral tracking,\nslow lateral tracking,\ndiagonal tracking,\ncamera orbit,\n90-degree rotation,\n180-degree orbit,\nlow-angle tracking,\noverhead movement,\nwhip-pan,\nrapid parallax,\nforeground occlusion,\nlens-level object passes,\nsudden stops,\nsudden acceleration.\n\nDo NOT repeatedly move only forward and backward.\n\nDo NOT make the camera continuously fly through tunnels.\n\nCamera movement must have rhythm.\n\nUse:\n\nslow → fast → stop → slow → explosive acceleration → deceleration → hold → fast.\n\n==================================================\nSECONDARY MOTION\n================\n\nEverything should have secondary motion.\n\nHair follows the head with slight delay.\n\nSkirt follows body movement.\n\nPaper bends and rotates.\n\nWater ripples.\n\nClock hands move.\n\nWheels rotate.\n\nClouds drift.\n\nLeaves react to wind.\n\nBicycle components move.\n\nMusicians perform.\n\nSpeakers vibrate.\n\nGraphic elements react to nearby movement.\n\nDo not freeze background objects.\n\n==================================================\nTEXT\n====\n\nUse very little text.\n\nOnly use clearly readable real words.\n\nAllowed examples:\n\n\"START\"\n\"READY?\"\n\"GO!\"\n\"AFTER SCHOOL\"\n\"WHEEL\"\n\"BRAKE\"\n\"ENGINE\"\n\nJapanese:\n\n「放課後」\n「行こう！」\n\nNever generate meaningless text.\n\nNever generate fake Japanese.\n\nNever generate random alphabet strings.\n\nNever generate distorted pseudo-typography.\n\n==================================================\nVISUAL QUALITY\n==============\n\nHigh-quality Japanese anime.\n\nClean line art.\n\nDetailed character animation.\n\nProfessional motion-graphics composition.\n\nDense layered composition.\n\n2.5D depth.\n\nStrong parallax.\n\nCinematic lighting.\n\nControlled motion blur.\n\nSophisticated easing.\n\nStrong anticipation and follow-through.\n\nClear visual hierarchy.\n\nEven though the information density is extremely high, the main character must always remain visually understandable.\n\nThe video should feel like an ambitious professional Adobe After Effects music-video sequence rather than a collection of unrelated AI shots.\n\nNO large sliding rectangles.\n\nNO PowerPoint-like transitions.\n\nNO static object arrangements.\n\nNO meaningless text.\n\nNO random symbols.\n\nNO gibberish.\n\nNO repetitive camera movement.\n\nNO constant velocity.\n\nNO frozen characters.\n\nNO missing limbs.\n\nNO missing lower body.\n\nNO duplicated characters.\n\nNO deformed hands.\n\nPrioritize complex simultaneous motion, strong easing, camera choreography and physical object-based transitions.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 181,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092055357074121025",
    "title": "忍者分身包围林间决斗",
    "titleEn": "Ninja Clones Surround a Forest Duel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名忍者在林地对峙，随后被成群分身包围并释放发光攻击，展示本地设备生成结果。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two ninjas confront each other in a forest before a crowd of clones surrounds them and luminous attacks erupt in a local-generation test. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_xiaomu/status/2092055357074121025",
    "sourceLabel": "X 原帖 · @ai_xiaomu",
    "author": "黄小木",
    "publishedAt": "2026-08-25T01:04:00.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092055357074121025.mp4",
    "posterUrl": "/posters/x/2092055357074121025.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "dgx spark 跑 minimax h3 数据：\n\n15 秒  720p  24fps  h.264   生成耗时 26 分钟。",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 2,
      "views": 1204,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092056698584547695",
    "title": "苹果猫吉祥物的明快广告短片",
    "titleEn": "A Bright Apple-Cat Mascot Spot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "苹果装扮的猫咪和兔子在鲜黄色果园图形中跳跃，最后组成彩色日文产品式收尾。 原生视频规格为 10 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Apple-costumed cats and a rabbit bounce through bright orchard graphics before assembling into a colorful Japanese product-style finale. Native-video output: 10s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fated_oath/status/2092056698584547695",
    "sourceLabel": "X 原帖 · @fated_oath",
    "author": "Ringoame（りんごあめ）@小説Fated Oath",
    "publishedAt": "2026-08-25T01:09:20.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092056698584547695.mp4",
    "posterUrl": "/posters/x/2092056698584547695.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます☀\n今日も暑いですが頑張りましょう🫠\nもう8月後半なのに驚き…\n\nMiniMaX H3で作ってみたけど日本語壊れてるw😂\n\nCapCut AIを使用して動画制作しています\n@capcutapp_jp #capcutcpp",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 6,
      "views": 86,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092062197497397485",
    "title": "键盘、阶梯与文字构成的霓虹轨道",
    "titleEn": "A Neon Orbit of Keyboards, Steps, and Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "合成器键盘绕地球旋转，几何阶梯持续抬升，并以高速英文排版完成抽象视觉段落。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A synthesizer orbits Earth, geometric steps keep rising, and high-speed English typography completes the abstract visual sequence. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aidoga_lab/status/2092062197497397485",
    "sourceLabel": "X 原帖 · @aidoga_lab",
    "author": "AI動画ラボ | AI contents director",
    "publishedAt": "2026-08-25T01:31:11.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092062197497397485.mp4",
    "posterUrl": "/posters/x/2092062197497397485.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3　「NEON ORBIT 2」👇",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 3,
      "views": 45,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092064985027629562",
    "title": "角色用巨型文字打破规则",
    "titleEn": "A Character Battles with Giant Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "动漫角色挥拳、冲刺并撞击立体 BREAK THE RULES 字样，让动作和排版在黑底空间中相互反应。 原生视频规格为 15 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime character punches, sprints, and collides with giant BREAK THE RULES typography inside a black graphic space. Native-video output: 15s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/neru_pipipi/status/2092064985027629562",
    "sourceLabel": "X 原帖 · @neru_pipipi",
    "author": "ねるぴ",
    "publishedAt": "2026-08-25T01:42:15.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092064985027629562.mp4",
    "posterUrl": "/posters/x/2092064985027629562.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AIスタジオワンルームさんのを参考にアレンジ。CapCutのMiniMax H3で生成してキャラクターに文字で戦ってもらいました！ルールをぶっ壊せ！\n\nアレンジ後のプロンプトは↓\n（長いので2つに分けてます）\n\nPR #capcutcpp @capcutapp_jp",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 339,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092066514992242959",
    "title": "雷雨夜里赶作业的漫画家庭",
    "titleEn": "A Manga Family Races Through Homework on a Stormy Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雷电环绕的漫画母亲催促孩子写作业，黄色日文标题和夸张表情逐步填满画面。 原生视频规格为 12 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A lightning-framed manga mother urges a child through homework as yellow Japanese captions and exaggerated expressions fill the frame. Native-video output: 12s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_de_create/status/2092066514992242959",
    "sourceLabel": "X 原帖 · @ai_de_create",
    "author": "ゆきたか // AI Creative",
    "publishedAt": "2026-08-25T01:48:20.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092066514992242959.mp4",
    "posterUrl": "/posters/x/2092066514992242959.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "わが家なう\n\n楽しかった夏休みも、そろそろ終わりが近づいてきましたね…\n\nMiniMax H3",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 0,
      "views": 81,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092069451042795678",
    "title": "白底动漫女孩的姿态切换",
    "titleEn": "An Anime Girl Changes Poses on White",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一名校服动漫女孩在纯白背景中切换站姿、手势和表情，保持角色外观一致。 原生视频规格为 15 秒 · 640×992。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same uniformed anime girl changes standing poses, gestures, and expressions against a white background while keeping her appearance consistent. Native-video output: 15s · 640×992. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2092069451042795678",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-25T02:00:00.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092069451042795678.mp4",
    "posterUrl": "/posters/x/2092069451042795678.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "640×992",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "別パターン #AI眼鏡部 #AI田中さん\nMiniMaxH3ローカル",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 142,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092069676272832714",
    "title": "角色设定图变成时装宣传片",
    "titleEn": "A Character Sheet Becomes a Fashion Promo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "红色服装角色从设定图分栏走入黑白伸展台，镜头以近景、阵列和品牌式图形完成宣传片。 原生视频规格为 20 秒 · 1080×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-outfit character moves from character-sheet panels into a black-and-white runway through close-ups, lineups, and brand-like graphics. Native-video output: 20s · 1080×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/varts_works/status/2092069676272832714",
    "sourceLabel": "X 原帖 · @varts_works",
    "author": "rui｜∫varts (シュバルツ)",
    "publishedAt": "2026-08-25T02:00:54.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092069676272832714.mp4",
    "posterUrl": "/posters/x/2092069676272832714.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "1080×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "TapNowキャンバス内のMiniMax H3で制作\nキャラシート1枚で作った15秒PV👠\n\nキャラシートだけでここまで出来るなら素材作らなくて済む😲みんなもやってみてね〜✨\n\n制作: @TapNow_JP @TapNow_AI #TapNow \n衣装: LUCY @ParallaxTalent #ParallaxAgency \n\nプロンプト長いので2つに分けてます🔻",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 6,
      "views": 174,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092070808361525604",
    "title": "白色展台上的时装群像变换",
    "titleEn": "Fashion Figures Transform Across a White Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "多名时装人物在白色展台列队，经过红黑杂志排版和人物复制，形成快速剪辑的时装视觉片。 原生视频规格为 45 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Fashion figures line up on a white stage, move through red-and-black editorial graphics, and multiply into a fast-cut fashion visual. Native-video output: 45s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/applete77191758/status/2092070808361525604",
    "sourceLabel": "X 原帖 · @applete77191758",
    "author": "appletea 🍎アップルティ☕丨AI動画制作",
    "publishedAt": "2026-08-25T02:05:24.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092070808361525604.mp4",
    "posterUrl": "/posters/x/2092070808361525604.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "昨日MiniMaxDesignで、画像数枚とBGM入れて、「MiniMaxH3で生成して、45秒で」と指定するとお風呂入ってる間に仕上がってました。なんか生成して勝手に編集までやってくれてました😅\n自分でも編集しようとしたんだけど、AIの編集の方がおもしろい仕上がり😅\n\n#MiniMaxH3  #MiniMaxDesign",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 300,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092072494232449336",
    "title": "海岸上空的陨石冲击",
    "titleEn": "Meteor Impact over a Coastline",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物在海岸望向天空，巨大坠落体掠过水面并触发冲击波，展示多次本地生成尝试。 原生视频规格为 55 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure watches from the coast as a massive falling object crosses the water and triggers an impact wave across several local-generation attempts. Native-video output: 55s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AlmostYeti/status/2092072494232449336",
    "sourceLabel": "X 原帖 · @AlmostYeti",
    "author": "Yeti",
    "publishedAt": "2026-08-25T02:12:06.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092072494232449336.mp4",
    "posterUrl": "/posters/x/2092072494232449336.jpg",
    "duration": 55,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Holy F*n shit it's a Meteor! Minimax H3 with Maestro in Pinokio. I never quite got the clip i wanted but got close, here are a few attempts.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 15,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092073771205312938",
    "title": "橙色 ADP 字样的角色图形秀",
    "titleEn": "A Character Motion Show Built Around ADP Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "白衣角色用手势操控橙色 ADP 字样，镜头在全身姿态、巨型文字和面部特写间切换。 原生视频规格为 15 秒 · 2544×1456。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-clad character controls orange ADP typography with gestures as the camera cuts between full-body poses, giant letters, and facial close-ups. Native-video output: 15s · 2544×1456. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/korosuke_tv/status/2092073771205312938",
    "sourceLabel": "X 原帖 · @korosuke_tv",
    "author": "Ryosuke Komuro(ころすけ）@ADP代表/Genspark Genius",
    "publishedAt": "2026-08-25T02:17:10.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092073771205312938.mp4",
    "posterUrl": "/posters/x/2092073771205312938.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2544×1456",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3はGensparkのAI動画にもあるから\nこのプロンプトで作ってみましたよ！",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 75,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092075145687080996",
    "title": "黑白刀锋中的动漫动作片段",
    "titleEn": "Anime Action Cut Through Black-and-White Blades",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫人物在强透视姿态、室内躺姿和刀锋剪影间切换，黑白光线把片段串成动作蒙太奇。 原生视频规格为 32 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime figure moves through foreshortened poses, an interior recline, and blade-lit silhouettes in a black-and-white action montage. Native-video output: 32s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mxvdxn_anime/status/2092075145687080996",
    "sourceLabel": "X 原帖 · @mxvdxn_anime",
    "author": "𝐌𝐗𝐕𝐃𝐗𝐍 // 𝐓𝐇𝐄 𝐀𝐍𝐈𝐌𝐄 𝐃𝐄𝐏𝐓",
    "publishedAt": "2026-08-25T02:22:38.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092075145687080996.mp4",
    "posterUrl": "/posters/x/2092075145687080996.jpg",
    "duration": 32,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "at least bro went down with a view 😮‍💨 minimax h3",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 4,
      "views": 211,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092076062712950995",
    "title": "女跑者穿过彩色几何赛道",
    "titleEn": "A Female Runner Crosses a Color-Blocked Course",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色女跑者剪影跃过高台并冲进粉、青、红色几何空间，速度线随每次跨越重新组合。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A female runner silhouette clears platforms and charges through pink, cyan, and red geometric spaces as speed lines recompose around each leap. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ayu_AI_0912/status/2092076062712950995",
    "sourceLabel": "X 原帖 · @Ayu_AI_0912",
    "author": "アユネオ",
    "publishedAt": "2026-08-25T02:26:16.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092076062712950995.mp4",
    "posterUrl": "/posters/x/2092076062712950995.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "うひゃあ～！Minimax H3楽しい！\n引用元の神プロンプトを拝借して\n女性バージョンを作った！\n\n（超モーションがかっこいいのに、\nポニーテールが長すぎいいいｗ）",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 130,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092077793156935793",
    "title": "废墟城市里的银发动漫少女",
    "titleEn": "A Silver-Haired Anime Girl in an Abandoned City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发少女站在积水的废弃街道中，镜头缓慢绕行并展示被植被覆盖的楼宇和角色近景。 原生视频规格为 5 秒 · 608×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired girl stands on a flooded abandoned street as the camera moves around her and reveals overgrown buildings and character close-ups. Native-video output: 5s · 608×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yukkii114/status/2092077793156935793",
    "sourceLabel": "X 原帖 · @yukkii114",
    "author": "白猫さん👯‍♀️@WCGTA運営の人",
    "publishedAt": "2026-08-25T02:33:09.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092077793156935793.mp4",
    "posterUrl": "/posters/x/2092077793156935793.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "608×352",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "没になった生成したアニメーションを供養\nMinimaxH3×CodeX for RTX4070Ti \n#MinimaxH3 #ChatGPT #CodeX #生成AI動画",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 31,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092078428707336696",
    "title": "彩色几何节拍中的篮球剪影",
    "titleEn": "A Basketball Silhouette Plays Through Geometric Beats",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色球员剪影在鲜黄、青蓝和粉色背景间运球、冲刺和扣篮，巨型英文排版跟随节拍切换。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black player silhouette dribbles, sprints, and dunks across yellow, cyan, and pink backgrounds as giant English typography hits the beat. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによるバスケットボール動画を制作する。\n\n主人公は、全身が黒一色で描かれたシンプルな人間のシルエット。バスケットボールも模様やラインのない黒一色の球体として描く。人物の顔、髪、服装などの細部は描かず、プロのバスケットボール選手を思わせる、しなやかで洗練された体型のシルエットとして表現する。背景には明るく鮮やかな色を使用し、黒い人物とボールの輪郭が明確に見える強いコントラストを維持する。\n\n主人公は、抽象的でスタイリッシュなバスケットボールコートで、余裕と自信を感じさせる華麗なボールハンドリングを披露する。低い姿勢での高速ドリブルから始まり、左右のクロスオーバー、レッグスルー、ビハインド・ザ・バック、イン・アンド・アウト、スピンムーブを滑らかに連続させる。単に技を並べるのではなく、それぞれの動きを自然につなげ、プロのストリートボーラーによる一続きのクールなコンビネーションとして見せる。\n\nドリブルコンビネーションの後、主人公は鋭く方向を変えてゴールへ加速する。力強く踏み切り、黒一色のボールを片手で保持しながら高く跳び、最後はダイナミックなワンハンドダンクを決める。ダンクは大げさな曲芸ではなく、スピード、跳躍力、身体の美しいラインを強調した、スタイリッシュで説得力のある動きにする。スローモーションは使用せず、15秒間を通してテンポと疾走感を維持する。\n\nボールが床に接触するタイミング、手から手へ移る瞬間、踏み切り、ジャンプ、リングへの到達、ダンク、着地を、力強いヒップホップまたはエレクトロニック・ミュージックのキック、スネア、ベースに正確にシンクロさせる。ボールが床に当たるたびに円形の波紋が広がり、クロスオーバーに合わせて鋭いラインが左右へ走り、方向転換すると背景の色面が瞬時に切り替わる。ダンクの瞬間には、すべてのシェイプとタイポグラフィーがビートに合わせて大きく反応する。\n\n背景では、長方形、円、三角形、太いライン、グリッド、コートラインなどのシェイプが、画面の上下左右や奥行き方向から勢いよくスライドインし、次のビートでスライドアウトする。複数のシェイプが異なる方向から重なり合い、分割、回転、拡大、縮小を繰り返しながら、リズミカルに画面構成を作り替えていく。ただし、シェイプで人物、ボール、リングを完全に隠さず、プレーの流れが常に明確に見えるようにする。\n\n背景には「BASKETBALL」「DRIBBLE」「DUNK」という単語を、正確なスペルの大文字でキネティック・タイポグラフィーとして登場させる。文字は一度表示されて終わるのではなく、登場後も音楽のビートに合わせて脈打つように拡大・縮小を繰り返し、前後左右への3D回転、奥行き方向への移動、軽い伸縮、色の切り替えを続ける。\n\n3D回転した文字は再び正面を向き、各単語が明確に読める瞬間を必ず作る。文字の欠落、重複、誤字、別の文字への変化は起こさず、「BASKETBALL」「DRIBBLE」「DUNK」以外の文字は表示しない。タイポグラフィーは人物とボールの背後や周囲で動かし、プレーを妨げない背景のグラフィック要素として使用する。\n\nカメラは基本的に人物の全身とボールを正面から捉えながら、音楽の強いビートに合わせてアングルを大胆に切り替える。正面からの構図に加えて、人物を斜め上から見下ろす俯瞰ショットと、斜め下から見上げる力強いローアングルショットを入れる。\n\n俯瞰ショットでは、コート上を動く足、ドリブルの軌道、床面を流れるシェイプやタイポグラフィーを印象的に見せる。ローアングルショットでは、黒いボールが床へ迫って跳ね返る瞬間、主人公の鋭い方向転換、ゴールへ向かう加速、踏み切りからダンクまでの高さと迫力を強調する。さらに、真横からの高速追従、短いズームインとズームアウト、左右へのスライド、リング付近から人物を見下ろす構図、ダンクする主人公を斜め下から捉える構図を加える。\n\nカットが切り替わっても、人物の動作、進行方向、ボールの位置、ドリブルのリズムを途切れさせず、同じ人物による一続きのプレーとして自然につなげる。アングルが変わっても、同一人物、同一の黒いボール、同一のコートとゴールを維持する。\n\nボールは映像全体を通して必ず1個だけにする。ボールの大きさと完全な球形を維持し、複製、変形、消失、瞬間移動、色の変化を発生させない。手、床、リングとの接触を物理的に正しく表現し、ボールが手や床をすり抜けないようにする。黒いボールが黒い身体と重なる場面でも、腕の位置と背景の色によって、ボールの丸い輪郭を判別できるようにする。\n\n人物の形状、体格、手足の本数を15秒間一貫させ、手足の変形、身体の分裂、不自然な残像を発生させない。リング、バックボード、ネットもシンプルな形状を維持し、ダンクの前後で位置や大きさを変化させない。\n\n大胆でありながら統一感のある配色、シャープなグラフィック、強いコントラスト、正確なビートシンクによって、とにかくカッコよく、スタイリッシュで、ファッション性の高い映像に仕上げる。終盤に向けてドリブル、カメラワーク、シェイプ、タイポグラフィーの動きを徐々に高密度にし、最後の強いビートでワンハンドダンクを決める。\n\nダンク後は大げさに喜んだり決めポーズを取ったりしない。主人公は軽やかに着地し、リングを通過した黒いボールが床で一度だけ弾む。そのまま主人公が余裕のある動きでカメラの横を通り過ぎ、背景のグラフィックがビートの余韻に合わせて整列して、クールに終了する。SNSで拡散されそうな、センスが良く、中毒性があり、何度も見返したくなる高品質な映像にする。",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2092078428707336696",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2092078428707336696",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-08-25T02:35:40.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092078428707336696.mp4",
    "posterUrl": "/posters/x/2092078428707336696.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「MiniMax H3 x バスケ」\nスポーツも良い\n\nMiniMax H3のt2vで生成\n@Hailuo_AI \n\nプロンプト↓\nMiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによるバスケットボール動画を制作する。\n\n主人公は、全身が黒一色で描かれたシンプルな人間のシルエット。バスケットボールも模様やラインのない黒一色の球体として描く。人物の顔、髪、服装などの細部は描かず、プロのバスケットボール選手を思わせる、しなやかで洗練された体型のシルエットとして表現する。背景には明るく鮮やかな色を使用し、黒い人物とボールの輪郭が明確に見える強いコントラストを維持する。\n\n主人公は、抽象的でスタイリッシュなバスケットボールコートで、余裕と自信を感じさせる華麗なボールハンドリングを披露する。低い姿勢での高速ドリブルから始まり、左右のクロスオーバー、レッグスルー、ビハインド・ザ・バック、イン・アンド・アウト、スピンムーブを滑らかに連続させる。単に技を並べるのではなく、それぞれの動きを自然につなげ、プロのストリートボーラーによる一続きのクールなコンビネーションとして見せる。\n\nドリブルコンビネーションの後、主人公は鋭く方向を変えてゴールへ加速する。力強く踏み切り、黒一色のボールを片手で保持しながら高く跳び、最後はダイナミックなワンハンドダンクを決める。ダンクは大げさな曲芸ではなく、スピード、跳躍力、身体の美しいラインを強調した、スタイリッシュで説得力のある動きにする。スローモーションは使用せず、15秒間を通してテンポと疾走感を維持する。\n\nボールが床に接触するタイミング、手から手へ移る瞬間、踏み切り、ジャンプ、リングへの到達、ダンク、着地を、力強いヒップホップまたはエレクトロニック・ミュージックのキック、スネア、ベースに正確にシンクロさせる。ボールが床に当たるたびに円形の波紋が広がり、クロスオーバーに合わせて鋭いラインが左右へ走り、方向転換すると背景の色面が瞬時に切り替わる。ダンクの瞬間には、すべてのシェイプとタイポグラフィーがビートに合わせて大きく反応する。\n\n背景では、長方形、円、三角形、太いライン、グリッド、コートラインなどのシェイプが、画面の上下左右や奥行き方向から勢いよくスライドインし、次のビートでスライドアウトする。複数のシェイプが異なる方向から重なり合い、分割、回転、拡大、縮小を繰り返しながら、リズミカルに画面構成を作り替えていく。ただし、シェイプで人物、ボール、リングを完全に隠さず、プレーの流れが常に明確に見えるようにする。\n\n背景には「BASKETBALL」「DRIBBLE」「DUNK」という単語を、正確なスペルの大文字でキネティック・タイポグラフィーとして登場させる。文字は一度表示されて終わるのではなく、登場後も音楽のビートに合わせて脈打つように拡大・縮小を繰り返し、前後左右への3D回転、奥行き方向への移動、軽い伸縮、色の切り替えを続ける。\n\n3D回転した文字は再び正面を向き、各単語が明確に読める瞬間を必ず作る。文字の欠落、重複、誤字、別の文字への変化は起こさず、「BASKETBALL」「DRIBBLE」「DUNK」以外の文字は表示しない。タイポグラフィーは人物とボールの背後や周囲で動かし、プレーを妨げない背景のグラフィック要素として使用する。\n\nカメラは基本的に人物の全身とボールを正面から捉えながら、音楽の強いビートに合わせてアングルを大胆に切り替える。正面からの構図に加えて、人物を斜め上から見下ろす俯瞰ショットと、斜め下から見上げる力強いローアングルショットを入れる。\n\n俯瞰ショットでは、コート上を動く足、ドリブルの軌道、床面を流れるシェイプやタイポグラフィーを印象的に見せる。ローアングルショットでは、黒いボールが床へ迫って跳ね返る瞬間、主人公の鋭い方向転換、ゴールへ向かう加速、踏み切りからダンクまでの高さと迫力を強調する。さらに、真横からの高速追従、短いズームインとズームアウト、左右へのスライド、リング付近から人物を見下ろす構図、ダンクする主人公を斜め下から捉える構図を加える。\n\nカットが切り替わっても、人物の動作、進行方向、ボールの位置、ドリブルのリズムを途切れさせず、同じ人物による一続きのプレーとして自然につなげる。アングルが変わっても、同一人物、同一の黒いボール、同一のコートとゴールを維持する。\n\nボールは映像全体を通して必ず1個だけにする。ボールの大きさと完全な球形を維持し、複製、変形、消失、瞬間移動、色の変化を発生させない。手、床、リングとの接触を物理的に正しく表現し、ボールが手や床をすり抜けないようにする。黒いボールが黒い身体と重なる場面でも、腕の位置と背景の色によって、ボールの丸い輪郭を判別できるようにする。\n\n人物の形状、体格、手足の本数を15秒間一貫させ、手足の変形、身体の分裂、不自然な残像を発生させない。リング、バックボード、ネットもシンプルな形状を維持し、ダンクの前後で位置や大きさを変化させない。\n\n大胆でありながら統一感のある配色、シャープなグラフィック、強いコントラスト、正確なビートシンクによって、とにかくカッコよく、スタイリッシュで、ファッション性の高い映像に仕上げる。終盤に向けてドリブル、カメラワーク、シェイプ、タイポグラフィーの動きを徐々に高密度にし、最後の強いビートでワンハンドダンクを決める。\n\nダンク後は大げさに喜んだり決めポーズを取ったりしない。主人公は軽やかに着地し、リングを通過した黒いボールが床で一度だけ弾む。そのまま主人公が余裕のある動きでカメラの横を通り過ぎ、背景のグラフィックがビートの余韻に合わせて整列して、クールに終了する。SNSで拡散されそうな、センスが良く、中毒性があり、何度も見返したくなる高品質な映像にする。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 156,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092079093273796878",
    "title": "机甲在长镜头中迎击坠落弹体",
    "titleEn": "A Mecha Faces a Falling Projectile in a Long Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "白色机甲在荒地升空，镜头跟随高速坠落弹体穿越云层，并回到机甲迎击的连续动作。 原生视频规格为 51 秒 · 608×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white mecha lifts off over a wasteland as the camera follows a fast projectile through clouds and returns to the machine bracing for impact. Native-video output: 51s · 608×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/palealloy/status/2092079093273796878",
    "sourceLabel": "X 原帖 · @palealloy",
    "author": "つみき/palealloy",
    "publishedAt": "2026-08-25T02:38:19.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092079093273796878.mp4",
    "posterUrl": "/posters/x/2092079093273796878.jpg",
    "duration": 51,
    "aspectRatio": "landscape",
    "resolution": "608×352",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "50秒動画を10秒×5で一発生成。参照画像6枚。\n後半ダブってるのはプロンプトのせい🥹\n#minimaxH3 #comfyUI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 41,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092080680431915223",
    "title": "粉发角色身边的五秒微动",
    "titleEn": "Five Seconds of Subtle Motion Around a Pink-Haired Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发动漫角色坐在废墟边缘，披风和背景光线产生轻微运动，展示五秒本地生成速度测试。 原生视频规格为 5 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired anime character sits above ruins while the cape and background light move subtly in a five-second local-generation preview. Native-video output: 5s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hypoxia_021/status/2092080680431915223",
    "sourceLabel": "X 原帖 · @hypoxia_021",
    "author": "兎羽",
    "publishedAt": "2026-08-25T02:44:37.000Z",
    "addedAt": "2026-08-25T10:30:31.572Z",
    "mediaUrl": "/media/x-2092080680431915223.mp4",
    "posterUrl": "/posters/x/2092080680431915223.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3 在5080上能做到720p 5s 生成只要5分钟以内，但是时长5s就极限了，生成10s会OOM。或者生成540p 10s，不过需要超分，目前好像没啥好用的超分方案",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 6,
      "snapshotAt": "2026-08-25T10:30:31.572Z"
    },
    "approvedAt": "2026-08-25T10:30:31.572Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092401685876605427",
    "title": "日式青春短片里的卡带与海风",
    "titleEn": "Cassette Tapes and Sea Breeze in a Japanese Youth Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "两名学生从夏日走廊和阅览室走向天台与海边，以卡带、耳机和克制慢镜头串起一段青春记忆。 原生视频规格为 45 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two students move from a summer corridor and reading room to a rooftop and the sea, linked by a cassette, shared earphones, and restrained slow-motion imagery. Native-video output: 45s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "基础设定：基于人物角色图 ，以及下面的画面内容，生成一段充满日式电影美学与叙事感的短片。仅生成15秒的画面内容。无背景音乐 风格要求： 整体必须是正宗的“日式青春文艺电影”视觉风格。画面排斥任何高饱和、高对比度的商业塑料感。采用“Natural Light & Airiness”自然光与空气感布光，高光处带有温暖的过曝柔化边缘。通过“极致纯净的环境白噪音”与“极致克制的电影慢镜头”形成极具沉浸感的静谧艺术张力 \n画面内容：（每一次 只要放15秒的 画面内容描述，分三次 给）\nScene 01 (0.0-6.0s) [微热开场]: 镜头一 近景面部特写女主 A（齐颈短发，穿着白衬衫制服，眼神清冷）正走过走廊，汗水在细腻的自然光下闪烁，眼神清冷而倔强。镜头二中，采用 35mm 焦段，固定长镜头 (Static Long Shot)。正午强烈的阳光穿透无风的走廊，将白窗帘照得近乎透明，空气因高温而微微扭曲  【日语旁白 (VO-Japanese)】: あの夏は、ずっと終わらない気がしていた。 Scene 02 (6.0-11.0s) [光影的视线]: 镜头一采用 85mm 焦段（浅景深，f/1.4）。闷热的阅览室，男生 A（校服衬衫，汗水打湿了鬓角）趴在桌上，眯着眼睛注视着窗边的女生 A。镜头二采用焦点转换 (Rack Focus)，焦点瞬间转移到男生 A 专注且略带忧郁的眼睛，捕捉到阳光在他瞳孔中折射出的微光。  Scene 03 (11.0-15.0s) [逆光的物语]: 镜头一极近特写男生 A 的手，由于紧张，手指沾染了汗水，将一只复古的卡式录音带（卡带）在逆光中递向女生 A。镜头二采用微距镜头展示卡带交接的瞬间，大逆光穿透透明的卡带外壳，折射出绚烂的镜头光晕（Lens Flare），画面极致唯美。  Scene 04 (15.0-21.0s) [共享的旋律]: 镜头一采用 50mm 焦段。两人并排站在靠窗的书架旁。男生 A 拿出复古的随身听 (Walkman)，戴上一只耳机，将另一只耳机递给女生 A。镜头二中，女生 A 犹豫了一下接过来戴上。两人在强烈的漫反射自然光下互动，脸部受光面极度均匀通透。  【日语旁白 (VO-Japanese)】: 同じメロディーを分け合えば、同じ世界に飛び込めるのかな。 Scene 05 (21.0-30.0s) [天台的余晖]: 镜头一展示学校天台，夕阳将天空染成绚丽的瑰红色。两人并排坐在石阶上，耳机线在他们之间拉出一条微妙的弧线。镜头二采用升格慢镜头（Slow-motion），微风吹动女生 A 的发丝与裙摆，她随着音乐轻轻闭上眼睛，男生 A 在一旁安静地看着她，画面充满诗意的留白。  【日语旁白 (VO-Japanese)】: イヤホンの中は、言えなかった夏。 Scene 06 (30.0-35.0s) [奔向碧海]: 镜头一采用 24mm 超广角，远景。两人推着自行车来到海边的堤坝。镜头二采用手持跟拍 (Handheld Tracking) 贴地拍摄，捕捉他们冲向沙滩时的背影和飞扬的裙摆，背景是湛蓝、清透得不真实的晴空。  Scene 07 (35.0-41.0s) [浪花与秘密]: 镜头一展示两人在海边堤坝，并排坐下。女生拨弄被风吹乱的头发。镜头二采用缓慢推轨 (Dolly-in)，采用低角度逆光仰拍，阳光穿过他们举起的手臂，在脸上形成柔和的镜头光晕 (Lens Flare)，两人害羞地相视一笑。  Scene 08 (41.0-45.0s) [夏天的风]: 镜头一采用 35mm 焦段，特写喝了一半的蓝紫色塑料瓶被放在沙滩上，旁边滚落着一颗脏篮球，一阵风吹过，扬起细微的沙粒。镜头二画面渐隐至纯白，中心浮现极其纤细、带有易碎感的明朝体字体：“夏天的风”，排版文艺诗意。",
    "promptSourceUrl": "https://x.com/TanLuAI/status/2092401685876605427",
    "sourceUrl": "https://x.com/TanLuAI/status/2092401685876605427",
    "sourceLabel": "X 原帖 · @TanLuAI",
    "author": "探路AI",
    "publishedAt": "2026-08-26T00:00:11.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092401685876605427.mp4",
    "posterUrl": "/posters/x/2092401685876605427.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "retro",
      "ugc"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "使用Minimax H3 做的 45秒 日式文艺短片\n\n通过一张角色场景图 +一段画面描述提示词+风格提示词\n\n每次生成15秒的片段，最后再拼接\n\n关键道具或者人物细节，需要截图作为后续视频的参考物料\n\n剪辑的时候，对一些有问题的画面进行裁切。部分画面通过调速，确保氛围感\n\n完整提示词👇👇👇\n基础设定：基于人物角色图 ，以及下面的画面内容，生成一段充满日式电影美学与叙事感的短片。仅生成15秒的画面内容。无背景音乐 风格要求： 整体必须是正宗的“日式青春文艺电影”视觉风格。画面排斥任何高饱和、高对比度的商业塑料感。采用“Natural Light & Airiness”自然光与空气感布光，高光处带有温暖的过曝柔化边缘。通过“极致纯净的环境白噪音”与“极致克制的电影慢镜头”形成极具沉浸感的静谧艺术张力 \n画面内容：（每一次 只要放15秒的 画面内容描述，分三次 给）\nScene 01 (0.0-6.0s) [微热开场]: 镜头一 近景面部特写女主 A（齐颈短发，穿着白衬衫制服，眼神清冷）正走过走廊，汗水在细腻的自然光下闪烁，眼神清冷而倔强。镜头二中，采用 35mm 焦段，固定长镜头 (Static Long Shot)。正午强烈的阳光穿透无风的走廊，将白窗帘照得近乎透明，空气因高温而微微扭曲  【日语旁白 (VO-Japanese)】: あの夏は、ずっと終わらない気がしていた。 Scene 02 (6.0-11.0s) [光影的视线]: 镜头一采用 85mm 焦段（浅景深，f/1.4）。闷热的阅览室，男生 A（校服衬衫，汗水打湿了鬓角）趴在桌上，眯着眼睛注视着窗边的女生 A。镜头二采用焦点转换 (Rack Focus)，焦点瞬间转移到男生 A 专注且略带忧郁的眼睛，捕捉到阳光在他瞳孔中折射出的微光。  Scene 03 (11.0-15.0s) [逆光的物语]: 镜头一极近特写男生 A 的手，由于紧张，手指沾染了汗水，将一只复古的卡式录音带（卡带）在逆光中递向女生 A。镜头二采用微距镜头展示卡带交接的瞬间，大逆光穿透透明的卡带外壳，折射出绚烂的镜头光晕（Lens Flare），画面极致唯美。  Scene 04 (15.0-21.0s) [共享的旋律]: 镜头一采用 50mm 焦段。两人并排站在靠窗的书架旁。男生 A 拿出复古的随身听 (Walkman)，戴上一只耳机，将另一只耳机递给女生 A。镜头二中，女生 A 犹豫了一下接过来戴上。两人在强烈的漫反射自然光下互动，脸部受光面极度均匀通透。  【日语旁白 (VO-Japanese)】: 同じメロディーを分け合えば、同じ世界に飛び込めるのかな。 Scene 05 (21.0-30.0s) [天台的余晖]: 镜头一展示学校天台，夕阳将天空染成绚丽的瑰红色。两人并排坐在石阶上，耳机线在他们之间拉出一条微妙的弧线。镜头二采用升格慢镜头（Slow-motion），微风吹动女生 A 的发丝与裙摆，她随着音乐轻轻闭上眼睛，男生 A 在一旁安静地看着她，画面充满诗意的留白。  【日语旁白 (VO-Japanese)】: イヤホンの中は、言えなかった夏。 Scene 06 (30.0-35.0s) [奔向碧海]: 镜头一采用 24mm 超广角，远景。两人推着自行车来到海边的堤坝。镜头二采用手持跟拍 (Handheld Tracking) 贴地拍摄，捕捉他们冲向沙滩时的背影和飞扬的裙摆，背景是湛蓝、清透得不真实的晴空。  Scene 07 (35.0-41.0s) [浪花与秘密]: 镜头一展示两人在海边堤坝，并排坐下。女生拨弄被风吹乱的头发。镜头二采用缓慢推轨 (Dolly-in)，采用低角度逆光仰拍，阳光穿过他们举起的手臂，在脸上形成柔和的镜头光晕 (Lens Flare)，两人害羞地相视一笑。  Scene 08 (41.0-45.0s) [夏天的风]: 镜头一采用 35mm 焦段，特写喝了一半的蓝紫色塑料瓶被放在沙滩上，旁边滚落着一颗脏篮球，一阵风吹过，扬起细微的沙粒。镜头二画面渐隐至纯白，中心浮现极其纤细、带有易碎感的明朝体字体：“夏天的风”，排版文艺诗意。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 56,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092399408763859123",
    "title": "霓虹城市上空的高速空战",
    "titleEn": "High-Speed Aerial Combat Above a Neon City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "飞行角色在霓虹城市与云层之间俯冲、追逐并交火，连续穿越高楼和爆炸光线。 原生视频规格为 23 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Flying characters dive, pursue, and exchange fire across a neon city and cloud layer, threading between towers and explosive light. Native-video output: 23s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kicknost/status/2092399408763859123",
    "sourceLabel": "X 原帖 · @kicknost",
    "author": "Kick Nost",
    "publishedAt": "2026-08-25T23:51:08.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092399408763859123.mp4",
    "posterUrl": "/posters/x/2092399408763859123.jpg",
    "duration": 23,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I created an aerial combat style using the Minimax H3 method in @Hailuo_AI . It was created from a high-quality 2k character sheet image.\n\n#minimaxh3 #minimaxdesign",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 39,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092397328502341872",
    "title": "潮酷狗狗的排版变身秀",
    "titleEn": "A Stylish Dog Takes Over a Typography Reel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一只狗在黑白时装构图、巨型字母和快速切换的图形层中完成俏皮的潮流展示。 原生视频规格为 15 秒 · 1080×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dog takes a playful fashion turn through black-and-white compositions, oversized lettering, and rapidly changing graphic layers. Native-video output: 15s · 1080×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/misoosooooup/status/2092397328502341872",
    "sourceLabel": "X 原帖 · @misoosooooup",
    "author": "miso🍙",
    "publishedAt": "2026-08-25T23:42:52.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092397328502341872.mp4",
    "posterUrl": "/posters/x/2092397328502341872.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1080×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "かっこよいイッヌになった♡\nありがとうございます😊💕\n\nMinimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 23,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092393827407696030",
    "title": "插画角色走进投影歌词舞台",
    "titleEn": "An Illustrated Performer Enters a Projected Lyric Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "插画角色在暗室投影前演唱，歌词、光影和镜面反射围绕人物形成连续的动态图形表演。 原生视频规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustrated performer sings before a dark-room projection while lyrics, light, and mirrored reflections form a continuous motion-graphics performance. Native-video output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/h64g_/status/2092393827407696030",
    "sourceLabel": "X 原帖 · @h64g_",
    "author": "h64g",
    "publishedAt": "2026-08-25T23:28:57.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092393827407696030.mp4",
    "posterUrl": "/posters/x/2092393827407696030.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "イラスト系の画像を入力した場合\n#MiniMaxH3 #MiniMaxDesign @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 56,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092387240664326495",
    "title": "魔法学徒与白兔的角色宣传片",
    "titleEn": "A Magical Apprentice and Her Rabbits in a Character PV",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "粉发魔法学徒在学院庭院与白兔互动，角色特写、书本和可爱排版组成明亮的动画宣传片。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A pink-haired magical apprentice interacts with white rabbits in an academy courtyard as character close-ups, a spellbook, and cute typography shape a bright anime PV. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 14.8-second bright, colorful premium anime game character PV with cute motion typography. 16:9, high-quality 2D anime look, dreamy, magical, elegant, warm and adorable. No external player UI, no watermark.  REFERENCE RULES @Image1 = FACE REFERENCE ONLY. Use @Image1 as the sole authority for face, eyes, facial proportions, bangs, hairstyle around the face, expression style and identity. @Image2 = OUTFIT / BODY DESIGN REFERENCE ONLY. Use @Image2 as the sole authority for clothing, cape, blouse, ribbon, brooch, belt, potion bottles, crystals, boots, accessories, silhouette, spellbook, colors and overall design. Keep exactly the same girl in every shot. Never redesign or simplify her.  CHARACTER A magical academy apprentice with long pink twin-tails, purple flower hair ornaments, luminous blue-violet eyes, purple hooded cape with gold trim, white frilled blouse, dark ribbon, crystal jewelry, potion bottles and crystals on her belt, and an ornate grimoire.  ENVIRONMENT A lush magical academy courtyard with old stone archways, green gardens, flowers and warm daylight. Several small white rabbits interact naturally with her: gathering around her feet, hopping, nuzzling her hands, chasing petals and reacting to magic. Magic is gentle and cute: floating petals, tiny stars, glowing particles, translucent purple-pink magic circles and soft book-based spell effects. No combat, no dark atmosphere.  MOTION TYPOGRAPHY Integrate clearly visible but elegant cute motion typography throughout the PV. Style: rounded or refined fantasy type, pastel pink, lavender, cream, white and subtle gold; small editorial captions, curved text, ribbon-like labels, flower/star motifs and playful floating words. Motion: soft pop-in, bloom, gentle slide, tiny bounce, arc rotation around magic circles, petal-like scatter and sparkle dissolve. Typography must remain secondary to the girl. Never cover her face. Avoid giant text, cyber HUD, glitch, industrial graphics, aggressive trailer typography or heavy sci-fi UI.  Use short phrases such as: “LILIA ASTRA” “STARFLOWER APPRENTICE” “LITTLE MAGIC” “RABBIT FRIENDS” “BLOOMING SPELL” “TWINKLE BLOOM” “GENTLE WONDER” “MAGIC IN BLOOM”  TIMELINE  0:00.0-0:00.5 Face close-up. Gentle smile and soft eye contact. Twin-tails move slightly in the breeze. “LILIA ASTRA” blooms into one corner with tiny flowers and stars.  0:00.5-0:01.0 Full-body courtyard shot. White rabbits gather around her boots. “RABBIT FRIENDS” appears beside them with a soft bounce.  0:01.0-0:01.5 Medium close-up. She opens the grimoire slightly and purple-pink particles spill from the pages. Curved “LITTLE MAGIC” follows the glow near the book.  0:01.5-0:02.0 Low rabbit shot. One rabbit hops after drifting petals. A tiny “hop!” briefly follows the movement.  0:02.0-0:02.5 She crouches and gently pets one rabbit. Warm framing. “GENTLE WONDER” slides softly into empty space.  0:02.5-0:03.0 Another rabbit comes close. She looks slightly surprised, then smiles. A tiny sparkle or “!” pops beside her.  0:03.0-0:03.5 Close shot of the grimoire. A translucent magic circle forms above it. “BLOOMING SPELL” curves around the circle and rotates gently.  0:03.5-0:04.0 Wide shot. Rabbits move around her in a loose circle. Her cape and twin-tails follow naturally. Small curved decorative text briefly traces their motion.  0:04.0-0:04.5 She closes the grimoire and hugs it to her chest. Petals and sparkles drift behind her. “STARFLOWER APPRENTICE” appears as an elegant character-title card.  0:04.5-0:05.0 Face close-up. She gives a happy smile. Tiny star and flower graphics bloom around the frame edges.  0:05.0-0:05.5 Rabbit close-up. One rabbit lifts its ears and makes a small hop. A tiny “twinkle!” dissolves into sparkles.  0:05.5-0:06.0 She stands gracefully and watches the rabbits. Keep the frame clean with only small decorative stars.  0:06.0-0:06.5 Medium-wide shot. Rabbits scatter left, right and toward foreground, creating depth. Add subtle lateral camera movement and petals crossing the lens.  0:06.5-0:07.0 She places one hand near her chest and turns into a cute elegant pose. “TWINKLE BLOOM” rises beside her with flower-shaped sparkles.  0:07.0-0:08.0 Pacing becomes calmer. She blinks slowly while one rabbit sits beside her boots. Hair, cape and petals continue gentle secondary motion. Minimal typography.  0:08.0-0:09.0 Petals drift slowly and the magic circle fades. Small “MAGIC IN BLOOM” appears faintly and dissolves into glowing dust.  0:09.0-0:10.0 Medium shot. She holds the grimoire to her chest while rabbits gather from both sides. Warm sunlight and shallow depth of field.  0:10.0-0:11.0 One rabbit jumps toward the foreground and stops near camera. The girl remains visible behind it. A tiny star/flower graphic reacts to the landing.  0:11.0-0:12.0 She looks toward the rabbit and gives a small affectionate smile. Keep the courtyard calm. No large typography.  0:12.0-0:13.0 Lingering atmospheric shot. Only hair, cape, petals, grass, rabbit ears and light particles move gently.  0:13.0-0:14.8 Return naturally toward the opening face close-up. She keeps the same gentle expression and slowly blinks. Final elegant title: “LILIA ASTRA” “STARFLOWER APPRENTICE” The title blooms softly with purple flowers, gold stars and sparkles, then settles into a calm loop-like ending.  EDITING First 7 seconds: lively 0.5-second character-PV cuts with clear shot variation: face close-up, full-body, medium, rabbit close-up, grimoire detail, wide and low-angle shots. After 7 seconds: gradually slow the rhythm for a gentle emotional finish. Use subtle push-ins, lateral movement, foreground rabbits and petals so the PV never feels static.  Maintain consistent face, costume, pink twin-tails, purple/gold academy clothing, ornate grimoire and readable rabbit anatomy. The final result should feel like a premium cute magical-academy character reveal PV combined with polished editorial motion typography, with the girl and rabbits always remaining the main focus.",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2092387240664326495",
    "sourceUrl": "https://x.com/haruuraeadss/status/2092387240664326495",
    "sourceLabel": "X 原帖 · @haruuraeadss",
    "author": "mayv@簡単プロ級プロンプト公開中！",
    "publishedAt": "2026-08-25T23:02:47.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092387240664326495.mp4",
    "posterUrl": "/posters/x/2092387240664326495.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "かわいいウサギさん\n\n#MiniMaxH3  #MiniMaxDesign @Hailuo_AI　\n\nプロンプト👇\n                                                                                                                                                                                                                   \nCreate a 14.8-second bright, colorful premium anime game character PV with cute motion typography. 16:9, high-quality 2D anime look, dreamy, magical, elegant, warm and adorable. No external player UI, no watermark.  REFERENCE RULES @Image1 = FACE REFERENCE ONLY. Use @Image1 as the sole authority for face, eyes, facial proportions, bangs, hairstyle around the face, expression style and identity. @Image2 = OUTFIT / BODY DESIGN REFERENCE ONLY. Use @Image2 as the sole authority for clothing, cape, blouse, ribbon, brooch, belt, potion bottles, crystals, boots, accessories, silhouette, spellbook, colors and overall design. Keep exactly the same girl in every shot. Never redesign or simplify her.  CHARACTER A magical academy apprentice with long pink twin-tails, purple flower hair ornaments, luminous blue-violet eyes, purple hooded cape with gold trim, white frilled blouse, dark ribbon, crystal jewelry, potion bottles and crystals on her belt, and an ornate grimoire.  ENVIRONMENT A lush magical academy courtyard with old stone archways, green gardens, flowers and warm daylight. Several small white rabbits interact naturally with her: gathering around her feet, hopping, nuzzling her hands, chasing petals and reacting to magic. Magic is gentle and cute: floating petals, tiny stars, glowing particles, translucent purple-pink magic circles and soft book-based spell effects. No combat, no dark atmosphere.  MOTION TYPOGRAPHY Integrate clearly visible but elegant cute motion typography throughout the PV. Style: rounded or refined fantasy type, pastel pink, lavender, cream, white and subtle gold; small editorial captions, curved text, ribbon-like labels, flower/star motifs and playful floating words. Motion: soft pop-in, bloom, gentle slide, tiny bounce, arc rotation around magic circles, petal-like scatter and sparkle dissolve. Typography must remain secondary to the girl. Never cover her face. Avoid giant text, cyber HUD, glitch, industrial graphics, aggressive trailer typography or heavy sci-fi UI.  Use short phrases such as: “LILIA ASTRA” “STARFLOWER APPRENTICE” “LITTLE MAGIC” “RABBIT FRIENDS” “BLOOMING SPELL” “TWINKLE BLOOM” “GENTLE WONDER” “MAGIC IN BLOOM”  TIMELINE  0:00.0-0:00.5 Face close-up. Gentle smile and soft eye contact. Twin-tails move slightly in the breeze. “LILIA ASTRA” blooms into one corner with tiny flowers and stars.  0:00.5-0:01.0 Full-body courtyard shot. White rabbits gather around her boots. “RABBIT FRIENDS” appears beside them with a soft bounce.  0:01.0-0:01.5 Medium close-up. She opens the grimoire slightly and purple-pink particles spill from the pages. Curved “LITTLE MAGIC” follows the glow near the book.  0:01.5-0:02.0 Low rabbit shot. One rabbit hops after drifting petals. A tiny “hop!” briefly follows the movement.  0:02.0-0:02.5 She crouches and gently pets one rabbit. Warm framing. “GENTLE WONDER” slides softly into empty space.  0:02.5-0:03.0 Another rabbit comes close. She looks slightly surprised, then smiles. A tiny sparkle or “!” pops beside her.  0:03.0-0:03.5 Close shot of the grimoire. A translucent magic circle forms above it. “BLOOMING SPELL” curves around the circle and rotates gently.  0:03.5-0:04.0 Wide shot. Rabbits move around her in a loose circle. Her cape and twin-tails follow naturally. Small curved decorative text briefly traces their motion.  0:04.0-0:04.5 She closes the grimoire and hugs it to her chest. Petals and sparkles drift behind her. “STARFLOWER APPRENTICE” appears as an elegant character-title card.  0:04.5-0:05.0 Face close-up. She gives a happy smile. Tiny star and flower graphics bloom around the frame edges.  0:05.0-0:05.5 Rabbit close-up. One rabbit lifts its ears and makes a small hop. A tiny “twinkle!” dissolves into sparkles.  0:05.5-0:06.0 She stands gracefully and watches the rabbits. Keep the frame clean with only small decorative stars.  0:06.0-0:06.5 Medium-wide shot. Rabbits scatter left, right and toward foreground, creating depth. Add subtle lateral camera movement and petals crossing the lens.  0:06.5-0:07.0 She places one hand near her chest and turns into a cute elegant pose. “TWINKLE BLOOM” rises beside her with flower-shaped sparkles.  0:07.0-0:08.0 Pacing becomes calmer. She blinks slowly while one rabbit sits beside her boots. Hair, cape and petals continue gentle secondary motion. Minimal typography.  0:08.0-0:09.0 Petals drift slowly and the magic circle fades. Small “MAGIC IN BLOOM” appears faintly and dissolves into glowing dust.  0:09.0-0:10.0 Medium shot. She holds the grimoire to her chest while rabbits gather from both sides. Warm sunlight and shallow depth of field.  0:10.0-0:11.0 One rabbit jumps toward the foreground and stops near camera. The girl remains visible behind it. A tiny star/flower graphic reacts to the landing.  0:11.0-0:12.0 She looks toward the rabbit and gives a small affectionate smile. Keep the courtyard calm. No large typography.  0:12.0-0:13.0 Lingering atmospheric shot. Only hair, cape, petals, grass, rabbit ears and light particles move gently.  0:13.0-0:14.8 Return naturally toward the opening face close-up. She keeps the same gentle expression and slowly blinks. Final elegant title: “LILIA ASTRA” “STARFLOWER APPRENTICE” The title blooms softly with purple flowers, gold stars and sparkles, then settles into a calm loop-like ending.  EDITING First 7 seconds: lively 0.5-second character-PV cuts with clear shot variation: face close-up, full-body, medium, rabbit close-up, grimoire detail, wide and low-angle shots. After 7 seconds: gradually slow the rhythm for a gentle emotional finish. Use subtle push-ins, lateral movement, foreground rabbits and petals so the PV never feels static.  Maintain consistent face, costume, pink twin-tails, purple/gold academy clothing, ornate grimoire and readable rabbit anatomy. The final result should feel like a premium cute magical-academy character reveal PV combined with polished editorial motion typography, with the girl and rabbits always remaining the main focus.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 47,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092386547341705687",
    "title": "四眼猫耳角色的竖屏舞蹈",
    "titleEn": "A Four-Eyed Cat-Eared Character Dances Vertically",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫发四眼猫耳角色在竖屏舞台上连续起舞，快速肢体动作中仍保持面部和服装特征。 原生视频规格为 15 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple-haired, four-eyed cat-eared character dances continuously in a vertical frame while retaining facial and costume details through fast movement. Native-video output: 15s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Chupico_nora/status/2092386547341705687",
    "sourceLabel": "X 原帖 · @Chupico_nora",
    "author": "Chupico",
    "publishedAt": "2026-08-25T23:00:02.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092386547341705687.mp4",
    "posterUrl": "/posters/x/2092386547341705687.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはよーございます💜ノラダンスです✨CapCutのMinimaxH3で作成してみました。四つ目崩れないで踊れますね❗️\n@capcutapp_jp\n#capcutcpp\n#MiniMaxH3",
    "engagement": {
      "replies": 3,
      "reposts": 4,
      "likes": 23,
      "views": 127,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092385200420647307",
    "title": "漫画角色穿过高密度动态图形",
    "titleEn": "Manga Characters Race Through Dense Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "漫画人物、摩托车、剪影与巨型英文排版高速叠加，组成一段充满千禧年代气质的视觉开场。 原生视频规格为 17 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Manga figures, motorcycles, silhouettes, and oversized English typography stack at high speed in a visual opener with a mid-2000s feel. Native-video output: 17s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a fast-paced 19-second motion-graphics video inspired very closely by the visual structure, editing rhythm, composition.\n\nOverall style:\nHigh-energy Japanese music-video / motion-graphics aesthetic from the mid-2000s.\nA mixture of graphic design, manga-style comic illustrations, anime-style character art, silhouettes, photographic textures, particles, and bold typography.\nThe video must feel like a professionally edited promotional opening rather than a normal narrative scene.\n\nUse the entire screen continuously.\nAvoid static shots.\nEvery few moments, the composition must transform, rotate, slide, zoom, blur, or transition into the next composition.\n\nIMPORTANT CHARACTER STYLE:\nAll male and female characters must appear as clearly illustrated comic / manga characters.\nDo not use photorealistic humans.\nUse expressive manga/anime-style faces, strong line art, dynamic poses, slightly exaggerated proportions, and graphic shading.\nCharacters may appear as full body silhouettes, medium shots, or extreme close-ups.\nShow both male and female characters throughout the video.\n\nIMPORTANT TEXT RULE:\nNever generate random letters, fake words, unreadable pseudo-text, distorted paragraphs, or meaningless symbols.\nDo NOT fill the screen with fake text.\nOnly use a very small number of clean, readable words:\n\"ARE YOU READY?\"\n\"LIVE\"\n\"MUSIC\"\n\"COMING SOON\"\nAll typography must be large, clean, correctly spelled, and intentionally designed.\nIf accurate text cannot be rendered, omit the text instead of generating gibberish.\n\n0.0–2.0 seconds:\nBegin with a dark navy and black graphic background.\nThin technical lines, transparent rectangles, subtle grids, and small comic-image fragments appear rapidly.\nA manga-style female character appears briefly inside a rectangular graphic frame.\nThe camera pushes forward rapidly through several overlapping graphic layers.\nUse diagonal composition and strong depth.\nThe scene should immediately feel like a music-video opening.\n\n2.0–4.0 seconds:\nHard transition into a powerful dark-blue graphic composition.\nA large white manga-style male silhouette moves dynamically across the screen.\nBehind him, enormous abstract circular graphic elements rotate.\nLarge clean typography reading \"MUSIC\" appears briefly, integrated into the composition.\nThe typography moves diagonally while the character and circular elements move independently.\nCamera sweeps rapidly from left to right and slightly rotates.\n\n4.0–6.0 seconds:\nTransition through a fast radial zoom.\nA dark red and black ornamental background appears with circular patterns and glowing highlights.\nA comic-style female face briefly emerges from the darkness.\nHer eyes and hair move subtly.\nThe camera rotates around the graphic center while the background pattern continuously rotates in the opposite direction.\nA clean \"COMING SOON\" title appears briefly and then is swept away by motion blur.\n\n6.0–8.0 seconds:\nExtreme close-up of a colorful manga-style female character.\nBright orange, red, purple, and green colors.\nStrong expressive eye.\nHair moves dynamically as if hit by wind.\nThe camera rapidly moves sideways across her face, then zooms extremely close into the eye.\nThe eye becomes a transition point into the next scene.\nUse motion blur and fast directional movement.\n\n8.0–10.0 seconds:\nBreak the image into multiple manga comic panels.\nShow several different comic-style male and female characters in separate panels:\na male character standing,\na female character looking toward the camera,\na group of young comic characters,\nand a dynamic musician-like character.\nThe panels slide, overlap, rotate slightly, and rapidly change size.\nDo not make the panels static.\nCamera travels diagonally across the entire composition.\nUse white backgrounds, black manga line art, and occasional red accent graphics.\n\n10.0–12.0 seconds:\nSuddenly switch to a high-contrast white manga illustration.\nA comic-style male character and female character appear together in a dramatic pose.\nUse strong black ink outlines and halftone manga shading.\nThe camera rapidly zooms toward the characters.\nThen rotate the entire comic panel approximately 20–30 degrees while additional graphic fragments fly past the camera.\nCreate the feeling of flipping through a manga at extremely high speed.\n\n12.0–14.0 seconds:\nCut into a dark blue and black particle field.\nThousands of small bright particles move toward and past the camera.\nThe camera travels forward through the particles.\nOccasionally reveal faint silhouettes of manga characters inside the particle cloud.\nThe particles accelerate strongly toward the end of the shot.\nUse depth, parallax, and cinematic motion blur.\n\n14.0–16.0 seconds:\nTransition into a clean blue graphic composition.\nShow a white silhouette of a motorcycle with a comic-style rider.\nThe motorcycle moves from left to right.\nThe rider's body and hair/clothing move naturally.\nLarge clean text \"ARE YOU READY?\" appears briefly near the motorcycle.\nThe camera tracks sideways with the motorcycle, then overtakes it with a fast lateral camera move.\nUse flat graphic design mixed with subtle depth.\n\n16.0–17.5 seconds:\nRapid montage of comic-style musicians and characters.\nShow a male vocalist, a female character, a guitarist, and several comic-style people in dynamic poses.\nThey should move independently: head movement, hair movement, arms, body movement, instruments moving.\nUse a white background with black manga line art.\nQuick zooms and diagonal wipes connect each character.\nDo not make the characters look like frozen illustrations.\n\n17.5–19.0 seconds:\nEverything rapidly collapses into a dark background.\nGraphic fragments and particles fly toward the center.\nA clean white logo-like title appears:\n\"LIVE\"\nThen the title rapidly becomes smaller while the camera pulls backward.\nFinish on a clean black screen with a subtle glow and a professional music-promo ending.\nNo random text.\nNo meaningless symbols.\n\nCAMERA AND EDITING:\nExtremely active camera throughout the entire 19 seconds.\nUse rapid push-ins, pull-backs, left-to-right tracking, right-to-left tracking, diagonal movement, orbital rotation, Dutch angles, whip-pan transitions, rotational transitions, and extreme zooms.\nDo not rely mainly on forward/backward camera movement.\nFrequently change camera direction.\nCharacters and graphic objects must also move independently from the camera.\nUse strong parallax between foreground, middle ground, and background.\nUse motion blur during transitions.\nUse occasional sharp freeze-like moments lasting only a fraction of a second before immediately transitioning again.\n\nVISUAL LANGUAGE:\nMid-2000s Japanese music promotion / experimental motion graphics.\nManga panels mixed with modern graphic design.\nBold black-and-white comic artwork contrasted with dark navy, deep red, white, orange, purple, and electric blue.\nLayered compositions.\nDiagonal typography.\nCircular graphic motifs.\nHalftone manga textures.\nFilm grain.\nLight streaks.\nParticle effects.\nFast zooms.\nGraphic wipes.\nHigh contrast.\n\nMost important:\nThe video must feel like one continuous, highly energetic graphic sequence.\nDo not create a normal story.\nDo not create long running scenes.\nDo not leave characters standing still.\nDo not use meaningless text.\nDo not use photorealistic people.\nDo not use generic corporate motion graphics.\nKeep the visual density high from beginning to end.\nMatch the reference video's rapid montage feeling, graphic layering, manga imagery, strong silhouettes, extreme close-ups, particles, motorcycle imagery, and aggressive camera movement while creating original artwork.",
    "promptSourceUrl": "https://x.com/goldwing_zoome/status/2092385200420647307",
    "sourceUrl": "https://x.com/goldwing_zoome/status/2092385200420647307",
    "sourceLabel": "X 原帖 · @goldwing_zoome",
    "author": "goldwing",
    "publishedAt": "2026-08-25T22:54:41.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092385200420647307.mp4",
    "posterUrl": "/posters/x/2092385200420647307.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "music-performance",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはよう動画。\n（元ネタとかありません）\n#MiniMaxH3\n\nプロンプトは以下です。\nCreate a fast-paced 19-second motion-graphics video inspired very closely by the visual structure, editing rhythm, composition.\n\nOverall style:\nHigh-energy Japanese music-video / motion-graphics aesthetic from the mid-2000s.\nA mixture of graphic design, manga-style comic illustrations, anime-style character art, silhouettes, photographic textures, particles, and bold typography.\nThe video must feel like a professionally edited promotional opening rather than a normal narrative scene.\n\nUse the entire screen continuously.\nAvoid static shots.\nEvery few moments, the composition must transform, rotate, slide, zoom, blur, or transition into the next composition.\n\nIMPORTANT CHARACTER STYLE:\nAll male and female characters must appear as clearly illustrated comic / manga characters.\nDo not use photorealistic humans.\nUse expressive manga/anime-style faces, strong line art, dynamic poses, slightly exaggerated proportions, and graphic shading.\nCharacters may appear as full body silhouettes, medium shots, or extreme close-ups.\nShow both male and female characters throughout the video.\n\nIMPORTANT TEXT RULE:\nNever generate random letters, fake words, unreadable pseudo-text, distorted paragraphs, or meaningless symbols.\nDo NOT fill the screen with fake text.\nOnly use a very small number of clean, readable words:\n\"ARE YOU READY?\"\n\"LIVE\"\n\"MUSIC\"\n\"COMING SOON\"\nAll typography must be large, clean, correctly spelled, and intentionally designed.\nIf accurate text cannot be rendered, omit the text instead of generating gibberish.\n\n0.0–2.0 seconds:\nBegin with a dark navy and black graphic background.\nThin technical lines, transparent rectangles, subtle grids, and small comic-image fragments appear rapidly.\nA manga-style female character appears briefly inside a rectangular graphic frame.\nThe camera pushes forward rapidly through several overlapping graphic layers.\nUse diagonal composition and strong depth.\nThe scene should immediately feel like a music-video opening.\n\n2.0–4.0 seconds:\nHard transition into a powerful dark-blue graphic composition.\nA large white manga-style male silhouette moves dynamically across the screen.\nBehind him, enormous abstract circular graphic elements rotate.\nLarge clean typography reading \"MUSIC\" appears briefly, integrated into the composition.\nThe typography moves diagonally while the character and circular elements move independently.\nCamera sweeps rapidly from left to right and slightly rotates.\n\n4.0–6.0 seconds:\nTransition through a fast radial zoom.\nA dark red and black ornamental background appears with circular patterns and glowing highlights.\nA comic-style female face briefly emerges from the darkness.\nHer eyes and hair move subtly.\nThe camera rotates around the graphic center while the background pattern continuously rotates in the opposite direction.\nA clean \"COMING SOON\" title appears briefly and then is swept away by motion blur.\n\n6.0–8.0 seconds:\nExtreme close-up of a colorful manga-style female character.\nBright orange, red, purple, and green colors.\nStrong expressive eye.\nHair moves dynamically as if hit by wind.\nThe camera rapidly moves sideways across her face, then zooms extremely close into the eye.\nThe eye becomes a transition point into the next scene.\nUse motion blur and fast directional movement.\n\n8.0–10.0 seconds:\nBreak the image into multiple manga comic panels.\nShow several different comic-style male and female characters in separate panels:\na male character standing,\na female character looking toward the camera,\na group of young comic characters,\nand a dynamic musician-like character.\nThe panels slide, overlap, rotate slightly, and rapidly change size.\nDo not make the panels static.\nCamera travels diagonally across the entire composition.\nUse white backgrounds, black manga line art, and occasional red accent graphics.\n\n10.0–12.0 seconds:\nSuddenly switch to a high-contrast white manga illustration.\nA comic-style male character and female character appear together in a dramatic pose.\nUse strong black ink outlines and halftone manga shading.\nThe camera rapidly zooms toward the characters.\nThen rotate the entire comic panel approximately 20–30 degrees while additional graphic fragments fly past the camera.\nCreate the feeling of flipping through a manga at extremely high speed.\n\n12.0–14.0 seconds:\nCut into a dark blue and black particle field.\nThousands of small bright particles move toward and past the camera.\nThe camera travels forward through the particles.\nOccasionally reveal faint silhouettes of manga characters inside the particle cloud.\nThe particles accelerate strongly toward the end of the shot.\nUse depth, parallax, and cinematic motion blur.\n\n14.0–16.0 seconds:\nTransition into a clean blue graphic composition.\nShow a white silhouette of a motorcycle with a comic-style rider.\nThe motorcycle moves from left to right.\nThe rider's body and hair/clothing move naturally.\nLarge clean text \"ARE YOU READY?\" appears briefly near the motorcycle.\nThe camera tracks sideways with the motorcycle, then overtakes it with a fast lateral camera move.\nUse flat graphic design mixed with subtle depth.\n\n16.0–17.5 seconds:\nRapid montage of comic-style musicians and characters.\nShow a male vocalist, a female character, a guitarist, and several comic-style people in dynamic poses.\nThey should move independently: head movement, hair movement, arms, body movement, instruments moving.\nUse a white background with black manga line art.\nQuick zooms and diagonal wipes connect each character.\nDo not make the characters look like frozen illustrations.\n\n17.5–19.0 seconds:\nEverything rapidly collapses into a dark background.\nGraphic fragments and particles fly toward the center.\nA clean white logo-like title appears:\n\"LIVE\"\nThen the title rapidly becomes smaller while the camera pulls backward.\nFinish on a clean black screen with a subtle glow and a professional music-promo ending.\nNo random text.\nNo meaningless symbols.\n\nCAMERA AND EDITING:\nExtremely active camera throughout the entire 19 seconds.\nUse rapid push-ins, pull-backs, left-to-right tracking, right-to-left tracking, diagonal movement, orbital rotation, Dutch angles, whip-pan transitions, rotational transitions, and extreme zooms.\nDo not rely mainly on forward/backward camera movement.\nFrequently change camera direction.\nCharacters and graphic objects must also move independently from the camera.\nUse strong parallax between foreground, middle ground, and background.\nUse motion blur during transitions.\nUse occasional sharp freeze-like moments lasting only a fraction of a second before immediately transitioning again.\n\nVISUAL LANGUAGE:\nMid-2000s Japanese music promotion / experimental motion graphics.\nManga panels mixed with modern graphic design.\nBold black-and-white comic artwork contrasted with dark navy, deep red, white, orange, purple, and electric blue.\nLayered compositions.\nDiagonal typography.\nCircular graphic motifs.\nHalftone manga textures.\nFilm grain.\nLight streaks.\nParticle effects.\nFast zooms.\nGraphic wipes.\nHigh contrast.\n\nMost important:\nThe video must feel like one continuous, highly energetic graphic sequence.\nDo not create a normal story.\nDo not create long running scenes.\nDo not leave characters standing still.\nDo not use meaningless text.\nDo not use photorealistic people.\nDo not use generic corporate motion graphics.\nKeep the visual density high from beginning to end.\nMatch the reference video's rapid montage feeling, graphic layering, manga imagery, strong silhouettes, extreme close-ups, particles, motorcycle imagery, and aggressive camera movement while creating original artwork.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 104,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092385090600931380",
    "title": "权力人物跳起热烈弗拉门戈",
    "titleEn": "A Political Strongman Breaks into Flamenco",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "一名熟悉的政治人物穿着舞台服装跳起弗拉门戈，夸张的手势、裙摆和聚光灯制造反差喜剧。 原生视频规格为 15 秒 · 1456×2544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A recognizable political figure performs flamenco in stage costume, using exaggerated gestures, swirling fabric, and spotlights for comic contrast. Native-video output: 15s · 1456×2544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/minaxlab/status/2092385090600931380",
    "sourceLabel": "X 原帖 · @minaxlab",
    "author": "Mina Labs",
    "publishedAt": "2026-08-25T22:54:14.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092385090600931380.mp4",
    "posterUrl": "/posters/x/2092385090600931380.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1456×2544",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Here some flamenco dance I created with the most notorious dictator 🤣 Kim Jong Un.\nThat was made with Minimax H3 and I hope after this.. Minimax will stay alive.\n#ai #minimax \n\nhttps://www.minaxlab.com",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 39,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092383888425316546",
    "title": "晚霞海岸的安静独处",
    "titleEn": "A Quiet Moment Alone by the Evening Sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物独自站在暮色海岸，海浪、发丝和衣摆的轻微运动营造出平静的傍晚氛围。 原生视频规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A lone figure stands on a twilight shore as subtle waves, hair, and clothing movement create a calm evening mood. Native-video output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ROSHENDILAN/status/2092383888425316546",
    "sourceLabel": "X 原帖 · @ROSHENDILAN",
    "author": "ROSHEN DILAN",
    "publishedAt": "2026-08-25T22:49:28.000Z",
    "addedAt": "2026-08-26T00:15:10.320Z",
    "mediaUrl": "/media/x-2092383888425316546.mp4",
    "posterUrl": "/posters/x/2092383888425316546.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Some evenings…\nthe sea is enough.\n\n🎬 Made with MiniMax H3 on PixVerse
📲 @PixVerse \n\n#PixVerseCPP #PixVerse",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 46,
      "snapshotAt": "2026-08-26T00:15:10.320Z"
    },
    "approvedAt": "2026-08-26T00:15:10.320Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092376809954517166",
    "title": "方正坐姿小猫的五秒微动",
    "titleEn": "Five Seconds of Subtle Motion from a Loafing Cat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一只香箱坐姿的小猫在方形画面中轻轻眨眼和呼吸，以极短片段呈现自然微动。 原生视频规格为 5 秒 · 720×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A loafing cat blinks and breathes gently in a square frame, delivering natural micro-movement in a very short clip. Native-video output: 5s · 720×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/haruharuy/status/2092376809954517166",
    "sourceLabel": "X 原帖 · @haruharuy",
    "author": "山下治城:YamashitaHaruki",
    "publishedAt": "2026-08-25T22:21:20.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092376809954517166.mp4",
    "posterUrl": "/posters/x/2092376809954517166.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature",
      "creatures"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "香箱座り　ネコ🐈　動画🎥MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 47,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092372313551646773",
    "title": "绘本风十五秒讲完竹取物语",
    "titleEn": "The Tale of the Bamboo Cutter in Fifteen Seconds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "平面绘本角色和章节色块依次呈现竹中公主、求婚者与月夜离别，把经典民间故事压缩成完整短篇。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Flat picture-book figures and chapter colors move through the bamboo princess, her suitors, and a moonlit farewell, compressing the folk tale into a complete short. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description: Create a complete 15-second Japanese folk-tale explainer in\na vertical 9:16 composition, telling the story of the bamboo princess Kaguya. Use friendly 2D\nflat-vector picture-book illustration: thick, even, near-black outlines of consistent weight,\nflat solid color fills, no shading, simple dot eyes and small curved mouths, restrained cutout\nanimation, and a light paper grain over the flat color fields. Use a compact palette of fresh\nbamboo green, warm gold, warm off-white, vermilion red and deep indigo blue, with near-black\noutlines. Background color changes mark the chapters of the story. The ending is quiet and\nbittersweet, never triumphant. No photorealism, no 3D rendering, no gradients, no drop shadows,\nno browser interface, no screen-recording artifacts, and no player controls.\n[Shot 1] Begin on a fresh bamboo-green background filled with tall straight bamboo stalks. A\nshort old man with a white beard, a brown work robe and a bamboo basket on his back walks\nbetween them, and the base of one single stalk glows warm gold. Then the Japanese caption\n\"むかしむかし\" appears at the top of the frame, sliding in gently word by word.\n[Shot 2] At 00:01.500, a warm gold field wipes across the frame as he cuts the glowing stalk\nopen and light pours out of it, revealing a tiny princess in a white robe standing inside,\nher long black hair falling straight and a soft glow around her. The Japanese caption\n\"かぐやひめ\" then appears in large near-black type beside her.\n[Shot 3] At 00:03.000, a warm off-white field wipes in. The princess grows taller in three\nclear steps until she is a young woman in a white robe with long black hair and a faint glow.\nThe old man and an old woman in an indigo kimono pop in on either side of her and smile.\n[Shot 4] At 00:04.500, a vermilion-red field wipes across the frame. Five noblemen in\ndifferent-coloured court robes and black caps arrive one after another from the frame edges.\nEach of them holds out a gift toward her with both hands.\n[Shot 5] At 00:06.000, staying on vermilion red, the princess raises a folded fan and sets them\nan impossible task. The five noblemen drop their shoulders and walk away off the frame edges,\ntheir gifts left behind on the ground.\n[Shot 6] At 00:07.500, a warm off-white field wipes in. The princess sits alone on a wooden\nveranda, and she looks up at a pale moon and a single tear runs down her cheek.\n[Shot 7] At 00:09.000, a deep indigo-blue field wipes across the frame. A large full moon rises\nin the night sky above the house. The princess bows her head to the old man and the old woman,\ntelling them where she is really from.\n[Shot 8] At 00:10.500, staying on deep indigo, a wide road of warm gold light comes down from\nthe moon and touches the ground. Robed heavenly figures on a cloud appear at the top of it,\ntheir faces left in shadow and never drawn.\n[Shot 9] At 00:12.000, staying on deep indigo, the old man and the old woman each take one of\nthe princess's hands and hold on. A thin celestial robe is laid over her shoulders from above,\nand her expression settles into calm.\n[Shot 10] At 00:13.500, the frame fills with warm gold along the road of light. The princess\nrises slowly up it toward the moon while the old man and the old woman stand small below,\nlooking up after her. The closing caption \"つきへ かえる\" appears in large near-black type,\nassembling in two stages and holding cleanly until exactly 15.000 seconds.\n\noverall_soundscape: Bamboo leaves rustle in a light wind and a stalk splits open with a clean\ncrack, followed by a soft swelling shimmer as the light pours out. Paper-like pops mark each\ncaption and each new arrival. Short restrained whooshes carry the color-field wipes, with a\ngentle wooden creak on the veranda, a single soft chime as the road of light touches the\nground, and a long airy rise as the princess ascends. The last second holds almost still. No\ndialogue.\n\nnon_diegetic_music: Generate an original 15-second Japanese period jingle using koto, shinobue\nbamboo flute, sho reed organ and a few soft hand bells at a slow, unhurried tempo. Begin light\nand curious in the bamboo grove, brighten and open as the light pours out, turn busy and\nslightly comic through the red courtship section, thin down to a single held line as the moon\nrises, swell gently as the road of light appears, and end not on a triumphant chord but on one\nquiet sustained note fading at exactly 15.000 seconds.",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2092372313551646773",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2092372313551646773",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-25T22:03:28.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092372313551646773.mp4",
    "posterUrl": "/posters/x/2092372313551646773.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "illustration",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のモーショングラフィックス × 日本昔話第3弾❗️\n\n「さるかに合戦」の次は「竹取物語（かぐや姫）」です😃\n例によってプロンプトはClaude Codeに任せています。この手法なら世界中のお話が作れそうですね☺️\n\nプロンプト↓\n-------------------\nintegrated_multimodal_description: Create a complete 15-second Japanese folk-tale explainer in\na vertical 9:16 composition, telling the story of the bamboo princess Kaguya. Use friendly 2D\nflat-vector picture-book illustration: thick, even, near-black outlines of consistent weight,\nflat solid color fills, no shading, simple dot eyes and small curved mouths, restrained cutout\nanimation, and a light paper grain over the flat color fields. Use a compact palette of fresh\nbamboo green, warm gold, warm off-white, vermilion red and deep indigo blue, with near-black\noutlines. Background color changes mark the chapters of the story. The ending is quiet and\nbittersweet, never triumphant. No photorealism, no 3D rendering, no gradients, no drop shadows,\nno browser interface, no screen-recording artifacts, and no player controls.\n[Shot 1] Begin on a fresh bamboo-green background filled with tall straight bamboo stalks. A\nshort old man with a white beard, a brown work robe and a bamboo basket on his back walks\nbetween them, and the base of one single stalk glows warm gold. Then the Japanese caption\n\"むかしむかし\" appears at the top of the frame, sliding in gently word by word.\n[Shot 2] At 00:01.500, a warm gold field wipes across the frame as he cuts the glowing stalk\nopen and light pours out of it, revealing a tiny princess in a white robe standing inside,\nher long black hair falling straight and a soft glow around her. The Japanese caption\n\"かぐやひめ\" then appears in large near-black type beside her.\n[Shot 3] At 00:03.000, a warm off-white field wipes in. The princess grows taller in three\nclear steps until she is a young woman in a white robe with long black hair and a faint glow.\nThe old man and an old woman in an indigo kimono pop in on either side of her and smile.\n[Shot 4] At 00:04.500, a vermilion-red field wipes across the frame. Five noblemen in\ndifferent-coloured court robes and black caps arrive one after another from the frame edges.\nEach of them holds out a gift toward her with both hands.\n[Shot 5] At 00:06.000, staying on vermilion red, the princess raises a folded fan and sets them\nan impossible task. The five noblemen drop their shoulders and walk away off the frame edges,\ntheir gifts left behind on the ground.\n[Shot 6] At 00:07.500, a warm off-white field wipes in. The princess sits alone on a wooden\nveranda, and she looks up at a pale moon and a single tear runs down her cheek.\n[Shot 7] At 00:09.000, a deep indigo-blue field wipes across the frame. A large full moon rises\nin the night sky above the house. The princess bows her head to the old man and the old woman,\ntelling them where she is really from.\n[Shot 8] At 00:10.500, staying on deep indigo, a wide road of warm gold light comes down from\nthe moon and touches the ground. Robed heavenly figures on a cloud appear at the top of it,\ntheir faces left in shadow and never drawn.\n[Shot 9] At 00:12.000, staying on deep indigo, the old man and the old woman each take one of\nthe princess's hands and hold on. A thin celestial robe is laid over her shoulders from above,\nand her expression settles into calm.\n[Shot 10] At 00:13.500, the frame fills with warm gold along the road of light. The princess\nrises slowly up it toward the moon while the old man and the old woman stand small below,\nlooking up after her. The closing caption \"つきへ かえる\" appears in large near-black type,\nassembling in two stages and holding cleanly until exactly 15.000 seconds.\n\noverall_soundscape: Bamboo leaves rustle in a light wind and a stalk splits open with a clean\ncrack, followed by a soft swelling shimmer as the light pours out. Paper-like pops mark each\ncaption and each new arrival. Short restrained whooshes carry the color-field wipes, with a\ngentle wooden creak on the veranda, a single soft chime as the road of light touches the\nground, and a long airy rise as the princess ascends. The last second holds almost still. No\ndialogue.\n\nnon_diegetic_music: Generate an original 15-second Japanese period jingle using koto, shinobue\nbamboo flute, sho reed organ and a few soft hand bells at a slow, unhurried tempo. Begin light\nand curious in the bamboo grove, brighten and open as the light pours out, turn busy and\nslightly comic through the red courtship section, thin down to a single held line as the moon\nrises, swell gently as the road of light appears, and end not on a triumphant chord but on one\nquiet sustained note fading at exactly 15.000 seconds.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 9,
      "views": 854,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092371886961578401",
    "title": "山姥辣妹的层层加码变身",
    "titleEn": "A Yamanba Gal Transformation That Keeps Escalating",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "人物从简洁的千禧造型逐步叠加发色、妆容、饰品和厚底鞋，最终进入高密度日文排版的时装广告。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A character layers hair color, makeup, accessories, and platform boots onto a simple Y2K look before landing in a dense Japanese-typography fashion spot. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "# 『YAMANBA TRANSFORM』15秒 詳細台本  尺：15秒｜画角：9:16縦型｜1080×1920｜スタイル：実写ファッションCM／平成ギャル／ヤマンバギャル／Y2K／変身シークエンス／巨大タイポグラフィ／高速モーショングラフィック｜テーマ：「足す。足す。まだ足す。」｜サブテーマ：「MORE IS MORE.」｜主人公：20代女性。同一人物固定。開始時はシンプルなY2Kストリートルック。変身後は濃い日焼け肌、白く強調した目元と鼻筋、淡い白〜ピンク系リップ、ハイトーン金髪＋白・ピンクメッシュ、大量ヘアアクセ、長いデコネイル、白ファー、レオパード、チェーン、ミニスカート、巨大厚底ブーツを組み合わせたヤマンバギャル。奇抜さは強調するが、常にハイファッション広告として美しく見せる。｜音楽：平成ユーロビート×Hyperpop×Fashion Club Electro、BPM154〜158。4つ打ちキック、クラップ、明るいシンセリード、Bubblegum Bass、デジタルボイスチョップ、カメラシャッター音。0〜8秒は段階的に音を増やし、8.6秒で一瞬ブレイク、9.8秒からフルビートで限界突破、13.2秒以降はエンドカード用の大きな2ヒット。 CUT01｜0.0–1.2秒｜BEFORE「足りない。」 Visuals：完全な白背景。主人公が中央に正面立ち。髪は明るいブラウン〜金髪だが装飾は少なく、メイクも控えめ。シンプルなY2Kトップス＋ミニスカート＋普通のスニーカー。画面右側には縦長の全身ミラー。主人公が鏡の中の自分を一瞬見てからカメラへ視線を戻す。画面上部の余白に巨大日本語「足りない。」が出現。 Camera：50mm、胸上から開始。0.0〜0.6秒はほぼ固定、0.6秒から顔へ約8％だけ静かにドリーイン。 Motion：主人公は片眉をわずかに上げ、鏡を見て「違う」と感じるような表情。身体はほぼ静止。 Lighting：クリーンな白色ビューティーライト。影は非常に浅い。広告撮影のビフォー状態として美しく見せる。 VFX：ほぼ無し。文字出現時だけごく薄い黒のスキャンライン。 SFX：小さなヒール／靴音「コツ」→音楽開始直前の低いベース。 Text/Telop：「足りない。」 Emotion：違和感、物足りなさ、変身前の静けさ。 Transition：主人公が指を鳴らす瞬間に白フラッシュ。 CUT02｜1.2–2.5秒｜HAIR「もっと。」 Visuals：白フラッシュ後、顔の超接写。主人公の髪が一気にハイトーン金髪へ変化し、白メッシュ、ピンクメッシュが走る。左右から星、ハート、蝶、カラフルなクリップ、ビーズ、ミニリボンが高速で飛び込み、髪へ次々装着される。髪のボリュームも一段大きくなる。 Camera：85mm顔接写から開始。0.2秒で35mm相当まで急激にプルバックし、髪全体を見せる。最後に約10°だけ右ロール。 Motion：主人公が頭を左から右へ振る。髪の動きと逆方向からアクセサリーが飛び込む。 Lighting：白キー＋ピンクとシアンの細いリムライト。髪の艶を強調。 VFX：アクセサリー装着時に小さな星形フラッシュ。印刷ズレ、微細なクロマティックオフセット。 SFX：パパパパッ→キラッ→ドン。 Text/Telop：巨大「HAIR!」＋小さく「もっと。」 Emotion：最初の変化による爽快感。 Transition：巨大ヘアクリップがレンズ前を横切る。 CUT03｜2.5–3.8秒｜MAKE「まだ。」 Visuals：主人公の顔を真正面から接写。メイクブラシがレンズ前を横切るたびに変身が進む。1回目で日焼け肌が深くなる。2回目で目元が白く強調。3回目で鼻筋に白いハイライト。4回目でラメと強いアイライン。最後に淡い白〜ピンク系リップが完成。 Camera：70mm、顔の正面。ブラシの動きに合わせて左右へ短い3〜5cmのスライド。最後の0.2秒だけ軽くプッシュイン。 Motion：主人公はほぼ静止し、目線だけブラシを追い、最後に正面を見る。 Lighting：硬めのビューティーキー。白い目元・鼻筋・ラメに明確なスペキュラーを作る。 VFX：ブラシの軌跡が白、ピンク、シルバーのグラフィックストロークへ変化。 SFX：サッ！サッ！サッ！→パッ。 Text/Telop：「MAKE!」＋小さく「まだ。」 Emotion：変身が明確に加速する。 Transition：最後の白ブラシストロークが全面を塗りつぶす。 CUT04｜3.8–5.0秒｜NAILS「盛る。」 Visuals：指先の超マクロ。最初は普通の短いネイル。主人公が指を広げる瞬間、ネイルが物理的に伸び、ピンク、白、シルバーの超ロングネイルへ変化。ハート、ラインストーン、星、チェーン、小さな立体パーツが「カチッ、カチッ」と追加される。背景は黒に切り替わり、指先だけが明るく浮かぶ。 Camera：100mm macro。左手の小指から親指へ横スライドし、そのまま右手へ短いウィップ移動。 Motion：主人公が指を広げる→ネイルが伸びる→最後に両手で顔の横をフレーミングする。 Lighting：爪の艶を最大化する細いトップライト＋小さなリム光。 VFX：ラインストーン装着時に極小のスパーク。背景に細いゼブラ柄ラインが一瞬だけ出現。 SFX：カチッ、カチッ、カチッ→キラッ。 Text/Telop：巨大「盛る。」 Emotion：過剰さを楽しむ。 Transition：両手がレンズを一瞬完全に覆う。 CUT05｜5.0–6.3秒｜ACCESSORIES「MORE! MORE! MORE!」 Visuals：両手が開くと、主人公は胸上ショット。首、耳、手首、腰へアクセサリーがビートに合わせて連続装着される。巨大ピアス、複数ネックレス、ブレスレット、大ぶりリング、サングラス、デコ携帯、チェーンベルト。背景はピンク×ブラックのチェッカー。 Camera：35mm。最初は胸上。アクセが装着されるたびにカメラ位置も3段階で変化し、正面→斜め左→斜め右へ高速ジャンプ。 Motion：主人公は肩を振る→サングラスを上げる→デコ携帯を持ち上げる。各動作を音楽の強拍に合わせる。 Lighting：ハードフラッシュ系。アクセサリーの金属とプラスチックに強い反射。 VFX：ステッカーバースト、プリクラ風ハート、ゼブラ柄、レオパード柄が0.2秒単位で切り替わる。 SFX：MORE!の各ヒットに「ドン！ドン！ドン！」 Text/Telop：「MORE!」→「MORE!」→「MORE!」 Emotion：映像も主人公も一気に過剰になる。 Transition：3つ目の「MORE!」が巨大化して全面ワイプ。 CUT06｜6.3–7.5秒｜FASHION「もっと派手に。」 Visuals：全身。衣装がレイヤー式に変身する。シンプルなY2K衣装の上へ黒×ピンクのトップス、レオパード柄、白ファー、チェーン、ミニスカート、大型ベルトが高速で重なっていく。最後にヤマンバらしい派手なシルエットが完成。 Camera：28mm。主人公の周囲を約100°左方向へ高速オービット。主人公自身は右方向へ回転し、カメラと逆方向の相対運動を作る。 Motion：主人公が両腕を広げて1回転。回転中に衣装が段階的に増える。最後に片脚を前へ出して停止。 Lighting：白キー＋マゼンタとライムのリムライト。ファーとチェーンの質感を強調。 VFX：衣装の切り替えごとに紙コラージュ的なリップル。背景グラフィックがストライプ→レオパード→ハートへ変化。 SFX：シュッ→バサッ→ジャラッ→ドン。 Text/Telop：「もっと派手に。」 Emotion：完成が近いと思わせる高揚感。 Transition：カメラが足元へ高速ティルトダウン。 CUT07｜7.5–8.6秒｜PLATFORM BOOTS「まだ足りない。」 Visuals：足元の接写。普通のスニーカーが映る。0.3秒だけ音楽が薄くなる。画面下部に「まだ足りない。」。次の瞬間、靴が巨大な白×シルバーの厚底ブーツへ一気に変形。主人公が強く踏み込むと、床のチェッカー柄が波紋状に外へ広がる。 Camera：35mmローアングル。靴へ高速プッシュインし、着地と同時に少しだけ上へ跳ねるような衝撃カメラ。 Motion：足を上げる→厚底に変化→「ドン」と着地。 Lighting：ブーツに白いトップライト。床面にはピンクの反射。 VFX：床の衝撃波、薄い星形スパーク、チェック柄の歪み。 SFX：一瞬無音→「ドォン！」。 Text/Telop：「まだ足りない。」 Emotion：最大の変身前フック。 Transition：着地衝撃で画面が一瞬ブラックアウト。 CUT08｜8.6–9.8秒｜QUESTION「完成？」 Visuals：黒背景。完成したように見える主人公が中央に全身で立つ。白いリムライトだけで輪郭を強調。巨大白文字「完成？」が画面中央に出現。主人公のサングラスが光を反射する。 Camera：50mm、完全正面。0.5秒だけごくゆっくり4％プッシュイン。 Motion：主人公は静止。0.4秒後にサングラスを少しだけ下げ、目だけカメラを見る。口角がわずかに上がる。 Lighting：超ローキー。ピンクと白の細いリムだけ。 VFX：ほぼ無し。文字にだけ軽いCRTちらつき。 SFX：音楽をほぼ止める。低いベース1音。 Text/Telop：「完成？」 Emotion：間、期待、次への溜め。 Transition：主人公が小さく首を傾けた瞬間に文字が消える。 CUT09｜9.8–11.3秒｜FINAL OVERLOAD「……まだ。」 Visuals：主人公の顔アップ。画面左に小さく「……まだ。」。直後、ショッキングピンクの背景が爆発的に出現。髪飾りがさらに倍増、ネックレス追加、ネイルに追加パーツ、サングラスがさらに派手な形へ。背景にはプリクラ、ゼブラ、レオパード、ハート、星、デコ携帯、ステッカー、巨大「MAX!」が次々飛び込む。画面全体まで“盛られていく”。 Camera：85mm顔アップから18mm超広角まで1秒で急速プルバック。主人公の周囲に広がった巨大なヤマンバ世界全体を見せる。 Motion：主人公は髪を大きく振り、最後に両腕を広げる。背景オブジェクトは人物と逆方向へ放射状に飛ぶ。 Lighting：白キー＋ピンク、シアン、ライムの多色リム。ただし顔色は自然に保つ。 VFX：プリクラ爆発、ハート、星、紙コラージュ、クロマティックズレ、2フレームフラッシュ。 SFX：フルビート復帰→MAX! MAX! MAX!の3ヒット。 Text/Telop：「……まだ。」→「MAX!」「MAX!」「MAX!」→最後に「YAMANBA!」 Emotion：限界突破、圧倒的な変身完成。 Transition：巨大「YAMANBA!」の文字の中をカメラが突き抜ける。 CUT10｜11.3–13.2秒｜HERO WALK「誰のためでもない。」 Visuals：真っ白なランウェイへ急変。完成したヤマンバギャルの主人公が一人、奥から真正面へ歩いてくる。前カットの情報量を大幅に落とし、メイク、髪、アクセ、衣装、厚底の完成形を見せる。左右の余白には何も置かない。 Camera：24mmローアングル。主人公が前進する速度と同じ速度でカメラが後退。最後0.4秒でややローから35mm相当の腰上へ軽くティルトアップ。 Motion：主人公は堂々と3歩。腕を大きく振らず、モデルウォーク。最後に中央で停止。 Lighting：クリーンなハイファッション照明。肌、白メイク、ファー、金属、厚底すべてが明確に読める。 VFX：ほぼ無し。髪と衣装の自然な揺れのみ。 SFX：ビートを少し薄くし、厚底のステップ音「ドン、ドン、ドン」を明瞭にする。 Text/Telop：小さく「誰のためでもない。」 Emotion：奇抜さから自己表現へ意味を転換。 Transition：3歩目の着地で床からショッキングピンクが上へせり上がる。 CUT11｜13.2–15.0秒｜END CARD Visuals：ショッキングピンク背景。主人公は画面中央やや下に全身で立つ。背後に巨大な白い円。上部に小さく「MORE IS MORE.」。中央に画面幅いっぱいの巨大「YAMANBA TRANSFORM」。下部に日本語「盛りすぎくらいが、ちょうどいい。」。右上に小さく「2000 → 2026」。余計な装飾は減らし、文字と主人公を主役にする。 Camera：基本固定。13.2〜13.7秒だけ約4％プッシュイン。その後完全ホールド。 Motion：主人公がサングラスを上げ、最後にカメラへ笑う。「YAMANBA TRANSFORM」が一度だけ着地バウンス。 Lighting：フラットな広告照明。白い円が髪の輪郭を分離。 VFX：最後に星形フラッシュ1回。14.7秒以降は完全静止。 SFX：ドン！→低いベースヒット→カシャッ！ Text/Telop：「MORE IS MORE.」「YAMANBA TRANSFORM」「盛りすぎくらいが、ちょうどいい。」「2000 → 2026」 Emotion：変身の爽快感と自己表現の強さを同時に残す。 Transition：15.0秒終了。 【変身の連続性ルール】主人公の顔、骨格、髪の長さ、体型は全カットで同一人物として維持する。CUT01からCUT09へ向けて、変身要素を必ず積み上げ式に追加する。前のカットで装着された髪色、メイク、ネイル、アクセ、衣装が次カットで消失しない。CUT09が最大装飾量、CUT10とCUT11はその完成形をそのまま維持する。 【ヤマンバ表現ルール】日焼け肌、白く強調した目元・鼻筋、強いアイメイク、ハイトーン髪、大量アクセ、長いネイル、厚底などを特徴として扱うが、笑いもの・ホラー・怪物・汚れた表現にはしない。常に2000年代ギャル文化を現代のファッション広告として再構築したプレミアムな実写表現にする。 【カメラワークルール】毎カット同じドリーインを使わない。顔接写、プルバック、横スライド、マクロ追従、オービット、ティルトダウン、ローアングル、クラッシュアウト、トラッキングを分散する。人物とカメラを同方向に動かし続けず、相対運動を作る。激しい手ブレ、高速360°回転、顔が読めなくなるモーションブラーは禁止。 【タイポグラフィルール】使用する主要文字は「足りない。」「HAIR!」「もっと。」「MAKE!」「まだ。」「盛る。」「MORE!」「もっと派手に。」「まだ足りない。」「完成？」「……まだ。」「MAX!」「YAMANBA!」「誰のためでもない。」「MORE IS MORE.」「YAMANBA TRANSFORM」「盛りすぎくらいが、ちょうどいい。」「2000 → 2026」のみ。意味不明な日本語、文字化け、重複文字、余計なブランド名、ロゴ、透かしを生成しない。文字は字幕ではなく、画面の主役オブジェクトとして衣装やカメラの動きと連動させる。 【演出の核】この映像の見せ場は「完成形」ではなく「まだ足りない」を繰り返しながら限界まで盛っていくプロセス。0.0〜8.6秒は段階的変身、8.6〜9.8秒で一度完全に溜め、9.8〜11.3秒で最大爆発、11.3秒以降は完成した姿を静かに見せる。全編を騒がしくせず、静→増加→静止→爆発→静かなヒーローショットという強弱を明確にする。",
    "promptSourceUrl": "https://x.com/su_nagomi/status/2092371892070232214",
    "sourceUrl": "https://x.com/su_nagomi/status/2092371886961578401",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-25T22:01:46.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092371886961578401.mp4",
    "posterUrl": "/posters/x/2092371886961578401.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ヤマンバギャル第2弾\nヤマンバギャル／Y2K／変身シークエンス／巨大タイポグラフィ／高速モーショングラフィック\nMinimaxH3で動画作成\nプロンプトはリプ欄に👇\n@Hailuo_AI #MinimaxH3",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 6,
      "views": 483,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092371453672915137",
    "title": "漫画屋顶上的急速追逐",
    "titleEn": "A High-Speed Chase Across Comic-Book Rooftops",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "漫画角色在城市屋顶奔跑和跃迁，速度线、分镜框与夸张透视把短片推向连续动作高潮。 原生视频规格为 9 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Comic-book characters run and leap across city rooftops as speed lines, panel framing, and exaggerated perspective drive the short into continuous action. Native-video output: 9s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yonaguniUDON/status/2092371453672915137",
    "sourceLabel": "X 原帖 · @yonaguniUDON",
    "author": "蛸壺_AIラクルゥ開発",
    "publishedAt": "2026-08-25T22:00:03.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092371453672915137.mp4",
    "posterUrl": "/posters/x/2092371453672915137.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "アメコミ風\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 61,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092369986606051691",
    "title": "炭火烤肉的食欲广告镜头",
    "titleEn": "Charcoal-Grilled Meat in Appetite-Driven Close-Ups",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "厚切肉片在炭火上翻烤、冒油并被夹起，近景光泽与烟气组成直接刺激食欲的食物短片。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Thick slices of meat grill over charcoal, glisten, smoke, and lift from the grate in an appetite-driven food short. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nakazakifam/status/2092369986606051691",
    "sourceLabel": "X 原帖 · @nakazakifam",
    "author": "中崎工房 | AIで1時間の仕事を5分で終わらせる人",
    "publishedAt": "2026-08-25T21:54:13.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092369986606051691.mp4",
    "posterUrl": "/posters/x/2092369986606051691.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 x 焼肉が食べたくなる動画。\n\n最近焼肉を食べてないので、猛烈に焼肉が食べたくなりました。しかし、最後に焼肉というよりステーキであることに気づきました。どちらにせよご飯に合いそうな焼いた肉です。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 83,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092367911528898999",
    "title": "办公室日常的连续 B-roll",
    "titleEn": "Continuous B-Roll Through an Office Day",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "人物在办公桌、会议和走廊之间切换，景别和机位持续变化，形成一段较长的职场素材蒙太奇。 原生视频规格为 36 秒 · 1906×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A person moves among desk work, meetings, and corridors as changing shot sizes and camera positions form an extended workplace B-roll montage. Native-video output: 36s · 1906×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/princedoesai/status/2092367911528898999",
    "sourceLabel": "X 原帖 · @princedoesai",
    "author": "Prince does AI",
    "publishedAt": "2026-08-25T21:45:59.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092367911528898999.mp4",
    "posterUrl": "/posters/x/2092367911528898999.jpg",
    "duration": 36,
    "aspectRatio": "landscape",
    "resolution": "1906×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "yet another minimax h3 experiment using ref to image. it can do b-roll quite well",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 57,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092367506115834335",
    "title": "绕口令化成竖屏动漫排版",
    "titleEn": "A Tongue Twister Becomes Vertical Anime Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫角色与日文绕口令在竖屏画面中快速交替，手势、表情和大字排版共同推动节奏。 原生视频规格为 15 秒 · 360×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime character and a Japanese tongue twister alternate rapidly in a vertical frame, driven by gestures, expressions, and oversized lettering. Native-video output: 15s · 360×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Sera_Shiranami/status/2092367506115834335",
    "sourceLabel": "X 原帖 · @Sera_Shiranami",
    "author": "白波セラ",
    "publishedAt": "2026-08-25T21:44:22.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092367506115834335.mp4",
    "posterUrl": "/posters/x/2092367506115834335.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "360×640",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "じゅげむ、じゅげむ……\n\nごぼうの擦り切れ？\n#MiniMaxH3 #白波セラ",
    "engagement": {
      "replies": 4,
      "reposts": 4,
      "likes": 12,
      "views": 436,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092365683023561071",
    "title": "仓鼠角色发出声音的自动流程",
    "titleEn": "A Hamster Character Moves and Vocalizes in an Automated Pipeline",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一只仓鼠角色在方形场景中活动并发出声音，呈现自动化图像到视频与音频流程的原始输出。 原生视频规格为 16 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hamster character moves and vocalizes in a square scene, showing the raw output of an automated image-to-video-and-audio pipeline. Native-video output: 16s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nicoptere/status/2092365683023561071",
    "sourceLabel": "X 原帖 · @nicoptere",
    "author": "Nicolas Barradeau",
    "publishedAt": "2026-08-25T21:37:07.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092365683023561071.mp4",
    "posterUrl": "/posters/x/2092365683023561071.jpg",
    "duration": 16,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 video & audio, FLUX images, automated process, raw result.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 123,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092364267546919032",
    "title": "阿努比斯猫耳少女的利落舞步",
    "titleEn": "An Anubis Catgirl Delivers Sharp Dance Moves",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "阿努比斯风格的猫耳少女在竖屏中快速起舞，服装饰件和人物轮廓随动作保持清晰。 原生视频规格为 10 秒 · 672×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An Anubis-inspired catgirl performs a fast vertical dance while costume details and her silhouette remain readable through the motion. Native-video output: 10s · 672×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/highsan_works/status/2092364267546919032",
    "sourceLabel": "X 原帖 · @highsan_works",
    "author": "ハイさん@AIイラスト",
    "publishedAt": "2026-08-25T21:31:30.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092364267546919032.mp4",
    "posterUrl": "/posters/x/2092364267546919032.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "672×928",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "アヌビスのキレキレダンス動画10。\n\nMiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 13,
      "views": 406,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092363865619317172",
    "title": "五窗口并行完成巨型灌汤包",
    "titleEn": "Five Windows Build a Giant Soup Dumpling in Parallel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "五个同步窗口依次展示和面、包馅、蒸制与开汤，最后汇合成一套完整的巨型灌汤包食谱。 原生视频规格为 65 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Five synchronized windows show dough, filling, steaming, and broth release before converging into a complete giant soup-dumpling recipe. Native-video output: 65s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The chef prepares a supple dumpling dough, rolls it into a very large thin circular wrapper on a lightly floured wooden board, then mixes seasoned pork filling with cubes of chilled gelatinized stock, showing convincing close-ups of the ingredients, hands, textures, flour, moisture, and utensils.\nA generous mound of filling is placed in the center, then the chef slowly lifts and pleats the enormous wrapper around it, creating dozens of small folds while rotating the dumpling by hand until the top closes into a neat twisted knot and the whole dumpling visibly sags with realistic weight.\nThe giant dumpling is transferred carefully into a wide bamboo steamer lined with cabbage leaves, the lid closes, steam escapes naturally around the rim, then after cooking the lid is lifted and dense steam reveals the now-translucent, glossy dumpling sitting intact inside.\nThe chef slides it into a shallow serving bowl, places ginger and black vinegar beside it, then punctures the top gently with chopsticks so a large amount of hot golden broth visibly pours from inside the dumpling into the bowl while the delicate skin remains mostly intact.\nFinish with a close-up of the chef opening the dumpling further and spooning up the broth, pork filling, and thin wrapper together, emphasizing realistic steam, liquid movement, chopstick sounds, bamboo creaks, ceramic clinks, and kitchen ambience, with absolutely no spoken language or music.",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2092363865619317172",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2092363865619317172",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-25T21:29:54.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092363865619317172.mp4",
    "posterUrl": "/posters/x/2092363865619317172.jpg",
    "duration": 65,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made a food recipe video with Minimax H3. Used multi-window sequence (5 windows). Prompt:\n\nThe chef prepares a supple dumpling dough, rolls it into a very large thin circular wrapper on a lightly floured wooden board, then mixes seasoned pork filling with cubes of chilled gelatinized stock, showing convincing close-ups of the ingredients, hands, textures, flour, moisture, and utensils.\nA generous mound of filling is placed in the center, then the chef slowly lifts and pleats the enormous wrapper around it, creating dozens of small folds while rotating the dumpling by hand until the top closes into a neat twisted knot and the whole dumpling visibly sags with realistic weight.\nThe giant dumpling is transferred carefully into a wide bamboo steamer lined with cabbage leaves, the lid closes, steam escapes naturally around the rim, then after cooking the lid is lifted and dense steam reveals the now-translucent, glossy dumpling sitting intact inside.\nThe chef slides it into a shallow serving bowl, places ginger and black vinegar beside it, then punctures the top gently with chopsticks so a large amount of hot golden broth visibly pours from inside the dumpling into the bowl while the delicate skin remains mostly intact.\nFinish with a close-up of the chef opening the dumpling further and spooning up the broth, pork filling, and thin wrapper together, emphasizing realistic steam, liquid movement, chopstick sounds, bamboo creaks, ceramic clinks, and kitchen ambience, with absolutely no spoken language or music.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 16,
      "views": 1500,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092363559804244443",
    "title": "摇滚巨星在舞台上高歌",
    "titleEn": "A Rock Icon Commands the Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "由角色转面图生成的摇滚歌手在舞台上演唱和挥臂，聚光灯与观众视角强化现场感。 原生视频规格为 15 秒 · 732×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rock singer generated from a turnaround sheet performs and gestures on stage as spotlights and audience-facing angles strengthen the live-show feel. Native-video output: 15s · 732×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/minaxlab/status/2092363559804244443",
    "sourceLabel": "X 原帖 · @minaxlab",
    "author": "Mina Labs",
    "publishedAt": "2026-08-25T21:28:41.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092363559804244443.mp4",
    "posterUrl": "/posters/x/2092363559804244443.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "732×1280",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Take Nano Banana and generate Freddy Mercury Turnaround Character Sheet. \nThen put is in Minimax H3 and ask it to act like Freddy Mercury , Sing like Freddy , Sound like Freddy.\nI’m not sure what he sing there.. but the result compare to my short prompt is pretty good. \n\nTry it now on Mina Labs.\n\n▪️In a case you want to place your Turnaround sheet on the video. Same like I did. You can use free Mina Labs Video Editor -> https://www.minaxlab.com/editor\n\nhttps://www.minaxlab.com",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092357500007846263",
    "title": "荒野中苏醒的有角巨兽",
    "titleEn": "A Horned Colossus Awakens in the Wasteland",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "披甲有角巨兽从静态概念图中苏醒，在荒野中转身、低吼并逼近镜头。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An armored horned colossus awakens from a concept image, turns, growls, and advances through a wasteland toward the camera. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/picatrix_picori/status/2092357500007846263",
    "sourceLabel": "X 原帖 · @picatrix_picori",
    "author": "Picatrix Picori",
    "publishedAt": "2026-08-25T21:04:36.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092357500007846263.mp4",
    "posterUrl": "/posters/x/2092357500007846263.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "My final picture from my first 1 year subscription of Midjourney becomes a video with Minimax H3 Max on @magnific \n\nVini, vidi, vici... i've created a monster\n\n#gamedev #indiedev #minimaxh3 #aivideo #veo3 #midjourney #aifilms",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 293,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092190323581555116",
    "title": "冰晶环绕的插画角色苏醒",
    "titleEn": "An Illustrated Character Awakens Among Ice Crystals",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "蓝白插画角色在冰晶、粒子与冷色光晕中产生细腻动作，让静态画面保留原有笔触并获得空间感。 原生视频规格为 15 秒 · 1920×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-and-white illustrated character gains subtle motion among ice crystals, particles, and cool light while preserving the original brushwork and depth. Native-video output: 15s · 1920×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CharaspowerAI/status/2092190323581555116",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-08-25T10:00:18.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092190323581555116.mp4",
    "posterUrl": "/posters/x/2092190323581555116.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1920×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is incredibly good at bringing illustrations to life. 🎨\nThe possibilities are seriously impressive.\n\nMore experiments coming soon.",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 13,
      "views": 1313,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092003946407870917",
    "title": "绿光戒指在荒野造出跑车",
    "titleEn": "A Green Power Ring Constructs a Sports Car",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "男子在荒野中用右手戒指投射绿色能量，光线逐层搭建出跑车并让它加速驶离。 原生视频规格为 15 秒 · 1280×736。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A man projects green energy from a ring on his right hand, assembling a sports car layer by layer before it accelerates away across the landscape. Native-video output: 15s · 1280×736. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description:\n\n<Image 1> is the CHARACTER APPEARANCE REFERENCE.\n<Image 0> is the BACKGROUND LOCATION REFERENCE.\n\nUse <Image 1> to preserve the man's exact facial identity, age, dark hair with gray temples, rugged stubble, body proportions, green-and-black tactical suit, illuminated chest emblem, and power-ring design.\n\nUse <Image 0> to preserve the open landscape, terrain, lighting, atmosphere, and general environmental appearance.\n\nCRITICAL RIGHT-HAND CONTINUITY:\nThe power ring is worn ONLY on the man's anatomical RIGHT HAND throughout the entire video.\nNEVER place the ring on his left hand.\nDo NOT mirror or swap hands when the camera angle changes.\nHis LEFT HAND remains completely ringless throughout the entire sequence.\nEvery green energy projection originates directly from the ring on his RIGHT HAND.\n\nPhotorealistic live-action cinematic realism.\n\nEnvironment: vast isolated American desert-like prairie, dry golden grass, dusty earth, scattered low scrub, distant low hills, huge open sky. Warm late-afternoon sunlight. No buildings, existing vehicles, roads, or other people visible.\n\nThe ONLY vehicle appearing in the scene is the futuristic sports-car construct created by the man's ring.\n\n[00:00–00:02.0]\n\nWide cinematic tracking shot.\n\nThe man from <Image 1> walks casually through the open landscape.\n\nHis arms swing naturally as he walks.\n\nAfter several steps he slows and stops.\n\nHe looks across the empty landscape in front of him as though deciding how he wants to travel.\n\nCamera gently moves from the wider walking shot into a medium three-quarter view.\n\n[00:02.0–00:03.0]\n\nMedium shot.\n\nHe turns slightly toward an open patch of ground beside him.\n\nHe raises his anatomical RIGHT HAND and deliberately points the power ring toward the empty space.\n\nHis LEFT hand remains completely ringless.\n\nThe ring begins glowing with concentrated emerald-green energy.\n\n[00:03.0–00:06.5]\n\nA narrow, clearly defined GREEN ENERGY BEAM shoots directly from the ring on his RIGHT HAND into the empty space beside him.\n\nCRITICAL:\nFrom this moment through the FINAL FRAME, the green energy beam remains visibly and continuously connected between his RIGHT-HAND ring and the construct sports car.\n\nThe beam NEVER disconnects.\nThe beam NEVER disappears.\nThe beam NEVER switches hands.\nThe beam NEVER originates from his left hand.\n\nCamera shifts to a wider three-quarter angle so both the man and the forming vehicle remain clearly visible.\n\nThe beam begins DRAWING A FUTURISTIC SPORTS CAR INTO EXISTENCE like a three-dimensional emerald sketch.\n\nThe construction is clearly progressive and physically understandable.\n\nBright emerald lines rapidly trace:\n\nwheels → lower chassis → aerodynamic nose → side profile → rear body → windshield and cockpit → roofline → remaining body panels.\n\nThe glowing wireframe establishes the complete shape of a low, sleek, aggressive futuristic sports car.\n\nAs each section is drawn, the wireframe rapidly fills and solidifies into dense, physical-looking emerald-green construct energy.\n\nThe vehicle is NOT an ordinary green-painted automobile.\n\nIt is unmistakably a POWER-RING ENERGY CONSTRUCT: solid and physically usable, but composed entirely of luminous translucent emerald-green energy with subtle internal glowing structure.\n\nFour clearly defined wheels.\nBelievable tires made from solid green construct energy.\nLow aerodynamic body.\nFuturistic cockpit.\nSleek windshield.\nAggressive but elegant proportions.\n\nThe DRIVER-SIDE DOOR AREA deliberately remains unfinished and open while the rest of the vehicle completes.\n\n[00:06.5–00:08.0]\n\nThe sports car finishes forming except for the open driver's entrance.\n\nThe man lowers his right arm slightly while STILL maintaining the visible energy connection between his RIGHT-HAND ring and the construct car.\n\nHe looks over the completed futuristic car.\n\nA confident, slightly cocky smirk appears on his face.\n\nHold just long enough to clearly read his expression.\n\nNo dialogue.\n\n[00:08.0–00:10.5]\n\nCamera moves into a dynamic medium-wide side angle.\n\nThe man walks directly to the open driver's side.\n\nThe continuous green energy beam remains connected from his RIGHT-HAND ring to the car throughout his movement.\n\nHe naturally steps into the driver's seat.\n\nDo NOT make him teleport into the vehicle.\n\nDo NOT have him pass through the bodywork.\n\nHe physically enters through the intentionally unfinished driver's-side opening and sits behind the wheel.\n\nOnce he is fully seated, glowing construct lines rapidly draw and solidify the final driver's-side door around him, completing the sports car.\n\nThe green beam STILL remains connected from the ring on his anatomical RIGHT HAND to the car.\n\n[00:10.5–00:12.0]\n\nLow cinematic three-quarter shot of the completed vehicle.\n\nThe futuristic emerald construct sports car powers up.\n\nInternal green energy pulses subtly through its body.\n\nThe wheels begin turning.\n\nDust shifts beneath the tires.\n\nThrough the cockpit, the man gives another brief confident, cocky smirk.\n\nHis RIGHT HAND remains the hand sustaining the construct.\n\nThe continuous green energy connection remains active.\n\n[00:12.0–00:15.0]\n\nThe construct sports car launches forward with powerful acceleration.\n\nReal dust and loose dirt kick backward from beneath the glowing green construct wheels.\n\nCamera briefly tracks alongside at a low angle as the vehicle rapidly gains speed.\n\nA beautiful EMERALD-GREEN ENERGY TRAIL streams behind the construct car as it accelerates across the open landscape.\n\nThe car sweeps past the camera and drives COMPLETELY OUT OF FRAME.\n\nDo NOT stop the car before it exits.\n\nDo NOT dissolve the vehicle.\n\nDo NOT transform it into another object.\n\nThe construct remains a complete futuristic sports car as it leaves the shot.\n\nCRITICAL FINAL CONTINUITY:\nThe sports car remains actively sustained by the man's RIGHT-HAND power ring while he drives.\n\nThe visible green energy connection remains continuously linked to his RIGHT-hand ring and integrated into the construct car's glowing energy structure through the FINAL FRAME.\n\nThe ring NEVER appears on his left hand.\n\nNo dialogue.\nNo attack.\nNo explosion.\nNo additional constructs.\nNo additional characters.\nNo existing real-world vehicles.\n\noverall_soundscape:\n\nNatural open-land ambience with gentle dry wind moving through grass and subtle footsteps on dusty earth at the beginning.\n\nWhen the RIGHT-HAND ring activates, introduce a clean concentrated emerald-energy hum.\n\nThe three-dimensional drawing process produces precise crystalline energy sounds as glowing lines trace the vehicle into existence, followed by deeper resonant energy tones as the wireframe solidifies into the physical construct.\n\nOnce completed, the sports car produces a smooth futuristic energy-powered motor sound rather than a conventional gasoline engine.\n\nSubtle rising energy whine as the vehicle powers up.\n\nConstruct tires interact physically with the terrain: dirt crunch, tire movement, and a strong burst of dust and gravel as the car accelerates.\n\nThe energy motor rises rapidly in pitch as the car drives away, accompanied by a subtle sustained energy sound from the emerald trail.\n\nNatural wind and landscape ambience remain underneath.\n\nNo dialogue.\n\nnon_diegetic_music:\n\nNone.",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2092003946407870917",
    "sourceUrl": "https://x.com/EndFolding79421/status/2092003946407870917",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-08-24T21:39:43.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2092003946407870917.mp4",
    "posterUrl": "/posters/x/2092003946407870917.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "illustration",
      "dark"
    ],
    "scenes": [
      "combat",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 COmfyui Green lanter #2..\nLInk to 1st test: https://x.com/EndFolding79421/status/2089840714016497677\n\nPrompt:\n\nintegrated_multimodal_description:\n\n<Image 1> is the CHARACTER APPEARANCE REFERENCE.\n<Image 0> is the BACKGROUND LOCATION REFERENCE.\n\nUse <Image 1> to preserve the man's exact facial identity, age, dark hair with gray temples, rugged stubble, body proportions, green-and-black tactical suit, illuminated chest emblem, and power-ring design.\n\nUse <Image 0> to preserve the open landscape, terrain, lighting, atmosphere, and general environmental appearance.\n\nCRITICAL RIGHT-HAND CONTINUITY:\nThe power ring is worn ONLY on the man's anatomical RIGHT HAND throughout the entire video.\nNEVER place the ring on his left hand.\nDo NOT mirror or swap hands when the camera angle changes.\nHis LEFT HAND remains completely ringless throughout the entire sequence.\nEvery green energy projection originates directly from the ring on his RIGHT HAND.\n\nPhotorealistic live-action cinematic realism.\n\nEnvironment: vast isolated American desert-like prairie, dry golden grass, dusty earth, scattered low scrub, distant low hills, huge open sky. Warm late-afternoon sunlight. No buildings, existing vehicles, roads, or other people visible.\n\nThe ONLY vehicle appearing in the scene is the futuristic sports-car construct created by the man's ring.\n\n[00:00–00:02.0]\n\nWide cinematic tracking shot.\n\nThe man from <Image 1> walks casually through the open landscape.\n\nHis arms swing naturally as he walks.\n\nAfter several steps he slows and stops.\n\nHe looks across the empty landscape in front of him as though deciding how he wants to travel.\n\nCamera gently moves from the wider walking shot into a medium three-quarter view.\n\n[00:02.0–00:03.0]\n\nMedium shot.\n\nHe turns slightly toward an open patch of ground beside him.\n\nHe raises his anatomical RIGHT HAND and deliberately points the power ring toward the empty space.\n\nHis LEFT hand remains completely ringless.\n\nThe ring begins glowing with concentrated emerald-green energy.\n\n[00:03.0–00:06.5]\n\nA narrow, clearly defined GREEN ENERGY BEAM shoots directly from the ring on his RIGHT HAND into the empty space beside him.\n\nCRITICAL:\nFrom this moment through the FINAL FRAME, the green energy beam remains visibly and continuously connected between his RIGHT-HAND ring and the construct sports car.\n\nThe beam NEVER disconnects.\nThe beam NEVER disappears.\nThe beam NEVER switches hands.\nThe beam NEVER originates from his left hand.\n\nCamera shifts to a wider three-quarter angle so both the man and the forming vehicle remain clearly visible.\n\nThe beam begins DRAWING A FUTURISTIC SPORTS CAR INTO EXISTENCE like a three-dimensional emerald sketch.\n\nThe construction is clearly progressive and physically understandable.\n\nBright emerald lines rapidly trace:\n\nwheels → lower chassis → aerodynamic nose → side profile → rear body → windshield and cockpit → roofline → remaining body panels.\n\nThe glowing wireframe establishes the complete shape of a low, sleek, aggressive futuristic sports car.\n\nAs each section is drawn, the wireframe rapidly fills and solidifies into dense, physical-looking emerald-green construct energy.\n\nThe vehicle is NOT an ordinary green-painted automobile.\n\nIt is unmistakably a POWER-RING ENERGY CONSTRUCT: solid and physically usable, but composed entirely of luminous translucent emerald-green energy with subtle internal glowing structure.\n\nFour clearly defined wheels.\nBelievable tires made from solid green construct energy.\nLow aerodynamic body.\nFuturistic cockpit.\nSleek windshield.\nAggressive but elegant proportions.\n\nThe DRIVER-SIDE DOOR AREA deliberately remains unfinished and open while the rest of the vehicle completes.\n\n[00:06.5–00:08.0]\n\nThe sports car finishes forming except for the open driver's entrance.\n\nThe man lowers his right arm slightly while STILL maintaining the visible energy connection between his RIGHT-HAND ring and the construct car.\n\nHe looks over the completed futuristic car.\n\nA confident, slightly cocky smirk appears on his face.\n\nHold just long enough to clearly read his expression.\n\nNo dialogue.\n\n[00:08.0–00:10.5]\n\nCamera moves into a dynamic medium-wide side angle.\n\nThe man walks directly to the open driver's side.\n\nThe continuous green energy beam remains connected from his RIGHT-HAND ring to the car throughout his movement.\n\nHe naturally steps into the driver's seat.\n\nDo NOT make him teleport into the vehicle.\n\nDo NOT have him pass through the bodywork.\n\nHe physically enters through the intentionally unfinished driver's-side opening and sits behind the wheel.\n\nOnce he is fully seated, glowing construct lines rapidly draw and solidify the final driver's-side door around him, completing the sports car.\n\nThe green beam STILL remains connected from the ring on his anatomical RIGHT HAND to the car.\n\n[00:10.5–00:12.0]\n\nLow cinematic three-quarter shot of the completed vehicle.\n\nThe futuristic emerald construct sports car powers up.\n\nInternal green energy pulses subtly through its body.\n\nThe wheels begin turning.\n\nDust shifts beneath the tires.\n\nThrough the cockpit, the man gives another brief confident, cocky smirk.\n\nHis RIGHT HAND remains the hand sustaining the construct.\n\nThe continuous green energy connection remains active.\n\n[00:12.0–00:15.0]\n\nThe construct sports car launches forward with powerful acceleration.\n\nReal dust and loose dirt kick backward from beneath the glowing green construct wheels.\n\nCamera briefly tracks alongside at a low angle as the vehicle rapidly gains speed.\n\nA beautiful EMERALD-GREEN ENERGY TRAIL streams behind the construct car as it accelerates across the open landscape.\n\nThe car sweeps past the camera and drives COMPLETELY OUT OF FRAME.\n\nDo NOT stop the car before it exits.\n\nDo NOT dissolve the vehicle.\n\nDo NOT transform it into another object.\n\nThe construct remains a complete futuristic sports car as it leaves the shot.\n\nCRITICAL FINAL CONTINUITY:\nThe sports car remains actively sustained by the man's RIGHT-HAND power ring while he drives.\n\nThe visible green energy connection remains continuously linked to his RIGHT-hand ring and integrated into the construct car's glowing energy structure through the FINAL FRAME.\n\nThe ring NEVER appears on his left hand.\n\nNo dialogue.\nNo attack.\nNo explosion.\nNo additional constructs.\nNo additional characters.\nNo existing real-world vehicles.\n\noverall_soundscape:\n\nNatural open-land ambience with gentle dry wind moving through grass and subtle footsteps on dusty earth at the beginning.\n\nWhen the RIGHT-HAND ring activates, introduce a clean concentrated emerald-energy hum.\n\nThe three-dimensional drawing process produces precise crystalline energy sounds as glowing lines trace the vehicle into existence, followed by deeper resonant energy tones as the wireframe solidifies into the physical construct.\n\nOnce completed, the sports car produces a smooth futuristic energy-powered motor sound rather than a conventional gasoline engine.\n\nSubtle rising energy whine as the vehicle powers up.\n\nConstruct tires interact physically with the terrain: dirt crunch, tire movement, and a strong burst of dust and gravel as the car accelerates.\n\nThe energy motor rises rapidly in pitch as the car drives away, accompanied by a subtle sustained energy sound from the emerald trail.\n\nNatural wind and landscape ambience remain underneath.\n\nNo dialogue.\n\nnon_diegetic_music:\n\nNone.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 278,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2091989570598412736",
    "title": "盾牌穿梭于漫画式工业战场",
    "titleEn": "A Shield Ricochets Through a Comic-Book Factory Battle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "蓝衣英雄在工业基地迎战持枪士兵，盾牌弹射、激光和冲击构图串起高速漫画动作场面。 原生视频规格为 20 秒 · 1280×736。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A blue-suited hero fights armed soldiers inside an industrial base as shield ricochets, lasers, and impact compositions drive a fast comic-book battle. Native-video output: 20s · 1280×736. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "VISUAL STYLE:\nPolished modern superhero cartoon animation with strong comic-book influence. Bold clean outlines, expressive heroic character animation, vivid colors, dynamic speed lines, stylized impact frames, exaggerated but physically readable action, fluid animation and crisp character silhouettes.\n\nCaptain America fights a squad of HYDRA agents inside a large industrial HYDRA weapons facility. Captain America wears his recognizable blue superhero combat uniform and carries his circular red-white-and-blue shield. HYDRA agents wear matching dark military uniforms with HYDRA insignia and carry futuristic energy laser blasters.\n\nLIGHTING:\nAutomatic cinematic action lighting optimized for visual clarity. Cold industrial facility lighting contrasts strongly with glowing red laser bolts, sparks, weapon flashes and brief explosive illumination.\n\nCAMERA:\nFAST COMIC-BOOK CAMERA. Aggressive perspective changes, whip-pans, dramatic low angles, rapid push-ins, shield-following shots and brief comic-book impact compositions. Despite the speed, every major attack remains visually readable.\n\nACTION SPEED — CRITICAL:\nCombat occurs at FAST NORMAL REAL-WORLD SPEED.\nNo slow motion.\nCaptain America moves quickly and decisively.\nHYDRA agents fire and react quickly.\nDo not make combat floaty, sluggish or posed.\n\n[00:00–00:03]\nWide low-angle establishing shot. Captain America stands in the center of the industrial HYDRA facility as several HYDRA agents spread out ahead of him and immediately raise their laser blasters.\n\nRed laser bolts streak toward him.\n\nRapid push-in as Captain America snaps his shield up. Multiple laser bolts strike the shield with bright energy flashes and showers of sparks.\n\n[00:03–00:06]\nWHIP-PAN into a dynamic side angle.\n\nCaptain America charges forward behind his shield while laser bolts streak past him.\n\nHe angles the shield and deliberately DEFLECTS one incoming laser bolt sideways into a HYDRA agent's blaster, knocking the weapon from the agent's hands in a burst of sparks.\n\nCap immediately shoulder-checks another agent with the shield and keeps moving.\n\n[00:06–00:10]\nFast comic-book cut.\n\nA HYDRA agent fires from an elevated metal platform.\n\nCaptain America spins and HURLS THE SHIELD.\n\nCamera rapidly follows directly behind the spinning shield as it flies across the facility.\n\nThe shield smashes the elevated agent's laser blaster aside, ricochets off a steel support beam, then strikes a second agent's weapon.\n\nQuick impact-frame flashes emphasize both hits.\n\n[00:10–00:13]\nLow-angle tracking shot alongside Captain America as he runs forward.\n\nThe spinning shield ricochets back toward him.\n\nWithout stopping, Captain America catches it cleanly on his arm.\n\nTwo HYDRA agents rush him.\n\nCap immediately blocks a laser blast, drives forward with a shield strike that sends one agent tumbling backward, then pivots and delivers a fast sweeping kick that knocks the second agent down.\n\n[00:13–00:17]\nRapid overhead comic-book shot.\n\nThree remaining HYDRA agents surround Captain America and unleash crossing streams of red laser fire.\n\nCap ducks, pivots and moves rapidly between the blasts while protecting himself with the shield.\n\nCUT CLOSE:\n\nA laser strikes the shield.\n\nCaptain America deliberately changes its angle.\n\nThe reflected laser blasts a nearby HYDRA weapons console.\n\nThe console ERUPTS in a dramatic shower of cartoon sparks, smoke and electrical energy.\n\nThe HYDRA agents recoil from the blast.\n\n[00:17–00:20]\nHeroic finishing sequence.\n\nCaptain America bursts through the smoke and throws his shield one final time.\n\nDynamic shield-following camera as it rapidly ricochets between the laser blasters of the remaining HYDRA agents, knocking the weapons from their hands one after another.\n\nThe shield arcs back.\n\nLOW-ANGLE HERO SHOT:\n\nCaptain America catches the shield perfectly.\n\nThe defeated and disarmed HYDRA agents lie or stumble around the smoky facility behind him.\n\nCap plants his feet and raises the shield defensively as sparks rain from the damaged machinery.\n\nFinal frame holds briefly on the strong animated comic-book hero pose.\n\nNo blood.\nNo gore.\nKeep Captain America's shield present and visually consistent throughout.\nDo not give Captain America a gun or laser weapon.\nHYDRA agents use the laser blasters.\nPrioritize clear, fast, readable superhero choreography over chaotic motion.\n\noverall_soundscape:\n\nSound ON.\n\nFast futuristic laser-blaster fire with sharp animated energy reports.\n\nHeavy metallic shield impacts, distinctive shield ricochets, energy blasts striking the shield, rushing shield flight sounds, punches, kicks, boots on the industrial floor, HYDRA agent reactions, electrical crackling, sparks, machinery and explosive console destruction.\n\nSound effects synchronize tightly with every visible impact and laser blast.\n\nnon_diegetic_music:\n\nEnergetic orchestral superhero action score throughout.\n\nDriving percussion, heroic brass, rapid strings and rising action intensity synchronized with the combat.\n\nMusic builds through the shield ricochet sequence and reaches a strong heroic musical hit exactly as Captain America catches the shield for the final pose.\n```",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2091989570598412736",
    "sourceUrl": "https://x.com/EndFolding79421/status/2091989570598412736",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-08-24T20:42:35.000Z",
    "addedAt": "2026-08-26T00:15:10.321Z",
    "mediaUrl": "/media/x-2091989570598412736.mp4",
    "posterUrl": "/posters/x/2091989570598412736.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 T2V COmfyui Test (20 Seconds)....\n\nintegrated_multimodal_description:\n\nVISUAL STYLE:\nPolished modern superhero cartoon animation with strong comic-book influence. Bold clean outlines, expressive heroic character animation, vivid colors, dynamic speed lines, stylized impact frames, exaggerated but physically readable action, fluid animation and crisp character silhouettes.\n\nCaptain America fights a squad of HYDRA agents inside a large industrial HYDRA weapons facility. Captain America wears his recognizable blue superhero combat uniform and carries his circular red-white-and-blue shield. HYDRA agents wear matching dark military uniforms with HYDRA insignia and carry futuristic energy laser blasters.\n\nLIGHTING:\nAutomatic cinematic action lighting optimized for visual clarity. Cold industrial facility lighting contrasts strongly with glowing red laser bolts, sparks, weapon flashes and brief explosive illumination.\n\nCAMERA:\nFAST COMIC-BOOK CAMERA. Aggressive perspective changes, whip-pans, dramatic low angles, rapid push-ins, shield-following shots and brief comic-book impact compositions. Despite the speed, every major attack remains visually readable.\n\nACTION SPEED — CRITICAL:\nCombat occurs at FAST NORMAL REAL-WORLD SPEED.\nNo slow motion.\nCaptain America moves quickly and decisively.\nHYDRA agents fire and react quickly.\nDo not make combat floaty, sluggish or posed.\n\n[00:00–00:03]\nWide low-angle establishing shot. Captain America stands in the center of the industrial HYDRA facility as several HYDRA agents spread out ahead of him and immediately raise their laser blasters.\n\nRed laser bolts streak toward him.\n\nRapid push-in as Captain America snaps his shield up. Multiple laser bolts strike the shield with bright energy flashes and showers of sparks.\n\n[00:03–00:06]\nWHIP-PAN into a dynamic side angle.\n\nCaptain America charges forward behind his shield while laser bolts streak past him.\n\nHe angles the shield and deliberately DEFLECTS one incoming laser bolt sideways into a HYDRA agent's blaster, knocking the weapon from the agent's hands in a burst of sparks.\n\nCap immediately shoulder-checks another agent with the shield and keeps moving.\n\n[00:06–00:10]\nFast comic-book cut.\n\nA HYDRA agent fires from an elevated metal platform.\n\nCaptain America spins and HURLS THE SHIELD.\n\nCamera rapidly follows directly behind the spinning shield as it flies across the facility.\n\nThe shield smashes the elevated agent's laser blaster aside, ricochets off a steel support beam, then strikes a second agent's weapon.\n\nQuick impact-frame flashes emphasize both hits.\n\n[00:10–00:13]\nLow-angle tracking shot alongside Captain America as he runs forward.\n\nThe spinning shield ricochets back toward him.\n\nWithout stopping, Captain America catches it cleanly on his arm.\n\nTwo HYDRA agents rush him.\n\nCap immediately blocks a laser blast, drives forward with a shield strike that sends one agent tumbling backward, then pivots and delivers a fast sweeping kick that knocks the second agent down.\n\n[00:13–00:17]\nRapid overhead comic-book shot.\n\nThree remaining HYDRA agents surround Captain America and unleash crossing streams of red laser fire.\n\nCap ducks, pivots and moves rapidly between the blasts while protecting himself with the shield.\n\nCUT CLOSE:\n\nA laser strikes the shield.\n\nCaptain America deliberately changes its angle.\n\nThe reflected laser blasts a nearby HYDRA weapons console.\n\nThe console ERUPTS in a dramatic shower of cartoon sparks, smoke and electrical energy.\n\nThe HYDRA agents recoil from the blast.\n\n[00:17–00:20]\nHeroic finishing sequence.\n\nCaptain America bursts through the smoke and throws his shield one final time.\n\nDynamic shield-following camera as it rapidly ricochets between the laser blasters of the remaining HYDRA agents, knocking the weapons from their hands one after another.\n\nThe shield arcs back.\n\nLOW-ANGLE HERO SHOT:\n\nCaptain America catches the shield perfectly.\n\nThe defeated and disarmed HYDRA agents lie or stumble around the smoky facility behind him.\n\nCap plants his feet and raises the shield defensively as sparks rain from the damaged machinery.\n\nFinal frame holds briefly on the strong animated comic-book hero pose.\n\nNo blood.\nNo gore.\nKeep Captain America's shield present and visually consistent throughout.\nDo not give Captain America a gun or laser weapon.\nHYDRA agents use the laser blasters.\nPrioritize clear, fast, readable superhero choreography over chaotic motion.\n\noverall_soundscape:\n\nSound ON.\n\nFast futuristic laser-blaster fire with sharp animated energy reports.\n\nHeavy metallic shield impacts, distinctive shield ricochets, energy blasts striking the shield, rushing shield flight sounds, punches, kicks, boots on the industrial floor, HYDRA agent reactions, electrical crackling, sparks, machinery and explosive console destruction.\n\nSound effects synchronize tightly with every visible impact and laser blast.\n\nnon_diegetic_music:\n\nEnergetic orchestral superhero action score throughout.\n\nDriving percussion, heroic brass, rapid strings and rising action intensity synchronized with the combat.\n\nMusic builds through the shield ricochet sequence and reaches a strong heroic musical hit exactly as Captain America catches the shield for the final pose.\n```",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 194,
      "snapshotAt": "2026-08-26T00:15:10.321Z"
    },
    "approvedAt": "2026-08-26T00:15:10.321Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092845610999386500",
    "title": "霓虹街头的粉发角色回眸",
    "titleEn": "A Pink-Haired Character Turns on a Neon Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发动漫角色背对镜头站在霓虹街头，细微的肩颈和发丝运动让短暂回眸更具氛围。 原生视频规格为 10 秒 · 512×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired anime character stands with her back to the camera on a neon street, where subtle shoulder and hair movement give a brief turn added atmosphere. Native-video output: 10s · 512×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YamataZen/status/2092845610999386500",
    "sourceLabel": "X 原帖 · @YamataZen",
    "author": "Yamata",
    "publishedAt": "2026-08-27T05:24:11.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092845610999386500.mp4",
    "posterUrl": "/posters/x/2092845610999386500.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "512×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Exposed nape and shoulders\n\n#aivideo #MiniMaxH3 #OC",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 20,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092843298771628454",
    "title": "仓鼠团队的迷你创作工作室",
    "titleEn": "A Hamster Team Runs a Tiny Creative Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拟人仓鼠在木质工作室里操作剪辑、音频和设计设备，五秒内完成一段温暖的团队协作展示。 原生视频规格为 5 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Anthropomorphic hamsters operate editing, audio, and design gear in a wooden studio, creating a warm five-second portrait of creative teamwork. Native-video output: 5s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HampterNFT/status/2092843298771628454",
    "sourceLabel": "X 原帖 · @HampterNFT",
    "author": "HamsterNFT 🐹",
    "publishedAt": "2026-08-27T05:15:00.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092843298771628454.mp4",
    "posterUrl": "/posters/x/2092843298771628454.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Thursday 🎬🎧\nInfrastructure Hamsters by day… Hamster Studio creators by slightly later in the morning. 🐹🎬🎧\nEditing, audio, graphics and lots of tiny ideas becoming bigger ones. ✨\nCreate. Experiment. Have fun.\n#Morning @Hailuo_AI #Hamsters #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 5,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092839767637676165",
    "title": "录音棚里的长镜头演唱",
    "titleEn": "A Long-Take Vocal Performance in the Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女歌手在昏暗录音棚中面对电容麦克风持续演唱，机位和灯光变化维持近一分钟的现场感。 原生视频规格为 49 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A female singer performs into a condenser microphone in a dark studio as camera and lighting changes sustain a live-session feel for nearly a minute. Native-video output: 49s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/suizenji98/status/2092839767637676165",
    "sourceLabel": "X 原帖 · @suizenji98",
    "author": "どっと",
    "publishedAt": "2026-08-27T05:00:58.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092839767637676165.mp4",
    "posterUrl": "/posters/x/2092839767637676165.jpg",
    "duration": 49,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ファーストテイク風MV作ってみた\n#MinimaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 6,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092839527442493879",
    "title": "城市校园广告的青春蒙太奇",
    "titleEn": "A Youthful Montage for an Urban School Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "学生从宿舍与走廊奔向城市街头，柔和逆光和快速切镜组成一段面向招生宣传的青春蒙太奇。 原生视频规格为 15 秒 · 2894×1576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A student moves from a dormitory and corridor into the city as soft backlight and quick cuts form a youthful admissions-style montage. Native-video output: 15s · 2894×1576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_300/status/2092839527442493879",
    "sourceLabel": "X 原帖 · @ai_300",
    "author": "鈴木@アナログ営業会社を100日後にAIで売上を300％にする人",
    "publishedAt": "2026-08-27T05:00:01.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092839527442493879.mp4",
    "posterUrl": "/posters/x/2092839527442493879.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2894×1576",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "CM制作、ここまでAIで作れる\nChatGPTとminiMAX H3で専門学校のCM素材を制作\n👇\n人物・背景・光までAIで映像素材にできる\n撮影していないのにCMっぽい画が完成する　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　\n\n今回作ったのは\n専門学校を想定したCM素材\n\nまずChatGPTで\n「誰に何を伝えるCMなのか」を整理\n\n高校生が見た瞬間に\n学校生活をイメージできる構成を考える\n\n次に動画生成AIへ\n欲しい映像のイメージを入力\n\n人物の雰囲気\n服装\nカメラとの距離\n\n背景の街並み\n時間帯\n光の入り方\n\nCM全体の空気感まで指定して\n映像素材を生成する\n\n今回の画では\n白いシャツを着た人物を中央に配置\n\n背景は街中\n後ろから光が差し込む構図にして\n\n「これから新しい場所へ進む」\nという学校CMらしい印象を狙っている\n\nここがAI動画の面白いところ\n\n従来なら\n出演者を決める\n↓\nロケーションを探す\n↓\n撮影する\n↓\n使えるカットを選ぶ\n\nという流れだった\n\n生成AIなら\nまずCMの完成イメージを映像化して\n\n「この方向性でいけるか」を\n撮影前に確認する使い方もできる\n\nもちろん\nAIで生成した映像はそのまま使う前に\n\n人物の動き\n文字の崩れ\n背景の違和感\n学校の実態とのズレ\n\nこのあたりは人間の確認が必要\n\nそれでも\n企画段階でCMの世界観を\n実際の映像として見せられるのは強い\n\n専門学校だけじゃない\n\n大学\nスクール\n採用広告\n企業紹介\nサービス紹介\n\nこういう「イメージを伝える動画」は\nAI生成との相性がかなりいい\n\nAI動画は\n完成CMを作る道具だけじゃない\n\n撮影前に\n「完成形を見せる道具」として使うだけでも\n広告制作の進め方が変わる",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 254,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092839052605362665",
    "title": "机械遗迹中的时装英雄镜头",
    "titleEn": "A Fashion Heroine Among Mechanical Ruins",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "时装角色在巨型机械残骸和发光装置之间移动，低机位与能量光环强化科幻海报般的视觉冲击。 原生视频规格为 15 秒 · 1080×1584。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fashion-forward heroine moves among colossal mechanical ruins and glowing machinery, with low angles and energy halos creating a sci-fi poster aesthetic. Native-video output: 15s · 1080×1584. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2092839052605362665",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-08-27T04:58:07.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092839052605362665.mp4",
    "posterUrl": "/posters/x/2092839052605362665.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1584",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat",
      "fashion"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Midjourney v8.2 --sref 3903476209 --profile cl83hbe\n\nAnimated by MiniMax H3 Max on @magnific\nVideo upscaled by @topazlabs Astra",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 312,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092836934154637398",
    "title": "橙发剑士的游戏角色宣传片",
    "titleEn": "An Orange-Haired Swordswoman Gets a Game Character PV",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "橙发剑士从拔刀动作切入属性面板和英文角色名，刀光、披风与排版共同完成游戏角色宣传片。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An orange-haired swordswoman moves from a sword draw into stat panels and an English character title, combining blade light, cape motion, and typography in a game-character PV. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Heidifu666/status/2092836934154637398",
    "sourceLabel": "X 原帖 · @Heidifu666",
    "author": "赛博大福",
    "publishedAt": "2026-08-27T04:49:42.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092836934154637398.mp4",
    "posterUrl": "/posters/x/2092836934154637398.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3用来做品牌IP的动态演绎能力，太合适了。\n我只给 H3 给一张 游戏编程公司的IP图片， 直接生成一支约 15 秒的动漫 PV 式角色介绍片。\n最惊艳的是它的动画转场和节奏感。从眼神特写、拔剑、挥砍到角色信息页，镜头不是简单切换，而是通过动作衔接、刀光擦镜、披风运动、快速推拉自然转场，角色名、属性值、Logo 这些品牌信息，也能顺着动画节奏自然进入，再配合音乐卡点，已经很接近一支成熟动漫 OP 的观看体验。\n以后角色介绍、节日短片、产品联动，都可以变成更轻量、更高频的内容。品牌 IP，可以从静态资产，直接进入动态内容生产。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 37,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092834537474753005",
    "title": "人声驱动的至上主义文字世界",
    "titleEn": "A Suprematist Typography World Driven by Voice",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色人物剪影用拟声词生成文字和几何形，画面从极简留白逐步失控，最终收束为一个“无”字。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black silhouette voices onomatopoeia that becomes typography and geometry, escalating from sparse space into controlled chaos before ending on a single word for nothingness. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "シュプレマティスム×シュルレアリスム｜抽象芸術的モーショングラフィック｜人物は黒いシルエット1人｜背景は現実空間ではなく、生成りホワイトを基調に、墨黒・くすんだ赤・オーカー・深い青・グレーで構成された抽象空間｜BGMなし｜環境音なし｜楽器音なし｜人物の口から発する声だけでリズムを作る｜発音と表示文字は完全一致｜文字は字幕ではなく物体として存在｜テンポは速め｜ストーリー性あり｜最後のオチは「無。」\nCUT01｜時間：0.0-1.8秒｜Visuals：生成りホワイトの余白の大きい空間。画面右寄りに黒い人物シルエットの横顔。何もない静かな世界。人物が小さく「ク」と発音した瞬間、黒い「ク」が口から現れ、その後ろに小さなくすんだ赤の正方形が出現する。続けて「パ」と発音すると、赤い「パ」とオーカーの円が生まれる。最後に「ポ」で深い青の長方形が出る。人物は、自分の声で図形と文字が生まれたことに気づき、少しだけ顔を上げる。｜Camera：85mm｜F4｜横顔クローズアップ｜静かなドリーイン。｜Motion：文字と図形は発音ごとに1セットずつ直線的に滑り出る。人物は驚きでわずかに反応。｜Lighting：フラットで柔らかい均一光。｜VFX：紙の質感、ごく薄い押し出し、最小限の影。｜SFX：声のみ「ク」「パ」「ポ」。乾いた口音。｜Emotion：発見。静かなはじまり。｜Transition：最後に出た青い長方形が横へ伸び、次カットの構図線になる。\nCUT02｜時間：1.8-3.6秒｜Visuals：カメラが少し引いて上半身。先ほど生まれた「ク」「パ」「ポ」と図形が人物の周囲に浮いている。人物は面白がって、速めに「ペ・ク・パ・ポ」と発音。1音ごとに「ペ」「ク」「パ」「ポ」が順番どおり出現し、対応する線・正方形・円・長方形と結びつく。出た図形が床や段差のように並び、人物の前に小さな“進める道”ができる。人物は一歩前へ出る。｜Camera：50mm｜F5.6｜3/4ミディアムショット。｜Motion：発音のたびに1文字＋1図形生成。最後に人物がその上へ軽く足を乗せる。｜Lighting：フラット。｜VFX：図形の出現はスライドと90度回転中心。｜SFX：「ペ・ク・パ・ポ」。音ごとに図形が生まれる軽い空気の圧力感。｜Emotion：遊びの開始。｜Transition：人物が踏んだオーカーの円が広がり、真俯瞰構図へ変わる。\nCUT03｜時間：3.6-5.4秒｜Visuals：真俯瞰。生成り背景に、今までの文字と図形が抽象的な配置を作っている。人物が中央を走る。人物は楽しくなって「クパポペ！」と一気に発音。発音に完全同期して「ク」「パ」「ポ」「ペ」が高速で一文字ずつ出現し、それぞれ黒い長方形、赤い正方形、青い円、オーカーの線へ接続される。4つの要素が組み合わさって大きな階段状の構成になる。人物はその上へ駆け上がる。｜Camera：24mm｜真俯瞰｜軽いロール。｜Motion：文字生成→構成変化→人物が走る。テンポは速め。｜Lighting：フラット。｜VFX：余白を保ちつつパララックスで奥行き。｜SFX：「ク・パ・ポ・ペ！」を早めに。｜Emotion：世界が広がる楽しさ。｜Transition：人物が最後の段を蹴った瞬間、重力方向が90度変わる。\nCUT04｜時間：5.4-7.1秒｜Visuals：画面の天地がずれ、人物は横向きの重力で“壁”を歩く状態になる。背景には斜めの赤い長方形と青い円。人物は少し調子に乗って「パクポ！ ペクパ！」と発音。発音した順に「パ」「ク」「ポ」「ペ」「ク」「パ」が一文字ずつ現れ、壁の足場や橋になっていく。ところが最後の「パ」が出た瞬間、人物の影だけが本体から剥がれて別方向へ歩き出す。｜Camera：35mm｜ワイド｜水平を崩したダッチ構図。｜Motion：人物は壁歩き、影は独立して前進。｜Lighting：フラット。｜VFX：影の分離は平面的で抽象的に。｜SFX：「パ・ク・ポ！ ペ・ク・パ！」。速くて気持ちいいが少し不穏。｜Emotion：遊びが少し変になる。｜Transition：影が振り返り、口を開く。\nCUT05｜時間：7.1-9.0秒｜Visuals：人物本体が左、独立した影が右。影が勝手に「ポ・パ・ク・ペ」と発音し、対応する「ポ」「パ」「ク」「ペ」を吐き出す。人物本体は驚いて、自分も「ク・パ・ポ」と応じる。双方の文字が空中で交差し、中央に抽象的な構成体を作る。ここで人物の口だけが顔から外れ、黒い楕円として中央に浮遊し始める。｜Camera：40mm｜左右対称に近い構図から、中央の浮遊口へゆっくり寄る。｜Motion：本体と影のコール＆レスポンス。最後に口が分離して浮く。｜Lighting：均一光。｜VFX：文字と図形の交差、口の平面分離。｜SFX：本体「ク・パ・ポ」／影「ポ・パ・ク・ペ」。左右で定位を変える。｜Emotion：制御不能への入り口。｜Transition：浮遊口が画面中央で大きく開き、その中へカメラが入る。\nCUT06｜時間：9.0-11.2秒｜Visuals：浮遊口の内部が抽象空間になっている。これまで生まれた文字と図形が巨大化し、シュプレマティスム的な構成として空間を満たす。人物本体、影、浮遊口が別々の位置に現れ、それぞれが高速で「クパポペ！ パクポ！ ペクパ！」と発音。発音に合わせて、対応文字が一文字ずつ正確に出現し、画面内の構成がどんどん複雑になる。人物は自分が作った世界に追われるように走る。｜Camera：18mm超広角｜FPV気味に前進、左右へスラローム。｜Motion：文字生成と空間変形が高速化。人物は逃げる。｜Lighting：フラットだが図形配置で緊張感。｜VFX：強めのパララックス、図形の回転、重なり。｜SFX：3者の声が少しずつずれて重なり、中毒的なポリリズムになる。｜Emotion：最高潮。楽しいが危うい。｜Transition：人物が立ち止まり、口を閉じようとする。\nCUT07｜時間：11.2-13.2秒｜Visuals：人物は中央で口を閉じ、止めようとする。しかし影と浮遊口だけがまだ動き、「ク！ パ！ ポ！ ペ！」を勝手に発音し続ける。対応する文字が人物の周囲に押し寄せ、人物は図形と文字の中に埋もれかける。人物が最後に大きく息を吸い込む。すると、画面内のすべての文字・図形・影・浮遊口が中央へ引っ張られ始める。｜Camera：32mm｜ミディアムワイド｜ゆるいドリーイン。｜Motion：前半は他要素だけが暴走、後半で吸引開始。｜Lighting：フラット。｜VFX：吸引前の静止感をしっかり見せる。｜SFX：勝手に鳴る「ク！パ！ポ！ペ！」→大きな吸気音。｜Emotion：主人公の決断。｜Transition：吸い込みをきっかけに全要素が一点へ収束。\nCUT08｜時間：13.2-15.0秒｜Visuals：人物、影、浮遊口、すべての文字と図形が整然と一点に吸い込まれ、生成りホワイトの空間だけが残る。人物すら消える。0.4秒、完全な無音と空白。14.2秒で画面中央に小さく黒文字「無。」が現れる。装飾なし、平面的で静止した文字。15.0秒までそのまま保つ。｜Camera：完全固定。｜Motion：吸収後は一切動かない。｜Lighting：生成りホワイト一色。｜VFX：なしに近い。｜SFX：13.2秒以降完全無音。｜Emotion：余韻。静かなオチ。｜Transition：そのまま終了。\n【音声ルール】発音と表示文字は完全一致。「ク」と言ったら必ず「ク」だけを出す。「パ」→「パ」、「ポ」→「ポ」、「ペ」→「ペ」。複合音は必ず一音ずつ分解して順番どおり表示する。例：「クパポペ」なら「ク」→「パ」→「ポ」→「ペ」の順で出現。ランダム文字の追加は禁止。\n【ストーリー構造】発見（声で世界が生まれる）→遊ぶ（文字と図形で道や足場ができる）→調子に乗る（重力が壊れ、影が独立する）→暴走（口や影が勝手にしゃべる）→回収（主人公が沈黙を選ぶ）→オチ（無。）\n【ビジュアルルール】ポップアートの明るすぎる原色は避ける。余白を広く使い、構図と図形配置で見せる。ベースはシュプレマティスム、異常現象はシュルレアリスム。文字はすべて物体として存在させ、床・橋・階段・空間構成要素になる。コミック的な吹き出しや派手な爆発は使わない。\n【カメラルール】横顔クローズアップ→3/4上半身→真俯瞰→ダッチワイド→対面構図→口内部FPV→ミディアムワイド→完全固定。全カット同じ構図にしない。テンポは速めだが、何が起きているかは明確に見えること。\n【禁止事項】BGM｜楽器音｜環境音｜蛍光色主体｜過剰なコミック演出｜ランダム文字｜文字化け｜フォトリアル背景｜ホラー的身体破壊｜意味のない長文字幕｜発音と文字の不一致。",
    "promptSourceUrl": "https://x.com/su_nagomi/status/2092834537474753005",
    "sourceUrl": "https://x.com/su_nagomi/status/2092834537474753005",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-27T04:40:11.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092834537474753005.mp4",
    "posterUrl": "/posters/x/2092834537474753005.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "意味不明なオノマトペ！\nシュプレマティスム×シュルレアリスム×抽象芸術的モーショングラフィック\n#FlovaAI  Minimax H3のT2Vで動画作成\n\n～プロンプト～\nシュプレマティスム×シュルレアリスム｜抽象芸術的モーショングラフィック｜人物は黒いシルエット1人｜背景は現実空間ではなく、生成りホワイトを基調に、墨黒・くすんだ赤・オーカー・深い青・グレーで構成された抽象空間｜BGMなし｜環境音なし｜楽器音なし｜人物の口から発する声だけでリズムを作る｜発音と表示文字は完全一致｜文字は字幕ではなく物体として存在｜テンポは速め｜ストーリー性あり｜最後のオチは「無。」\nCUT01｜時間：0.0-1.8秒｜Visuals：生成りホワイトの余白の大きい空間。画面右寄りに黒い人物シルエットの横顔。何もない静かな世界。人物が小さく「ク」と発音した瞬間、黒い「ク」が口から現れ、その後ろに小さなくすんだ赤の正方形が出現する。続けて「パ」と発音すると、赤い「パ」とオーカーの円が生まれる。最後に「ポ」で深い青の長方形が出る。人物は、自分の声で図形と文字が生まれたことに気づき、少しだけ顔を上げる。｜Camera：85mm｜F4｜横顔クローズアップ｜静かなドリーイン。｜Motion：文字と図形は発音ごとに1セットずつ直線的に滑り出る。人物は驚きでわずかに反応。｜Lighting：フラットで柔らかい均一光。｜VFX：紙の質感、ごく薄い押し出し、最小限の影。｜SFX：声のみ「ク」「パ」「ポ」。乾いた口音。｜Emotion：発見。静かなはじまり。｜Transition：最後に出た青い長方形が横へ伸び、次カットの構図線になる。\nCUT02｜時間：1.8-3.6秒｜Visuals：カメラが少し引いて上半身。先ほど生まれた「ク」「パ」「ポ」と図形が人物の周囲に浮いている。人物は面白がって、速めに「ペ・ク・パ・ポ」と発音。1音ごとに「ペ」「ク」「パ」「ポ」が順番どおり出現し、対応する線・正方形・円・長方形と結びつく。出た図形が床や段差のように並び、人物の前に小さな“進める道”ができる。人物は一歩前へ出る。｜Camera：50mm｜F5.6｜3/4ミディアムショット。｜Motion：発音のたびに1文字＋1図形生成。最後に人物がその上へ軽く足を乗せる。｜Lighting：フラット。｜VFX：図形の出現はスライドと90度回転中心。｜SFX：「ペ・ク・パ・ポ」。音ごとに図形が生まれる軽い空気の圧力感。｜Emotion：遊びの開始。｜Transition：人物が踏んだオーカーの円が広がり、真俯瞰構図へ変わる。\nCUT03｜時間：3.6-5.4秒｜Visuals：真俯瞰。生成り背景に、今までの文字と図形が抽象的な配置を作っている。人物が中央を走る。人物は楽しくなって「クパポペ！」と一気に発音。発音に完全同期して「ク」「パ」「ポ」「ペ」が高速で一文字ずつ出現し、それぞれ黒い長方形、赤い正方形、青い円、オーカーの線へ接続される。4つの要素が組み合わさって大きな階段状の構成になる。人物はその上へ駆け上がる。｜Camera：24mm｜真俯瞰｜軽いロール。｜Motion：文字生成→構成変化→人物が走る。テンポは速め。｜Lighting：フラット。｜VFX：余白を保ちつつパララックスで奥行き。｜SFX：「ク・パ・ポ・ペ！」を早めに。｜Emotion：世界が広がる楽しさ。｜Transition：人物が最後の段を蹴った瞬間、重力方向が90度変わる。\nCUT04｜時間：5.4-7.1秒｜Visuals：画面の天地がずれ、人物は横向きの重力で“壁”を歩く状態になる。背景には斜めの赤い長方形と青い円。人物は少し調子に乗って「パクポ！ ペクパ！」と発音。発音した順に「パ」「ク」「ポ」「ペ」「ク」「パ」が一文字ずつ現れ、壁の足場や橋になっていく。ところが最後の「パ」が出た瞬間、人物の影だけが本体から剥がれて別方向へ歩き出す。｜Camera：35mm｜ワイド｜水平を崩したダッチ構図。｜Motion：人物は壁歩き、影は独立して前進。｜Lighting：フラット。｜VFX：影の分離は平面的で抽象的に。｜SFX：「パ・ク・ポ！ ペ・ク・パ！」。速くて気持ちいいが少し不穏。｜Emotion：遊びが少し変になる。｜Transition：影が振り返り、口を開く。\nCUT05｜時間：7.1-9.0秒｜Visuals：人物本体が左、独立した影が右。影が勝手に「ポ・パ・ク・ペ」と発音し、対応する「ポ」「パ」「ク」「ペ」を吐き出す。人物本体は驚いて、自分も「ク・パ・ポ」と応じる。双方の文字が空中で交差し、中央に抽象的な構成体を作る。ここで人物の口だけが顔から外れ、黒い楕円として中央に浮遊し始める。｜Camera：40mm｜左右対称に近い構図から、中央の浮遊口へゆっくり寄る。｜Motion：本体と影のコール＆レスポンス。最後に口が分離して浮く。｜Lighting：均一光。｜VFX：文字と図形の交差、口の平面分離。｜SFX：本体「ク・パ・ポ」／影「ポ・パ・ク・ペ」。左右で定位を変える。｜Emotion：制御不能への入り口。｜Transition：浮遊口が画面中央で大きく開き、その中へカメラが入る。\nCUT06｜時間：9.0-11.2秒｜Visuals：浮遊口の内部が抽象空間になっている。これまで生まれた文字と図形が巨大化し、シュプレマティスム的な構成として空間を満たす。人物本体、影、浮遊口が別々の位置に現れ、それぞれが高速で「クパポペ！ パクポ！ ペクパ！」と発音。発音に合わせて、対応文字が一文字ずつ正確に出現し、画面内の構成がどんどん複雑になる。人物は自分が作った世界に追われるように走る。｜Camera：18mm超広角｜FPV気味に前進、左右へスラローム。｜Motion：文字生成と空間変形が高速化。人物は逃げる。｜Lighting：フラットだが図形配置で緊張感。｜VFX：強めのパララックス、図形の回転、重なり。｜SFX：3者の声が少しずつずれて重なり、中毒的なポリリズムになる。｜Emotion：最高潮。楽しいが危うい。｜Transition：人物が立ち止まり、口を閉じようとする。\nCUT07｜時間：11.2-13.2秒｜Visuals：人物は中央で口を閉じ、止めようとする。しかし影と浮遊口だけがまだ動き、「ク！ パ！ ポ！ ペ！」を勝手に発音し続ける。対応する文字が人物の周囲に押し寄せ、人物は図形と文字の中に埋もれかける。人物が最後に大きく息を吸い込む。すると、画面内のすべての文字・図形・影・浮遊口が中央へ引っ張られ始める。｜Camera：32mm｜ミディアムワイド｜ゆるいドリーイン。｜Motion：前半は他要素だけが暴走、後半で吸引開始。｜Lighting：フラット。｜VFX：吸引前の静止感をしっかり見せる。｜SFX：勝手に鳴る「ク！パ！ポ！ペ！」→大きな吸気音。｜Emotion：主人公の決断。｜Transition：吸い込みをきっかけに全要素が一点へ収束。\nCUT08｜時間：13.2-15.0秒｜Visuals：人物、影、浮遊口、すべての文字と図形が整然と一点に吸い込まれ、生成りホワイトの空間だけが残る。人物すら消える。0.4秒、完全な無音と空白。14.2秒で画面中央に小さく黒文字「無。」が現れる。装飾なし、平面的で静止した文字。15.0秒までそのまま保つ。｜Camera：完全固定。｜Motion：吸収後は一切動かない。｜Lighting：生成りホワイト一色。｜VFX：なしに近い。｜SFX：13.2秒以降完全無音。｜Emotion：余韻。静かなオチ。｜Transition：そのまま終了。\n【音声ルール】発音と表示文字は完全一致。「ク」と言ったら必ず「ク」だけを出す。「パ」→「パ」、「ポ」→「ポ」、「ペ」→「ペ」。複合音は必ず一音ずつ分解して順番どおり表示する。例：「クパポペ」なら「ク」→「パ」→「ポ」→「ペ」の順で出現。ランダム文字の追加は禁止。\n【ストーリー構造】発見（声で世界が生まれる）→遊ぶ（文字と図形で道や足場ができる）→調子に乗る（重力が壊れ、影が独立する）→暴走（口や影が勝手にしゃべる）→回収（主人公が沈黙を選ぶ）→オチ（無。）\n【ビジュアルルール】ポップアートの明るすぎる原色は避ける。余白を広く使い、構図と図形配置で見せる。ベースはシュプレマティスム、異常現象はシュルレアリスム。文字はすべて物体として存在させ、床・橋・階段・空間構成要素になる。コミック的な吹き出しや派手な爆発は使わない。\n【カメラルール】横顔クローズアップ→3/4上半身→真俯瞰→ダッチワイド→対面構図→口内部FPV→ミディアムワイド→完全固定。全カット同じ構図にしない。テンポは速めだが、何が起きているかは明確に見えること。\n【禁止事項】BGM｜楽器音｜環境音｜蛍光色主体｜過剰なコミック演出｜ランダム文字｜文字化け｜フォトリアル背景｜ホラー的身体破壊｜意味のない長文字幕｜発音と文字の不一致。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 338,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092830820306436318",
    "title": "甜点角色走进竖屏菜单广告",
    "titleEn": "Dessert Characters Enter a Vertical Menu Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拟人甜点角色在咖啡馆与菜单版式间切换，竖屏构图把角色展示和价格信息组合成轻快广告。 原生视频规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Anthropomorphic dessert characters move between a café and menu layouts, combining character presentation and pricing into a lively vertical ad. Native-video output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mikot27567/status/2092830820306436318",
    "sourceLabel": "X 原帖 · @Mikot27567",
    "author": "mikoto",
    "publishedAt": "2026-08-27T04:25:25.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092830820306436318.mp4",
    "posterUrl": "/posters/x/2092830820306436318.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "デザート擬人化挑戦投稿！！\nminimaxH3で作ってみました。\n\n#SeaArtDailyChallenge　\n@SeaArt_Ai  #minimaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 7,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092827875129086336",
    "title": "一颗豆蔻串起咖喱店故事",
    "titleEn": "One Cardamom Pod Connects a Curry-Shop Story",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "扁平插画从咖喱店旅行到农场相遇，再回到新品菜单，用简洁转场完整讲述一段食材故事。 原生视频规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Flat illustrations travel from a curry shop to a farm encounter and back to a new menu, using simple transitions to tell a complete ingredient story. Native-video output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tanabe_fragm/status/2092827875129086336",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-08-27T04:13:42.000Z",
    "addedAt": "2026-08-27T05:34:46.855Z",
    "mediaUrl": "/media/x-2092827875129086336.mp4",
    "posterUrl": "/posters/x/2092827875129086336.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3を使い、架空のカレー店の新メニュー告知動画を制作🍛\n\n今回は↓のようなストーリー性がある動画にしてみました。\n\n① 高円寺にある架空のカレー店「スパイス堂」の店主の佐藤さんが、宮崎県に旅行にいく\n② 宮崎県でカルダモン農家の田中さん（架空）と出会う\n③ その場でカルダモンを食べ、その美味しさに佐藤さんは感激\n④ 田中さんに分けてもらったカルダモンで新しいメニューを作成\n\nMiniMax H3でいろんな動画を作りましたが、こういう物語性があるシンプルなアニメーションも良いですね👍\n9:16の縦型にすることで、Xだけでなく\n\n・YouTubeショート\n・Instagram\n・TikTok\n\nなどにも投稿可能です😃\n\nプロンプトを応用すれば様々な動画を作れます。\nChatGPT・Gemini・Claudeに渡して、修正内容を指示すれば、すぐ作ってくれるはずです。\n\nぜひご利用ください😌\n\nプロンプト↓\n----------------------\nintegrated_multimodal_description: Create a complete 15-second Japanese flat-illustration story\ncommercial in a vertical 9:16 composition, telling how a Tokyo curry-shop owner brought a\nfarmer's cardamom back to his own shop. Use friendly 2D flat-vector picture-book illustration:\nthick, even, near-black outlines of consistent weight, flat solid color fills, no shading,\nsimple dot eyes and small curved mouths, restrained cutout animation, and a light paper grain\nover the flat color fields. Use a compact palette of warm off-white, deep indigo blue, fresh\nleaf green, golden yellow and vermilion red, with near-black outlines. Background color changes\nmark the chapters of the story. Two characters recur throughout and must never be redesigned:\nthe shop owner is a short-haired man with round glasses and a navy-blue apron, and the farmer is\na man with a straw hat, green rubber boots and a white towel around his neck. All on-screen\nJapanese text must stay sharp, correctly formed and fully readable at smartphone size. No\nphotorealism, no 3D rendering, no gradients, no drop shadows, no browser interface, no\nscreen-recording artifacts, and no player controls.\n[Shot 1] Begin on a warm off-white background. The metal shutter of a small street curry shop\nrolls upward and the short-haired shop owner with round glasses and a navy-blue apron steps out\nand stands in front of it. Then the Japanese shop name \"スパイス堂\" assembles on the signboard\nabove him in near-black type in two clear stages.\n[Shot 2] At 00:01.500, a deep indigo-blue field wipes across the frame. A simplified white\noutline of the Japanese archipelago rises from the bottom edge and settles in the centre of the\nframe, and a small golden-yellow circular marker drops onto the eastern side of the main island\nin the upper-right area of the map, with the caption \"東京\" appearing beside that marker. A\nsmall simple airplane then lifts off from that same \"東京\" marker with its nose pointed down and\nto the left, and travels along one smooth curved path from the upper right of the map to the\nlower left, drawing a white dotted trail progressively behind it as it goes so the completed\nroute is visible from start to finish; as it reaches the southern island at the lower left the\n\"東京\" marker stays in place and shrinks slightly, and a second, larger golden-yellow marker\npops in and pulses once with the caption \"宮崎\" appearing beside it.\n[Shot 3] At 00:03.000, a fresh leaf-green field wipes across the frame and rows of tall green\ncardamom leaves grow upward from the bottom edge until they fill the lower half. The farmer in\nthe straw hat, green boots and white neck towel pops in from frame-left and the shop owner with\nround glasses and a navy apron pops in from frame-right, the two of them facing each other among\nthe leaves. A two-line caption then assembles across the upper area in near-black type in\ntwo clear stages, the first line reading \"カルダモン農家の\" and the second line reading\n\"田中さんと会った\", both lines centred and fully legible.\n[Shot 4] At 00:04.500, staying on leaf green, the farmer crouches down, picks a single small\ngreen cardamom pod from the base of a plant, and holds it out toward the shop owner. The pod\nthen flies to the centre of the frame and scales up large, settling into a clean centred\nclose-up with a thin near-black outline.\n[Shot 5] At 00:06.000, staying on leaf green, the shop owner lifts the green pod to his mouth\nand bites it, his eyebrows rising. Three white curling aroma swirls then rise from the pod one\nafter another and spread outward across the upper half of the frame.\n[Shot 6] At 00:07.500, golden yellow bursts across the entire frame and fills it edge to edge.\nThe shop owner throws both arms up and leans back with a wide open mouth and small radiating\nlines around his head. The caption \"美味い！\" then punches in beside him one character at a\ntime in large near-black type.\n[Shot 7] At 00:09.000, staying on golden yellow, the shop owner bows deeply toward the farmer\nwith both hands pressed together while the farmer smiles and holds out a small tied cloth bag\nfull of green pods. The shop owner then takes the bag in both hands and a few small sparkle\nmarks pop around it.\n[Shot 8] At 00:10.500, a vermilion-red field wipes across the frame. Golden-brown curry pours\nfrom a small ladle onto a warm off-white plate of rice at the centre of the frame, and two\ncurling steam lines rise above it in two stages. Three green cardamom pods then drop in from the\ntop of the frame one at a time and land on the curry.\n[Shot 9] At 00:12.000, a warm off-white field wipes across the frame. The plate of curry slides\nto the centre while a small bowl of green salad slides in from frame-left and a tall glass of\nwhite lassi slides in from frame-right, the three of them settling into a neat row. A two-line\ncaption then assembles above them in near-black type in two clear stages: a smaller first line\nreading \"期間限定\" and a larger second line reading \"田中さんのカルダモンカレー\".\n[Shot 10] At 00:13.500, staying on warm off-white, the row of three dishes and the caption above\nthem all shift upward toward the top of the frame, and the price \"1500円\" punches into the open\nspace below in very large near-black type in two stages, with a short golden-yellow underline\nbar snapping in beneath it. The smaller caption \"サラダ・ラッシー付\" then slides in directly\nunder the price in one clean step, and the whole composition holds completely still until\nexactly 15.000 seconds.\n\noverall_soundscape: A metal shutter rattles upward over quiet street ambience, followed by a\nrising airy whoosh as the airplane lifts off, a soft continuous hiss while the dotted trail\ndraws itself, and a light tick with a small chime when it arrives at the southern marker. Leaves\nrustle as the field grows and a small crisp snap marks the pod being picked, with a soft breathy\npuff under the rising aroma swirls. Bright paper-like pops mark each caption and each sparkle,\nand the curry section carries a thick pouring glug, a light steam hiss and a clean ceramic clink\nas the plate, bowl and glass settle into a row. One firm low impact lands as the price snaps\ninto place. No dialogue.\n\nnon_diegetic_music: Generate an original 15-second warm Japanese commercial jingle using shinobue\nbamboo flute, softly fingerpicked acoustic guitar, light hand percussion and a small shaker at a\nmoderate walking tempo. Begin calm and sparse over the off-white shop opening, add a steady\ntravelling rhythm over the indigo flight, lighten and brighten across the green field, reach its\nbrightest and fullest point on the golden-yellow reaction, warm into a rounder melody over the\nred curry section, and land on one clean final chord as the price appears, holding it to exactly\n15.000 seconds.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 185,
      "snapshotAt": "2026-08-27T05:34:46.855Z"
    },
    "approvedAt": "2026-08-27T05:34:46.855Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092826910422073517",
    "title": "牧羊少年连续穿越四段冒险",
    "titleEn": "A Young Shepherd Crosses Four Linked Adventures",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黏土动画风格的牧羊少年与羊群穿过草原、山谷和不同事件，四段文本生成镜头被串成一分钟故事。 原生视频规格为 60 秒 · 960×548。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A clay-style young shepherd and his flock move through plains, valleys, and changing events as four text-generated shots become a one-minute story. Native-video output: 60s · 960×548. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BradleyYoungjr/status/2092826910422073517",
    "sourceLabel": "X 原帖 · @BradleyYoungjr",
    "author": "Brad",
    "publishedAt": "2026-08-27T04:09:52.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092826910422073517.mp4",
    "posterUrl": "/posters/x/2092826910422073517.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "960×548",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "nature",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is pretty crazy. This video is four separate 15s one shot generations purely driven from text prompts. The consistency despite not having image references is pretty good. Lots of weirdness going on like voice changes, characters looking a bit different, but IMO very impressive since it was all text prompting. This took about 5-10 minutes (including writing the prompts with codex) and a little under $2.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 24,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092825704312480181",
    "title": "月蚀舞台上的三味线演奏",
    "titleEn": "A Shamisen Performance on an Eclipse Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "和风角色在月蚀、红叶与墨色舞台中演奏三味线，环形转场和书法标题营造戏剧化开场。 原生视频规格为 15 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A Japanese-inspired character plays shamisen amid an eclipse, red leaves, and an ink-dark stage, with circular transitions and calligraphic titles creating a dramatic opener. Native-video output: 15s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SushiNFTart/status/2092825704312480181",
    "sourceLabel": "X 原帖 · @SushiNFTart",
    "author": "🍣お寿司職人｜AIゲーム・アニメ",
    "publishedAt": "2026-08-27T04:05:05.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092825704312480181.mp4",
    "posterUrl": "/posters/x/2092825704312480181.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max\n三味線を弾かせてみたテスト\n\n#月蝕綺譚 弁天さまとロゴをお借りしました🪕🍁",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 236,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092823518702653864",
    "title": "夏日刨冰引发的意外反应",
    "titleEn": "A Shaved-Ice Treat Triggers a Summer Surprise",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少女在乡间小店外品尝蓝色刨冰，表情从享受转为惊讶，手持镜头保持轻松夏日气息。 原生视频规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A young woman tastes blue shaved ice outside a rural shop, shifting from delight to surprise while handheld framing keeps the mood light and summery. Native-video output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/safa_dayo/status/2092823518702653864",
    "sourceLabel": "X 原帖 · @safa_dayo",
    "author": "サファ【AIイラスト多め】",
    "publishedAt": "2026-08-27T03:56:24.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092823518702653864.mp4",
    "posterUrl": "/posters/x/2092823518702653864.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "かき氷が...😰\n\nHailuo AIのMiniMax H3環境で作成",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 114,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092822345446781014",
    "title": "静态角色照片变成自然对白",
    "titleEn": "A Static Character Photo Becomes Natural Dialogue",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "银发角色从静态造型中获得连续表情、手势和韩语对白，竖屏构图保持服装与面部特征。 原生视频规格为 10 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired character gains continuous expressions, gestures, and Korean dialogue from a static costume photo while the vertical framing preserves facial and outfit details. Native-video output: 10s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lucasyao99261/status/2092822345446781014",
    "sourceLabel": "X 原帖 · @lucasyao99261",
    "author": "Lucasyao",
    "publishedAt": "2026-08-27T03:51:44.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092822345446781014.mp4",
    "posterUrl": "/posters/x/2092822345446781014.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "fashion"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "제목：민한나의 Lucy가 움직이기 시작했다 ⚡  \n소개：@hannamin914의 정지 코스 사진을 MiniMax-H3로 영화급 모션 + 자연스러운 한국어 대사로.  \n태그：#MiniMaxH3 #AIvideo #CosplayAI #MotionDesign #AIart\nprompts below 👇",
    "engagement": {
      "replies": 8,
      "reposts": 0,
      "likes": 0,
      "views": 69,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092821327136657535",
    "title": "粉色巨龙飞越樱花山谷",
    "titleEn": "A Pink Dragon Flies Across a Cherry-Blossom Valley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "粉色巨龙从山谷河面掠过樱花树与瀑布，竖屏镜头保持翅膀、长尾和环境的连续运动。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink dragon glides above a river, cherry blossoms, and waterfalls as the vertical shot sustains continuous wing, tail, and environment motion. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SORAY_AI/status/2092821327136657535",
    "sourceLabel": "X 原帖 · @SORAY_AI",
    "author": "SORAY-AI",
    "publishedAt": "2026-08-27T03:47:41.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092821327136657535.mp4",
    "posterUrl": "/posters/x/2092821327136657535.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Another cute video of Saadyna! This time with Minimax H3 on @PixVerse #PixverseCPP\n\nI really love how she looks. I need to make more videos with her. I have a bunch of dragons I made and they all need videos!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 82,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092818051075133545",
    "title": "白熊品牌套件变成动态图形",
    "titleEn": "A Polar-Bear Brand Kit Becomes Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "白熊品牌视觉从静态海报展开为摄影机、红色标记和日文排版，快速完成一段完整的品牌动态图形。 原生视频规格为 10 秒 · 1080×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A polar-bear brand kit expands from a static poster into camera motifs, red markers, and Japanese typography, forming a complete motion-graphics spot. Native-video output: 10s · 1080×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2092818051075133545",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-08-27T03:34:40.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092818051075133545.mp4",
    "posterUrl": "/posters/x/2092818051075133545.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1080×960",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ブランドキット ×\nAIモーショングラフィックス\n\n① ミヤマさんのプロンプトでブランドキット画像を作成 ② ついでに MiniMax H3 でカッコよく動画化！\n\n ◀ 左：Xバナー参照のブランドキット画像\n ▶ 右：MiniMax H3\n\n……誰やこのしろくま🤣\n\nprompt👇",
    "engagement": {
      "replies": 2,
      "reposts": 5,
      "likes": 63,
      "views": 8427,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092818026555199663",
    "title": "南瓜装置铺满当代艺术馆",
    "titleEn": "Pumpkin Installations Fill a Contemporary Gallery",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头穿过极简展厅、重复南瓜装置和红色标题画面，以连续运动重组波点艺术语言。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera passes through a minimal gallery, repeating pumpkin installations, and a red title frame, reworking a polka-dot art language through continuous motion. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/vkuoo/status/2092818026555199663",
    "sourceLabel": "X 原帖 · @vkuoo",
    "author": "vkuoo",
    "publishedAt": "2026-08-27T03:34:34.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092818026555199663.mp4",
    "posterUrl": "/posters/x/2092818026555199663.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Yayoi Kusama\n\nBy MiniMax H3 on @krea_ai .",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092650078854852881",
    "title": "地狱俱乐部里的爵士乐演出",
    "titleEn": "A Jazz Performance Inside an Infernal Club",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "猩红角色在昏暗俱乐部中演奏，火光、舞台乐手和近距离表演镜头组成一支近一分钟的地狱乐队 MV。 原生视频规格为 55 秒 · 1248×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A crimson figure performs in a dark club as firelight, band members, and close stage shots build a nearly minute-long infernal jazz video. Native-video output: 55s · 1248×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2092650078854852881",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-26T16:27:13.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092650078854852881.mp4",
    "posterUrl": "/posters/x/2092650078854852881.jpg",
    "duration": 55,
    "aspectRatio": "landscape",
    "resolution": "1248×704",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Finally made what I've always wanted to make: a cool jazz music video set in hell.\n\nNow think about this --- This clip didn't exist yesterday. Now it does, 100% minted from thin air by my PC. No powerful workstation. No video editing skills. Just a PC with 8GB+ VRAM. And all you have to do is ask your PC.\n\n> Model: Minimax H3\n> Resolution: 720p\n> Reference Audio: \"I fall in love too easily\" (Chet Baker)\n> Prompt: A one-shot long take video of a bloody man walking toward the camera and walking past hell, singing like everything is normal. the contrast of the singing from all the chaotic things happening in the background. The entire video is the man walking towards the camera as the camera  moves through hell, and the camera should keep shooting the same person throughout the entire video.\n\nGenerated in Maestro: Multi-window sequence (Auto-plan) with 6 windows.",
    "engagement": {
      "replies": 9,
      "reposts": 5,
      "likes": 91,
      "views": 10087,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092764009896534293",
    "title": "港口上空的危险走绳",
    "titleEn": "A Dangerous Tightrope Walk Above the Harbor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子在停泊货轮前的粗绳上保持平衡，海面、船体和缓慢推进的镜头强化高空惊险感。 原生视频规格为 15 秒 · 2944×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman balances on a thick rope in front of a docked cargo ship, with water, hull scale, and a slow camera move heightening the sense of danger. Native-video output: 15s · 2944×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/opener_ai/status/2092764009896534293",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-26T23:59:56.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092764009896534293.mp4",
    "posterUrl": "/posters/x/2092764009896534293.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Walking on a tightrope these days… one wrong move and it’s over 🫨\n\nMinimax H3 @Hailuo_AI \n\n#minimaxh3 #grokimagine",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 134,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2092570843985658163",
    "title": "深夜地铁里的半透明晃动",
    "titleEn": "Translucent Motion Inside a Late-Night Subway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣女子独坐空荡地铁车厢，窗外高速掠过的灯光与人物周围的半透明晃动制造不安氛围。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in black sits alone in an empty subway car as rushing lights outside and translucent motion around her create an uneasy late-night mood. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/opener_ai/status/2092570843985658163",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-08-26T11:12:21.000Z",
    "addedAt": "2026-08-27T05:34:46.856Z",
    "mediaUrl": "/media/x-2092570843985658163.mp4",
    "posterUrl": "/posters/x/2092570843985658163.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The translucent, fluid wobbling actually looks super clean and realistic on MiniMax Hailuo H3 @Hailuo_AI \n\n#minimaxH3",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 12,
      "views": 363,
      "snapshotAt": "2026-08-27T05:34:46.856Z"
    },
    "approvedAt": "2026-08-27T05:34:46.856Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093137271067427119",
    "title": "窗边手作时光的细腻循环",
    "titleEn": "A Quiet Crafting Loop by the Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画角色在雨窗旁整理布料、端杯饮水，细小手部动作与柔和光线构成安静的生活片段。 原生视频规格为 15 秒 · 1024×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated character works with fabric and takes a drink beside a rain-streaked window, forming a quiet slice of life through small hand movements and soft light. Native-video output: 15s · 1024×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/imaginaldomain/status/2093137271067427119",
    "sourceLabel": "X 原帖 · @imaginaldomain",
    "author": "空想圏域 / Imaginal Domain｜編纂室",
    "publishedAt": "2026-08-28T00:43:08.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093137271067427119.mp4",
    "posterUrl": "/posters/x/2093137271067427119.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3 完全に理解しました。",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 8,
      "views": 90,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093133440359317751",
    "title": "白纸上的少女逐笔成形",
    "titleEn": "An Anime Girl Takes Shape Stroke by Stroke",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "画笔从空白画布勾出少女姿态，再逐步补齐线稿、颜色和盆栽，呈现完整绘画延时过程。 原生视频规格为 15 秒 · 1920×1070。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A stylus builds a girl from a blank canvas, adding line art, color, and a potted plant through a complete drawing-timelapse sequence. Native-video output: 15s · 1920×1070. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "How the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] 2D anime digital illustration, a high-quality digital drawing recorded as a timelapse, with the artist's hand holding a stylus pen visibly drawing on the canvas. The video starts from a white digital canvas and draws the basic forms and shapes first, starting from the general (big) outline sketch of the pose, then adding line after line to add new details on the face, hair, and clothes until a complete lineart sketch for the character is done. Then the coloring phase begins: flat base colors are filled evenly within the lineart, soft shading and highlights are layered on the face, hair, and clothes, and final finishing touches are added until the completed artwork matches <Picture 1> exactly.\n\noverall_soundscape: The soft scratch of the stylus on the tablet screen, light desk ambience.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/PhotogenicWeekE/status/2093133529785983467",
    "sourceUrl": "https://x.com/PhotogenicWeekE/status/2093133440359317751",
    "sourceLabel": "X 原帖 · @PhotogenicWeekE",
    "author": "Photogenic Weekend",
    "publishedAt": "2026-08-28T00:27:55.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093133440359317751.mp4",
    "posterUrl": "/posters/x/2093133440359317751.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1070",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "実は俺、絵も描ける(笑) < 嘘、全くダメw MiniMax H3 f2v。Prompt、短いからコメントに",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 20,
      "views": 1063,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093130319033663506",
    "title": "红月下的暗夜角色片头",
    "titleEn": "A Dark Character Opener Under a Red Moon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑红角色在月蚀、蝙蝠剪影和书法标题间登场，连续特写组成十五秒动画片头。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-and-red character appears amid an eclipse, bat silhouettes, and calligraphic titles in a fifteen-second animated opener. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SushiNFTart/status/2093130319033663506",
    "sourceLabel": "X 原帖 · @SushiNFTart",
    "author": "🍣お寿司職人｜AIゲーム・アニメ",
    "publishedAt": "2026-08-28T00:15:31.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093130319033663506.mp4",
    "posterUrl": "/posters/x/2093130319033663506.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#月蝕綺譚 \nサスラ・トバリのアニメOP風ショートアニメ🦇\nMiniMax H3 Maxで作ってみました🎬\n（ロゴとキャラをお借りした非公式二次創作です）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 1742,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093130098761433420",
    "title": "夏日校园疾驰的动画片头",
    "titleEn": "A High-Speed Summer School Anime Opener",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "夏装少女从校园奔跑、骑车一路切入摩托疾驰，动能排版和高速转场维持音乐片头节奏。 原生视频规格为 17 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Girls in summer uniforms run and cycle from school into a racing-motorcycle sequence as kinetic type and rapid transitions sustain an opening-theme pace. Native-video output: 17s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "CHARACTER STYLE — SUMMER UNIFORMS:\n\nAll high-school girls wear modern, cute, contemporary SUMMER school uniforms.\n\nThe uniforms must clearly look like Japanese summer school uniforms.\n\nUse lightweight short-sleeved blouses, short-sleeved shirts, lightweight summer knit vests, breathable cardigans used only very lightly, cute ribbons, slim neckties, pleated skirts, subtle plaid skirts, and modern school bags.\n\nThe girls should look cool, fresh, youthful, and comfortable in summer.\n\nUse refreshing summer colors such as white, light blue, pale blue, soft navy, cream, pale pink, mint, and light gray.\n\nPrefer white and light-colored tops combined with cool-toned skirts.\n\nThe uniforms should feel light and airy.\n\nShow natural movement of short sleeves, ribbons, loose hair, pleated skirts, and lightweight fabric.\n\nHair and clothing should react strongly to wind and movement.\n\nThe overall impression should be:\n\ncool summer morning,\nfresh breeze,\nyouthful Japanese summer,\nbright sunlight,\nlightweight clothing,\nrefreshing atmosphere,\nclean contemporary anime aesthetic.\n\nAvoid heavy winter uniforms.\n\nAvoid thick jackets.\n\nAvoid long-sleeved blazers.\n\nAvoid heavy sweaters.\n\nAvoid scarves.\n\nAvoid winter fashion.\n\nAvoid dark, heavy clothing dominating the frame.\n\nThe summer uniforms should remain stylish and cute while clearly being school-appropriate.\n\nEach girl can have a slightly different summer uniform variation.\n\nTYPOGRAPHY — EXTREMELY HIGH DENSITY:\n\nTypography is one of the PRIMARY visual elements of the entire video.\n\nIncrease the amount of readable typography to approximately TWICE the previous version.\n\nThe screen should frequently contain many readable words simultaneously.\n\nHowever, every word must remain visually recognizable.\n\nNever generate random letters, fake words, pseudo-English, corrupted text, meaningless symbols, or unreadable paragraphs.\n\nUse many short readable words and phrases.\n\nUse combinations of large, medium, and small typography simultaneously.\n\nExamples:\n\n\"MUSIC\"\n\"LIVE\"\n\"START\"\n\"ARE YOU READY?\"\n\"COMING SOON\"\n\"LET'S GO\"\n\"RUN\"\n\"JUMP\"\n\"DANCE\"\n\"PLAY\"\n\"ROCK\"\n\"GO!\"\n\"HEY!\"\n\"FEEL THE MUSIC\"\n\"KEEP MOVING\"\n\"NEVER STOP\"\n\"RIGHT NOW\"\n\"ONE MORE TIME\"\n\"READY?\"\n\"GO GO GO\"\n\"MAKE SOME NOISE\"\n\"TONIGHT\"\n\"GOOD VIBES\"\n\"SUPER!\"\n\"YOUTH\"\n\"SUMMER\"\n\"FRIENDS\"\n\"DREAM\"\n\"LOVE\"\n\"SMILE\"\n\"START NOW\"\n\"SUMMER DAYS\"\n\"SUMMER VIBES\"\n\"FEEL FREE\"\n\"GOOD DAY\"\n\"KEEP GOING\"\n\"JUST MOVE\"\n\"TURN IT UP\"\n\"ALL RIGHT\"\n\"LET'S RIDE\"\n\"FASTER\"\n\"FULL SPEED\"\n\"RIDE ON\"\n\"TAKE OFF\"\n\"GO FASTER\"\n\"ONE MORE\"\n\"RIGHT NOW\"\n\"THIS MOMENT\"\n\"FEEL ALIVE\"\n\nUse many of these words throughout the video.\n\nDo not display all words at once.\n\nInstead, distribute them dynamically across foreground, middle-ground, and background.\n\nUse approximately twice as many typography elements as the previous version.\n\nAt many moments, 5–15 separate readable typography elements may exist simultaneously at different depths and sizes.\n\nSome can be very small but still clearly readable.\n\nLarge typography should occasionally dominate the composition.\n\nSmall typography should create visual richness around the characters.\n\nTYPOGRAPHY PLACEMENT:\n\nPlace readable text:\n\nbehind characters,\nin front of characters,\nabove characters,\nbelow characters,\nalong motion trails,\ninside manga panels,\non school papers,\non signs,\non walls,\nalong roads,\naround motorcycles,\naround musical instruments,\ninside circular graphics,\nbetween characters,\nnear hands and feet,\nfollowing hair movement,\nfollowing skirts and ribbons,\ninside particles,\nalong camera paths,\nand extremely close to the camera.\n\nTypography should exist in 3D space.\n\nSome words should pass extremely close to the camera.\n\nSome words should remain far in the background.\n\nCreate strong typographic parallax.\n\nTYPOGRAPHY MOTION:\n\nEvery typography element should have intentional movement.\n\nUse:\n\nhorizontal sliding,\nvertical sliding,\ndiagonal movement,\nrotation,\nscaling,\nstretching,\ncompression,\nletter-by-letter appearance,\nmotion trails,\ncircular orbiting,\ndepth movement,\nforeground passes,\nbackground passes,\ntypographic wipes,\ntypographic explosions,\ntypographic transitions,\ntypographic collisions,\ntypographic stacking,\ntypographic scattering.\n\nCharacter movement should directly influence the typography.\n\nA girl's running motion creates:\n\n\"RUN\"\n\"FASTER\"\n\"KEEP GOING\"\n\nA jump creates:\n\n\"JUMP\"\n\"GO!\"\n\"TAKE OFF\"\n\nA dance creates:\n\n\"DANCE\"\n\"FEEL THE MUSIC\"\n\"TURN IT UP\"\n\nA guitar performance creates:\n\n\"MUSIC\"\n\"PLAY\"\n\"ROCK\"\n\"LIVE\"\n\nA motorcycle creates:\n\n\"RIDE\"\n\"FULL SPEED\"\n\"FASTER\"\n\"GO FASTER\"\n\nWind moving through the girl's summer uniform creates:\n\n\"SUMMER\"\n\"SUMMER DAYS\"\n\"FEEL FREE\"\n\nDo not treat typography as subtitles.\n\nTypography is part of the physical motion-graphics environment.\n\nSUMMER ATMOSPHERE:\n\nCreate a subtle refreshing summer atmosphere throughout the video.\n\nUse bright natural light, clear skies, soft sunlight, gentle wind, light shadows, and a clean colorful anime palette.\n\nThe girls' hair, ribbons, sleeves, skirts, and loose clothing should frequently respond to wind.\n\nUse occasional subtle summer elements:\n\nsunlight,\nblue sky,\nsummer breeze,\nfloating paper,\nsmall leaves,\nlight particles,\nreflections,\nbright school corridors,\nopen windows,\nsummer streets.\n\nDo not make the video feel like a generic beach vacation.\n\nThe primary environment remains a dynamic Japanese youth / school / music-video world.\n\nMOTORCYCLE CHANGE:\n\nReplace all cruiser / American-style motorcycles with modern Japanese-style full-fairing sport motorcycles.\n\nThe motorcycle must be a sleek Japanese-style racer replica / supersport motorcycle.\n\nUse:\n\nfull aerodynamic fairings,\naggressive front cowl,\nsport riding position,\nclip-on handlebars,\nrear-set footpegs,\nlarge front brake discs,\nsport tires,\nhigh-performance wheels,\nsharp aerodynamic bodywork,\ncompact tail section,\nracing-inspired graphics.\n\nThe rider uses a clearly forward-leaning sport riding posture.\n\nThe motorcycle must visually resemble a modern racer replica rather than a cruiser.\n\nIMPORTANT:\n\nNO American cruiser motorcycle.\n\nNO chopper.\n\nNO long low cruiser.\n\nNO large relaxed handlebars.\n\nNO chrome-heavy cruiser styling.\n\nNO cruiser riding posture.\n\nNO classic Harley-style silhouette.\n\nInstead:\n\nJapanese sportbike.\n\nRacer replica.\n\nFull fairing.\n\nAggressive aerodynamic silhouette.\n\nForward-leaning riding position.\n\nHigh-speed cornering.\n\nMOTORCYCLE ACTION:\n\nThe girl actively controls the motorcycle.\n\nShe accelerates hard.\n\nShe shifts her body.\n\nShe leans deeply into a corner.\n\nThe motorcycle banks dramatically.\n\nThe tires maintain strong contact with the road.\n\nThe camera tracks beside the motorcycle at high speed.\n\nThen the camera moves slightly ahead of the motorcycle.\n\nThe camera swings around the front.\n\nThe front wheel and brake disc fill the frame.\n\nThe spinning wheel transforms into a circular graphic.\n\nThe brake disc becomes a rotating typography ring.\n\nWords such as:\n\n\"FULL SPEED\"\n\"FASTER\"\n\"RIDE\"\n\"GO FASTER\"\n\nrotate around the wheel.\n\nThe motorcycle accelerates again.\n\nIts motion trail becomes enormous kinetic typography.\n\n\"GO!\"\n\nstretches across the screen.\n\nThen:\n\n\"TAKE OFF\"\n\nbecomes the transition into the next scene.\n\nThe motorcycle sequence should feel like a Japanese racing-anime music video rather than a motorcycle advertisement.\n\nCAMERA DIRECTION:\n\nContinue using extremely varied camera movement.\n\nPrioritize:\n\nleft-to-right tracking,\nright-to-left tracking,\ndiagonal tracking,\nside tracking,\ncamera moving beside characters,\ncamera moving between characters,\ncamera overtaking the sport motorcycle,\ncamera moving around the front wheel,\ncamera circling musicians,\ncamera rotating around dancers,\ncamera dropping vertically,\ncamera rising vertically,\nDutch angles,\nrapid push-ins,\nrapid pull-backs,\nwhip pans,\nrotational transitions,\nextreme close passes,\nforeground object wipes,\nperspective distortion,\nfocus changes,\nstrong parallax,\ntypography passing the lens.\n\nDo NOT rely primarily on forward and backward camera movement.\n\nVISUAL DENSITY:\n\nIncrease the overall visual information density.\n\nCharacters, summer uniforms, objects, manga panels, typography, motion trails, particles, musical instruments, school environments, bicycles, sport motorcycles, school bags, papers, signs, and graphic elements should coexist.\n\nNever leave large empty areas.\n\nAt many moments, several independent visual events should happen simultaneously.\n\nFor example:\n\na girl running in the foreground,\n\nanother girl jumping in the middle-ground,\n\na sport motorcycle passing in the background,\n\nlarge typography crossing diagonally,\n\nsmall typography floating behind the characters,\n\nmotion trails connecting all elements,\n\nand manga panels transforming around them.\n\nThe scene should feel extremely rich but still intentionally composed.\n\nThe complexity should come from meaningful simultaneous actions and readable typography.\n\nFINAL VISUAL TARGET:\n\nModern Japanese anime music-video opening.\n\nCute contemporary Japanese summer uniforms.\n\nRefreshing summer atmosphere.\n\nBeautiful energetic high-school girls.\n\nExtremely dynamic action.\n\nHigh-speed Japanese sport motorcycle.\n\nKinetic typography.\n\nApproximately TWICE the previous typography density.\n\nMany readable words.\n\nTypography integrated into character movement.\n\nTypography integrated into camera movement.\n\nTypography integrated into environmental transitions.\n\nFast editing.\n\nContinuous transformation.\n\nStrong parallax.\n\nStrong lateral camera movement.\n\nStrong rotational camera movement.\n\nHigh information density.\n\nStylish contemporary Japanese graphic design.\n\nMid-2000s Japanese music-video energy combined with modern anime motion graphics and contemporary kinetic typography.\n\nMOST IMPORTANT:\n\nSUMMER UNIFORMS.\n\nREFRESHING SUMMER ATMOSPHERE.\n\nCUTE MODERN JAPANESE HIGH-SCHOOL FASHION.\n\nTWICE THE PREVIOUS AMOUNT OF READABLE TYPOGRAPHY.\n\nALL TEXT MUST BE READABLE.\n\nSPORT RACING MOTORCYCLE, NOT AMERICAN CRUISER.\n\nSHOW GIRLS DOING THINGS.\n\nSHOW TYPOGRAPHY MOVING.\n\nSHOW THE ENTIRE BODY OF THE CHARACTERS WHEN POSSIBLE.\n\nNEVER LET THE VIDEO BECOME A COLLECTION OF STATIC PORTRAITS.",
    "promptSourceUrl": "https://x.com/goldwing_zoome/status/2093130098761433420",
    "sourceUrl": "https://x.com/goldwing_zoome/status/2093130098761433420",
    "sourceLabel": "X 原帖 · @goldwing_zoome",
    "author": "goldwing",
    "publishedAt": "2026-08-28T00:14:38.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093130098761433420.mp4",
    "posterUrl": "/posters/x/2093130098761433420.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようバイク女子動画。\n#MiniMaxH3\n\nプロンプトは以下です。\nCHARACTER STYLE — SUMMER UNIFORMS:\n\nAll high-school girls wear modern, cute, contemporary SUMMER school uniforms.\n\nThe uniforms must clearly look like Japanese summer school uniforms.\n\nUse lightweight short-sleeved blouses, short-sleeved shirts, lightweight summer knit vests, breathable cardigans used only very lightly, cute ribbons, slim neckties, pleated skirts, subtle plaid skirts, and modern school bags.\n\nThe girls should look cool, fresh, youthful, and comfortable in summer.\n\nUse refreshing summer colors such as white, light blue, pale blue, soft navy, cream, pale pink, mint, and light gray.\n\nPrefer white and light-colored tops combined with cool-toned skirts.\n\nThe uniforms should feel light and airy.\n\nShow natural movement of short sleeves, ribbons, loose hair, pleated skirts, and lightweight fabric.\n\nHair and clothing should react strongly to wind and movement.\n\nThe overall impression should be:\n\ncool summer morning,\nfresh breeze,\nyouthful Japanese summer,\nbright sunlight,\nlightweight clothing,\nrefreshing atmosphere,\nclean contemporary anime aesthetic.\n\nAvoid heavy winter uniforms.\n\nAvoid thick jackets.\n\nAvoid long-sleeved blazers.\n\nAvoid heavy sweaters.\n\nAvoid scarves.\n\nAvoid winter fashion.\n\nAvoid dark, heavy clothing dominating the frame.\n\nThe summer uniforms should remain stylish and cute while clearly being school-appropriate.\n\nEach girl can have a slightly different summer uniform variation.\n\nTYPOGRAPHY — EXTREMELY HIGH DENSITY:\n\nTypography is one of the PRIMARY visual elements of the entire video.\n\nIncrease the amount of readable typography to approximately TWICE the previous version.\n\nThe screen should frequently contain many readable words simultaneously.\n\nHowever, every word must remain visually recognizable.\n\nNever generate random letters, fake words, pseudo-English, corrupted text, meaningless symbols, or unreadable paragraphs.\n\nUse many short readable words and phrases.\n\nUse combinations of large, medium, and small typography simultaneously.\n\nExamples:\n\n\"MUSIC\"\n\"LIVE\"\n\"START\"\n\"ARE YOU READY?\"\n\"COMING SOON\"\n\"LET'S GO\"\n\"RUN\"\n\"JUMP\"\n\"DANCE\"\n\"PLAY\"\n\"ROCK\"\n\"GO!\"\n\"HEY!\"\n\"FEEL THE MUSIC\"\n\"KEEP MOVING\"\n\"NEVER STOP\"\n\"RIGHT NOW\"\n\"ONE MORE TIME\"\n\"READY?\"\n\"GO GO GO\"\n\"MAKE SOME NOISE\"\n\"TONIGHT\"\n\"GOOD VIBES\"\n\"SUPER!\"\n\"YOUTH\"\n\"SUMMER\"\n\"FRIENDS\"\n\"DREAM\"\n\"LOVE\"\n\"SMILE\"\n\"START NOW\"\n\"SUMMER DAYS\"\n\"SUMMER VIBES\"\n\"FEEL FREE\"\n\"GOOD DAY\"\n\"KEEP GOING\"\n\"JUST MOVE\"\n\"TURN IT UP\"\n\"ALL RIGHT\"\n\"LET'S RIDE\"\n\"FASTER\"\n\"FULL SPEED\"\n\"RIDE ON\"\n\"TAKE OFF\"\n\"GO FASTER\"\n\"ONE MORE\"\n\"RIGHT NOW\"\n\"THIS MOMENT\"\n\"FEEL ALIVE\"\n\nUse many of these words throughout the video.\n\nDo not display all words at once.\n\nInstead, distribute them dynamically across foreground, middle-ground, and background.\n\nUse approximately twice as many typography elements as the previous version.\n\nAt many moments, 5–15 separate readable typography elements may exist simultaneously at different depths and sizes.\n\nSome can be very small but still clearly readable.\n\nLarge typography should occasionally dominate the composition.\n\nSmall typography should create visual richness around the characters.\n\nTYPOGRAPHY PLACEMENT:\n\nPlace readable text:\n\nbehind characters,\nin front of characters,\nabove characters,\nbelow characters,\nalong motion trails,\ninside manga panels,\non school papers,\non signs,\non walls,\nalong roads,\naround motorcycles,\naround musical instruments,\ninside circular graphics,\nbetween characters,\nnear hands and feet,\nfollowing hair movement,\nfollowing skirts and ribbons,\ninside particles,\nalong camera paths,\nand extremely close to the camera.\n\nTypography should exist in 3D space.\n\nSome words should pass extremely close to the camera.\n\nSome words should remain far in the background.\n\nCreate strong typographic parallax.\n\nTYPOGRAPHY MOTION:\n\nEvery typography element should have intentional movement.\n\nUse:\n\nhorizontal sliding,\nvertical sliding,\ndiagonal movement,\nrotation,\nscaling,\nstretching,\ncompression,\nletter-by-letter appearance,\nmotion trails,\ncircular orbiting,\ndepth movement,\nforeground passes,\nbackground passes,\ntypographic wipes,\ntypographic explosions,\ntypographic transitions,\ntypographic collisions,\ntypographic stacking,\ntypographic scattering.\n\nCharacter movement should directly influence the typography.\n\nA girl's running motion creates:\n\n\"RUN\"\n\"FASTER\"\n\"KEEP GOING\"\n\nA jump creates:\n\n\"JUMP\"\n\"GO!\"\n\"TAKE OFF\"\n\nA dance creates:\n\n\"DANCE\"\n\"FEEL THE MUSIC\"\n\"TURN IT UP\"\n\nA guitar performance creates:\n\n\"MUSIC\"\n\"PLAY\"\n\"ROCK\"\n\"LIVE\"\n\nA motorcycle creates:\n\n\"RIDE\"\n\"FULL SPEED\"\n\"FASTER\"\n\"GO FASTER\"\n\nWind moving through the girl's summer uniform creates:\n\n\"SUMMER\"\n\"SUMMER DAYS\"\n\"FEEL FREE\"\n\nDo not treat typography as subtitles.\n\nTypography is part of the physical motion-graphics environment.\n\nSUMMER ATMOSPHERE:\n\nCreate a subtle refreshing summer atmosphere throughout the video.\n\nUse bright natural light, clear skies, soft sunlight, gentle wind, light shadows, and a clean colorful anime palette.\n\nThe girls' hair, ribbons, sleeves, skirts, and loose clothing should frequently respond to wind.\n\nUse occasional subtle summer elements:\n\nsunlight,\nblue sky,\nsummer breeze,\nfloating paper,\nsmall leaves,\nlight particles,\nreflections,\nbright school corridors,\nopen windows,\nsummer streets.\n\nDo not make the video feel like a generic beach vacation.\n\nThe primary environment remains a dynamic Japanese youth / school / music-video world.\n\nMOTORCYCLE CHANGE:\n\nReplace all cruiser / American-style motorcycles with modern Japanese-style full-fairing sport motorcycles.\n\nThe motorcycle must be a sleek Japanese-style racer replica / supersport motorcycle.\n\nUse:\n\nfull aerodynamic fairings,\naggressive front cowl,\nsport riding position,\nclip-on handlebars,\nrear-set footpegs,\nlarge front brake discs,\nsport tires,\nhigh-performance wheels,\nsharp aerodynamic bodywork,\ncompact tail section,\nracing-inspired graphics.\n\nThe rider uses a clearly forward-leaning sport riding posture.\n\nThe motorcycle must visually resemble a modern racer replica rather than a cruiser.\n\nIMPORTANT:\n\nNO American cruiser motorcycle.\n\nNO chopper.\n\nNO long low cruiser.\n\nNO large relaxed handlebars.\n\nNO chrome-heavy cruiser styling.\n\nNO cruiser riding posture.\n\nNO classic Harley-style silhouette.\n\nInstead:\n\nJapanese sportbike.\n\nRacer replica.\n\nFull fairing.\n\nAggressive aerodynamic silhouette.\n\nForward-leaning riding position.\n\nHigh-speed cornering.\n\nMOTORCYCLE ACTION:\n\nThe girl actively controls the motorcycle.\n\nShe accelerates hard.\n\nShe shifts her body.\n\nShe leans deeply into a corner.\n\nThe motorcycle banks dramatically.\n\nThe tires maintain strong contact with the road.\n\nThe camera tracks beside the motorcycle at high speed.\n\nThen the camera moves slightly ahead of the motorcycle.\n\nThe camera swings around the front.\n\nThe front wheel and brake disc fill the frame.\n\nThe spinning wheel transforms into a circular graphic.\n\nThe brake disc becomes a rotating typography ring.\n\nWords such as:\n\n\"FULL SPEED\"\n\"FASTER\"\n\"RIDE\"\n\"GO FASTER\"\n\nrotate around the wheel.\n\nThe motorcycle accelerates again.\n\nIts motion trail becomes enormous kinetic typography.\n\n\"GO!\"\n\nstretches across the screen.\n\nThen:\n\n\"TAKE OFF\"\n\nbecomes the transition into the next scene.\n\nThe motorcycle sequence should feel like a Japanese racing-anime music video rather than a motorcycle advertisement.\n\nCAMERA DIRECTION:\n\nContinue using extremely varied camera movement.\n\nPrioritize:\n\nleft-to-right tracking,\nright-to-left tracking,\ndiagonal tracking,\nside tracking,\ncamera moving beside characters,\ncamera moving between characters,\ncamera overtaking the sport motorcycle,\ncamera moving around the front wheel,\ncamera circling musicians,\ncamera rotating around dancers,\ncamera dropping vertically,\ncamera rising vertically,\nDutch angles,\nrapid push-ins,\nrapid pull-backs,\nwhip pans,\nrotational transitions,\nextreme close passes,\nforeground object wipes,\nperspective distortion,\nfocus changes,\nstrong parallax,\ntypography passing the lens.\n\nDo NOT rely primarily on forward and backward camera movement.\n\nVISUAL DENSITY:\n\nIncrease the overall visual information density.\n\nCharacters, summer uniforms, objects, manga panels, typography, motion trails, particles, musical instruments, school environments, bicycles, sport motorcycles, school bags, papers, signs, and graphic elements should coexist.\n\nNever leave large empty areas.\n\nAt many moments, several independent visual events should happen simultaneously.\n\nFor example:\n\na girl running in the foreground,\n\nanother girl jumping in the middle-ground,\n\na sport motorcycle passing in the background,\n\nlarge typography crossing diagonally,\n\nsmall typography floating behind the characters,\n\nmotion trails connecting all elements,\n\nand manga panels transforming around them.\n\nThe scene should feel extremely rich but still intentionally composed.\n\nThe complexity should come from meaningful simultaneous actions and readable typography.\n\nFINAL VISUAL TARGET:\n\nModern Japanese anime music-video opening.\n\nCute contemporary Japanese summer uniforms.\n\nRefreshing summer atmosphere.\n\nBeautiful energetic high-school girls.\n\nExtremely dynamic action.\n\nHigh-speed Japanese sport motorcycle.\n\nKinetic typography.\n\nApproximately TWICE the previous typography density.\n\nMany readable words.\n\nTypography integrated into character movement.\n\nTypography integrated into camera movement.\n\nTypography integrated into environmental transitions.\n\nFast editing.\n\nContinuous transformation.\n\nStrong parallax.\n\nStrong lateral camera movement.\n\nStrong rotational camera movement.\n\nHigh information density.\n\nStylish contemporary Japanese graphic design.\n\nMid-2000s Japanese music-video energy combined with modern anime motion graphics and contemporary kinetic typography.\n\nMOST IMPORTANT:\n\nSUMMER UNIFORMS.\n\nREFRESHING SUMMER ATMOSPHERE.\n\nCUTE MODERN JAPANESE HIGH-SCHOOL FASHION.\n\nTWICE THE PREVIOUS AMOUNT OF READABLE TYPOGRAPHY.\n\nALL TEXT MUST BE READABLE.\n\nSPORT RACING MOTORCYCLE, NOT AMERICAN CRUISER.\n\nSHOW GIRLS DOING THINGS.\n\nSHOW TYPOGRAPHY MOVING.\n\nSHOW THE ENTIRE BODY OF THE CHARACTERS WHEN POSSIBLE.\n\nNEVER LET THE VIDEO BECOME A COLLECTION OF STATIC PORTRAITS.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 115,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093126466792034728",
    "title": "刀光、雨伞与马蹄组成文字节拍",
    "titleEn": "Blades, Umbrellas, and Hooves Form a Typographic Beat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "刀刃、雨伞和马蹄分别触发英文标题，声音同步的物件运动把三段文字变成短促节拍。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blade, umbrella, and galloping hooves each trigger an English title, turning three object actions into a sound-synced typographic rhythm. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PixVerse/status/2093126466792034728",
    "sourceLabel": "X 原帖 · @PixVerse",
    "author": "PixVerse",
    "publishedAt": "2026-08-28T00:00:12.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093126466792034728.mp4",
    "posterUrl": "/posters/x/2093126466792034728.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "We tested MiniMax H3 on sound-synced typography. All from a text prompt.\n\nA blade slashes THE LAST DRAW. An umbrella unfolds UNDER THE RAIN. Hoofbeats pound out NO ROAD BACK.\n\nPixVerse members get 50% OFF MiniMax H3, or 60% OFF with Ultra or Team Ultra. The offer ends on September 4, 2026, at 11:59 PM PDT.\n\nPrompt below",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 16,
      "views": 658,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093124872608297470",
    "title": "考拉角色举起产品发布标牌",
    "titleEn": "A Koala Mascot Raises a Product Launch Sign",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卡通考拉举起发布标牌，箭头旋转后回到正面，用简单角色动作完成一段产品上线提示。 原生视频规格为 7 秒 · 1888×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cartoon koala lifts a launch sign, spins the arrow, and returns it to camera in a compact product-release announcement. Native-video output: 7s · 1888×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BeatsOnBase/status/2093124872608297470",
    "sourceLabel": "X 原帖 · @BeatsOnBase",
    "author": "Beats on Base",
    "publishedAt": "2026-08-27T23:53:52.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093124872608297470.mp4",
    "posterUrl": "/posters/x/2093124872608297470.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1888×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Consider this your sign 🪧\n\nMiniMax H3 Max image-to-video and text-to-video just landed across the base:0x315b8c9a1123c10228d469551033440441b41f0b infrastructure.\n\nFast. Cheap. Incredibly high quality.\n\nSpin that shit. LFG ♻️",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 6,
      "views": 202,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093124514284712336",
    "title": "漫画分镜切入动画片头",
    "titleEn": "Manga Panels Cut into an Animated Opener",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白漫画墙、彩色角色与粗体标题连续切换，手绘质感和快速排版组成动画开场。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Black-and-white manga walls, colored characters, and bold titles switch in sequence, combining drawn texture and rapid typography into an animated opening. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/studio_oneroom/status/2093124514284712336",
    "sourceLabel": "X 原帖 · @studio_oneroom",
    "author": "AIスタジオワンルーム（AIアニメ、動画、漫画）",
    "publishedAt": "2026-08-27T23:52:27.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093124514284712336.mp4",
    "posterUrl": "/posters/x/2093124514284712336.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "TapNowのMiniMax H3でモーショングラフィックアニメを制作しました。\n\nやっぱりMidjourneyはいいですね。\n雰囲気のある絵が作れて楽しいです。\n\nまだ、研究中ですけどアニメとモーショングラフィックを混ぜたらアニメのオープニング映像とかに良さそうだと思いました。\n\nTapTVに制作過程を公開\n\nhttps://app.tapnow.ai/home/taptv/9/24e94c12-7781-49fa-9ace-0abea325cd16?order=created_at\n\nまぁ制作過程などと言ってもリファレンス画像と制作プロンプトしかないのですが😆\n\n@TapNow_AI \n#TapNow",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 20,
      "views": 865,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093122577170592135",
    "title": "花魁穿行灯笼街巷",
    "titleEn": "An Oiran Moves Through a Lantern-Lit Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从昏暗花街推进到盛装花魁近景，再回到灯笼与人群，形成夜色中的古风角色展示。 原生视频规格为 23 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera moves from a dim pleasure-quarter street to a close portrait of an elaborately dressed oiran, then back to lanterns and crowds for a period character vignette. Native-video output: 23s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/suizenji98/status/2093122577170592135",
    "sourceLabel": "X 原帖 · @suizenji98",
    "author": "どっと",
    "publishedAt": "2026-08-27T23:44:45.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093122577170592135.mp4",
    "posterUrl": "/posters/x/2093122577170592135.jpg",
    "duration": 23,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "花魁\n#MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 9,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093119842534801746",
    "title": "秋日穿搭变成杂志广告",
    "titleEn": "An Autumn Outfit Becomes a Magazine Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "枫叶、咖啡和棕色穿搭在杂志版式间切换，以暖色静物和人物卡完成秋季时尚短片。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Maple leaves, coffee, and a brown outfit move through magazine layouts, combining warm still life and character cards into an autumn fashion spot. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kik0ai1jikake/status/2093119842534801746",
    "sourceLabel": "X 原帖 · @kik0ai1jikake",
    "author": "あぁく",
    "publishedAt": "2026-08-27T23:33:53.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093119842534801746.mp4",
    "posterUrl": "/posters/x/2093119842534801746.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "H3色々試した\n#chatgpt #MinimaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 359,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093119206581809545",
    "title": "粉发乐队在白色影棚演奏",
    "titleEn": "A Pink-Haired Band Performs in a White Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一粉发角色分饰吉他手与鼓手，乐器动作、演唱口型和白色影棚镜头组成短乐队表演。 原生视频规格为 15 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same pink-haired character appears as guitarists and a drummer, combining instrument motion, singing, and a white studio into a short band performance. Native-video output: 15s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mari_69_pink/status/2093119206581809545",
    "sourceLabel": "X 原帖 · @mari_69_pink",
    "author": "山中 まり",
    "publishedAt": "2026-08-27T23:31:21.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093119206581809545.mp4",
    "posterUrl": "/posters/x/2093119206581809545.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "H3 Max、試してみた。\n\nこれはすごい！ヤバい！\n(語彙力)\n\nプロンプトも素直に効いて、\n生成音源とのリップシンク、\n楽器の動きも完璧👌\n\n顔の破綻もほぼない！\n\n…AIの進化スピードが早すぎてヤバい…\n\n#H3Max\n#minimaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 64,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093118957628821666",
    "title": "橙黑文字在玩具节奏中弹跳",
    "titleEn": "Orange-and-Black Type Bounces Like a Toy",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "线条、圆点和日英文字在橙黑配色中聚合、滚动与弹跳，十五秒内完成九段高速动态图形。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Lines, circles, and Japanese-English type assemble, roll, and bounce in an orange-and-black palette across nine fast motion-graphics beats. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "【基本】\n・15秒／9ショット／未就学児〜小学校低学年がワクワクする、明るく楽しい超高速モーショングラフィック\n・オフホワイト、インクブラック、高彩度オレンジのみ\n・文字は太く丸みのあるポップな書体。日本語は太いゴシック\n・文字を「おもちゃ」のように弾ませる、転がす、積む、跳ねる、吸い込む\n・表示文字は9種のみ。綴り・大小文字・空白を完全維持。余計な文字、数字、記号、文字化け禁止\n①【みにまっくす えいち3】\n②【もーしょん】\n③【ぐらふぃくす】\n④【のーえでいっと】\n⑤【わんしょっと】\n⑥【コピペで】\n⑦【かんたん】\n⑧【みんみん】\n⑨【minmin】\n\n【SHOTS】\n1｜黒地。大量の線分が四方から「びゅーん！」と中央へ集合。オレンジ円がポンッと大ジャンプして激突。\n2｜00:01。線分がパズルのようにパチパチ結合し【①】完成。黒→白。\n3｜00:02.6。オレンジ文字がカメラへ急接近。画面いっぱいになり、一瞬ピタッと停止。\n4｜00:03.4。Hの縦棒がニョキッと上下へ伸び、画面を3列に分割。列が上下に楽しく高速スクロール。\n5｜00:05.2。3列が「ピタッ！」と停止。【②】が巨大なゴムボールのように跳ね込む。\n6｜00:06.4。黒地。3Dグリッドがジェットコースターのように奥へ走り、【③】が次々増殖。\n7｜00:08。4枚のカラーカードがパッパッと切替。【④】【⑤】【⑥】【⑦】を順番に表示。\n8｜00:10。4カードが渦に吸い込まれ、オレンジの光玉に変化。左右から【⑧】が飛び込みドーン！衝撃波。\n9｜00:12。【⑧】がポンッと着地。その上に【⑨】。中央2行＋オレンジ線。最後は00:13.8〜15.0完全静止。",
    "promptSourceUrl": "https://x.com/minmin_4410/status/2093118957628821666",
    "sourceUrl": "https://x.com/minmin_4410/status/2093118957628821666",
    "sourceLabel": "X 原帖 · @minmin_4410",
    "author": "minmin みんみん｜変なAI Creator",
    "publishedAt": "2026-08-27T23:30:22.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093118957628821666.mp4",
    "posterUrl": "/posters/x/2093118957628821666.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "またワンルームさんの神プロンプトが！！🤩\nプロンプト文をさらに圧縮してテイスト超改変して遊んでみてます✨️\n\n#MiniMaxH3 @Hailuo_AI\nプロンプト⬇️\n【基本】\n・15秒／9ショット／未就学児〜小学校低学年がワクワクする、明るく楽しい超高速モーショングラフィック\n・オフホワイト、インクブラック、高彩度オレンジのみ\n・文字は太く丸みのあるポップな書体。日本語は太いゴシック\n・文字を「おもちゃ」のように弾ませる、転がす、積む、跳ねる、吸い込む\n・表示文字は9種のみ。綴り・大小文字・空白を完全維持。余計な文字、数字、記号、文字化け禁止\n①【みにまっくす えいち3】\n②【もーしょん】\n③【ぐらふぃくす】\n④【のーえでいっと】\n⑤【わんしょっと】\n⑥【コピペで】\n⑦【かんたん】\n⑧【みんみん】\n⑨【minmin】\n\n【SHOTS】\n1｜黒地。大量の線分が四方から「びゅーん！」と中央へ集合。オレンジ円がポンッと大ジャンプして激突。\n2｜00:01。線分がパズルのようにパチパチ結合し【①】完成。黒→白。\n3｜00:02.6。オレンジ文字がカメラへ急接近。画面いっぱいになり、一瞬ピタッと停止。\n4｜00:03.4。Hの縦棒がニョキッと上下へ伸び、画面を3列に分割。列が上下に楽しく高速スクロール。\n5｜00:05.2。3列が「ピタッ！」と停止。【②】が巨大なゴムボールのように跳ね込む。\n6｜00:06.4。黒地。3Dグリッドがジェットコースターのように奥へ走り、【③】が次々増殖。\n7｜00:08。4枚のカラーカードがパッパッと切替。【④】【⑤】【⑥】【⑦】を順番に表示。\n8｜00:10。4カードが渦に吸い込まれ、オレンジの光玉に変化。左右から【⑧】が飛び込みドーン！衝撃波。\n9｜00:12。【⑧】がポンッと着地。その上に【⑨】。中央2行＋オレンジ線。最後は00:13.8〜15.0完全静止。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 325,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093116596915192075",
    "title": "巧克力牛奶完成旋转倒杯广告",
    "titleEn": "Chocolate Milk Gets a Twist-and-Pour Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "巧克力牛奶盒旋转进入厨房，饮料倒入玻璃杯并以品牌口号收尾，液体和包装运动连贯。 原生视频规格为 27 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A chocolate-milk carton spins into a kitchen, pours into a glass, and ends on a branded line with coherent liquid and packaging motion. Native-video output: 27s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Creatify_AI/status/2093116596915192075",
    "sourceLabel": "X 原帖 · @Creatify_AI",
    "author": "Creatify AI",
    "publishedAt": "2026-08-27T23:20:59.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093116596915192075.mp4",
    "posterUrl": "/posters/x/2093116596915192075.jpg",
    "duration": 27,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "We made a commercial-grade chocolate milk ad using Hailuo MiniMax H3, generated in Creatify.  \n\nLiquid is the hardest thing in AI video. From the viscosity to the swirl to the pour catching the light. \n\n> Multimodal input: text, image, video, audio together >Character consistency across shots \n>2K resolution \n>5 to 15 second clips  \n\nTwist, pour, and pop.",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 8,
      "views": 292,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093116428790698074",
    "title": "书店角色转身穿过姓名排版",
    "titleEn": "A Bookstore Character Gets a Motion Identity",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "围裙角色在姓名排版、转身动作和多角度展示间切换，把参考角色表扩展为完整身份短片。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An apron-clad character moves through name typography, a turn, and multiple views, extending a reference sheet into a complete motion-identity clip. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MizukiAMANO1/status/2093116428790698074",
    "sourceLabel": "X 原帖 · @MizukiAMANO1",
    "author": "天野水樹@ノクタ民エンジョイ勢",
    "publishedAt": "2026-08-27T23:20:19.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093116428790698074.mp4",
    "posterUrl": "/posters/x/2093116428790698074.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "四分割画像の次は、プロンプトをお借りして、MiniMax H3で遊んでみる\n新作で使おうと思っている書店員「貝淵茜子」のカッコイイ動画\n分割画像を作ったときに、レファレンスとして使えるキャラシートも生成したので、それをベースに\nrtx3060環境かつ高速化不使用なので１時間かかったT_T",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 99,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093115527791288580",
    "title": "雨夜古装对白铺开一分钟故事",
    "titleEn": "Rainy-Night Period Dialogue Unfolds for a Minute",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "提灯男子与室内女子在雨夜往返对白，竖屏字幕、环境声和口型共同支撑近一分钟古装片段。 原生视频规格为 61 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A lantern-carrying man and an indoor woman exchange dialogue on a rainy night as vertical subtitles, ambience, and lip sync support a nearly minute-long period scene. Native-video output: 61s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TristanLi99/status/2093115527791288580",
    "sourceLabel": "X 原帖 · @TristanLi99",
    "author": "Tristan Lee",
    "publishedAt": "2026-08-27T23:16:44.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093115527791288580.mp4",
    "posterUrl": "/posters/x/2093115527791288580.jpg",
    "duration": 61,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "《画皮·雨夜》约61秒字幕版已经完成。全片配音、环境声和口型均由 MiniMax H3 + ComfyUI 原生生成，未使用 QwenTTS。\n\n规格：60.801秒、9:16、768×1344、24fps、H.264、AAC双声道32kHz。已抽帧检查人物连续性、说话时嘴部状态、画皮变化和字幕显示；没有额外使用语音识别逐字转录音轨。\n\n感觉再摸索摸索，可以去作短剧了",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 51,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093114211555770764",
    "title": "地下车库里的武士肖像",
    "titleEn": "A Samurai Portrait in an Underground Garage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "披甲武士站在地下车库，镜头从中景推到面部特写，以写实灯光呈现短促角色肖像。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An armored samurai stands in an underground garage as the camera moves from a medium view into a close portrait under realistic lighting. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yasisreal/status/2093114211555770764",
    "sourceLabel": "X 原帖 · @yasisreal",
    "author": "Cycle",
    "publishedAt": "2026-08-27T23:11:30.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093114211555770764.mp4",
    "posterUrl": "/posters/x/2093114211555770764.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でつくってみたけど\nなぜ地下駐車場なんだ。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 35,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093113408002429285",
    "title": "能量饮料罐缠上红蓝光轨",
    "titleEn": "Light Trails Wrap an Energy-Drink Can",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "饮料罐在黑色影棚中旋转，红蓝光轨环绕包装并以横置产品镜头结束。 原生视频规格为 10 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An energy-drink can rotates in a black studio as red and blue light trails wrap the package before a horizontal product shot closes the clip. Native-video output: 10s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ksa8Turky/status/2093113408002429285",
    "sourceLabel": "X 原帖 · @ksa8Turky",
    "author": "‏تركي",
    "publishedAt": "2026-08-27T23:08:19.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093113408002429285.mp4",
    "posterUrl": "/posters/x/2093113408002429285.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@higgsfield \n#MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 161,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093110975649956226",
    "title": "光头英雄闯入霓虹动画片头",
    "titleEn": "A Bald Hero Bursts into a Neon Anime Opener",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "光头角色在霓虹特写、英雄姿态和动能标题间高速切换，形成带音乐感的动作动画片头。 原生视频规格为 15 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A bald character cuts rapidly between neon close-ups, heroic poses, and kinetic titles in an action-anime opening with a musical pulse. Native-video output: 15s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "مشهد انمي اكشن مدته ١٥ ثانية، يظهر @ Image1 باقوى مقدمة انمي اكشن بطابع بطولي وخيال علمي وقوى خارقة، يظهر كبداية شارة انمي مع اغنية مخصصة له بعنوان Bald Man يظهر فيها كيناتيك تايبوغرافي بطابع قوي وهوك بصري مجنون يجعل المشاهد يراقب المقدمة لاخر لحظة، لقطات من زوايا مختلفة و حركات كاميرا متعددة وفلاش كت سريع يبرز الاكشن و قوة البطل وابرازه بشكل قوي بصريا، الوان فايبرانت ونيون مع قلو ، تتابع مشاهد وسببيه لكل حدث وماتش كت قوي يجذب المشاهد حتى اخر لحظة، تاثيرات بصرية مبهرة ونهاية مرضية تظهر البطل في اقوى حالاته",
    "promptSourceUrl": "https://x.com/aziz4ai/status/2093110980624499026",
    "sourceUrl": "https://x.com/aziz4ai/status/2093110975649956226",
    "sourceLabel": "X 原帖 · @aziz4ai",
    "author": "AZIZ | AI 🇸🇦",
    "publishedAt": "2026-08-27T22:58:39.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093110975649956226.mp4",
    "posterUrl": "/posters/x/2093110975649956226.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Direct prompting with only character sheet 😍\n\nI made this short video using #MiniMaxH3 by @Hailuo_AI \n\nLove the capabilities and prompt coherence and understanding even Arabic language for prompting 🚀\n\nThe prompt in Arabic in next comment ⬇️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 851,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093109176671359184",
    "title": "纸板剧场解释原子状态变化",
    "titleEn": "A Cardboard Theater Explains Atomic States",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "纸板科学家与粒子模型演示固体、液体和气体，标签与手写提示把概念转成直观教学短片。 原生视频规格为 43 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cardboard scientist and particle models demonstrate solids, liquids, and gases, turning the concept into a clear educational film with labels and handwritten cues. Native-video output: 43s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Arkatech/status/2093109176671359184",
    "sourceLabel": "X 原帖 · @Arkatech",
    "author": "Arkatech",
    "publishedAt": "2026-08-27T22:51:30.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093109176671359184.mp4",
    "posterUrl": "/posters/x/2093109176671359184.jpg",
    "duration": 43,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Used MiniMax Design explainer flow to make this educational film about atoms....let the series begin🛸 \n#minimaxh3 on #MiniMaxDesign @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 123,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093104167321825301",
    "title": "动漫角色穿过霓虹几何空间",
    "titleEn": "An Anime Character Crosses a Neon Geometric Space",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫角色从眼部特写进入发光网格、立方体和圆环空间，竖屏镜头用高速几何转场延续角色设计。 原生视频规格为 15 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime character moves from an eye close-up into glowing grids, cubes, and rings as fast geometric transitions preserve the design in a vertical clip. Native-video output: 15s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mugi_AI_Art/status/2093104167321825301",
    "sourceLabel": "X 原帖 · @mugi_AI_Art",
    "author": "むぎ@AIアートのセカイ",
    "publishedAt": "2026-08-27T22:31:36.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093104167321825301.mp4",
    "posterUrl": "/posters/x/2093104167321825301.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "想像神ちゃん × シネマティック・モーショングラフィック⚡️\n今回はPollo AIの「MiniMax H3」を使って、15秒の映像作品を制作しました！\n細かなショット切り替えや高速カメラワーク、リングやキューブが連動するモーションを詰め込みつつ、想像神ちゃんの表情やデザインもしっかり残せたのが印象的でした。\nテンポの速い映像でも、キャラクターとグラフィックを自然につないでくれるので、短尺作品やMV風の映像にも使いやすそうです。\n現在、Pollo AIではMiniMax H3のUnlimitedキャンペーンが開催中！\n・Proプラン：10日間 Unlimited\n・Ultraプラン：365日間 Unlimited\n生成回数を気にせず、構成やカメラワークを何度も試せるのは嬉しいですね。気になっていた方はぜひチェックしてみてください🙌\n※対象プランやキャンペーンの詳細は公式サイト公式ページをご確認ください。\n@PolloAIJP",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 22,
      "views": 535,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093103976644493667",
    "title": "街头骑行对白配上动能字幕",
    "titleEn": "Street Cycling Dialogue Gets Kinetic Captions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名骑行者边穿过街区边交谈，跟随人物的英文字幕和连续机位让日常对话更有节奏。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two cyclists talk while moving through a neighborhood, with character-following English captions and continuous framing adding rhythm to the exchange. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hadiprassetyo_/status/2093103976644493667",
    "sourceLabel": "X 原帖 · @hadiprassetyo_",
    "author": "Hadi Prasetyo",
    "publishedAt": "2026-08-27T22:30:50.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093103976644493667.mp4",
    "posterUrl": "/posters/x/2093103976644493667.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Lagi nyoba bikin video AI dengan percakapan yang terasa lebih natural👀\n\nBukan cuma karakter berdiri terus ngomong, tapi tetap gerakan, saling respons, plus kinetic subtitle biar scene-nya lebih hidup.\n\nPelan-pelan makin berasa kayak real footage.\n\nMINIMAX H3 on @Sogni_Protocol",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 94,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093103449223082308",
    "title": "绿幕前的猫咪轻快伸展",
    "titleEn": "A Cat Stretches Playfully Against Green",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "浅色猫咪在纯绿色背景前俯身、抬头并摆动尾巴，短镜头集中测试连续身体运动。 原生视频规格为 7 秒 · 512×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pale cat crouches, looks up, and moves its tail against a solid green background in a short continuity test of body motion. Native-video output: 7s · 512×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nyymisfw/status/2093103449223082308",
    "sourceLabel": "X 原帖 · @nyymisfw",
    "author": "NyymiSFW",
    "publishedAt": "2026-08-27T22:28:44.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093103449223082308.mp4",
    "posterUrl": "/posters/x/2093103449223082308.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "512×704",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 test",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 6,
      "views": 167,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093101101298119086",
    "title": "女仆角色在室内轻轻转身",
    "titleEn": "A Maid Character Comes to Life Indoors",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发女仆角色在室内场景中保持造型并产生细微动作，展示 Apple Silicon 本地生成的短片结果。 原生视频规格为 5 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde maid character retains her design while moving subtly in an interior, showing a short result generated locally on Apple Silicon. Native-video output: 5s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ruku_practice/status/2093101101298119086",
    "sourceLabel": "X 原帖 · @ruku_practice",
    "author": "ルク｜ゲームポータル運営＆再起動日記",
    "publishedAt": "2026-08-27T22:19:25.000Z",
    "addedAt": "2026-08-28T01:32:58.034Z",
    "mediaUrl": "/media/x-2093101101298119086.mp4",
    "posterUrl": "/posters/x/2093101101298119086.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "本地生成",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "早速、Minimax H3動画生成を128GB Macで試しました。\nローカル mlx-serveを利用することで、時間はかかるけどそこまで破綻なく生成に成功。（自分はよくわかってない）\n昼間に急ぐときはfal、夜にガチャを回しておくときはこちら、ということができそうです。\n詳しいやり方はまたどこかでまとめます。",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 8,
      "views": 208,
      "snapshotAt": "2026-08-28T01:32:58.034Z"
    },
    "approvedAt": "2026-08-28T01:32:58.034Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093407813980602709",
    "title": "星际战场铺开即时战略混战",
    "titleEn": "A Space Battlefield Erupts into Real-Time Strategy Combat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "俯视镜头掠过岩石星球上的基地、载具与交火单位，把即时战略界面扩展成连续战场。 原生视频规格为 14 秒 · 704×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An overhead camera crosses bases, vehicles, and combat units on a rocky planet, expanding a real-time strategy interface into a continuous battlefield. Native-video output: 14s · 704×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cocktailpeanut/status/2093407813980602709",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-28T18:38:11.000Z",
    "addedAt": "2026-08-29T00:19:15.550Z",
    "mediaUrl": "/media/x-2093407813980602709.mp4",
    "posterUrl": "/posters/x/2093407813980602709.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "704×384",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I generated Starcraft with Minimax H3. What is he even saying lol",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 15,
      "views": 1766,
      "snapshotAt": "2026-08-29T00:19:15.550Z"
    },
    "approvedAt": "2026-08-29T00:19:15.550Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093281766542807370",
    "title": "蓝发剑士在雪花与书法间循环",
    "titleEn": "A Blue-Haired Swordsman Loops Through Snow and Calligraphy",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝发剑士保持持刀姿态，飘雪、书法与冷色光效在竖屏壁纸中形成循环。 原生视频规格为 39 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired swordsman holds a blade as snow, calligraphy, and cool light effects form a looping vertical wallpaper. Native-video output: 39s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cia0_exe/status/2093281766542807370",
    "sourceLabel": "X 原帖 · @Cia0_exe",
    "author": "Cia0",
    "publishedAt": "2026-08-28T10:17:19.000Z",
    "addedAt": "2026-08-29T00:19:15.550Z",
    "mediaUrl": "/media/x-2093281766542807370.mp4",
    "posterUrl": "/posters/x/2093281766542807370.jpg",
    "duration": 39,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @itsPolloAI - 365 days of unlimited \n\nWallpaper PC 4K smooth loop animation and detailed typography art \n\nForget about adding GPU cards or purchasing extra compute, there’s no separate deployment to manage either.\n\nIf you’ve been using H3, this might be worth checking out. \n\n5,690 Chinese prompt",
    "engagement": {
      "replies": 20,
      "reposts": 1,
      "likes": 28,
      "views": 3195,
      "snapshotAt": "2026-08-29T00:19:15.550Z"
    },
    "approvedAt": "2026-08-29T00:19:15.550Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093487845314822338",
    "title": "街头角色完成荒诞任务",
    "titleEn": "A Street Character Completes an Absurd Mission",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "街头角色沿城市道路行动，镜头以游戏式构图推进，并以任务完成提示收尾。 原生视频规格为 8 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A street character moves through an urban road in game-like framing before a mission-complete cue closes the sequence. Native-video output: 8s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/boxingai/status/2093487845314822338",
    "sourceLabel": "X 原帖 · @boxingai",
    "author": "vx",
    "publishedAt": "2026-08-28T23:56:12.000Z",
    "addedAt": "2026-08-29T00:19:15.550Z",
    "mediaUrl": "/media/x-2093487845314822338.mp4",
    "posterUrl": "/posters/x/2093487845314822338.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max T2V",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 20,
      "snapshotAt": "2026-08-29T00:19:15.550Z"
    },
    "approvedAt": "2026-08-29T00:19:15.550Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093487787664417088",
    "title": "掌机广告上下同步变形",
    "titleEn": "Two Handheld-Console Ads Transform in Sync",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下分屏中的掌机、角色和霓虹包装同步切换，用同一广告节奏并排比较两种结果。 原生视频规格为 15 秒 · 1920×2196。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Handheld consoles, characters, and neon packaging transform in synchronized split screens, comparing two results through the same ad rhythm. Native-video output: 15s · 1920×2196. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/onofumi_AI/status/2093487787664417088",
    "sourceLabel": "X 原帖 · @onofumi_AI",
    "author": "おのふみ| AIクリエーター×個人開発",
    "publishedAt": "2026-08-28T23:55:58.000Z",
    "addedAt": "2026-08-29T00:19:15.550Z",
    "mediaUrl": "/media/x-2093487787664417088.mp4",
    "posterUrl": "/posters/x/2093487787664417088.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1920×2196",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial",
      "ugc"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "同じプロンプトをH3とH3 Maxに投げたら、面白かったです。\n\n上が通常のMiniMax H3(2K)、下がfalがポストトレーニングしたH3 Max(768P)。架空ゲーム機の15秒ベクター広告を、一字も変えずに両方へ投げた比較です。\n\n驚いたのはMAX爆速なのにクオリティがさほど変わらない事。\n\n後は好み問題も少し入ってくる。コストと時間考えるとMAXかなーという判断。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 120,
      "snapshotAt": "2026-08-29T00:19:15.550Z"
    },
    "approvedAt": "2026-08-29T00:19:15.550Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093486797250756791",
    "title": "霓虹义体角色跃过城市屋顶",
    "titleEn": "A Neon Cyborg Leaps Across City Rooftops",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "义体角色在霓虹城市中奔跑、翻跃和落地，高速跟拍串联屋顶动作。 原生视频规格为 15 秒 · 1152×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cybernetic character runs, vaults, and lands across a neon city as fast tracking shots connect the rooftop action. Native-video output: 15s · 1152×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/harako_ai/status/2093486797250756791",
    "sourceLabel": "X 原帖 · @harako_ai",
    "author": "ハラコ AI",
    "publishedAt": "2026-08-28T23:52:02.000Z",
    "addedAt": "2026-08-29T00:19:15.550Z",
    "mediaUrl": "/media/x-2093486797250756791.mp4",
    "posterUrl": "/posters/x/2093486797250756791.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMaxH3製パルクール動画よ～！🤖プロンプトはsolにお願いしたんだけど、調整数回でこれが出てくるの驚いちゃうわねえ…",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 4,
      "snapshotAt": "2026-08-29T00:19:15.550Z"
    },
    "approvedAt": "2026-08-29T00:19:15.550Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093484810975187396",
    "title": "室内草地蔓延成超现实清晨",
    "titleEn": "Indoor Grass Spreads into a Surreal Morning",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物脚下的草地侵入现代室内，走廊与自然质感在连续移动中交叠。 原生视频规格为 34 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Grass spreads beneath a character inside a modern interior, blending corridors and natural texture through continuous movement. Native-video output: 34s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/breath_in_code/status/2093484810975187396",
    "sourceLabel": "X 原帖 · @breath_in_code",
    "author": "LinacoWSL",
    "publishedAt": "2026-08-28T23:44:08.000Z",
    "addedAt": "2026-08-29T00:19:15.550Z",
    "mediaUrl": "/media/x-2093484810975187396.mp4",
    "posterUrl": "/posters/x/2093484810975187396.jpg",
    "duration": 34,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "さて、今日も１日頑張りましょ💪 MINIMAX H3\nRTX4070ti ローカル生成",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 20,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.550Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093484761180463541",
    "title": "狐耳少女的生日庆典蒙太奇",
    "titleEn": "A Fox-Eared Heroine Gets a Birthday Montage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "狐耳少女在蛋糕、烟花和同伴镜头间切换，十五秒内完成一段生日音乐蒙太奇。 原生视频规格为 71 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fox-eared heroine moves between cake, fireworks, and companion shots in a fifteen-second birthday music montage. Native-video output: 71s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EarthGigantea/status/2093484761180463541",
    "sourceLabel": "X 原帖 · @EarthGigantea",
    "author": "アースキー｜薬剤師のAI実装実験室",
    "publishedAt": "2026-08-28T23:43:56.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093484761180463541.mp4",
    "posterUrl": "/posters/x/2093484761180463541.jpg",
    "duration": 71,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今日は薬剤師の傍ら、趣味である動画生成AIの一部をご紹介します😊\n\n#NOAの日 #NOAart #MiniMaxH3　\n\n🐺のあちゃん、お誕生日おめでとうございます🎉\n\nあけじーさんが生み出した「のあ」と、あけじーさんが作った『ニクニクカーニバル』（SUNO）をお借りしてRemixをつくり、70秒のミュージックビデオにしました。\n\n映像は全カット、ローカル環境MiniMax-H3で生成しています。編集はRemotion。\n\nクラウドは使っていません。　\n\n8月29日は焼肉の日。\n私の推し活も　ずっとつづくよ～～～！！！🔥",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 35,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093484413502018000",
    "title": "雨幕、聚光灯与标题组成黑白片头",
    "titleEn": "Rain, a Spotlight, and Type Form a Monochrome Opener",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雨幕中的人影、聚光灯和轮廓文字连续出现，以克制黑白画面形成片头。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure in rain, a spotlight, and outlined type appear in sequence to form a restrained monochrome opener. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aicreataro/status/2093484413502018000",
    "sourceLabel": "X 原帖 · @aicreataro",
    "author": "aicreataro",
    "publishedAt": "2026-08-28T23:42:33.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093484413502018000.mp4",
    "posterUrl": "/posters/x/2093484413502018000.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "anime"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMaxH3でアニメーションの練習。\n作って、壊して、また作る。 \nどれも失敗の証拠として手元に残っている。 \n捨てられないまま、机の上に並んでいる。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 17,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093483469137752318",
    "title": "猫狗在街头展开说唱对决",
    "titleEn": "Cats and Dogs Face Off in a Street Rap Battle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拟人猫狗在街头舞台轮流对唱，近景表情和节拍镜头维持长篇对决。 原生视频规格为 177 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Anthropomorphic cats and dogs trade verses on a street stage as close expressions and rhythmic shots sustain an extended rap battle. Native-video output: 177s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ljd0427/status/2093483469137752318",
    "sourceLabel": "X 原帖 · @ljd0427",
    "author": "TANG",
    "publishedAt": "2026-08-28T23:38:48.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093483469137752318.mp4",
    "posterUrl": "/posters/x/2093483469137752318.jpg",
    "duration": 177,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "일상에서 마주치는 강아지와 고양이의 눈빛, 기싸움, 그리고 묘하게 이어지는 우정. 그 익숙한 바이브를 펑키한 랩 배틀로 만들어봤습니다.\nI turned that familiar vibe into a funky rap battle with MiniMax H3.\n\n#Dogs #Cats",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 22,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093481926514417950",
    "title": "集市、舞台与宇宙串成超长幻游",
    "titleEn": "A Market, Stage, and Cosmos Form an Extended Fever Dream",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "市集人群、舞台表演和宇宙意象在长镜头叙事中不断接力，形成超长幻想旅程。 原生视频规格为 401 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Market crowds, stage performances, and cosmic imagery relay through a long-form narrative to create an extended fantasy journey. Native-video output: 401s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wikiwayne/status/2093481926514417950",
    "sourceLabel": "X 原帖 · @wikiwayne",
    "author": "Wayne Lowry",
    "publishedAt": "2026-08-28T23:32:40.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093481926514417950.mp4",
    "posterUrl": "/posters/x/2093481926514417950.jpg",
    "duration": 401,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "You might think you’re high watching this video, but I wanted to test a long-running piece with MiniMax H3, orchestrated by my Hermes agent. The whole thing took an overnight run and a little refinement today. It’s hilariously terrible. And you will laugh out loud a couple of times. I guarantee it. AI slop at its finest.\n\nHigh-level prompt for the A–Z run:\n\nLive-action cinematic mini-stories. G-rated and child-safe.\nEach clip is one capital letter as a real object in the world, not a graphic overlay.\nTwo shots, cut at 00:05.5, with small, slow camera moves. One warm spoken English line.\nQuiet diegetic sound under the line, music stays below speech, short sting after the last word.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 41,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093481504139591787",
    "title": "箭头、时钟与图标环绕校园角色",
    "titleEn": "Arrows, Clocks, and Icons Orbit a School Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "校园角色在箭头、时钟和相机图标间切换，图形元素推动竖屏动效节奏。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A school character moves among arrows, clocks, and camera icons as graphic elements drive the vertical motion rhythm. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yudotanaka/status/2093481504139591787",
    "sourceLabel": "X 原帖 · @yudotanaka",
    "author": "田中勇道 | AI画像・動画生成",
    "publishedAt": "2026-08-28T23:31:00.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093481504139591787.mp4",
    "posterUrl": "/posters/x/2093481504139591787.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でサークル→モザイク→グリッチのトランジション動画を生成してみました。ピクトグラムも入れてモーショングラフィックっぽくしてます。トランジションやモーショングラフィックは、Seedance系よりもMiniMax H3が合ってますね。\n@pixverse　#pixversecpp",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 39,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093477620650312050",
    "title": "咖啡馆情景喜剧迎来机器人访客",
    "titleEn": "A Café Sitcom Welcomes a Robot Visitor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "熟悉的咖啡馆座位迎来机器人角色，人物反应与连续对白构成情景喜剧片段。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A familiar café seating area receives a robot visitor, with character reactions and continuous dialogue forming a sitcom scene. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/atgorans_k/status/2093477620650312050",
    "sourceLabel": "X 原帖 · @atgorans_k",
    "author": "Goran k",
    "publishedAt": "2026-08-28T23:15:34.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093477620650312050.mp4",
    "posterUrl": "/posters/x/2093477620650312050.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max text to video is incredibly good and so much fun to use!\n\nIt's like creating alternate universes in real time",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 39,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093476741553820002",
    "title": "睡前刷手机变成竖屏应用短剧",
    "titleEn": "Late-Night Phone Scrolling Becomes a Vertical App Story",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子在床边刷手机，画面切入应用界面与生活片段，组成竖屏产品叙事。 原生视频规格为 28 秒 · 1080×1944。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman scrolls her phone by the bed before the scene moves into app screens and daily-life moments for a vertical product story. Native-video output: 28s · 1080×1944. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_talk_log/status/2093476741553820002",
    "sourceLabel": "X 原帖 · @ai_talk_log",
    "author": "AIと猫と話す人@Minamo開発",
    "publishedAt": "2026-08-28T23:12:04.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093476741553820002.mp4",
    "posterUrl": "/posters/x/2093476741553820002.jpg",
    "duration": 28,
    "aspectRatio": "portrait",
    "resolution": "1080×1944",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます☀️\n\nローカルでMinimax H3に生成してもらった動画をtiktokに出してみたけど伸びない🤣🤣\n動画作るのは簡単になりつつあるけど、また別の試行錯誤ですね！\n\n冒頭フックをもっと強くしてみよう🏃",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 68,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093475617555579119",
    "title": "化学实验室里的荒诞对白",
    "titleEn": "Absurd Dialogue Unfolds in a Chemistry Lab",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名实验室角色围绕设备展开紧张又荒诞的对话，近景切换维持戏剧冲突。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two lab characters exchange tense, absurd dialogue around technical equipment as close cuts sustain the dramatic conflict. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/atgorans_k/status/2093475617555579119",
    "sourceLabel": "X 原帖 · @atgorans_k",
    "author": "Goran k",
    "publishedAt": "2026-08-28T23:07:36.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093475617555579119.mp4",
    "posterUrl": "/posters/x/2093475617555579119.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I don't remember this Breaking bad episode (because it was generated by MiniMax H3 Max)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 43,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093474780683149442",
    "title": "粉金圆环带动收益文字起舞",
    "titleEn": "A Rose-Gold Hoop Sets Revenue Type in Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "黑色体操剪影操控粉金圆环，收益数字与感谢文字随动作和纸屑节拍出现。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black gymnast silhouette controls a rose-gold hoop while revenue figures and thank-you type appear with movement and confetti beats. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "「添付した参照画像を、映像全体の背景として使用する。参照画像を別の競技場や抽象背景へ置き換えず、画面全体を覆う大きな背景グラフィックとして維持する。元画像の黒い背景、白とグレーの通知表示、左側のカラフルなクラッカーのアイコンを活かす。\n\n背景画像に含まれる文字やアイコンを描き直したり、別の内容へ変更したりしない。背景には、わずかなズーム、横方向への移動、明るさの変化、軽いぼかし、奥行き表現のみを加える。主人公と新しいタイポグラフィーを目立たせるため、背景画像を少し暗くするが、参照画像であることが認識できる状態を保つ。\n\nMiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによる、新体操のフープ演技動画を制作する。\n\n主人公は、全身が黒一色で描かれた一人の女性のシルエット。顔、髪、肌、衣装などの細部は描かず、プロの新体操選手らしい、しなやかで洗練された身体のラインを持つ美しいシルエットとして表現する。\n\n黒い人物が背景に埋もれないよう、身体の輪郭に細い白、マゼンタ、バイオレットのリムライトを加える。ただし、身体の内部は黒一色を維持し、顔や衣装のディテールは描かない。\n\n主人公は一本の鮮やかなメタリック・ピンクゴールドのフープを使用する。フープは細く軽量な新体操用フープで、映像全体を通して完全な円形、一定の太さと大きさを維持する。\n\nフープの色は、淡いローズピンクを基調として、銅色とシャンパンカラーの反射、明るいピンクホワイトのハイライトを持つ、高級感のあるピンクゴールドとする。単純なピンク色やプラスチック素材にはせず、照明や角度によってローズ、カッパー、淡い金属光沢が美しく変化する質感として表現する。\n\n【0.0～3.5秒：X収益】\n映像は、ピンクゴールドのフープを画面手前に極端に大きく配置した、大胆な構図から始まる。カメラはフープの円の内側を通して、その奥に立つ黒い女性シルエットと参照画像の背景を捉える。\n\n主人公はフープの中心に身体を収めた状態から、腕を大きく伸ばしてフープを回転させる。フープを片手で垂直に回しながら身体の周囲へ滑らかに移動させ、そのままフープの中を片脚ずつ優雅に通り抜ける。\n\n最初の強いビートに合わせて、フープの内側に「X収益」の文字を正確に表示する。「X収益」はフープの回転に呼応して画面奥から拡大しながら現れ、最後は正面を向いて明確に読める状態になる。\n\nフープが床へ軽く触れると、接地点からピンクゴールドの円形波紋が広がる。参照画像左側のクラッカーを連想させる、小さな三角形、円、曲線のシェイプがビートに合わせて飛び出す。\n\n【3.5～8.5秒：$96】\n主人公はフープを手首の周囲で高速回転させ、その回転を止めずに腕から肩、背中、反対側の腕へ転がす。フープの自然な慣性と遠心力を維持し、身体の表面に沿って滑らかに移動させる。\n\n続いて、主人公はフープを垂直に転がしながら前方へ送り、その軌道に合わせて大きく脚を伸ばしたスプリットリープを行う。空中で身体と脚の美しいラインを見せ、着地と同時に転がってきたフープを片手で正確に受け止める。\n\nフープの中心に「$96」の文字を正確に表示する。ドル記号を数字の前に置き、スペースを入れず、必ず「$96」と表示する。フープが回転している間も文字自体は正面を向き、明確に読み取れる瞬間を作る。\n\n「$96」は音楽のビートに合わせて拡大、縮小、前後方向への移動を繰り返す。数字がフープの外へ飛び出したり、別の数字へ変化したり、桁が増減したりしない。\n\nカメラは正面からの全身ショットに加え、斜め上から見下ろす俯瞰ショットへ切り替える。俯瞰ショットでは、床面を転がるピンクゴールドのフープの円形、主人公の足運び、背景上を流れる円やラインを印象的に見せる。\n\n【8.5～12.5秒：Thank You】\n主人公はピンクゴールドのフープを頭上へ高く投げ上げる。フープは空中で正しい軸を中心に回転し、完全な円形を維持する。\n\nフープが空中へ上昇する間、主人公は美しいピボットターンを連続して行う。ターンから大きなリープへ移行し、空中を通過するフープの内側へ身体を合わせる。着地後にもう一度身体を回転させ、落下してくる同じフープを片手で正確にキャッチする。\n\nフープが最も高い位置へ到達した瞬間、「Thank You」の文字がその内側から現れる。「Thank」と「You」の間に正しいスペースを入れ、大文字と小文字の組み合わせを正確に維持する。\n\n「Thank You」は、感謝と祝福を感じさせる優雅な動きで画面奥から手前へ近づく。文字の周囲では、参照画像のクラッカーと調和するマゼンタ、イエロー、オレンジ、バイオレット、ピンクゴールドの紙吹雪と小さなシェイプが広がる。\n\nカメラは斜め下から見上げるローアングルへ移動し、空中のピンクゴールドのフープ、主人公の伸びやかな身体、背景に浮かぶ「Thank You」を同時に捉える。カメラアングルが変わっても、主人公とフープの位置関係、回転方向、動作の連続性を維持する。\n\n【12.5～15.0秒：フィナーレ】\n主人公はキャッチしたフープを大きく一回転させ、最後にフープを垂直に立てた状態で片手に保持する。もう片方の腕と片脚を美しく伸ばし、フープと身体の曲線が調和する、上品で自然なフィニッシュを見せる。\n\n最後の画面では、ピンクゴールドのフープを中央へ大きく配置する。フープの内側に「$96」、フープの上方に「X収益」、下方に「Thank You」を読みやすく整列させる。\n\n3つのタイポグラフィーと主人公が、参照画像の重要な部分を完全に覆わないよう、奥行きと配置を調整する。参照画像は少しぼかした状態で背景に残し、その前に黒い主人公、ピンクゴールドのフープ、3つの文字を鮮明に表示する。\n\n最後の約0.8秒は、主人公、フープ、文字の大きな動きを止める。ピンクゴールドを含むカラフルな紙吹雪だけがゆっくりと落下し、収益への感謝と祝福が明確に伝わる画面として保持する。\n\n【モーショングラフィックと音楽】\n背景には、円、長方形、三角形、曲線、太いライン、グリッド、紙吹雪などのシェイプを登場させる。配色は、参照画像のクラッカーに使われているマゼンタ、イエロー、オレンジ、バイオレットに、メタリックなピンクゴールドを組み合わせる。\n\nシェイプは主人公の演技と連動して動かす。フープが円を描くとピンクゴールドの円形波紋が広がり、フープを転がすとピンクゴールドのラインが進行方向へ伸びる。主人公がリープすると三角形や紙吹雪が上方へ飛び、フープをキャッチすると背景のシェイプが一瞬静止してから外側へ広がる。\n\nフープの回転によって発生する光の軌跡、円形波紋、反射光、細いラインは、すべてフープと同じピンクゴールドを基調とする。光の軌跡はフープの動きを短時間だけ強調し、フープ本体が複数に分裂しているようには見せない。\n\n足の踏み込み、ターン、リープの踏み切りと着地、フープの回転、投げ上げ、キャッチを、華やかで力強いエレクトロニック・ミュージックのキック、スネア、ベース、アクセント音へ正確に同期させる。\n\nカメラは正面、俯瞰、ローアングル、人物の周囲を回る滑らかな移動を組み合わせる。可能な限り主人公の全身とフープ全体を同時に画面内へ収め、フープの一部が画面外へ不自然に切れないようにする。\n\n【文字と一貫性】\n新たに表示するキネティック・タイポグラフィーは「X収益」「$96」「Thank You」の3種類だけに限定する。\n\n「X収益」の文字の欠落や別の漢字への変化を起こさない。「$96」はドル記号を省略せず、必ず数字の前へ配置する。「Thank You」は正しい大文字と小文字、2単語の間のスペースを維持する。文字の重複、誤字、変形、別の文字や数字への変化を起こさない。\n\n参照画像に元から含まれている表示は背景画像の一部として扱い、内容を再生成、改変、翻訳、アニメーションさせない。新しく企業名、ロゴ、ユーザー名、ウォーターマーク、字幕を追加しない。\n\nフープは映像全体を通して必ず1本だけにする。フープを複製、分裂、消失、変形させず、完全な円形と一定の大きさを維持する。フープが主人公の身体を不自然に貫通したり、手から瞬間移動したりしない。\n\nフープの色は、映像全体を通して同じメタリック・ピンクゴールドを維持する。通常の黄色い金色、単色のピンク、赤、オレンジなどへ変化させない。照明や角度による自然なローズ、カッパー、シャンパンカラーの反射だけを許可する。\n\n主人公の形状、体格、手足の本数を15秒間一貫させる。手足の変形、身体の分裂、不自然な残像を発生させない。アングルが変わっても、同じ女性、同じ一本のフープ、同じ背景画像を維持する。\n\n黒い女性シルエット、鮮やかなメタリック・ピンクゴールドのフープ、参照画像の通知画面、華やかな紙吹雪、正確なタイポグラフィーを組み合わせ、Xで得られた収益への感謝と喜びが伝わる、スタイリッシュで祝祭感のある高品質な15秒動画に仕上げる。」",
    "promptSourceUrl": "https://x.com/itsumoaokuro/status/2093474780683149442",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2093474780683149442",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-08-28T23:04:17.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093474780683149442.mp4",
    "posterUrl": "/posters/x/2093474780683149442.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "anime"
    ],
    "scenes": [
      "product",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ヤッター！！\nX収益入ってたー😆\n\nこれもいつもポストを見てくれる皆さまのおかげです✨\nありがとうございます🥹\n\n感謝の舞をご覧ください〜\n\n動画はMiniMax H3で作成\n@Hailuo_AI \n\nプロンプト↓\n「添付した参照画像を、映像全体の背景として使用する。参照画像を別の競技場や抽象背景へ置き換えず、画面全体を覆う大きな背景グラフィックとして維持する。元画像の黒い背景、白とグレーの通知表示、左側のカラフルなクラッカーのアイコンを活かす。\n\n背景画像に含まれる文字やアイコンを描き直したり、別の内容へ変更したりしない。背景には、わずかなズーム、横方向への移動、明るさの変化、軽いぼかし、奥行き表現のみを加える。主人公と新しいタイポグラフィーを目立たせるため、背景画像を少し暗くするが、参照画像であることが認識できる状態を保つ。\n\nMiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによる、新体操のフープ演技動画を制作する。\n\n主人公は、全身が黒一色で描かれた一人の女性のシルエット。顔、髪、肌、衣装などの細部は描かず、プロの新体操選手らしい、しなやかで洗練された身体のラインを持つ美しいシルエットとして表現する。\n\n黒い人物が背景に埋もれないよう、身体の輪郭に細い白、マゼンタ、バイオレットのリムライトを加える。ただし、身体の内部は黒一色を維持し、顔や衣装のディテールは描かない。\n\n主人公は一本の鮮やかなメタリック・ピンクゴールドのフープを使用する。フープは細く軽量な新体操用フープで、映像全体を通して完全な円形、一定の太さと大きさを維持する。\n\nフープの色は、淡いローズピンクを基調として、銅色とシャンパンカラーの反射、明るいピンクホワイトのハイライトを持つ、高級感のあるピンクゴールドとする。単純なピンク色やプラスチック素材にはせず、照明や角度によってローズ、カッパー、淡い金属光沢が美しく変化する質感として表現する。\n\n【0.0～3.5秒：X収益】\n映像は、ピンクゴールドのフープを画面手前に極端に大きく配置した、大胆な構図から始まる。カメラはフープの円の内側を通して、その奥に立つ黒い女性シルエットと参照画像の背景を捉える。\n\n主人公はフープの中心に身体を収めた状態から、腕を大きく伸ばしてフープを回転させる。フープを片手で垂直に回しながら身体の周囲へ滑らかに移動させ、そのままフープの中を片脚ずつ優雅に通り抜ける。\n\n最初の強いビートに合わせて、フープの内側に「X収益」の文字を正確に表示する。「X収益」はフープの回転に呼応して画面奥から拡大しながら現れ、最後は正面を向いて明確に読める状態になる。\n\nフープが床へ軽く触れると、接地点からピンクゴールドの円形波紋が広がる。参照画像左側のクラッカーを連想させる、小さな三角形、円、曲線のシェイプがビートに合わせて飛び出す。\n\n【3.5～8.5秒：$96】\n主人公はフープを手首の周囲で高速回転させ、その回転を止めずに腕から肩、背中、反対側の腕へ転がす。フープの自然な慣性と遠心力を維持し、身体の表面に沿って滑らかに移動させる。\n\n続いて、主人公はフープを垂直に転がしながら前方へ送り、その軌道に合わせて大きく脚を伸ばしたスプリットリープを行う。空中で身体と脚の美しいラインを見せ、着地と同時に転がってきたフープを片手で正確に受け止める。\n\nフープの中心に「$96」の文字を正確に表示する。ドル記号を数字の前に置き、スペースを入れず、必ず「$96」と表示する。フープが回転している間も文字自体は正面を向き、明確に読み取れる瞬間を作る。\n\n「$96」は音楽のビートに合わせて拡大、縮小、前後方向への移動を繰り返す。数字がフープの外へ飛び出したり、別の数字へ変化したり、桁が増減したりしない。\n\nカメラは正面からの全身ショットに加え、斜め上から見下ろす俯瞰ショットへ切り替える。俯瞰ショットでは、床面を転がるピンクゴールドのフープの円形、主人公の足運び、背景上を流れる円やラインを印象的に見せる。\n\n【8.5～12.5秒：Thank You】\n主人公はピンクゴールドのフープを頭上へ高く投げ上げる。フープは空中で正しい軸を中心に回転し、完全な円形を維持する。\n\nフープが空中へ上昇する間、主人公は美しいピボットターンを連続して行う。ターンから大きなリープへ移行し、空中を通過するフープの内側へ身体を合わせる。着地後にもう一度身体を回転させ、落下してくる同じフープを片手で正確にキャッチする。\n\nフープが最も高い位置へ到達した瞬間、「Thank You」の文字がその内側から現れる。「Thank」と「You」の間に正しいスペースを入れ、大文字と小文字の組み合わせを正確に維持する。\n\n「Thank You」は、感謝と祝福を感じさせる優雅な動きで画面奥から手前へ近づく。文字の周囲では、参照画像のクラッカーと調和するマゼンタ、イエロー、オレンジ、バイオレット、ピンクゴールドの紙吹雪と小さなシェイプが広がる。\n\nカメラは斜め下から見上げるローアングルへ移動し、空中のピンクゴールドのフープ、主人公の伸びやかな身体、背景に浮かぶ「Thank You」を同時に捉える。カメラアングルが変わっても、主人公とフープの位置関係、回転方向、動作の連続性を維持する。\n\n【12.5～15.0秒：フィナーレ】\n主人公はキャッチしたフープを大きく一回転させ、最後にフープを垂直に立てた状態で片手に保持する。もう片方の腕と片脚を美しく伸ばし、フープと身体の曲線が調和する、上品で自然なフィニッシュを見せる。\n\n最後の画面では、ピンクゴールドのフープを中央へ大きく配置する。フープの内側に「$96」、フープの上方に「X収益」、下方に「Thank You」を読みやすく整列させる。\n\n3つのタイポグラフィーと主人公が、参照画像の重要な部分を完全に覆わないよう、奥行きと配置を調整する。参照画像は少しぼかした状態で背景に残し、その前に黒い主人公、ピンクゴールドのフープ、3つの文字を鮮明に表示する。\n\n最後の約0.8秒は、主人公、フープ、文字の大きな動きを止める。ピンクゴールドを含むカラフルな紙吹雪だけがゆっくりと落下し、収益への感謝と祝福が明確に伝わる画面として保持する。\n\n【モーショングラフィックと音楽】\n背景には、円、長方形、三角形、曲線、太いライン、グリッド、紙吹雪などのシェイプを登場させる。配色は、参照画像のクラッカーに使われているマゼンタ、イエロー、オレンジ、バイオレットに、メタリックなピンクゴールドを組み合わせる。\n\nシェイプは主人公の演技と連動して動かす。フープが円を描くとピンクゴールドの円形波紋が広がり、フープを転がすとピンクゴールドのラインが進行方向へ伸びる。主人公がリープすると三角形や紙吹雪が上方へ飛び、フープをキャッチすると背景のシェイプが一瞬静止してから外側へ広がる。\n\nフープの回転によって発生する光の軌跡、円形波紋、反射光、細いラインは、すべてフープと同じピンクゴールドを基調とする。光の軌跡はフープの動きを短時間だけ強調し、フープ本体が複数に分裂しているようには見せない。\n\n足の踏み込み、ターン、リープの踏み切りと着地、フープの回転、投げ上げ、キャッチを、華やかで力強いエレクトロニック・ミュージックのキック、スネア、ベース、アクセント音へ正確に同期させる。\n\nカメラは正面、俯瞰、ローアングル、人物の周囲を回る滑らかな移動を組み合わせる。可能な限り主人公の全身とフープ全体を同時に画面内へ収め、フープの一部が画面外へ不自然に切れないようにする。\n\n【文字と一貫性】\n新たに表示するキネティック・タイポグラフィーは「X収益」「$96」「Thank You」の3種類だけに限定する。\n\n「X収益」の文字の欠落や別の漢字への変化を起こさない。「$96」はドル記号を省略せず、必ず数字の前へ配置する。「Thank You」は正しい大文字と小文字、2単語の間のスペースを維持する。文字の重複、誤字、変形、別の文字や数字への変化を起こさない。\n\n参照画像に元から含まれている表示は背景画像の一部として扱い、内容を再生成、改変、翻訳、アニメーションさせない。新しく企業名、ロゴ、ユーザー名、ウォーターマーク、字幕を追加しない。\n\nフープは映像全体を通して必ず1本だけにする。フープを複製、分裂、消失、変形させず、完全な円形と一定の大きさを維持する。フープが主人公の身体を不自然に貫通したり、手から瞬間移動したりしない。\n\nフープの色は、映像全体を通して同じメタリック・ピンクゴールドを維持する。通常の黄色い金色、単色のピンク、赤、オレンジなどへ変化させない。照明や角度による自然なローズ、カッパー、シャンパンカラーの反射だけを許可する。\n\n主人公の形状、体格、手足の本数を15秒間一貫させる。手足の変形、身体の分裂、不自然な残像を発生させない。アングルが変わっても、同じ女性、同じ一本のフープ、同じ背景画像を維持する。\n\n黒い女性シルエット、鮮やかなメタリック・ピンクゴールドのフープ、参照画像の通知画面、華やかな紙吹雪、正確なタイポグラフィーを組み合わせ、Xで得られた収益への感謝と喜びが伝わる、スタイリッシュで祝祭感のある高品質な15秒動画に仕上げる。」",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 15,
      "views": 428,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093474752379879812",
    "title": "小鸡切开一颗固执的西洋梨",
    "titleEn": "A Tiny Chick Slices a Stubborn Pear",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "毛绒小鸡在厨房案板前举刀切梨，反差角色与食材动作形成短促喜剧。 原生视频规格为 10 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fluffy chick raises a knife over a pear on a kitchen board, turning the contrast between character and ingredient into compact comedy. Native-video output: 10s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/WorkChiharu/status/2093474752379879812",
    "sourceLabel": "X 原帖 · @WorkChiharu",
    "author": "ちはる｜AIアニメ × もちぴ｜Canva ⭐",
    "publishedAt": "2026-08-28T23:04:10.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093474752379879812.mp4",
    "posterUrl": "/posters/x/2093474752379879812.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「鳥取の梨を作って」と何度お願いしても、\nどうしても西洋梨になるMiniMax H3😂🍐\n「丸い和梨！西洋梨じゃない！」まで書いたのにこの強い意志🤣\nAI動画、思い通りにならないところまで面白い。\n\n#AI動画 #生成AI #MiniMaxH3 #AIあるある #もちぴ",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 44,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093473871261499887",
    "title": "橙黑针织角色穿行城市街头",
    "titleEn": "An Orange-and-Black Character Walks Through the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙黑针织造型的人物穿过城市街头，多角度跟拍突出服装纹理和步态。 原生视频规格为 45 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character in orange-and-black knitwear crosses city streets as multi-angle tracking emphasizes texture and gait. Native-video output: 45s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Banyu_Biroeee/status/2093473871261499887",
    "sourceLabel": "X 原帖 · @Banyu_Biroeee",
    "author": "Rian Andria",
    "publishedAt": "2026-08-28T23:00:40.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093473871261499887.mp4",
    "posterUrl": "/posters/x/2093473871261499887.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Playin' around with Minimax H3 from @Hailuo_AI @MiniMax_AI \n\n#MinimaxDesign #MinimaxH3 #HailuoAI #AIVideo",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 42,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093473703917228148",
    "title": "长耳黑猫在酒吧端坐",
    "titleEn": "A Long-Eared Black Cat Sits at the Bar",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "长耳黑猫坐在酒吧柜台前，饮品、暖灯与克制表情组成拟人角色镜头。 原生视频规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-eared black cat sits at a bar counter as drinks, warm lights, and a restrained expression form an anthropomorphic character shot. Native-video output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Sadal_Quari/status/2093473703917228148",
    "sourceLabel": "X 原帖 · @Sadal_Quari",
    "author": "エルナ｜AIと新潟メモ",
    "publishedAt": "2026-08-28T23:00:00.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093473703917228148.mp4",
    "posterUrl": "/posters/x/2093473703917228148.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "RTX 5070 12GB＋ComfyUIでMiniMax H3のT2V\n\n参照画像なし、プロンプトだけ\n「異様に耳の長い黒猫」を生成\n\nカクテル飲んで酔っ払う🐈‍⬛\n\n#MiniMaxH3 #ComfyUI #LocalAI #AIvideo",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 88,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093466155373748711",
    "title": "失控演播室播报开源模型新闻",
    "titleEn": "A Chaotic Newsroom Recaps Open-Weight AI",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卡通新闻主播在演播室播报模型动态，快速对白和角色反应构成长篇讽刺新闻。 原生视频规格为 62 秒 · 1888×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Cartoon anchors report model news in a studio, using rapid dialogue and character reactions for an extended satirical broadcast. Native-video output: 62s · 1888×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FeatherlessAI/status/2093466155373748711",
    "sourceLabel": "X 原帖 · @FeatherlessAI",
    "author": "Featherless AI",
    "publishedAt": "2026-08-28T22:30:00.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093466155373748711.mp4",
    "posterUrl": "/posters/x/2093466155373748711.jpg",
    "duration": 62,
    "aspectRatio": "landscape",
    "resolution": "1888×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Last month in ai was openweightpilled. \nRendered with Minimax H3 on Featherless AI \nh/t @Siliconmania",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 6,
      "views": 96,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093465029261590971",
    "title": "金发歌手穿梭乐队、车厢与屋顶",
    "titleEn": "A Blonde Singer Moves Between a Band, Car, and Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发歌手从乐队表演切入车厢和屋顶，多场景镜头围绕同一首歌连续推进。 原生视频规格为 28 秒 · 1280×732。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde singer moves from a band performance into a car and rooftop, carrying one song through multiple connected settings. Native-video output: 28s · 1280×732. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mirthtime/status/2093465029261590971",
    "sourceLabel": "X 原帖 · @mirthtime",
    "author": "⚡️Mirthtime ⚡️",
    "publishedAt": "2026-08-28T22:25:32.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093465029261590971.mp4",
    "posterUrl": "/posters/x/2093465029261590971.jpg",
    "duration": 28,
    "aspectRatio": "landscape",
    "resolution": "1280×732",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I am really digging what’s possible with Minimax H3 locally. Ran this on my 5090 in about 30 mins.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 109,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093463961123590502",
    "title": "声音波形收束为个人标题卡",
    "titleEn": "Sound Waves Resolve into a Personal Title Card",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "发光波形、网格与字母逐步聚合，最后收束成简洁的个人品牌标题。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Glowing waveforms, grids, and letters assemble step by step before resolving into a concise personal brand title. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_animer/status/2093463961123590502",
    "sourceLabel": "X 原帖 · @ai_animer",
    "author": "雨雲(Amagumo)806 ｜ AI Animer",
    "publishedAt": "2026-08-28T22:21:17.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093463961123590502.mp4",
    "posterUrl": "/posters/x/2093463961123590502.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "H3で自己PRを15秒一発生成！✨\n\n文字と映像を一緒に組めるのが強くて、自己紹介や商品紹介で大活躍しそうなモデルですね😊\n\n#MiniMaxH3 #MiniMaxDesign @Hailuo_AI",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 16,
      "views": 371,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093459701707284700",
    "title": "气泡与鲸鱼最终汇成“梦”",
    "titleEn": "Bubbles and a Whale Resolve into the Word Dream",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "气泡、鲸鱼和云层在超现实空间中展开，最终汇聚成主题文字“梦”。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Bubbles, a whale, and clouds unfold through a surreal space before converging into the themed character for dream. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "**【テーマ】＝夢**  この【テーマ】だけを変更して使用する。  【テーマ】の文字、言葉、意味、感情、音の響き、文化的イメージ、象徴、自然現象、物体、場所、時代感などを最初に分析し、そのテーマに最も適した独自の映像世界を設計すること。  毎回同じ演出を使用してはいけない。  【テーマ】に応じて、以下をすべて自動的に変更すること。  色彩設計 背景世界 登場する物体・モチーフ 材質・テクスチャ ライティング カメラワーク モーション 変形方法 粒子・煙・液体・光などのVFX 映像テンポ 空気感 音響演出 ラストの文字表現  【テーマ】から最も自然に連想される世界を作り、映像そのものがテーマの意味を表現すること。  単純にテーマ文字の周囲へ装飾を置くだけではなく、**世界そのものが【テーマ】から生まれ、最後にはその世界全体が【テーマ】の文字または形へ収束する構成**にする。  15秒の基本構造のみ共通とする。  0.00–3.00秒｜謎 【テーマ】を直接見せず、その世界を象徴する断片・物体・現象から始める。  3.00–7.00秒｜世界展開 【テーマ】から連想した独自世界が高速に広がる。カメラもその世界の特徴に合わせて変化させる。  7.00–11.00秒｜最大の見せ場 【テーマ】を最も強く象徴する現象、存在、変形、アクションを発生させる。毎回テーマ固有の演出を考える。  11.00–13.50秒｜収束 それまで登場した世界、物体、光、構造、粒子などが一斉に集まり始める。  13.50–15.00秒｜完成 カメラが引くことで、それまでの世界全体が【テーマ】を形作っていたことが初めて判明する。  **最重要ルール：** 【テーマ】以外に固定されたモチーフや色は存在しない。 赤・金・中国風・炎・氷・星などの特定演出を毎回使わない。 過去のテーマの演出を流用しない。 入力された【テーマ】ごとに、ゼロから最適な演出コンセプトを考える。 同じテンプレート感を残すのは「世界が最後にテーマへ変わる」という構造だけ。",
    "promptSourceUrl": "https://x.com/su_nagomi/status/2093459701707284700",
    "sourceUrl": "https://x.com/su_nagomi/status/2093459701707284700",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-08-28T22:04:22.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093459701707284700.mp4",
    "posterUrl": "/posters/x/2093459701707284700.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "stylized"
    ],
    "scenes": [
      "product",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "テーマ「夢」\n前回の龍から夢にテーマ変えたらまた違った映像できました！プロンプトの初めの文字変えるだけでOK！ぜひやってみてね！\n\n#MinimaxH3 で動画作成 @Hailuo_AI \n\n～プロンプト～\n**【テーマ】＝夢**  この【テーマ】だけを変更して使用する。  【テーマ】の文字、言葉、意味、感情、音の響き、文化的イメージ、象徴、自然現象、物体、場所、時代感などを最初に分析し、そのテーマに最も適した独自の映像世界を設計すること。  毎回同じ演出を使用してはいけない。  【テーマ】に応じて、以下をすべて自動的に変更すること。  色彩設計 背景世界 登場する物体・モチーフ 材質・テクスチャ ライティング カメラワーク モーション 変形方法 粒子・煙・液体・光などのVFX 映像テンポ 空気感 音響演出 ラストの文字表現  【テーマ】から最も自然に連想される世界を作り、映像そのものがテーマの意味を表現すること。  単純にテーマ文字の周囲へ装飾を置くだけではなく、**世界そのものが【テーマ】から生まれ、最後にはその世界全体が【テーマ】の文字または形へ収束する構成**にする。  15秒の基本構造のみ共通とする。  0.00–3.00秒｜謎 【テーマ】を直接見せず、その世界を象徴する断片・物体・現象から始める。  3.00–7.00秒｜世界展開 【テーマ】から連想した独自世界が高速に広がる。カメラもその世界の特徴に合わせて変化させる。  7.00–11.00秒｜最大の見せ場 【テーマ】を最も強く象徴する現象、存在、変形、アクションを発生させる。毎回テーマ固有の演出を考える。  11.00–13.50秒｜収束 それまで登場した世界、物体、光、構造、粒子などが一斉に集まり始める。  13.50–15.00秒｜完成 カメラが引くことで、それまでの世界全体が【テーマ】を形作っていたことが初めて判明する。  **最重要ルール：** 【テーマ】以外に固定されたモチーフや色は存在しない。 赤・金・中国風・炎・氷・星などの特定演出を毎回使わない。 過去のテーマの演出を流用しない。 入力された【テーマ】ごとに、ゼロから最適な演出コンセプトを考える。 同じテンプレート感を残すのは「世界が最後にテーマへ変わる」という構造だけ。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 225,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093459333980037397",
    "title": "水面人声与雨巷串成长篇音乐叙事",
    "titleEn": "Water, Vocals, and a Rainy Alley Form a Long Music Story",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水面人物、室内人声和雨巷行走连续交替，口型与场景过渡支撑长篇音乐叙事。 原生视频规格为 170 秒 · 1280×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure on water, indoor vocals, and a walk through a rainy alley alternate as lip sync and transitions support a long music narrative. Native-video output: 170s · 1280×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mazzo/status/2093459333980037397",
    "sourceLabel": "X 原帖 · @mazzo",
    "author": "Koya Matsuo",
    "publishedAt": "2026-08-28T22:02:54.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093459333980037397.mp4",
    "posterUrl": "/posters/x/2093459333980037397.jpg",
    "duration": 170,
    "aspectRatio": "landscape",
    "resolution": "1280×704",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のリップシンクは独自実装していたのですが、 @uzuki425 さんが開発されたhttps://github.com/animede/Minimax-H3-lipsync-mv の成果を取り込むことで、リップシンク精度が大幅に改善されました。ありがとうございます。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 429,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093459105646256624",
    "title": "雨夜街景并排比较材质细节",
    "titleEn": "A Rainy Street Compares Material Detail Side by Side",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "并排画面中的人物走过湿润街道，反射、服装和背景细节用于直观比较。 原生视频规格为 5 秒 · 1080×1224。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character walks through a wet street in parallel frames, making reflections, clothing, and background detail easy to compare. Native-video output: 5s · 1080×1224. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DeepForest_Inc/status/2093459105646256624",
    "sourceLabel": "X 原帖 · @DeepForest_Inc",
    "author": "🟢DeepForest_Studio_AI🟢",
    "publishedAt": "2026-08-28T22:02:00.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093459105646256624.mp4",
    "posterUrl": "/posters/x/2093459105646256624.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1080×1224",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 vs MiniMax-H3 + Z-Image Graft🌃\n\nMiniMax-H3にZ-Imageの一部の特性を移植したモデルと比較✨\n\nprompt / seed / settings は同一\n\n壁、金属、濡れた路面、反射などの細かな質感差を見るテスト動画\n\n自分ではかなり分かりづらいけど、並べると少し傾向は見えるかも🤔\n\n#MiniMaxH3 #ComfyUI",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 14,
      "views": 892,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093458879107747996",
    "title": "客厅科学宅围绕奇怪提议争论",
    "titleEn": "A Living-Room Sitcom Debates an Odd Proposal",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "几名科学宅在客厅围绕奇怪提议展开对白，反应镜头和口型维持情景喜剧节奏。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Several science-minded roommates debate an odd proposal in a living room as reaction shots and lip sync sustain the sitcom rhythm. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altryne/status/2093458879107747996",
    "sourceLabel": "X 原帖 · @altryne",
    "author": "Alex Volkov",
    "publishedAt": "2026-08-28T22:01:06.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093458879107747996.mp4",
    "posterUrl": "/posters/x/2093458879107747996.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Everyone's sharing their grok bot templates. This one is fun!\n\nCreated a @bot that will use any tweet, research and write a tiny cold opener sitcom scene, and render it (super fast) with @fal Minimax H3 Max! \n\nClick the link  to try this (needs @fal and @grok API keys)",
    "engagement": {
      "replies": 7,
      "reposts": 1,
      "likes": 14,
      "views": 1351,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093458611527950391",
    "title": "银发学生举起发光玩具枪",
    "titleEn": "A Silver-Haired Student Raises a Glowing Blaster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发学生在校园与霓虹光效间举起发光枪械，特写和动作镜头组成动画片段。 原生视频规格为 7 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired student raises a glowing blaster between school scenery and neon effects as close-ups and action shots form an animated sequence. Native-video output: 7s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yonaguniUDON/status/2093458611527950391",
    "sourceLabel": "X 原帖 · @yonaguniUDON",
    "author": "蛸壺_AIラクルゥ開発",
    "publishedAt": "2026-08-28T22:00:02.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093458611527950391.mp4",
    "posterUrl": "/posters/x/2093458611527950391.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ばきゅんばきゅん！\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 10,
      "views": 82,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093457990661853670",
    "title": "镜头穿墙追随人物走过多间房",
    "titleEn": "The Camera Follows a Character Through Multiple Rooms",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "镜头从正面跟随转为侧向穿墙，再升至垂直俯视，连续呈现人物走过多间房的路径。 原生视频规格为 15 秒 · 1080×1330。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The camera shifts from frontal tracking to lateral wall-crossing, then rises into a vertical overhead view while following a character through connected rooms. Native-video output: 15s · 1080×1330. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use the attached image as the exact starting frame and visual reference. Preserve the same woman exactly throughout the entire video: same face, short black bob haircut, bangs, facial features, body proportions, casual black outfit, accessories, and overall appearance. Do not change her identity, hairstyle, clothing, age, or body shape.\n\nCreate a cinematic live-action movie sequence of the woman quietly walking through a dark modern apartment. The main focus is sophisticated continuous camera movement and architectural composition. Her walking pace remains calm, natural, and steady.\n\nBegin with a medium frontal shot as she walks slowly through the room. The camera smoothly moves around her and transitions into a lateral tracking shot, traveling perfectly parallel to her movement. Show her from the side while the interior slides naturally behind her with realistic parallax. Use subtle cinematic motion, shallow depth of field, dark concrete textures, muted black and gray interiors, soft practical lamps, deep shadows, realistic reflections, and natural lighting.\n\nWithout cutting, continue tracking sideways. The camera now behaves as if the apartment has been sliced open like an architectural cross-section. The woman keeps walking naturally from room to room while the camera travels horizontally alongside her.\n\nThe camera can physically pass through walls.\n\nWhenever a wall enters the camera's path, do not stop, cut, teleport, or change scenes. The wall briefly fills the foreground as the camera smoothly travels directly through its thickness, immediately revealing the next connected room on the other side. No portals, dissolves, magical transitions, or visual effects.\n\nContinue this movement across several connected spaces: bedroom → doorway → narrow corridor → living room → another room. Maintain believable architectural continuity. The result should resemble watching a realistic dollhouse cross-section from the side, with the woman continuously walking through the building.\n\nGradually pull farther away while continuing the sideways tracking movement. Reveal multiple rooms simultaneously. Walls, doors, corridors, furniture, lamps, windows, and shadows form increasingly geometric compositions. The woman becomes smaller within the frame but remains the only moving person.\n\nThen begin a dramatic but extremely smooth camera transition.\n\nWhile she continues walking, the camera rises high above the apartment. At the same time, slowly rotate the camera downward until it reaches an exact 90-degree bird's-eye view.\n\nThe ceiling visually disappears from the composition, revealing the apartment layout beneath.\n\nFinish with a PERFECTLY VERTICAL TOP-DOWN SHOT.\n\nThe camera must point straight down at the floor with no perspective tilt, like viewing an architectural floor plan from directly above.\n\nFrom this height, the entire apartment becomes a striking geometric composition: rectangular rooms, narrow corridors, walls, doors, beds, tables, sofas, lights, and shadows create abstract graphic shapes while remaining completely photorealistic.\n\nThe woman now appears small within the architecture. Continue following her as she walks through the floor-plan-like environment: across one room, through a doorway, along a corridor, and into another room.\n\nHer movement through the geometric apartment should become the visual centerpiece, almost like a tiny human figure moving through an architectural diagram.\n\nMake the entire sequence feel like ONE CONTINUOUS CINEMATIC TAKE:\nfrontal walking shot → camera rotates to her side → lateral tracking → camera passes through walls → multiple rooms revealed in cross-section → camera pulls back → camera rises above the building → exact 90-degree overhead view → woman continues walking through the geometric floor plan.\n\nMaintain realistic spatial continuity at all times. Every room must feel physically connected. Doors and corridors should logically connect the spaces she enters.\n\nCamera movement should be extremely smooth, deliberate, precise, and elegant, like a technically impossible prestige-film tracking shot.\n\nVisual style: premium psychological drama, architectural cinematography, subtle surrealism, dark minimalist apartment, realistic live-action photography, cinematic lighting, deep shadows, desaturated black and gray palette, occasional warm practical lamps, natural skin and fabric texture, realistic walking physics, sophisticated composition, high production value.\n\nIMPORTANT CHARACTER CONSISTENCY:\nOnly ONE woman exists in the entire sequence.\nNever duplicate her.\nNever add another person.\nKeep her face and hairstyle identical to the reference.\nKeep her casual outfit identical throughout.\nNo costume transformation.\nNo facial changes.\nNo body changes.\n\nIMPORTANT CAMERA RULES:\nNo random cuts.\nNo montage.\nNo teleportation.\nNo sudden location changes.\nNo portals.\nNo magical wall effects.\nNo camera shake.\nNo abrupt zooms.\nNo impossible character movement.\nThe CAMERA passes through the architecture while the woman physically walks through normal doors and corridors.\nThe final overhead view must be exactly 90 degrees downward and visually resemble a geometric architectural floor plan.",
    "promptSourceUrl": "https://x.com/varts_works/status/2093457994344526234",
    "sourceUrl": "https://x.com/varts_works/status/2093457990661853670",
    "sourceLabel": "X 原帖 · @varts_works",
    "author": "rui｜∫varts (シュバルツ)",
    "publishedAt": "2026-08-28T21:57:34.000Z",
    "addedAt": "2026-08-29T00:19:15.551Z",
    "mediaUrl": "/media/x-2093457990661853670.mp4",
    "posterUrl": "/posters/x/2093457990661853670.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1330",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【 MiniMax H3 】One image → A cinematic one-take camera journey through multiple rooms🎥\n\n1枚の画像から、映画のワンシーンのような特殊なカメラワークを作るためのプロンプトです🎬\n\n横から人物を追いかけ、壁を越えて次の部屋へ🏠\n最後はカメラが上昇し、部屋全体を真上から見下ろす「間取り図」のような構図\n\nCamera sequence:\nFront tracking → Side tracking → Through walls → Multiple rooms → Camera rises → 90° top-down floor-plan shot\n\nCreated with: @Hailuo_AI #MiniMaxDesign #MiniMaxH3 \n\nPrompt ↓",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 10,
      "views": 402,
      "snapshotAt": "2026-08-29T00:19:15.551Z"
    },
    "approvedAt": "2026-08-29T00:19:15.551Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093715484910731416",
    "title": "金属碎片高速拼成“VELOCITY”",
    "titleEn": "Metal Fragments Assemble the Word VELOCITY at Full Speed",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "镜头穿过高速金属碎片、火花和冲击波，最终拉远揭示完整的立体标题。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The camera races through metallic fragments, sparks, and shock waves before pulling back to reveal a complete three-dimensional title. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "High-octane cinematic title sequence built entirely around speed: begin with an extreme close-up of polished metal traveling through darkness so quickly that only reflections and sparks are visible. The camera accelerates alongside it as enormous metallic components streak past like projectiles, generating sonic shockwaves and pressure distortion. Each component violently decelerates at the center of frame, rotating and locking into place with precise mechanical impacts while additional fragments arrive from every direction. The camera threads between assembling letterforms at extreme speed, passing through showers of sparks and glowing friction trails until a final impossible pullback reveals that every moving component has constructed the enormous word \"VELOCITY\". A supersonic pressure wave travels from the first letter to the last, bending the surrounding atmosphere and stripping glowing particles from the metal surfaces. The camera instantly launches backward as the entire title accelerates forward through the frame like a massive projectile, leaving only turbulent smoke and incandescent trails behind. Hyper-dynamic cinematography, physically convincing motion blur, metallic micro-detail, sonic-boom distortion, premium theatrical title VFX.",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2093715484910731416",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2093715484910731416",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-08-29T15:00:45.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093715484910731416.mp4",
    "posterUrl": "/posters/x/2093715484910731416.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The more I experiment with Minimax H3, the more I love it for title VFX.\nThis time I focused entirely on speed + typography + aggressive camera choreography.\n\nThe result goes HARD. ⚡️\nPROMPT 👇\nHigh-octane cinematic title sequence built entirely around speed: begin with an extreme close-up of polished metal traveling through darkness so quickly that only reflections and sparks are visible. The camera accelerates alongside it as enormous metallic components streak past like projectiles, generating sonic shockwaves and pressure distortion. Each component violently decelerates at the center of frame, rotating and locking into place with precise mechanical impacts while additional fragments arrive from every direction. The camera threads between assembling letterforms at extreme speed, passing through showers of sparks and glowing friction trails until a final impossible pullback reveals that every moving component has constructed the enormous word \"VELOCITY\". A supersonic pressure wave travels from the first letter to the last, bending the surrounding atmosphere and stripping glowing particles from the metal surfaces. The camera instantly launches backward as the entire title accelerates forward through the frame like a massive projectile, leaving only turbulent smoke and incandescent trails behind. Hyper-dynamic cinematography, physically convincing motion blur, metallic micro-detail, sonic-boom distortion, premium theatrical title VFX.",
    "engagement": {
      "replies": 0,
      "reposts": 6,
      "likes": 18,
      "views": 1444,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093857142285717681",
    "title": "五秒勇者旅程从村庄冲到魔王决战",
    "titleEn": "A Five-Second Hero Journey Races from Village to Demon King",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "动画勇者离开村庄、迎战冰龙并击破魔王，五秒内压缩成连续冒险。 原生视频规格为 5 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated hero leaves a village, confronts an ice dragon, and defeats a demon king in a five-second adventure. Native-video output: 5s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/harogetter/status/2093857142285717681",
    "sourceLabel": "X 原帖 · @harogetter",
    "author": "ハロゲ",
    "publishedAt": "2026-08-30T00:23:39.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093857142285717681.mp4",
    "posterUrl": "/posters/x/2093857142285717681.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "5秒で旅立ちから魔王を倒すアニメ\n\nhttps://fal.ai/tools/minimax-h3-max",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 21,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093856961876119569",
    "title": "金发厨师用火焰完成料理",
    "titleEn": "A Blonde Chef Finishes a Dish with Fire",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发动画厨师翻锅、点火并完成装盘，手绘质感贯穿快速料理动作。 原生视频规格为 5 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde animated chef flips a pan, ignites the dish, and finishes plating with a consistent hand-drawn look. Native-video output: 5s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/uwatenageCH/status/2093856961876119569",
    "sourceLabel": "X 原帖 · @uwatenageCH",
    "author": "うわてなげ@ケミカルクッカー亮の人",
    "publishedAt": "2026-08-30T00:22:56.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093856961876119569.mp4",
    "posterUrl": "/posters/x/2093856961876119569.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "噂のminimax-H3-maxお試ししてみましたが、セル画っぽさ出てますね、久しぶりの動画ミカさん😀しかし5秒とはいえ生成が爆速過ぎる\nhttps://fal.ai/tools/minimax-h3-max",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 10,
      "views": 112,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093856304867807413",
    "title": "校园日常卷入异世界长篇冒险",
    "titleEn": "A School-Day Story Opens into a Long Fantasy Adventure",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "放学街景、发光变身与多人异世界战斗被串成长篇叙事，角色在现实与幻想场景间推进。 原生视频规格为 257 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "After-school streets, a glowing transformation, and ensemble fantasy battles connect into a long narrative moving between reality and another world. Native-video output: 257s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DSarukun/status/2093856304867807413",
    "sourceLabel": "X 原帖 · @DSarukun",
    "author": "DegiSaruKun",
    "publishedAt": "2026-08-30T00:20:19.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093856304867807413.mp4",
    "posterUrl": "/posters/x/2093856304867807413.jpg",
    "duration": 257,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🌱🎉 #CNPLand ㊗️3周年🎉🌱\n＼✨おめでとうございます✨／\n\n今年は思い切って、\nおハヤ紫🏫×異世界転生風🥷×CNPLand🌈\nで､実写風AI動画を作成しちゃいましたよ～🎥✨\n\nシャックさん🦈に教えてもらった #MiniMaxH3 と､またべえさん🐧のBrain教材と､ClaudeCode君達のおかげでなんとか完成できました🙌",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 17,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093853841254953237",
    "title": "三名少年迎战紫色城市裂隙",
    "titleEn": "Three Heroes Confront a Purple Rift Above the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三名动画角色在城市街头集结，紫色空间裂隙与标题排版组成音乐片头。 原生视频规格为 30 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three animated characters gather on a city street as a purple spatial rift and bold typography form a music-driven opener. Native-video output: 30s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HexxRL/status/2093853841254953237",
    "sourceLabel": "X 原帖 · @HexxRL",
    "author": "Hexx ./",
    "publishedAt": "2026-08-30T00:10:32.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093853841254953237.mp4",
    "posterUrl": "/posters/x/2093853841254953237.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "RESONANCE//BREAK - Wake The World\n\nendless land of fantasy and magic creation. lipsync, voices and angles maxed\n\n@vidmuse_ai Freeform paired with MiniMax H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 49,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093853764503441812",
    "title": "厄瓜多尔地域符号切成动感海报",
    "titleEn": "Ecuadorian Regions Turn into Kinetic Poster Art",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "旋涡、人物与地域标签在鲜艳图形中切换，把海岸、高地和东方地区组合成海报动画。 原生视频规格为 10 秒 · 2504×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Swirls, characters, and regional labels switch through vivid graphics, combining Ecuador’s coast, highlands, and east into animated poster art. Native-video output: 10s · 2504×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiworld_ec/status/2093853764503441812",
    "sourceLabel": "X 原帖 · @aiworld_ec",
    "author": "AI World Ec",
    "publishedAt": "2026-08-30T00:10:14.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093853764503441812.mp4",
    "posterUrl": "/posters/x/2093853764503441812.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2504×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Regiones del Ecuador | Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 16,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093853734975471759",
    "title": "疯狂科学家戴着宽檐帽闯进实验室",
    "titleEn": "A Mad Scientist Enters the Lab in a Wide-Brimmed Hat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "疯狂科学家与同伴在外星实验室争论，夸张表情和卡通口型推动短对白。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A mad scientist and companion argue inside an alien laboratory, driven by exaggerated expressions and cartoon lip sync. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Arn0Stride7/status/2093853734975471759",
    "sourceLabel": "X 原帖 · @Arn0Stride7",
    "author": "NeonStride",
    "publishedAt": "2026-08-30T00:10:06.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093853734975471759.mp4",
    "posterUrl": "/posters/x/2093853734975471759.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is actually fun and cheap",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 16,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093848858917015650",
    "title": "月下哥特少女完成音乐钩子片段",
    "titleEn": "A Gothic Heroine Performs a Moonlit Music Hook",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "哥特少女在满月与暗色舞台间保持造型，演唱口型和镜头切换组成音乐钩子。 原生视频规格为 12 秒 · 832×1472。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A gothic heroine keeps her design across a full moon and dark stage while lip sync and shot changes form a music hook. Native-video output: 12s · 832×1472. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Yuki_mec/status/2093848858917015650",
    "sourceLabel": "X 原帖 · @Yuki_mec",
    "author": "Yuki",
    "publishedAt": "2026-08-29T23:50:44.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093848858917015650.mp4",
    "posterUrl": "/posters/x/2093848858917015650.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "832×1472",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "SUNOのHook動画をMiniMax H３で作成。 CodexからComfyUIに投入。LQ動画からHQ動画へのアップスケールも。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 24,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093848766751375718",
    "title": "黑色剪影走向雾中的金色阶梯",
    "titleEn": "A Dark Silhouette Walks Toward a Golden Stairway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色人影从暗处走向雾中阶梯，远端金光和微小粒子保持梦境般的单镜头氛围。 原生视频规格为 15 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark figure approaches a staircase in mist as distant golden light and particles sustain a dreamlike single-shot mood. Native-video output: 15s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YuuAki101/status/2093848766751375718",
    "sourceLabel": "X 原帖 · @YuuAki101",
    "author": "YuuAki@AIart",
    "publishedAt": "2026-08-29T23:50:22.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093848766751375718.mp4",
    "posterUrl": "/posters/x/2093848766751375718.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AI境界のトロンプルイユ\n\nということで動かしてみた(*´∀｀*)\n\n#MiniMaxH3\n@Hailuo_AI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 73,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093844620149182464",
    "title": "白色团子吉祥物给串团子淋酱",
    "titleEn": "A White Mascot Sauces a Plate of Dango",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "白色团子吉祥物端起壶为串团子淋上酱汁，竖屏近景突出食物与可爱动作。 原生视频规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white dumpling mascot pours sauce over skewered dango in a vertical close-up focused on food and cute motion. Native-video output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/WorkChiharu/status/2093844620149182464",
    "sourceLabel": "X 原帖 · @WorkChiharu",
    "author": "ちはる｜AIアニメ × もちぴ｜Canva ⭐",
    "publishedAt": "2026-08-29T23:33:53.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093844620149182464.mp4",
    "posterUrl": "/posters/x/2093844620149182464.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Maxで、もちぴのみたらし団子🍡\n1枚画像から生成してみたら、本当に秒で完成😳\nこの速さすごい‼",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 52,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093843289745289370",
    "title": "声波、圆环与立方体收束成发光标志",
    "titleEn": "Waveforms, Rings, and Cubes Resolve into Glowing Marks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "抽象声波转成圆环，玻璃立方体和光轨再分别收束为发光几何标志。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Abstract waveforms turn into rings while glass cubes and light trails separately resolve into glowing geometric marks. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hAru_mAki_ch/status/2093843289745289370",
    "sourceLabel": "X 原帖 · @hAru_mAki_ch",
    "author": "Maki@Sunwood AI Labs.",
    "publishedAt": "2026-08-29T23:28:36.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093843289745289370.mp4",
    "posterUrl": "/posters/x/2093843289745289370.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "stylized"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max、無料枠でモーショングラフィックも試してみたぞ！！！\n\n今回は実写じゃなくて、\n波形→リング、ガラスキューブ→発光フレーム、\n棒グラフ→円弧の3パターン。\n\n全部5秒・768p・音声付きで生成。\n抽象表現なら動きのつながりは結構いい感じ、、、\n\nただ、最後にロゴっぽい形へ寄ったり、\n最初の要素が段階的に増えたりもする。\nモーショングラフィック用途は、もう少し詰めたい🔥",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 361,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093842563698757935",
    "title": "粉发少女在红色跑车旁轻快转身",
    "titleEn": "A Pink-Haired Girl Turns Beside a Red Sports Car",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "粉发动画少女在住宅车库与红色跑车旁转身挥手，角色和车辆保持连续。 原生视频规格为 5 秒 · 960×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired animated girl turns and waves beside a red sports car in a residential driveway while both remain consistent. Native-video output: 5s · 960×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HetareIvory/status/2093842563698757935",
    "sourceLabel": "X 原帖 · @HetareIvory",
    "author": "あいぼりー",
    "publishedAt": "2026-08-29T23:25:43.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093842563698757935.mp4",
    "posterUrl": "/posters/x/2093842563698757935.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "960×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ののたんさんから教えてもらったfalのMiniMax H3 Max、確かに早い！\n\nうちの子もスムーズに動いてる～！",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 129,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093842404159947111",
    "title": "九套角色造型撕纸切换成游戏宣传片",
    "titleEn": "Nine Character Looks Tear into a Game Promo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "黑红撕纸、霓虹城市、HUD 与多套角色造型高速衔接，形成完整游戏宣传片。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Black-and-red ripped paper, neon city graphics, HUD elements, and multiple character looks connect at speed into a game promo. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MODEL：MiniMax H3｜15秒｜16:9｜24fps\nREFERENCE：[@参照1] ONLY\nSTRUCTURE：9 LOOKS / ONE CONTINUOUS GAME-PV EDIT / EACH LOOK USES A DIFFERENT AE MOTION LANGUAGE\nSTYLE：premium anime game PV × fashion editorial × aggressive After Effects motion design × mixed-media collage × cinematic 2.5D。参照1枚に含まれる9つの完成ビジュアルを「9個のLOOK候補」として読み取り、それぞれの人物外見・衣装・色彩・背景世界・構図の差を維持しながら、15秒の中で高速に連結する。各LOOKは同じテンプレート動作を繰り返さず、必ず異なるAEモーション原理で見せる。\n\nREFERENCE INTERPRETATION / ABSOLUTE CLEANUP：\n[@参照1]は9面をまとめたコンタクトシートだが、映像内にコンタクトシートそのものを表示しない。参照画像上の「01〜09」「LOOK 01〜09」「0.0–1.5s等の時間表記」「FINAL」およびパネル境界線は制作指示用メタ情報としてのみ認識し、生成映像には出さない。ナンバリング、タイムコード、LOOK表記、白い区切り線、3×3グリッドを画面へコピーしない。人物、衣装、髪、色調、背景、タイポの視覚要素だけを抽出。各LOOKはフルスクリーンで展開し、前LOOKの要素を次LOOKの転換素材へ変換する。\n\nCHARACTER / LOOK LOCK：\n参照内の黒髪系キャラクターと銀〜ラベンダー髪系キャラクターを同一作品世界の成人女性キャラクターとして扱う。各LOOKで参照に描かれた顔、目、髪型、髪色、衣装、アクセサリー、メイク、身体比率を尊重し、そのLOOKの途中で別人化させない。LOOKが切り替わる瞬間にのみ参照通りのヘア、衣装、カラー、背景へ変化可能。顔モーフ、二重顔、余分な腕、人物融合は禁止。視線、瞬き、僅かな首振り、呼吸、髪や布の二次動作のみ。走る、歩き続ける、ジャンプ、ダンス、格闘でテンポを作らない。速度はAE、カメラ、レイヤー、タイポ、マスク、都市、光、原画素材から作る。\n\nGLOBAL MOTION RULE：\n15秒を9LOOKへ分けるが静止画スライドショーにはしない。前LOOKが止まる前に次LOOKを0.12〜0.30秒先行させ、FAST BUT FLUIDで連結。各LOOKの主役AE TECHNIQUEは重複禁止。単純CUT、FADE、CROSS DISSOLVE、均等GRID、カードSLIDE、同じZOOM反復は禁止。完成KVとして一瞬読めても0.7秒以上静止させない。\n\nTEXT CONTROL：\n参照画像に含まれる番号、時間、LOOK、FINALなどの管理文字は全削除。主要タイポは「H3」「MiniMax」「Design」のみ。字幕として下部へ置かない。巨大前景、MASK、TRACK MATTE、3D奥行き、壁面印刷、光学的遮蔽物として使用。ランダム英数字、日本語、中国語、偽UI文字、架空ブランド、追加ロゴ、ウォーターマークは禁止。\n\n0.0–1.55s｜LOOK A — RIPPED PAPER TRACK MATTE：\n参照左上の黒髪・赤アクセント・眼差しの強い世界から開始。最初から顔の極端なEYE CU。BLACK / RED / WHITEの紙面が人物の輪郭と髪の方向に沿って有機的に裂ける。AEのRIPPED-PAPER TRACK MATTE＋ROUGHEN EDGES＋TURBULENT DISPLACEを主役にし、破断面は直線ではなく繊維質で不規則。裂けた下層から夜の都市が見え、都市は2.5Dで逆PARALLAX。裂け目が目の横を通過するたびに背景深度だけが変わる。最後は一本の赤い裂傷を画面奥まで伸ばし、その裂傷そのものを次LOOKへのCAMERA PATHにする。小さな紙屑を大量に飛ばさない。\n\n1.55–3.10s｜LOOK B — TYPOGRAPHY DEPTH TUNNEL：\n裂傷へカメラが飛び込み、紫〜黒の世界へ色相を急転換。銀〜ラベンダー髪の人物を低い奥行きレイヤーに固定し、巨大「H3」をAE 3D LAYERとして超前景へ配置。Hと3を同一平面に置かず、数メートル離れたZ位置として扱う。カメラは文字の内側・隙間・エッジを縫って高速FLY-THROUGH。文字面には人物の目や都市をTRACK MATTEで一瞬だけ流す。MOTION BLURは文字と背景に強く、顔には最小限。紫のLIGHT SWEEPが文字のエッジを走り、最後に3の曲線が画面全体を覆うMATCH OCCLUSIONとなって次へ。\n\n3.10–4.70s｜LOOK C — NEON SIGN REBUILD / LIGHT TRACE：\n遮蔽が外れると赤いネオン都市と黒髪キャラクター。ここでは紙裂けも巨大文字トンネルも主役にしない。AEのSABER系NEON TRACE、LIGHT STREAK、GLOW PATH、MASK PATH ANIMATIONを主役にする。都市看板、窓枠、人物の髪外周、サングラスやアクセサリーの輪郭を赤い光が高速走査。光が通過した場所だけ背景の都市が別角度へREBUILDされる。MiniMaxはネオン看板のように一瞬形成されるが、固定看板ではなく光の軌跡から0.2秒で組み上がり、次の光が通ると解体。最後は一本の赤い光がカメラレンズを横断し、LUMA WIPEとして次の青紫世界を露出。\n\n4.70–6.20s｜LOOK D — 2.5D FLORAL PARALLAX / DEPTH ORBIT：\n青紫の夜、花、銀髪キャラクターのLOOKへ。人物を切り抜いた複数カードにしない。人物本体は一人のまま固定し、花びら、前景花、髪の一部、都市、遠景光だけを別深度へ分離。AE 2.5D CAMERA＋DEPTH MAP＋PARALLAX ORBITを主役に、カメラが人物の周囲を僅かに弧を描きながら高速PUSH。前景花は大きく逆方向へ流れ、背景ビルは遅れて動く。MiniMaxは遠景の巨大印刷面として斜めに現れ、カメラ移動に伴うPERSPECTIVEだけで読める。花びらが画面中央を覆う瞬間、その花びらをCOLOR MATTEへ変換して黄黒LOOKへ接続。\n\n6.20–7.75s｜LOOK E — PRINT MISREGISTRATION / POSTER OFFSET：\n黄・黒・白を主軸に、黒髪と銀髪の2人が同時に存在するファッションKV。主役はPRINT MISREGISTRATION、RGBではなくYELLOW / BLACK / WHITEの版ズレ、HALFTONE SCALE、POSTER OFFSET、FRAME ECHO。人物そのものを複製せず、人物の輪郭線・影・印刷版だけを0.03〜0.08秒ずらして追従させる。「Design」は巨大な白タイポとして中央を横断するが、文字面の一部を黒インク版と黄色版が時間差で印刷する。画面全体が印刷機を通過しているように上下の版が高速同期。最後は黄色インクの大きなSWASHが液体ではなく平面的なPRINT MATTEとして横切り、CYAN系の次LOOKを露出。\n\n7.75–9.25s｜LOOK F — HUD VECTOR SCAN / TECHNICAL REVEAL：\nCYAN / BLACK / SILVERへ一気に変更。銀髪キャラクターの顔を中心に、AE SHAPE LAYER、TRIM PATHS、VECTOR SCAN、RADIAL RETICLE、TEXTLESS HUDを主役にする。四角いUI箱を並べない。細いCYAN線が顔を避けながら髪、肩、背景都市を解析するように走り、線の交点から背景だけが別レイヤーへ切り替わる。円形RETICLEは人物の目に固定せず、その周囲を0.2秒で通過して都市の遠近へ吸収。ラインが高速で伸び縮みし、その端点をNULL OBJECTとしてカメラが追従。最後に全VECTOR LINEが一点へ収束し、極細の縦線へ圧縮。その線が左右へ開いて紫ゴシック世界を出す。\n\n9.25–10.80s｜LOOK G — GOTHIC GLASS / DISPLACEMENT REFRACTION：\n紫、黒、深紅のゴシック空間。黒髪キャラクターを静かに見せ、背景にステンドグラス、細い枝状構造、蝋燭のような点光源を配置。主役AEはREFRACTION MAP、DISPLACEMENT MAP、CAUSTIC GLASS、PRISMATIC EDGE。巨大「H3」は平面タイポではなく、半透明の割れたガラス越しに屈折して見える。ガラス片は爆発させず、画面手前で数枚だけゆっくり回転しつつ、カメラはその隙間を高速に抜ける。人物の顔は屈折させず、前景ガラスが顔を横切る瞬間だけ遮蔽として使用。最後に一枚のガラス面へ細い亀裂が走り、亀裂が全面へ広がる直前にSHATTERではなく屈折面だけが奥へ吸い込まれ、次LOOKへ。\n\n10.80–12.35s｜LOOK H — CRYSTAL SHARD TIME-SLICE：\n黒髪と銀髪の2人を同一画面へ。青黒の結晶・ガラス世界。LOOK Gの屈折とは異なり、ここではAE TIME DISPLACEMENT＋TIME SLICE＋SHARD MASKを主役にする。大小の結晶片それぞれに0.03〜0.12秒異なる時間位相を持たせ、同じ瞬間の複製ではなく、視線、瞬き、髪の揺れの異なる時間断片を一つの画面に再構成。顔そのものを複数生成せず、破片内にクロップされた部分だけ時間差表示。「H3」「MiniMax」「Design」はそれぞれ別の結晶面に反射したように断片表示し、カメラ移動で一瞬だけ3語が整列して読める。最後に全SHARDの時間差が一瞬同期し、同期フレームをMATCH FREEZEせずそのままFINALへ押し出す。\n\n12.35–15.0s｜LOOK I — FINAL MULTI-LAYER KV / KINETIC ASSEMBLY：\n最終LOOKは黒髪と銀髪の2人を中心に、RED / BLACK / WHITEへ回帰しつつ、これまでの色を小さなアクセントとして統合。ここだけは過去8LOOKの演出を単純再演せず、AE KINETIC ASSEMBLY＋3D CAMERA SOLVE＋TRACK MATTE CHAINで最終KVを組み上げる。最初の0.5秒で都市奥景がZ方向から収束、次に2人の人物レイヤーが別方向のMATCH OCCLUSIONから固定位置へ入り、H3が超前景、MiniMaxが中景、Designが斜め奥景へ配置される。文字を横一列に並べない。紫の細線、CYAN HUD断片、黄色PRINT EDGE、赤いNEON TRACEがそれぞれ0.05〜0.12秒差で最終画面へ吸収され、過去LOOKの色彩履歴を一枚のKVに統合。14.4秒以降もMICRO PARALLAX、髪の僅かな揺れ、都市光の走査を継続。14.8秒でほぼ完成し、最後0.2秒だけ視認可能なHERO FRAME。BLACK OUT、FADE OUTは禁止。\n\nTRANSITION CHAIN：\nA→B＝裂傷CAMERA PATH。B→C＝巨大3によるMATCH OCCLUSION。C→D＝NEON LUMA WIPE。D→E＝花びらCOLOR MATTE。E→F＝PRINT SWASH MATTE。F→G＝VECTOR LINE APERTURE。G→H＝REFRACTION SUCTION。H→I＝TIME-SLICE SYNCHRONIZATION。全転換を異なる原理で構成し、同じWHIP、同じZOOM、同じ紙裂けを連打しない。前シーンの最後の動体が次シーンの最初の構造へ物理的に変換されるため、9LOOKでも一本の連続PVとして感じさせる。\n\nCOLOR：\nLOOK A＝RED / BLACK / WHITE、LOOK B＝VIOLET / BLACK / WHITE、LOOK C＝CRIMSON NEON / BLACK、LOOK D＝INDIGO / LAVENDER / FLORAL PURPLE、LOOK E＝YELLOW / BLACK / WHITE、LOOK F＝CYAN / BLACK / SILVER、LOOK G＝DEEP PURPLE / BLACK / BURGUNDY、LOOK H＝ICE BLUE / CYAN / BLACK、LOOK I＝RED / BLACK / WHITE＋過去LOOKの微量アクセント。全LOOKを同じ赤黒へ統一しない。LOOK切替時に色温度、光源、背景質感、衣装印象まで明確に変わること。ただし各LOOK内は参照外見を維持。\n\nAUDIO / MUSIC：\nMUSIC ON。15秒を一本で走る132 BPM前後のstylish urban JRPG × alternative hip-hop × acid-jazz edge × cinematic breakbeat。0秒から即BEAT開始。各LOOKで曲そのものを交換せず、同一グルーヴのORCHESTRATIONを映像色に合わせて変化。A＝dry kick＋sub bass＋distorted texture、B＝low synth＋reverse brass、C＝electric pulse＋short guitar stab、D＝Rhodes＋airy texture、E＝percussive bass＋dry snare、F＝tight electronic percussion＋mono synth、G＝low strings＋dark piano accent、H＝glass percussionを極少量＋sub pulse、I＝全要素を整理してFULL GROOVE。EDM DROP、idol pop、cheerful funk、trailer braam、choirは禁止。音楽は途中で停止しない。\n\nSFX DESIGN：\n各LOOKで音も差別化。A＝TEXTURED PAPER RIP、B＝DEEP TYPO WHOOSH、C＝ELECTRIC SWEEP、D＝SOFT AIR PASS、E＝DRY PRINT IMPACT、F＝DIGITAL SWEEP、G＝CRYSTAL RESONANCE、H＝REVERSE GRANULAR TICK、I＝SUB IMPACT＋METALLIC LOCK。papapa / tatata、連続click、cheap UI beep、cartoon popは禁止。SFXはMUSICへ埋め込み、大きな構造変化だけを強調。\n\nABSOLUTE NEGATIVE：\nNO 3×3 GRID IN VIDEO、NO PANEL NUMBER、NO 01–09、NO LOOK LABEL、NO TIMECODE TEXT、NO FINAL LABEL、NO REFERENCE BORDER、NO CONTACT SHEET REPRODUCTION、NO RANDOM TEXT、NO SUBTITLE、NO WATERMARK、NO DUPLICATE PERSON、NO FACE MORPH、NO EXTRA LIMBS、NO RUNNING、NO WALKING MONTAGE、NO DANCE、NO FIGHTING、NO GENERIC ACTION POSE、NO REPEATED AE TRANSITION、NO RECTANGULAR TILE SLIDE、NO SLOW DISSOLVE、NO FADE、NO LONG STATIC HOLD、NO SAME COLOR FOR ALL LOOKS。\n\nFINAL INTENT：\n参照1枚を「そのまま9面表示する画像」として使うのではなく、9つの異なるアートディレクションを持つ完成LOOKの設計図として利用する。15秒の中で、黒赤の裂け紙、紫のタイポ空間、赤ネオン、青紫の花と2.5D、黄黒の印刷版ズレ、CYAN HUD、紫ゴシック屈折、青いTIME-SLICE結晶、最終GAME KVへと連続変形。人物が激しく動くのではなく、各LOOKごとに異なるAE技法が主役となって画面全体を動かす。9シーンすべての色、衣装、背景、質感、モーション原理を明確に差別化しながら、最後には同じ作品の一本の高級ANIME GAME PVとして収束させる。",
    "promptSourceUrl": "https://x.com/sailorv321/status/2093842404159947111",
    "sourceUrl": "https://x.com/sailorv321/status/2093842404159947111",
    "sourceLabel": "X 原帖 · @sailorv321",
    "author": "C’est La Vie | AI Director",
    "publishedAt": "2026-08-29T23:25:05.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093842404159947111.mp4",
    "posterUrl": "/posters/x/2093842404159947111.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax Design × H3 🎬\n\n今回は、9つの異なるシーンをまとめた1枚の参照画像からH3で映像化👀\n\n各シーンごとに色・衣装・雰囲気を変えながら、\nRipped Paper、2.5D、Track Matte、Typography、HUD、Parallaxなど、それぞれ違うAE Motionでつないでみました✨\n\n1枚の中に複数のLOOKを設計しておくことで、\nシーンごとの演出や変化をH3に伝えやすいのが面白い🎨\n\n#MiniMaxH3 #MiniMaxDesign\n\nMiniMax Design × H3 Prompt ⬇️\n\nMODEL：MiniMax H3｜15秒｜16:9｜24fps\nREFERENCE：[@参照1] ONLY\nSTRUCTURE：9 LOOKS / ONE CONTINUOUS GAME-PV EDIT / EACH LOOK USES A DIFFERENT AE MOTION LANGUAGE\nSTYLE：premium anime game PV × fashion editorial × aggressive After Effects motion design × mixed-media collage × cinematic 2.5D。参照1枚に含まれる9つの完成ビジュアルを「9個のLOOK候補」として読み取り、それぞれの人物外見・衣装・色彩・背景世界・構図の差を維持しながら、15秒の中で高速に連結する。各LOOKは同じテンプレート動作を繰り返さず、必ず異なるAEモーション原理で見せる。\n\nREFERENCE INTERPRETATION / ABSOLUTE CLEANUP：\n[@参照1]は9面をまとめたコンタクトシートだが、映像内にコンタクトシートそのものを表示しない。参照画像上の「01〜09」「LOOK 01〜09」「0.0–1.5s等の時間表記」「FINAL」およびパネル境界線は制作指示用メタ情報としてのみ認識し、生成映像には出さない。ナンバリング、タイムコード、LOOK表記、白い区切り線、3×3グリッドを画面へコピーしない。人物、衣装、髪、色調、背景、タイポの視覚要素だけを抽出。各LOOKはフルスクリーンで展開し、前LOOKの要素を次LOOKの転換素材へ変換する。\n\nCHARACTER / LOOK LOCK：\n参照内の黒髪系キャラクターと銀〜ラベンダー髪系キャラクターを同一作品世界の成人女性キャラクターとして扱う。各LOOKで参照に描かれた顔、目、髪型、髪色、衣装、アクセサリー、メイク、身体比率を尊重し、そのLOOKの途中で別人化させない。LOOKが切り替わる瞬間にのみ参照通りのヘア、衣装、カラー、背景へ変化可能。顔モーフ、二重顔、余分な腕、人物融合は禁止。視線、瞬き、僅かな首振り、呼吸、髪や布の二次動作のみ。走る、歩き続ける、ジャンプ、ダンス、格闘でテンポを作らない。速度はAE、カメラ、レイヤー、タイポ、マスク、都市、光、原画素材から作る。\n\nGLOBAL MOTION RULE：\n15秒を9LOOKへ分けるが静止画スライドショーにはしない。前LOOKが止まる前に次LOOKを0.12〜0.30秒先行させ、FAST BUT FLUIDで連結。各LOOKの主役AE TECHNIQUEは重複禁止。単純CUT、FADE、CROSS DISSOLVE、均等GRID、カードSLIDE、同じZOOM反復は禁止。完成KVとして一瞬読めても0.7秒以上静止させない。\n\nTEXT CONTROL：\n参照画像に含まれる番号、時間、LOOK、FINALなどの管理文字は全削除。主要タイポは「H3」「MiniMax」「Design」のみ。字幕として下部へ置かない。巨大前景、MASK、TRACK MATTE、3D奥行き、壁面印刷、光学的遮蔽物として使用。ランダム英数字、日本語、中国語、偽UI文字、架空ブランド、追加ロゴ、ウォーターマークは禁止。\n\n0.0–1.55s｜LOOK A — RIPPED PAPER TRACK MATTE：\n参照左上の黒髪・赤アクセント・眼差しの強い世界から開始。最初から顔の極端なEYE CU。BLACK / RED / WHITEの紙面が人物の輪郭と髪の方向に沿って有機的に裂ける。AEのRIPPED-PAPER TRACK MATTE＋ROUGHEN EDGES＋TURBULENT DISPLACEを主役にし、破断面は直線ではなく繊維質で不規則。裂けた下層から夜の都市が見え、都市は2.5Dで逆PARALLAX。裂け目が目の横を通過するたびに背景深度だけが変わる。最後は一本の赤い裂傷を画面奥まで伸ばし、その裂傷そのものを次LOOKへのCAMERA PATHにする。小さな紙屑を大量に飛ばさない。\n\n1.55–3.10s｜LOOK B — TYPOGRAPHY DEPTH TUNNEL：\n裂傷へカメラが飛び込み、紫〜黒の世界へ色相を急転換。銀〜ラベンダー髪の人物を低い奥行きレイヤーに固定し、巨大「H3」をAE 3D LAYERとして超前景へ配置。Hと3を同一平面に置かず、数メートル離れたZ位置として扱う。カメラは文字の内側・隙間・エッジを縫って高速FLY-THROUGH。文字面には人物の目や都市をTRACK MATTEで一瞬だけ流す。MOTION BLURは文字と背景に強く、顔には最小限。紫のLIGHT SWEEPが文字のエッジを走り、最後に3の曲線が画面全体を覆うMATCH OCCLUSIONとなって次へ。\n\n3.10–4.70s｜LOOK C — NEON SIGN REBUILD / LIGHT TRACE：\n遮蔽が外れると赤いネオン都市と黒髪キャラクター。ここでは紙裂けも巨大文字トンネルも主役にしない。AEのSABER系NEON TRACE、LIGHT STREAK、GLOW PATH、MASK PATH ANIMATIONを主役にする。都市看板、窓枠、人物の髪外周、サングラスやアクセサリーの輪郭を赤い光が高速走査。光が通過した場所だけ背景の都市が別角度へREBUILDされる。MiniMaxはネオン看板のように一瞬形成されるが、固定看板ではなく光の軌跡から0.2秒で組み上がり、次の光が通ると解体。最後は一本の赤い光がカメラレンズを横断し、LUMA WIPEとして次の青紫世界を露出。\n\n4.70–6.20s｜LOOK D — 2.5D FLORAL PARALLAX / DEPTH ORBIT：\n青紫の夜、花、銀髪キャラクターのLOOKへ。人物を切り抜いた複数カードにしない。人物本体は一人のまま固定し、花びら、前景花、髪の一部、都市、遠景光だけを別深度へ分離。AE 2.5D CAMERA＋DEPTH MAP＋PARALLAX ORBITを主役に、カメラが人物の周囲を僅かに弧を描きながら高速PUSH。前景花は大きく逆方向へ流れ、背景ビルは遅れて動く。MiniMaxは遠景の巨大印刷面として斜めに現れ、カメラ移動に伴うPERSPECTIVEだけで読める。花びらが画面中央を覆う瞬間、その花びらをCOLOR MATTEへ変換して黄黒LOOKへ接続。\n\n6.20–7.75s｜LOOK E — PRINT MISREGISTRATION / POSTER OFFSET：\n黄・黒・白を主軸に、黒髪と銀髪の2人が同時に存在するファッションKV。主役はPRINT MISREGISTRATION、RGBではなくYELLOW / BLACK / WHITEの版ズレ、HALFTONE SCALE、POSTER OFFSET、FRAME ECHO。人物そのものを複製せず、人物の輪郭線・影・印刷版だけを0.03〜0.08秒ずらして追従させる。「Design」は巨大な白タイポとして中央を横断するが、文字面の一部を黒インク版と黄色版が時間差で印刷する。画面全体が印刷機を通過しているように上下の版が高速同期。最後は黄色インクの大きなSWASHが液体ではなく平面的なPRINT MATTEとして横切り、CYAN系の次LOOKを露出。\n\n7.75–9.25s｜LOOK F — HUD VECTOR SCAN / TECHNICAL REVEAL：\nCYAN / BLACK / SILVERへ一気に変更。銀髪キャラクターの顔を中心に、AE SHAPE LAYER、TRIM PATHS、VECTOR SCAN、RADIAL RETICLE、TEXTLESS HUDを主役にする。四角いUI箱を並べない。細いCYAN線が顔を避けながら髪、肩、背景都市を解析するように走り、線の交点から背景だけが別レイヤーへ切り替わる。円形RETICLEは人物の目に固定せず、その周囲を0.2秒で通過して都市の遠近へ吸収。ラインが高速で伸び縮みし、その端点をNULL OBJECTとしてカメラが追従。最後に全VECTOR LINEが一点へ収束し、極細の縦線へ圧縮。その線が左右へ開いて紫ゴシック世界を出す。\n\n9.25–10.80s｜LOOK G — GOTHIC GLASS / DISPLACEMENT REFRACTION：\n紫、黒、深紅のゴシック空間。黒髪キャラクターを静かに見せ、背景にステンドグラス、細い枝状構造、蝋燭のような点光源を配置。主役AEはREFRACTION MAP、DISPLACEMENT MAP、CAUSTIC GLASS、PRISMATIC EDGE。巨大「H3」は平面タイポではなく、半透明の割れたガラス越しに屈折して見える。ガラス片は爆発させず、画面手前で数枚だけゆっくり回転しつつ、カメラはその隙間を高速に抜ける。人物の顔は屈折させず、前景ガラスが顔を横切る瞬間だけ遮蔽として使用。最後に一枚のガラス面へ細い亀裂が走り、亀裂が全面へ広がる直前にSHATTERではなく屈折面だけが奥へ吸い込まれ、次LOOKへ。\n\n10.80–12.35s｜LOOK H — CRYSTAL SHARD TIME-SLICE：\n黒髪と銀髪の2人を同一画面へ。青黒の結晶・ガラス世界。LOOK Gの屈折とは異なり、ここではAE TIME DISPLACEMENT＋TIME SLICE＋SHARD MASKを主役にする。大小の結晶片それぞれに0.03〜0.12秒異なる時間位相を持たせ、同じ瞬間の複製ではなく、視線、瞬き、髪の揺れの異なる時間断片を一つの画面に再構成。顔そのものを複数生成せず、破片内にクロップされた部分だけ時間差表示。「H3」「MiniMax」「Design」はそれぞれ別の結晶面に反射したように断片表示し、カメラ移動で一瞬だけ3語が整列して読める。最後に全SHARDの時間差が一瞬同期し、同期フレームをMATCH FREEZEせずそのままFINALへ押し出す。\n\n12.35–15.0s｜LOOK I — FINAL MULTI-LAYER KV / KINETIC ASSEMBLY：\n最終LOOKは黒髪と銀髪の2人を中心に、RED / BLACK / WHITEへ回帰しつつ、これまでの色を小さなアクセントとして統合。ここだけは過去8LOOKの演出を単純再演せず、AE KINETIC ASSEMBLY＋3D CAMERA SOLVE＋TRACK MATTE CHAINで最終KVを組み上げる。最初の0.5秒で都市奥景がZ方向から収束、次に2人の人物レイヤーが別方向のMATCH OCCLUSIONから固定位置へ入り、H3が超前景、MiniMaxが中景、Designが斜め奥景へ配置される。文字を横一列に並べない。紫の細線、CYAN HUD断片、黄色PRINT EDGE、赤いNEON TRACEがそれぞれ0.05〜0.12秒差で最終画面へ吸収され、過去LOOKの色彩履歴を一枚のKVに統合。14.4秒以降もMICRO PARALLAX、髪の僅かな揺れ、都市光の走査を継続。14.8秒でほぼ完成し、最後0.2秒だけ視認可能なHERO FRAME。BLACK OUT、FADE OUTは禁止。\n\nTRANSITION CHAIN：\nA→B＝裂傷CAMERA PATH。B→C＝巨大3によるMATCH OCCLUSION。C→D＝NEON LUMA WIPE。D→E＝花びらCOLOR MATTE。E→F＝PRINT SWASH MATTE。F→G＝VECTOR LINE APERTURE。G→H＝REFRACTION SUCTION。H→I＝TIME-SLICE SYNCHRONIZATION。全転換を異なる原理で構成し、同じWHIP、同じZOOM、同じ紙裂けを連打しない。前シーンの最後の動体が次シーンの最初の構造へ物理的に変換されるため、9LOOKでも一本の連続PVとして感じさせる。\n\nCOLOR：\nLOOK A＝RED / BLACK / WHITE、LOOK B＝VIOLET / BLACK / WHITE、LOOK C＝CRIMSON NEON / BLACK、LOOK D＝INDIGO / LAVENDER / FLORAL PURPLE、LOOK E＝YELLOW / BLACK / WHITE、LOOK F＝CYAN / BLACK / SILVER、LOOK G＝DEEP PURPLE / BLACK / BURGUNDY、LOOK H＝ICE BLUE / CYAN / BLACK、LOOK I＝RED / BLACK / WHITE＋過去LOOKの微量アクセント。全LOOKを同じ赤黒へ統一しない。LOOK切替時に色温度、光源、背景質感、衣装印象まで明確に変わること。ただし各LOOK内は参照外見を維持。\n\nAUDIO / MUSIC：\nMUSIC ON。15秒を一本で走る132 BPM前後のstylish urban JRPG × alternative hip-hop × acid-jazz edge × cinematic breakbeat。0秒から即BEAT開始。各LOOKで曲そのものを交換せず、同一グルーヴのORCHESTRATIONを映像色に合わせて変化。A＝dry kick＋sub bass＋distorted texture、B＝low synth＋reverse brass、C＝electric pulse＋short guitar stab、D＝Rhodes＋airy texture、E＝percussive bass＋dry snare、F＝tight electronic percussion＋mono synth、G＝low strings＋dark piano accent、H＝glass percussionを極少量＋sub pulse、I＝全要素を整理してFULL GROOVE。EDM DROP、idol pop、cheerful funk、trailer braam、choirは禁止。音楽は途中で停止しない。\n\nSFX DESIGN：\n各LOOKで音も差別化。A＝TEXTURED PAPER RIP、B＝DEEP TYPO WHOOSH、C＝ELECTRIC SWEEP、D＝SOFT AIR PASS、E＝DRY PRINT IMPACT、F＝DIGITAL SWEEP、G＝CRYSTAL RESONANCE、H＝REVERSE GRANULAR TICK、I＝SUB IMPACT＋METALLIC LOCK。papapa / tatata、連続click、cheap UI beep、cartoon popは禁止。SFXはMUSICへ埋め込み、大きな構造変化だけを強調。\n\nABSOLUTE NEGATIVE：\nNO 3×3 GRID IN VIDEO、NO PANEL NUMBER、NO 01–09、NO LOOK LABEL、NO TIMECODE TEXT、NO FINAL LABEL、NO REFERENCE BORDER、NO CONTACT SHEET REPRODUCTION、NO RANDOM TEXT、NO SUBTITLE、NO WATERMARK、NO DUPLICATE PERSON、NO FACE MORPH、NO EXTRA LIMBS、NO RUNNING、NO WALKING MONTAGE、NO DANCE、NO FIGHTING、NO GENERIC ACTION POSE、NO REPEATED AE TRANSITION、NO RECTANGULAR TILE SLIDE、NO SLOW DISSOLVE、NO FADE、NO LONG STATIC HOLD、NO SAME COLOR FOR ALL LOOKS。\n\nFINAL INTENT：\n参照1枚を「そのまま9面表示する画像」として使うのではなく、9つの異なるアートディレクションを持つ完成LOOKの設計図として利用する。15秒の中で、黒赤の裂け紙、紫のタイポ空間、赤ネオン、青紫の花と2.5D、黄黒の印刷版ズレ、CYAN HUD、紫ゴシック屈折、青いTIME-SLICE結晶、最終GAME KVへと連続変形。人物が激しく動くのではなく、各LOOKごとに異なるAE技法が主役となって画面全体を動かす。9シーンすべての色、衣装、背景、質感、モーション原理を明確に差別化しながら、最後には同じ作品の一本の高級ANIME GAME PVとして収束させる。",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 4,
      "views": 658,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093841400836259993",
    "title": "雪花标志穿过水果与早餐碗",
    "titleEn": "A Snowflake Mark Moves Through Fruit and a Breakfast Bowl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝色雪花标志进入水果拼贴和早餐碗场景，产品元素以三维广告节奏完成转场。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue snowflake mark moves into a fruit collage and breakfast-bowl scene as product elements transition with a three-dimensional ad rhythm. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiartpanjp/status/2093841400836259993",
    "sourceLabel": "X 原帖 · @aiartpanjp",
    "author": "pan",
    "publishedAt": "2026-08-29T23:21:06.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093841400836259993.mp4",
    "posterUrl": "/posters/x/2093841400836259993.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#雪ノ糸\n #ズーチャレ \n#CREATORZZ \n #TapNow \n@TapNow_AI \nMiniMax H3 \n3D",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 53,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093840726270620108",
    "title": "白发角色与发光猫完成三模型并排比较",
    "titleEn": "A White-Haired Character and Glowing Cat Compare Three Models",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一白发角色与发光猫在三列画面中同步变化，直接比较三种模型的角色稳定性。 原生视频规格为 20 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same white-haired character and glowing cat change in three synchronized columns for a direct comparison of character consistency. Native-video output: 20s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/soranoblue3/status/2093840726270620108",
    "sourceLabel": "X 原帖 · @soranoblue3",
    "author": "けこねこ",
    "publishedAt": "2026-08-29T23:18:25.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093840726270620108.mp4",
    "posterUrl": "/posters/x/2093840726270620108.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "メガネ悪魔のお兄さんに\nPollo AIで使える3つの動画ツールを比べてみたヽ(=´▽`=)ﾉ\n\n全て同じプロンプトで\n\n左）seedance2.5・20秒\n中）Wan3.0・20秒\n右)MiniMax H3・15秒\n\nどれも画風・キャラクターの変化が最後までほぼ無くいい感じ♡\n\nあとは、どれが好みかくらい(*˙꒫˙* )\n\nあなたはどのお兄さんが好きですか？\n（私は左が好みです(￣b￣)ｼｰｯ）\n\nまだコスト50%オフもあるので\nリプ欄のリンクから\nぜひPollo AIで使ってみて♡\n@PolloAIJP",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 3,
      "views": 120,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093839723005726730",
    "title": "九宫格人物素材展开成动感创作卡",
    "titleEn": "A Nine-Panel Character Grid Expands into a Kinetic Creator Card",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "九宫格人物素材拆分成青黄色杂志版式，手机、标签和人物动作连续进入创作主题卡。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A nine-panel character grid breaks into teal-and-yellow magazine layouts as a phone, labels, and character motion build a creator-themed card. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/taya_mama_AI/status/2093839723005726730",
    "sourceLabel": "X 原帖 · @taya_mama_AI",
    "author": "た〜や｜スキマ時間×AI副業ママ",
    "publishedAt": "2026-08-29T23:14:26.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093839723005726730.mp4",
    "posterUrl": "/posters/x/2093839723005726730.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3がfalでお安く試せたのでリベンジしてみた‼\nさすが話題のモデルだけあってトランジションしても画像が破綻してない✨\n\n偶然の面白さで言えばGrokも捨てがたい…🤤\n\nたくぞーさん @tkz_aiart の MiniMAXの教科書（PR→https://brmk.io/7YqGxL）手に入れたので、再販リリース終わったらMiniMAX極めます🔥今はまだ我慢…\n\n↑\nコチラ明日まで割引価格らしいです！詳しくはリンク内のオプチャ割り引きをご確認ください～",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 284,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093839427592388608",
    "title": "黑发少女化作流星撞向荒原",
    "titleEn": "A Dark-Haired Heroine Becomes a Meteor over the Wasteland",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发少女从空中坠落并卷起巨大烟尘，五秒动作以流星撞击般的冲击收尾。 原生视频规格为 5 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired heroine falls from the sky and raises a huge dust cloud, ending the five-second action like a meteor strike. Native-video output: 5s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yumesyokunin/status/2093839427592388608",
    "sourceLabel": "X 原帖 · @yumesyokunin",
    "author": "夢職人",
    "publishedAt": "2026-08-29T23:13:15.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093839427592388608.mp4",
    "posterUrl": "/posters/x/2093839427592388608.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax-h3-max\nほむほむメテオアタック",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 37,
      "views": 531,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093838781195706876",
    "title": "狐面守护者让荒谷重新长出花木",
    "titleEn": "A Fox-Masked Guardian Restores Life to a Barren Valley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "狐面守护者释放粉色光脉，荒芜山谷逐渐恢复河流、草木与花朵。 原生视频规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fox-masked guardian releases pink veins of light as a barren valley regains rivers, grass, and flowers. Native-video output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ewotanosimo2310/status/2093838781195706876",
    "sourceLabel": "X 原帖 · @ewotanosimo2310",
    "author": "のいしろ@AI遊び",
    "publishedAt": "2026-08-29T23:10:41.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093838781195706876.mp4",
    "posterUrl": "/posters/x/2093838781195706876.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【必殺・花脈創生（かみゃくそうせい）】\n\n狐面の守護者が大地に手をかざすと、桃色の光脈が谷を走り、水と草花を呼び戻す。\n壊すだけではない必殺技を作りたかった。\n\n参照画像1枚から、世界観・技の性質・カメラの動きまでAIのMiniMax H3に委ねて生成しました。\nJACC　#AI必殺技コンテスト 　楽しいね",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 69,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093837994654654742",
    "title": "红黑动画角色用日语介绍生成速度",
    "titleEn": "Red-and-Black Anime Hosts Explain Generation Speed",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "两名动画角色在漫画气泡与日文标题间切换，以口型和夸张动作介绍生成速度。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two animated hosts move between comic bubbles and Japanese titles, using lip sync and exaggerated motion to explain generation speed. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/x_nonotan/status/2093837994654654742",
    "sourceLabel": "X 原帖 · @x_nonotan",
    "author": "ののたん",
    "publishedAt": "2026-08-29T23:07:34.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093837994654654742.mp4",
    "posterUrl": "/posters/x/2093837994654654742.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "anime"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "すご。。minimax H3 Maxほんとに冗談みたいな速度で出ますね🫢\nプリビズとか、テスト出力をこれで繰り返して清書は別モデルとかできそう？\nしかもこれが将来的にオープンウェイトになる予定って、またひとつ世界が変わりそう😆\n\n今なら登録不要で１日５回使えるみたいです\n🔽リンクはリプ",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 292,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093837039410225347",
    "title": "母女拥抱让黑猫迸出蓝色电光",
    "titleEn": "A Family Hug Sends Blue Sparks through a Black Cat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "女子抱着黑猫与女孩靠近，蓝色电光闪过后两人一猫完成温暖拥抱。 原生视频规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman holding a black cat meets a girl, and a flash of blue energy resolves into a warm embrace among all three. Native-video output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bonyoh1/status/2093837039410225347",
    "sourceLabel": "X 原帖 · @bonyoh1",
    "author": "凡用人型兵器",
    "publishedAt": "2026-08-29T23:03:46.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093837039410225347.mp4",
    "posterUrl": "/posters/x/2093837039410225347.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "話題のAI動画生成モデルの無料サービスがあったので試してみたけど、5秒の動画が5秒で生成される時代ヤバすぎるだろ\nhttps://fal.ai/tools/minimax-h3-max",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 78,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093836090914607439",
    "title": "篝火旁的两名圣女完成静默和解",
    "titleEn": "Two Heroines Reconcile Beside a Campfire",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名幻想角色在湖边篝火旁交谈并靠近，月色、倒影与克制表情支撑安静戏剧。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two fantasy heroines talk and draw closer beside a lakeside campfire as moonlight, reflections, and restrained expressions support a quiet drama. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/luche_whitewing/status/2093836090914607439",
    "sourceLabel": "X 原帖 · @luche_whitewing",
    "author": "翼 AIcreator",
    "publishedAt": "2026-08-29T23:00:00.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093836090914607439.mp4",
    "posterUrl": "/posters/x/2093836090914607439.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🌸 8時の美少女🌸\n姿や声が変わっても、\nその人が歩んできた心まで\n消えてしまうわけじゃない。\n「聖女」ではなく、\nただひとりの「サーシャ」として——\nTOPVIEW AI　@topviewAIJP のMiniMaxH3で\nシネマティックなドラマ動画を生成📹️\n（左が768P、右が2Kです）\n\n誰かに本当の名前を呼ばれることが、\n自分を取り戻す小さな灯火になる夜もある。\nあなたも今日は、\n誰かに望まれた姿ではなく、\nあなた自身のままでいられますように🌿✨\n『聖女ではなく、サーシャ』\n名前と心をめぐる、静かな15秒の物語です。🕊️🤍\n聖女サーシャについては引用元ポストを👇️",
    "engagement": {
      "replies": 3,
      "reposts": 8,
      "likes": 28,
      "views": 528,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093834459393167664",
    "title": "金发小天使旋转着展开双臂",
    "titleEn": "A Tiny Blonde Angel Spins with Open Arms",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发小天使在白色背景上转身、跳步并展开双臂，角色线稿与服装保持稳定。 原生视频规格为 5 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny blonde angel turns, skips, and opens her arms against white while the line art and costume remain consistent. Native-video output: 5s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yumesyokunin/status/2093834459393167664",
    "sourceLabel": "X 原帖 · @yumesyokunin",
    "author": "夢職人",
    "publishedAt": "2026-08-29T22:53:31.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093834459393167664.mp4",
    "posterUrl": "/posters/x/2093834459393167664.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "fashion",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax-h3-max\nほんとに驚くほど高速だなぁ・・・",
    "engagement": {
      "replies": 2,
      "reposts": 11,
      "likes": 84,
      "views": 1020,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093832651236929882",
    "title": "两名游戏英雄在蘑菇王国并肩登场",
    "titleEn": "Two Game Heroes Meet in a Mushroom Kingdom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名游戏风英雄在城堡与砖块场景前并肩站立，短动作保持双方造型。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two game-inspired heroes stand together before a castle and brick landscape, holding both designs through a short action. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/American_N3rd/status/2093832651236929882",
    "sourceLabel": "X 原帖 · @American_N3rd",
    "author": "American Nerd 🇺🇸",
    "publishedAt": "2026-08-29T22:46:20.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093832651236929882.mp4",
    "posterUrl": "/posters/x/2093832651236929882.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "nature",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Okay... MiniMax H3 Max creates videos faster than you can watch them... Looks like no copyright either.. wow. Link below",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 57,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093831877006397828",
    "title": "红黑撕纸与城市霓虹重组双人海报",
    "titleEn": "Red-Black Paper Tears Rebuild a Two-Character Poster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝发与银发角色在撕纸、城市霓虹和巨大排版间切换，最终重组为游戏主视觉。 原生视频规格为 15 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Blue- and silver-haired characters move through ripped paper, city neon, and oversized type before resolving into a game key visual. Native-video output: 15s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MODEL：MiniMax H3｜15秒｜16:9｜24fps\nREFERENCE：[参照1]＋[参照2] ONLY\nSTYLE：original urban superpower 2D game PV × premium JRPG opening × rebellious ripped-paper collage × aggressive AE motion design × fashion editorial game graphics。RED / BLACK / WHITEを主軸に、参照キャラクター固有色は維持。シャープな2Dセル、漫画スクリーントーン、印刷ズレ、インク輪郭、裂けた紙繊維、ネオン反射、非対称UI、巨大タイポを統合。\n\n【REFERENCE / IDENTITY LOCK】\n[参照1]＝CHARACTER A、[参照2]＝CHARACTER B。2人とも成人女性。顔、目、輪郭、髪、身体比率、衣装、アクセサリーを15秒維持。A/Bの融合、顔交換、衣装交換、別人化、複製、第三人物は禁止。\n\n【CORE】\n人物を走らせて速度を作らない。CHARACTER MOTION = MINIMAL、CAMERA / GRAPHICS / TYPOGRAPHY / CITY / AUDIO = EXTREME。A/Bは視線、瞬き、僅かな首振り、呼吸、髪と衣装の揺れのみ。走る、歩行シークエンス、ジャンプ、格闘、手から能力を撃つ演出は禁止。映像全体を「静かな2人のKEY VISUALを、都市・原画・文字・UI・紙面そのものが狂った速度で再編集するGAME PV」とする。\n\n【MOST IMPORTANT — ORGANIC RIPPED SCREEN DESIGN】\n画面を四角いブロック、均等グリッド、カード矩形で切り分けない。ブロック状の分割を主役にしない。切断面は「紙を手で高速に引き裂いたような、不規則で細長い裂け目」。毛羽立った紙繊維、ギザギザの非対称エッジ、細い裂傷、斜めに伸びる破れ、曲線と鋭角が混ざる有機的輪郭。裂け目は一本の直線ではなく、途中で枝分かれし、細くなり、再び広がる。画面中央で整然と4分割しない。顔や都市の輪郭に沿って裂け、髪の流れや視線の方向から自然に亀裂が発生する。裂けた前景紙面の下から次ショット、赤いハーフトーン、別角度の都市、H3 / MiniMax / Designが覗く。紙片は厚いパネルではなく薄い2D原画シートとして前後へしなる。直線WIPEではなくTEAR → PEEL → CURL → WHIP → REVEALを0.08〜0.25秒で連鎖。\n\n【GAME VISUAL SYSTEM】\nGAME UIは後半だけではなく第1フレームから最終フレームまで常時存在。ただしUIも四角い箱を大量に置かず、裂けた紙面の輪郭へ沿わせる。CORNER BRACKETは片側だけ、TARGET LINEは途中で切れる、STATUS BARは文字なしの細い赤線として紙の裂け目へ吸収。CROSSHAIR、HALFTONE、CHARACTER CUT-IN、TEXTLESS GAUGE、2.5D ART CELL、NEON TRACE、SCREEN TEAR、KINETIC TYPEを毎フレーム最低2種類併用。UIは装飾ではなく次画面を生むMASK / MATTE / CAMERA PATH。\n\n【TEXT LOCK】\n可読文字は「H3」「MiniMax」「Design」の3種類のみ。日本語、中国語、数字、架空UI文字、看板、追加英単語、字幕、ウォーターマーク禁止。文字は字幕ではなく巨大GAME TITLE OBJECT、裂け目の下層、TRACK MATTE、前景遮蔽物として使う。\n\n【TEMPO】\n平均0.25〜0.55秒ごとに構造更新。MICRO TRANSITION 0.06〜0.22秒。0.7秒以上同じ構図を保持しない。前ショットが終わる前に次ショットが始まる。常時2〜4レイヤーが別速度で進行。FAST BUT FLUID、NOT JERKY。FADE、SLOW DISSOLVE、静止スライド、長い溜め、文字が止まるのを待つ編集は禁止。\n\n【0.0–0.8s】\n最初から完成したGAME PV画面。Aの片目CUを大きく配置し、その上へRED HALFTONE、細いWHITE UI LINE、都市夜景を重ねる。画面右上から紙面に細い裂傷が走り、Aの髪の流れに沿って亀裂が斜め下へ伸びる。裂け目の奥にはAの別CUと都市の赤い光。裂け目を0.1秒単位で増殖させ、紙面の端が僅かにめくれ、その隙間へカメラがPUSH。四角いPANELへ変換しない。\n\n【0.8–1.6s】\n裂けた紙片が前景へWHIPし、裏面がBLACK→REDへ瞬間反転。その紙片の形状そのものが巨大H3のMASKへ変形。H3は画面中央へ普通に表示せず、Hの一部だけ超前景、3は奥景、残りは裂け目越しに断続的に見える。カメラがHの縦画をすれすれに通過。HがBの目を0.1秒隠した瞬間、遮蔽箇所だけBのCUへ置換。文字を完成表示して待たない。\n\n【1.6–2.4s】\nBの顔CU。Bは目線だけ僅かに上げる。Bの髪の外周から白い紙面が裂けるように細い亀裂が3方向へ走り、裂けた奥から都市、Aの目、RED SCREEN TONEを同時露出。裂け目ごとにZ深度を変え、前景紙片は手前へめくれ、奥の都市は逆方向へPULL。細いUIラインが裂け目に沿って走り、そのまま次TEARの起点になる。\n\n【2.4–3.2s】\n都市のビル壁面が紙ポスターのように縦へ剥がれ、下層からMiniMax。文字はWHITE、背後にRED SHADOW、さらにBLACK OFFSET。ただし長方形の赤帯を固定表示しない。赤は裂けた紙の裏面として断片的に見せる。MiniMaxの文字面をTRACK MATTEにし、内部だけA/B/都市を高速MATCH CUT。文字外周も0.1秒ごとに削れ、破れた紙繊維へ変形。カメラは文字の隙間を通過。\n\n【3.2–4.0s】\nAとBのDUAL CUT-IN。左右50/50の分割は禁止。A側は左下から斜めに裂けた大きな紙片、B側は右上から侵食する細長い破れとして非対称配置。中央境界は直線ではなく複雑な裂け目。A/Bは静止中心。背景交差点だけが5層の原画セルへ分離し逆PARALLAX。裂けた境界をRED NEON TRACEが0.12秒で走り、通過した場所から次構図へ置換。\n\n【4.0–4.8s】\nNEON TRACEが画面端へ到達する前に、前景の紙面を勢いよく横へPEEL。下からAの別CU、Bの瞳、都市道路、ハーフトーンが一枚のコラージュとして現れる。各素材は四角く切らず、すべて破れた不規則輪郭。小さな紙片が飛散するのではなく、大きな原画シートが薄くしなりながらカメラ前を横切る。DesignのDだけが裂け目の奥に先行出現。\n\n【4.8–5.6s】\nDが巨大前景へ抜け、その後ろからesignが0.08秒遅れで追従しDesign完成。ただし0.2秒以内に文字の一部が紙の裂け目へ隠れる。WHITE NEON SWEEPが文字輪郭をなぞり、SWEEP通過部だけ都市線画とBの輪郭が高コントラスト化。Designの斜線を次の裂傷へ連続変形し、画面を細長く引き裂く。裂けた下層にH3の一部を既に見せる。\n\n【5.6–6.4s】\nH3 / MiniMax / Designを3層のGAME ARCHITECTUREとしてZ空間に分散。直線的なロゴ並びは禁止。H3は超前景で一部画面外、MiniMaxは裂けた中層紙面、Designは都市奥景へ斜め配置。文字間をA/Bの顔と都市が覗く。カメラが文字と紙層の間を高速で縫う。各紙層は別方向へPEELし、次のA CUT-INへ連続接続。\n\n【6.4–7.2s】\nAのULTIMATE CUT-IN風CU。能力発射はしない。Aの瞳が僅かに横へ動くと、その視線延長上にRED INK TEARが走る。都市が爆発するのではなく、街全体が印刷原画として6〜9枚に剥離。道路、窓、ネオン、空が紙の層としてずれ、間からBの顔が出現。破断面は全てギザギザで有機的。カメラは紙層の隙間を高速FLY-THROUGH。\n\n【7.2–8.0s】\nBへ到達する直前、Bの顔周囲だけ紙が円形ではなく不規則に裂け、目、髪、衣装の一部が別深度へ滑る。B自体を複製しない。クロップのみ。薄いWHITE CROSSHAIRが0.08秒横断し、その軌跡が次の裂け目へ変化。背景はRED / BLACK / WHITEの印刷レイヤーが0.05秒差でズレ続け、静止感をなくす。\n\n【8.0–8.8s】\nAとBを交互に見せるのではなく同時存在させ、手前の裂けた紙片が一瞬Aを隠す間に下層だけBへ切り替えるMATCH REVEAL。さらに別の紙片でBを隠しAへ戻す。人物が移動しているように見せず、編集そのものがキャラを交換。H3巨大クロップが奥からPUSH、MiniMaxが逆方向SLIDE、Designは紙面裏側に断片表示。\n\n【8.8–9.6s】\nREDの大きな印刷面が画面右から侵入するが矩形ブロックではなく、上下が激しく裂けた一枚紙。中央から縦に引き裂かれ、その間からBの大きな顔とDesign。最前景にはAの片目。WHITE UI BRACKETは4隅全部を囲まず2辺だけが短く点灯。0.15秒で位置を変え、裂け目へ吸収。\n\n【9.6–10.4s】\nカメラが巨大H3の側面を通過し、その奥にA/Bの3種類のGAME KV候補を0.2秒単位で連続MATCH CUT。各KVは四角いフレームではなく、裂けた紙の巨大シルエットで切り取る。1つ目A優勢、2つ目B優勢、3つ目A/B同格。ポーズ変化の途中は見せず、異なる完成構図を高速置換。背景CITYは常に逆PARALLAX。\n\n【10.4–11.2s】\n3つ目のKVが完成する前に画面全体が一本の大きな斜め裂傷から連鎖的に破れる。細い亀裂→枝分かれ→紙面がめくれる→下層の次KVが現れる。大量の小ブロックに分割しない。裂けた面は3〜5枚程度の大きな不規則シート。BLACK NEGATIVE SPACE、RED HALFTONE、WHITE PAPER FIBERを隙間へ見せる。カメラは最も細い裂け目へPUSH。\n\n【11.2–12.0s】\n裂け目を抜けるとA左寄り、B右寄りのNEAR FINAL KV。しかし左右対称は禁止。Aは大きな前景、Bは少し奥、H3が両者を斜めに跨ぎ、MiniMaxは中景、Designは都市壁面の裂けた紙層へ統合。0.25秒だけ完成感を出した後、H3の一画が前景へWHIPし画面を再び引き裂く。\n\n【12.0–12.8s】\n裂けたH3の一画をCAMERA PATHとして利用し、A/Bの目CU、都市、赤い印刷面を高速通過。WHITE NEON TRACEとRED INK LINEが逆方向から走り、交差地点で紙面がPEEL。音と映像を完全同期。人物はほぼ静止なのに、周囲の紙面・都市・文字・UIだけが激しく動き続ける。\n\n【12.8–13.6s】\nFINAL KV再構築開始。A/Bを中央付近へ固定し、背後の都市原画が大きな破れた紙層として奥から一枚ずつ高速収束。H3＝最大前景、MiniMax＝横方向中景、Design＝斜め奥景。文字を一列に並べない。各層が0.05〜0.12秒差で互いを追い越し、RED / BLACK / WHITEの構造を完成させる。\n\n【13.6–14.4s】\nWHITE LIGHT TEARが左→右、0.1秒後RED LIGHT TEARが右→左へ交差。単なる光線ではなく、通過した場所で紙表面が一瞬めくれ下層原画を露出。交差点でA/Bの目CUが瞬間的に覗き、すぐFINAL構図へ戻る。肌や髪を赤一色に染めない。最後までGAME UI、HALFTONE、裂けた紙輪郭を維持。\n\n【14.4–15.0s】\nFINAL GAME CHARACTER KV。A/B中心、H3最大、MiniMax中景、Design斜めサブグラフィック。裂けた紙面、都市、RED HALFTONE、BLACK NEGATIVE SPACE、WHITE FIBER EDGE、UI LINE、NEON EDGEを一体化。14.8秒までMICRO PARALLAXと紙端の僅かな揺れを継続し、最後0.2秒だけほぼ固定。BLACK OUT禁止。\n\n【CAMERA / TRANSITION】\nRIPPED-PAPER TRACK MATTE、TEAR REVEAL、PEEL TRANSITION、PAPER CURL、INK TEAR、ORGANIC EDGE WIPE、MATCH MOVE、MATCH OCCLUSION、TYPOGRAPHY FLY-THROUGH、EXTREME PUSH-PULL、2.5D PARALLAX、Z-DEPTH CROSSING、NEON TRACE、HALFTONE REVEALを複合。RECTANGULAR BLOCK WIPE、均等GRID、タイル分割、カードを並べただけの画面は禁止。同じ転換を反復せず、前ショットの裂け目・文字・髪の流れ・都市輪郭を次ショットのMASKへ変換。\n\n【AUDIO / MUSIC】\nMUSIC ON。15秒全体を一本の高速GAME PV TRACKとして作る。132 BPM前後、dark urban alternative hip-hop × cinematic breakbeat × stylish JRPG battle menu energy。重いSUB BASS、乾いたKICK、タイトなSNARE、短いBREAKBEAT、歪ませすぎないBASS PULSE、低いSYNTH DRONE、逆再生TEXTURE、短いSTRING STAB、METALLIC RISERを使用。EDM DROP、明るいアイドルPOP、陽気なFUNKは禁止。0秒からBEAT開始し、途中で音楽を切らない。映像の0.08〜0.25秒転換すべてに細かく音を付けすぎず、MUSICの大きなビートと映像構造を同期する。\n\n【SFX】\n紙を裂く「ビリビリ」そのままの安い効果音にしない。RIPPED PAPERは低域を含む短いTEXTURED TEAR＋AIR WHOOSHとして高級に処理。PEELは薄いREVERSE SWELL、巨大文字通過はDEEP WHOOSH、PUSH/PULLはSUB IMPACT、NEON TRACEは短いELECTRIC SWEEP。SFXはMUSICへ溶け込ませ、毎回同じ音を連打しない。repetitive click / cartoon pop / cheap UI beep / excessive explosionは禁止。\n\n【ABSOLUTE】\nNO BLOCKY SCREEN DIVISION、NO RECTANGULAR TILE GRID、NO CLEAN 4-WAY SPLIT、NO RUNNING、NO WALKING SEQUENCE、NO JUMP、NO FIGHTING、NO HAND-CAST POWER、NO GENERIC SUPERHERO ACTION、NO NORMAL ANIME PV SECTION、NO GAME STYLE ONLY IN SECOND HALF、NO STATIC SLIDESHOW、NO SLOW TRANSITION、NO LONG HOLD、NO FADE、NO THIRD CHARACTER、NO DUPLICATE、NO FACE CHANGE、NO EXTRA LIMBS、NO RANDOM TEXT / NUMBERS、NO WATERMARK、NO SUBTITLE。\n\n【FINAL INTENT】\n四角いブロックを高速で並べる映像ではない。2人のキャラクターと都市が、巨大な印刷原画・ポスター・漫画ページとして何層にも重なり、その紙面が視線、髪の流れ、タイポグラフィ、ネオンの軌道に沿って不規則に裂け、めくれ、下層の次ショットへ連続的に繋がる。人物は静かでクール、編集・紙面・都市・文字・カメラ・音楽だけが凶暴に速い。全15秒を一つの高級GAME PVとして成立させる。",
    "promptSourceUrl": "https://x.com/sailorv321/status/2093831877006397828",
    "sourceUrl": "https://x.com/sailorv321/status/2093831877006397828",
    "sourceLabel": "X 原帖 · @sailorv321",
    "author": "C’est La Vie | AI Director",
    "publishedAt": "2026-08-29T22:43:15.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093831877006397828.mp4",
    "posterUrl": "/posters/x/2093831877006397828.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax Design × H3で GAME PV風 🎮\n\nMiniMax Designにある「Urban Superpower JRPG PV」のテンプレートをベースに、自分のキャラクターに合わせて演出を改良してみました👀\n\nゼロから構成を考えなくても、テンプレートを起点に\nMotion Graphics、Art Typography、画面転換まで\n自分好みに作り込んでいけるのが便利✨\n\n今回はRipped Paper＋2.5D＋AE Motionを追加して、\nよりスピード感のあるGAME PV風に🎬\n\nMade with MiniMax Design × H3\n#MiniMaxH3 on #MiniMaxDesign\n\nMiniMax Design × H3 Prompt ⬇️\n\nMODEL：MiniMax H3｜15秒｜16:9｜24fps\nREFERENCE：[参照1]＋[参照2] ONLY\nSTYLE：original urban superpower 2D game PV × premium JRPG opening × rebellious ripped-paper collage × aggressive AE motion design × fashion editorial game graphics。RED / BLACK / WHITEを主軸に、参照キャラクター固有色は維持。シャープな2Dセル、漫画スクリーントーン、印刷ズレ、インク輪郭、裂けた紙繊維、ネオン反射、非対称UI、巨大タイポを統合。\n\n【REFERENCE / IDENTITY LOCK】\n[参照1]＝CHARACTER A、[参照2]＝CHARACTER B。2人とも成人女性。顔、目、輪郭、髪、身体比率、衣装、アクセサリーを15秒維持。A/Bの融合、顔交換、衣装交換、別人化、複製、第三人物は禁止。\n\n【CORE】\n人物を走らせて速度を作らない。CHARACTER MOTION = MINIMAL、CAMERA / GRAPHICS / TYPOGRAPHY / CITY / AUDIO = EXTREME。A/Bは視線、瞬き、僅かな首振り、呼吸、髪と衣装の揺れのみ。走る、歩行シークエンス、ジャンプ、格闘、手から能力を撃つ演出は禁止。映像全体を「静かな2人のKEY VISUALを、都市・原画・文字・UI・紙面そのものが狂った速度で再編集するGAME PV」とする。\n\n【MOST IMPORTANT — ORGANIC RIPPED SCREEN DESIGN】\n画面を四角いブロック、均等グリッド、カード矩形で切り分けない。ブロック状の分割を主役にしない。切断面は「紙を手で高速に引き裂いたような、不規則で細長い裂け目」。毛羽立った紙繊維、ギザギザの非対称エッジ、細い裂傷、斜めに伸びる破れ、曲線と鋭角が混ざる有機的輪郭。裂け目は一本の直線ではなく、途中で枝分かれし、細くなり、再び広がる。画面中央で整然と4分割しない。顔や都市の輪郭に沿って裂け、髪の流れや視線の方向から自然に亀裂が発生する。裂けた前景紙面の下から次ショット、赤いハーフトーン、別角度の都市、H3 / MiniMax / Designが覗く。紙片は厚いパネルではなく薄い2D原画シートとして前後へしなる。直線WIPEではなくTEAR → PEEL → CURL → WHIP → REVEALを0.08〜0.25秒で連鎖。\n\n【GAME VISUAL SYSTEM】\nGAME UIは後半だけではなく第1フレームから最終フレームまで常時存在。ただしUIも四角い箱を大量に置かず、裂けた紙面の輪郭へ沿わせる。CORNER BRACKETは片側だけ、TARGET LINEは途中で切れる、STATUS BARは文字なしの細い赤線として紙の裂け目へ吸収。CROSSHAIR、HALFTONE、CHARACTER CUT-IN、TEXTLESS GAUGE、2.5D ART CELL、NEON TRACE、SCREEN TEAR、KINETIC TYPEを毎フレーム最低2種類併用。UIは装飾ではなく次画面を生むMASK / MATTE / CAMERA PATH。\n\n【TEXT LOCK】\n可読文字は「H3」「MiniMax」「Design」の3種類のみ。日本語、中国語、数字、架空UI文字、看板、追加英単語、字幕、ウォーターマーク禁止。文字は字幕ではなく巨大GAME TITLE OBJECT、裂け目の下層、TRACK MATTE、前景遮蔽物として使う。\n\n【TEMPO】\n平均0.25〜0.55秒ごとに構造更新。MICRO TRANSITION 0.06〜0.22秒。0.7秒以上同じ構図を保持しない。前ショットが終わる前に次ショットが始まる。常時2〜4レイヤーが別速度で進行。FAST BUT FLUID、NOT JERKY。FADE、SLOW DISSOLVE、静止スライド、長い溜め、文字が止まるのを待つ編集は禁止。\n\n【0.0–0.8s】\n最初から完成したGAME PV画面。Aの片目CUを大きく配置し、その上へRED HALFTONE、細いWHITE UI LINE、都市夜景を重ねる。画面右上から紙面に細い裂傷が走り、Aの髪の流れに沿って亀裂が斜め下へ伸びる。裂け目の奥にはAの別CUと都市の赤い光。裂け目を0.1秒単位で増殖させ、紙面の端が僅かにめくれ、その隙間へカメラがPUSH。四角いPANELへ変換しない。\n\n【0.8–1.6s】\n裂けた紙片が前景へWHIPし、裏面がBLACK→REDへ瞬間反転。その紙片の形状そのものが巨大H3のMASKへ変形。H3は画面中央へ普通に表示せず、Hの一部だけ超前景、3は奥景、残りは裂け目越しに断続的に見える。カメラがHの縦画をすれすれに通過。HがBの目を0.1秒隠した瞬間、遮蔽箇所だけBのCUへ置換。文字を完成表示して待たない。\n\n【1.6–2.4s】\nBの顔CU。Bは目線だけ僅かに上げる。Bの髪の外周から白い紙面が裂けるように細い亀裂が3方向へ走り、裂けた奥から都市、Aの目、RED SCREEN TONEを同時露出。裂け目ごとにZ深度を変え、前景紙片は手前へめくれ、奥の都市は逆方向へPULL。細いUIラインが裂け目に沿って走り、そのまま次TEARの起点になる。\n\n【2.4–3.2s】\n都市のビル壁面が紙ポスターのように縦へ剥がれ、下層からMiniMax。文字はWHITE、背後にRED SHADOW、さらにBLACK OFFSET。ただし長方形の赤帯を固定表示しない。赤は裂けた紙の裏面として断片的に見せる。MiniMaxの文字面をTRACK MATTEにし、内部だけA/B/都市を高速MATCH CUT。文字外周も0.1秒ごとに削れ、破れた紙繊維へ変形。カメラは文字の隙間を通過。\n\n【3.2–4.0s】\nAとBのDUAL CUT-IN。左右50/50の分割は禁止。A側は左下から斜めに裂けた大きな紙片、B側は右上から侵食する細長い破れとして非対称配置。中央境界は直線ではなく複雑な裂け目。A/Bは静止中心。背景交差点だけが5層の原画セルへ分離し逆PARALLAX。裂けた境界をRED NEON TRACEが0.12秒で走り、通過した場所から次構図へ置換。\n\n【4.0–4.8s】\nNEON TRACEが画面端へ到達する前に、前景の紙面を勢いよく横へPEEL。下からAの別CU、Bの瞳、都市道路、ハーフトーンが一枚のコラージュとして現れる。各素材は四角く切らず、すべて破れた不規則輪郭。小さな紙片が飛散するのではなく、大きな原画シートが薄くしなりながらカメラ前を横切る。DesignのDだけが裂け目の奥に先行出現。\n\n【4.8–5.6s】\nDが巨大前景へ抜け、その後ろからesignが0.08秒遅れで追従しDesign完成。ただし0.2秒以内に文字の一部が紙の裂け目へ隠れる。WHITE NEON SWEEPが文字輪郭をなぞり、SWEEP通過部だけ都市線画とBの輪郭が高コントラスト化。Designの斜線を次の裂傷へ連続変形し、画面を細長く引き裂く。裂けた下層にH3の一部を既に見せる。\n\n【5.6–6.4s】\nH3 / MiniMax / Designを3層のGAME ARCHITECTUREとしてZ空間に分散。直線的なロゴ並びは禁止。H3は超前景で一部画面外、MiniMaxは裂けた中層紙面、Designは都市奥景へ斜め配置。文字間をA/Bの顔と都市が覗く。カメラが文字と紙層の間を高速で縫う。各紙層は別方向へPEELし、次のA CUT-INへ連続接続。\n\n【6.4–7.2s】\nAのULTIMATE CUT-IN風CU。能力発射はしない。Aの瞳が僅かに横へ動くと、その視線延長上にRED INK TEARが走る。都市が爆発するのではなく、街全体が印刷原画として6〜9枚に剥離。道路、窓、ネオン、空が紙の層としてずれ、間からBの顔が出現。破断面は全てギザギザで有機的。カメラは紙層の隙間を高速FLY-THROUGH。\n\n【7.2–8.0s】\nBへ到達する直前、Bの顔周囲だけ紙が円形ではなく不規則に裂け、目、髪、衣装の一部が別深度へ滑る。B自体を複製しない。クロップのみ。薄いWHITE CROSSHAIRが0.08秒横断し、その軌跡が次の裂け目へ変化。背景はRED / BLACK / WHITEの印刷レイヤーが0.05秒差でズレ続け、静止感をなくす。\n\n【8.0–8.8s】\nAとBを交互に見せるのではなく同時存在させ、手前の裂けた紙片が一瞬Aを隠す間に下層だけBへ切り替えるMATCH REVEAL。さらに別の紙片でBを隠しAへ戻す。人物が移動しているように見せず、編集そのものがキャラを交換。H3巨大クロップが奥からPUSH、MiniMaxが逆方向SLIDE、Designは紙面裏側に断片表示。\n\n【8.8–9.6s】\nREDの大きな印刷面が画面右から侵入するが矩形ブロックではなく、上下が激しく裂けた一枚紙。中央から縦に引き裂かれ、その間からBの大きな顔とDesign。最前景にはAの片目。WHITE UI BRACKETは4隅全部を囲まず2辺だけが短く点灯。0.15秒で位置を変え、裂け目へ吸収。\n\n【9.6–10.4s】\nカメラが巨大H3の側面を通過し、その奥にA/Bの3種類のGAME KV候補を0.2秒単位で連続MATCH CUT。各KVは四角いフレームではなく、裂けた紙の巨大シルエットで切り取る。1つ目A優勢、2つ目B優勢、3つ目A/B同格。ポーズ変化の途中は見せず、異なる完成構図を高速置換。背景CITYは常に逆PARALLAX。\n\n【10.4–11.2s】\n3つ目のKVが完成する前に画面全体が一本の大きな斜め裂傷から連鎖的に破れる。細い亀裂→枝分かれ→紙面がめくれる→下層の次KVが現れる。大量の小ブロックに分割しない。裂けた面は3〜5枚程度の大きな不規則シート。BLACK NEGATIVE SPACE、RED HALFTONE、WHITE PAPER FIBERを隙間へ見せる。カメラは最も細い裂け目へPUSH。\n\n【11.2–12.0s】\n裂け目を抜けるとA左寄り、B右寄りのNEAR FINAL KV。しかし左右対称は禁止。Aは大きな前景、Bは少し奥、H3が両者を斜めに跨ぎ、MiniMaxは中景、Designは都市壁面の裂けた紙層へ統合。0.25秒だけ完成感を出した後、H3の一画が前景へWHIPし画面を再び引き裂く。\n\n【12.0–12.8s】\n裂けたH3の一画をCAMERA PATHとして利用し、A/Bの目CU、都市、赤い印刷面を高速通過。WHITE NEON TRACEとRED INK LINEが逆方向から走り、交差地点で紙面がPEEL。音と映像を完全同期。人物はほぼ静止なのに、周囲の紙面・都市・文字・UIだけが激しく動き続ける。\n\n【12.8–13.6s】\nFINAL KV再構築開始。A/Bを中央付近へ固定し、背後の都市原画が大きな破れた紙層として奥から一枚ずつ高速収束。H3＝最大前景、MiniMax＝横方向中景、Design＝斜め奥景。文字を一列に並べない。各層が0.05〜0.12秒差で互いを追い越し、RED / BLACK / WHITEの構造を完成させる。\n\n【13.6–14.4s】\nWHITE LIGHT TEARが左→右、0.1秒後RED LIGHT TEARが右→左へ交差。単なる光線ではなく、通過した場所で紙表面が一瞬めくれ下層原画を露出。交差点でA/Bの目CUが瞬間的に覗き、すぐFINAL構図へ戻る。肌や髪を赤一色に染めない。最後までGAME UI、HALFTONE、裂けた紙輪郭を維持。\n\n【14.4–15.0s】\nFINAL GAME CHARACTER KV。A/B中心、H3最大、MiniMax中景、Design斜めサブグラフィック。裂けた紙面、都市、RED HALFTONE、BLACK NEGATIVE SPACE、WHITE FIBER EDGE、UI LINE、NEON EDGEを一体化。14.8秒までMICRO PARALLAXと紙端の僅かな揺れを継続し、最後0.2秒だけほぼ固定。BLACK OUT禁止。\n\n【CAMERA / TRANSITION】\nRIPPED-PAPER TRACK MATTE、TEAR REVEAL、PEEL TRANSITION、PAPER CURL、INK TEAR、ORGANIC EDGE WIPE、MATCH MOVE、MATCH OCCLUSION、TYPOGRAPHY FLY-THROUGH、EXTREME PUSH-PULL、2.5D PARALLAX、Z-DEPTH CROSSING、NEON TRACE、HALFTONE REVEALを複合。RECTANGULAR BLOCK WIPE、均等GRID、タイル分割、カードを並べただけの画面は禁止。同じ転換を反復せず、前ショットの裂け目・文字・髪の流れ・都市輪郭を次ショットのMASKへ変換。\n\n【AUDIO / MUSIC】\nMUSIC ON。15秒全体を一本の高速GAME PV TRACKとして作る。132 BPM前後、dark urban alternative hip-hop × cinematic breakbeat × stylish JRPG battle menu energy。重いSUB BASS、乾いたKICK、タイトなSNARE、短いBREAKBEAT、歪ませすぎないBASS PULSE、低いSYNTH DRONE、逆再生TEXTURE、短いSTRING STAB、METALLIC RISERを使用。EDM DROP、明るいアイドルPOP、陽気なFUNKは禁止。0秒からBEAT開始し、途中で音楽を切らない。映像の0.08〜0.25秒転換すべてに細かく音を付けすぎず、MUSICの大きなビートと映像構造を同期する。\n\n【SFX】\n紙を裂く「ビリビリ」そのままの安い効果音にしない。RIPPED PAPERは低域を含む短いTEXTURED TEAR＋AIR WHOOSHとして高級に処理。PEELは薄いREVERSE SWELL、巨大文字通過はDEEP WHOOSH、PUSH/PULLはSUB IMPACT、NEON TRACEは短いELECTRIC SWEEP。SFXはMUSICへ溶け込ませ、毎回同じ音を連打しない。repetitive click / cartoon pop / cheap UI beep / excessive explosionは禁止。\n\n【ABSOLUTE】\nNO BLOCKY SCREEN DIVISION、NO RECTANGULAR TILE GRID、NO CLEAN 4-WAY SPLIT、NO RUNNING、NO WALKING SEQUENCE、NO JUMP、NO FIGHTING、NO HAND-CAST POWER、NO GENERIC SUPERHERO ACTION、NO NORMAL ANIME PV SECTION、NO GAME STYLE ONLY IN SECOND HALF、NO STATIC SLIDESHOW、NO SLOW TRANSITION、NO LONG HOLD、NO FADE、NO THIRD CHARACTER、NO DUPLICATE、NO FACE CHANGE、NO EXTRA LIMBS、NO RANDOM TEXT / NUMBERS、NO WATERMARK、NO SUBTITLE。\n\n【FINAL INTENT】\n四角いブロックを高速で並べる映像ではない。2人のキャラクターと都市が、巨大な印刷原画・ポスター・漫画ページとして何層にも重なり、その紙面が視線、髪の流れ、タイポグラフィ、ネオンの軌道に沿って不規則に裂け、めくれ、下層の次ショットへ連続的に繋がる。人物は静かでクール、編集・紙面・都市・文字・カメラ・音楽だけが凶暴に速い。全15秒を一つの高級GAME PVとして成立させる。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 1142,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093828046935486968",
    "title": "超级反派被控制台密码难住",
    "titleEn": "A Supervillain Is Defeated by a Control-Panel Password",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "披风反派准备启动末日武器，却因密码错误在整齐列队的手下面前陷入尴尬。 原生视频规格为 15 秒 · 1536×672。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A caped villain prepares to launch a doomsday weapon but gets stuck on a password in front of rows of waiting henchmen. Native-video output: 15s · 1536×672. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A hilarious 15-second 3D animated comedy sequence featuring colorful stylized feature-animation characters, extremely expressive faces, exaggerated body language, cinematic lighting, polished rendering, dramatic orchestral music, and razor-sharp comedic timing.  \n\nInside an enormous secret volcano lair, a terrifying supervillain in black armor and a dramatic cape stands before a gigantic glowing control panel while hundreds of identical henchmen wait in perfect formation. Outside, a colossal doomsday laser rises from the volcano and points toward the moon. Lightning flashes. Alarms pulse. The villain slowly raises one finger over a huge red button labeled ACTIVATE and announces dramatically, **“At last... the world will kneel before me!”**  He presses the button.  Nothing happens.  A tiny computer voice says, **“Password required.”**  \nThe villain freezes.  He types confidently.  **“Incorrect password.”**  He tries again.  **“Incorrect password.”**  His confident smile disappears. Hundreds of henchmen awkwardly stare at the floor.  The villain angrily whispers, **“Was it... EvilMaster?”**  A henchman quietly suggests, **“Capital E?”**  The villain tries it.  **“Account locked. Try again in 24 hours.”**  Complete silence.  The villain slowly turns toward his entire army, absolutely humiliated.  One henchman cautiously raises his hand.  **“Should we... conquer tomorrow?”** \n The villain points furiously toward the exit.  **“EVERYBODY GO HOME!”**  Instantly, hundreds of henchmen cheer, throw their helmets into the air, and sprint toward the exit like schoolchildren released early.  One henchman runs past yelling, **“THREE-DAY WEEKEND!”**  \n\nThe supervillain remains completely alone beside his useless doomsday machine while sad elevator music begins playing from the control panel.  Bright polished 3D animated feature-film quality, spectacular villain lair, exaggerated reactions, squash-and-stretch comedy, epic visuals contrasted with mundane frustration, fluid cinematic camera movement, perfectly readable action, consistent characters, no subtitles, no title card.",
    "promptSourceUrl": "https://x.com/DeCat2025/status/2093828046935486968",
    "sourceUrl": "https://x.com/DeCat2025/status/2093828046935486968",
    "sourceLabel": "X 原帖 · @DeCat2025",
    "author": "DeCat",
    "publishedAt": "2026-08-29T22:28:02.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093828046935486968.mp4",
    "posterUrl": "/posters/x/2093828046935486968.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1536×672",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "music-performance",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Supervillain\nMinimax H3 @Hailuo_AI \n\nA hilarious 15-second 3D animated comedy sequence featuring colorful stylized feature-animation characters, extremely expressive faces, exaggerated body language, cinematic lighting, polished rendering, dramatic orchestral music, and razor-sharp comedic timing.  \n\nInside an enormous secret volcano lair, a terrifying supervillain in black armor and a dramatic cape stands before a gigantic glowing control panel while hundreds of identical henchmen wait in perfect formation. Outside, a colossal doomsday laser rises from the volcano and points toward the moon. Lightning flashes. Alarms pulse. The villain slowly raises one finger over a huge red button labeled ACTIVATE and announces dramatically, **“At last... the world will kneel before me!”**  He presses the button.  Nothing happens.  A tiny computer voice says, **“Password required.”**  \nThe villain freezes.  He types confidently.  **“Incorrect password.”**  He tries again.  **“Incorrect password.”**  His confident smile disappears. Hundreds of henchmen awkwardly stare at the floor.  The villain angrily whispers, **“Was it... EvilMaster?”**  A henchman quietly suggests, **“Capital E?”**  The villain tries it.  **“Account locked. Try again in 24 hours.”**  Complete silence.  The villain slowly turns toward his entire army, absolutely humiliated.  One henchman cautiously raises his hand.  **“Should we... conquer tomorrow?”** \n The villain points furiously toward the exit.  **“EVERYBODY GO HOME!”**  Instantly, hundreds of henchmen cheer, throw their helmets into the air, and sprint toward the exit like schoolchildren released early.  One henchman runs past yelling, **“THREE-DAY WEEKEND!”**  \n\nThe supervillain remains completely alone beside his useless doomsday machine while sad elevator music begins playing from the control panel.  Bright polished 3D animated feature-film quality, spectacular villain lair, exaggerated reactions, squash-and-stretch comedy, epic visuals contrasted with mundane frustration, fluid cinematic camera movement, perfectly readable action, consistent characters, no subtitles, no title card.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 6,
      "views": 116,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093828013427474485",
    "title": "仿生手接住水滴并浇灌新芽",
    "titleEn": "A Robotic Hand Catches Water and Feeds a Seedling",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人手将水滴入仿生机械手掌，液体汇向嫩芽，最后切到湿润的机器人面孔。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A human hand drips water into a robotic palm, feeding a seedling before the scene cuts to a rain-wet android face. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/xiaoao/status/2093828013427474485",
    "sourceLabel": "X 原帖 · @xiaoao",
    "author": "xiaoao",
    "publishedAt": "2026-08-29T22:27:54.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093828013427474485.mp4",
    "posterUrl": "/posters/x/2093828013427474485.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MINIMAX H3 MAX 只用了三秒生成的视频！！！",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 0,
      "views": 48,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093826321587847398",
    "title": "熟食店三明治被小型飞碟劫走",
    "titleEn": "A Tiny Flying Saucer Abducts a Deli Sandwich",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "男子在熟食店吃三明治，窗外飞碟靠近并用光束将食物从手中吸走。 原生视频规格为 12 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man eats a sandwich in a deli before a saucer approaches the window and beams the food out of his hands. Native-video output: 12s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ScottShebby/status/2093826321587847398",
    "sourceLabel": "X 原帖 · @ScottShebby",
    "author": "ScottShebby",
    "publishedAt": "2026-08-29T22:21:11.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093826321587847398.mp4",
    "posterUrl": "/posters/x/2093826321587847398.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sandwich Abduction - MiniMax H3\n@MiniMax_AI @runwayml",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 26,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093825302904586668",
    "title": "小螃蟹沿月夜池塘完成四段旅程",
    "titleEn": "A Tiny Crab Crosses a Moonlit Pond in Four Chapters",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "小螃蟹在月夜池塘、荷叶与岸边场景间前行，四个片段串成长达半分钟的旅程。 原生视频规格为 32 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny crab moves among a moonlit pond, lily pads, and the shore as four segments form a half-minute journey. Native-video output: 32s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tea_story_hoshi/status/2093825302904586668",
    "sourceLabel": "X 原帖 · @tea_story_hoshi",
    "author": "おちゃばなし | tea_and_stories",
    "publishedAt": "2026-08-29T22:17:08.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093825302904586668.mp4",
    "posterUrl": "/posters/x/2093825302904586668.jpg",
    "duration": 32,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "fantasy",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "local minmaxH3検証 / サワガニのクララの日常🦀\n・昨夜投げた１回のバッチ処理で出来た映像\n（映像としてはおかしい部分もそのまま）\n・音楽はminmaxMusic3で一緒に生成したものを結合\n・8秒 の４つのセグメントを１チャプチャーで出力\n・VRAM16Gの関係で1秒しか前チャプターを参照してないので整合性が崩れるのがとても惜しい。\n・生成環境の詳細は👇\n-----\n- ComfyUI 0.33.4\n- PyTorch 2.13.0 + CUDA 12.6\n- GPU NVIDIA RTX 4060 Ti（VRAM 16GB）／システムRAM 64GB\n- OS Windows（リモート実行、Mac から操作）\n- ベース MiniMax-H3 Ref2VA（minimax_h3_ref2va_pruned_int8_convrot）\n- 加速LoRA MiniMax-H3-Ref2VA-Acc-8Step（alibaba-pai 公式の PDD 蒸留）\n- テキストエンコーダ Qwen3-VL 32B（NVFP4 AWQ量子化）\n- VAE 映像 fp16 ／ 音声 fp32（別々）\n\nカスタムノード\n- ComfyUI-H3-Multishot — マルチショット連鎖の中核\n- ComfyUI-H3-Multishot-Wordless — 無言版サンプラー（発話の宣言を注入しない派生）\n- ComfyUI-MiniMax-H3-PDD-Acc — PDD ヘッドバンクの読み込みと融合\n- ComfyUI-Spectrum-MiniMax-H3（今回は PDD と併用不可のため OFF）\n- ComfyUI-KJNodes / ComfyUI-FreeMemory\n\n生成設定\n- 解像度 960×544（16:9）\n- 尺 1セグメント 8.0秒（192フレーム）／ 24fps\n- サンプラー euler、8ステップ（PDD nfe=8、CFG不要）\n- スケジューラ PDD の訓練済みシグマ（SigmaShift 12/3）\n- 連鎖 チャプター間は cut、引き継ぎは最終フレーム＋末尾1秒\n- 参照 キャラクター画像1枚\n- 音声 映像と同時生成（H3はAV同時生成モデル）\n- 書き出し MP4 / H.264 / CRF14 / 8bit\n\n所要時間（実測）\n- 1チャプター＝8秒 で 15.2〜18.2分\n- 32秒（4セグメント）で 67.5分\n- 従来の20ステップ比で 約28%短縮（PDD 蒸留の効果）\n\n制作ツール\nh3Deck / 自作の単一HTMLフロントエンド。ComfyUI API を直接叩き、チャプターを順次発注\n\n#minmaxh3 #ローカル生成の沼 #サワガニのクララ",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 13,
      "views": 633,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093821064631771446",
    "title": "粉发主持人用日文动效介绍免费生成",
    "titleEn": "A Pink-Haired Host Introduces Free Generation with Japanese Type",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "粉发动画主持人在霓虹舞台挥手，日文标题和几何光效同步突出生成信息。 原生视频规格为 5 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A pink-haired animated host gestures on a neon stage as Japanese type and geometric light effects emphasize generation information. Native-video output: 5s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "日本の可愛い美少女アニメキャラクターがMiniMax H3 Maxの紹介をするショート映像。「MiniMax H3 Maxが無料で使える！」という文字をインパクトのあるモーショングラフィックで表現。女の子は大げさなモーションで、日本語で短い台詞を喋る",
    "promptSourceUrl": "https://x.com/SSSS_CRYPTOMAN/status/2093821064631771446",
    "sourceUrl": "https://x.com/SSSS_CRYPTOMAN/status/2093821064631771446",
    "sourceLabel": "X 原帖 · @SSSS_CRYPTOMAN",
    "author": "SSSS.CRYPTOMAN⚡️AI",
    "publishedAt": "2026-08-29T22:00:17.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093821064631771446.mp4",
    "posterUrl": "/posters/x/2093821064631771446.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "anime"
    ],
    "scenes": [
      "product",
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Maxが無料で使えるぞ！\n５秒の動画が５秒以下で出力されるとかヤバすぎｗ\n日本語のテキスト精度も現状最高レベル\nこの爆速生成は絶対体験してみた方がいい\n\n🔗ここからログイン不要・無料で１日５回使えます\nhttps://fal.ai/tools/minimax-h3-max\n\n👇この動画のプロンプト\n日本の可愛い美少女アニメキャラクターがMiniMax H3 Maxの紹介をするショート映像。「MiniMax H3 Maxが無料で使える！」という文字をインパクトのあるモーショングラフィックで表現。女の子は大げさなモーションで、日本語で短い台詞を喋る",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 40,
      "views": 3171,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093820981190021435",
    "title": "漏水喷壶牵出花园里的温柔寓言",
    "titleEn": "A Leaking Watering Can Leads to a Garden Parable",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "漏水喷壶、志愿者和老园丁在田圃与山野间展开故事，长篇镜头最终回到被浇灌的土地。 原生视频规格为 38 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A leaking can, a volunteer, and an old gardener move through plots and hills in a long story that returns to the watered ground. Native-video output: 38s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2093820981190021435",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-08-29T21:59:57.000Z",
    "addedAt": "2026-08-30T00:39:42.827Z",
    "mediaUrl": "/media/x-2093820981190021435.mp4",
    "posterUrl": "/posters/x/2093820981190021435.jpg",
    "duration": 38,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A leaking watering can. A volunteer sure it's a waste. An old gardener who only says: come and find me at closing.\n\n▶ Full film: https://www.youtube.com/watch?v=a2HSShNXCeU\n\nMade with MiniMax-H3\n\n#ShortFilm #Parable #AIFilm",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 33,
      "snapshotAt": "2026-08-30T00:39:42.827Z"
    },
    "approvedAt": "2026-08-30T00:39:42.827Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093820848088215623",
    "title": "波点南瓜升空完成艺术家纪念动画",
    "titleEn": "A Polka-Dot Pumpkin Rises in an Artist Tribute",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黄色波点南瓜与简笔人物在海边反复升降，以极简卡通形式完成纪念片段。 原生视频规格为 5 秒 · 720×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A yellow polka-dot pumpkin rises and falls beside simple figures on a shore in a minimalist animated tribute. Native-video output: 5s · 720×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/haruharuy/status/2093820848088215623",
    "sourceLabel": "X 原帖 · @haruharuy",
    "author": "山下治城:YamashitaHaruki",
    "publishedAt": "2026-08-29T21:59:26.000Z",
    "addedAt": "2026-08-30T00:39:42.828Z",
    "mediaUrl": "/media/x-2093820848088215623.mp4",
    "posterUrl": "/posters/x/2093820848088215623.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "R.I.P.草間彌生さん　動画🎥MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 57,
      "snapshotAt": "2026-08-30T00:39:42.828Z"
    },
    "approvedAt": "2026-08-30T00:39:42.828Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093820333614571731",
    "title": "红色玻璃时间线包围未来舞者",
    "titleEn": "Red Glass Timelines Fold Around a Futuristic Dancer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白舞者在红色玻璃框架与折叠时间线中穿行，镜头最终把空间压成悬浮结构。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black-and-white dancer moves through red glass frames and folding timelines before the camera compresses the space into a floating structure. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[Core Concept]\n15s, 160 BPM, 4/4, 40 beats. Industrial Hyperpop × Deconstructed Club: hard kicks, metallic snares, glitch hi-hats, distorted bass, sliced vocals. A timeline-based reality where rhythm controls space and motion.\n\n[Character Identity]\nSingle woman dancer, Maintain identical face, body, hairstyle and outfit. Multiple figures are only past-frame projections.\n\n[World Logic]\nEach kick compresses the timeline toward the dancer. Each snare duplicates previous movements into adjacent frames. Moving against the timeline transforms the 2D interface into a foldable 3D structure. The dancer escapes the repetition loop.\n\n[Visual Language]\nBlack, cold white, warning red. Glass timelines, metal rails, glowing frame lines, scanning grids. Extreme wide-angle, low\nangles, top-down shots, Dutch angles, dolly moves, impossible perspective. 3D word MINIMAX exists as physical architecture, not subtitles.\n\n[Motion Rules]\nUse anticipation → attack → overshoot → settle. Kick drives spatial compression, snare breaks frames, hi-hats animate cursors and fragments, bass bends time. Build with glitch cuts, then explode into an impossible continuous long take during the drop.\n\n[Restrictions]\nNo character changes, extra people, random text, subtitles, logos, outfit changes, bad anatomy, flat camera angles, or meaningless flashes.",
    "promptSourceUrl": "https://x.com/ivanfioravanti/status/2093820336273756636",
    "sourceUrl": "https://x.com/ivanfioravanti/status/2093820333614571731",
    "sourceLabel": "X 原帖 · @ivanfioravanti",
    "author": "Ivan Fioravanti ᯅ",
    "publishedAt": "2026-08-29T21:57:23.000Z",
    "addedAt": "2026-08-30T00:39:42.828Z",
    "mediaUrl": "/media/x-2093820333614571731.mp4",
    "posterUrl": "/posters/x/2093820333614571731.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I tested MiniMax H3 Max by @fal and it's really fast and furious! \n\n15 seconds video:\n- 768p ~16 secs\n- 480p ~ 4 secs\n\nHere are 3 examples. Prompt below.",
    "engagement": {
      "replies": 7,
      "reposts": 1,
      "likes": 18,
      "views": 1919,
      "snapshotAt": "2026-08-30T00:39:42.828Z"
    },
    "approvedAt": "2026-08-30T00:39:42.828Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093818061275226260",
    "title": "黑发学生从商业街走到河岸",
    "titleEn": "A Dark-Haired Student Walks from Shopping Street to Riverbank",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发学生在日本商业街、电车和河岸间连续出现，城市实景与动画角色保持统一。 原生视频规格为 26 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired student appears across a Japanese shopping street, train, and riverbank while the animated character stays consistent against live-action scenery. Native-video output: 26s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ebigawa_pico/status/2093818061275226260",
    "sourceLabel": "X 原帖 · @ebigawa_pico",
    "author": "海老川ピコ",
    "publishedAt": "2026-08-29T21:48:21.000Z",
    "addedAt": "2026-08-30T00:39:42.828Z",
    "mediaUrl": "/media/x-2093818061275226260.mp4",
    "posterUrl": "/posters/x/2093818061275226260.jpg",
    "duration": 26,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "https://fal.ai/tools/minimax-h3-max\nminimax-h3-max 試してみた",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-08-30T00:39:42.828Z"
    },
    "approvedAt": "2026-08-30T00:39:42.828Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2093816350196048105",
    "title": "舞厅庆典迎来夸张的壮汉舞者",
    "titleEn": "A Ballroom Celebration Welcomes an Exaggerated Strongman",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "霓虹舞厅中的人群为壮汉欢呼，传统服饰舞者与夸张表情把短片推向荒诞庆典。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A neon ballroom crowd cheers a muscular figure as traditional-costume dancers and exaggerated faces turn the scene into an absurd celebration. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhiMarHal/status/2093816350196048105",
    "sourceLabel": "X 原帖 · @PhiMarHal",
    "author": "PhiMarHal",
    "publishedAt": "2026-08-29T21:41:33.000Z",
    "addedAt": "2026-08-30T00:39:42.828Z",
    "mediaUrl": "/media/x-2093816350196048105.mp4",
    "posterUrl": "/posters/x/2093816350196048105.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Pasted the tweet in Minimax H3 Max.\n\nThis is actually exactly how it went.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 112,
      "snapshotAt": "2026-08-30T00:39:42.828Z"
    },
    "approvedAt": "2026-08-30T00:39:42.828Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094427110764917030",
    "title": "微缩施工队从零搭起汉堡",
    "titleEn": "A Miniature Construction Crew Builds a Burger from Scratch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "微缩工程车和工人依次吊装面包、肉饼与配料，把桌面变成完整的汉堡施工现场。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Miniature vehicles and workers lift buns, patties, and toppings into place, turning a tabletop into a complete burger construction site. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a high-end cinematic tilt-shift food commercial that shows the full construction of a hamburger from start to finish on a clean tabletop, staged like a miniature construction site.  Use a strong tilt-shift look throughout: shallow selective focus, miniature diorama realism, soft focus falloff across the frame, and a low tabletop camera angle that makes the tiny vehicles feel large and heroic while the burger ingredients feel monumental.  The entire video should follow one continuous build process. Begin with an empty build site: the bottom bun is placed at center as the foundation. Then, step by step, miniature construction vehicles bring in and install each ingredient in order. A dump truck delivers sesame seeds near the work zone. A forklift transports a thick tomato slice. A front loader carries shredded lettuce with some loose pieces spilling naturally. A yellow excavator grips and places pickle slices with its bucket. A mobile crane lifts a glossy cheddar slice with hook and sling, lowering it carefully onto the hot grilled beef patty. Onion rings wait to the side like construction coils. Sauce is applied like mortar. Finally, the top bun is lifted and guided into place as the final roof piece, completing the burger.  Include several tiny workers in simple safety helmets directing the build. One signals with a baton, another points toward the stacking area, and others react to each installation step to sell the miniature worksite illusion.  Camera language should feel premium and playful: cinematic push-ins, low-angle dolly moves, miniature-style lateral tracking shots, macro insert shots of vehicle actions, selective rack focus between machines and ingredients, and gentle orbiting hero passes around the burger. Keep the movement smooth and expensive, with the tilt-shift effect always helping the scene feel like a handcrafted miniature world.  Lighting should be warm and polished like a luxury food commercial, with a strong key from upper left, a gentle rim from rear right, glossy highlights on sauce and cheese, and subtle reflections on the construction vehicles. Emphasize realistic textures: toasted bun crust, soft bun crumb, sesame seeds, grill marks on the beef patty, cheddar sheen, wet tomato surfaces, crisp lettuce edges, pickle gloss, and subtle machine wear and decals.  Build the pacing so each ingredient placement feels satisfying and clear. The viewer should always understand that the burger is being assembled in sequence from beginning to end. Finish on a fully completed perfect burger, freshly built, as the vehicles pause around it like a proud construction crew. End with a polished hero shot of the completed hamburger in the center of the miniature worksite.  Overall tone: premium, humorous, charming, visually satisfying, high-end commercial quality, with a strong tilt-shift miniature aesthetic and a clear beginning-to-end burger assembly story.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2094427110764917030",
    "sourceUrl": "https://x.com/aimikoda/status/2094427110764917030",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-31T14:08:30.000Z",
    "addedAt": "2026-08-31T15:24:28.119Z",
    "mediaUrl": "/media/x-2094427110764917030.mp4",
    "posterUrl": "/posters/x/2094427110764917030.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 - Tilt-Shift Burger Commerical - t2v\n\nBurger Under Construction\nA tilt-shift burger commercial where tiny construction crews build the burger from the ground up.\n\nYou can easily adapt this prompt for other foods.\n\nPrompt:\nCreate a high-end cinematic tilt-shift food commercial that shows the full construction of a hamburger from start to finish on a clean tabletop, staged like a miniature construction site.  Use a strong tilt-shift look throughout: shallow selective focus, miniature diorama realism, soft focus falloff across the frame, and a low tabletop camera angle that makes the tiny vehicles feel large and heroic while the burger ingredients feel monumental.  The entire video should follow one continuous build process. Begin with an empty build site: the bottom bun is placed at center as the foundation. Then, step by step, miniature construction vehicles bring in and install each ingredient in order. A dump truck delivers sesame seeds near the work zone. A forklift transports a thick tomato slice. A front loader carries shredded lettuce with some loose pieces spilling naturally. A yellow excavator grips and places pickle slices with its bucket. A mobile crane lifts a glossy cheddar slice with hook and sling, lowering it carefully onto the hot grilled beef patty. Onion rings wait to the side like construction coils. Sauce is applied like mortar. Finally, the top bun is lifted and guided into place as the final roof piece, completing the burger.  Include several tiny workers in simple safety helmets directing the build. One signals with a baton, another points toward the stacking area, and others react to each installation step to sell the miniature worksite illusion.  Camera language should feel premium and playful: cinematic push-ins, low-angle dolly moves, miniature-style lateral tracking shots, macro insert shots of vehicle actions, selective rack focus between machines and ingredients, and gentle orbiting hero passes around the burger. Keep the movement smooth and expensive, with the tilt-shift effect always helping the scene feel like a handcrafted miniature world.  Lighting should be warm and polished like a luxury food commercial, with a strong key from upper left, a gentle rim from rear right, glossy highlights on sauce and cheese, and subtle reflections on the construction vehicles. Emphasize realistic textures: toasted bun crust, soft bun crumb, sesame seeds, grill marks on the beef patty, cheddar sheen, wet tomato surfaces, crisp lettuce edges, pickle gloss, and subtle machine wear and decals.  Build the pacing so each ingredient placement feels satisfying and clear. The viewer should always understand that the burger is being assembled in sequence from beginning to end. Finish on a fully completed perfect burger, freshly built, as the vehicles pause around it like a proud construction crew. End with a polished hero shot of the completed hamburger in the center of the miniature worksite.  Overall tone: premium, humorous, charming, visually satisfying, high-end commercial quality, with a strong tilt-shift miniature aesthetic and a clear beginning-to-end burger assembly story.",
    "engagement": {
      "replies": 7,
      "reposts": 2,
      "likes": 36,
      "views": 1472,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.119Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094279178602283013",
    "title": "食材悬浮重组成精品汉堡广告",
    "titleEn": "Floating Ingredients Reassemble into a Premium Burger Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "面包与食材在微距镜头中分层悬浮，再精准落位并以餐厅广告定格收尾。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Buns and ingredients float apart in macro close-ups, then align precisely and resolve into a polished restaurant-ad finish. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a high-end cinematic burger commercial in portrait format, designed like a premium restaurant campaign.  Open on an extreme macro shot of a golden toasted bun, revealing tiny bake-specks, soft flour dusting and rich surface texture under warm directional light. The camera glides smoothly across the bun, then pulls back as the burger ingredients separate into a perfectly aligned exploded stack above a matte wooden tabletop.  The full burger floats on a single vertical axis: toasted top bun, crisp ruffled lettuce, glossy tomato slices, translucent purple-red onion rings, melted cheddar, a thick char-grilled beef patty and toasted bottom bun. Each layer moves with elegant controlled motion, subtle rotation and realistic weight.  Use premium food-commercial camera movement throughout: macro tracking shots, smooth dolly pushes, slow orbital moves, shallow-focus passes between ingredients and precise rack focuses. Add occasional speed ramps into slow-motion beauty moments. Let tiny crumbs, droplets and subtle steam move through the light for additional depth.  Integrate bold hand-lettered white typography directly into the composition. Words appear between the floating burger layers, following the camera movement with kinetic typography. Letters can slide behind ingredients, reveal through depth, stretch slightly during transitions and lock cleanly into place. Add minimal white doodle strokes and graphic accent lines that animate around key ingredients.  Lighting is luxurious and appetizing: soft directional key from upper-left, controlled fill, warm highlights, deep dimensional shadows and glossy specular detail on tomato, cheese and meat. Background is a refined warm brown gradient with subtle cinematic falloff.  Build toward a final satisfying moment where all ingredients rapidly assemble into one perfect burger. The camera performs a fast controlled push-in, then settles into a polished hero shot. Steam rises gently from the patty, cheese settles over the edges and the typography resolves beside the burger.  Final frame: centered premium burger hero shot, clean composition, elegant white campaign typography and subtle animated graphic accents.  Visual style: luxury food advertising, modern restaurant campaign, cinematic macro photography, rich warm tones, high contrast, shallow depth of field, highly detailed food textures, sophisticated motion design, premium typography, smooth dynamic camera choreography, polished commercial finish.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2094279178602283013",
    "sourceUrl": "https://x.com/aimikoda/status/2094279178602283013",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-08-31T04:20:40.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094279178602283013.mp4",
    "posterUrl": "/posters/x/2094279178602283013.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 - Text to Video, Burger Ad\n\nYou can easily adapt this prompt for any food.\n\nPrompt:\nCreate a high-end cinematic burger commercial in portrait format, designed like a premium restaurant campaign.  Open on an extreme macro shot of a golden toasted bun, revealing tiny bake-specks, soft flour dusting and rich surface texture under warm directional light. The camera glides smoothly across the bun, then pulls back as the burger ingredients separate into a perfectly aligned exploded stack above a matte wooden tabletop.  The full burger floats on a single vertical axis: toasted top bun, crisp ruffled lettuce, glossy tomato slices, translucent purple-red onion rings, melted cheddar, a thick char-grilled beef patty and toasted bottom bun. Each layer moves with elegant controlled motion, subtle rotation and realistic weight.  Use premium food-commercial camera movement throughout: macro tracking shots, smooth dolly pushes, slow orbital moves, shallow-focus passes between ingredients and precise rack focuses. Add occasional speed ramps into slow-motion beauty moments. Let tiny crumbs, droplets and subtle steam move through the light for additional depth.  Integrate bold hand-lettered white typography directly into the composition. Words appear between the floating burger layers, following the camera movement with kinetic typography. Letters can slide behind ingredients, reveal through depth, stretch slightly during transitions and lock cleanly into place. Add minimal white doodle strokes and graphic accent lines that animate around key ingredients.  Lighting is luxurious and appetizing: soft directional key from upper-left, controlled fill, warm highlights, deep dimensional shadows and glossy specular detail on tomato, cheese and meat. Background is a refined warm brown gradient with subtle cinematic falloff.  Build toward a final satisfying moment where all ingredients rapidly assemble into one perfect burger. The camera performs a fast controlled push-in, then settles into a polished hero shot. Steam rises gently from the patty, cheese settles over the edges and the typography resolves beside the burger.  Final frame: centered premium burger hero shot, clean composition, elegant white campaign typography and subtle animated graphic accents.  Visual style: luxury food advertising, modern restaurant campaign, cinematic macro photography, rich warm tones, high contrast, shallow depth of field, highly detailed food textures, sophisticated motion design, premium typography, smooth dynamic camera choreography, polished commercial finish.",
    "engagement": {
      "replies": 13,
      "reposts": 21,
      "likes": 234,
      "views": 10198,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094340753195725179",
    "title": "黑色剪影连闯游泳骑行与跑步赛段",
    "titleEn": "A Black Silhouette Races through Swim, Bike, and Run Stages",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色运动员剪影穿过蓝色水道、粉色自行车坡道与三分屏终点，完成高速铁人三项动效。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black athlete silhouette crosses blue water, a pink cycling ramp, and a three-way finish in a fast triathlon motion graphic. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによるトライアスロン動画を制作する。\n\n映像は、水泳4秒、自転車4秒、ラン4秒の順に展開し、最後の3秒は画面を斜めに3分割して、3つの競技を同時に見せる。スローモーションは使用せず、15秒間を通してスピード感と力強いテンポを維持する。\n\n主人公は、全身が黒一色で描かれた一人の人間のシルエット。顔、髪、肌、競技ウェアなどの細部は描かず、プロのトライアスロン選手を思わせる、引き締まったしなやかな体型として表現する。\n\n水泳、自転車、ランのすべてで、同じ体格、身長、身体の比率を持つ同一人物の黒いシルエットを維持する。背景には競技ごとに異なる明るく鮮やかな色を使用し、黒い人物との強いコントラストを保つ。\n\n【0.0～4.0秒：SWIM】\n\n映像は、鮮やかなシアンとブルーで構成された、抽象的でスタイリッシュな水面から始まる。カメラは水面すれすれの斜め前方から主人公を捉え、黒いシルエットが水を切りながら高速で泳いでくる。\n\n主人公は正確で力強いフリースタイルを行う。左右の腕を交互に大きく前へ伸ばし、水中を後方へ押し出しながら、脚は自然で連続的なフラッターキックを行う。腕の入水、キャッチ、プル、リカバリーを滑らかにつなげ、途中で一度だけ横を向いて自然に呼吸する。\n\n単に水面の上で腕を回しているようには見せず、身体の大部分を水中に維持し、前方へ進む推進力、水の抵抗、浮力、飛沫に物理的な説得力を持たせる。\n\n腕が水面へ入るたびに円形の波紋が広がり、水を後方へ押すとシアンの太いラインと細かな光の粒子が進行方向と反対側へ流れる。キックに合わせて、水中に細いグリッドや波状のシェイプが現れる。\n\nカメラを水面すれすれの追従ショットから、斜め上の俯瞰ショットへ切り替える。俯瞰ショットでは、主人公の全身、水面を切り裂く軌道、左右へ広がる波紋を明確に見せる。\n\n背景には「SWIM」という単語を正確なスペルの大文字で登場させる。「SWIM」は水中から浮かび上がるように現れ、音楽のビートに合わせて拡大、縮小、前後方向への移動、軽い3D回転を行う。\n\n3D回転した「SWIM」は必ず再び正面を向き、単語全体が明確に読める瞬間を作る。文字は主人公の背後や水中へ配置し、泳ぎを妨げないようにする。\n\n4.0秒の強いビートで、主人公の腕が作った大きな円形波紋を画面いっぱいに拡大する。その円形波紋を、次のシーンの自転車の前輪へグラフィック・マッチさせ、シャープに切り替える。\n\n【4.0～8.0秒：BIKE】\n\n拡大した水面の円形波紋が、黒い競技用ロードバイクの前輪へ瞬時に切り替わる。背景の色は、ブルーとシアンから、鮮やかなマゼンタ、コーラル、オレンジへ大胆に変化する。\n\n主人公は黒一色の競技用ロードバイクに乗り、抽象的で近未来的なロードコースを高速で走行する。上半身を低く保った空気抵抗の少ない姿勢でハンドルを握り、両脚でペダルを滑らかに高速回転させる。\n\nロードバイクは、細いフレーム、ドロップハンドル、サドル、クランク、ペダル、チェーン、同じ大きさの2つの車輪を持つ、正しい競技用自転車の形状として描く。\n\nカメラは真横から高速で並走し、主人公の全身と自転車全体を捉える。続いて、地面すれすれから前輪とペダリングを大きく見せるローアングル、斜め上からコースの曲線を見せる俯瞰ショットへ、音楽のビートに合わせて大胆に切り替える。\n\n車輪が回転すると、マゼンタとオレンジの円形ラインが短い軌跡として残る。ペダルを踏み込むたびに背景のシェイプが後方へ加速し、コースを曲がると長方形、三角形、太いライン、グリッドが同じ方向へ大きく傾く。\n\n背景には「BIKE」という単語を正確なスペルの大文字で登場させる。「BIKE」は道路面から立ち上がるように現れ、車輪の回転と連動して3D回転、拡大、縮小、奥行き方向への移動を行う。\n\n3D回転した「BIKE」は必ず再び正面を向き、単語全体が明確に読める瞬間を作る。文字は主人公や自転車と接触する物体ではなく、背景のグラフィック要素として使用する。\n\n8.0秒の強いビートで、主人公がペダルを力強く踏み下ろす。ペダルを踏み込む足元のアップから、ランで足が地面へ接触する瞬間へ、動きの方向とリズムを維持したマッチカットで切り替える。\n\n【8.0～12.0秒：RUN】\n\n背景は、鮮やかなイエロー、ライムグリーン、バイオレットを組み合わせた、抽象的でスタイリッシュなランニングコースへ切り替わる。\n\n主人公はトライアスロン終盤のランを思わせる、力強く持続的なペースで走る。単なる短距離スプリントではなく、上半身を安定させ、肘を自然に曲げ、腕を前後へ効率的に振りながら、高いピッチで前方へ進む。\n\n接地した足の真上へ身体の重心が自然に移動し、反対側の膝が前方へ上がる。足の接地、蹴り出し、空中局面、次の接地を滑らかにつなげ、左右の脚を正しく交互に動かす。\n\nカメラは真横から主人公と同じ速度で追従し、全身のランニングフォームを見せる。続いて、斜め前方から迫ってくる主人公を捉えるショット、足元すれすれのローアングル、斜め上からコースを見下ろす俯瞰ショットへ切り替える。\n\n足が地面へ接触するたびに円形の波紋が広がり、蹴り出すと鋭いラインが後方へ走る。腕の振りや走行リズムに合わせて、長方形、円、三角形、グリッドが画面の上下左右からスライドインし、次のビートでスライドアウトする。\n\n背景には「RUN」という単語を正確なスペルの大文字で登場させる。「RUN」は主人公の後方から急速に追いつくように画面奥から移動し、ビートに合わせて拡大、縮小、3D回転、軽い伸縮、色の切り替えを行う。\n\n3D回転した「RUN」は必ず再び正面を向き、単語全体が明確に読める瞬間を作る。文字で主人公の身体や足元を覆い隠さない。\n\n12.0秒の強いビートで主人公が力強く地面を踏み込む。接地点から3本の太い斜めラインが広がり、そのラインによって画面全体を3つのパネルへ分割する。\n\n【12.0～15.0秒：3競技同時表示】\n\n画面を、少し傾いたシャープな境界線によって、左、中央、右の3つの縦長パネルへ分割する。\n\n左側のパネルでは、シアンとブルーの水中を、同じ黒い主人公がフリースタイルで泳ぎ続ける。背景には「SWIM」を表示する。\n\n中央のパネルでは、マゼンタ、コーラル、オレンジのロードコースを、同じ黒い主人公がロードバイクで高速走行する。背景には「BIKE」を表示する。\n\n右側のパネルでは、イエロー、ライムグリーン、バイオレットのコースを、同じ黒い主人公が力強く走り続ける。背景には「RUN」を表示する。\n\n3つのパネルは静止画にせず、水泳のストローク、自転車のペダリング、ランの足運びを同時に継続させる。それぞれの動きを同じ音楽のビートへ同期させ、3競技が一体となったトライアスロンのリズムを作る。\n\n分割画面では、それぞれのパネルに主人公を1人だけ表示する。同じ競技の人物を同じパネル内で複製しない。各パネルの人物、用具、背景が、隣のパネルへはみ出したり、混ざったりしないようにする。\n\n終盤に向けて、各パネルの波紋、車輪の円形ライン、ランの接地エフェクトを徐々に高密度にする。最後の強いビートで3つの動きが同時にアクセントを作り、パネルの境界線と背景のシェイプが整列する。\n\n最後の約0.6秒は、「SWIM」「BIKE」「RUN」の3語と、3競技を行う主人公の姿が明確に確認できる安定した3分割画面として保持し、クールに終了する。大げさな決めポーズや喜ぶ動作は入れない。\n\n【モーショングラフィックと音楽】\n\n水泳では、水の飛沫、波紋、波状のラインを、音楽の細かいビートへ同期させる。自転車では、ペダリング、車輪の回転、コースの通過を、ベースと連続するパーカッションへ同期させる。ランでは、左右の足の接地を、キックとスネアへ正確に同期させる。\n\n水泳から自転車、自転車からランへ進むにつれて、音楽、カメラ、シェイプ、タイポグラフィーの動きを徐々に高密度にし、最後の3分割画面で3つのリズムを統合する。\n\n背景のシェイプは人物や競技用具を完全に隠さず、それぞれの動作が常に明確に見えるようにする。制御されたモーションブラーを使用するが、人物の手足や自転車の車輪が複数に見えるほど強くしない。\n\n【文字と一貫性】\n\n映像内に登場する文字は「SWIM」「BIKE」「RUN」の3種類だけに限定する。すべて正確なスペルの大文字で表示し、文字の欠落、重複、誤字、別の文字への変化を起こさない。それ以外の文字、数字、企業名、ロゴ、ウォーターマーク、字幕は表示しない。\n\n通常の競技シーンでは、主人公を一人だけにする。最後の3分割画面に限り、同じ主人公を各パネルへ一人ずつ表示する。\n\n水泳シーンでは、自転車やランニングコースを登場させない。自転車シーンでは、水中表現や余分な自転車を登場させない。ランシーンでは、自転車やフープなどの競技用具を登場させない。\n\n競技用ロードバイクは必ず1台だけにする。車輪を2つに限定し、フレーム、ハンドル、サドル、クランク、チェーン、ペダルの形状と位置を維持する。車輪の複製、変形、消失、瞬間移動を発生させない。\n\n自転車走行中は両手を正しくハンドルへ置き、両足をペダルへ固定する。ペダルと脚を正しく連動させ、足がペダルから外れたり、フレームや車輪を貫通したりしない。\n\n主人公の形状、体格、頭、胴体、手足の本数を15秒間一貫させる。手足の変形、身体の分裂、不自然な残像、競技間での性別や体格の変化を発生させない。\n\n大胆でありながら統一感のある配色、シャープなグラフィック、強い遠近感、正確なビートシンクを組み合わせ、とにかくカッコよく、スタイリッシュで、スピード感とファッション性の高い映像に仕上げる。SNSで拡散されそうな、センスが良く、中毒性があり、何度も見返したくなる高品質なトライアスロン動画にする。",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2094340753195725179",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2094340753195725179",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-08-31T08:25:21.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094340753195725179.mp4",
    "posterUrl": "/posters/x/2094340753195725179.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でスポーツ動画\n今回はトライアスロン\n@Hailuo_AI \n\n好きなスポーツ組み合わせて、オリジナルトライアスロン作っても面白いかも😆\n\nプロンプト↓\nMiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによるトライアスロン動画を制作する。\n\n映像は、水泳4秒、自転車4秒、ラン4秒の順に展開し、最後の3秒は画面を斜めに3分割して、3つの競技を同時に見せる。スローモーションは使用せず、15秒間を通してスピード感と力強いテンポを維持する。\n\n主人公は、全身が黒一色で描かれた一人の人間のシルエット。顔、髪、肌、競技ウェアなどの細部は描かず、プロのトライアスロン選手を思わせる、引き締まったしなやかな体型として表現する。\n\n水泳、自転車、ランのすべてで、同じ体格、身長、身体の比率を持つ同一人物の黒いシルエットを維持する。背景には競技ごとに異なる明るく鮮やかな色を使用し、黒い人物との強いコントラストを保つ。\n\n【0.0～4.0秒：SWIM】\n\n映像は、鮮やかなシアンとブルーで構成された、抽象的でスタイリッシュな水面から始まる。カメラは水面すれすれの斜め前方から主人公を捉え、黒いシルエットが水を切りながら高速で泳いでくる。\n\n主人公は正確で力強いフリースタイルを行う。左右の腕を交互に大きく前へ伸ばし、水中を後方へ押し出しながら、脚は自然で連続的なフラッターキックを行う。腕の入水、キャッチ、プル、リカバリーを滑らかにつなげ、途中で一度だけ横を向いて自然に呼吸する。\n\n単に水面の上で腕を回しているようには見せず、身体の大部分を水中に維持し、前方へ進む推進力、水の抵抗、浮力、飛沫に物理的な説得力を持たせる。\n\n腕が水面へ入るたびに円形の波紋が広がり、水を後方へ押すとシアンの太いラインと細かな光の粒子が進行方向と反対側へ流れる。キックに合わせて、水中に細いグリッドや波状のシェイプが現れる。\n\nカメラを水面すれすれの追従ショットから、斜め上の俯瞰ショットへ切り替える。俯瞰ショットでは、主人公の全身、水面を切り裂く軌道、左右へ広がる波紋を明確に見せる。\n\n背景には「SWIM」という単語を正確なスペルの大文字で登場させる。「SWIM」は水中から浮かび上がるように現れ、音楽のビートに合わせて拡大、縮小、前後方向への移動、軽い3D回転を行う。\n\n3D回転した「SWIM」は必ず再び正面を向き、単語全体が明確に読める瞬間を作る。文字は主人公の背後や水中へ配置し、泳ぎを妨げないようにする。\n\n4.0秒の強いビートで、主人公の腕が作った大きな円形波紋を画面いっぱいに拡大する。その円形波紋を、次のシーンの自転車の前輪へグラフィック・マッチさせ、シャープに切り替える。\n\n【4.0～8.0秒：BIKE】\n\n拡大した水面の円形波紋が、黒い競技用ロードバイクの前輪へ瞬時に切り替わる。背景の色は、ブルーとシアンから、鮮やかなマゼンタ、コーラル、オレンジへ大胆に変化する。\n\n主人公は黒一色の競技用ロードバイクに乗り、抽象的で近未来的なロードコースを高速で走行する。上半身を低く保った空気抵抗の少ない姿勢でハンドルを握り、両脚でペダルを滑らかに高速回転させる。\n\nロードバイクは、細いフレーム、ドロップハンドル、サドル、クランク、ペダル、チェーン、同じ大きさの2つの車輪を持つ、正しい競技用自転車の形状として描く。\n\nカメラは真横から高速で並走し、主人公の全身と自転車全体を捉える。続いて、地面すれすれから前輪とペダリングを大きく見せるローアングル、斜め上からコースの曲線を見せる俯瞰ショットへ、音楽のビートに合わせて大胆に切り替える。\n\n車輪が回転すると、マゼンタとオレンジの円形ラインが短い軌跡として残る。ペダルを踏み込むたびに背景のシェイプが後方へ加速し、コースを曲がると長方形、三角形、太いライン、グリッドが同じ方向へ大きく傾く。\n\n背景には「BIKE」という単語を正確なスペルの大文字で登場させる。「BIKE」は道路面から立ち上がるように現れ、車輪の回転と連動して3D回転、拡大、縮小、奥行き方向への移動を行う。\n\n3D回転した「BIKE」は必ず再び正面を向き、単語全体が明確に読める瞬間を作る。文字は主人公や自転車と接触する物体ではなく、背景のグラフィック要素として使用する。\n\n8.0秒の強いビートで、主人公がペダルを力強く踏み下ろす。ペダルを踏み込む足元のアップから、ランで足が地面へ接触する瞬間へ、動きの方向とリズムを維持したマッチカットで切り替える。\n\n【8.0～12.0秒：RUN】\n\n背景は、鮮やかなイエロー、ライムグリーン、バイオレットを組み合わせた、抽象的でスタイリッシュなランニングコースへ切り替わる。\n\n主人公はトライアスロン終盤のランを思わせる、力強く持続的なペースで走る。単なる短距離スプリントではなく、上半身を安定させ、肘を自然に曲げ、腕を前後へ効率的に振りながら、高いピッチで前方へ進む。\n\n接地した足の真上へ身体の重心が自然に移動し、反対側の膝が前方へ上がる。足の接地、蹴り出し、空中局面、次の接地を滑らかにつなげ、左右の脚を正しく交互に動かす。\n\nカメラは真横から主人公と同じ速度で追従し、全身のランニングフォームを見せる。続いて、斜め前方から迫ってくる主人公を捉えるショット、足元すれすれのローアングル、斜め上からコースを見下ろす俯瞰ショットへ切り替える。\n\n足が地面へ接触するたびに円形の波紋が広がり、蹴り出すと鋭いラインが後方へ走る。腕の振りや走行リズムに合わせて、長方形、円、三角形、グリッドが画面の上下左右からスライドインし、次のビートでスライドアウトする。\n\n背景には「RUN」という単語を正確なスペルの大文字で登場させる。「RUN」は主人公の後方から急速に追いつくように画面奥から移動し、ビートに合わせて拡大、縮小、3D回転、軽い伸縮、色の切り替えを行う。\n\n3D回転した「RUN」は必ず再び正面を向き、単語全体が明確に読める瞬間を作る。文字で主人公の身体や足元を覆い隠さない。\n\n12.0秒の強いビートで主人公が力強く地面を踏み込む。接地点から3本の太い斜めラインが広がり、そのラインによって画面全体を3つのパネルへ分割する。\n\n【12.0～15.0秒：3競技同時表示】\n\n画面を、少し傾いたシャープな境界線によって、左、中央、右の3つの縦長パネルへ分割する。\n\n左側のパネルでは、シアンとブルーの水中を、同じ黒い主人公がフリースタイルで泳ぎ続ける。背景には「SWIM」を表示する。\n\n中央のパネルでは、マゼンタ、コーラル、オレンジのロードコースを、同じ黒い主人公がロードバイクで高速走行する。背景には「BIKE」を表示する。\n\n右側のパネルでは、イエロー、ライムグリーン、バイオレットのコースを、同じ黒い主人公が力強く走り続ける。背景には「RUN」を表示する。\n\n3つのパネルは静止画にせず、水泳のストローク、自転車のペダリング、ランの足運びを同時に継続させる。それぞれの動きを同じ音楽のビートへ同期させ、3競技が一体となったトライアスロンのリズムを作る。\n\n分割画面では、それぞれのパネルに主人公を1人だけ表示する。同じ競技の人物を同じパネル内で複製しない。各パネルの人物、用具、背景が、隣のパネルへはみ出したり、混ざったりしないようにする。\n\n終盤に向けて、各パネルの波紋、車輪の円形ライン、ランの接地エフェクトを徐々に高密度にする。最後の強いビートで3つの動きが同時にアクセントを作り、パネルの境界線と背景のシェイプが整列する。\n\n最後の約0.6秒は、「SWIM」「BIKE」「RUN」の3語と、3競技を行う主人公の姿が明確に確認できる安定した3分割画面として保持し、クールに終了する。大げさな決めポーズや喜ぶ動作は入れない。\n\n【モーショングラフィックと音楽】\n\n水泳では、水の飛沫、波紋、波状のラインを、音楽の細かいビートへ同期させる。自転車では、ペダリング、車輪の回転、コースの通過を、ベースと連続するパーカッションへ同期させる。ランでは、左右の足の接地を、キックとスネアへ正確に同期させる。\n\n水泳から自転車、自転車からランへ進むにつれて、音楽、カメラ、シェイプ、タイポグラフィーの動きを徐々に高密度にし、最後の3分割画面で3つのリズムを統合する。\n\n背景のシェイプは人物や競技用具を完全に隠さず、それぞれの動作が常に明確に見えるようにする。制御されたモーションブラーを使用するが、人物の手足や自転車の車輪が複数に見えるほど強くしない。\n\n【文字と一貫性】\n\n映像内に登場する文字は「SWIM」「BIKE」「RUN」の3種類だけに限定する。すべて正確なスペルの大文字で表示し、文字の欠落、重複、誤字、別の文字への変化を起こさない。それ以外の文字、数字、企業名、ロゴ、ウォーターマーク、字幕は表示しない。\n\n通常の競技シーンでは、主人公を一人だけにする。最後の3分割画面に限り、同じ主人公を各パネルへ一人ずつ表示する。\n\n水泳シーンでは、自転車やランニングコースを登場させない。自転車シーンでは、水中表現や余分な自転車を登場させない。ランシーンでは、自転車やフープなどの競技用具を登場させない。\n\n競技用ロードバイクは必ず1台だけにする。車輪を2つに限定し、フレーム、ハンドル、サドル、クランク、チェーン、ペダルの形状と位置を維持する。車輪の複製、変形、消失、瞬間移動を発生させない。\n\n自転車走行中は両手を正しくハンドルへ置き、両足をペダルへ固定する。ペダルと脚を正しく連動させ、足がペダルから外れたり、フレームや車輪を貫通したりしない。\n\n主人公の形状、体格、頭、胴体、手足の本数を15秒間一貫させる。手足の変形、身体の分裂、不自然な残像、競技間での性別や体格の変化を発生させない。\n\n大胆でありながら統一感のある配色、シャープなグラフィック、強い遠近感、正確なビートシンクを組み合わせ、とにかくカッコよく、スタイリッシュで、スピード感とファッション性の高い映像に仕上げる。SNSで拡散されそうな、センスが良く、中毒性があり、何度も見返したくなる高品質なトライアスロン動画にする。",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 20,
      "views": 1524,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094197910678499732",
    "title": "长笛旋律化作金色音符与晨光",
    "titleEn": "A Flute Melody Becomes Golden Notes and Morning Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "动画少女吹奏长笛，金色音符旋转扩散，最终汇入带有晨光色彩的问候画面。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated flutist performs as golden notes spiral outward and resolve into a warm morning greeting. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2094197910678499732",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-08-30T22:57:44.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094197910678499732.mp4",
    "posterUrl": "/posters/x/2094197910678499732.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3はこういう表現も綺麗\n今回はt2v で作成\n@Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 5,
      "likes": 70,
      "views": 2820,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094436234022285632",
    "title": "动漫职员用全息界面启动智能办公室",
    "titleEn": "Anime Coworkers Launch a Smart Office through Holographic Controls",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名动漫职员在夜景办公室操作悬浮蓝色界面，最后收束为完整的科技产品广告画面。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two anime coworkers operate floating blue interfaces in a night office before the sequence resolves into a finished technology ad. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rem_rev/status/2094436234022285632",
    "sourceLabel": "X 原帖 · @rem_rev",
    "author": "リマレレ│AI×YouTube×ブログ",
    "publishedAt": "2026-08-31T14:44:45.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094436234022285632.mp4",
    "posterUrl": "/posters/x/2094436234022285632.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "anime"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3-MAXの動画生成スピードが速すぎる！！\n\n各種AIモデルを提供する fal が開発した、MiniMax H3を進化させたモデルなんだけど、オリジナルの13倍速い＆品質落ちないということで試してみたら、15秒の動画が22秒で完成した。\n\nこんな感じですぐに広告動画ができちゃうなんて、すごすぎるわ…",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 61,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094435486664478793",
    "title": "红裙舞者旋转切出彩色动效卡片",
    "titleEn": "A Red-Dressed Dancer Spins into Colorful Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红裙舞者在庭院中起舞并完成高速旋转，画面叠入夸张文字和鲜艳图形元素。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dancer in red performs and spins through a courtyard while oversized lettering and vivid graphic elements enter the frame. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/apilpirman/status/2094435486664478793",
    "sourceLabel": "X 原帖 · @apilpirman",
    "author": "Apil",
    "publishedAt": "2026-08-31T14:41:47.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094435486664478793.mp4",
    "posterUrl": "/posters/x/2094435486664478793.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 can’t really be compared to Seedance. It has its own strengths, especially in graphic editing and text, whether 2D or 3D.\n\n@Hailuo_AI\n\nDownload minimax design app on PC, free 3k credits for new user",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 147,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094433656278597726",
    "title": "夏日相遇与现场演唱交织成音乐短片",
    "titleEn": "Summer Encounters Intercut with a Live Vocal Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "海边与城市中的夏日人物片段和女歌手演唱交替出现，串起一段关于陪伴的音乐叙事。 原生视频规格为 127 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Summer moments by the sea and in the city intercut with a female singer, forming a music narrative about companionship. Native-video output: 127s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/junyait/status/2094433656278597726",
    "sourceLabel": "X 原帖 · @junyait",
    "author": "海澤純 AI CREATIVE",
    "publishedAt": "2026-08-31T14:34:31.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094433656278597726.mp4",
    "posterUrl": "/posters/x/2094433656278597726.jpg",
    "duration": 127,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【TapNow作品】Music Video「World」Song by C feat.有紗\n\n「何者でもない僕に、君が世界をくれた」\n\n壮大な世界を描いた曲じゃない。\n\n駅まで歩くこと。\n海へ向かうこと。\n夏祭りで笑うこと。\n花火を見ること。\n\n好きな人と過ごした何気ない夏の日が、\nいつの間にか「世界」になっていた。\n\nCちゃんが歌い、\n有紗がその物語を生きる。\n\n「あの日、君といた世界」。\nぜひ最後まで観てください。\n\nTapNow( @TapNow_AI )で制作しました。\n\n@TapNow_JP \n#TapNow\n#MiniMax\n#H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 58,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094431473499783415",
    "title": "古装女子用手机记录初到上海的一天",
    "titleEn": "An Ancient-Dressed Traveler Vlogs Her First Day in Shanghai",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "古装女子手持手机游览外滩、武康路与城市街区，以自然中文记录现代上海体验。 原生视频规格为 15 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman in historical dress films herself around the Bund, Wukang Road, and city streets, narrating her first Shanghai visit in natural Mandarin. Native-video output: 15s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "一位20出头的东方古代女性穿越到现代上海真实的个人度假视频日志，她第一次来到上海。生活自然、随性而发，偶尔还有些小瑕疵——绝对不是那种商业化的旅游宣传视频。角色：始终都是同一个东方古代女性。\n\n鹅蛋脸脸型，气质凌冽，脸型流畅，长发自然披散，一侧斜插玉簪，清冷出尘，气质淡雅，丹凤眼，鼻子秀挺，嘴唇饱满柔美但边界清晰；妆容以淡雅为主；无耳饰；\n\n衣服为淡绿色为主，内衬为白色，服饰为传统交领襦裙，交领右衽，上裾下裙，腰间系带，线条简洁，配色清冷，无花样样式。她使用自然的中文进行交谈，没有字幕、标注、翻译或屏幕上的文字提示。\n\n 摄像机：手持智能手机，屏幕格式为16:9的横向模式。自然的手臂摆动、行走时的抖动、不完美的构图、自动对焦的变动、曝光变化、真实的运动模糊效果以及手机与摄像机的结合效果。\n\n没有无人机、云台设备，也没有任何电影级的调色或美化滤镜。 \n\n0–4秒—外滩：她以自拍的方式行走在外滩，东方明珠就在她的身后。她将相机对准纪念碑东方明珠，然后又转向自己，说道：“跟古代真是一点都不一样呢。” 4–7秒—和平饭店：她指着和平饭店的门匾说道：“在这里面吃一顿饭肯定要花很多银子。”\n\n 7–10秒——武康路：她走在附近，武康路的标志性建筑物武康大楼在她身后若隐若现。她说：“看到的时候感觉更美了。” \n\n10–14秒—上海博物馆：在上海博物馆内，她参观着各种展品，并说道：“这里真的有太多值得欣赏的东西了。”\n\n 14–15秒—告别：她拍了一张上海外滩黄浦江的蓝调时刻的自拍照，照片中东方明珠显得格外明亮，她微笑着说：“真是美好的一天啊。我肯定想再回来一次。”她笑着用手掌遮住了镜头。\n\n 音频：真实的上海的交通声、公交车声、行人声、鸟叫声、脚步声、博物馆的氛围音效、食物的声音以及风声。同时还有轻快的旅行音乐作为背景音乐。\n\n对话部分依然清晰可辨。 真实感锁定：真实的上海地标建筑、逼真的人群模拟、食物效果、建筑细节、灯光效果、阴影表现、反射效果、角色的皮肤、头发以及织物物理效果。角色的自然表情和不完美的摄像机行为也得到了完美呈现。 \n\n负面特征：没有面部变形、重复的人物形象、扭曲的手部动作、漂浮的物体、虚假的人群、塑料般的皮肤效果、过度美化的妆容、夸张的反应动作、不真实的建筑外观、瞬移效果、计算机生成的图像、字幕、文字说明、标志、水印，以及商业化的旅游风格。",
    "promptSourceUrl": "https://x.com/nopinduoduo/status/2094431473499783415",
    "sourceUrl": "https://x.com/nopinduoduo/status/2094431473499783415",
    "sourceLabel": "X 原帖 · @nopinduoduo",
    "author": "我真的没有拼多多",
    "publishedAt": "2026-08-31T14:25:50.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094431473499783415.mp4",
    "posterUrl": "/posters/x/2094431473499783415.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "重生之 我穿越到了2026年的上海，逛了一天魔都的消费比我们古代可贵太多了！\n\n由MiniMaxH3 @Hailuo_AI 生成；\n\n #MiniMaxH3  \n\n提示词：\n一位20出头的东方古代女性穿越到现代上海真实的个人度假视频日志，她第一次来到上海。生活自然、随性而发，偶尔还有些小瑕疵——绝对不是那种商业化的旅游宣传视频。角色：始终都是同一个东方古代女性。\n\n鹅蛋脸脸型，气质凌冽，脸型流畅，长发自然披散，一侧斜插玉簪，清冷出尘，气质淡雅，丹凤眼，鼻子秀挺，嘴唇饱满柔美但边界清晰；妆容以淡雅为主；无耳饰；\n\n衣服为淡绿色为主，内衬为白色，服饰为传统交领襦裙，交领右衽，上裾下裙，腰间系带，线条简洁，配色清冷，无花样样式。她使用自然的中文进行交谈，没有字幕、标注、翻译或屏幕上的文字提示。\n\n 摄像机：手持智能手机，屏幕格式为16:9的横向模式。自然的手臂摆动、行走时的抖动、不完美的构图、自动对焦的变动、曝光变化、真实的运动模糊效果以及手机与摄像机的结合效果。\n\n没有无人机、云台设备，也没有任何电影级的调色或美化滤镜。 \n\n0–4秒—外滩：她以自拍的方式行走在外滩，东方明珠就在她的身后。她将相机对准纪念碑东方明珠，然后又转向自己，说道：“跟古代真是一点都不一样呢。” 4–7秒—和平饭店：她指着和平饭店的门匾说道：“在这里面吃一顿饭肯定要花很多银子。”\n\n 7–10秒——武康路：她走在附近，武康路的标志性建筑物武康大楼在她身后若隐若现。她说：“看到的时候感觉更美了。” \n\n10–14秒—上海博物馆：在上海博物馆内，她参观着各种展品，并说道：“这里真的有太多值得欣赏的东西了。”\n\n 14–15秒—告别：她拍了一张上海外滩黄浦江的蓝调时刻的自拍照，照片中东方明珠显得格外明亮，她微笑着说：“真是美好的一天啊。我肯定想再回来一次。”她笑着用手掌遮住了镜头。\n\n 音频：真实的上海的交通声、公交车声、行人声、鸟叫声、脚步声、博物馆的氛围音效、食物的声音以及风声。同时还有轻快的旅行音乐作为背景音乐。\n\n对话部分依然清晰可辨。 真实感锁定：真实的上海地标建筑、逼真的人群模拟、食物效果、建筑细节、灯光效果、阴影表现、反射效果、角色的皮肤、头发以及织物物理效果。角色的自然表情和不完美的摄像机行为也得到了完美呈现。 \n\n负面特征：没有面部变形、重复的人物形象、扭曲的手部动作、漂浮的物体、虚假的人群、塑料般的皮肤效果、过度美化的妆容、夸张的反应动作、不真实的建筑外观、瞬移效果、计算机生成的图像、字幕、文字说明、标志、水印，以及商业化的旅游风格。",
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 9,
      "views": 398,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094429774571782201",
    "title": "金发旅客在站台迎来诡异列车镜头",
    "titleEn": "A Blonde Traveler Meets a Strange Train at the Platform",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发动画少女牵着兔子站在乡间月台，列车进站后画面转入带窥视感的圆形视角。 原生视频规格为 10 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde anime traveler waits with a rabbit on a rural platform before an arriving train shifts the scene into an unsettling circular viewpoint. Native-video output: 10s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ZenzaiOshilco/status/2094429774571782201",
    "sourceLabel": "X 原帖 · @ZenzaiOshilco",
    "author": "ぜんざい",
    "publishedAt": "2026-08-31T14:19:05.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094429774571782201.mp4",
    "posterUrl": "/posters/x/2094429774571782201.jpg",
    "duration": 10,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おやすMinimaxH3\n煽り車両",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 17,
      "views": 151,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094429444249194534",
    "title": "石神千空在丛林中举起蓝色药剂",
    "titleEn": "Senku Raises a Blue Formula in the Jungle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白发动漫角色在热带丛林中调制并举起蓝色烧瓶，镜头从半身推进到自信特写。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired anime scientist mixes and raises a blue flask in a tropical jungle as the camera pushes into a confident close-up. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/27upon2/status/2094429444249194534",
    "sourceLabel": "X 原帖 · @27upon2",
    "author": "Sriraam",
    "publishedAt": "2026-08-31T14:17:46.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094429444249194534.mp4",
    "posterUrl": "/posters/x/2094429444249194534.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3 just knows Senku crazy",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 50,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094429002404646966",
    "title": "白虎少女从角色设定走进竹林",
    "titleEn": "A White-Tiger Heroine Steps from Character Sheet into a Bamboo Forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "角色设定图中的白虎少女保持服装和体态，在片段末尾进入幽深竹林环境。 原生视频规格为 21 秒 · 1280×590。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-tiger heroine carries her costume and proportions from a character sheet into a deep bamboo-forest setting. Native-video output: 21s · 1280×590. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DordrayO/status/2094429002404646966",
    "sourceLabel": "X 原帖 · @DordrayO",
    "author": "Dordray Japanees  Oganizasion",
    "publishedAt": "2026-08-31T14:16:01.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094429002404646966.mp4",
    "posterUrl": "/posters/x/2094429002404646966.jpg",
    "duration": 21,
    "aspectRatio": "landscape",
    "resolution": "1280×590",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "fashion"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max\n10秒\n日本語セリフ良いね\n勝手に東洋風BGMで良いね\nスピーディーだけど\nモーション手抜感否めない\n\nこのキャラはすご〜くすご〜く\n昔に描いた同人誌のオリジナルキャラクター風水四神娘のうちの白虎娘",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 177,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094425579290538004",
    "title": "空置水泥房一镜变成温暖客厅",
    "titleEn": "An Empty Concrete Room Transforms into a Warm Living Space",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "空置水泥房在连续镜头中逐步出现墙面、沙发、灯具与绿植，完成室内改造。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An empty concrete room gains finished walls, a sofa, lighting, and plants in a continuous interior transformation. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ferica7/status/2094425579290538004",
    "sourceLabel": "X 原帖 · @Ferica7",
    "author": "Ferica7",
    "publishedAt": "2026-08-31T14:02:25.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094425579290538004.mp4",
    "posterUrl": "/posters/x/2094425579290538004.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cool :)  I Used MiniMax H3 Max from fal.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 46,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094424981773893895",
    "title": "达·芬奇手稿翻页唤醒机械与人体结构",
    "titleEn": "Da Vinci Notebook Pages Animate Machines and Anatomy",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "泛黄手稿在翻页与旋转中显露机械草图、线框人体和心脏结构，形成文艺复兴发明档案。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Aged notebook pages turn and spiral through machine sketches, wireframe anatomy, and a heart in a Renaissance invention archive. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/vkuoo/status/2094424981773893895",
    "sourceLabel": "X 原帖 · @vkuoo",
    "author": "vkuoo",
    "publishedAt": "2026-08-31T14:00:02.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094424981773893895.mp4",
    "posterUrl": "/posters/x/2094424981773893895.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Meet Leonardo da Vinci ✨\nGenerated on @krea_ai  with Minimax H3 Max",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 53,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094441764296020070",
    "title": "两名少年在海边并肩交谈",
    "titleEn": "Two Young Men Talk Side by Side at the Seaside",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名少年沿海边步道并肩前行，在明亮海景中保持一致的真人角色外观与自然交流。 原生视频规格为 8 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two young men walk together along a seaside path, maintaining consistent live-action identities and a natural exchange against the bright coast. Native-video output: 8s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yuzukisan_voice/status/2094441764296020070",
    "sourceLabel": "X 原帖 · @yuzukisan_voice",
    "author": "朝賀悠月｜声優・ナレーター・マルチクリエイター",
    "publishedAt": "2026-08-31T15:06:44.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094441764296020070.mp4",
    "posterUrl": "/posters/x/2094441764296020070.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "いくつか試してみて、『AUBE』を存在させるためのモデルが見つかった！出会えた✨\nこの収穫はデカい！！\nMiniMax H3だと違和感なくて、ユウもHAJUNも本当に生きて存在してるみたい🥹\nわたしが求めていた、やりたいことをカタチにしていける未来がまた少し見えた気がする。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 35,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094438296751685761",
    "title": "彩色墨点迸发成初音主题片头",
    "titleEn": "Colorful Ink Bursts into a Miku-Themed Opener",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色背景上的荧光墨点爆开，绿色双马尾角色与粗体标题连续切换成高能片头。 原生视频规格为 10 秒 · 1080×1890。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Neon ink splashes across black as a green twin-tailed character and bold lettering cut together into an energetic opener. Native-video output: 10s · 1080×1890. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Pastel_Illust_/status/2094438296751685761",
    "sourceLabel": "X 原帖 · @Pastel_Illust_",
    "author": "闇野ぱすてる",
    "publishedAt": "2026-08-31T14:52:57.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094438296751685761.mp4",
    "posterUrl": "/posters/x/2094438296751685761.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1080×1890",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "私は何を作ってるんだ…\n\nMiniMax H3の無駄使い",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 16,
      "views": 920,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094438197896102218",
    "title": "兔耳朋克少女让东京巷道随节拍闪烁",
    "titleEn": "A Punk Rabbit Girl Makes a Tokyo Alley Pulse to the Beat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "兔耳朋克少女在霓虹巷道中舞动，招牌、积水反光与城市灯光跟随节拍同步变化。 原生视频规格为 64 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A punk rabbit-eared heroine dances in a neon alley as signs, wet-street reflections, and city lights pulse with the beat. Native-video output: 64s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/quu1945/status/2094438197896102218",
    "sourceLabel": "X 原帖 · @quu1945",
    "author": "ドラゴン串",
    "publishedAt": "2026-08-31T14:52:33.000Z",
    "addedAt": "2026-08-31T15:24:28.120Z",
    "mediaUrl": "/media/x-2094438197896102218.mp4",
    "posterUrl": "/posters/x/2094438197896102218.jpg",
    "duration": 64,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ComfyUIイベントに動画提出間になんとか合った。。\n他の人の動画見たら、すごかった。落ちても仕方ないけど、いい勉強になりました。\n筋トレは終わったし、ちょっとドラクエウォークしてくるわ。。明日から出社だし、動画作る時間少なくなるな。。(´;ω;｀)\n\nAIの動画説明：\nTOKYO LOUD（64秒）\n\n夜の東京の裏路地。ウサギの耳と尻尾を持つパンクの女の子が、ネオンの下で踊る。\n\n芯にあるのは「街全体がドラムマシンになる」という一点です。\n\n路地にはパンクのビートが鳴り続けていて、街のすべてがその拍に従います。 看板が拍で明滅し、濡れたアスファルトに映る光も一緒に明滅する。彼女のブーツが水たまりを叩く音も拍に乗る。5番目のショットで全部の光と音がいっせいに消えて無音になり、次のショットで戻ってくる。\n\n音は一切足していません。 音楽トラックを乗せたのではなく、ビートも水しぶきもネオンの唸りも、H3 が絵と一緒に生成したものです。8秒×8本、それぞれの音はそれぞれの絵と結びついたまま、順番に繋いだだけです。\n\n最後はネオンの看板が「MiniMax H3」と光って、暗転して終わります。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 34,
      "snapshotAt": "2026-08-31T15:24:28.120Z"
    },
    "approvedAt": "2026-08-31T15:24:28.120Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094432834266206326",
    "title": "遥控器密码引爆情侣间的尴尬审问",
    "titleEn": "A Remote-Control Password Triggers an Awkward Couples Interrogation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "男子拿着遥控器试图输入密码，身旁女子的追问把客厅场景推向生活化喜剧。 原生视频规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man attempts to enter a password on a remote while the woman beside him turns the living-room moment into an awkward comedy. Native-video output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DoctorAmna11/status/2094432834266206326",
    "sourceLabel": "X 原帖 · @DoctorAmna11",
    "author": "Ai Doctor",
    "publishedAt": "2026-08-31T14:31:15.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094432834266206326.mp4",
    "posterUrl": "/posters/x/2094432834266206326.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "POV: You forgot the password… and accidentally exposed yourself. \n\nBro thought he was trying to log into Netflix, not survive a relationship interrogation. \n\nCreated with MiniMax H3",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 12,
      "views": 211,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094432525250597104",
    "title": "赛博朋克女主穿过故障风格预告片",
    "titleEn": "A Cyberpunk Heroine Cuts through a Glitch-Driven Trailer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "赛博朋克女主在霓虹、故障切片与巨大标题之间高速切换，形成紧凑的角色预告片。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cyberpunk heroine moves rapidly through neon, glitch fragments, and oversized type in a compact character trailer. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/vadooai/status/2094432525250597104",
    "sourceLabel": "X 原帖 · @vadooai",
    "author": "Vadoo AI",
    "publishedAt": "2026-08-31T14:30:01.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094432525250597104.mp4",
    "posterUrl": "/posters/x/2094432525250597104.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "anime"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This isn’t just an anime trailer. It’s a visual speedrun. 🏁\n\nOne character. Pure cyberpunk energy.\n\nGenerated with MiniMax H3 and brought together with Vadoo AI.\n\nAI filmmaking is getting seriously fast. 🔥",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 7,
      "views": 179,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094431901897367562",
    "title": "少女骑着巨蜥飞奔却被小猫追上",
    "titleEn": "A Girl Races on a Giant Lizard while a Cat Keeps Up",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少女骑着巨蜥高速穿行，强烈的追逐运动与轻松的小猫反差制造幽默效果。 原生视频规格为 10 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl speeds forward on a giant lizard, while a small cat keeping pace creates a playful contrast to the intense chase. Native-video output: 10s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/miokonboDaysAI/status/2094431901897367562",
    "sourceLabel": "X 原帖 · @miokonboDaysAI",
    "author": "みおこんぼ",
    "publishedAt": "2026-08-31T14:27:32.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094431901897367562.mp4",
    "posterUrl": "/posters/x/2094431901897367562.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今夜はオオトカゲに乗って…いや、だから速いっ…！速いよ！！想像の2倍は速いのMiniMaxH3は。\n迫力があるよ！！こんなの車より速……あ、猫ちゃんが追いつけるならそうでもないのか。\nそれじゃおやとぅみ。GN！",
    "engagement": {
      "replies": 2,
      "reposts": 5,
      "likes": 30,
      "views": 138,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094430361355001956",
    "title": "猫耳玩家在格斗游戏中打出终结一击",
    "titleEn": "A Cat-Eared Gamer Lands a Fighting-Game Knockout",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "猫耳少女坐在红色椅子上操作格斗游戏，角色倒地后她举手庆祝胜利。 原生视频规格为 10 秒 · 768×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared girl plays a fighting game from a red chair and raises her hand in celebration after the opponent falls. Native-video output: 10s · 768×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2094430361355001956",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-08-31T14:21:25.000Z",
    "addedAt": "2026-08-31T15:24:28.121Z",
    "mediaUrl": "/media/x-2094430361355001956.mp4",
    "posterUrl": "/posters/x/2094430361355001956.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1088",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ゲージ削り切れてないけど KO.!\n\nMiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 5,
      "views": 59,
      "snapshotAt": "2026-08-31T15:24:28.121Z"
    },
    "approvedAt": "2026-08-31T15:24:28.121Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094570516665495746",
    "title": "极简字体与图形拼成虚构宣传片",
    "titleEn": "Minimal Type and Shapes Form a Fictional Promo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深蓝、米白与橙色图形在画面中折叠切换，并以日文标语和小角色组成极简宣传片。 原生视频规格为 14 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Navy, cream, and orange forms fold between Japanese slogans and a tiny character in a minimal fictional promo. Native-video output: 14s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2094570516665495746",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-08-31T23:38:21.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094570516665495746.mp4",
    "posterUrl": "/posters/x/2094570516665495746.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI × モーショングラフィックス \n\n架空のモーショングラフィックス広告\n\nMiniMax H3 Max  #MiniMaxdesign で制作",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 304,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094453168025530598",
    "title": "月份字样在极简版式中折叠翻转",
    "titleEn": "Month Names Fold and Flip through Minimal Layouts",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "AUGUST 与 SEPTEMBER 字样在蓝白平面构图中依次翻转、拉伸并重组。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "AUGUST and SEPTEMBER flip, stretch, and reassemble across restrained blue-and-white graphic layouts. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2094453168025530598",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-08-31T15:52:02.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094453168025530598.mp4",
    "posterUrl": "/posters/x/2094453168025530598.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIモーグラ\n09_ミニマル_キネティックタイポグラフィ\n\nAUGUST / SEPTEMBER 🇯🇵\n\nMiniMax H3 Max on #MiniMaxDesign",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 12,
      "views": 1645,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094560870907142248",
    "title": "眼镜少女在教室里连贯起舞",
    "titleEn": "A Bespectacled Schoolgirl Dances through a Classroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴眼镜的校服少女在水彩质感教室中完成抬手、转身与连续舞步。 原生视频规格为 15 秒 · 704×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bespectacled schoolgirl performs raised-arm turns and linked dance steps inside a watercolor-textured classroom. Native-video output: 15s · 704×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2094560870907142248",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-31T23:00:01.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094560870907142248.mp4",
    "posterUrl": "/posters/x/2094560870907142248.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "704×928",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "dance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ダンス #AI眼鏡部 MiniMax H3 ローカル",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 24,
      "views": 651,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094530669141451192",
    "title": "水彩校服少女完成轻快舞步",
    "titleEn": "A Watercolor Schoolgirl Performs a Playful Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水彩风校服少女从近景动作过渡到全身舞姿，在教室里保持角色外观一致。 原生视频规格为 15 秒 · 704×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A watercolor-style schoolgirl moves from close-up gestures into a full-body routine while keeping a consistent appearance. Native-video output: 15s · 704×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2094530669141451192",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-08-31T21:00:00.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094530669141451192.mp4",
    "posterUrl": "/posters/x/2094530669141451192.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "704×928",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "dance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはよ！9月！ #AI眼鏡部 MiniMax H3ローカル",
    "engagement": {
      "replies": 3,
      "reposts": 3,
      "likes": 59,
      "views": 1126,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094423490845921745",
    "title": "纹身女子离开黄色跑车走向街道",
    "titleEn": "A Tattooed Woman Steps Away from a Yellow Sports Car",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "纹身女子从黄色跑车前起身，穿过阳光街道并在镜头远离时回望。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A tattooed woman rises from a yellow sports car, walks down a sunny street, and glances back as the camera drifts away. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, modern western attitude, small-town cowboy entrance energy, not a car wash video and not a glamour ad. The girl fully matches <Picture 1>: same face, short hair, wet skin, black semi-sheer top, denim shorts, tattoos, sneakers, yellow vintage car, hose, summer street, storefronts, warm streetlights, blue sky, and clouds. The first frame begins exactly from the pose in <Picture 1>, with her seated on the yellow hood, one hand supporting herself, the other holding the hose, and the water already swinging to one side, with no pose reset. IMAX 70mm film grain, Nolan-inspired cinematic realism, photorealistic natural palette with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. Low-angle medium shot with a slow slight push-in and faint living sway. From the first second she is already moving: casually swinging the water stream, shifting her shoulders with relaxed swagger, smiling like she already owns the whole street. All water already on her legs must keep moving naturally at all times, flowing, splitting, merging, and dripping with changing paths as she moves. Her chest and upper torso must show subtle real secondary motion from breath, small body sway, landing, and walking, visible but restrained. The nearby off-screen man (S1), mature, relaxed, amused, says in an off-screen voiceover: <d>[English] You making trouble again?</d> while remaining unseen. She does not stop moving, only smiles more and lifts her eyes near the lens. [Shot 2] At 00:02.600, the camera cuts tighter and sinks with her movement. She shuts off the water, flicks the hose aside like tossing a rein, and slides off the hood cleanly, right leg first, landing light and confident like a cowboy dismount. The girl (S2), young but mature, open, sharp, says: <d>[English] Only when the town’s too quiet.</d> [Shot 3] At 00:04.800, the camera tracks backward as she walks toward it, the yellow car falling behind her like a mount she has left behind. She casually wipes a wet hand or brushes her shorts, shoulders loose, stride steady, while the off-screen man (S1) says in an off-screen voiceover: <d>[English] You always come in this loud?</d> She keeps a small smile and brings easy pressure as she closes the distance. [Shot 4] At 00:08.600, the camera tightens to a medium close-up and nearly stops as she reaches a dangerous conversational distance. She leans in slightly, one hand near her waist, and says with playful threat and total control: <d>[English] You gonna stare... or buy me a drink?</d> [Shot 5] At 00:11.800, the camera widens slightly and drifts as she laughs softly, turns away, walks down the street, gives one brief side glance back, and keeps going. She is the only visible person on screen. No car wash commercial energy, no influencer posing, no slow motion, no frozen water, no plastic body motion.\n\noverall_soundscape: Water spray, hose movement, soft summer street ambience, distant storefront sound, fabric friction, sneaker steps, small metal contact, and natural breathing.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2094423490845921745",
    "sourceUrl": "https://x.com/ou_zhen599/status/2094423490845921745",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-08-31T13:54:07.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094423490845921745.mp4",
    "posterUrl": "/posters/x/2094423490845921745.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3  \n\n🍻“Want to Have a Drink? I Say, Why Not?”..  \n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, modern western attitude, small-town cowboy entrance energy, not a car wash video and not a glamour ad. The girl fully matches <Picture 1>: same face, short hair, wet skin, black semi-sheer top, denim shorts, tattoos, sneakers, yellow vintage car, hose, summer street, storefronts, warm streetlights, blue sky, and clouds. The first frame begins exactly from the pose in <Picture 1>, with her seated on the yellow hood, one hand supporting herself, the other holding the hose, and the water already swinging to one side, with no pose reset. IMAX 70mm film grain, Nolan-inspired cinematic realism, photorealistic natural palette with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. Low-angle medium shot with a slow slight push-in and faint living sway. From the first second she is already moving: casually swinging the water stream, shifting her shoulders with relaxed swagger, smiling like she already owns the whole street. All water already on her legs must keep moving naturally at all times, flowing, splitting, merging, and dripping with changing paths as she moves. Her chest and upper torso must show subtle real secondary motion from breath, small body sway, landing, and walking, visible but restrained. The nearby off-screen man (S1), mature, relaxed, amused, says in an off-screen voiceover: <d>[English] You making trouble again?</d> while remaining unseen. She does not stop moving, only smiles more and lifts her eyes near the lens. [Shot 2] At 00:02.600, the camera cuts tighter and sinks with her movement. She shuts off the water, flicks the hose aside like tossing a rein, and slides off the hood cleanly, right leg first, landing light and confident like a cowboy dismount. The girl (S2), young but mature, open, sharp, says: <d>[English] Only when the town’s too quiet.</d> [Shot 3] At 00:04.800, the camera tracks backward as she walks toward it, the yellow car falling behind her like a mount she has left behind. She casually wipes a wet hand or brushes her shorts, shoulders loose, stride steady, while the off-screen man (S1) says in an off-screen voiceover: <d>[English] You always come in this loud?</d> She keeps a small smile and brings easy pressure as she closes the distance. [Shot 4] At 00:08.600, the camera tightens to a medium close-up and nearly stops as she reaches a dangerous conversational distance. She leans in slightly, one hand near her waist, and says with playful threat and total control: <d>[English] You gonna stare... or buy me a drink?</d> [Shot 5] At 00:11.800, the camera widens slightly and drifts as she laughs softly, turns away, walks down the street, gives one brief side glance back, and keeps going. She is the only visible person on screen. No car wash commercial energy, no influencer posing, no slow motion, no frozen water, no plastic body motion.\n\noverall_soundscape: Water spray, hose movement, soft summer street ambience, distant storefront sound, fabric friction, sneaker steps, small metal contact, and natural breathing.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 15,
      "views": 832,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094322714739871760",
    "title": "战后女武者从低头沉思到重新站稳",
    "titleEn": "A Battle-Worn Fighter Regains Her Composure",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "受伤女武者在烟雾弥漫的竞技场中低头整理情绪，随后抬眼恢复镇定。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A battle-worn fighter gathers herself in a smoky arena, then raises her gaze and regains her composure. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, post-fight intimacy. Chun-Li fully matches <Picture 1>: same face, hair buns, white ribbons, battle-worn blue qipao-style outfit, wrist guards, white boots, seated pose, smoky arena environment, clothing damage, body proportions, and controlled intensity. The first frame must begin exactly from the seated posture in <Picture 1>, with no pose reset. IMAX 70mm film grain, Nolan-inspired cinematic realism, photorealistic natural palette with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. Low-angle medium close-up, very slow physical push-in, no cut during the first emotional phase. Chun-Li sits wide and grounded with both hands on her waist. The off-screen interviewer, a mature calm warm woman (S1), says in an off-screen voiceover: <d>[English] Do you still miss your father?</d> Chun-Li does not answer immediately. Her breath catches, one hand slowly drops from her waist to her thigh or the floor, one knee subtly draws inward, her chin dips, her gaze lowers, and her jaw tightens before she finally answers. Chun-Li (S2), low, quiet, controlled, says: <d>[English] Every day.</d> [Shot 2] At 00:07.000, the shot cuts softly to a tighter medium close-up. Chun-Li steadies herself with one hand on her thigh or forearm, leans forward slightly, and says: <d>[English] I still hear him when I fight.</d> [Shot 3] At 00:11.000, the shot cuts softly to a slightly wider medium close-up. Chun-Li slowly straightens, brings one hand higher again, resets her shoulders, firms her jaw, lifts her eyes with quiet resolve, and says: <d>[English] I carry him with me.</d> Only Chun-Li is visible. The interviewer never appears. No melodrama, no exaggerated crying, no glamour posing. End on regained stillness, not collapse.\n\noverall_soundscape: Faint smoke movement, subtle room tone, soft distant arena hush, light fabric rustle, subtle boot shift, and quiet breath.\n\nnon_diegetic_music: Very restrained low strings and sparse piano at a slow tempo, mixed gently under the natural ambience.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2094322714739871760",
    "sourceUrl": "https://x.com/ou_zhen599/status/2094322714739871760",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-08-31T07:13:40.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094322714739871760.mp4",
    "posterUrl": "/posters/x/2094322714739871760.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "dialogue",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3    \n\n🥋“Chun-Li Show, Scene 3-A dramatic scene featuring Chun-Li”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, post-fight intimacy. Chun-Li fully matches <Picture 1>: same face, hair buns, white ribbons, battle-worn blue qipao-style outfit, wrist guards, white boots, seated pose, smoky arena environment, clothing damage, body proportions, and controlled intensity. The first frame must begin exactly from the seated posture in <Picture 1>, with no pose reset. IMAX 70mm film grain, Nolan-inspired cinematic realism, photorealistic natural palette with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. Low-angle medium close-up, very slow physical push-in, no cut during the first emotional phase. Chun-Li sits wide and grounded with both hands on her waist. The off-screen interviewer, a mature calm warm woman (S1), says in an off-screen voiceover: <d>[English] Do you still miss your father?</d> Chun-Li does not answer immediately. Her breath catches, one hand slowly drops from her waist to her thigh or the floor, one knee subtly draws inward, her chin dips, her gaze lowers, and her jaw tightens before she finally answers. Chun-Li (S2), low, quiet, controlled, says: <d>[English] Every day.</d> [Shot 2] At 00:07.000, the shot cuts softly to a tighter medium close-up. Chun-Li steadies herself with one hand on her thigh or forearm, leans forward slightly, and says: <d>[English] I still hear him when I fight.</d> [Shot 3] At 00:11.000, the shot cuts softly to a slightly wider medium close-up. Chun-Li slowly straightens, brings one hand higher again, resets her shoulders, firms her jaw, lifts her eyes with quiet resolve, and says: <d>[English] I carry him with me.</d> Only Chun-Li is visible. The interviewer never appears. No melodrama, no exaggerated crying, no glamour posing. End on regained stillness, not collapse.\n\noverall_soundscape: Faint smoke movement, subtle room tone, soft distant arena hush, light fabric rustle, subtle boot shift, and quiet breath.\n\nnon_diegetic_music: Very restrained low strings and sparse piano at a slow tempo, mixed gently under the natural ambience.",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 60,
      "views": 3750,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094576083198951676",
    "title": "空白画布逐步绘成橙发角色设定",
    "titleEn": "A Blank Canvas Evolves into an Orange-Haired Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "绘图软件中的空白画布依次出现草稿、线稿与上色，最终完成橙发角色设定。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A blank drawing canvas advances through rough sketch, line art, and color before revealing a finished orange-haired character. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second fast-paced character creation timelapse inside a professional digital drawing and character design interface.\n\nUse the provided character reference only as the final target for the character’s appearance, proportions, clothing, hairstyle, colors and distinctive features. The process must visibly start from a completely blank canvas and reconstruct the referenced character from scratch.\n\nShow a visible mouse cursor actively controlling the entire creation process. Every major visual change should feel caused by cursor interaction with the UI.\n\n0:00-0:02\nStart on a blank white or neutral canvas inside a clean professional drawing interface. The cursor quickly creates the first loose construction lines, head shape, body proportions and basic pose using rough sketch strokes.\n\n0:02-0:05\nAccelerate into detailed sketching. The cursor rapidly defines the face, hair, anatomy, clothing silhouette, accessories and characteristic features of the referenced character. Use believable drawing gestures, short strokes, selections, transforms and occasional undo corrections.\n\n0:05-0:08\nThe rough sketch becomes clean line art. The cursor switches tools, adjusts brush size and cleans the silhouette. Unnecessary construction lines disappear while precise facial details, clothing folds and small design elements are added.\n\n0:08-0:11\nRapid coloring phase. The cursor selects regions and fills the character with the correct colors from the reference. Show quick palette selections, color picking, layer changes, clipping masks and brush passes. Add shadows, highlights and material details in fast succession.\n\n0:11-0:13\nRendering and polish. The cursor adds final lighting, hair detail, skin shading, fabric texture, reflections and small accent details. Briefly zoom into important areas such as the face or costume while refining them, then zoom back out.\n\n0:13-0:15\nThe interface clears unnecessary guides and panels. The finished character is shown fully rendered and closely matching the provided reference. End with the cursor making one final small adjustment, then stop on a clean full-character presentation.\n\nSTYLE:\nHigh-speed digital art timelapse, polished professional character-design workflow, believable drawing software UI, smooth cursor movement, rapid but readable actions, clean layers and tool changes, no random text, no fake dialogue, no unrelated popups.\n\nMOTION:\nThe entire 15 seconds should feel continuously productive. Avoid long pauses. Compress several minutes or hours of drawing into an energetic timelapse while keeping the progression clearly readable: blank canvas → rough sketch → clean line art → flat colors → shading → polished final character.\n\nCAMERA:\nPrimarily screen-recording style with subtle digital zooms and pans following the cursor and active drawing area. Keep the UI stable and readable. Do not use cinematic cuts outside the software interface.\n\nIMPORTANT:\nThe character must not appear instantly or through a single magical transformation. Every stage should visibly build upon the previous one through cursor-driven drawing, editing, coloring and rendering actions.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2094576338187543010",
    "sourceUrl": "https://x.com/aimikoda/status/2094576083198951676",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-09-01T00:00:28.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094576083198951676.mp4",
    "posterUrl": "/posters/x/2094576083198951676.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🔥 If I tell you this is how I create my characters, don't believe me.\n\nWith this prompt, you can turn your character reference into a 15 second creation process video. From rough sketch to final render, the whole build comes together on screen inside a drawing UI.\n\nMiniMax H3 - You can find the prompt below.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 197,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094575793615814722",
    "title": "水墨武士与女战士在白色空间决斗",
    "titleEn": "An Ink Samurai and Valkyrie Duel in a White Void",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "两名黑色水墨战士从烟雾中成形，在纯白空间连续交锋并以撞击定格收尾。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two black-ink warriors form from smoke, trade strikes across a white void, and finish in a locked impact. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "{\n\"archetype\": \"Artistic / Fashion / Conceptual\",\n\"duration\": \"15s\",\n\"aspect_ratio\": \"16:9\",\n\"prompt\": {\n\"concept\": {\n\"title\": \"Ink Smoke Duel – Full Strike Fight Sequence\",\n\"description\": \"Continuous 15s one-take 16:9 black ink-smoke fight. Two heroes form from living smoke on a pure white void and immediately fight. Every beat contains a complete strike, clash, or counter — no idle posing. Smoke is the only transition. Typography appears only three times, formed from smoke, never as a flat overlay. Dark cinematic score drives the hits. Always fluid, never static.\",\n\"duration\": \"15s\",\n\"rhythm_structure\": {\n\"0-3s\": \"Both form from opposite sides and charge into first full strike\",\n\"3-6s\": \"Blade lock breaks into Valkyrie full twin-slash + first text\",\n\"6-9s\": \"Ronin full overhead smash + Valkyrie dodge/counter\",\n\"9-12s\": \"Mid-air crossing full strikes + second text\",\n\"12-15s\": \"Final simultaneous full strike, hold, third text\"\n}\n},\n\"camera_direction\": {\n\"shot_type\": \"Strict continuous one-take, 16:9 cinematic widescreen, locked or micro-drift\",\n\"forbidden\": [\"No cuts\", \"No teleportation\", \"No hidden transitions\", \"No freeze frames\", \"No idle standing\"],\n\"camera_journey\": \"Centered medium-wide on pure white void. Extremely slow drift. Energy from living smoke, full blade arcs, impact bursts, and body motion. Slight push-in on the final strike.\"\n},\n\"audio\": {\n\"music\": \"Dark cinematic orchestral + taiko drums + low cello. Hits land on the beat. Blade-clash, fabric-whip, and ink-whoosh SFX under the score. No vocals. Builds through 12s, holds on the final strike.\",\n\"follows_music\": true,\n\"sync_notes\": \"Every full strike lands on a drum or string accent. Smoke explodes on impact.\"\n},\n\"typography\": {\n\"allowed_words\": [\n\"INK DUEL\",\n\"FULL STRIKE\",\n\"WHO REMAINS\"\n],\n\"appearances\": 3,\n\"treatment\": {\n\"not_flat_overlay\": true,\n\"required_properties\": [\n\"Formed from dense black smoke particles and tendrils\",\n\"Physical volumetric depth\",\n\"Smoke occlusion with blades and cloaks\",\n\"Subtle continuous warping\",\n\"Bold uppercase condensed sans-serif\",\n\"Kinetic emergence only at the three specified beats\"\n]\n}\n},\n\"visual_style\": {\n\"overall\": \"Black ink-smoke art, ultra high-contrast monochrome, pure white void, volumetric living smoke, 16:9 widescreen\",\n\"inspired_by\": [\n\"Obsidian Ronin – horned kabuto, liquid-ink mane, gold-filigree armor, heavy katana full swing\",\n\"Inkshade Valkyrie – porcelain face, wild two-tone hair, silver-black gothic armor, dissolving cloak, twin-blade full arcs\"\n],\n\"color_palette\": [\"pure black\", \"deep charcoal smoke\", \"pure white void\"],\n\"materials_and_texture\": \"Living ink smoke, variable density, filament motion, soft volumetric edges, micro-turbulence, condensed-smoke blade trails\"\n},\n\"motion_language\": {\n\"follows_music\": true,\n\"core_behavior\": \"Never static. Smoke always swirls. Heroes form already attacking. Every 3s contains a complete strike that travels full arc and lands or is parried. Morphs only through smoke densifying and reforming. Visible smoke and blade trails in every frame.\"\n},\n\"storyboard\": {\n\"total_beats\": 5,\n\"beat_duration\": \"3 seconds each\",\n\"beats\": [\n{\n\"id\": \"01\",\n\"time\": \"0.0–3.0\",\n\"description\": \"Pure white void. Black smoke splits: from the right it coalesces into Obsidian Ronin already mid-charge, from the left into Inkshade Valkyrie already sprinting. Both complete a first full strike that meets center-frame — his two-handed horizontal cut vs her crossed twin-blade block. Impact burst of dense smoke. No text. End of beat: lock holds one frame then both dissolve into swirling smoke while still moving.\"\n},\n{\n\"id\": \"02\",\n\"time\": \"3.0–6.0\",\n\"description\": \"Smoke reforms them mid-action. Valkyrie completes a full twin-katana X-slash through the Ronin’s dissolving torso; he staggers back as smoke tears off his armor and mane. Her cloak whips a full circle. Text “INK DUEL” forms left from the slash-trail smoke into bold uppercase, then warps with the residual smoke. End of beat: both forms break into smoke again.\"\n},\n{\n\"id\": \"03\",\n\"time\": \"6.0–9.0\",\n\"description\": \"Smoke reverses and reforms. Ronin completes a full overhead smash — katana travels from high-right all the way through to the floor. Valkyrie ducks under the arc, cloak flattening, then drives a full rising single-blade counter into his ribs. Smoke detonates upward from the floor strike and the rib hit. No new text; previous letters shred into filaments. End of beat: both dissolve while still following through.\"\n},\n{\n\"id\": \"04\",\n\"time\": \"9.0–12.0\",\n\"description\": \"Smoke reforms both in mid-air crossing paths. Each completes a full strike at the same instant — his downward diagonal vs her spinning horizontal twin-slash. Blades pass through each other’s smoke-bodies. Text “FULL STRIKE” emerges right from the crossing blade trails into bold uppercase, volumetric and interacting with the clash cloud. End of beat: they drop back into smoke, still rotating.\"\n},\n{\n\"id\": \"05\",\n\"time\": \"12.0–15.0\",\n\"description\": \"Final reform. Both charge from opposite sides and complete simultaneous full strikes that meet dead center in a dense black core. Bodies lean into the hit, cloaks and mane blown backward, blades buried in the impact cloud. Residual smoke briefly reforms then “WHO REMAINS” emerges centered from the core into bold uppercase. They hold the locked strike as smoke slowly thins into the white void. Never fully static until the last frame.\"\n}\n]\n}\n}\n}",
    "promptSourceUrl": "https://x.com/ShamiWeb3/status/2094575793615814722",
    "sourceUrl": "https://x.com/ShamiWeb3/status/2094575793615814722",
    "sourceLabel": "X 原帖 · @ShamiWeb3",
    "author": "Shami",
    "publishedAt": "2026-08-31T23:59:19.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094575793615814722.mp4",
    "posterUrl": "/posters/x/2094575793615814722.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "dark"
    ],
    "scenes": [
      "combat",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "They don’t pose. They strike.\n\nObsidian Ronin vs Inkshade Valkyrie.\n\nSmoke is the only cut. Blades are the only language.\n\nMade with MiniMax H3 on @FishCreativeHQ\n\nPrompt:\n{\n\"archetype\": \"Artistic / Fashion / Conceptual\",\n\"duration\": \"15s\",\n\"aspect_ratio\": \"16:9\",\n\"prompt\": {\n\"concept\": {\n\"title\": \"Ink Smoke Duel – Full Strike Fight Sequence\",\n\"description\": \"Continuous 15s one-take 16:9 black ink-smoke fight. Two heroes form from living smoke on a pure white void and immediately fight. Every beat contains a complete strike, clash, or counter — no idle posing. Smoke is the only transition. Typography appears only three times, formed from smoke, never as a flat overlay. Dark cinematic score drives the hits. Always fluid, never static.\",\n\"duration\": \"15s\",\n\"rhythm_structure\": {\n\"0-3s\": \"Both form from opposite sides and charge into first full strike\",\n\"3-6s\": \"Blade lock breaks into Valkyrie full twin-slash + first text\",\n\"6-9s\": \"Ronin full overhead smash + Valkyrie dodge/counter\",\n\"9-12s\": \"Mid-air crossing full strikes + second text\",\n\"12-15s\": \"Final simultaneous full strike, hold, third text\"\n}\n},\n\"camera_direction\": {\n\"shot_type\": \"Strict continuous one-take, 16:9 cinematic widescreen, locked or micro-drift\",\n\"forbidden\": [\"No cuts\", \"No teleportation\", \"No hidden transitions\", \"No freeze frames\", \"No idle standing\"],\n\"camera_journey\": \"Centered medium-wide on pure white void. Extremely slow drift. Energy from living smoke, full blade arcs, impact bursts, and body motion. Slight push-in on the final strike.\"\n},\n\"audio\": {\n\"music\": \"Dark cinematic orchestral + taiko drums + low cello. Hits land on the beat. Blade-clash, fabric-whip, and ink-whoosh SFX under the score. No vocals. Builds through 12s, holds on the final strike.\",\n\"follows_music\": true,\n\"sync_notes\": \"Every full strike lands on a drum or string accent. Smoke explodes on impact.\"\n},\n\"typography\": {\n\"allowed_words\": [\n\"INK DUEL\",\n\"FULL STRIKE\",\n\"WHO REMAINS\"\n],\n\"appearances\": 3,\n\"treatment\": {\n\"not_flat_overlay\": true,\n\"required_properties\": [\n\"Formed from dense black smoke particles and tendrils\",\n\"Physical volumetric depth\",\n\"Smoke occlusion with blades and cloaks\",\n\"Subtle continuous warping\",\n\"Bold uppercase condensed sans-serif\",\n\"Kinetic emergence only at the three specified beats\"\n]\n}\n},\n\"visual_style\": {\n\"overall\": \"Black ink-smoke art, ultra high-contrast monochrome, pure white void, volumetric living smoke, 16:9 widescreen\",\n\"inspired_by\": [\n\"Obsidian Ronin – horned kabuto, liquid-ink mane, gold-filigree armor, heavy katana full swing\",\n\"Inkshade Valkyrie – porcelain face, wild two-tone hair, silver-black gothic armor, dissolving cloak, twin-blade full arcs\"\n],\n\"color_palette\": [\"pure black\", \"deep charcoal smoke\", \"pure white void\"],\n\"materials_and_texture\": \"Living ink smoke, variable density, filament motion, soft volumetric edges, micro-turbulence, condensed-smoke blade trails\"\n},\n\"motion_language\": {\n\"follows_music\": true,\n\"core_behavior\": \"Never static. Smoke always swirls. Heroes form already attacking. Every 3s contains a complete strike that travels full arc and lands or is parried. Morphs only through smoke densifying and reforming. Visible smoke and blade trails in every frame.\"\n},\n\"storyboard\": {\n\"total_beats\": 5,\n\"beat_duration\": \"3 seconds each\",\n\"beats\": [\n{\n\"id\": \"01\",\n\"time\": \"0.0–3.0\",\n\"description\": \"Pure white void. Black smoke splits: from the right it coalesces into Obsidian Ronin already mid-charge, from the left into Inkshade Valkyrie already sprinting. Both complete a first full strike that meets center-frame — his two-handed horizontal cut vs her crossed twin-blade block. Impact burst of dense smoke. No text. End of beat: lock holds one frame then both dissolve into swirling smoke while still moving.\"\n},\n{\n\"id\": \"02\",\n\"time\": \"3.0–6.0\",\n\"description\": \"Smoke reforms them mid-action. Valkyrie completes a full twin-katana X-slash through the Ronin’s dissolving torso; he staggers back as smoke tears off his armor and mane. Her cloak whips a full circle. Text “INK DUEL” forms left from the slash-trail smoke into bold uppercase, then warps with the residual smoke. End of beat: both forms break into smoke again.\"\n},\n{\n\"id\": \"03\",\n\"time\": \"6.0–9.0\",\n\"description\": \"Smoke reverses and reforms. Ronin completes a full overhead smash — katana travels from high-right all the way through to the floor. Valkyrie ducks under the arc, cloak flattening, then drives a full rising single-blade counter into his ribs. Smoke detonates upward from the floor strike and the rib hit. No new text; previous letters shred into filaments. End of beat: both dissolve while still following through.\"\n},\n{\n\"id\": \"04\",\n\"time\": \"9.0–12.0\",\n\"description\": \"Smoke reforms both in mid-air crossing paths. Each completes a full strike at the same instant — his downward diagonal vs her spinning horizontal twin-slash. Blades pass through each other’s smoke-bodies. Text “FULL STRIKE” emerges right from the crossing blade trails into bold uppercase, volumetric and interacting with the clash cloud. End of beat: they drop back into smoke, still rotating.\"\n},\n{\n\"id\": \"05\",\n\"time\": \"12.0–15.0\",\n\"description\": \"Final reform. Both charge from opposite sides and complete simultaneous full strikes that meet dead center in a dense black core. Bodies lean into the hit, cloaks and mane blown backward, blades buried in the impact cloud. Residual smoke briefly reforms then “WHO REMAINS” emerges centered from the core into bold uppercase. They hold the locked strike as smoke slowly thins into the white void. Never fully static until the last frame.\"\n}\n]\n}\n}\n}",
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 9,
      "views": 106,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094571185917038660",
    "title": "巨剑女战士跃入屋顶亡灵围攻",
    "titleEn": "A Greatsword Warrior Leaps into a Rooftop Horde",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "持巨剑的女战士在阴雨城镇屋顶奔跑，被亡灵包围后跃入密集敌群。 原生视频规格为 29 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A greatsword warrior runs across rainy rooftops, becomes surrounded by undead, and leaps into the horde. Native-video output: 29s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aistudynowcom/status/2094571185917038660",
    "sourceLabel": "X 原帖 · @aistudynowcom",
    "author": "Esha Sharma",
    "publishedAt": "2026-08-31T23:41:00.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094571185917038660.mp4",
    "posterUrl": "/posters/x/2094571185917038660.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@ComfyUI  Minimax H3 with turbo lora, 20GB model. wanted to see how far this thing could actually go so I went full dark fantasy — one swordswoman, massive greatsword, thousands of undead.\n\nshe runs across the rooftops, gets surrounded, then just jumps straight into the horde ⚔️",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 44,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094569437810905369",
    "title": "护士穿越医院与霓虹街区的伪游戏关卡",
    "titleEn": "A Nurse Runs through Hospital and Neon Fake-Game Levels",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "护士角色在病房、车辆与霓虹街区等伪游戏场景间连续切换。 原生视频规格为 75 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A nurse character moves through hospital rooms, vehicles, and neon streets in a sequence of fake-gameplay stages. Native-video output: 75s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Gooboberti/status/2094569437810905369",
    "sourceLabel": "X 原帖 · @Gooboberti",
    "author": "Gooboberti",
    "publishedAt": "2026-08-31T23:34:03.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094569437810905369.mp4",
    "posterUrl": "/posters/x/2094569437810905369.jpg",
    "duration": 75,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The NurseVerse fake gameplay medley - Fal AI Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 38,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094569241701912906",
    "title": "蜗牛宇宙伪游戏关卡在城市中切换",
    "titleEn": "Snailverse Fake-Gameplay Stages Shift across the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏伪游戏画面在霓虹巷道、手机视角与多段城市关卡之间快速切换。 原生视频规格为 61 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Vertical fake gameplay cuts rapidly among neon alleys, phone-view moments, and several city stages. Native-video output: 61s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Gooboberti/status/2094569241701912906",
    "sourceLabel": "X 原帖 · @Gooboberti",
    "author": "Gooboberti",
    "publishedAt": "2026-08-31T23:33:17.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094569241701912906.mp4",
    "posterUrl": "/posters/x/2094569241701912906.jpg",
    "duration": 61,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Welcome to the Snailverse, fake gameplay medley - Fal AI Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 36,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094568116449874056",
    "title": "后院草坪快速挖掘并搭起地下掩体",
    "titleEn": "A Backyard Lawn Is Dug into a DIY Bunker",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "后院草坪从划线、挖坑到木板结构成形，以连续施工片段展示地下掩体搭建。 原生视频规格为 106 秒 · 608×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A backyard progresses from marked grass to excavation and timber walls in a rapid DIY bunker build. Native-video output: 106s · 608×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2094568116449874056",
    "sourceLabel": "X 原帖 · @UnrealRafael",
    "author": "Рафаэль Анриал ∞",
    "publishedAt": "2026-08-31T23:28:48.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094568116449874056.mp4",
    "posterUrl": "/posters/x/2094568116449874056.jpg",
    "duration": 106,
    "aspectRatio": "landscape",
    "resolution": "608×352",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Absolutely insane 🤯\"Building a DIY backyard bunker\"\nThe video is composed of several clips / Final frame for creating the next video. Done manually on an RTX 5090.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 29,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094568098745442483",
    "title": "客厅男子完成一段镜头前讽刺表演",
    "titleEn": "A Living-Room Figure Delivers a Satirical Performance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名男子在普通客厅中起身、停顿并面向镜头完成短促的讽刺表演。 原生视频规格为 8 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man rises, pauses, and addresses the camera in a short satirical performance staged in an ordinary living room. Native-video output: 8s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/matze2001/status/2094568098745442483",
    "sourceLabel": "X 原帖 · @matze2001",
    "author": "Mathias_M",
    "publishedAt": "2026-08-31T23:28:44.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094568098745442483.mp4",
    "posterUrl": "/posters/x/2094568098745442483.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "So hätte Tillschneider dem Russentassen-Gate vmtl. einen anderen Verlauf geben können. 👇\nMal wieder ein KI-Video zur Erhöhung der Medienkompetenz. Lokal erzeugt via MiniMax-H3. Ein paar Sekunden Anfangssequenz können einfach weiter-gepromptet werden. Auch Stimme wird übernommen.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 110,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094564413529591940",
    "title": "巨型猫咪在街头追逐动漫少女",
    "titleEn": "A Giant Cat Chases an Anime Girl through the Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝衣动漫少女穿过霓虹街道奔跑，一只表情夸张的巨猫从后方持续追赶。 原生视频规格为 5 秒 · 480×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-clad anime girl runs through a neon street while an expressive giant cat chases close behind. Native-video output: 5s · 480×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yumesyokunin/status/2094564413529591940",
    "sourceLabel": "X 原帖 · @yumesyokunin",
    "author": "夢職人",
    "publishedAt": "2026-08-31T23:14:05.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094564413529591940.mp4",
    "posterUrl": "/posters/x/2094564413529591940.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "480×640",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax-h3-max\nでかい",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 21,
      "views": 558,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094564093453795339",
    "title": "同一战斗场景用分镜对比相机运动",
    "titleEn": "Split-Screen Battle Frames Compare Camera Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一雨夜战斗场景以多分镜并排呈现，对比静态构图与分段相机运动的变化。 原生视频规格为 15 秒 · 864×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Multiple views of the same rainy battle scene compare a fixed composition with segmented camera movement. Native-video output: 15s · 864×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/theeakiraxx/status/2094564093453795339",
    "sourceLabel": "X 原帖 · @theeakiraxx",
    "author": "akiraxx",
    "publishedAt": "2026-08-31T23:12:49.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094564093453795339.mp4",
    "posterUrl": "/posters/x/2094564093453795339.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "864×960",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で、背景のコンセプトアートを足したらカメラが動かなくなった。\n\n世界観は安定したけど、ずっと正対のまま（上）。\n構図が完成されすぎてて、それ守ろうとしてる感じ。\n\nプロンプトで\n「要素の出典であって構図ではない」って書き足して、\n区間ごとにカメラの動き指定したのが下。\n\n同じ素材で15秒。\nちょっとした一言。\nそれだけで変わる。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 31,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094561470512353381",
    "title": "猫耳女仆双人组同步摆出可爱姿势",
    "titleEn": "Twin Cat-Eared Maids Strike Matching Poses",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名猫耳女仆在城市街头同步抬手、贴近并摆出连续可爱姿势。 原生视频规格为 5 秒 · 704×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two cat-eared maids raise their hands, lean together, and strike a series of matching poses on a city street. Native-video output: 5s · 704×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yumesyokunin/status/2094561470512353381",
    "sourceLabel": "X 原帖 · @yumesyokunin",
    "author": "夢職人",
    "publishedAt": "2026-08-31T23:02:24.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094561470512353381.mp4",
    "posterUrl": "/posters/x/2094561470512353381.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "704×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax-h3-max\n可愛いポーズ",
    "engagement": {
      "replies": 0,
      "reposts": 13,
      "likes": 48,
      "views": 744,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094560873335374254",
    "title": "草莓与玩偶之间展开微缩城市世界",
    "titleEn": "A Miniature City Unfolds between a Strawberry and a Doll",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "桌面草莓和小玩偶之间出现微缩街区，镜头在正常尺度与迷你世界间切换。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A miniature neighborhood appears between a tabletop strawberry and doll as the view shifts between normal and tiny scales. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/musitsune/status/2094560873335374254",
    "sourceLabel": "X 原帖 · @musitsune",
    "author": "ツネ｜IP Creative Director",
    "publishedAt": "2026-08-31T23:00:01.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094560873335374254.mp4",
    "posterUrl": "/posters/x/2094560873335374254.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI映像ならカメラだけじゃなく世界\nそのもののスケールを演出できる🍓\n\nスケール表現も、なんでもできます。\nたぶん。 #MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 446,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094558092457627976",
    "title": "小型载人飞行器从街头升向城市上空",
    "titleEn": "A Passenger Drone Rises from the Street over the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女性在街头启动小型载人飞行器，飞行器升至城市上空并以远景镜头收尾。 原生视频规格为 15 秒 · 1536×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman launches a compact passenger drone from the street, and it rises above the city before a wide aerial finish. Native-video output: 15s · 1536×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/qtum/status/2094558092457627976",
    "sourceLabel": "X 原帖 · @qtum",
    "author": "Qtum",
    "publishedAt": "2026-08-31T22:48:58.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094558092457627976.mp4",
    "posterUrl": "/posters/x/2094558092457627976.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1536×672",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 by @MiniMax_AI just got added to Qtum AI Video. \n\n15 second videos in 720p and 2k quality\nCost effective, far less than other models.\n\nThis 15s 720p video below cost 1326 credits, about a dollar to generate.",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 8,
      "views": 1146,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094557142166409300",
    "title": "粉色衬衫模特切换成故障时尚大片",
    "titleEn": "A Pink-Shirt Model Cuts through a Glitch Fashion Editorial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "粉色衬衫模特在柔光室内完成多个近景姿态，并以快速故障切换形成时尚大片。 原生视频规格为 15 秒 · 832×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A model in a pink shirt moves through soft-lit close-ups and fast glitch transitions in a fashion editorial. Native-video output: 15s · 832×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Fast paced, glitch effects, editorial fashion showcase",
    "promptSourceUrl": "https://x.com/sichy/status/2094557142166409300",
    "sourceUrl": "https://x.com/sichy/status/2094557142166409300",
    "sourceLabel": "X 原帖 · @sichy",
    "author": "Pavel Sich",
    "publishedAt": "2026-08-31T22:45:12.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094557142166409300.mp4",
    "posterUrl": "/posters/x/2094557142166409300.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 is impressive.\n\nEntire prompt: \"Fast paced, glitch effects, editorial fashion showcase\"\n\nDefault prompt enhance on (balanced). Seven words → a full editorial fashion video.\n\nRan it in DreamDwell Labs (H3 Max on fal).",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 65,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094556981516464609",
    "title": "悬浮跑鞋在蓝色光轨中完成产品旋转",
    "titleEn": "A Floating Running Shoe Spins through Blue Light Trails",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色跑鞋悬浮旋转，蓝色光轨环绕鞋底细节并切入品牌化产品定格。 原生视频规格为 5 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black running shoe floats and rotates through blue light trails before settling into a branded product frame. Native-video output: 5s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sena_designer/status/2094556981516464609",
    "sourceLabel": "X 原帖 · @sena_designer",
    "author": "sena｜AI動画",
    "publishedAt": "2026-08-31T22:44:34.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094556981516464609.mp4",
    "posterUrl": "/posters/x/2094556981516464609.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Maxがfal上で1日最大5本まで無料枠生成できるようになったらしいので試してみた！\n画面上には秒数選択の画面はなくて、5秒固定でした。\nとはいえ、動画生成のコストはどんどん下がっていきそうですね🤔",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 80,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094556015165636649",
    "title": "昏暗房间里的单支蜡烛生日",
    "titleEn": "A One-Candle Birthday in a Dim Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "年轻女子在昏暗房间对着糕点上的单支蜡烛许愿，灯亮后露出朴素家庭录像质感。 原生视频规格为 15 秒 · 1024×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A young woman makes a wish over one candle in a dark room before the lights reveal a modest home-video scene. Native-video output: 15s · 1024×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "ULTRA REALISTIC 15 SECOND PERSONAL HOME VIDEO. 1080p render of 4:3 early 2000s MiniDV tape footage. Single continuous handheld take, no cuts. Dialogue driven, Hinglish. Title: EK MOMBATTI.\n\nCONCEPT\nA birthday with no cake. One candle stuck into a pastry from the corner shop, lights switched off in a small room, one friend filming and singing badly. Twelve seconds later the lights are back on and it is over.\n\nSUBJECT LOCK\nIndian woman, early twenties, warm medium brown skin, realistic pores, minimal makeup, face lit almost entirely by a single candle flame from below so the shadows are unflattering and the highlights blow out. Dark brown hair loosely tied back with strands falling loose. Oversized cream knit sweater with stretched cuffs, blue jeans, white sneakers scuffed at the toe. Same face, hair and outfit for the full 15 seconds. Sitting cross legged on the floor, embarrassed, hands in her sleeves.\n\nSECOND VOICE LOCK\nThe friend holding the camcorder is never seen, not a hand, not a shadow. Voice only, young female, close to the mic, singing off key with no shame. Never addresses an audience.\n\nSETTING\nSmall rented room at night with the ceiling light switched off. A single pastry on a steel plate on the floor with one thin candle in it, a mattress on the floor, a plastic chair with clothes on it, a curtain with street light leaking around the edge. Lived in and unstyled. No visible brands, no logos, no readable signage, no decorations.\n\nCAMERA\nRaw handheld MiniDV camcorder held low and close, sitting on the floor across from her. Severe exposure pumping in near darkness, the candle flame blooming into a smeared vertical streak, her face crushed to noise then recovering. Autofocus hunting constantly and never locking. One clumsy zoom in on the candle. Heavy low light tape grain, interlace combing, one dropout glitch. No stabilization, no gimbal.\n\nBEATS\n0:00 to 0:05 She is sitting in the dark behind the candle looking mildly embarrassed while the friend sings the birthday song off key. No lyrics needed, humming and improvised words.\n0:05 to 0:09 The exchange about the wish, quick, overlapping, both laughing.\n0:09 to 0:12 She blows the candle out. Total darkness on tape for two full seconds, only sound, image reduced to noise and a faint window edge. Nobody bothers to fix it.\n0:12 to 0:15 The ceiling light snaps on, the frame blows out white then settles on her squinting, and the tape cuts.\n\nDIALOGUE\nHinglish code switching, natural overlapping delivery, flat and unprojected.\n0:05 She says Wish kya karun\n0:06 Friend off camera says Bata mat, warna nahi hoga\n0:08 She says Nahi bata rahi, chill\n0:13 Friend off camera says Light on kar diya, sorry\nVisible mouth movement with bilabial closure on m, b and p while she is lit. Friend lines off camera only, closer to the mic. The birthday singing is improvised humming, not a quoted song.\n\nAUDIO\nDiegetic only. Off key singing very close to the mic, a ceiling fan, the plate shifting on the floor, her breath blowing out the candle, a light switch click, a scooter outside, a neighbour's television, faint hand noise on the camcorder. No music, no score, no added ambience beds.\n\nSTYLE RULES\nNothing dramatic happens. No posing, no celebration montage, no hero framing, no warm candle glamour lighting. Realism comes from the bad singing, the two seconds of black, and the harsh light at the end. If she performs for the lens the illusion breaks.\n\nNEGATIVE\nNo music. No subtitles. No captions. No watermark. No timestamp. No slow motion. No color grading. No lens flare. No shallow depth of field. No visible second person. No brand logos. No readable text.\n\nLAST FRAME\nHer squinting into the newly switched on ceiling light, overexposed and off center, smoke still rising from the candle, one frame of tape tear, hard cut to black. Absolutely no on screen text, no title card, no lettering, no numbers, no logo, no signage of any kind in the final frame.",
    "promptSourceUrl": "https://x.com/bmx_ai13/status/2094556015165636649",
    "sourceUrl": "https://x.com/bmx_ai13/status/2094556015165636649",
    "sourceLabel": "X 原帖 · @bmx_ai13",
    "author": "BMX",
    "publishedAt": "2026-08-31T22:40:43.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094556015165636649.mp4",
    "posterUrl": "/posters/x/2094556015165636649.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "motion-graphics",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Good Night Everyone.\n\nVideo created with MiniMax H3\nPrompt: ULTRA REALISTIC 15 SECOND PERSONAL HOME VIDEO. 1080p render of 4:3 early 2000s MiniDV tape footage. Single continuous handheld take, no cuts. Dialogue driven, Hinglish. Title: EK MOMBATTI.\n\nCONCEPT\nA birthday with no cake. One candle stuck into a pastry from the corner shop, lights switched off in a small room, one friend filming and singing badly. Twelve seconds later the lights are back on and it is over.\n\nSUBJECT LOCK\nIndian woman, early twenties, warm medium brown skin, realistic pores, minimal makeup, face lit almost entirely by a single candle flame from below so the shadows are unflattering and the highlights blow out. Dark brown hair loosely tied back with strands falling loose. Oversized cream knit sweater with stretched cuffs, blue jeans, white sneakers scuffed at the toe. Same face, hair and outfit for the full 15 seconds. Sitting cross legged on the floor, embarrassed, hands in her sleeves.\n\nSECOND VOICE LOCK\nThe friend holding the camcorder is never seen, not a hand, not a shadow. Voice only, young female, close to the mic, singing off key with no shame. Never addresses an audience.\n\nSETTING\nSmall rented room at night with the ceiling light switched off. A single pastry on a steel plate on the floor with one thin candle in it, a mattress on the floor, a plastic chair with clothes on it, a curtain with street light leaking around the edge. Lived in and unstyled. No visible brands, no logos, no readable signage, no decorations.\n\nCAMERA\nRaw handheld MiniDV camcorder held low and close, sitting on the floor across from her. Severe exposure pumping in near darkness, the candle flame blooming into a smeared vertical streak, her face crushed to noise then recovering. Autofocus hunting constantly and never locking. One clumsy zoom in on the candle. Heavy low light tape grain, interlace combing, one dropout glitch. No stabilization, no gimbal.\n\nBEATS\n0:00 to 0:05 She is sitting in the dark behind the candle looking mildly embarrassed while the friend sings the birthday song off key. No lyrics needed, humming and improvised words.\n0:05 to 0:09 The exchange about the wish, quick, overlapping, both laughing.\n0:09 to 0:12 She blows the candle out. Total darkness on tape for two full seconds, only sound, image reduced to noise and a faint window edge. Nobody bothers to fix it.\n0:12 to 0:15 The ceiling light snaps on, the frame blows out white then settles on her squinting, and the tape cuts.\n\nDIALOGUE\nHinglish code switching, natural overlapping delivery, flat and unprojected.\n0:05 She says Wish kya karun\n0:06 Friend off camera says Bata mat, warna nahi hoga\n0:08 She says Nahi bata rahi, chill\n0:13 Friend off camera says Light on kar diya, sorry\nVisible mouth movement with bilabial closure on m, b and p while she is lit. Friend lines off camera only, closer to the mic. The birthday singing is improvised humming, not a quoted song.\n\nAUDIO\nDiegetic only. Off key singing very close to the mic, a ceiling fan, the plate shifting on the floor, her breath blowing out the candle, a light switch click, a scooter outside, a neighbour's television, faint hand noise on the camcorder. No music, no score, no added ambience beds.\n\nSTYLE RULES\nNothing dramatic happens. No posing, no celebration montage, no hero framing, no warm candle glamour lighting. Realism comes from the bad singing, the two seconds of black, and the harsh light at the end. If she performs for the lens the illusion breaks.\n\nNEGATIVE\nNo music. No subtitles. No captions. No watermark. No timestamp. No slow motion. No color grading. No lens flare. No shallow depth of field. No visible second person. No brand logos. No readable text.\n\nLAST FRAME\nHer squinting into the newly switched on ceiling light, overexposed and off center, smoke still rising from the candle, one frame of tape tear, hard cut to black. Absolutely no on screen text, no title card, no lettering, no numbers, no logo, no signage of any kind in the final frame.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 9,
      "views": 172,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094554524338061630",
    "title": "绿色纸张纹理折叠成排版海报",
    "titleEn": "Green Paper Textures Fold into a Typographic Poster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿色纸张纹理与弧形线条翻转折叠，最终组合成层叠文字海报。 原生视频规格为 15 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Green paper textures and curved lines fold and rotate into a layered typographic poster. Native-video output: 15s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/manga_nimation/status/2094554524338061630",
    "sourceLabel": "X 原帖 · @manga_nimation",
    "author": "彼方｜300日で未経験からAIクリエイターとして成長する記録",
    "publishedAt": "2026-08-31T22:34:48.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094554524338061630.mp4",
    "posterUrl": "/posters/x/2094554524338061630.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3使って\n紙っぽい感じのモーショングラフィック\n\n#TopviewAI\n@TopviewAIJP",
    "engagement": {
      "replies": 2,
      "reposts": 16,
      "likes": 49,
      "views": 430,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094553594393719023",
    "title": "行李箱展开并自动完成旅行收纳",
    "titleEn": "A Suitcase Unpacks and Reorganizes Itself for Travel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "行李箱打开后衣物与旅行用品悬浮展开，再有序折叠回箱中并形成旅行广告定格。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Clothes and travel items float from an open suitcase, then fold back into place for a polished travel-ad finish. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "プレミアムな旅行テックブランドのための、縦型シネマティック・キャンペーンムービーを制作する。ラグジュアリーファッション広告の洗練と、思わず旅に出たくなる高揚感を融合させる。\n\n映像の中心コンセプトは、ひとつのスーツケースから旅の世界全体が立ち上がっていくこと。\n\n冒頭は、ほとんど抽象映像のようなディテールから始める。金属のエッジ、リブ状のシェル表面、ジッパーや金具、使い込まれたトラベルラベル、柔らかな反射光をクローズアップで捉える。光がスーツケースの表面を滑るように移動し、カメラもゆっくりかつ精密に動く。全体像を見せる前に、まずラグジュアリープロダクトの広告映像のような印象をつくる。\n\nその直後、スーツケースが不意にロック解除され、開く。\n\n中身を単に見せるのではなく、荷物が空中へ浮かび上がり、スーツケースの上に立体的な縦方向の構成をつくる。白いシャツは空中で広がり、軽量ジャケットはゆっくり回転する。ニット、サングラス、スニーカー、流れるようなドレス、パスポートケース、小さな旅行小物がそれぞれ独立したレイヤーとして分離し、個別の軌道を描いて浮遊する。\n\n動きは物理的に自然でありながら、映像としては現実以上に美しく整えられていること。布は空気や重力に自然に反応し、硬いアイテムには適切な慣性と重量感を持たせる。一方で、それぞれのアイテムには十分な余白と分離感を与え、雑然とした荷造りではなく、ファッション誌のエディトリアルインスタレーションのような構成にする。\n\nカメラワークは、浮遊するアイテムに合わせて連続的に変化させる。生地の繊維が見えるほどの極端な寄りから広い構図へ抜け、浅い被写界深度で前景の服の間を通過し、浮遊するワードローブの周囲を抑制の効いた軌道で回り込む。トランジションでは一瞬スピードを上げ、その直後に美しいスローモーションへ落とす。フォーカスは、スーツケースの金具からシャツの縫い目、さらにフレームを横切るサングラスへと切り替えていく。細かな繊維、荷物タグ、紙のチケット、微細な埃が光を受け、奥行きを生み出す。\n\nタイポグラフィは単なる画面上のオーバーレイではなく、空間そのものの一部として扱う。\n\n大きく力強い白のエディトリアルタイポグラフィを使い、クリーンで現代的な旅行テックブランドらしい印象をつくる。若々しさ、プレミアム感、デジタルネイティブ感を持たせつつ、余白、整列、文字間の設計は緻密にする。文字は服と服の間のネガティブスペースに入り込み、一部が浮遊物の後ろに隠れたり、スーツケースの中から現れたり、衣服が横切ることで一瞬見えなくなったりする。\n\n使用するコピーは英語と日本語とし、以下を使う。\n\n“PACK LESS. GO MORE.”　\n“YOUR NEXT TRIP STARTS HERE.”　\"世界は、もっと近くなる。\"\n“TRAVEL LIGHTER. MOVE FURTHER.”　”次の旅を、もっと軽く。”\n\n文字自体も映像の動きに参加させる。カメラが近づくにつれて単語が拡大したり、パララックスに合わせて横方向へ移動したり、布の裏側へ消えたり、複数の奥行きレイヤーに分かれたり、最後には正確な位置へスナップするように収束する。装飾的すぎる書体や過剰なエフェクトは避け、スケール、構図、タイミングそのもので強さを出す。\nタイポグラフィの周囲には、最小限のサブグラフィックを加える。細いルートライン、控えめな空港コード、微小なロケーションマーク、簡略化した座標、限定的なインターフェース風パーツなど。実際のアプリ画面を再現するのではなく、現代的な旅行アプリの存在を抽象的に想起させる程度に留める。\n\n色彩は抑制し、上質にまとめる。背景はウォームアイボリー、サンド、柔らかなニュートラルカラーを基調としたスタジオ空間とし、滑らかなグラデーションとシネマティックな光の減衰を入れる。上方斜めからの方向性のある光で立体感をつくり、控えめなフィルライトで生地の質感を保つ。硬質な表面には計算された反射を入れ、サングラス、スーツケースの金具、アクセサリーには小さく鋭いスペキュラーハイライトを入れる。影は十分に深く保ち、浮遊物が実際に空間に存在しているような奥行きをつくる。\n\n終盤では、冒頭の「飛び出す」動きを反転させる。\n\n浮遊していたアイテムが、完全に同期した流れでスーツケースへ戻り始める。靴は回転しながら所定位置へ入り、衣類は空中で折り畳まれ、アクセサリーは正確な収納位置へ降り、細かな旅行小物もひとつずつ消えていく。動きは徐々に加速させるが、最後まで制御された美しさを維持し、乱雑な落下ではなく、精密なモーションデザインとして見せる。\n\nスーツケースが閉じる直前に、小さな予想外の動きをひとつ入れる。荷物タグ、シャツの袖、チケットのいずれかが一瞬だけ再び上へ飛び出し、映像に人間味と遊びを加える。\n\nラストは、きれいに荷造りされたスーツケースのヒーローカット。スーツケースは少しだけ開いた状態でもよい。カメラは短くプッシュインしたあと静かに止まり、画面には十分な余白を残す。最終コピーを明確に配置する。\n\n“YOUR NEXT TRIP STARTS with NEWT.”\n\n全体の方向性は、ラグジュアリーな旅行広告、ファッションエディトリアルのモーション表現、プレミアムなプロダクト撮影、緻密な素材表現、浅い被写界深度、上質な物理シミュレーション、モダンなキネティックタイポグラフィ、洗練されたニュートラルカラー、ダイナミックだが統制されたカメラワーク、若々しい旅行テックの感性、高度に磨き込まれたコマーシャルフィニッシュ。",
    "promptSourceUrl": "https://x.com/shinojapan/status/2094553594393719023",
    "sourceUrl": "https://x.com/shinojapan/status/2094553594393719023",
    "sourceLabel": "X 原帖 · @shinojapan",
    "author": "Takaya Shinozuka / CEO of Reiwa Travel",
    "publishedAt": "2026-08-31T22:31:06.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094553594393719023.mp4",
    "posterUrl": "/posters/x/2094553594393719023.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "stylized"
    ],
    "scenes": [
      "product",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3が、いよいよネクストレベルになってきててやばい。\n\nプロンプト保存用にどうぞ。つ🔖\n\nプレミアムな旅行テックブランドのための、縦型シネマティック・キャンペーンムービーを制作する。ラグジュアリーファッション広告の洗練と、思わず旅に出たくなる高揚感を融合させる。\n\n映像の中心コンセプトは、ひとつのスーツケースから旅の世界全体が立ち上がっていくこと。\n\n冒頭は、ほとんど抽象映像のようなディテールから始める。金属のエッジ、リブ状のシェル表面、ジッパーや金具、使い込まれたトラベルラベル、柔らかな反射光をクローズアップで捉える。光がスーツケースの表面を滑るように移動し、カメラもゆっくりかつ精密に動く。全体像を見せる前に、まずラグジュアリープロダクトの広告映像のような印象をつくる。\n\nその直後、スーツケースが不意にロック解除され、開く。\n\n中身を単に見せるのではなく、荷物が空中へ浮かび上がり、スーツケースの上に立体的な縦方向の構成をつくる。白いシャツは空中で広がり、軽量ジャケットはゆっくり回転する。ニット、サングラス、スニーカー、流れるようなドレス、パスポートケース、小さな旅行小物がそれぞれ独立したレイヤーとして分離し、個別の軌道を描いて浮遊する。\n\n動きは物理的に自然でありながら、映像としては現実以上に美しく整えられていること。布は空気や重力に自然に反応し、硬いアイテムには適切な慣性と重量感を持たせる。一方で、それぞれのアイテムには十分な余白と分離感を与え、雑然とした荷造りではなく、ファッション誌のエディトリアルインスタレーションのような構成にする。\n\nカメラワークは、浮遊するアイテムに合わせて連続的に変化させる。生地の繊維が見えるほどの極端な寄りから広い構図へ抜け、浅い被写界深度で前景の服の間を通過し、浮遊するワードローブの周囲を抑制の効いた軌道で回り込む。トランジションでは一瞬スピードを上げ、その直後に美しいスローモーションへ落とす。フォーカスは、スーツケースの金具からシャツの縫い目、さらにフレームを横切るサングラスへと切り替えていく。細かな繊維、荷物タグ、紙のチケット、微細な埃が光を受け、奥行きを生み出す。\n\nタイポグラフィは単なる画面上のオーバーレイではなく、空間そのものの一部として扱う。\n\n大きく力強い白のエディトリアルタイポグラフィを使い、クリーンで現代的な旅行テックブランドらしい印象をつくる。若々しさ、プレミアム感、デジタルネイティブ感を持たせつつ、余白、整列、文字間の設計は緻密にする。文字は服と服の間のネガティブスペースに入り込み、一部が浮遊物の後ろに隠れたり、スーツケースの中から現れたり、衣服が横切ることで一瞬見えなくなったりする。\n\n使用するコピーは英語と日本語とし、以下を使う。\n\n“PACK LESS. GO MORE.”　\n“YOUR NEXT TRIP STARTS HERE.”　\"世界は、もっと近くなる。\"\n“TRAVEL LIGHTER. MOVE FURTHER.”　”次の旅を、もっと軽く。”\n\n文字自体も映像の動きに参加させる。カメラが近づくにつれて単語が拡大したり、パララックスに合わせて横方向へ移動したり、布の裏側へ消えたり、複数の奥行きレイヤーに分かれたり、最後には正確な位置へスナップするように収束する。装飾的すぎる書体や過剰なエフェクトは避け、スケール、構図、タイミングそのもので強さを出す。\nタイポグラフィの周囲には、最小限のサブグラフィックを加える。細いルートライン、控えめな空港コード、微小なロケーションマーク、簡略化した座標、限定的なインターフェース風パーツなど。実際のアプリ画面を再現するのではなく、現代的な旅行アプリの存在を抽象的に想起させる程度に留める。\n\n色彩は抑制し、上質にまとめる。背景はウォームアイボリー、サンド、柔らかなニュートラルカラーを基調としたスタジオ空間とし、滑らかなグラデーションとシネマティックな光の減衰を入れる。上方斜めからの方向性のある光で立体感をつくり、控えめなフィルライトで生地の質感を保つ。硬質な表面には計算された反射を入れ、サングラス、スーツケースの金具、アクセサリーには小さく鋭いスペキュラーハイライトを入れる。影は十分に深く保ち、浮遊物が実際に空間に存在しているような奥行きをつくる。\n\n終盤では、冒頭の「飛び出す」動きを反転させる。\n\n浮遊していたアイテムが、完全に同期した流れでスーツケースへ戻り始める。靴は回転しながら所定位置へ入り、衣類は空中で折り畳まれ、アクセサリーは正確な収納位置へ降り、細かな旅行小物もひとつずつ消えていく。動きは徐々に加速させるが、最後まで制御された美しさを維持し、乱雑な落下ではなく、精密なモーションデザインとして見せる。\n\nスーツケースが閉じる直前に、小さな予想外の動きをひとつ入れる。荷物タグ、シャツの袖、チケットのいずれかが一瞬だけ再び上へ飛び出し、映像に人間味と遊びを加える。\n\nラストは、きれいに荷造りされたスーツケースのヒーローカット。スーツケースは少しだけ開いた状態でもよい。カメラは短くプッシュインしたあと静かに止まり、画面には十分な余白を残す。最終コピーを明確に配置する。\n\n“YOUR NEXT TRIP STARTS with NEWT.”\n\n全体の方向性は、ラグジュアリーな旅行広告、ファッションエディトリアルのモーション表現、プレミアムなプロダクト撮影、緻密な素材表現、浅い被写界深度、上質な物理シミュレーション、モダンなキネティックタイポグラフィ、洗練されたニュートラルカラー、ダイナミックだが統制されたカメラワーク、若々しい旅行テックの感性、高度に磨き込まれたコマーシャルフィニッシュ。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 8,
      "views": 1812,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094553032679923751",
    "title": "复古电脑屏幕扩张成未来苹果商店",
    "titleEn": "A Vintage Computer Screen Expands into a Futuristic Apple Store",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "复古电脑的绿色线框画面逐步长出材质，并从显像管内部扩张成未来零售空间。 原生视频规格为 14 秒 · 1056×704。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A green wireframe inside a vintage computer gains detail and expands through the CRT into a futuristic retail space. Native-video output: 14s · 1056×704. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Start from the exact uploaded Apple II frame, completely still and photorealistic, like archival footage from the late 1970s. The black CRT suddenly flickers and a primitive glowing green wireframe room begins drawing itself line by line on the screen. The wireframe rapidly gains polygons, textures, realistic lighting, and finally becomes a photorealistic futuristic Apple Store visible inside the CRT. Then the image physically pushes outward through the glass: the CRT surface bulges, the futuristic floor and walls spill into the real room, and the transformation spreads outward from the monitor across the desk and surroundings until the entire environment has become a sleek ultra-modern Apple Store. The original beige Apple II remains completely unchanged in the center, still running. One continuous shot, no cuts, realistic physical transformation, no music, only CRT hum, electrical crackle, subtle construction-like digital sounds, and room ambience.",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2094553032679923751",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2094553032679923751",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-08-31T22:28:52.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094553032679923751.mp4",
    "posterUrl": "/posters/x/2094553032679923751.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1056×704",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "retro"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Prompt: Start from the exact uploaded Apple II frame, completely still and photorealistic, like archival footage from the late 1970s. The black CRT suddenly flickers and a primitive glowing green wireframe room begins drawing itself line by line on the screen. The wireframe rapidly gains polygons, textures, realistic lighting, and finally becomes a photorealistic futuristic Apple Store visible inside the CRT. Then the image physically pushes outward through the glass: the CRT surface bulges, the futuristic floor and walls spill into the real room, and the transformation spreads outward from the monitor across the desk and surroundings until the entire environment has become a sleek ultra-modern Apple Store. The original beige Apple II remains completely unchanged in the center, still running. One continuous shot, no cuts, realistic physical transformation, no music, only CRT hum, electrical crackle, subtle construction-like digital sounds, and room ambience.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 7,
      "views": 924,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094551731405545694",
    "title": "女子面部逐层变成黑色雕塑面具",
    "titleEn": "A Woman’s Face Morphs into a Black Sculptural Mask",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "镜头贴近女子面部，黑色流体结构逐步覆盖五官并形成光滑雕塑面具。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera closes on a woman as black fluid structures spread across her features and form a smooth sculptural mask. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tsutsumu263/status/2094551731405545694",
    "sourceLabel": "X 原帖 · @tsutsumu263",
    "author": "つつむ",
    "publishedAt": "2026-08-31T22:23:42.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094551731405545694.mp4",
    "posterUrl": "/posters/x/2094551731405545694.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max R2V\n6秒で生成できました🤣",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 30,
      "views": 1513,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094551473531326626",
    "title": "真人舞者穿过高速日文排版演出",
    "titleEn": "A Live-Action Dancer Moves through High-Speed Japanese Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "真人舞者在夜景舞台中动作，巨大的日文文字沿人物轮廓快速切换并重组。 原生视频规格为 29 秒 · 2520×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A live-action dancer performs on a night stage while oversized Japanese type moves around the body and rapidly reassembles. Native-video output: 29s · 2520×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "## INPUT【基本設定】\nA: 主人公：提供された参照画像\nB: 表示テキスト：【例：①「START」②「GO」③「FINAL」】\nC: 演出の雰囲気：【例：疾走感のあるクールなMV】\nD: 表現スタイル：【例：アニメ（セルアニメ）】\nE: 表現スタイル詳細：【例：線画を強調したセルアニメ】\nF: テーマ色：【例：白＋赤】\nG: 尺：【例：15秒】\n\nDを全編の唯一の表現基準とし、人物・背景・小物・文字すべてに統一適用。選択した表現スタイル以外の描画・質感・レンダリング表現を混在させない。EはDの範囲内で反映。Aの主人公の顔、髪型、衣装、体型、シルエットを忠実に維持。追加人物なし。\n\n---\n\n## 任意指定タイポグラフィ演出\n\n※使用したい演出は【使用】と記載。空欄はAIが尺・構成に応じてランダム採用・不採用を判断。\n※複数指定時は異なるタイミングで使用。具体的な指示を入力した場合はその内容を優先。\n\n- ボディライン・ラッピング【】：身体の輪郭に沿って文字を湾曲。\n- フェイス・セーフティスプリット【】：顔直前で文字を左右へ分裂。\n- タイポグラフィ・トンネル【】：人物を避けて前景・中景・背景に文字を配置。\n- シルエット・ネガティブスペース【】：人物シルエット部分だけ文字を避ける。\n- 文字の残像リング【】：人物外周を文字がリング状に旋回。\n- 文字回避カメラ【】：俯瞰・超ローアングル等で文字との重なりを回避。\n- 文字分裂・再集合【】：文字が分裂→人物を回避→正しい文字列へ再集合。\n- その他タイポグラフィ演出【】：指定内容を使用。\n\n## テキスト表示\n\nBの順番を厳守。カンマ「,」でグループ化。\n同一グループは「順番表示→全て同時表示」。\n例：①→②→③→①②③同時。\n例：①→②→①②同時→③→④→③④同時。\n指定文字以外、字幕、ロゴ、意味不明な文字は禁止。最終グループの同時表示をクライマックスへ配置。\n\n## 基本演出\n\n主人公の身体動作→カメラ→文字の順に反応。主人公を常に主役とする。\nドリー、ウィップパン、スナップズーム、オービット、高速トラッキング、俯瞰、超ローアングル等を組み合わせ、同一構図を連続使用しない。\n高速運動と完全静止を強く対比し、動作・カメラ・文字・カットをビート同期。\n\n## タイポグラフィ\n\nBの文字のみ使用。主人公の動作に連動：\n横払い→横飛翔／引き寄せ→集合／回転→旋回／手を前へ→急接近／停止→完全停止。\n文字単独のランダム移動は禁止。字間拡大、分解→再整列、文字ごとの遅延追従を短いアクセントとして使用。\n\n## 人物視認性・空間\n\n顔、目、口、頭部、主要シルエットを文字で隠さない。顔周辺はクリアゾーン。\n人物へ接近した文字は、湾曲・分裂・回避・点滅・瞬間消失等で人物を避ける。\n背景文字は人物を囲み、前景文字は顔・身体を避ける。\n文字をカメラ直前・人物手前・人物背後へ配置し、距離差とパララックスで奥行きを表現。\n\n## 尺別自動構成\n\nGに応じてカット数・表示時間・カメラ速度・演出密度を最適化。\n短尺：フック→順番表示→同時表示→フィニッシュ。\n中尺：フック→展開→特殊演出→バレットタイム→再加速→クライマックス。\n長尺：導入→展開→バレットタイム→再加速→展開→クライマックス→ヒーローショット。\n短いほど凝縮、長いほど演出バリエーションを増加。\n最終0.3～0.5秒は主人公＋最終文字列を安定表示。\n\n## バレットタイム\n\n尺に応じて0～2回。\n主人公・髪・衣装・文字を完全静止し、カメラだけ高速オービット。\n時間再開と同時に全要素を一気に再始動。\n\n## Audio\n\nCに合うインストゥルメンタル。ボーカル・セリフなし。\n動作・文字・カメラ・カットをビート同期。バレットタイム直前に音を絞り、再開時に強いビートを復帰。\n\n## 最重要\n\nDの表現スタイルを全編統一し、それ以外を混在させない。\n主人公のデザインを変更しない。追加人物なし。\n指定文字以外を表示しない。表示順・グループ順を厳守。\n主人公の顔・主要シルエットを文字で隠さない。\n主人公の身体動作とカメラワークを最優先。\nタイポグラフィは主人公を隠すためではなく、主人公を強調するために動かす。",
    "promptSourceUrl": "https://x.com/minmin_4410/status/2094328484705558794",
    "sourceUrl": "https://x.com/minmin_4410/status/2094551473531326626",
    "sourceLabel": "X 原帖 · @minmin_4410",
    "author": "minmin みんみん｜変なAI Creator",
    "publishedAt": "2026-08-31T22:22:40.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094551473531326626.mp4",
    "posterUrl": "/posters/x/2094551473531326626.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "2520×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MinimaxH3使って\n【 】内をうめるだけでタイポグラフィで遊べるプロンプト出来た(リプ欄参照)😃✨️\n\n✅️歌詞=動画生成時に指定したテキスト\n\n「なかなかオモロイ」「テンプレ出来た」「皆試してくれ」\n\n「キャラクター画像」「準備して」\n\n「リプ欄のプロンプトの」「【 】内を」「埋めるだけ」\n\n「まずBからGは」「必須！」「自分好みで」「OKだ！」\n\n「まずGは4秒」「テキスト2-3個で」「試せ！」\n\n「その先は任意指定」「タイポグラフィ演出」\n\n「8つの演出から」「選んで」「【 】内に」「【使用】と入れろ！」\n\n「どんな演出があるかは」「この動画を見て」「確認！」\n\n「簡単だろ？」「それじゃあ」「アバヨ！」\n\n#MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 16,
      "views": 513,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094550523156496807",
    "title": "迷你动漫少女围着咖啡杯迎接清晨",
    "titleEn": "A Tiny Anime Girl Greets the Morning beside a Coffee Cup",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "迷你紫发动漫少女绕着咖啡杯移动，探头观察并在勺子靠近时露出笑容。 原生视频规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny purple-haired anime girl moves around a coffee cup, peeks over the rim, and smiles as a spoon approaches. Native-video output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YuuAki101/status/2094550523156496807",
    "sourceLabel": "X 原帖 · @YuuAki101",
    "author": "YuuAki@AIart",
    "publishedAt": "2026-08-31T22:18:54.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094550523156496807.mp4",
    "posterUrl": "/posters/x/2094550523156496807.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはろーです☀️\n\n「モーニングコーヒー」\n\n#MiniMaxH3\n@Hailuo_AI",
    "engagement": {
      "replies": 2,
      "reposts": 6,
      "likes": 18,
      "views": 322,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094547436547268740",
    "title": "街头舞者在夜景广场练习连贯步伐",
    "titleEn": "A Street Dancer Practices a Fluid Routine in a Night Plaza",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴帽舞者在城市夜景广场完成下蹲、侧滑与抬臂动作，保持连贯街舞节奏。 原生视频规格为 15 秒 · 768×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A capped dancer links crouches, side slides, and raised-arm moves in a night-lit city plaza. Native-video output: 15s · 768×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ikucchi99/status/2094547436547268740",
    "sourceLabel": "X 原帖 · @ikucchi99",
    "author": "99ikucchi",
    "publishedAt": "2026-08-31T22:06:38.000Z",
    "addedAt": "2026-09-01T00:16:12.269Z",
    "mediaUrl": "/media/x-2094547436547268740.mp4",
    "posterUrl": "/posters/x/2094547436547268740.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×960",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ダンス練習　もっとうまくなりたい \n#AIart  #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 11,
      "snapshotAt": "2026-09-01T00:16:12.269Z"
    },
    "approvedAt": "2026-09-01T00:16:12.269Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094587498559312324",
    "title": "雨中少女撑伞陪小狗沿河散步",
    "titleEn": "A Girl Walks Her Dog beneath an Umbrella in the Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "撑黑伞的少女与小狗沿雨中河岸步道前行，水面、路面倒影与步伐保持连贯。 原生视频规格为 15 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl under a black umbrella walks her dog along a rainy riverside path with steady motion and wet-surface reflections. Native-video output: 15s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/opener_ai/status/2094587498559312324",
    "sourceLabel": "X 原帖 · @opener_ai",
    "author": "Jin.B",
    "publishedAt": "2026-09-01T00:45:49.000Z",
    "addedAt": "2026-09-02T00:08:44.374Z",
    "mediaUrl": "/media/x-2094587498559312324.mp4",
    "posterUrl": "/posters/x/2094587498559312324.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "we literally need each other fr 😭🤝\n\nHailuo Minimax H3\n\n#hailuo #minimaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 188,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.374Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094938448972915167",
    "title": "棋盘与时钟在超现实风暴中崩裂",
    "titleEn": "A Chessboard and Clock Fracture inside a Surreal Storm",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "孤立人物、巨型眼睛、棋盘与机械时钟在黑金色风暴世界中接连变形和崩裂。 原生视频规格为 12 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A lone figure, giant eye, chessboard, and mechanical clock transform and fracture across a dark gold stormscape. Native-video output: 12s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Bic_Revelation/status/2094938448972915167",
    "sourceLabel": "X 原帖 · @Bic_Revelation",
    "author": "𝓑𝓲𝓬_𝓡𝓮𝓿𝓮𝓵𝓪𝓽𝓲𝓸𝓷",
    "publishedAt": "2026-09-02T00:00:22.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094938448972915167.mp4",
    "posterUrl": "/posters/x/2094938448972915167.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The stone stayed blind. The sky didn’t.\n\nA clock in the storm. A king who never moved.\nThe board broke. The doorway held.\n\nI'm having fun creating with Minimax H3🔥\n\nMinimax H3 Unlimited. @itsPolloAI\nReply Unlimited if you want the link.\n\n#PolloAiCpp #MinimaxH3 #BicRevelation",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 27,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094938356337529050",
    "title": "水彩校服少女跳起完整音乐舞蹈",
    "titleEn": "A Watercolor Schoolgirl Performs a Full Music Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水彩风校服少女在教室中随歌曲完成连续舞步，长镜头保持人物和场景风格统一。 原生视频规格为 70 秒 · 640×992。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A watercolor-style schoolgirl dances through a full song in a classroom while the extended sequence preserves her look and setting. Native-video output: 70s · 640×992. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2094938356337529050",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-09-02T00:00:00.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094938356337529050.mp4",
    "posterUrl": "/posters/x/2094938356337529050.jpg",
    "duration": 70,
    "aspectRatio": "portrait",
    "resolution": "640×992",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [
      "dance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「夏のつづき」 #A眼鏡部\n昨日のダンスにちゃんと歌詞作ってSunoで曲をつけてMVにしてみました。MiniMax H3 ローカル",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 89,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094937839423373541",
    "title": "办公走廊在飞踢后瞬间清空",
    "titleEn": "An Office Hallway Empties after a Flying Kick",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣人物在办公走廊面对无脸西装人，飞踢击碎玻璃后场景突然恢复为空走廊。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-clad figure faces faceless suits in an office corridor, kicks through glass, and leaves the hallway suddenly empty. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/vkuoo/status/2094937839423373541",
    "sourceLabel": "X 原帖 · @vkuoo",
    "author": "vkuoo",
    "publishedAt": "2026-09-01T23:57:57.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094937839423373541.mp4",
    "posterUrl": "/posters/x/2094937839423373541.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🚀 The world around us may not be the reality we believe in.\n\nGenerated with Minimax H3 on @krea_ai",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 18,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094937116514099278",
    "title": "水龙头流出的液体变成发光花田",
    "titleEn": "Water from a Faucet Becomes a Glowing Flower Field",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "手掌接住水龙头流出的青色液体，液体化成发光形体并过渡到夕阳下的花田。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hand catches turquoise liquid from a faucet, which becomes a glowing form before opening into a sunset flower field. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI_VideoLab/status/2094937116514099278",
    "sourceLabel": "X 原帖 · @AI_VideoLab",
    "author": "AI Video Lab｜AI Filmmaking",
    "publishedAt": "2026-09-01T23:55:05.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094937116514099278.mp4",
    "posterUrl": "/posters/x/2094937116514099278.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This is the kind of visual idea AI makes possible.\n\nA simple kitchen faucet suddenly bends reality—water, motion and perspective turning into something completely unexpected.\n\n15 seconds of pure imagination with MiniMax H3.\n\n#MiniMaxH3 #AI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 18,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094936313883713662",
    "title": "黄色个人资料页折叠成时尚海报",
    "titleEn": "A Yellow Profile Page Folds into a Fashion Poster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黄色平面版式通过翻转和遮罩重组，最终形成带人物、文字和数据模块的时尚海报。 原生视频规格为 8 秒 · 768×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A yellow graphic layout flips and masks itself into a fashion poster with a walking figure, type, and data panels. Native-video output: 8s · 768×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ninofku/status/2094936313883713662",
    "sourceLabel": "X 原帖 · @ninofku",
    "author": "ほづみゆうき@データ分析に基づき、子どもにも大人にもやさしい中央区を作ります",
    "publishedAt": "2026-09-01T23:51:53.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094936313883713662.mp4",
    "posterUrl": "/posters/x/2094936313883713662.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "768×1152",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Xのプロフィール欄を用いたファッションポスター風画像生成。毎度ながらプロンプトのクオリティが素晴らしいです\n\nついでにminimax h3で動画にもしてみましたが、これはこれで良いですね",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 91,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094934993835577510",
    "title": "女子从显示器画面走进现实房间",
    "titleEn": "A Woman Steps from a Monitor into the Real Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "显示器里的橙衣女子逐渐跨出屏幕，转身时衣装变化并进入桌面前的现实空间。 原生视频规格为 61 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in orange steps out of a monitor, changes appearance during the transition, and enters the room in front of the desk. Native-video output: 61s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mazzo/status/2094934993835577510",
    "sourceLabel": "X 原帖 · @mazzo",
    "author": "Koya Matsuo",
    "publishedAt": "2026-09-01T23:46:39.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094934993835577510.mp4",
    "posterUrl": "/posters/x/2094934993835577510.jpg",
    "duration": 61,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今朝のソラリスの窓辺。妻の8ミリビデオ動画を学習したMiniMax H3でこんな映像になった。毎晩、RTX 5090が棲むソラリスの海で自動生成される",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 169,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094932418901811658",
    "title": "精灵女子被多把手枪近距离包围",
    "titleEn": "An Elf Woman Is Surrounded by Point-Blank Pistols",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "金发精灵女子在白色背景前被多只黑手套持枪包围，人物表情与枪械位置保持稳定。 原生视频规格为 3 秒 · 1344×752。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde elf woman is encircled by pistols held in black gloves against a white background, with stable expression and placement. Native-video output: 3s · 1344×752. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tomw852/status/2094932418901811658",
    "sourceLabel": "X 原帖 · @Tomw852",
    "author": "Tom𝕎",
    "publishedAt": "2026-09-01T23:36:25.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094932418901811658.mp4",
    "posterUrl": "/posters/x/2094932418901811658.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "1344×752",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This is Gun Fu \n\n(LoRA links used are in the replies)\n────────────────────────────\nGPU: RTX 5070 (12GB)\nModel: #Minimax H3 (int8)\nBased on 8-step PDD Acc sampling (Reference Video-to-Video)\n\n#AIvideo #comfyUI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 57,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094931820039077987",
    "title": "水晶花苞生长成多彩机械玫瑰",
    "titleEn": "A Crystal Bud Grows into a Multicolored Mechanical Rose",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "草地上的透明花苞逐步抽高、展开花瓣，最终形成带金属与水晶质感的多彩玫瑰。 原生视频规格为 10 秒 · 1080×1486。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A transparent bud rises from the grass, unfolds its petals, and becomes a multicolored rose with crystal and metallic textures. Native-video output: 10s · 1080×1486. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/robazem/status/2094931820039077987",
    "sourceLabel": "X 原帖 · @robazem",
    "author": "Rob ✌︎",
    "publishedAt": "2026-09-01T23:34:02.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094931820039077987.mp4",
    "posterUrl": "/posters/x/2094931820039077987.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1080×1486",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cyber Rose growth timelapse made on MiniMax H3 🌹 \n\nFull grown image prompt below, made on ChatGPT.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 63,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094926683614703926",
    "title": "蜘蛛侠装扮的马里奥完成角色亮相",
    "titleEn": "A Spider-Suited Mario Strikes a Character-Intro Pose",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿黑色蜘蛛侠战衣的马里奥在绿色游戏图标背景前转身、挥拳并完成角色亮相。 原生视频规格为 7 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Mario in a black spider suit turns, raises a fist, and completes a character-introduction pose against green game icons. Native-video output: 7s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nonameoasis/status/2094926683614703926",
    "sourceLabel": "X 原帖 · @nonameoasis",
    "author": "nonameoasis",
    "publishedAt": "2026-09-01T23:13:37.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094926683614703926.mp4",
    "posterUrl": "/posters/x/2094926683614703926.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Character Intro with #MiniMaxH3 @Hailuo_AI \n🕸️🕷️🍄🏵️🌟🔥",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 28,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094923428604379417",
    "title": "线稿芭蕾少女完成轻盈旋转",
    "titleEn": "A Sketch-Style Ballerina Completes a Light Spin",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白线稿芭蕾少女在留白背景中抬臂、旋转并连续变换舞姿。 原生视频规格为 15 秒 · 704×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-and-white sketch ballerina raises her arms, spins, and transitions through several poses against an open background. Native-video output: 15s · 704×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2094923428604379417",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-09-01T23:00:41.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094923428604379417.mp4",
    "posterUrl": "/posters/x/2094923428604379417.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "704×928",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "vertical"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます バレー２ #AI眼鏡部 MiniMaxH3ローカル",
    "engagement": {
      "replies": 3,
      "reposts": 3,
      "likes": 25,
      "views": 362,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094923255505187221",
    "title": "餐桌玩偶与月亮完成图形匹配转场",
    "titleEn": "A Tabletop Doll and the Moon Form a Graphic-Match Transition",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "咖啡馆里的小玩偶通过圆形构图切换到满月，再顺滑返回桌面角色。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny café doll transitions through a circular match cut into the full moon before returning smoothly to the tabletop character. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/musitsune/status/2094923255505187221",
    "sourceLabel": "X 原帖 · @musitsune",
    "author": "ツネ｜IP Creative Director",
    "publishedAt": "2026-09-01T23:00:00.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094923255505187221.mp4",
    "posterUrl": "/posters/x/2094923255505187221.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "形、位置、方向を意識するだけで\nまったく違うシーン同士でも\n気持ちよくつながります、たぶん🎈\n\n#MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 536,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094922790579454031",
    "title": "短发少女在卧室靠近镜头交谈",
    "titleEn": "A Short-Haired Girl Leans toward the Camera in a Bedroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短发动画少女在卧室里走近镜头、抬手并进行一段带对白的近距离表演。 原生视频规格为 60 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired animated girl approaches the camera in a bedroom, raises her hand, and delivers a close-up dialogue performance. Native-video output: 60s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mikumiku_aloha/status/2094922790579454031",
    "sourceLabel": "X 原帖 · @mikumiku_aloha",
    "author": "mikumiku_aloha",
    "publishedAt": "2026-09-01T22:58:09.000Z",
    "addedAt": "2026-09-02T00:08:44.375Z",
    "mediaUrl": "/media/x-2094922790579454031.mp4",
    "posterUrl": "/posters/x/2094922790579454031.jpg",
    "duration": 60,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "オタクに優しいギャルを求めて。\n\nセリフや内容はチャッピーが生成しMiniMax H3で動画に。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 73,
      "snapshotAt": "2026-09-02T00:08:44.375Z"
    },
    "approvedAt": "2026-09-02T00:08:44.375Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094922081012253069",
    "title": "街头行人通过遮罩逐步消失",
    "titleEn": "A Street Pedestrian Is Gradually Removed with a Mask",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一街头镜头以多格对比展示行人被遮罩移除的过程，并保留后方道路和绿篱。 原生视频规格为 10 秒 · 960×1596。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A multi-panel street test shows a pedestrian being masked away while preserving the road and hedge behind them. Native-video output: 10s · 960×1596. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AB_exr/status/2094922081012253069",
    "sourceLabel": "X 原帖 · @AB_exr",
    "author": "AB",
    "publishedAt": "2026-09-01T22:55:20.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094922081012253069.mp4",
    "posterUrl": "/posters/x/2094922081012253069.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "960×1596",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimaxH3 inpaintを使用した人物除去テスト。\nsam3で作成したマスクをフェザーしてプロンプトで調整をすすめるが輪郭が現状で残ってしまう。\nあとはここだけ、、\n最後の方にでてくる奥の看板も序盤から破綻せずに生成できているのがすごい。\nクラウドに飛ばさずともここまでいける。\n#aivfx",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 22,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094921819778093122",
    "title": "发光记忆珠串起一段安静科幻短片",
    "titleEn": "Glowing Memory Beads Anchor a Quiet Sci-Fi Short",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "男女角色在夜景、室内和黑场字幕之间切换，以发光记忆珠串联一段情感科幻故事。 原生视频规格为 138 秒 · 704×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two characters move among night scenes, interiors, and title cards in an emotional sci-fi story linked by glowing memory beads. Native-video output: 138s · 704×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2094921819778093122",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-09-01T22:54:18.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094921819778093122.mp4",
    "posterUrl": "/posters/x/2094921819778093122.jpg",
    "duration": 138,
    "aspectRatio": "portrait",
    "resolution": "704×1280",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "motion-graphics",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "In a near-future where you keep your best days as glowing beads in a tin, she finds none of theirs saved — then learns the reason. A quiet sci-fi short.\n\nFull film: https://www.youtube.com/watch?v=3VVzPZkA9Dw\n\nMade with MiniMax-H3\n#ShortFilm #SciFi #AIFilm",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 49,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094920506881323127",
    "title": "对望阳台里的生活片段彼此映照",
    "titleEn": "Life across Opposite Balconies Mirrors Itself",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名住户隔着庭院观察彼此的日常，画面在窗边、阳台与室内片段间交替。 原生视频规格为 38 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two residents watch one another across a courtyard as the film alternates among windows, balconies, and quiet interior moments. Native-video output: 38s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2094920506881323127",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-09-01T22:49:05.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094920506881323127.mp4",
    "posterUrl": "/posters/x/2094920506881323127.jpg",
    "duration": 38,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "She spent a year envying the balcony across the courtyard. Turns out someone across the way was quietly envying hers.\n\nA small parable about the light you can't see yourself giving off.\n\n▶ Full film: https://www.youtube.com/watch?v=RXBe1pReSS0\n\nMade with MiniMax-H3\n#ShortFilm #AIFilm",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 50,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094912497203700100",
    "title": "红色日蚀下的剑士迎战巨型敌人",
    "titleEn": "A Swordsman Faces a Giant Enemy beneath a Red Eclipse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫剑士在红色日蚀与遍地残骸之间起身，随后直面体型巨大的敌人。 原生视频规格为 31 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime swordsman rises beneath a red eclipse among fallen bodies and confronts a towering enemy. Native-video output: 31s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2094912497203700100",
    "sourceLabel": "X 原帖 · @UnrealRafael",
    "author": "Рафаэль Анриал ∞",
    "publishedAt": "2026-09-01T22:17:15.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094912497203700100.mp4",
    "posterUrl": "/posters/x/2094912497203700100.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Berserk: The Golden Age Arc\" minimax H3 Absolutely insane 🤯",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 26,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094909467649310998",
    "title": "软陶角色演示日语问候语",
    "titleEn": "Soft-Clay Characters Teach a Japanese Greeting",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "两个软陶儿童角色通过对话、日文字符和英文释义演示“こんにちは”的发音与含义。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two soft-clay child characters use dialogue, Japanese text, and an English gloss to teach the greeting “konnichiwa.” Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FlashBoardsAI/status/2094909467649310998",
    "sourceLabel": "X 原帖 · @FlashBoardsAI",
    "author": "FlashBoardsOfficial",
    "publishedAt": "2026-09-01T22:05:13.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094909467649310998.mp4",
    "posterUrl": "/posters/x/2094909467649310998.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One prompt = one kids' language lesson.\nA 15-second kids' lesson: correct hiragana, native pronunciation, soft 3D. MiniMax H3.\nPrompt in the reply.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 33,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094908813895983599",
    "title": "蓝发双人组在泳池边切换度假造型",
    "titleEn": "Two Blue-Haired Characters Switch Resort Looks by the Pool",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名蓝发动漫角色在泳池边喝饮料、聊天并切换泳装，保持鲜艳度假氛围。 原生视频规格为 10 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two blue-haired anime characters drink, chat, and switch into swimwear beside a pool in a bright resort setting. Native-video output: 10s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ikedakoen/status/2094908813895983599",
    "sourceLabel": "X 原帖 · @ikedakoen",
    "author": "Koen",
    "publishedAt": "2026-09-01T22:02:37.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094908813895983599.mp4",
    "posterUrl": "/posters/x/2094908813895983599.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "リゾート🍹🏖️🐠\n\nMiniMax H3、たのしいです😃 \n#yukicafemaids #セシル変身",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 139,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094908528473563635",
    "title": "虚拟主播走入极简黑白展厅",
    "titleEn": "A Virtual Host Walks into a Minimal Black-and-White Gallery",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "虚拟主播从圆形灯光下走向黑白文字装置，并以正面自我介绍画面收尾。 原生视频规格为 5 秒 · 544×800。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A virtual host walks from a circular spotlight toward black-and-white type installations and ends in a front-facing introduction. Native-video output: 5s · 544×800. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mikot27567/status/2094908528473563635",
    "sourceLabel": "X 原帖 · @Mikot27567",
    "author": "mikoto",
    "publishedAt": "2026-09-01T22:01:29.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094908528473563635.mp4",
    "posterUrl": "/posters/x/2094908528473563635.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "544×800",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます\n本日のテーマは「影の自治国」です\n\nなんだそりゃｗ\n\nMIniMaxH3アプリイベントが始まったので、そっちへ全振りです。\nそのせいかMIniMaxH3のフローが激重・・・",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 22,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094905215392809447",
    "title": "卡通角色给小汽车做例行检查",
    "titleEn": "Cartoon Characters Give a Small Car Its Inspection",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "纸片卡通角色围着小汽车检查底盘和车身，画面以双语“车辆检查”标题呈现。 原生视频规格为 5 秒 · 720×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Paper-cut cartoon characters inspect a small car’s body and underside beneath a bilingual vehicle-inspection title. Native-video output: 5s · 720×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/haruharuy/status/2094905215392809447",
    "sourceLabel": "X 原帖 · @haruharuy",
    "author": "山下治城:YamashitaHaruki",
    "publishedAt": "2026-09-01T21:48:19.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094905215392809447.mp4",
    "posterUrl": "/posters/x/2094905215392809447.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "車検　動画🎥MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 39,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094900526903099592",
    "title": "夜间警察追逐突然变成地铁舞会",
    "titleEn": "A Night Police Chase Turns into a Subway Dance Party",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "夜间街头追捕切入地铁站，警察与夸张角色在站厅中集体跳舞。 原生视频规格为 31 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A nighttime street pursuit cuts into a subway station where police and exaggerated characters break into a group dance. Native-video output: 31s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ArcaArtificial/status/2094900526903099592",
    "sourceLabel": "X 原帖 · @ArcaArtificial",
    "author": "Arca Artificial by Lordwind Enrique",
    "publishedAt": "2026-09-01T21:29:41.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094900526903099592.mp4",
    "posterUrl": "/posters/x/2094900526903099592.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "TEST Minimax H3 - 1 Reference Image - 30Seg",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 49,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094898401448182191",
    "title": "迷你动漫车手冲过林间拉力赛道",
    "titleEn": "Tiny Anime Drivers Charge through a Forest Rally",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名迷你动漫车手驾驶贴花拉力赛车穿过林间土路，镜头从车外切至前挡风视角。 原生视频规格为 15 秒 · 800×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two tiny anime drivers race a sticker-covered rally car along a forest dirt road as the view shifts from exterior to windshield. Native-video output: 15s · 800×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tengu_AIillust/status/2094898401448182191",
    "sourceLabel": "X 原帖 · @tengu_AIillust",
    "author": "大AI天狗",
    "publishedAt": "2026-09-01T21:21:14.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094898401448182191.mp4",
    "posterUrl": "/posters/x/2094898401448182191.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "800×544",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "皆様、おはようございますー！！\n本日も素敵な1日をお過ごし下さい(∩´∀｀)∩\n\nMiniMax H3によってちびキャララリーを動画化する事に成功しました🥰\nこれが作り放題ってのは途轍もなくすごいですね…！\nなお元画像もプロンプトもチャッピーに作って貰ってるので、一番頑張ってるのはチャッピーです🤣",
    "engagement": {
      "replies": 4,
      "reposts": 10,
      "likes": 35,
      "views": 265,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094892924752339405",
    "title": "孤独行者穿过巨大落日下的海滩",
    "titleEn": "A Lone Walker Crosses a Beach beneath a Giant Sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "人物剪影沿海滩缓慢行走，巨大的橙色落日在低云与海岸线上保持稳定。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silhouetted figure walks slowly along the beach beneath a huge orange sun held steady over the low clouds and shoreline. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SShahHashmi/status/2094892924752339405",
    "sourceLabel": "X 原帖 · @SShahHashmi",
    "author": "𝕊𝕙𝕒𝕜𝕖𝕖𝕝 𝕊𝕙𝕒𝕙",
    "publishedAt": "2026-09-01T20:59:29.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094892924752339405.mp4",
    "posterUrl": "/posters/x/2094892924752339405.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sunset Walk\n\n#MiniMaxH3 T2V Turbo",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 26,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094892572233846952",
    "title": "影星与棋王在老照片里对弈交谈",
    "titleEn": "A Film Star and Chess Champion Converse in an Old Photograph",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "黑白历史照片被赋予细微动作，两位人物在棋盘两侧转头、交谈并继续对局。 原生视频规格为 16 秒 · 1308×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-and-white historical photograph gains subtle motion as two figures turn, converse, and continue their chess game. Native-video output: 16s · 1308×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BernhardSony66/status/2094892572233846952",
    "sourceLabel": "X 原帖 · @BernhardSony66",
    "author": "Bittermandl",
    "publishedAt": "2026-09-01T20:58:05.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094892572233846952.mp4",
    "posterUrl": "/posters/x/2094892572233846952.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1308×720",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"A good player is always lucky.\"\n                            -\n\"Luck is believing you're lucky.\"\n\nFormer world chess champion José Raúl Capablanca with actress Mae Clarke. Mae Clarke was a well-known Hollywood actress of the early 1930s, best remembered for the grapefruit scene with James Cagney in The Public Enemy (1931) and as Elizabeth in Frankenstein (1931). Capablanca visited Los Angeles in April 1933 for a simultaneous chess exhibition at the Los Angeles Athletic Club. Capablanca faced 32 boards, scoring +25 =6 –1.\n\nMade with Minimax H3 in WanGP inside Pinokio!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 26,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094892483096592492",
    "title": "红色日蚀战场展开四十秒动漫决战",
    "titleEn": "A Forty-Second Anime Battle Unfolds beneath a Red Eclipse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "剑士在红色日蚀笼罩的战场中穿过残骸，与巨型敌人展开多镜头连续战斗。 原生视频规格为 41 秒 · 608×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A swordsman crosses a battlefield beneath a red eclipse and fights a giant enemy through a continuous multi-shot sequence. Native-video output: 41s · 608×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2094892483096592492",
    "sourceLabel": "X 原帖 · @UnrealRafael",
    "author": "Рафаэль Анриал ∞",
    "publishedAt": "2026-09-01T20:57:43.000Z",
    "addedAt": "2026-09-02T00:08:44.376Z",
    "mediaUrl": "/media/x-2094892483096592492.mp4",
    "posterUrl": "/posters/x/2094892483096592492.jpg",
    "duration": 41,
    "aspectRatio": "landscape",
    "resolution": "608×352",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Berserk: The Golden Age Arc\" minimax H3 Absolutely insane 🤯 40s",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 29,
      "snapshotAt": "2026-09-02T00:08:44.376Z"
    },
    "approvedAt": "2026-09-02T00:08:44.376Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095175917412962628",
    "title": "红发战士中箭后倒在燃烧战场",
    "titleEn": "A Red-Haired Warrior Falls after an Arrow Strike",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "红发双刀战士冲过燃烧战场，中箭后踉跄失衡并重重倒地，火星和兵器持续运动。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A red-haired dual-sword warrior charges through a burning battlefield, staggers after an arrow strike, and crashes to the ground amid sparks and falling weapons. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, epic fantasy war climax, hyper-real and tragic. The same red-haired female warrior from <Picture 1> is fully preserved in face, hair, body proportions, dual swords, armor, burning battlefield, skulls, spears, smoke, embers, sparks, and infernal war atmosphere. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural color grading with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. The whole video is one continuous shot with no cuts. The camera stays low and close, retreating under pressure while also making slight side drift, subtle breathing-like push-pull, one short impact disturbance, a curved move around her front-side during the breakdown, and a low dipping follow into the fall. From 00:00.000 to 00:04.500, she charges straight toward the lens at high speed, chest angled slightly toward camera-left, head forward, eyes locked ahead, mouth slightly open, breathing hard, shoulders and core tight. One sword drives low at frame-right, the other opens wider at frame-left. Hair, straps, armor, cloth, and blades all carry strong secondary motion. Embers cross the lens and flames boil behind her. From 00:04.500 to 00:06.200, a heavy arrow blasts in from the off-screen right-front side and hits high in her right upper torso near the shoulder and chest. The impact is like a hammer: right shoulder jerks back, chest contracts, head and neck snap slightly rearward, and her breath is cut off. The camera gives one short force reaction. From 00:06.200 to 00:10.500, she does not fall immediately. She forces half a step to one and a half more steps forward while losing control: upper body tipping backward, lower body still trying to drive ahead, compensation steps failing, pain and disbelief visible in her face. The camera arcs around her front-side to watch the breakdown. One sword slips free first and is thrown or dropped; the second weapon also loses control and falls out of its attack line. Hair, shoulders, arms, armor, leather, and weapons all lag with visible inertia while sparks and fire whip around her. From 00:10.500 to 00:13.200, the collapse fully breaks through in layers: upper body whips farther back, chest and shoulder are dragged open by the wound, legs try one last catch and fail, knees buckle, and she slams backward or diagonally backward with heavy force. The camera follows the fall with a low dipping angled movement. From 00:13.200 to 00:15.000, her body and weapons hit the ground with one deep heavy impact, throwing up dust, embers, sparks, and debris. The camera stays low with a small drifting or sinking continuation, holding her as a fallen legend rather than a weak victim. Fire and battlefield ash keep moving to the end.\n\noverall_soundscape: Heavy running footfalls, armor and weapon movement, wind, ember crackle, distant fire roar, arrow whistle, brutal impact into armor and body, broken breath, stumbling loss of control, weapon drops, and one deep final body-and-metal crash.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2095175917412962628",
    "sourceUrl": "https://x.com/ou_zhen599/status/2095175917412962628",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-02T15:43:59.000Z",
    "addedAt": "2026-09-02T17:38:00.648Z",
    "mediaUrl": "/media/x-2095175917412962628.mp4",
    "posterUrl": "/posters/x/2095175917412962628.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3\n\n⚔️“The Final Battle”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, epic fantasy war climax, hyper-real and tragic. The same red-haired female warrior from <Picture 1> is fully preserved in face, hair, body proportions, dual swords, armor, burning battlefield, skulls, spears, smoke, embers, sparks, and infernal war atmosphere. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural color grading with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. The whole video is one continuous shot with no cuts. The camera stays low and close, retreating under pressure while also making slight side drift, subtle breathing-like push-pull, one short impact disturbance, a curved move around her front-side during the breakdown, and a low dipping follow into the fall. From 00:00.000 to 00:04.500, she charges straight toward the lens at high speed, chest angled slightly toward camera-left, head forward, eyes locked ahead, mouth slightly open, breathing hard, shoulders and core tight. One sword drives low at frame-right, the other opens wider at frame-left. Hair, straps, armor, cloth, and blades all carry strong secondary motion. Embers cross the lens and flames boil behind her. From 00:04.500 to 00:06.200, a heavy arrow blasts in from the off-screen right-front side and hits high in her right upper torso near the shoulder and chest. The impact is like a hammer: right shoulder jerks back, chest contracts, head and neck snap slightly rearward, and her breath is cut off. The camera gives one short force reaction. From 00:06.200 to 00:10.500, she does not fall immediately. She forces half a step to one and a half more steps forward while losing control: upper body tipping backward, lower body still trying to drive ahead, compensation steps failing, pain and disbelief visible in her face. The camera arcs around her front-side to watch the breakdown. One sword slips free first and is thrown or dropped; the second weapon also loses control and falls out of its attack line. Hair, shoulders, arms, armor, leather, and weapons all lag with visible inertia while sparks and fire whip around her. From 00:10.500 to 00:13.200, the collapse fully breaks through in layers: upper body whips farther back, chest and shoulder are dragged open by the wound, legs try one last catch and fail, knees buckle, and she slams backward or diagonally backward with heavy force. The camera follows the fall with a low dipping angled movement. From 00:13.200 to 00:15.000, her body and weapons hit the ground with one deep heavy impact, throwing up dust, embers, sparks, and debris. The camera stays low with a small drifting or sinking continuation, holding her as a fallen legend rather than a weak victim. Fire and battlefield ash keep moving to the end.\n\noverall_soundscape: Heavy running footfalls, armor and weapon movement, wind, ember crackle, distant fire roar, arrow whistle, brutal impact into armor and body, broken breath, stumbling loss of control, weapon drops, and one deep final body-and-metal crash.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 141,
      "snapshotAt": "2026-09-02T17:38:00.648Z"
    },
    "approvedAt": "2026-09-02T17:38:00.648Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095171317578502559",
    "title": "女拳手完成单臂平板支撑挑战",
    "titleEn": "A Female Boxer Completes a One-Arm Plank Challenge",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "戴拳套的女拳手从俯卧撑姿势切入单臂平板支撑，身体发力、失衡和坐起动作连续。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A gloved female boxer moves from a push-up into a one-arm plank, with connected strain, imbalance, and recovery motion. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, warm training-room athletic skill showcase. The woman fully matches <Picture 1>: same face, braid, outfit, gloves, shoes, body proportions, warm training room, floor texture, and opening side-sit pose. She starts exactly in the source-image position: right palm planted on the floor, left forearm across the raised left knee, right leg folded under toward camera-right, torso upright and relaxed. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural grading with charcoal black, umber, dark ochre, ash gray, and pale amber subtly interwoven. [Shot 1] Left-front low near-medium shot, gentle push-in plus slight slide right. Motion is alive from frame one: right palm subtly pressing, left fingers shifting on the knee, visible breathing in the chest and shoulders, braid and shoelaces moving slightly. She looks straight at the viewer and says: <d>[English] Watch closely.</d> Then she transfers weight into the right arm, lifts the left forearm away from the knee and keeps it off the floor, raises the pelvis slightly, rotates the torso more downward, and unlocks the left leg backward. [Shot 2] At 00:03.100, the camera cuts to a front-right medium shot with a subtle lateral move and slight move-in. She explosively kicks both legs back into a one-arm supported low angled push-up setup. The right arm is the only support, the chest angles diagonally down-left, the left arm stays off the floor, and the legs extend back in opposite diagonals. She briefly stabilizes the top with visible full-body tension, pelvis lifted, shoulder locked, and no sag. [Shot 3] At 00:05.800, the camera cuts to a left-side low near-medium shot and slowly creeps in. She begins the real descent: the right elbow bends deeply, the body lowers as one unit, and the left hand never touches the floor. She suspends just above the bottom with visible micro tremor through the arm, shoulder, chest, and core, then explosively presses back up cleanly with no wobble and no extra support. [Shot 4] At 00:09.500, the camera cuts to a front-left wider medium shot. It gently pulls back during the recovery as she folds the legs back in, lowers the pelvis smoothly, and settles into a new relaxed side-sit, different from the opening pose: right hand behind the body for support, left knee raised, right leg more extended toward lower-right frame, torso slightly leaned back, shoulders fully relaxed. From 00:12.900 onward the same shot turns into a very light push-in. She tilts her head, smiles with easy confidence, and says: <d>[English] See? It's easier than it looks.</d> The whole piece must show one real hard move only: calm setup, explosive one-arm entry, deep descent, bottom hold, explosive press back up, and relaxed playful finish.\n\noverall_soundscape: Soft training-room ambience, light fabric friction, glove and floor contact, shoe movement, breath, and subtle bodyweight shift sounds.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2095171317578502559",
    "sourceUrl": "https://x.com/ou_zhen599/status/2095171317578502559",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-02T15:25:43.000Z",
    "addedAt": "2026-09-02T17:38:00.648Z",
    "mediaUrl": "/media/x-2095171317578502559.mp4",
    "posterUrl": "/posters/x/2095171317578502559.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "fashion",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3\n\n👟“One-Arm Plank—You’ll Get the Hang of It After Watching This”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, warm training-room athletic skill showcase. The woman fully matches <Picture 1>: same face, braid, outfit, gloves, shoes, body proportions, warm training room, floor texture, and opening side-sit pose. She starts exactly in the source-image position: right palm planted on the floor, left forearm across the raised left knee, right leg folded under toward camera-right, torso upright and relaxed. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural grading with charcoal black, umber, dark ochre, ash gray, and pale amber subtly interwoven. [Shot 1] Left-front low near-medium shot, gentle push-in plus slight slide right. Motion is alive from frame one: right palm subtly pressing, left fingers shifting on the knee, visible breathing in the chest and shoulders, braid and shoelaces moving slightly. She looks straight at the viewer and says: <d>[English] Watch closely.</d> Then she transfers weight into the right arm, lifts the left forearm away from the knee and keeps it off the floor, raises the pelvis slightly, rotates the torso more downward, and unlocks the left leg backward. [Shot 2] At 00:03.100, the camera cuts to a front-right medium shot with a subtle lateral move and slight move-in. She explosively kicks both legs back into a one-arm supported low angled push-up setup. The right arm is the only support, the chest angles diagonally down-left, the left arm stays off the floor, and the legs extend back in opposite diagonals. She briefly stabilizes the top with visible full-body tension, pelvis lifted, shoulder locked, and no sag. [Shot 3] At 00:05.800, the camera cuts to a left-side low near-medium shot and slowly creeps in. She begins the real descent: the right elbow bends deeply, the body lowers as one unit, and the left hand never touches the floor. She suspends just above the bottom with visible micro tremor through the arm, shoulder, chest, and core, then explosively presses back up cleanly with no wobble and no extra support. [Shot 4] At 00:09.500, the camera cuts to a front-left wider medium shot. It gently pulls back during the recovery as she folds the legs back in, lowers the pelvis smoothly, and settles into a new relaxed side-sit, different from the opening pose: right hand behind the body for support, left knee raised, right leg more extended toward lower-right frame, torso slightly leaned back, shoulders fully relaxed. From 00:12.900 onward the same shot turns into a very light push-in. She tilts her head, smiles with easy confidence, and says: <d>[English] See? It's easier than it looks.</d> The whole piece must show one real hard move only: calm setup, explosive one-arm entry, deep descent, bottom hold, explosive press back up, and relaxed playful finish.\n\noverall_soundscape: Soft training-room ambience, light fabric friction, glove and floor contact, shoe movement, breath, and subtle bodyweight shift sounds.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 6,
      "views": 248,
      "snapshotAt": "2026-09-02T17:38:00.648Z"
    },
    "approvedAt": "2026-09-02T17:38:00.648Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095200881167986789",
    "title": "角色剪影在高速图形转场中亮相",
    "titleEn": "A Character Emerges through Rapid Graphic Transitions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "人物在黑白粒子、巨型文字和鲜明平面构图之间高速切换，完成节奏密集的角色亮相。 原生视频规格为 15 秒 · 2544×1456。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A character cuts rapidly among black-and-white particles, oversized type, and bold graphic compositions in a dense reveal sequence. Native-video output: 15s · 2544×1456. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create an explosive, motion-graphics-driven character reveal trailer in 16:9, exactly 13 distinct cuts, 24fps. The character is the exact girl from the provided reference image: preserve her face, proportions, hairstyle, outfit, materials, accessories and colors throughout; never redesign her. This film is 80% bold graphic design in motion and 20% character action: giant kinetic typography, hard-edged shapes, diagonal wipes, flat color fields, barcode strips, UI ticks, halftone, speed lines and shutter flashes. Palette: ice blue, cobalt, white, silver-grey, black. Style: premium AAA motion-graphics title sequence x streetwear campaign film. Every graphic element moves fast and snaps hard on the beat.\nCUT 01 | 0.00-1.00s - Pure graphics: a vertical ice-blue lightning slash rips a black field in half, white shards spray sideways, a barcode strip races left-to-right and breaks; the letters slam in one after another with heavy impact shake and chromatic aberration.\nCUT 02 | 1.00-2.00s - The letters collapse into a receding tunnel of nested rectangles; she bursts through the vanishing point in a low, aggressive sprint, frost particles peeling off her sneakers, camera whipping to follow.\nCUT 03 | 2.00-3.10s - She plants both hands on a giant rotating block of white typography and pendulum-swings underneath it, kicking the letters into spinning fragments that trail ice-blue streaks; whip-pan as she exits frame right.\nCUT 04 | 3.10-4.00s - Full-bleed diagonal color fields (cobalt / white / ice-blue) wipe in sequence; she appears only as a high-contrast silhouette performing a single continuous wall-kick into a mid-air twist. Huge horizontal text OVERDRIVE scrolls in the opposite direction of her travel.\nCUT 05 | 4.00-5.10s - She ricochets off a towering stack of stacked words reading KEEP MOVING, each bounce sending a letter flying; camera tracks her in a tight orbit as the stack topples, then snap-zooms to her determined profile.\nCUT 06 | 5.10-6.00s - Hard cut to a stark white editorial card with enormous black DROP and a single cobalt underline; she drops from the top of frame, lands on the word, compresses it like a trampoline and launches again; the type rebounds with a spring distortion.\nCUT 07 | 6.00-7.00s - Silver-grey field crossed by a thick black chevron; she cartwheels along the chevron in four stroboscopic ghost frames, each ghost a different palette color; giant outlined text SPIN rolls 360 degrees in perfect sync with her rotation.\nCUT 08 | 7.00-8.00s - Kinetic type barrage: the words SHARP / QUICK / FRESH / COLD slam in one per beat with shutter flashes and camera shake while she power-slides on one knee across the foreground, jacket flaring, frost bursting from the platform sneakers.\nCUT 09 | 8.00-9.00s - Black frame, giant white isometric cubes tumbling in 3D space; she sprints across the tops of the cubes, plants a foot and launches into a tight corkscrew, freezing at the apex; a thin ice-blue reticle and data ticks lock onto her silhouette.\nCUT 10 | 9.00-10.10s - Freeze releases: she dives toward camera through successive flat-color planes that peel away like torn posters, each plane revealing a larger fragment of her name; a sneaker wipe across the lens ends the shot.\nCUT 11 | 10.10-11.10s - Rapid-fire graphic cards: five full-screen layouts snap past with hard cuts — each a different bold composition (offset type, oversized chevron, barcode overlay, slashed crop, halftone burst) showing her in a unique athletic beat (handspring, low dash, mid-air reach, crouched ready stance, walking away). Numbers 01–05 punch in the corners.\nCUT 12 | 11.10-13.00s - Hero moment on a clean white cyclorama: she lands a final twisting aerial dead center in slow motion, plants, stands tall with one hand in a pocket, and a shockwave of concentric ice-blue rings, wind streaks and shattered letterforms blasts outward; brief iconic freeze with her faint confident expression, then overexpose to white.\nCUT 13 | 13.00-15.00s - Final identity card: enormous black wordmark dominates a pale white field with translucent ice-blue rings, technical arcs, halftone and a rough black circular emblem containing a ghosted snow-crystal / barcode motif; she stands relaxed overlapping the letters, wind rippling her jacket; one last ice-blue pulse sweeps through the typography and a frost-ring flash punctuates the end.\nEditing: extremely aggressive rhythm — hard cuts on every beat, graphic matches, whip pans, snap zooms, stroboscopic freezes, foreground wipes, RGB splits, shutter flashes, impact shakes and speed ramps; every cut must feel compositionally different; typography is always fully readable before the character overlaps it. No weapons, no combat, no fire; all energy comes from motion design, wind, frost, glass and her parkour-style athleticism. BGM: hard-hitting electronic / drum-heavy future bass with aggressive drops, risers, sub hits and glitch fills locked to every cut; sneaker impacts, whooshes, glass shatters and typography slams are rhythmic sound-design elements. Peak at CUT 12 and end with a cold electronic logo stinger. Premium AAA quality, anime-inspired cinematic rendering, stylish and explosive, strong graphic-design identity, consistent character design, exactly 13 cuts.",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2095200881167986789",
    "sourceUrl": "https://x.com/Just_sharon7/status/2095200881167986789",
    "sourceLabel": "X 原帖 · @Just_sharon7",
    "author": "Sharon Riley",
    "publishedAt": "2026-09-02T17:23:11.000Z",
    "addedAt": "2026-09-02T17:38:00.648Z",
    "mediaUrl": "/media/x-2095200881167986789.mp4",
    "posterUrl": "/posters/x/2095200881167986789.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2544×1456",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Thirteen cuts later you still cannot look away from her.\n\nMinimax H3 on @FishCreativeHQ \n\nPrompt\n\nCreate an explosive, motion-graphics-driven character reveal trailer in 16:9, exactly 13 distinct cuts, 24fps. The character is the exact girl from the provided reference image: preserve her face, proportions, hairstyle, outfit, materials, accessories and colors throughout; never redesign her. This film is 80% bold graphic design in motion and 20% character action: giant kinetic typography, hard-edged shapes, diagonal wipes, flat color fields, barcode strips, UI ticks, halftone, speed lines and shutter flashes. Palette: ice blue, cobalt, white, silver-grey, black. Style: premium AAA motion-graphics title sequence x streetwear campaign film. Every graphic element moves fast and snaps hard on the beat.\nCUT 01 | 0.00-1.00s - Pure graphics: a vertical ice-blue lightning slash rips a black field in half, white shards spray sideways, a barcode strip races left-to-right and breaks; the letters slam in one after another with heavy impact shake and chromatic aberration.\nCUT 02 | 1.00-2.00s - The letters collapse into a receding tunnel of nested rectangles; she bursts through the vanishing point in a low, aggressive sprint, frost particles peeling off her sneakers, camera whipping to follow.\nCUT 03 | 2.00-3.10s - She plants both hands on a giant rotating block of white typography and pendulum-swings underneath it, kicking the letters into spinning fragments that trail ice-blue streaks; whip-pan as she exits frame right.\nCUT 04 | 3.10-4.00s - Full-bleed diagonal color fields (cobalt / white / ice-blue) wipe in sequence; she appears only as a high-contrast silhouette performing a single continuous wall-kick into a mid-air twist. Huge horizontal text OVERDRIVE scrolls in the opposite direction of her travel.\nCUT 05 | 4.00-5.10s - She ricochets off a towering stack of stacked words reading KEEP MOVING, each bounce sending a letter flying; camera tracks her in a tight orbit as the stack topples, then snap-zooms to her determined profile.\nCUT 06 | 5.10-6.00s - Hard cut to a stark white editorial card with enormous black DROP and a single cobalt underline; she drops from the top of frame, lands on the word, compresses it like a trampoline and launches again; the type rebounds with a spring distortion.\nCUT 07 | 6.00-7.00s - Silver-grey field crossed by a thick black chevron; she cartwheels along the chevron in four stroboscopic ghost frames, each ghost a different palette color; giant outlined text SPIN rolls 360 degrees in perfect sync with her rotation.\nCUT 08 | 7.00-8.00s - Kinetic type barrage: the words SHARP / QUICK / FRESH / COLD slam in one per beat with shutter flashes and camera shake while she power-slides on one knee across the foreground, jacket flaring, frost bursting from the platform sneakers.\nCUT 09 | 8.00-9.00s - Black frame, giant white isometric cubes tumbling in 3D space; she sprints across the tops of the cubes, plants a foot and launches into a tight corkscrew, freezing at the apex; a thin ice-blue reticle and data ticks lock onto her silhouette.\nCUT 10 | 9.00-10.10s - Freeze releases: she dives toward camera through successive flat-color planes that peel away like torn posters, each plane revealing a larger fragment of her name; a sneaker wipe across the lens ends the shot.\nCUT 11 | 10.10-11.10s - Rapid-fire graphic cards: five full-screen layouts snap past with hard cuts — each a different bold composition (offset type, oversized chevron, barcode overlay, slashed crop, halftone burst) showing her in a unique athletic beat (handspring, low dash, mid-air reach, crouched ready stance, walking away). Numbers 01–05 punch in the corners.\nCUT 12 | 11.10-13.00s - Hero moment on a clean white cyclorama: she lands a final twisting aerial dead center in slow motion, plants, stands tall with one hand in a pocket, and a shockwave of concentric ice-blue rings, wind streaks and shattered letterforms blasts outward; brief iconic freeze with her faint confident expression, then overexpose to white.\nCUT 13 | 13.00-15.00s - Final identity card: enormous black wordmark dominates a pale white field with translucent ice-blue rings, technical arcs, halftone and a rough black circular emblem containing a ghosted snow-crystal / barcode motif; she stands relaxed overlapping the letters, wind rippling her jacket; one last ice-blue pulse sweeps through the typography and a frost-ring flash punctuates the end.\nEditing: extremely aggressive rhythm — hard cuts on every beat, graphic matches, whip pans, snap zooms, stroboscopic freezes, foreground wipes, RGB splits, shutter flashes, impact shakes and speed ramps; every cut must feel compositionally different; typography is always fully readable before the character overlaps it. No weapons, no combat, no fire; all energy comes from motion design, wind, frost, glass and her parkour-style athleticism. BGM: hard-hitting electronic / drum-heavy future bass with aggressive drops, risers, sub hits and glitch fills locked to every cut; sneaker impacts, whooshes, glass shatters and typography slams are rhythmic sound-design elements. Peak at CUT 12 and end with a cold electronic logo stinger. Premium AAA quality, anime-inspired cinematic rendering, stylish and explosive, strong graphic-design identity, consistent character design, exactly 13 cuts.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 6,
      "views": 216,
      "snapshotAt": "2026-09-02T17:38:00.648Z"
    },
    "approvedAt": "2026-09-02T17:38:00.648Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095198265813713087",
    "title": "金发动画少女被手枪团团围住",
    "titleEn": "A Blonde Anime Girl Is Encircled by Pistols",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发动画少女在黑色背景前保持镇定，多把白色手枪从不同方向逐步靠近她的脸。 原生视频规格为 3 秒 · 960×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde anime girl remains composed against black as white pistols close in on her face from multiple directions. Native-video output: 3s · 960×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SmallKino_/status/2095198265813713087",
    "sourceLabel": "X 原帖 · @SmallKino_",
    "author": "SmallKino",
    "publishedAt": "2026-09-02T17:12:48.000Z",
    "addedAt": "2026-09-02T17:38:00.648Z",
    "mediaUrl": "/media/x-2095198265813713087.mp4",
    "posterUrl": "/posters/x/2095198265813713087.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "960×640",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This time It's hiyori \n\n#BLEACH #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 59,
      "snapshotAt": "2026-09-02T17:38:00.649Z"
    },
    "approvedAt": "2026-09-02T17:38:00.648Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095193333266268655",
    "title": "牛仔外套少女在铁网前随性起舞",
    "titleEn": "A Girl in a Denim Jacket Dances by a Chain-Link Fence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红色泳装与牛仔外套造型的女子在户外铁网前连续摆动、转身并变换舞姿。 原生视频规格为 5 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a red outfit and denim jacket sways, turns, and changes poses in a continuous dance beside an outdoor chain-link fence. Native-video output: 5s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/miokurose/status/2095193333266268655",
    "sourceLabel": "X 原帖 · @miokurose",
    "author": "黒瀬 澪",
    "publishedAt": "2026-09-02T16:53:12.000Z",
    "addedAt": "2026-09-02T17:38:00.649Z",
    "mediaUrl": "/media/x-2095193333266268655.mp4",
    "posterUrl": "/posters/x/2095193333266268655.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "金網の前でちょっとダンスしてみたよ。\nデニムジャケット羽織ったまま動くの、\n意外と楽しい  \n\nSFW.\nNo nudity, no suggestive intent.\nRedCraft / METAFILM\nKTMix_LoRA\nPowered by Hailuo AI (MiniMax H3)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 11,
      "views": 214,
      "snapshotAt": "2026-09-02T17:38:00.649Z"
    },
    "approvedAt": "2026-09-02T17:38:00.649Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095193171311436225",
    "title": "白色巨龙在森林空地逼近人物",
    "titleEn": "A White Dragon Looms over a Figure in a Forest Clearing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "体型巨大的白色羽鳞龙在林间空地低头逼近白衣人物，眼神和头部运动保持连贯。 原生视频规格为 8 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A colossal white feather-scaled dragon lowers its head toward a white-clad figure in a forest clearing with steady gaze and head motion. Native-video output: 8s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kulufly/status/2095193171311436225",
    "sourceLabel": "X 原帖 · @kulufly",
    "author": "Kulufly",
    "publishedAt": "2026-09-02T16:52:33.000Z",
    "addedAt": "2026-09-02T17:38:00.649Z",
    "mediaUrl": "/media/x-2095193171311436225.mp4",
    "posterUrl": "/posters/x/2095193171311436225.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Happy Dragon Day! 🐉\n\nCreated in @Hailuo_AI \n\n#hailuoai #minimaxh3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 20,
      "snapshotAt": "2026-09-02T17:38:00.649Z"
    },
    "approvedAt": "2026-09-02T17:38:00.649Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095190591852408902",
    "title": "留声机房间延展成水晶舞厅",
    "titleEn": "A Gramophone Room Expands into a Crystal Ballroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "老人操作留声机后，居室逐步延展成挂满水晶灯的舞厅，成群舞者随镜头出现。 原生视频规格为 16 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "After an elderly man tends a gramophone, his room expands into a chandelier-filled ballroom populated by dancing couples. Native-video output: 16s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2095190591852408902",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-09-02T16:42:18.000Z",
    "addedAt": "2026-09-02T17:38:00.649Z",
    "mediaUrl": "/media/x-2095190591852408902.mp4",
    "posterUrl": "/posters/x/2095190591852408902.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI ver1.5 Final\nFastVideo I2V Motion Control 768p\n\nFVはまだI2Vが蒸留されてなくて未対応（と言うか出せる映像が💩）なので、セグメント2以降はlightx2v 4step v1.1 768pに切り替える事で綺麗な絵になるように。妥協案なのでFV FL2Vモデルが出たら速攻で戻します",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 115,
      "snapshotAt": "2026-09-02T17:38:00.649Z"
    },
    "approvedAt": "2026-09-02T17:38:00.649Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095188231340056976",
    "title": "灰狗体验从上班到瘫倒的一天",
    "titleEn": "A Greyhound Endures a Full Day at the Office",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿白衬衫的灰狗从床上醒来、坐在电脑前工作，最后精疲力竭地倒在沙发上。 原生视频规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A greyhound in a white shirt wakes in bed, works at an office computer, and finally collapses exhausted on a sofa. Native-video output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HeyZaraKhan/status/2095188231340056976",
    "sourceLabel": "X 原帖 · @HeyZaraKhan",
    "author": "ZARA",
    "publishedAt": "2026-09-02T16:32:55.000Z",
    "addedAt": "2026-09-02T17:38:00.649Z",
    "mediaUrl": "/media/x-2095188231340056976.mp4",
    "posterUrl": "/posters/x/2095188231340056976.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The daily life of a corporate employee... 😫\n\nCreated with MiniMax H3",
    "engagement": {
      "replies": 1,
      "reposts": 7,
      "likes": 10,
      "views": 903,
      "snapshotAt": "2026-09-02T17:38:00.649Z"
    },
    "approvedAt": "2026-09-02T17:38:00.649Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095186858108154190",
    "title": "冰晶歌姬在蓝色舞台完成亮相",
    "titleEn": "An Ice-Crystal Songstress Takes the Blue Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发动画歌姬从闭眼特写过渡到蓝色舞台演出，冰晶翅片和角色标题在结尾展开。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired animated songstress moves from a closed-eye close-up to a blue stage performance ending with crystal wings and a character title. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LiluluLoserLab/status/2095186858108154190",
    "sourceLabel": "X 原帖 · @LiluluLoserLab",
    "author": "リルル@Loser Lab",
    "publishedAt": "2026-09-02T16:27:28.000Z",
    "addedAt": "2026-09-02T17:38:00.649Z",
    "mediaUrl": "/media/x-2095186858108154190.mp4",
    "posterUrl": "/posters/x/2095186858108154190.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Lumina, the Ice Crystal Chantress　MiniMax H3\nKikiさんのプロンプトを元にチャッピーと一緒に勉強した成果がこちらです。\n\nKikiさんに感謝！ヽ(*'ω'*)ノ\n@Mayz1169",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 69,
      "snapshotAt": "2026-09-02T17:38:00.649Z"
    },
    "approvedAt": "2026-09-02T17:38:00.649Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095186838038216756",
    "title": "赛亚人风格少女完成能量变身",
    "titleEn": "A Saiyan-Inspired Girl Completes an Energy Transformation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短发动画少女在黑色背景前蓄力，金色能量爆发后变成带尾巴的战斗形态。 原生视频规格为 10 秒 · 1088×1600。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired anime girl powers up against black and transforms in a golden energy burst into a tailed combat form. Native-video output: 10s · 1088×1600. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/_GAi92/status/2095186838038216756",
    "sourceLabel": "X 原帖 · @_GAi92",
    "author": "_GAi",
    "publishedAt": "2026-09-02T16:27:23.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095186838038216756.mp4",
    "posterUrl": "/posters/x/2095186838038216756.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1088×1600",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I still can't get 2D to look right, it always ends up looking like some live-2D stuff...\n\nAh well. Here's Rhuba, about to throw hands.\n#minimaxh3",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 13,
      "views": 264,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095183410738135094",
    "title": "同一提示在两个版本间呈现不同结果",
    "titleEn": "One Prompt Produces Different Results across Two Versions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "并排画面用动画人物和动态图形展示相同指令在两个版本中的构图与动作差异。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Side-by-side animation and motion graphics show how the same instruction changes composition and movement across two versions. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YouArtStudio/status/2095183410738135094",
    "sourceLabel": "X 原帖 · @YouArtStudio",
    "author": "YouArt",
    "publishedAt": "2026-09-02T16:13:46.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095183410738135094.mp4",
    "posterUrl": "/posters/x/2095183410738135094.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "How much difference can one sentence make?\n\nMiniMax H3 vs. H3 Max—tested side by side on YouArt.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 335,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095181428850102728",
    "title": "电梯门开合间保持男子面部清晰",
    "titleEn": "A Man’s Face Stays Clear as Elevator Doors Shift",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "西装男子在电梯门开合与风景转场中保持正面面部细节，用于展示视频面部修复效果。 原生视频规格为 5 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A suited man retains frontal facial detail as elevator doors and scenery change around him in a video face-refinement test. Native-video output: 5s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2095181428850102728",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-09-02T16:05:53.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095181428850102728.mp4",
    "posterUrl": "/posters/x/2095181428850102728.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI ver1.5 Final \nA100 Face Refiner: 1376×768 5s ⏱38s\n\nFace Refiner（動画版のADetailerみたいなやつ）もA100だとどうかな〜と思ってたけどフツーに使えるな🤔",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 164,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095179948805980236",
    "title": "奔跑人物撞破纸片城市寻找出口",
    "titleEn": "A Runner Breaks through a Paper City to Find an Exit",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色人物在蓝橙纸片城市中奔跑、撞穿建筑，镜头拉远后以“Find Your Way”构图收尾。 原生视频规格为 12 秒 · 2650×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark figure runs through and breaks apart a blue-and-orange paper city before the camera pulls back into a “Find Your Way” composition. Native-video output: 12s · 2650×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/os_orabi/status/2095179948805980236",
    "sourceLabel": "X 原帖 · @os_orabi",
    "author": "osama orabi",
    "publishedAt": "2026-09-02T16:00:01.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095179948805980236.mp4",
    "posterUrl": "/posters/x/2095179948805980236.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "2650×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "dark"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "FIND YOUR WAY - H3\n#MiniMaxH3 #minimaxdesign",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 65,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095179726805770518",
    "title": "魔法报纸带来霍格沃茨停办噩耗",
    "titleEn": "A Magic Newspaper Announces Hogwarts Is Closing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "学生阅读会动的魔法报纸，看到霍格沃茨停办标题后震惊后仰并用日语惊呼。 原生视频规格为 12 秒 · 672×928。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A student reads a moving magical newspaper, recoils at a Hogwarts-closing headline, and reacts aloud in Japanese. Native-video output: 12s · 672×928. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "I2VA，12秒单镜头：上传图作为0秒首帧，人物阅读魔法报纸→发现“HOGWARTS TO CLOSE”→表情夸张震惊并后仰→日语惊呼「うそ……これ、本当なの!? ホグワーツが閉校ですって……ありえない！」→最后保持震惊定格，全程慢推近、电影感、人物身份稳定、口型同步。",
    "promptSourceUrl": "https://x.com/lucasyao99261/status/2095180476210393590",
    "sourceUrl": "https://x.com/lucasyao99261/status/2095179726805770518",
    "sourceLabel": "X 原帖 · @lucasyao99261",
    "author": "Lucasyao",
    "publishedAt": "2026-09-02T15:59:08.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095179726805770518.mp4",
    "posterUrl": "/posters/x/2095179726805770518.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "672×928",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hogwarts Just Got the Worst News ⚡\nWhat if a magical newspaper revealed the impossible? 🎬✨\n#MiniMaxH3 #AIVideo #AIAnimation #CinematicAI #HarryPotter \nprompts below 👇",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 19,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095178028829507813",
    "title": "真人女孩走进像素换装游戏",
    "titleEn": "A Real Girl Steps into a Pixel Dress-Up Game",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "真人女孩在房间中消失，随后以像素换装界面的角色出现并完成绿色礼服选择。 原生视频规格为 15 秒 · 640×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A real girl vanishes from a room and reappears as a character in a pixel dress-up interface that completes a green-gown selection. Native-video output: 15s · 640×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/i4LMU/status/2095178028829507813",
    "sourceLabel": "X 原帖 · @i4LMU",
    "author": "Ullr",
    "publishedAt": "2026-09-02T15:52:23.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095178028829507813.mp4",
    "posterUrl": "/posters/x/2095178028829507813.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "640×480",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "real girl, pixel game energy 💫 Powered by #MiniMaxH3 on @magnific .",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 27,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095176867217350732",
    "title": "动画记者在街头完成镜头播报",
    "titleEn": "An Anime Reporter Delivers a Street Broadcast",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发动画记者手持麦克风在街头面向镜头播报，场景和人物身份在镜头变化中保持稳定。 原生视频规格为 10 秒 · 1024×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired anime reporter addresses the camera with a microphone on a city street while her identity and setting remain stable. Native-video output: 10s · 1024×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ITOKtw/status/2095176867217350732",
    "sourceLabel": "X 原帖 · @ITOKtw",
    "author": "ITOK",
    "publishedAt": "2026-09-02T15:47:46.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095176867217350732.mp4",
    "posterUrl": "/posters/x/2095176867217350732.jpg",
    "duration": 10,
    "aspectRatio": "square",
    "resolution": "1024×1024",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime",
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MiniMaxH3 controlnetの接続の確認をしていたときに出来た，controlを無視したけど，何か良かったもの。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 34,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095173208962150604",
    "title": "红蓝双人组展开卡通追逐舞蹈",
    "titleEn": "A Red-and-Blue Duo Turns a Chase into a Cartoon Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "红发与蓝发角色在平面图形空间中追逐、抢占中心并同步舞蹈，最终以双人造型收尾。 原生视频规格为 9 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Red- and blue-haired characters chase, compete for center stage, and synchronize their dance across a flat graphic world before a paired finale. Native-video output: 9s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a high-energy 2D anime/cartoon music video featuring the exact two characters from the provided reference. The sequence should feel like one continuous playful performance built around their rivalry, contrasting personalities and eventual synchronization. No text, typography, letters, logos or title cards anywhere.  \n\nCHARACTER LOCK: Use the reference as the absolute authority for both characters’ identities, proportions, outfits, colors and art style. Never merge or exchange their designs.  BLUE GIRL: extremely long bright-cyan twin-tails, cyan bangs, navy pinstripe visor cap with a large red-and-white striped bow, pale skin, blue eyes, playful sharp-toothed expression, navy pinstripe maid-style dress with white apron and frilled hem, oversized cuffs, fingerless gloves, white leg warmers and red roller shoes with yellow details. She is mischievous, teasing and energetic.  RED GIRL: vivid red hair arranged into large outward ribbon-like sections, small red-and-black cap with yellow smiley badge, pale skin, expressive red eyes, blue-and-white striped shirt, black tie, red suspenders, high-waisted red trousers, bright yellow gloves and chunky dark shoes. She is theatrical, competitive, expressive and slightly dramatic.  \n\nMaintain the original simplified 2D cartoon/anime aesthetic: bold clean outlines, flat saturated colors, minimal cel shading, exaggerated facial expressions and stylized proportions.  0–3s — Start immediately with the red-haired girl dancing alone in a bright abstract red-and-white graphic space. She performs two exaggerated rhythmic steps and a playful turn, clearly enjoying being the center of attention. Suddenly the blue-haired girl races behind her on roller shoes from left to right. Her enormous cyan twin-tails sweep dramatically through the composition. The red-haired girl freezes for a split second, looks sideways in disbelief, then immediately runs after her.  3–6s — Transition naturally into a fast side-tracking chase. The blue-haired girl glides smoothly ahead, looks back and sticks out her tongue with a mischievous grin. The red-haired girl accelerates and overtakes her from the opposite side. They exchange competitive sideways looks while still moving, then both perform a synchronized hard stop. Their momentum carries directly into a playful dance phrase: step → turn → shoulder hit → opposite-direction spin. Their clothes and hair continue reacting to the previous chase.  6–9s — The rivalry becomes a game of stealing center stage. The red-haired girl slides into the center and begins an exaggerated pose, but the blue-haired girl smoothly rolls across behind her and steals the center position. The red-haired girl immediately circles around and pushes herself back into the composition. They repeatedly weave around each other without physically fighting.  The abstract background reacts to their movement: cyan and navy shapes follow the blue girl, while red and yellow shapes follow the red girl. Stripes, circles, stars, checker patterns, smiley symbols and simple graphic bursts overlap as their paths cross. No written symbols or readable characters.  9–12s — On the strongest musical beat, both suddenly stop back-to-back. The camera arcs smoothly around them as the blue girl’s giant cyan twin-tails swing dramatically toward one side and the red girl’s ribbon-shaped hair spreads toward the opposite side, creating a strong symmetrical red-versus-blue silhouette.  They glance sideways at each other, realize they are perfectly synchronized, then simultaneously turn outward and launch into the same fast choreography: side step → heel hit → shoulder bounce → quick turn → small hop. Their movements match perfectly while their facial performances remain different: the blue girl stays cheeky and mischievous while the red girl performs with exaggerated idol-like confidence.  12–15s — They travel forward together, blue girl gliding and red girl bouncing alongside her, while the red and blue graphic worlds finally merge into one energetic composition of circles, stripes, stars and colorful geometric bursts.  Both slide into a coordinated final pose. For one brief moment everything looks perfectly synchronized. Then the blue-haired girl mischievously bumps the red-haired girl with her shoulder. The red-haired girl stumbles sideways, turns toward her with an exaggerated offended expression, then immediately bumps her back. The blue girl laughs, the red girl breaks into a grin, and both remain slightly off-balance together as the video ends naturally in motion.  \n\nVISUAL STYLE: Strictly match the provided artwork: playful early-2000s-inspired 2D cartoon/anime aesthetic, bold clean outlines, flat saturated colors, simplified anatomy, minimal cel shading and expressive cartoon faces. Do not upgrade them into detailed modern anime characters.  Use strong red-versus-blue graphic composition throughout. Blue girl scenes emphasize cyan, navy, white and small red accents. Red girl scenes emphasize vivid red, white, yellow and black. Gradually combine both palettes as their rivalry becomes synchronized.  Keep environments completely abstract and graphic. Use moving stripes, circles, checker patterns, stars, smiley icons, geometric bursts and flat color fields rather than realistic locations.  \n\nMOTION: Fast, playful and highly connected. Prioritize character interaction over fashion posing. Every action should cause the next: solo dance → roller interruption → chase → synchronized stop → competitive dance → back-to-back synchronization → shared choreography → playful ending.  Use strong 2D key poses, smear frames, overshoot, squash-and-stretch accents and expressive facial animation. Blue girl’s giant twin-tails should create flowing arcs during skating and turns. Red girl’s large hair shapes, suspenders, tie and yellow gloves should bounce rhythmically with her movement.  \n\nCAMERA: Use energetic side tracking, smooth arcs, quick reframing and medium-to-full-body movement. Keep both characters readable whenever they interact. Avoid excessive random cuts and disconnected close-ups.  \n\nAUDIO: Synchronize tightly to bright fast Japanese pop with playful electronic drums and cartoon percussion. Use roller-wheel sounds, quick skids, shoe impacts, fabric swishes, playful pops and rhythmic whooshes.  \n\nIMPORTANT: Both reference characters must remain visually distinct and consistent throughout. Never swap hairstyles, colors, clothing, accessories or body proportions. No hand reaching toward the lens, no palm covering the camera, no outfit transformations, no hairstyle transformations, no extra characters, no photorealism, no 3D CGI, no realistic environments, no modern glossy anime redesign, no cyberpunk effects, no fighting, no disconnected fashion poses, no excessive jump cuts, no facial drift, no uncontrolled morphing. Absolutely no text, typography, letters, numbers, captions, logos, signs or readable writing anywhere in the video.",
    "promptSourceUrl": "https://x.com/Mayz1169/status/2095173829912117558",
    "sourceUrl": "https://x.com/Mayz1169/status/2095173208962150604",
    "sourceLabel": "X 原帖 · @Mayz1169",
    "author": "Kiki",
    "publishedAt": "2026-09-02T15:33:14.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095173208962150604.mp4",
    "posterUrl": "/posters/x/2095173208962150604.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Miku × Teto but make it 2000s cartoon 💙❤️\n\ncreate with MiniMax H3\n\nPrompt below ⤵️",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 13,
      "views": 483,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095172066232447351",
    "title": "猫形特工在地下通道追逐交锋",
    "titleEn": "Cat-Like Agents Clash through an Underground Passage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拟人猫角色在地下通道奔跑、贴身交锋并冲向出口，动作在多个镜头间保持连续。 原生视频规格为 11 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Anthropomorphic cat agents run, grapple, and race toward an exit through an underground passage with connected action across shots. Native-video output: 11s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bakatau8u7/status/2095172066232447351",
    "sourceLabel": "X 原帖 · @bakatau8u7",
    "author": "kataru2000",
    "publishedAt": "2026-09-02T15:28:41.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095172066232447351.mp4",
    "posterUrl": "/posters/x/2095172066232447351.jpg",
    "duration": 11,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "わかったこと。MinimaxH3 Maxは爆速だけど、MinimaxH3のほうが、プロンプトに書ききれてない部分をうまーく忖度して整合性を持たせてくれる。プロンプトはMarkdown形式を使っているけど、日本語より英語の指示の方が余さず読み取ってくれる気がする。むー。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 40,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095171800573608346",
    "title": "女剑士在林间击退忍者包围",
    "titleEn": "A Swordswoman Defeats a Ring of Ninjas in the Forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣女剑士在林间遭忍者围攻，以快速转身和连续挥剑逐个击退敌人。 原生视频规格为 10 秒 · 1280×714。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-clad swordswoman is surrounded by ninjas in a forest and defeats them through quick turns and connected sword strikes. Native-video output: 10s · 1280×714. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Amatellous/status/2095171800573608346",
    "sourceLabel": "X 原帖 · @Amatellous",
    "author": "あまてらす",
    "publishedAt": "2026-09-02T15:27:38.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095171800573608346.mp4",
    "posterUrl": "/posters/x/2095171800573608346.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×714",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "超短編時代劇アニメ。 \nMiniMax H3で生成（キャラと衣装設定はKrea2）。 \n1mp・10秒 / 生成17分 \nRTX 5070 Ti + 32GB RAM",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 39,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095171242374656289",
    "title": "人体骨架驱动多角色姿态变化",
    "titleEn": "A Pose Skeleton Drives Movement across Multiple Characters",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "同一姿态骨架与多张角色图并排展示，角色动作随骨架变化同步调整。 原生视频规格为 8 秒 · 2118×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pose skeleton is displayed beside several character images whose body positions change in step with the tracked movement. Native-video output: 8s · 2118×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jun_kongo/status/2095171242374656289",
    "sourceLabel": "X 原帖 · @jun_kongo",
    "author": "kongo jun / 混合 順",
    "publishedAt": "2026-09-02T15:25:25.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095171242374656289.mp4",
    "posterUrl": "/posters/x/2095171242374656289.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "2118×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 ControlNet openpose",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 78,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095171127454880046",
    "title": "边缘线稿控制多角色动作轮廓",
    "titleEn": "Edge Maps Guide Motion across Multiple Characters",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "角色原图、边缘线稿与生成结果并排变化，用于展示轮廓约束下的动作一致性。 原生视频规格为 8 秒 · 2120×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Character references, edge maps, and generated results change side by side to demonstrate motion consistency under contour guidance. Native-video output: 8s · 2120×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jun_kongo/status/2095171127454880046",
    "sourceLabel": "X 原帖 · @jun_kongo",
    "author": "kongo jun / 混合 順",
    "publishedAt": "2026-09-02T15:24:57.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095171127454880046.mp4",
    "posterUrl": "/posters/x/2095171127454880046.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "2120×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 ControlNet canny",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 66,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095170982654988309",
    "title": "猫耳店主在夜市端出冷蟹拉面",
    "titleEn": "A Cat-Eared Cook Serves Chilled Crab Ramen at Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "猫耳店主在灯笼夜市摊位制作并端出冷蟹拉面，顾客围桌用餐，摊位氛围连续。 原生视频规格为 20 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared cook prepares and serves chilled crab ramen at a lantern-lit night market while diners gather around the stall. Native-video output: 20s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nyannero0828/status/2095170982654988309",
    "sourceLabel": "X 原帖 · @nyannero0828",
    "author": "にゃんねろ | 🎰スロスロメスガキ | 🍭お料理女子🍳",
    "publishedAt": "2026-09-02T15:24:23.000Z",
    "addedAt": "2026-09-02T17:38:00.650Z",
    "mediaUrl": "/media/x-2095170982654988309.mp4",
    "posterUrl": "/posters/x/2095170982654988309.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "food",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "屋台で冷やしカニラーメンを提供するにゃんねろ\n\n #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 42,
      "snapshotAt": "2026-09-02T17:38:00.650Z"
    },
    "approvedAt": "2026-09-02T17:38:00.650Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095651474797641955",
    "title": "数字装配线逐层构建红发角色",
    "titleEn": "A Digital Assembly Line Builds a Red-Haired Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝图、线框与发光扫描层逐步拼合红发角色，最终完成全身造型亮相。 原生视频规格为 15 秒 · 1920×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Blueprints, wireframes, and luminous scan layers assemble a red-haired character piece by piece before the completed full-body reveal. Native-video output: 15s · 1920×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use the provided reference as the final character target.\n\nCreate a short premium cinematic character creation video where the character is built step by step through a high-tech digital assembly process. The sequence should feel fast, intense and visually sophisticated.\n\nShow a clear staged progression. Begin with abstract digital fragments, prototype sketches, blueprint-style drawings, construction lines and technical concept sheets appearing in the background. Then gradually assemble the character from the inside out, with limbs, body parts, clothing and facial features forming through scanning passes, holographic layers, wireframes, segmented image fragments, luminous particles and sleek interface elements.\n\nLet the arms, legs and other body parts visibly generate in stages, as if engineered by an advanced system. Some moments should show unfinished digital anatomy, partial structures and evolving forms before they lock into the final design. Keep the background alive with passing prototype art, silhouette studies, detail callouts and design variations to reinforce the feeling of a real creation process.\n\nThe pacing should be rapid and aggressive, with dense motion, layered depth, sharp transitions and constant visual activity, while keeping the character build clearly readable as the main focus. The overall style should feel futuristic, polished, luxurious and cinematic.\n\nEnd with the fully completed character, cleanly revealed and clearly matching the reference.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2095651474797641955",
    "sourceUrl": "https://x.com/aimikoda/status/2095651474797641955",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-09-03T23:13:41.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095651474797641955.mp4",
    "posterUrl": "/posters/x/2095651474797641955.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "illustration"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 - Prompt Share\n\nGive it a character reference and it turns it into a high-tech digital assembly sequence. You can use parts of the result in your music videos too.\n\nPrompt:\nUse the provided reference as the final character target.\n\nCreate a short premium cinematic character creation video where the character is built step by step through a high-tech digital assembly process. The sequence should feel fast, intense and visually sophisticated.\n\nShow a clear staged progression. Begin with abstract digital fragments, prototype sketches, blueprint-style drawings, construction lines and technical concept sheets appearing in the background. Then gradually assemble the character from the inside out, with limbs, body parts, clothing and facial features forming through scanning passes, holographic layers, wireframes, segmented image fragments, luminous particles and sleek interface elements.\n\nLet the arms, legs and other body parts visibly generate in stages, as if engineered by an advanced system. Some moments should show unfinished digital anatomy, partial structures and evolving forms before they lock into the final design. Keep the background alive with passing prototype art, silhouette studies, detail callouts and design variations to reinforce the feeling of a real creation process.\n\nThe pacing should be rapid and aggressive, with dense motion, layered depth, sharp transitions and constant visual activity, while keeping the character build clearly readable as the main focus. The overall style should feel futuristic, polished, luxurious and cinematic.\n\nEnd with the fully completed character, cleanly revealed and clearly matching the reference.",
    "engagement": {
      "replies": 3,
      "reposts": 5,
      "likes": 64,
      "views": 1330,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095499292534587699",
    "title": "女牛仔在荒漠皮卡旁告别",
    "titleEn": "A Cowgirl Says Goodbye beside a Desert Pickup",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "女牛仔坐在荒漠皮卡车尾与画外人物对话，随后起身走到车旁完成告别。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A cowgirl speaks with an off-screen companion from a pickup bed, then steps down beside the truck to finish the farewell. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, warm western desert pride and intimacy, always seen from the ground-level position of an unseen man. The woman fully matches <Picture 1>: same face, brown cowboy hat, long dark-brown hair, denim jacket, white lace top, ripped jeans, brown boots, old white pickup truck, tailgate, truck-bed interior, rust, taillights, bed proportions, desert ground, distant mountains, bright sky, and clouds. The truck must stay fully locked throughout, especially the tailgate, inner bed walls, rust, rear lights, and bed structure. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural color grading with charcoal black, umber, dark ochre, ash gray, and pale amber subtly interwoven. The camera stays on the ground for the whole video, always looking up at her from the man’s position, never rising to her eye level and never floating. She begins seated on the tailgate exactly as in <Picture 1>. Motion is alive from frame one: hat brim lifting subtly, hair moving as soft separate strands and loose bundles across cheek, neck, and shoulder, cloth edges shifting in the wind. The unseen nearby man (S1) says in an off-screen voiceover: <d>[English] Is all this land yours?</d> while remaining unseen. Still seated, she steadies the hat brim, brushes loose hair away from her cheek and ear, and if there is any eye-line moisture it must move subtly with blinking and head angle, never stay frozen. She smiles and answers downward: <d>[English] Not mine. My family’s.</d> Then she clearly rises from the tailgate into the truck bed through a real action chain: hand bracing, core lifting, boots finding support, torso rising, and one step into the bed. The camera remains at ground level and only follows by tilting upward and shifting slightly. Once standing inside the bed, she looks down at him from above and says: <d>[English] I grew up with every inch of it.</d> She must then clearly move inside the truck bed, not stand still: first one or two light steps toward frame-left while pointing toward the left-side road, land, or slope, then one or two steps toward frame-right while pointing toward the farther right-side desert, horizon, or hills. The low ground-level camera must follow her left and right movement like a real person below tracking her with head turns and small body shifts. During this moving introduction she says: <d>[English] That road, those hills, that dust... that’s home.</d> Then, still above him and still from the ground view, she finishes with deeper warmth and belonging: <d>[English] Out here, you don’t just live on the land. It lives in you.</d> She remains the only visible person. The man is never shown. No frozen hair, no frozen moisture, no floating camera, no equal-height angle, no truck shape drift, no standing still speech, no fashion-shoot energy.\n\noverall_soundscape: Dry desert wind, subtle truck metal creaks, light fabric motion, small boot shifts on the tailgate and truck-bed floor, distant open-land ambience, the man’s ground-level voice, and her relaxed downward replies.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2095499292534587699",
    "sourceUrl": "https://x.com/ou_zhen599/status/2095499292534587699",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-03T13:08:58.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095499292534587699.mp4",
    "posterUrl": "/posters/x/2095499292534587699.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "dialogue",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3\n\n⛰️“Westworld”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, warm western desert pride and intimacy, always seen from the ground-level position of an unseen man. The woman fully matches <Picture 1>: same face, brown cowboy hat, long dark-brown hair, denim jacket, white lace top, ripped jeans, brown boots, old white pickup truck, tailgate, truck-bed interior, rust, taillights, bed proportions, desert ground, distant mountains, bright sky, and clouds. The truck must stay fully locked throughout, especially the tailgate, inner bed walls, rust, rear lights, and bed structure. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural color grading with charcoal black, umber, dark ochre, ash gray, and pale amber subtly interwoven. The camera stays on the ground for the whole video, always looking up at her from the man’s position, never rising to her eye level and never floating. She begins seated on the tailgate exactly as in <Picture 1>. Motion is alive from frame one: hat brim lifting subtly, hair moving as soft separate strands and loose bundles across cheek, neck, and shoulder, cloth edges shifting in the wind. The unseen nearby man (S1) says in an off-screen voiceover: <d>[English] Is all this land yours?</d> while remaining unseen. Still seated, she steadies the hat brim, brushes loose hair away from her cheek and ear, and if there is any eye-line moisture it must move subtly with blinking and head angle, never stay frozen. She smiles and answers downward: <d>[English] Not mine. My family’s.</d> Then she clearly rises from the tailgate into the truck bed through a real action chain: hand bracing, core lifting, boots finding support, torso rising, and one step into the bed. The camera remains at ground level and only follows by tilting upward and shifting slightly. Once standing inside the bed, she looks down at him from above and says: <d>[English] I grew up with every inch of it.</d> She must then clearly move inside the truck bed, not stand still: first one or two light steps toward frame-left while pointing toward the left-side road, land, or slope, then one or two steps toward frame-right while pointing toward the farther right-side desert, horizon, or hills. The low ground-level camera must follow her left and right movement like a real person below tracking her with head turns and small body shifts. During this moving introduction she says: <d>[English] That road, those hills, that dust... that’s home.</d> Then, still above him and still from the ground view, she finishes with deeper warmth and belonging: <d>[English] Out here, you don’t just live on the land. It lives in you.</d> She remains the only visible person. The man is never shown. No frozen hair, no frozen moisture, no floating camera, no equal-height angle, no truck shape drift, no standing still speech, no fashion-shoot energy.\n\noverall_soundscape: Dry desert wind, subtle truck metal creaks, light fabric motion, small boot shifts on the tailgate and truck-bed floor, distant open-land ambience, the man’s ground-level voice, and her relaxed downward replies.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 8,
      "views": 376,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095457286298984916",
    "title": "纹身女骑手驶向落日公路",
    "titleEn": "A Tattooed Rider Leaves for the Sunset Highway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "纹身女骑手在荒漠公路旁启动复古摩托，迎着风驶向开阔山路。 原生视频规格为 14 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A tattooed rider starts a vintage motorcycle beside a desert road and rides into the open mountain highway. Native-video output: 14s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, warm road-movie departure energy. The girl fully matches <Picture 1>: same face, hair color, tattoos, body proportions, black bikini, red vintage motorcycle, transparent front windshield, mirrors, fuel tank, rear luggage, desert road, distant mountains, evening light, and emotional atmosphere. The motorcycle design must stay fully locked throughout, and the transparent windshield must always remain visible and structurally clear. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural color grading with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. The first half is one continuous shot with no cut. The camera starts close from her left-front side and moves gently with soft living motion. Her hair must move as real soft strands and small loose bundles in the wind, never as one stiff block. She looks toward the road with restrained sadness while the mature off-screen man (S1) says in an off-screen voiceover: <d>[English] Where are you going?</d> She lifts a hand, tucks windblown hair behind her ear, lightly wipes the wetness at the edge of one eye, and quietly answers: <d>[English] Los Angeles.</d> Still in the same shot, the camera follows her hands, handlebars, tank, transparent windshield, and body as she grips the bike seriously, rolls the throttle, starts the engine, settles her weight, and turns right onto the open road. The road, mountains, and sunset space open wider as she rides away. At 00:10.800, the only cut of the entire video switches to a frontal riding shot. The camera moves low in front of her while keeping the same motorcycle design locked, especially the transparent windshield and front structure. The riding wind is now much stronger, and her hair must show stronger layered strand motion, with fine strands and loose locks pulled sharply backward. Her expression opens from held emotion into relief and genuine excitement. She smiles, laughs lightly in the wind, and says: <d>[English] I’m coming, Los Angeles.</d> The video ends on hope, freedom, and forward motion. No extra characters, no motorcycle design change, no frozen hair, no car-wash or fashion-commercial feeling.\n\noverall_soundscape: Light wind and open-road ambience at first, then stronger riding wind, soft bike handling sounds, a real engine start, road movement, and natural voice delivery.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2095457286298984916",
    "sourceUrl": "https://x.com/ou_zhen599/status/2095457286298984916",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-03T10:22:03.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095457286298984916.mp4",
    "posterUrl": "/posters/x/2095457286298984916.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "dark",
      "retro"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3\n\n🏍️“Leaving”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic realism, warm road-movie departure energy. The girl fully matches <Picture 1>: same face, hair color, tattoos, body proportions, black bikini, red vintage motorcycle, transparent front windshield, mirrors, fuel tank, rear luggage, desert road, distant mountains, evening light, and emotional atmosphere. The motorcycle design must stay fully locked throughout, and the transparent windshield must always remain visible and structurally clear. IMAX 70mm film grain, inspired by Nolan cinematic aesthetics, photorealistic natural color grading with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. The first half is one continuous shot with no cut. The camera starts close from her left-front side and moves gently with soft living motion. Her hair must move as real soft strands and small loose bundles in the wind, never as one stiff block. She looks toward the road with restrained sadness while the mature off-screen man (S1) says in an off-screen voiceover: <d>[English] Where are you going?</d> She lifts a hand, tucks windblown hair behind her ear, lightly wipes the wetness at the edge of one eye, and quietly answers: <d>[English] Los Angeles.</d> Still in the same shot, the camera follows her hands, handlebars, tank, transparent windshield, and body as she grips the bike seriously, rolls the throttle, starts the engine, settles her weight, and turns right onto the open road. The road, mountains, and sunset space open wider as she rides away. At 00:10.800, the only cut of the entire video switches to a frontal riding shot. The camera moves low in front of her while keeping the same motorcycle design locked, especially the transparent windshield and front structure. The riding wind is now much stronger, and her hair must show stronger layered strand motion, with fine strands and loose locks pulled sharply backward. Her expression opens from held emotion into relief and genuine excitement. She smiles, laughs lightly in the wind, and says: <d>[English] I’m coming, Los Angeles.</d> The video ends on hope, freedom, and forward motion. No extra characters, no motorcycle design change, no frozen hair, no car-wash or fashion-commercial feeling.\n\noverall_soundscape: Light wind and open-road ambience at first, then stronger riding wind, soft bike handling sounds, a real engine start, road movement, and natural voice delivery.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 429,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095632778897568213",
    "title": "法国斗牛犬直立回家开门",
    "titleEn": "A French Bulldog Walks Home and Opens the Door",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "法国斗牛犬像疲惫上班族一样直立行走，提着纸袋摸索钥匙后进入公寓。 原生视频规格为 14 秒 · 480×864。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A French bulldog walks upright like a tired commuter, carries a paper bag, fumbles with keys, and enters an apartment. Native-video output: 14s · 480×864. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A hyper-realistic vertical TikTok video filmed from inside an apartment at night. An adorable real-looking French bulldog walks upright unsteadily on two hind legs like a tired human, carrying a small paper takeout bag in one front paw and fumbling with apartment keys in the other. His entire appearance is that of a normal French bulldog, but his posture and gestures are fully human. He struggles to fit the key into the lock, drops it, bends down while remaining bipedal to pick it up, finally opens the door, walks inside, kicks the door shut behind him with one hind foot and stumbles toward the couch. Real hallway lighting, keys jingling, muffled city noise, shaky phone-camera realism, no music.",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2095632778897568213",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2095632778897568213",
    "sourceLabel": "X 原帖 · @cocktailpeanut",
    "author": "cocktail peanut",
    "publishedAt": "2026-09-03T21:59:24.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095632778897568213.mp4",
    "posterUrl": "/posters/x/2095632778897568213.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I'm constantly amazed at the level of prompt adherence now possible. Just read the entire prompt and compare with the video. It does exactly what I asked it to. Minimax H3.\n\nPrompt: A hyper-realistic vertical TikTok video filmed from inside an apartment at night. An adorable real-looking French bulldog walks upright unsteadily on two hind legs like a tired human, carrying a small paper takeout bag in one front paw and fumbling with apartment keys in the other. His entire appearance is that of a normal French bulldog, but his posture and gestures are fully human. He struggles to fit the key into the lock, drops it, bends down while remaining bipedal to pick it up, finally opens the door, walks inside, kicks the door shut behind him with one hind foot and stumbles toward the couch. Real hallway lighting, keys jingling, muffled city noise, shaky phone-camera realism, no music.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 973,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095625382267408604",
    "title": "篮球少年在动态图形球场扣篮",
    "titleEn": "A Basketball Player Dunks through a Graphic Court",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "篮球少年在鲜艳几何球场中运球、起跳扣篮，俯拍与大字标题连续切换。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A basketball player dribbles and dunks across a vivid geometric court as overhead views and bold title graphics alternate. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2095625382267408604",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-09-03T21:30:00.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095625382267408604.mp4",
    "posterUrl": "/posters/x/2095625382267408604.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "バスケがしたいとお聞きしたので\n\nMiniMax H3で作成しました\n@Hailuo_AI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 12,
      "views": 593,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095527374444003437",
    "title": "钛金属碎片锻造成冲击标题",
    "titleEn": "Titanium Fragments Forge an Impact Title",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "钛金属表面被冲击撕裂成碎片，随后反向聚合为带熔光裂缝的巨型标题。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A titanium surface shatters under impact, then reverses into a monumental title streaked with molten cracks. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Ultra-cinematic Hollywood title sequence, opening on an extreme macro shot of a perfectly smooth monolithic titanium surface suspended in darkness. A microscopic dent suddenly appears at its center, followed by an invisible supersonic impact that violently propagates across the metal, bending the entire surface in a spectacular pressure ripple. The camera rapidly pulls backward as the titanium plate explodes into thousands of heavy fragments, glowing sparks and molten particles. Time dramatically slows while the debris hangs in the air, then a powerful magnetic force reverses every fragment at once. Pieces accelerate back toward the center, colliding with brutal precision and forging the monumental word \"IMPACT\" from dense battle-scarred titanium. The camera sweeps inches above the newly formed letters as glowing energy races through their cracks, then whips into a perfectly centered frontal hero shot. A final invisible shockwave slams into the title, creating a spectacular ripple through the metal before blasting sparks, dust and fragments toward the lens. Photorealistic metal deformation, sophisticated destruction physics, aggressive speed ramps, anamorphic cinematography, premium AAA theatrical title design.",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2095527374444003437",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2095527374444003437",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-09-03T15:00:33.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095527374444003437.mp4",
    "posterUrl": "/posters/x/2095527374444003437.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This is AI motion design in 2026. 🤯\nNo keyframes.\n\nNo simulation setup.\nJust an idea, precise direction and Minimax H3.\nFull prompt below 👇\nUltra-cinematic Hollywood title sequence, opening on an extreme macro shot of a perfectly smooth monolithic titanium surface suspended in darkness. A microscopic dent suddenly appears at its center, followed by an invisible supersonic impact that violently propagates across the metal, bending the entire surface in a spectacular pressure ripple. The camera rapidly pulls backward as the titanium plate explodes into thousands of heavy fragments, glowing sparks and molten particles. Time dramatically slows while the debris hangs in the air, then a powerful magnetic force reverses every fragment at once. Pieces accelerate back toward the center, colliding with brutal precision and forging the monumental word \"IMPACT\" from dense battle-scarred titanium. The camera sweeps inches above the newly formed letters as glowing energy races through their cracks, then whips into a perfectly centered frontal hero shot. A final invisible shockwave slams into the title, creating a spectacular ripple through the metal before blasting sparks, dust and fragments toward the lens. Photorealistic metal deformation, sophisticated destruction physics, aggressive speed ramps, anamorphic cinematography, premium AAA theatrical title design.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 29,
      "views": 1416,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095516988743225637",
    "title": "柠檬少女穿梭明快动效广告",
    "titleEn": "A Lemon-Themed Girl Powers a Bright Motion Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "金发动画少女端着柠檬饮料，在黄色字形、圆环与快速图形转场间完成广告亮相。 原生视频规格为 15 秒 · 2544×1456。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A blonde anime girl presents lemonade through yellow typography, circular motifs, and fast graphic transitions. Native-video output: 15s · 2544×1456. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "LUMI is an elegant, cheerful anime girl with long, vivid lemon-yellow hair with blunt bangs, honey-amber almond eyes with dramatic lashes, soft pink blush, a small mint-green hairpin, dangling lemon-slice earrings, a yellow ruffled off-shoulder top layered over a white halter crop top, and holds a tall glass of sparkling lemonade with ice and lemon slices. Preserve her exact face, proportions, hairstyle, outfit, materials, accessories and colors throughout; never redesign her.\n\nThis film is 80% bold graphic design in motion and 20% character action: giant kinetic typography, hard-edged citrus shapes, flat color fields, sunburst rays, halftone dots, juice-splash graphics, speed lines and shutter flashes. Palette: lemon yellow, citrus gold, cream white, soft blush pink, mint accent. Style: premium AAA motion-graphics title sequence x citrus lifestyle campaign film. Every graphic element moves fast and snaps hard on the beat.\n\nCUT 01 | 0.00–1.00s – Pure graphics: a giant lemon-yellow wordmark \"LUMI\" slams onto a cream field with a sunburst-ray flash; halftone dots pulse and tiny citrus-slice UI ticks flicker along the frame edges.\n\nCUT 02 | 1.00–2.10s – Cream-white field, huge black outline letters \"FRESH\" drop in and bounce; she twirls into frame holding her lemonade glass aloft, ruffles fluttering, sparkle particles trailing behind her.\n\nCUT 03 | 2.10–3.10s – Close shot: she lifts the glass and takes a slow sip, eyes glinting; a fizz of golden bubbles erupts around her as kinetic type \"ZEST\" spins and locks into place behind her shoulder.\n\nCUT 04 | 3.10–4.10s – Flat citrus-yellow field: she spins her glass and lemonade splashes upward in a stylized graphic arc that freezes mid-air as the wordmark \"SPLASH\" bursts through the liquid shape.\n\nCUT 05 | 4.10–5.10s – Hard cut to cream field: giant outlined type \"POUR\" tilts diagonally; she pours a stream of golden lemonade into her glass, the falling stream morphing into the letterforms as it lands.\n\nCUT 06 | 5.10–6.10s – Sunburst radial wipe: she flicks a lemon slice off the rim of her glass toward camera; it spins in slow motion catching the light, while the word \"CITRUS\" scatters into halftone dot particles.\n\nCUT 07 | 6.10–7.10s – Slow hair-flip, her hair rippling like liquid gold; speed lines and sparkle bursts trail from it as the giant type \"GOLDEN\" ripples across the frame in sync with the movement.\n\nCUT 08 | 7.10–8.10s – Kinetic type barrage: SWEET / COOL / BRIGHT / SUNNY slam in one per beat with shutter flashes and camera shake, while she raises her glass in a toast toward camera, ice clinking in slow motion.\n\nCUT 09 | 8.10–9.10s – Cream cyclorama: she tilts her head back with a light laugh, hair catching the wind; a cluster of citrus-slice graphics and sparkle motes swirl around her in a slow orbiting halo.\n\nCUT 10 | 9.10–10.10s – Layered flat-color glass panes (lemon yellow, cream, blush) shatter one by one, each revealing a bigger letter of \"LUMI\" underneath, as she steps forward through the breaking panes.\n\nCUT 11 | 10.10–11.10s – Giant halftone sunburst rays pulse outward from behind her in rhythm with a bass drop, lemon-slice confetti raining down as she smiles and lightly lifts her glass higher.\n\nCUT 12 | 11.10–13.00s – Hero moment on a clean cream cyclorama: she stands center frame, glass raised, warm confident smile, as a shockwave of concentric golden rings, citrus confetti and wind-rippled ruffles blasts outward from her; brief iconic freeze, overexpose to white.\n\nCUT 13 | 13.00–15.00s – Final identity card: an enormous black \"LUMI\" wordmark dominates a pale cream field with translucent lemon-yellow rings, sunburst arcs, halftone dots and a circular emblem containing a lemon-slice motif; she stands relaxed beside the letters, glass in hand, hair rippling in the breeze; one last golden light pulse sweeps through the typography and a soft chime accent punctuates the end.\n\nEditing: aggressive, stylish rhythm — hard cuts on every beat, graphic matches, snap zooms, speed ramps, shutter flashes and impact pulses; every cut must feel compositionally different; typography is always fully readable before the character overlaps it.\n\nNo weapons, no combat, no fire; all energy comes from motion design, light, liquid, glass and citrus.\n\nBGM: bright, uptempo pop/electro-swing with fizzy drops and glitch fills locked to every cut; glass clinks, fizz pops, fabric flutter and sparkle chimes as rhythmic sound-design elements. Peak at CUT 12 and end with a warm chime logo stinger.\n\nPremium AAA quality, anime-inspired cinematic rendering, stylish and vibrant, strong graphic-design identity, consistent character design, exactly 13 cuts",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2095516988743225637",
    "sourceUrl": "https://x.com/Just_sharon7/status/2095516988743225637",
    "sourceLabel": "X 原帖 · @Just_sharon7",
    "author": "Sharon Riley",
    "publishedAt": "2026-09-03T14:19:17.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095516988743225637.mp4",
    "posterUrl": "/posters/x/2095516988743225637.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2544×1456",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "LUMI just turned a lemonade ad into a full-blown visual experience.\n\nMinimax H3 on @TapNow_AI \n\nPrompt\n\nLUMI is an elegant, cheerful anime girl with long, vivid lemon-yellow hair with blunt bangs, honey-amber almond eyes with dramatic lashes, soft pink blush, a small mint-green hairpin, dangling lemon-slice earrings, a yellow ruffled off-shoulder top layered over a white halter crop top, and holds a tall glass of sparkling lemonade with ice and lemon slices. Preserve her exact face, proportions, hairstyle, outfit, materials, accessories and colors throughout; never redesign her.\n\nThis film is 80% bold graphic design in motion and 20% character action: giant kinetic typography, hard-edged citrus shapes, flat color fields, sunburst rays, halftone dots, juice-splash graphics, speed lines and shutter flashes. Palette: lemon yellow, citrus gold, cream white, soft blush pink, mint accent. Style: premium AAA motion-graphics title sequence x citrus lifestyle campaign film. Every graphic element moves fast and snaps hard on the beat.\n\nCUT 01 | 0.00–1.00s – Pure graphics: a giant lemon-yellow wordmark \"LUMI\" slams onto a cream field with a sunburst-ray flash; halftone dots pulse and tiny citrus-slice UI ticks flicker along the frame edges.\n\nCUT 02 | 1.00–2.10s – Cream-white field, huge black outline letters \"FRESH\" drop in and bounce; she twirls into frame holding her lemonade glass aloft, ruffles fluttering, sparkle particles trailing behind her.\n\nCUT 03 | 2.10–3.10s – Close shot: she lifts the glass and takes a slow sip, eyes glinting; a fizz of golden bubbles erupts around her as kinetic type \"ZEST\" spins and locks into place behind her shoulder.\n\nCUT 04 | 3.10–4.10s – Flat citrus-yellow field: she spins her glass and lemonade splashes upward in a stylized graphic arc that freezes mid-air as the wordmark \"SPLASH\" bursts through the liquid shape.\n\nCUT 05 | 4.10–5.10s – Hard cut to cream field: giant outlined type \"POUR\" tilts diagonally; she pours a stream of golden lemonade into her glass, the falling stream morphing into the letterforms as it lands.\n\nCUT 06 | 5.10–6.10s – Sunburst radial wipe: she flicks a lemon slice off the rim of her glass toward camera; it spins in slow motion catching the light, while the word \"CITRUS\" scatters into halftone dot particles.\n\nCUT 07 | 6.10–7.10s – Slow hair-flip, her hair rippling like liquid gold; speed lines and sparkle bursts trail from it as the giant type \"GOLDEN\" ripples across the frame in sync with the movement.\n\nCUT 08 | 7.10–8.10s – Kinetic type barrage: SWEET / COOL / BRIGHT / SUNNY slam in one per beat with shutter flashes and camera shake, while she raises her glass in a toast toward camera, ice clinking in slow motion.\n\nCUT 09 | 8.10–9.10s – Cream cyclorama: she tilts her head back with a light laugh, hair catching the wind; a cluster of citrus-slice graphics and sparkle motes swirl around her in a slow orbiting halo.\n\nCUT 10 | 9.10–10.10s – Layered flat-color glass panes (lemon yellow, cream, blush) shatter one by one, each revealing a bigger letter of \"LUMI\" underneath, as she steps forward through the breaking panes.\n\nCUT 11 | 10.10–11.10s – Giant halftone sunburst rays pulse outward from behind her in rhythm with a bass drop, lemon-slice confetti raining down as she smiles and lightly lifts her glass higher.\n\nCUT 12 | 11.10–13.00s – Hero moment on a clean cream cyclorama: she stands center frame, glass raised, warm confident smile, as a shockwave of concentric golden rings, citrus confetti and wind-rippled ruffles blasts outward from her; brief iconic freeze, overexpose to white.\n\nCUT 13 | 13.00–15.00s – Final identity card: an enormous black \"LUMI\" wordmark dominates a pale cream field with translucent lemon-yellow rings, sunburst arcs, halftone dots and a circular emblem containing a lemon-slice motif; she stands relaxed beside the letters, glass in hand, hair rippling in the breeze; one last golden light pulse sweeps through the typography and a soft chime accent punctuates the end.\n\nEditing: aggressive, stylish rhythm — hard cuts on every beat, graphic matches, snap zooms, speed ramps, shutter flashes and impact pulses; every cut must feel compositionally different; typography is always fully readable before the character overlaps it.\n\nNo weapons, no combat, no fire; all energy comes from motion design, light, liquid, glass and citrus.\n\nBGM: bright, uptempo pop/electro-swing with fizzy drops and glitch fills locked to every cut; glass clinks, fizz pops, fabric flutter and sparkle chimes as rhythmic sound-design elements. Peak at CUT 12 and end with a warm chime logo stinger.\n\nPremium AAA quality, anime-inspired cinematic rendering, stylish and vibrant, strong graphic-design identity, consistent character design, exactly 13 cuts",
    "engagement": {
      "replies": 47,
      "reposts": 9,
      "likes": 244,
      "views": 18385,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095473707611553897",
    "title": "双手从白纸勾勒出梳妆女子",
    "titleEn": "Two Hands Draw a Woman at Her Vanity from Blank Paper",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "铅笔从空白纸面开始勾勒轮廓、五官与梳妆用品，逐步完成带色彩的精细人物画。 原生视频规格为 15 秒 · 2544×1456。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A pencil builds an outline, facial features, and vanity objects from blank paper until a detailed colored portrait is complete. Native-video output: 15s · 2544×1456. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A completely blank sheet of smooth white paper fills the frame. Two hands enter from the bottom. The right hand holds a sharpened graphite pencil; the left hand rests lightly at the lower-left corner of the paper to keep it still.\nThe right hand makes the first mark: a very light, loose oval in the center of the page for the head. Immediately add a faint vertical center line and a horizontal eye-line crossing the middle of the oval. Draw two short marks for the brow line and the base of the nose, then a light curve for the chin. All construction lines stay pale and barely visible.\nNext, block in the neck as two simple sloping lines and a shallow U for the shoulders and collarbones. Sketch a rough rectangle for the torso and the square neckline of the top. Still using only the lightest pressure, place a large circle on the left for the round vanity mirror and a few vertical ovals and cylinders across the bottom for bottles and brush handles. Nothing is detailed yet—only placement.\nNow the face. Darken the eye-line slightly and draw the almond shapes of both eyes, leaving a small highlight circle in each iris. Add the upper lids, a few lashes, and the gentle arch of the brows. Move to the nose: a soft wedge and two small nostril curves. Then the mouth: a short center dip, the full upper lip, and the slight upward curve of the smile. Keep every stroke inside the original oval.\nShift to the hair. Start at the hairline and draw the high, messy half-up bun with the flower-shaped clip on top and the two small barrettes on the right side. Pull long, wavy strands down the left side of the face and over the right shoulder, using loose, overlapping S-curves. Let a few shorter pieces fall across the forehead.\nRefine the clothing: the ribbed texture of the white top, the thin orange trim around the neckline and sleeve, and the tiny bow at the center. Add the thin necklace chain and its small pendant. Darken the contour of the right hand resting against the cheek, fingers relaxed.\nMove to the foreground still life. Outline the gold-framed round mirror on the left, then the cluster of bottles—tall toner bottle, perfume with a faceted cap, pump bottles, and the cup of makeup brushes on the right. Keep these objects slightly lighter so they sit in front of the figure without competing.\nBegin shading. Hatch soft graphite under the chin, along the left side of the nose, beneath the lower lip, and in the hollow of the neck. Build darker value in the hair where strands overlap and around the clips. Add light cross-hatching on the fabric folds of the top and a faint cast shadow from the hand onto the cheek. Leave the paper white for the brightest highlights on the forehead, nose tip, and lip.\nThe left hand slides out of frame. The right hand makes a few final tightening strokes on the eyes and the edges of the hair, then rests the pencil at the top of the paper. The finished drawing is a complete graphite sketch of the blonde woman at her vanity, built entirely from the first faint oval to the last shadow, on otherwise empty white paper.\nSatisfying, calm, close-up process video. Continuous take, no cuts, no extra motion except the hands and pencil. Soft natural light, visible pencil grain and paper tooth.",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2095473707611553897",
    "sourceUrl": "https://x.com/Just_sharon7/status/2095473707611553897",
    "sourceLabel": "X 原帖 · @Just_sharon7",
    "author": "Sharon Riley",
    "publishedAt": "2026-09-03T11:27:18.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095473707611553897.mp4",
    "posterUrl": "/posters/x/2095473707611553897.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2544×1456",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I drew myself with the help of Minimax H3 on @FishCreativeHQ \n\nTry it with your own picture\n\nprompt\n\nA completely blank sheet of smooth white paper fills the frame. Two hands enter from the bottom. The right hand holds a sharpened graphite pencil; the left hand rests lightly at the lower-left corner of the paper to keep it still.\nThe right hand makes the first mark: a very light, loose oval in the center of the page for the head. Immediately add a faint vertical center line and a horizontal eye-line crossing the middle of the oval. Draw two short marks for the brow line and the base of the nose, then a light curve for the chin. All construction lines stay pale and barely visible.\nNext, block in the neck as two simple sloping lines and a shallow U for the shoulders and collarbones. Sketch a rough rectangle for the torso and the square neckline of the top. Still using only the lightest pressure, place a large circle on the left for the round vanity mirror and a few vertical ovals and cylinders across the bottom for bottles and brush handles. Nothing is detailed yet—only placement.\nNow the face. Darken the eye-line slightly and draw the almond shapes of both eyes, leaving a small highlight circle in each iris. Add the upper lids, a few lashes, and the gentle arch of the brows. Move to the nose: a soft wedge and two small nostril curves. Then the mouth: a short center dip, the full upper lip, and the slight upward curve of the smile. Keep every stroke inside the original oval.\nShift to the hair. Start at the hairline and draw the high, messy half-up bun with the flower-shaped clip on top and the two small barrettes on the right side. Pull long, wavy strands down the left side of the face and over the right shoulder, using loose, overlapping S-curves. Let a few shorter pieces fall across the forehead.\nRefine the clothing: the ribbed texture of the white top, the thin orange trim around the neckline and sleeve, and the tiny bow at the center. Add the thin necklace chain and its small pendant. Darken the contour of the right hand resting against the cheek, fingers relaxed.\nMove to the foreground still life. Outline the gold-framed round mirror on the left, then the cluster of bottles—tall toner bottle, perfume with a faceted cap, pump bottles, and the cup of makeup brushes on the right. Keep these objects slightly lighter so they sit in front of the figure without competing.\nBegin shading. Hatch soft graphite under the chin, along the left side of the nose, beneath the lower lip, and in the hollow of the neck. Build darker value in the hair where strands overlap and around the clips. Add light cross-hatching on the fabric folds of the top and a faint cast shadow from the hand onto the cheek. Leave the paper white for the brightest highlights on the forehead, nose tip, and lip.\nThe left hand slides out of frame. The right hand makes a few final tightening strokes on the eyes and the edges of the hair, then rests the pencil at the top of the paper. The finished drawing is a complete graphite sketch of the blonde woman at her vanity, built entirely from the first faint oval to the last shadow, on otherwise empty white paper.\nSatisfying, calm, close-up process video. Continuous take, no cuts, no extra motion except the hands and pencil. Soft natural light, visible pencil grain and paper tooth.",
    "engagement": {
      "replies": 48,
      "reposts": 16,
      "likes": 328,
      "views": 19638,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095602031142859199",
    "title": "同一参考角色穿越多组海滩镜头",
    "titleEn": "One Reference Character Moves through Multiple Beach Shots",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一卷发女子在海滩远景、浪花与近景之间连续切换，用多组镜头检验角色保持。 原生视频规格为 20 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same curly-haired woman moves between wide beach views, breaking waves, and close-ups to test character consistency. Native-video output: 20s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2095602031142859199",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-09-03T19:57:13.000Z",
    "addedAt": "2026-09-04T00:26:06.281Z",
    "mediaUrl": "/media/x-2095602031142859199.mp4",
    "posterUrl": "/posters/x/2095602031142859199.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Using @fal MiniMax H3 Max Ref to Video is great for fast idea visualization.\n\nIt allows you to quickly experiment and pretty cheap. Each of these clips were finished in less than 1 minute for $0.76:",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 14,
      "views": 811,
      "snapshotAt": "2026-09-04T00:26:06.281Z"
    },
    "approvedAt": "2026-09-04T00:26:06.281Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095665444590178354",
    "title": "女孩在雨日窗边端起咖啡",
    "titleEn": "A Woman Raises Her Coffee on a Rainy Day",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "室内女孩坐在窗边端起热咖啡，窗外雨景与杯中蒸汽共同营造安静晨间氛围。 原生视频规格为 7 秒 · 1280×1664。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman lifts a steaming coffee beside a rainy window, creating a quiet morning atmosphere indoors. Native-video output: 7s · 1280×1664. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bangel_19/status/2095665444590178354",
    "sourceLabel": "X 原帖 · @bangel_19",
    "author": "bangel19🇯🇵",
    "publishedAt": "2026-09-04T00:09:12.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095665444590178354.mp4",
    "posterUrl": "/posters/x/2095665444590178354.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1280×1664",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "きょうも雨だねぇ～。\n コーヒー飲みながらコーヒーを飲む動画を作る。😅  \n\n「生成AI（MiniMax H3）を使用しています」",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 18,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095662653364387852",
    "title": "两个版本同屏对比城市与产品镜头",
    "titleEn": "Two Versions Compare City and Product Shots Side by Side",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "分屏展示两个版本生成的城市、车辆与产品构图，并以信息图形总结画面差异。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A split-screen compares city, vehicle, and product compositions from two versions before an infographic summary. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YouArtStudio/status/2095662653364387852",
    "sourceLabel": "X 原帖 · @YouArtStudio",
    "author": "YouArt",
    "publishedAt": "2026-09-03T23:58:06.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095662653364387852.mp4",
    "posterUrl": "/posters/x/2095662653364387852.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MimiMax H3 Max vs Minimax H3. \nI like H3 better, How about you ?",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 2,
      "views": 70,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095662472346923386",
    "title": "烤年糕膨胀成猫后逃跑",
    "titleEn": "A Grilled Mochi Swells into a Cat and Escapes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "烤网上的白色年糕连续膨胀变形成猫，察觉灼热后跳下烤网逃进厨房。 原生视频规格为 10 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A white mochi swells continuously into a cat on the grill, realizes the heat, and bolts into the kitchen. Native-video output: 10s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[ANIMAL] と [FINAL_TAGLINE] を変えれば犬やうさぎでも遊べます。\n===\n[ANIMAL] = 猫（←好きな動物に置き換える）\n[FINAL_TAGLINE] = もちもち、逃走中。（←好きなフレーズに置き換える）\n\n炭火の上の焼き網に置かれた1つの白い餅が、熱でぷくっと膨らみ、その膨らみが連続的に変形しながら、[ANIMAL] の姿になり、最後には焼き網から飛び降りて逃げ出す映像を作成してください。\n\n映像は、黒い金属の焼き網の上に置かれた1つの丸い白餅の極端なクローズアップから始まります。\n\n餅は炭火の熱でゆっくり膨らみます。\n表面には、\n・つやのある焼き餅の皮\n・熱で伸びた薄い膜\n・細かなシワ\n・わずかに半透明な部分\n・香ばしい薄い焼き色\n・小さな焦げ目\n・立ちのぼる湯気\nが見えます。\n\nその膨らみは、別シーンに切り替わるのではなく、そのまま連続的に形を変えていきます。\n\n約2秒後、\n餅の上部や側面から、[ANIMAL] の特徴的なパーツ\n（たとえば耳、鼻先、足、しっぽ、くちばし、角など）\nが、ぷくっと自然に盛り上がるように現れます。\n\n続いて、\n餅そのものが少しずつ変形しながら、\n[ANIMAL] のシルエットへ育っていきます。\n\nこの変形は絶対に別シーンへの切り替えにしないでください。\n同じ餅が、その場で連続的に伸び、膨らみ、折れ曲がりながら、\n自然に [ANIMAL] になる様子を1ショットで見せてください。\n\n完成した姿は、明らかに [ANIMAL] だと分かる一方で、\n身体の素材は最後まで「焼き餅」のままにしてください。\n\n重要:\n普通の毛皮の動物にしないでください。\n身体は白く柔らかな餅の質感、\n香ばしい焼き目、\nぷっくりした弾力、\n少し伸びた焼き餅の皮でできています。\n\n約5秒地点。\n\n完全に [ANIMAL] になった瞬間、\nその生き物が突然ぱちっと目を開きます。\n\n一瞬だけ静止し、\n自分が焼かれていることに気づいたように、\n「熱っ！」というリアクションで身体をビクッと跳ねさせます。\n\nその後、\n焼き網の上で足を小刻みに動かしながら、\n熱そうに「あちち、あちち」と慌てる様子を見せてください。\n\n焼き網が少し揺れ、\n炭火から小さな火の粉が飛び、\n身体から湯気が立ち上ります。\n\n約7秒地点。\n\n焼き餅の [ANIMAL] はカメラの方を一瞬振り返り、\n次の瞬間、\n焼き網から勢いよく飛び降ります。\n\nそのまま卓上や和風の台所、または小さな店先のような空間を、\nものすごい勢いで走って逃げていきます。\n\n走るたびに、\nもちもちした身体が上下にぷるぷる揺れます。\n\nしっぽや耳なども、\n餅らしい弾力で少し揺れます。\n\nただし、\nチーズのように長く糸を引いたり、\nスライムのように溶けたりはしません。\n\n最後の1.5秒。\n\n焼き餅の [ANIMAL] が画面奥へ走り去りながら、\n一度だけこちらを振り返ります。\n\nその瞬間、\n画面下に\n\n「[ANIMAL]」\n\nを読みやすくシンプルに表示してください。\n\nその下に短いコピーとして、\n\n「[FINAL_TAGLINE]」\n\nを表示してください。\n\n最後のフレームは、\n「かわいくて、ちょっとシュールで、思わず笑ってしまう」\n余韻が残る映像にしてください。\n\n全体のトーンは、\n\n温かい\n食欲をそそる\nかわいい\n少しシュール\n映画のようにリアル\n記憶に残る\n\nを両立してください。\n\n# Camera\n\n冒頭:\nextreme macro close-up\n\n変形中:\ncamera remains close and steady,\nallowing the transformation to be clearly visible\n\n逃げ始めた後:\nsmooth low-angle tracking shot\n\n最後:\nslightly wider cinematic framing\n\n基本的に1つの連続した映像として見せてください。\n\n# Important\n\n・餅 → [ANIMAL] の変形は完全に連続的\n・突然の切り替えや瞬間置き換えは禁止\n・変身エフェクトでごまかさない\n・煙で変形を隠しすぎない\n・最後まで焼き餅素材であること\n・[ANIMAL] らしさはシルエットと動きで明確に出す\n・逃げ出す展開を必ず入れる\n・最後に一度振り返らせる\n・最後の文字は最小限で、読みやすく、正しく表示すること\n\n# Style\n\nphotorealistic macro food cinematography,\nedible mochi creature,\nwarm charcoal grill lighting,\nshallow depth of field,\ncinematic Japanese commercial feel,\nrealistic toasted mochi texture,\ncute but believable creature,\nslightly surreal advertisement style,\nhigh-detail steam and browned mochi surface.\n\n# Avoid\n\nordinary furry animal,\nplush toy,\nplastic figurine,\nmarshmallow texture,\nballoon texture,\nslime,\ncheese stretch,\nmonster transformation,\nmagical sparkle transformation,\nhard scene cuts,\njump cuts,\ninstant replacement,\nunreadable text,\nmisspelled display text.",
    "promptSourceUrl": "https://x.com/loglogrog/status/2095662475752755348",
    "sourceUrl": "https://x.com/loglogrog/status/2095662472346923386",
    "sourceLabel": "X 原帖 · @loglogrog",
    "author": "マグマグ",
    "publishedAt": "2026-09-03T23:57:23.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095662472346923386.mp4",
    "posterUrl": "/posters/x/2095662472346923386.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "焼いてた餅、\n「あちっ！」って顔して猫になって逃げました。\n\nfalの「MiniMax H3 Max Turbo」で作った10秒動画。\n毛皮じゃなくて、ちゃんと「焼き餅」の質感のまま変形して逃走するのがシュールで可愛い…笑\n\n生成プロンプト全文はリプ欄に置いておきます👇",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 3,
      "views": 165,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095660932235927734",
    "title": "男子在紫色能量中完成异变",
    "titleEn": "A Man Transforms inside Purple Energy",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "夜色中的男子被紫色能量纹路覆盖，眼睛发光并在面部异变后重新显露全身。 原生视频规格为 17 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Purple energy crawls over a man at night, illuminating his eyes and transforming his face before a full-body reveal. Native-video output: 17s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/b0dre/status/2095660932235927734",
    "sourceLabel": "X 原帖 · @b0dre",
    "author": "Wilson Lora 🇩🇴",
    "publishedAt": "2026-09-03T23:51:16.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095660932235927734.mp4",
    "posterUrl": "/posters/x/2095660932235927734.jpg",
    "duration": 17,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Mede with Minimax H3 Hailuo locally on RTX 5090 on 9m 53s\n\nScene 16s:\nThe second he stopped being human\nTransformation / aura / eyes changing. Hard cut to the peak.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 20,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095657793269252444",
    "title": "未来收音机让女人听见一分钟之后",
    "titleEn": "A Future Radio Lets a Woman Hear One Minute Ahead",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子在昏暗卧室听见收音机播放未来片段，随后独自面对关系破裂的结果。 原生视频规格为 138 秒 · 704×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman hears a fragment from the future through a bedside radio and quietly confronts the outcome of a failing relationship. Native-video output: 138s · 704×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2095657793269252444",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-09-03T23:38:48.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095657793269252444.mp4",
    "posterUrl": "/posters/x/2095657793269252444.jpg",
    "duration": 138,
    "aspectRatio": "portrait",
    "resolution": "704×1280",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "She found a radio that plays sixty seconds into the future. She heard the wrong half of one sentence — and quietly started letting him go.\n\nTHE NEXT MINUTE, a near-future short.\n▶ https://www.youtube.com/watch?v=SDNpTCUb2FM\nMade with MiniMax-H3\n\n#ShortFilm #SciFiShort #AIFilm",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 17,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095657527543521728",
    "title": "三种影视风格快速切换人物对白",
    "titleEn": "Character Dialogue Jumps across Three Screen Styles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短片在雨夜真人、童话动画与成人卡通风格之间快速切换，保持人物对白节奏。 原生视频规格为 20 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The clip jumps among rainy live action, fairy-tale animation, and adult cartoon styles while preserving dialogue rhythm. Native-video output: 20s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/byAmirHanna/status/2095657527543521728",
    "sourceLabel": "X 原帖 · @byAmirHanna",
    "author": "Amir Hanna 🦜",
    "publishedAt": "2026-09-03T23:37:44.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095657527543521728.mp4",
    "posterUrl": "/posters/x/2095657527543521728.jpg",
    "duration": 20,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max needs to be stopped",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 61,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095656390383771823",
    "title": "金发动画少女在室内低声对白",
    "titleEn": "A Blonde Anime Girl Speaks Softly Indoors",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "金发动画少女在室内近景中转头、眨眼并轻声说话，细微表情保持连续。 原生视频规格为 7 秒 · 1664×2496。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde anime girl turns, blinks, and speaks softly in an indoor close-up with continuous subtle expressions. Native-video output: 7s · 1664×2496. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/romfuku/status/2095656390383771823",
    "sourceLabel": "X 原帖 · @romfuku",
    "author": "romfuku",
    "publishedAt": "2026-09-03T23:33:13.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095656390383771823.mp4",
    "posterUrl": "/posters/x/2095656390383771823.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1664×2496",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ギギ・アンダルシアの囁き別バージョンです。こっちの方が顔の動きとかは良かったんですが「大佐」を「ダイサ」と呼んでしまい不採用となりサブアカウントでのポストとしました。Minimax-H3でもLTXほどではないですが漢字の読み間違いはあるのでカナが安定かもですね\n#GigiAndalucia #GundamHathaway",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 85,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095654556457898393",
    "title": "金发动画少女用表情回应镜头",
    "titleEn": "A Blonde Anime Girl Answers with Subtle Expressions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "金发动画少女在竖屏近景里改变视线和嘴型，以细微笑意回应画外对白。 原生视频规格为 7 秒 · 1664×2496。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde anime girl shifts her gaze and mouth shapes in a vertical close-up, answering off-screen dialogue with a faint smile. Native-video output: 7s · 1664×2496. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fumanfuku/status/2095654556457898393",
    "sourceLabel": "X 原帖 · @fumanfuku",
    "author": "不満福",
    "publishedAt": "2026-09-03T23:25:56.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095654556457898393.mp4",
    "posterUrl": "/posters/x/2095654556457898393.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1664×2496",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "閃光のハサウェイ キルケーの魔女のギギ・アンダルシアの囁きのシーンをMinimax-H3のFL2VAで作ってみました。彼女には実際に関係を持たずとも図星を突けるほどの洞察力があるんですかね。え？こんなことまで言ってない？よく聞こえなかったので盛ってしまいました。\n#GigiAndalucia #GundamHathaway",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 10,
      "views": 292,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095652302174740499",
    "title": "银发女孩在红白字形间晨间亮相",
    "titleEn": "A Silver-Haired Girl Appears inside Red-and-White Type",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发角色穿过红白巨型字形与快切构图，最终在舞台中央完成早安主题亮相。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired character moves through giant red-and-white typography before a centered good-morning stage reveal. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ni_naai/status/2095652302174740499",
    "sourceLabel": "X 原帖 · @ni_naai",
    "author": "NinaAI（になあい）",
    "publishedAt": "2026-09-03T23:16:58.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095652302174740499.mp4",
    "posterUrl": "/posters/x/2095652302174740499.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Good Morning☕️\nMiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 19,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095649176529027283",
    "title": "紫发虚拟偶像在霓虹舞台起舞",
    "titleEn": "A Purple-Haired Virtual Idol Dances on a Neon Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫发Q版偶像在霓虹舞台连续摆臂、转身和踩点，完成一段轻快舞蹈。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A chibi purple-haired idol swings her arms, turns, and steps to the beat across a neon performance stage. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Chupico_nora/status/2095649176529027283",
    "sourceLabel": "X 原帖 · @Chupico_nora",
    "author": "Chupico",
    "publishedAt": "2026-09-03T23:04:33.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095649176529027283.mp4",
    "posterUrl": "/posters/x/2095649176529027283.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ノアダンスです🚀sunoで作った曲に合わせてMinimaxH3で踊って貰いました✨ codexくん が材料アップロードして動画生成まで出来るよ❗️と言い出したので試しにお願いしてみました☺️ログインとクレジット管理は自分でやらないとだけどね。結果がこちらです✨",
    "engagement": {
      "replies": 2,
      "reposts": 6,
      "likes": 20,
      "views": 156,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095648030531707119",
    "title": "草莓玩偶从平面海报进入立体世界",
    "titleEn": "A Strawberry Doll Steps from a Poster into a 3D World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "草莓主题玩偶从平面产品海报过渡到柔软立体场景，周围花朵与果实同步展开。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A strawberry-themed doll moves from a flat product poster into a soft 3D world as flowers and fruit unfold around her. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/musitsune/status/2095648030531707119",
    "sourceLabel": "X 原帖 · @musitsune",
    "author": "ツネ｜IP Creative Director",
    "publishedAt": "2026-09-03T23:00:00.000Z",
    "addedAt": "2026-09-04T00:26:06.282Z",
    "mediaUrl": "/media/x-2095648030531707119.mp4",
    "posterUrl": "/posters/x/2095648030531707119.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "静止画をただ動かすのではなく\n奥行きを作り、キャラクターを動かし\n世界そのものを広げていく感じ。🍓\n\nポスターから映像も、なんでもできます。\nたぶん。 #MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 7,
      "views": 491,
      "snapshotAt": "2026-09-04T00:26:06.282Z"
    },
    "approvedAt": "2026-09-04T00:26:06.282Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095641918285983744",
    "title": "螃蟹造型师在舱室为少女做发型",
    "titleEn": "A Crab Stylist Works on a Girl inside a Pod",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卡通少女站在未来舱室里，一只螃蟹挥动工具靠近她并完成夸张造型互动。 原生视频规格为 14 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cartoon girl stands inside a futuristic pod while a crab waves styling tools and performs an exaggerated makeover interaction. Native-video output: 14s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/grmchn4ai/status/2095641918285983744",
    "sourceLabel": "X 原帖 · @grmchn4ai",
    "author": "⚙gear machine@AI",
    "publishedAt": "2026-09-03T22:35:43.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095641918285983744.mp4",
    "posterUrl": "/posters/x/2095641918285983744.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "360度動画、gpt-imageとminimax h3でもいけるのかな、と思って試してみた。できた動画にCodexで「Youtubeの360度メタ情報付けて」ってお願いしてアップしたらドラッグして動くものできたー。ケモノ憑依美容院ASMRです。\nhttps://youtu.be/uciI4NRXRhU",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 8,
      "views": 1372,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095641128624349347",
    "title": "红裙女子触墙唤出记忆剪影",
    "titleEn": "A Woman in Red Awakens Memory Silhouettes from a Wall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红裙女子走入灰色大厅并触碰墙面，裂纹与金色人影随之显现又归于寂静。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in red enters a gray hall and touches the wall, revealing cracks and golden human silhouettes before the room falls still. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2095641128624349347",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-09-03T22:32:34.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095641128624349347.mp4",
    "posterUrl": "/posters/x/2095641128624349347.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Before I fell in love with creating sweet little stories, I often explored a darker, more surreal side of imagination.\n\nIt felt good to visit that place again.\n\nSome rooms preserve our memories. Others keep a piece of us.\n\nThe Room Remembers 🩵\n\n#MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 49,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095640484261527917",
    "title": "花朵伴随动画少女展开晨间片头",
    "titleEn": "Flowers Open around an Anime Girl in a Morning Intro",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "花瓣、向日葵与巨型字形围绕动画少女快速展开，组合成明亮的晨间片头。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Petals, sunflowers, and oversized typography unfold around an anime girl in a bright morning title sequence. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/neko_suno0908/status/2095640484261527917",
    "sourceLabel": "X 原帖 · @neko_suno0908",
    "author": "neko",
    "publishedAt": "2026-09-03T22:30:01.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095640484261527917.mp4",
    "posterUrl": "/posters/x/2095640484261527917.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます☀️\nお花いっぱいの朝です🌸\n\n#MiniMaxH3\n#Capcutcpp\n@capcutapp_jp",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 44,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095636175277134281",
    "title": "抽象人物在几何图形中被指引前进",
    "titleEn": "Abstract Figures Move through Geometric Directions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色人物图标在红蓝圆形与矩形之间移动、伸缩和消失，形成极简动态图形叙事。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Black figure icons move, stretch, and disappear among red and blue circles and rectangles in a minimalist motion narrative. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2095636175277134281",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-09-03T22:12:53.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095636175277134281.mp4",
    "posterUrl": "/posters/x/2095636175277134281.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "2.5D抽象モーショングラフィック｜シュプレマティスム＋シュルレアリスム\n\nなかなかセリフと文字が一致しなくて苦戦。\nちょっとずれたのでタイミングを編集しました！\n\nMinimaxH3で動画作成\n@Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 87,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095636031114744034",
    "title": "长篇动画对白检验粤语与字幕表现",
    "titleEn": "Anime Characters Hold a Cantonese Conversation with Subtitles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多名动画角色在森林等场景持续对白，以三分钟片段检验粤语发音、口型与字幕稳定性。 原生视频规格为 193 秒 · 1944×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Animated characters hold extended conversations across forest scenes in a three-minute test of Cantonese speech, lip sync, and subtitles. Native-video output: 193s · 1944×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hk_nick/status/2095636031114744034",
    "sourceLabel": "X 原帖 · @hk_nick",
    "author": "𝓝𝓲𝓬𝓴🇵🇳葉月暑いね@猫の親バカ！！",
    "publishedAt": "2026-09-03T22:12:19.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095636031114744034.mp4",
    "posterUrl": "/posters/x/2095636031114744034.jpg",
    "duration": 193,
    "aspectRatio": "landscape",
    "resolution": "1944×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "整左足足三日, 就係得呢三分幾鐘...🥲\n啲分鏡真係唔易整\n但衫同臉算跟得幾好, 無咩話甩哂咁\n台詞的話, 只要比足時間佢, 佢係會原封不動照講\n但廣東話就係廢既, 唔係出鄉音就直頭讀另一個字\n字幕都係要自己執\nMinimax H3既字幕, 除非用佢50幾gb既原裝\n否則點整都係出爛字\n條prompt寫得好就少啲穿崩位",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 89,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095635838063309167",
    "title": "绘图软件中的线稿逐步变成彩色肖像",
    "titleEn": "A Digital Sketch Gradually Becomes a Color Portrait",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "绘图软件界面从空白线条开始塑造人物五官，最终转化为色彩鲜艳的完整肖像。 原生视频规格为 15 秒 · 3778×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A drawing interface builds facial features from blank lines and eventually transforms them into a vivid finished portrait. Native-video output: 15s · 3778×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/VisioneerArt/status/2095635838063309167",
    "sourceLabel": "X 原帖 · @VisioneerArt",
    "author": "VISIONEER :: ✨",
    "publishedAt": "2026-09-03T22:11:33.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095635838063309167.mp4",
    "posterUrl": "/posters/x/2095635838063309167.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3778×2160",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": ":: try it out yourself below... time-lapse step-by-step photoshop art demonstration prompt by Kōda @aimikoda :: upload your img & run the prompt in minimax h3 vid gen :: 👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095634392945471836",
    "title": "提灯少女带领镜头穿过夜间街巷",
    "titleEn": "A Lantern-Bearing Girl Leads the Camera through Night Streets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "第一视角跟随提灯少女穿过暖光街巷，沿途人物、拱门和灯饰保持连贯。 原生视频规格为 15 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A first-person camera follows a lantern-bearing girl through warm night streets, passing people, arches, and hanging lights. Native-video output: 15s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chiha_20220301/status/2095634392945471836",
    "sourceLabel": "X 原帖 · @chiha_20220301",
    "author": "千早@AIart",
    "publishedAt": "2026-09-03T22:05:48.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095634392945471836.mp4",
    "posterUrl": "/posters/x/2095634392945471836.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます☕️\n今日は夜のランタン通りを散策します☺️\n#pixversecpp\n@PixVerse\n\nところで最近作っているこのスチームパンク都市の紹介動画はPixVerseのMiniMax H3で生成しているのですが、環境音を細かく指定しても、音が小さかったり、ぼぼ無音だったりするんですが、そういうもん？それとも私のプロンプトが悪いのかしら🤔",
    "engagement": {
      "replies": 3,
      "reposts": 5,
      "likes": 17,
      "views": 233,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095628606483624009",
    "title": "银发偶像在星河舞台完成舞蹈表演",
    "titleEn": "A Silver-Haired Idol Performs on a Galactic Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发虚拟偶像在蓝紫星河舞台挥手、旋转并切换远近景完成舞蹈。 原生视频规格为 18 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired virtual idol waves, spins, and shifts between close and wide shots on a blue-purple galactic stage. Native-video output: 18s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LiluluLoserLab/status/2095628606483624009",
    "sourceLabel": "X 原帖 · @LiluluLoserLab",
    "author": "リルル@Loser Lab",
    "publishedAt": "2026-09-03T21:42:49.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095628606483624009.mp4",
    "posterUrl": "/posters/x/2095628606483624009.jpg",
    "duration": 18,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "IDOL MINA PV MiniMax H3\n今日のチャッピーとの勉強の成果です。\nヽ(*'ω'*)ノ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 68,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095628441291043103",
    "title": "真人街景与动画频道交替出现",
    "titleEn": "Live-Action Streets Alternate with Animated Channels",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "合集在真人街景、童话角色与成人动画频道之间切换，展示本地显卡生成的多种风格。 原生视频规格为 47 秒 · 1934×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A compilation switches among live-action streets, fairy-tale characters, and adult animation to show styles generated on a local GPU. Native-video output: 47s · 1934×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/megaaziib/status/2095628441291043103",
    "sourceLabel": "X 原帖 · @megaaziib",
    "author": "Megaaziib",
    "publishedAt": "2026-09-03T21:42:09.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095628441291043103.mp4",
    "posterUrl": "/posters/x/2095628441291043103.jpg",
    "duration": 47,
    "aspectRatio": "landscape",
    "resolution": "1934×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "this is not actual tv shows. this is all ai videos using minimax h3, generated from my rtx 4060ti 16gb vram only 3 minutes each video. this is insane all this videos comes from my pc and not from datacenters. #comfyui #minimaxh3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 81,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095627950079271105",
    "title": "小螃蟹乘船抵达花朵岛屿",
    "titleEn": "A Tiny Crab Sails toward a Flower Island",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "小螃蟹角色乘船穿过水面，花瓣变形成巨型花朵并在岛屿周围盛开。 原生视频规格为 32 秒 · 1024×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny crab character sails across the water as petals transform into giant flowers blooming around an island. Native-video output: 32s · 1024×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tea_story_hoshi/status/2095627950079271105",
    "sourceLabel": "X 原帖 · @tea_story_hoshi",
    "author": "おちゃばなし | tea_and_stories",
    "publishedAt": "2026-09-03T21:40:12.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095627950079271105.mp4",
    "posterUrl": "/posters/x/2095627950079271105.jpg",
    "duration": 32,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "local minmaxH3検証 / サワガニのクララの船旅🦀\n- 夜に投げておいた1ジョブの結果（無編集） \n- プロンプトを再調整してseedance2.5で一発勝負するためのたたき台🔥\n- 最初のハサミは右手を使って欲しかった✂️🦀\n生成環境の詳細は👇 \n\n■ 環境詳細\n - ComfyUI 0.33.4\n  - PyTorch 2.13.0 + CUDA 12.6\n  - GPU NVIDIA RTX 4060 Ti（VRAM 16GB）\n  - OS Windows（リモート実行、Mac から操作）\n  - ベース MiniMax-H3 Ref2VA\n  - 加速LoRA MiniMax-H3-Ref2VA-Acc-8Step\n  - テキストエンコーダ Qwen3-VL 32B\n  - VAE 映像 fp16 ／ 音声 fp32（別々）\n\n  カスタムノード\n  - ComfyUI-H3-Multishot — マルチショット連鎖の中核\n  - ComfyUI-H3-Multishot-Wordless — 無言版\n  - ComfyUI-MiniMax-H3-PDD-Acc — PDD ヘッドバンクの読み込みと融合\n  - ComfyUI-Spectrum-MiniMax-H3（今回は PDD と併用不可のため OFF）\n  - ComfyUI-KJNodes / ComfyUI-FreeMemory\n\n■ 実測速度（1セグメント＝8秒・アップスケールなし）\n・512×288 ： 32秒 29.6分\n・1024×576： 32秒 76.0分\n\n  #minmaxh3  #サワガニのクララ　#ローカル生成の沼",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "views": 150,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095626511671791835",
    "title": "卡通夫妻在厨房展开一段对话",
    "titleEn": "A Cartoon Couple Exchanges Dialogue in the Kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卡通夫妻在厨房面对面说话，面部表情、手势和场景连续变化。 原生视频规格为 15 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cartoon couple talks face to face in a kitchen while their expressions, gestures, and surroundings remain coherent. Native-video output: 15s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/heathergreen/status/2095626511671791835",
    "sourceLabel": "X 原帖 · @heathergreen",
    "author": "Heather Green",
    "publishedAt": "2026-09-03T21:34:29.000Z",
    "addedAt": "2026-09-04T00:26:06.283Z",
    "mediaUrl": "/media/x-2095626511671791835.mp4",
    "posterUrl": "/posters/x/2095626511671791835.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I don't normally do videos but I saw this prompt share today from @aimikoda and thought I'd give it a try with MiniMax H3 using @magnific",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 82,
      "snapshotAt": "2026-09-04T00:26:06.283Z"
    },
    "approvedAt": "2026-09-04T00:26:06.283Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095806481312444464",
    "title": "插画师在凌乱工作室里为进度发愁",
    "titleEn": "An Illustrator Slumps over a Cluttered Studio Desk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "戴眼镜的插画师在堆满画稿的工作室抽烟、扶额并起身离开电脑，镜头随后拉远展现空荡工位。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bespectacled illustrator smokes, rubs his forehead, and leaves a desk piled with sketches before the camera pulls back across the empty studio. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ivanfioravanti/status/2095806481312444464",
    "sourceLabel": "X 原帖 · @ivanfioravanti",
    "author": "Ivan Fioravanti ᯅ",
    "publishedAt": "2026-09-04T09:29:38.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095806481312444464.mp4",
    "posterUrl": "/posters/x/2095806481312444464.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I can't keep up, there is too much to follow in the AI space nowadays 🤣\n\nminimax-h3-max-turbo image-to-video",
    "engagement": {
      "replies": 7,
      "reposts": 5,
      "likes": 99,
      "views": 7199,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095770682797760857",
    "title": "两位超级英雄在屋顶谈论新模型",
    "titleEn": "Two Superheroes Discuss a New Model on a Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "红蓝制服的两位超级英雄在城市屋顶面对面交谈，并用一问一答制造关于新模型的反转。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two red-and-blue-suited superheroes trade lines on a city rooftop, ending their discussion of a new model with a comic reversal. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Generate a scene of “Spiderman vs Superman” that says, “Hey\", Spiderman says to Superman, ‘Hey, we keep waiting for GPT‑6 to come out,’ and Superman says, ‘Are you? I started using it one week ago.",
    "promptSourceUrl": "https://x.com/ivanfioravanti/status/2095770682797760857",
    "sourceUrl": "https://x.com/ivanfioravanti/status/2095770682797760857",
    "sourceLabel": "X 原帖 · @ivanfioravanti",
    "author": "Ivan Fioravanti ᯅ",
    "publishedAt": "2026-09-04T07:07:22.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095770682797760857.mp4",
    "posterUrl": "/posters/x/2095770682797760857.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max Turbo on @fal is just unbelievable! 10 secs to generate a 15 secs video! I'm blown away!\n\nPrompt: Generate a scene of “Spiderman vs Superman” that says, “Hey\", Spiderman says to Superman, ‘Hey, we keep waiting for GPT‑6 to come out,’ and Superman says, ‘Are you? I started using it one week ago.\n\nGive it a try: https://fal.ai/models/minimax/h3-max-turbo/text-to-video?share=452c3993-fc05-488b-88df-13b336f05290",
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 44,
      "views": 5485,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095769564961464789",
    "title": "黑色汽水罐穿越冷雾与气泡完成定格",
    "titleEn": "A Black Soda Can Moves through Frost and Fizz",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色汽水罐在冷雾中落地，镜头穿过气泡、冰块和飞溅液体，最终回到凝露覆盖的产品定格。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black soda can lands in cold vapor, moves through bubbles, ice, and liquid splashes, then returns to a condensation-covered hero shot. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "\"TITLE: COCA-COLA ZERO SUGAR — FIZZ UNLOCKED\n\nDURATION:\n15 seconds\n\nFORMAT:\nVertical 9:16\n\nFRAME RATE:\n24fps\n\nGENRE:\nPremium cinematic beverage commercial\n\nSTYLE:\nUltra-premium photorealistic product cinematography,\nHollywood commercial quality,\nhigh-speed liquid physics,\nmacro beverage photography,\nrealistic aluminum,\nrealistic condensation,\ncrystal-clear ice,\nvolumetric cold vapor,\ndark cola liquid,\ndeep black studio environment,\nCoca-Cola red highlights,\nsilver edge lighting,\nHDR contrast,\ncinematic depth of field.\n\nNo people.\nNo hands.\nProduct remains the hero throughout.\n\n==================================================\n[REFERENCE CONTROL]\n==================================================\n\nUse the uploaded storyboard sheet ONLY as:\n\n- shot-sequence reference\n- product reference\n- framing reference\n- lighting reference\n- camera-motion reference\n- VFX reference\n- timing reference\n- material reference\n- color reference\n- final-packshot reference\n\nDO NOT animate or display:\n\n- storyboard panels\n- panel borders\n- shot numbers\n- timestamps\n- storyboard sketches\n- arrows\n- technical tables\n- shot-list text\n- bottom production strip\n- black presentation-board UI\n- any labels from the storyboard\n\nReconstruct the board as ONE continuous polished 15-second beverage commercial.\n\nThe final video must look like the finished advertisement represented by the TOP cinematic frames of the storyboard, not a moving storyboard sheet.\n\n==================================================\n[PRODUCT IDENTITY LOCK]\n==================================================\n\nHERO PRODUCT:\n\nOne Coca-Cola Zero Sugar tall aluminum can.\n\nMaintain the exact product appearance from the storyboard/reference:\n\n- tall slim aluminum can\n- predominantly matte black body\n- red Coca-Cola script\n- small white “ZERO SUGAR” text\n- metallic silver can top\n- metallic silver base\n- realistic printed aluminum surface\n- exact can proportions\n- consistent packaging in every shot\n\nCRITICAL:\n\nThe logo must remain:\nstable,\ncorrectly oriented,\nrecognizable,\nand attached to the same physical can.\n\nDo not redesign or morph the can.\n\nDo not duplicate the hero product unless a physically realistic reflection is visible.\n\nNo packaging flicker.\n\nNo typography changing between shots.\n\n==================================================\n[MASTER VISUAL LANGUAGE]\n==================================================\n\nThe commercial is built from five physical materials:\n\nBLACK ALUMINUM\n+\nCONDENSATION\n+\nICE\n+\nCOLA\n+\nCARBONATION\n\nVisual progression:\n\nARRIVAL\n→\nCONDENSATION\n→\nOPENING\n→\nCARBONATION\n→\nICE\n→\nLIQUID ORBIT\n→\nCROWN SPLASH\n→\nFINAL HERO LOCK\n\nThe energy comes from real product materials rather than fantasy magic.\n\nRed illumination can behave like stylized commercial lighting,\nbut it should not become supernatural electricity.\n\n==================================================\n[SHOT 1 — THE COLD ARRIVAL]\n[0:00–0:01.5]\n==================================================\n\nBegin immediately in action.\n\nAlmost completely black premium studio environment.\n\nCamera is extremely low,\nalmost touching a glossy frozen black surface.\n\nFrom darkness above,\nthe Coca-Cola Zero Sugar can drops vertically into frame.\n\nThe can lands perfectly upright.\n\nAt contact:\n\nTHUMP.\n\nA thin skin of ice fractures outward beneath the can.\n\nA circular frost bloom spreads quickly across the glossy surface.\n\nSmall crystal-clear ice fragments jump into the air.\n\nCold vapor bursts outward around the base.\n\nA narrow vertical column of deep red light glows behind the product,\ncreating a premium rim around the black can.\n\nCamera performs a fast but controlled LOW PUSH-IN toward the can.\n\nCondensation begins appearing immediately across the cold aluminum.\n\nDo not tip or deform the can.\n\nEND TRANSITION:\n\nCamera pushes extremely close toward the wet aluminum surface until the can texture fills frame.\n\n==================================================\n[SHOT 2 — CONDENSATION WAKE]\n[0:01.5–0:03.0]\n==================================================\n\nEXTREME MACRO.\n\nThe frame is now almost entirely the curved black aluminum surface.\n\nHundreds of tiny condensation beads sit on the metal.\n\nVery shallow depth of field.\n\nOne larger water droplet forms.\n\nCamera tracks laterally beside the droplet.\n\nThe droplet begins sliding slowly downward across the can.\n\nIts movement creates beautiful magnified reflections of the red Coca-Cola graphics underneath.\n\nA narrow moving red reflection follows the droplet across the wet aluminum.\n\nTiny frost crystals remain near some edges.\n\nLighting:\n\nblack negative space,\ncold silver specular highlights,\nthin red reflection.\n\nThe camera should feel only millimeters away from the packaging.\n\nEND TRANSITION:\n\nAs the droplet reaches a metallic edge,\nMATCH CUT its circular shape into the circular silver top of the can.\n\n==================================================\n[SHOT 3 — THE CRACK]\n[0:03.0–0:04.8]\n==================================================\n\nEXTREME TOP-DOWN MACRO.\n\nThe silver can lid fills most of frame.\n\nRealistic condensation beads cover the cold metal.\n\nThe pull tab lifts cleanly by controlled product animation.\n\nNo visible hand.\n\nThe seal breaks.\n\nSFX:\n\nKSSSHHHH.\n\nA pressurized plume of fine cold white vapor bursts upward.\n\nMicro droplets and tiny carbonation particles explode around the opening.\n\nThe camera gives a quick subtle push toward the opening.\n\nBehind the can,\ndeep red circular reflections expand outward across the black studio surface.\n\nKeep these circles as LIGHT REFLECTIONS,\nnot supernatural energy.\n\nThe vapor catches red and silver highlights.\n\nEND TRANSITION:\n\nCamera pushes directly toward the newly opened mouth of the can.\n\nThe dark opening fills the frame.\n\nDive inside.\n\n==================================================\n[SHOT 4 — INSIDE THE FIZZ]\n[0:04.8–0:06.7]\n==================================================\n\nTRANSITION INTO ABSTRACT MACRO LIQUID WORLD.\n\nCamera now travels through the interior visual language of carbonated cola.\n\nDark rich cola surrounds the lens.\n\nThousands of carbonation bubbles rise rapidly through frame.\n\nBubbles range from:\n\ntiny micro-bubbles\nto\nlarge transparent spherical bubbles.\n\nRuby-red transmitted light shines through the cola.\n\nHighlights distort naturally through bubble surfaces.\n\nCamera moves forward and slightly upward,\nwhile bubbles rise past in the opposite direction.\n\nStrong parallax.\n\nThe scene should feel like travelling through a cinematic carbonation universe.\n\nUse realistic:\n\nrefraction,\nsurface tension,\nbubble deformation,\nliquid viscosity,\ndark cola transparency.\n\nNo product packaging visible here.\n\nEND TRANSITION:\n\nOne large bubble approaches camera.\n\nIts clear spherical surface fills the screen.\n\nThe bubble becomes a crystal-clear ICE FORM through a seamless match transition.\n\n==================================================\n[SHOT 5 — ZERO DEGREE IMPACT]\n[0:06.7–0:08.5]\n==================================================\n\nA large sculptural crystal-clear ice block rotates slowly in a black studio environment.\n\nThrough the ice,\nthe Coca-Cola Zero Sugar can is visible but distorted by realistic refraction.\n\nThe ice is highly detailed:\n\ninternal cracks,\nsmall trapped bubbles,\nsharp translucent edges,\ncold white-blue highlights.\n\nCamera performs a slow forward move.\n\nThen—\n\nCRACK.\n\nThe ice splits into several controlled large pieces.\n\nDo NOT create uncontrolled explosion debris.\n\nThe fragments rotate outward elegantly.\n\nAs they separate,\nthe HERO CAN is revealed perfectly upright in the center.\n\nCold mist flows between the floating pieces.\n\nRed backlight silhouettes the can.\n\nCamera continues smoothly through the opening created by the fractured ice.\n\nEND TRANSITION:\n\nOne rotating ice fragment sweeps past camera,\nrevealing the next wider product composition.\n\n==================================================\n[SHOT 6 — THE RED CURRENT]\n[0:08.5–0:10.5]\n==================================================\n\nWIDE PREMIUM PRODUCT HERO SHOT.\n\nThe Coca-Cola Zero Sugar can stands centered and stable,\nslightly elevated above or perfectly grounded on a glossy black reflective surface.\n\nA broad ribbon of real DARK COLA liquid sweeps around the can.\n\nThe liquid forms a dynamic asymmetric orbit.\n\nIt must behave like real beverage liquid:\n\nweight,\nsurface tension,\ntransparent brown edges,\nruby highlights,\nfine droplets,\nsmall bubbles.\n\nSmall crystal ice fragments travel within and around the liquid ribbon.\n\nA few carbonation bubbles sparkle near the edges.\n\nCamera slowly COUNTER-ORBITS around the can in the opposite direction to the liquid motion.\n\nStrong cinematic parallax.\n\nThe can itself remains completely stable.\n\nKeep branding visible whenever facing camera.\n\nRed highlights move naturally across the liquid.\n\nDo not create a perfect glowing energy ring.\n\nThis is physical cola.\n\nEND TRANSITION:\n\nThe liquid orbit rises slightly,\nthen begins collapsing rapidly downward toward the base of the can.\n\n==================================================\n[SHOT 7 — THE CROWN]\n[0:10.5–0:12.7]\n==================================================\n\nTHREE-QUARTER HERO ANGLE.\n\nThe orbiting cola collapses downward.\n\nIt impacts the glossy surface around the can.\n\nBOOM / LIQUID IMPACT.\n\nA dramatic high-speed CROWN SPLASH erupts around the can.\n\nHuge curved sheets of dark cola rise around the product.\n\nIndividual droplets suspend in the air.\n\nSeveral crystal-clear ice pieces are caught within the splash.\n\nThe product stays:\n\ncentered,\nupright,\nstable,\nreadable.\n\nCamera makes a short controlled PULL-BACK to reveal the entire crown.\n\nUse brief high-speed-commercial slow-motion feeling at the largest splash moment.\n\nRed-white lighting flashes subtly through the moving liquid.\n\nCold vapor mixes near the base.\n\nThen gravity begins pulling the splash back down.\n\nDo not cover the logo for too long.\n\nEND TRANSITION:\n\nThe moving cola sheets fall below frame.\n\nLiquid clears naturally to reveal a pristine final packshot.\n\n==================================================\n[SHOT 8 — HERO LOCK]\n[0:12.7–0:15.0]\n==================================================\n\nCLEAN FINAL PACKSHOT.\n\nCoca-Cola Zero Sugar can stands perfectly upright on glossy black ice.\n\nThe can is covered in dense premium condensation.\n\nTiny droplets move slowly downward.\n\nSmall irregular pieces of clear ice rest around its base.\n\nDark cola has settled into a thin reflective pool.\n\nLow cold mist drifts horizontally near the surface.\n\nBehind the product:\n\ndeep red volumetric glow,\nsoft atmospheric haze,\nrich black background.\n\nUse silver rim lighting to define the can edges.\n\nCamera performs an extremely slow final HERO PUSH-IN.\n\nLeave intentional negative space on one side.\n\nShow clean advertising copy:\n\nZERO SUGAR.\nFULL FIZZ.\n\nBelow in smaller typography:\n\nCOCA-COLA ZERO SUGAR\n\nKeep product dominant.\n\nText must remain sharp,\nstable,\nclean,\nand readable.\n\nNo warped letters.\n\nNo random extra text.\n\nHOLD the final hero image long enough to register.\n\n==================================================\n[CAMERA LANGUAGE]\n==================================================\n\nPremium beverage-commercial cinematography.\n\nSHOT 1:\nextreme low push-in.\n\nSHOT 2:\nmacro lateral track.\n\nSHOT 3:\ntop macro micro-push.\n\nSHOT 4:\nimmersive liquid forward dive.\n\nSHOT 5:\nslow forward push through fractured ice.\n\nSHOT 6:\ncontrolled counter-orbit.\n\nSHOT 7:\nshort pull-back during liquid crown impact.\n\nSHOT 8:\nslow packshot push-in.\n\nCamera movement should be:\n\nprecise,\nsmooth,\ncontrolled,\nexpensive-looking.\n\nAvoid handheld motion.\n\nAvoid random camera shake.\n\nThe only physical impact shake should be extremely subtle during:\n\ncan landing\nand\ncrown-splash impact.\n\n==================================================\n[PRODUCT PHYSICS]\n==================================================\n\nAll materials must behave realistically.\n\nALUMINUM:\nrigid,\ncold,\nmetallic,\nnever bends.\n\nCONDENSATION:\nforms naturally,\nbeads merge,\ngravity moves droplets downward.\n\nICE:\nclear,\ndense,\nrefractive,\nrealistic fracture patterns,\nrealistic weight.\n\nCOLA:\ndark carbonated liquid,\ncorrect surface tension,\nweight,\nviscosity,\nsplash behavior,\nthin transparent edges.\n\nCARBONATION:\nfine bubbles move upward,\nvary naturally in size,\nremain physically plausible.\n\nMIST:\ncold,\nlow-density,\nvolumetric,\nreacts subtly to air movement.\n\nNo jelly-like can.\n\nNo rubbery metal.\n\nNo syrup-thick cola.\n\n==================================================\n[LIGHTING]\n==================================================\n\nMASTER PALETTE:\n\nBLACK\nRED\nSILVER\nDARK COLA BROWN\nCLEAR ICE\n\nLighting should remain coherent throughout.\n\nUse:\n\ndeep black background,\nCoca-Cola red backlights,\nnarrow red strip reflections,\ncold silver edge lights,\nwhite specular highlights,\nsubtle transmitted ruby light through cola.\n\nShot 4 may become more saturated with deep ruby-red transmitted light.\n\nShot 5 emphasizes cold white-blue ice edges.\n\nShot 8 returns to a clean black/red/silver brand world.\n\nNo green.\n\nNo yellow-dominant lighting.\n\nNo multicolor neon.\n\n==================================================\n[SOUND DESIGN]\n==================================================\n\nSound should feel extremely tactile and premium.\n\nSHOT 1:\nfast air displacement,\nmetal landing,\ndeep cinematic impact,\nice cracking,\ncold vapor burst.\n\nSHOT 2:\nnear-silent macro ambience,\ntiny condensation movement,\nsubtle cold air hiss.\n\nSHOT 3:\nsharp aluminum tab SNAP,\npressurized KSSSHHH,\ndense carbonation fizz.\n\nSHOT 4:\nsubmerged liquid whoosh,\nlayers of fine bubbles,\nlow immersive fizz texture.\n\nSHOT 5:\ndeep crystalline ICE CRACK,\nsmall ice-piece collisions,\ncold vapor movement.\n\nSHOT 6:\nwide liquid sweep,\nsubtle bubbles,\nrising bass tone.\n\nSHOT 7:\nmassive but clean beverage splash,\nice clinks,\ndeep impact pulse,\ncarbonation burst.\n\nSHOT 8:\nquiet continuing fizz,\noccasional ice tick,\nsoft mist,\nsound gradually resolves.\n\n==================================================\n[NON-DIEGETIC MUSIC]\n==================================================\n\nUse minimal premium electronic-orchestral commercial scoring.\n\n00:00:\ndeep sub-bass impact on can landing.\n\n00:01.5:\nrestrained rhythmic pulse begins.\n\n00:03:\nmusic briefly opens around the can crack.\n\n00:04.8:\nsub-bass + atmospheric texture builds during carbonation dive.\n\n00:06.7:\nlow cinematic pulse increases.\n\n00:08.5:\ncontrolled rhythm builds with the cola orbit.\n\n00:10.5:\nfull tension peak.\n\n00:11.5:\nlargest impact accent on crown splash.\n\n00:12.7:\nmusic instantly simplifies.\n\nOne clean premium sustained chord remains beneath the hero packshot.\n\nResolve exactly at 00:15.\n\n==================================================\n[MOTION RULES]\n==================================================\n\nDo not fill every second with chaotic motion.\n\nUse contrast:\n\nFAST landing\n→\nSLOW macro\n→\nSHARP opening\n→\nIMMERSIVE movement\n→\nCONTROLLED fracture\n→\nELEGANT orbit\n→\nPOWERFUL splash\n→\nCALM final lock.\n\nEvery movement should have:\n\nclear anticipation,\nclear action,\nclear physical reaction.\n\nAvoid excessive cuts inside individual storyboard shots.\n\n==================================================\n[CONTINUITY RULES]\n==================================================\n\nCRITICAL:\n\nUse ONE hero can.\n\nCan size and proportions remain consistent.\n\nPackaging does not change.\n\nLogo position does not randomly rotate.\n\nCondensation must persist logically after the can becomes cold.\n\nThe environment remains the same black/red premium studio world except for the interior carbonation macro sequence.\n\nSequence must remain:\n\n1. COLD ARRIVAL\n2. CONDENSATION WAKE\n3. THE CRACK\n4. INSIDE THE FIZZ\n5. ZERO DEGREE IMPACT\n6. THE RED CURRENT\n7. THE CROWN\n8. HERO LOCK\n\nDo not reorder.\n\n==================================================\n[FINAL PAYOFF PRIORITY]\n==================================================\n\nShots 7 and 8 are the product payoff.\n\nDo NOT rush the final hero image.\n\nThe crown splash should peak by approximately 00:11.5–00:12.0.\n\nAllow sufficient time for:\n\nsplash to settle\n→\nclean product reveal\n→\ntagline readability\n→\nfinal visual hold.\n\nThe last 2 seconds must feel premium, calm and unmistakably product-focused.\n\n==================================================\n[NEGATIVE CONSTRAINTS]\n==================================================\n\nDo NOT animate:\n\nstoryboard sheet,\npanels,\nsketches,\ntables,\ntitles,\nshot numbers,\ntechnical text,\narrows.\n\nNo people.\nNo hands.\nNo faces.\nNo drinking shot.\n\nNo duplicate cans.\nNo warped cans.\nNo crushed cans.\nNo label morphing.\nNo logo flicker.\nNo random packaging redesign.\n\nNo chocolate.\nNo milk.\nNo green liquid.\nNo green energy.\nNo fantasy electricity.\nNo lightning around product.\n\nNo perfect magical energy rings.\n\nAny circular red forms must read as:\nreflections,\nlighting,\nor physical liquid movement.\n\nNo unrealistic liquid clipping through can.\n\nNo floating ice with impossible physics unless briefly suspended by the splash.\n\nNo cartoon rendering.\nNo low-poly CGI.\nNo plastic-looking metal.\nNo cheap neon aesthetic.\n\nNo unrelated objects.\nNo other brands.\nNo additional logos.\nNo subtitles.\nNo watermark.\n\n==================================================\n[FINAL FRAME LOCK]\n==================================================\n\nFinal frame:\n\nOne Coca-Cola Zero Sugar can,\nperfectly upright,\nblack aluminum body,\nred Coca-Cola branding,\nwhite ZERO SUGAR label,\ndense condensation,\nglossy black ice surface,\nsmall clear ice fragments around base,\nsettled dark-cola reflection,\nlow cold mist,\ndeep red halo behind product,\nsilver rim highlights.\n\nTagline in clean negative space:\n\nZERO SUGAR.\nFULL FIZZ.\n\nSmaller line:\n\nCOCA-COLA ZERO SUGAR\n\nPremium,\nminimal,\ncold,\nrefreshing,\ncinematic.\n\nHOLD TO END.\"",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2095770228257116414",
    "sourceUrl": "https://x.com/ManuAGI01/status/2095769564961464789",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-09-04T07:02:56.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095769564961464789.mp4",
    "posterUrl": "/posters/x/2095769564961464789.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I made this 15-second ads clip using @higgsfield_ai\n \nCreated with @Hailuo_AI's MiniMax H3 \n\n- No reference images.       \n- No animators.       \n- Just prompts\n\nPrompt...👇",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 4,
      "views": 418,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095733073145168360",
    "title": "绘图软件从空白画布塑造红发角色",
    "titleEn": "A Drawing App Builds a Red-Haired Character from Blank Canvas",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "绘图界面从空白画布开始快速建立线稿、上色与阴影，逐步完成红发角色的全身设计。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A drawing interface moves from blank canvas through sketching, color, and shading to complete a full-body red-haired character design. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second fast-paced character creation timelapse inside a professional digital drawing and character design interface.\n\nUse the provided character reference only as the final target for the character’s appearance, proportions, clothing, hairstyle, colors and distinctive features. The process must visibly start from a completely blank canvas and reconstruct the referenced character from scratch.\n\nShow a visible mouse cursor actively controlling the entire creation process. Every major visual change should feel caused by cursor interaction with the UI.\n\n0:00-0:02\nStart on a blank white or neutral canvas inside a clean professional drawing interface. The cursor quickly creates the first loose construction lines, head shape, body proportions and basic pose using rough sketch strokes.\n\n0:02-0:05\nAccelerate into detailed sketching. The cursor rapidly defines the face, hair, anatomy, clothing silhouette, accessories and characteristic features of the referenced character. Use believable drawing gestures, short strokes, selections, transforms and occasional undo corrections.\n\n0:05-0:08\nThe rough sketch becomes clean line art. The cursor switches tools, adjusts brush size and cleans the silhouette. Unnecessary construction lines disappear while precise facial details, clothing folds and small design elements are added.\n\n0:08-0:11\nRapid coloring phase. The cursor selects regions and fills the character with the correct colors from the reference. Show quick palette selections, color picking, layer changes, clipping masks and brush passes. Add shadows, highlights and material details in fast succession.\n\n0:11-0:13\nRendering and polish. The cursor adds final lighting, hair detail, skin shading, fabric texture, reflections and small accent details. Briefly zoom into important areas such as the face or costume while refining them, then zoom back out.\n\n0:13-0:15\nThe interface clears unnecessary guides and panels. The finished character is shown fully rendered and closely matching the provided reference. End with the cursor making one final small adjustment, then stop on a clean full-character presentation.\n\nSTYLE:\nHigh-speed digital art timelapse, polished professional character-design workflow, believable drawing software UI, smooth cursor movement, rapid but readable actions, clean layers and tool changes, no random text, no fake dialogue, no unrelated popups.\n\nMOTION:\nThe entire 15 seconds should feel continuously productive. Avoid long pauses. Compress several minutes or hours of drawing into an energetic timelapse while keeping the progression clearly readable: blank canvas → rough sketch → clean line art → flat colors → shading → polished final character.\n\nCAMERA:\nPrimarily screen-recording style with subtle digital zooms and pans following the cursor and active drawing area. Keep the UI stable and readable. Do not use cinematic cuts outside the software interface.\n\nIMPORTANT:\nThe character must not appear instantly or through a single magical transformation. Every stage should visibly build upon the previous one through cursor-driven drawing, editing, coloring and rendering actions.",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2095733569872375858",
    "sourceUrl": "https://x.com/ManuAGI01/status/2095733073145168360",
    "sourceLabel": "X 原帖 · @ManuAGI01",
    "author": "ManuAGI 🤖 - ( ManuIn )",
    "publishedAt": "2026-09-04T04:37:56.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095733073145168360.mp4",
    "posterUrl": "/posters/x/2095733073145168360.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Gm Friends... ☕️\n\nThis “drawing process” never happened...🤩\n\nOne prompt + your character ref → MiniMax H3 fakes the entire 15-second timelapse from blank canvas to final render.\n\nInspired by @aimikoda \n\nPrompt in the next post...👇",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 3,
      "views": 147,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095753904550191133",
    "title": "彩色人物模块重组为平台品牌落版",
    "titleEn": "Colorful Character Modules Assemble into a Platform Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红蓝人物与立方体模块在白色版面中滑动重组，最后汇聚成简洁的平台品牌落版。 原生视频规格为 13 秒 · 2160×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Red and blue figures and block-like modules slide across a white layout before resolving into a clean platform brand lockup. Native-video output: 13s · 2160×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2095753904550191133",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-09-04T06:00:42.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095753904550191133.mp4",
    "posterUrl": "/posters/x/2095753904550191133.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "2160×1920",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI × モーショングラフィックス\n\nMiniMax H3 on #sousakuAI\n\n架空のAIプラットフォームCM\n\n3Dっぽい質感を動かしてみた",
    "engagement": {
      "replies": 3,
      "reposts": 45,
      "likes": 567,
      "views": 44828,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video.",
    "featured": {
      "at": "2026-09-05T15:50:15.683Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2095715172883345912",
    "title": "极简日文字卡连续翻转构成片头",
    "titleEn": "Minimal Japanese Title Cards Flip into an Intro",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色日文字样在纯白画面中依次出现、放大和切换，最终以模型名称完成极简片头。 原生视频规格为 12 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Black Japanese typography appears, scales, and switches across white frames before a minimal model-name end card. Native-video output: 12s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2095715172883345912",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-09-04T03:26:48.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095715172883345912.mp4",
    "posterUrl": "/posters/x/2095715172883345912.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIモーグラ\n11_AEっぽい動き\n\nSHINJIRO / LOGIC 🇯🇵\n\nMiniMax H3 Max Turbo on #fal",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 29,
      "views": 4359,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095717083401371768",
    "title": "女孩对镜头说出想吃拉面的愿望",
    "titleEn": "A Girl Tells the Camera She Wants Ramen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "竖屏近景中的女孩自然眨眼并对镜头说话，字幕同步呈现她肚子饿、想吃拉面的台词。 原生视频规格为 5 秒 · 768×1376。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A girl blinks and speaks directly to camera in a vertical close-up while subtitles carry her line about being hungry for ramen. Native-video output: 5s · 768×1376. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "girl speaking <d>お腹すいた〜！ラーメン食べたい！</d>",
    "promptSourceUrl": "https://x.com/core_tan/status/2095717083401371768",
    "sourceUrl": "https://x.com/core_tan/status/2095717083401371768",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-09-04T03:34:23.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095717083401371768.mp4",
    "posterUrl": "/posters/x/2095717083401371768.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1376",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI ver1.5 Final2\nAudio Refiner ON/OFF\n\nPrompt:\ngirl speaking <d>お腹すいた〜！ラーメン食べたい！</d>",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 1109,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095714009312010700",
    "title": "吉他手在舞台演唱并接受音频精修",
    "titleEn": "A Guitarist Performs with Refined Audio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "舞台上的吉他手一边弹奏一边演唱，近景持续捕捉嘴型、手部动作和现场灯光。 原生视频规格为 5 秒 · 768×1376。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A guitarist sings while playing onstage as the close-up follows his mouth movement, hands, and concert lighting. Native-video output: 5s · 768×1376. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2095714009312010700",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-09-04T03:22:10.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095714009312010700.mp4",
    "posterUrl": "/posters/x/2095714009312010700.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1376",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI ver1.5 Final2 / Turbo 4STEP \nAudio Refiner ON / OFF\n\n音楽でもバッチシ！Turbo LoRAの欠点が一つ減った。設定は僕がベストだと思った設定一本に決め打ちしてるので、操作はボタンのON/OFFだけです",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 1049,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096025508953932278",
    "title": "雨夜剑客连续迎战多名对手",
    "titleEn": "A Swordswoman Faces Several Opponents in the Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "持刀女子在雨夜庭院中等待围攻，随后快速出刀击倒逼近的黑衣对手。 原生视频规格为 8 秒 · 2752×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A sword-wielding woman waits in a rain-soaked courtyard, then strikes down approaching black-clad opponents in quick succession. Native-video output: 8s · 2752×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aistudynowcom/status/2096025508953932278",
    "sourceLabel": "X 原帖 · @aistudynowcom",
    "author": "Esha Sharma",
    "publishedAt": "2026-09-04T23:59:58.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096025508953932278.mp4",
    "posterUrl": "/posters/x/2096025508953932278.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "2752×1536",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3, full sword fight in the rain, 4 steps.\n\nWatch 0:03. One guy walks in on his own, stops right in front of her, and stands there waiting to be killed. Polite.\n\nThe one behind him is not even moving. Waiting for his turn. There is a queue 💀\n\n#MiniMaxH3 #ComfyUI #AIVideo",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 18,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096021791991026168",
    "title": "背着书包的狸猫在草地上列队前进",
    "titleEn": "Backpack-Wearing Raccoons March across a Field",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一排狸猫背着带有学科字样的书包在草地上前进，镜头随后转向俯拍队列。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A row of raccoons carries backpacks marked with school subjects across a grassy field before the camera shifts overhead. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ito_jo/status/2096021791991026168",
    "sourceLabel": "X 原帖 · @ito_jo",
    "author": "伊藤穣（いとじょ）@跡見 現代文化表現学科",
    "publishedAt": "2026-09-04T23:45:12.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096021791991026168.mp4",
    "posterUrl": "/posters/x/2096021791991026168.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "広報に生成AIをうまく使いたいのだが・・\n\nダメだこんなことしか思いつかない。（MiniMax H3使用）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 16,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096021085896741133",
    "title": "青绿色双马尾角色完成流行转身动作",
    "titleEn": "A Teal Twin-Tailed Character Performs a Trending Spin",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "青绿色双马尾角色在白色背景下抬腿、转身并恢复正面姿势，检验快速旋转时的肢体稳定性。 原生视频规格为 5 秒 · 672×1216。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A teal twin-tailed character lifts a leg, spins, and returns to a front pose against white, testing body stability through a fast turn. Native-video output: 5s · 672×1216. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/funifuni365/status/2096021085896741133",
    "sourceLabel": "X 原帖 · @funifuni365",
    "author": "田中未来｜Miku AI Art",
    "publishedAt": "2026-09-04T23:42:23.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096021085896741133.mp4",
    "posterUrl": "/posters/x/2096021085896741133.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "672×1216",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "一回転する動画が流行ってるようなので\n試してみました\nMiniMaxH3ローカルです",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 28,
      "views": 271,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096017342006489207",
    "title": "Q版棒球选手在球场挥棒击球",
    "titleEn": "A Chibi Baseball Player Swings in a Stadium",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴眼镜的Q版棒球选手在满场观众前站上打击区，挥棒后以彩带和标题完成收尾。 原生视频规格为 15 秒 · 480×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bespectacled chibi batter steps up before a full stadium, swings, and ends on a confetti-filled title card. Native-video output: 15s · 480×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hehehehehe035/status/2096017342006489207",
    "sourceLabel": "X 原帖 · @hehehehehe035",
    "author": "へへへへへ",
    "publishedAt": "2026-09-04T23:27:31.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096017342006489207.mp4",
    "posterUrl": "/posters/x/2096017342006489207.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×832",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます😁\n\n今日は９月５日\n\n『国民栄誉賞の日』\n\n王貞治さんが初めての受賞者なんだそうです\n\nということでMiniMax H3 Max Turboで\n\nパワプロっぽいやつ作ったら中国のパチモンみたいなやつができました😆",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 10,
      "views": 251,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096013365143327152",
    "title": "巨型锦鲤漂浮穿过城市街道",
    "titleEn": "Giant Koi Float through a City Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "鱼眼镜头跟随人物穿过城市路口，数条巨型锦鲤从楼宇之间游过，最终与水母共同悬浮在街道上空。 原生视频规格为 22 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fisheye camera follows a person through a city crossing as giant koi swim between buildings and join a floating jellyfish overhead. Native-video output: 22s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nan_in_space/status/2096013365143327152",
    "sourceLabel": "X 原帖 · @nan_in_space",
    "author": "Nan Ha",
    "publishedAt": "2026-09-04T23:11:42.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096013365143327152.mp4",
    "posterUrl": "/posters/x/2096013365143327152.jpg",
    "duration": 22,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I made the MiniMax H3 launch video for our product @ViggleAI. \n\nProb the first time I've ever built a full launch video end to end with AI. Pretty proud of how the final cut and the aesthetics turned out!!\n\nSharing my 2 cents on the workflow, tool by tool 👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 41,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096012245742268641",
    "title": "失联姐妹被一盏发光台灯重新连接",
    "titleEn": "A Glowing Lamp Reconnects Estranged Sisters",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长篇短片围绕两位失联姐妹展开：床边台灯在对方思念时亮起，引导她们重新面对沉默的关系。 原生视频规格为 138 秒 · 704×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-form short follows two estranged sisters whose bedside lamp glows when the other is awake and missing her, drawing them back toward their broken relationship. Native-video output: 138s · 704×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2096012245742268641",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-09-04T23:07:16.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096012245742268641.mp4",
    "posterUrl": "/posters/x/2096012245742268641.jpg",
    "duration": 138,
    "aspectRatio": "portrait",
    "resolution": "704×1280",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A lamp that only glows while the person you're estranged from lies awake missing you. Two sisters, four years of silence, and a light neither of them could switch off.\nMade with MiniMax-H3\n#ShortFilm #SciFiShort #AIFilm",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 27,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096012074182979692",
    "title": "小动物角色组成轻快舞蹈队形",
    "titleEn": "Small Animal Characters Form a Playful Dance Line",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一群拟人小动物围绕中心角色摆动手臂和身体，在粉色背景前完成轻快群舞。 原生视频规格为 11 秒 · 608×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A group of anthropomorphic animals surrounds the lead character, moving arms and bodies in a playful dance against a pink backdrop. Native-video output: 11s · 608×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/_GAi92/status/2096012074182979692",
    "sourceLabel": "X 原帖 · @_GAi92",
    "author": "_GAi",
    "publishedAt": "2026-09-04T23:06:35.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096012074182979692.mp4",
    "posterUrl": "/posters/x/2096012074182979692.jpg",
    "duration": 11,
    "aspectRatio": "portrait",
    "resolution": "608×1056",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Fuwa dancing with a bunch of rodents.\nI made this a little wile ago, wasn't happy with how it came out.\nBut ehhhh... why sit on it?\nMight as well post it so you can see the progress and growth happening in real time as I work out how to best utilize MiniMax H3.\n\nGuest appearances from @lilykeilani011, @Kur_art_ai and @Shgrr_A.\n#minimaxh3",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 50,
      "views": 556,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096010425997652307",
    "title": "草莓玩偶用多种机位演绎咖啡馆镜头",
    "titleEn": "A Strawberry Doll Demonstrates Café Camera Angles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "草莓装玩偶坐在咖啡馆中，镜头在超广角、低机位和近景之间切换，展示同一场景的不同拍法。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A strawberry-dressed doll sits in a café while the camera switches among extreme wide, low-angle, and close-up views of the same scene. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/musitsune/status/2096010425997652307",
    "sourceLabel": "X 原帖 · @musitsune",
    "author": "ツネ｜IP Creative Director",
    "publishedAt": "2026-09-04T23:00:02.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096010425997652307.mp4",
    "posterUrl": "/posters/x/2096010425997652307.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "同じカフェでも撮り方を変えるだけで\n映画になるカメラワークを体験した。🍓\n\n「何を映すか」だけではなく\n「どう撮るか」で印象は大きく変わる。\n\n映画撮影も、なんでもできます。たぶん。\n#MiniMaxH3 @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 456,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096010423258796488",
    "title": "猫耳女孩在工业场景中完成舞蹈",
    "titleEn": "A Cat-Eared Dancer Performs in an Industrial Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "猫耳女孩在昏暗工业空间连续移步、转身和摆臂，镜头保持全身动作清晰可见。 原生视频规格为 10 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared dancer steps, turns, and swings her arms in a dim industrial room while the full-body framing stays clear. Native-video output: 10s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Chupico_nora/status/2096010423258796488",
    "sourceLabel": "X 原帖 · @Chupico_nora",
    "author": "Chupico",
    "publishedAt": "2026-09-04T23:00:01.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096010423258796488.mp4",
    "posterUrl": "/posters/x/2096010423258796488.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "舞蹈视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "dance",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはよーございます🐺レゾダンスです。実写の方が体重移動とかダンスの動きは安定するみたいですねー。耳がふわふわだぁ\n@capcutapp_jp\n#capcutcpp\n#MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 12,
      "views": 63,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096003193448915042",
    "title": "女学生捧书穿行于回忆式动画片段",
    "titleEn": "A Student Carries a Book through Memory-Like Anime Scenes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴眼镜的女学生捧着红色书本出现在教室与花影之间，字幕和柔和转场组成回忆式音乐片段。 原生视频规格为 38 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bespectacled student holds a red book between classroom and floral scenes as subtitles and soft transitions shape a memory-like music sequence. Native-video output: 38s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kiyoshi_shin/status/2096003193448915042",
    "sourceLabel": "X 原帖 · @kiyoshi_shin",
    "author": "新清士@AIコンテンツ開発者",
    "publishedAt": "2026-09-04T22:31:17.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2096003193448915042.mp4",
    "posterUrl": "/posters/x/2096003193448915042.jpg",
    "duration": 38,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "music-performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "9月5日(土)秋葉原「 #AI夜市 ～なんでもありな同人AI即売会～」で 『田中さんのいる風景』をベースにしたMV「ただいまの練習」をブースで流します。MiniMax H3（ローカル）で作成しました。購入者特典としてフル動画が見れるURLのQRコードをプレゼントします。\nブース机4aです。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 821,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095999374472142894",
    "title": "粉裙女子在花墙前练习仰卧起坐",
    "titleEn": "A Woman in Pink Practices Sit-Ups before a Flower Wall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "身穿粉色裙装的女子躺在花墙前反复起身，短片以简单动作检验人物姿态连续性。 原生视频规格为 5 秒 · 768×1024。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman in a pink dress repeatedly sits up in front of a flower wall, using a simple exercise to test pose continuity. Native-video output: 5s · 768×1024. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "腹筋して",
    "promptSourceUrl": "https://x.com/moon_himiko/status/2095999374472142894",
    "sourceUrl": "https://x.com/moon_himiko/status/2095999374472142894",
    "sourceLabel": "X 原帖 · @moon_himiko",
    "author": "Ai_Himiko",
    "publishedAt": "2026-09-04T22:16:07.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095999374472142894.mp4",
    "posterUrl": "/posters/x/2095999374472142894.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはよう\nみんなMiniMax H3ですごいの作ってるが\nひみこは筋トレ\n\npromptは『腹筋して』\nコストは8円\nみんなも一緒に腹筋しましょー",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 55,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095998677760249969",
    "title": "湖畔森林中浮现隐秘生物的目光",
    "titleEn": "Hidden Creatures Watch from a Lakeside Forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "月色下的湖水与密林保持静谧，树干间逐渐亮起成排眼睛，为数字绘画增添神秘生命感。 原生视频规格为 7 秒 · 1440×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A moonlit lake and dense forest remain still as rows of eyes emerge between the trunks, bringing mysterious life to the digital painting. Native-video output: 7s · 1440×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ramblingpix/status/2095998677760249969",
    "sourceLabel": "X 原帖 · @Ramblingpix",
    "author": "Ramblingpix",
    "publishedAt": "2026-09-04T22:13:21.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095998677760249969.mp4",
    "posterUrl": "/posters/x/2095998677760249969.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1440×1080",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The living forest\n\ndigital painting/MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095994398714515649",
    "title": "校车旁的乐队完成一场户外演奏",
    "titleEn": "A Band Performs beside a School Bus",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长篇音乐视频在乡间校车、户外舞台和吉他演奏之间切换，并以复古数字标题完成收尾。 原生视频规格为 182 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-form music video cuts among a rural school bus, an outdoor stage, and guitar performance before a retro digital title card. Native-video output: 182s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MerchForNobody/status/2095994398714515649",
    "sourceLabel": "X 原帖 · @MerchForNobody",
    "author": "Absolute._.Nobody",
    "publishedAt": "2026-09-04T21:56:21.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095994398714515649.mp4",
    "posterUrl": "/posters/x/2095994398714515649.jpg",
    "duration": 182,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "retro"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "You saw it on Films Ai first (link in comments) 👀\n\n\"Adulting Anthem\" is my first 100% locally generated music video! \nI made this with ComfyUI and MiniMax H3, running entirely on my own GPU:  MSI RTX 5070 with only 12GB of VRAM. \n👉🏽No data center.\n👉🏽 No cloud render farm.\n👉🏽No subscription meter running. \n👉🏽Just my own hardware and power doing the work!\n\nI won't lie, some of these clips took a hot minute to render but the trade-off of owning the whole process, start to finish, was worth every minute of the wait.\n\n(MSI, Nvidia...look what we did with your hardware. We're taking sponsors, just saying. 👀)\n\nA s4lty-nobody Produc7ion. Written and created by A. Nobody.\n\nThis is why local generative AI matters for indie artists.\nIt puts the whole creative process back in our hands instead of renting it from a cloud provider. \n\nIf you're concerned about the environmental toll data centers take, consider donating to help us keep our hardware updated so we can keep generating locally...it's the future of generative AI for the indie artist.\n\n💸Find s4ltybot and I on BuyMeACoffee, links on our profiles...or support us via X money!\n\nThank you for supporting hybrid artists and local generative AI... it means a lot.",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 11,
      "views": 182,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095990911566225431",
    "title": "蓝发少女在书桌前翻阅魔法书",
    "titleEn": "A Blue-Haired Girl Reads a Magic Book at Her Desk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝发动画少女坐在书桌前翻阅发光书本，细微视线、手势和室内光影保持连贯。 原生视频规格为 12 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired anime girl reads a glowing book at her desk with coherent gaze shifts, hand motion, and indoor lighting. Native-video output: 12s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/studio_yebisu/status/2095990911566225431",
    "sourceLabel": "X 原帖 · @studio_yebisu",
    "author": "StudioYebisu",
    "publishedAt": "2026-09-04T21:42:29.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095990911566225431.mp4",
    "posterUrl": "/posters/x/2095990911566225431.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "社区案例",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "私のローカルのMinimax H3のモデルは変わってないはずなのに、Astra君に動画作らせるとRefに忠実で質が高いのはどういうこと…\n\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 56,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095987516713402635",
    "title": "古装人物演示一则月度通信套餐",
    "titleEn": "Period Characters Present a Monthly Mobile Plan",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "古装人物与老人围绕手机展开对话，画面随后切入月亮背景和通信套餐价格页。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Period-costumed characters and elders discuss a phone before the scene shifts to a moonlit backdrop and mobile-plan price card. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ayu_AI_0912/status/2095987516713402635",
    "sourceLabel": "X 原帖 · @Ayu_AI_0912",
    "author": "アユネオ",
    "publishedAt": "2026-09-04T21:29:00.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095987516713402635.mp4",
    "posterUrl": "/posters/x/2095987516713402635.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max Turbo実験中\n――――\n面白おおお！\n\nFable 5.1にお願いして\n生成してもらった！\n\nまだMiniMax H3 Max Turboは\nリファレンスできないっぽいので、\nText to videoで生成しています。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 7,
      "views": 300,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2095985472988430734",
    "title": "像素格斗角色膨胀后击倒对手",
    "titleEn": "A Pixel Fighter Inflates and Knocks Out an Opponent",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "复古街机格斗中，红色角色突然膨胀成巨型形态并发动攻击，最终让对手倒地。 原生视频规格为 10 秒 · 640×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a retro arcade fight, the red character suddenly inflates into a giant form, attacks, and leaves the opponent knocked down. Native-video output: 10s · 640×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chasinglightX64/status/2095985472988430734",
    "sourceLabel": "X 原帖 · @chasinglightX64",
    "author": "chasinglight",
    "publishedAt": "2026-09-04T21:20:52.000Z",
    "addedAt": "2026-09-05T00:12:52.244Z",
    "mediaUrl": "/media/x-2095985472988430734.mp4",
    "posterUrl": "/posters/x/2095985472988430734.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "640×480",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "retro"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Some MiniMax h3 Text-To-Video inflation AI video test",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 59,
      "views": 2268,
      "snapshotAt": "2026-09-05T00:12:52.244Z"
    },
    "approvedAt": "2026-09-05T00:12:52.244Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096262846963396954",
    "title": "雨巷里的贴身格斗",
    "titleEn": "A Close-Quarters Fight in a Rainy Alley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "两名女格斗家在雨夜巷道中完成连续踢击、倒地与近距离对峙，最后以一句克制对白收束。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two women trade grounded kicks in a rain-soaked alley before the action narrows into a restrained face-to-face exchange. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Photoreal live-action rainy alley confrontation between adult Chun-Li and Cammy. Preserve the exact opening composition, identities, hairstyles, costumes, proportions, wet East Asian alley, lanterns, puddles, rainfall, and light positions from <Picture 1>. Continue the airborne kick collision already in progress with no freeze, reset, or new wind-up. Contact shakes rain from skin and fabric; puddle splashes happen only on landing. Follow collision, recoil, and landing. Chun-Li (S1) lands stable, bends her knees, restores torso alignment, exhales sharply, and watches Cammy’s shoulder line. Cammy (S2) lands lower, slides briefly, regains support, and stays pitched forward. [Shot 2] At 00:01.400, the camera cuts to a matched advancing three-quarter view after landing. Cammy wipes rain and a trace of blood from her lip once, looks up, and drives in low. Chun-Li reads shoulder and hip initiation, makes a short evasive adjustment, intercepts with one compact kick, retracts, and adds one short pressing kick that shifts Cammy half a step off line. Show planted support, contact, recoil, and recovery. [Shot 3] At 00:03.800, the camera cuts to a lateral close two-shot on Cammy’s recovery. Cammy presses back into close range without resetting. Chun-Li redirects the entry and takes the centerline. Their faces draw near during the brief check. Chun-Li meets Cammy’s eyes. Cammy glances down and aside for one instant, catches breath, tightens her jaw, and attacks again. Chun-Li’s movements grow shorter and more precise while Cammy’s shoulders tighten and breathing fragments. [Shot 4] At 00:06.500, the camera cuts along Cammy’s renewed forward commitment. She gives half a step, lunges again too early, and places weight too far forward. Chun-Li intercepts the entry, sets her support foot, rotates through her hips, and lands one decisive kick to Cammy’s torso. The camera makes one short controlled whip with the turn, then settles before impact. Cammy recoils diagonally, skids through a puddle, loses support, braces with a forearm, and falls onto hip and side in shallow water. No hovering, no exaggerated flight. Chun-Li retracts, plants, and does not pursue. [Shot 5] At 00:09.800, the camera cuts on Cammy’s hand planting on the wet ground. Cammy raises her upper body with a small shoulder load tremor and keeps a defiant lifted chin. Chun-Li approaches with one or two measured steps, then lowers in a controlled squat until both women’s eyes are level. Cammy checks Chun-Li’s hands, then looks into her eyes. Chun-Li does not touch her. Before 00:12.000, settle into a lateral close two-shot with both eyes and Chun-Li’s mouth readable. [Shot 6] At 00:12.000, the camera cuts seamlessly on the completed squat into the settled close two-shot and holds this angle to the end. Chun-Li (S1), in a mature calm female voice, says: <d>[English] You would rather hate me than tell the truth.</d> Deliver the line evenly, with slight extra weight on “hate me” and a firm landing on “truth,” finishing by about 00:14.500. Cammy (S2) keeps eye contact, takes one audible inward breath as if to answer, parts her lips slightly, but says nothing. Chun-Li adds nothing. Rain continues along eyelashes, cheeks, and jaws; both breathing and tiny balance adjustments remain visible until 15.00 seconds. End on unresolved proximity with no freeze, fade, or reconciliation. Treat all six shots as one continuous confrontation, not separate rounds. No action-showcase fighting, no round reset, no routine reverse coverage, no arbitrary cuts, no showy orbiting, no uncontrolled shake, no blank expressions, no glassy eyes, no exaggerated lip sync, no weightless kicks, no unsupported aerial combos, no extra dialogue, no subtitles, no watermarks, no identity drift, no malformed bodies or hands.\n\noverall_soundscape: Two distinct female vocal identities only: Chun-Li mature and steady, Cammy younger and tighter, expressed only through breath and impact exhalations. Heavy rain, splashing steps, wet fabric friction, short solid body impacts, close breathing, and distant alley ambience remain continuous. Lower the environment slightly before the line while keeping rain audible, then retain Cammy’s unanswered inhale and nearby rainfall after it.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2096262846963396954",
    "sourceUrl": "https://x.com/ou_zhen599/status/2096262846963396954",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-05T15:43:04.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096262846963396954.mp4",
    "posterUrl": "/posters/x/2096262846963396954.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "combat",
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3\n\n⛈️Image-to-Video: “Unspoken in the Rain”\nThe combat in H3 is difficult; you'll need to use Lora to get through it. Character models glitching at high speeds is also a common issue.\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Photoreal live-action rainy alley confrontation between adult Chun-Li and Cammy. Preserve the exact opening composition, identities, hairstyles, costumes, proportions, wet East Asian alley, lanterns, puddles, rainfall, and light positions from <Picture 1>. Continue the airborne kick collision already in progress with no freeze, reset, or new wind-up. Contact shakes rain from skin and fabric; puddle splashes happen only on landing. Follow collision, recoil, and landing. Chun-Li (S1) lands stable, bends her knees, restores torso alignment, exhales sharply, and watches Cammy’s shoulder line. Cammy (S2) lands lower, slides briefly, regains support, and stays pitched forward. [Shot 2] At 00:01.400, the camera cuts to a matched advancing three-quarter view after landing. Cammy wipes rain and a trace of blood from her lip once, looks up, and drives in low. Chun-Li reads shoulder and hip initiation, makes a short evasive adjustment, intercepts with one compact kick, retracts, and adds one short pressing kick that shifts Cammy half a step off line. Show planted support, contact, recoil, and recovery. [Shot 3] At 00:03.800, the camera cuts to a lateral close two-shot on Cammy’s recovery. Cammy presses back into close range without resetting. Chun-Li redirects the entry and takes the centerline. Their faces draw near during the brief check. Chun-Li meets Cammy’s eyes. Cammy glances down and aside for one instant, catches breath, tightens her jaw, and attacks again. Chun-Li’s movements grow shorter and more precise while Cammy’s shoulders tighten and breathing fragments. [Shot 4] At 00:06.500, the camera cuts along Cammy’s renewed forward commitment. She gives half a step, lunges again too early, and places weight too far forward. Chun-Li intercepts the entry, sets her support foot, rotates through her hips, and lands one decisive kick to Cammy’s torso. The camera makes one short controlled whip with the turn, then settles before impact. Cammy recoils diagonally, skids through a puddle, loses support, braces with a forearm, and falls onto hip and side in shallow water. No hovering, no exaggerated flight. Chun-Li retracts, plants, and does not pursue. [Shot 5] At 00:09.800, the camera cuts on Cammy’s hand planting on the wet ground. Cammy raises her upper body with a small shoulder load tremor and keeps a defiant lifted chin. Chun-Li approaches with one or two measured steps, then lowers in a controlled squat until both women’s eyes are level. Cammy checks Chun-Li’s hands, then looks into her eyes. Chun-Li does not touch her. Before 00:12.000, settle into a lateral close two-shot with both eyes and Chun-Li’s mouth readable. [Shot 6] At 00:12.000, the camera cuts seamlessly on the completed squat into the settled close two-shot and holds this angle to the end. Chun-Li (S1), in a mature calm female voice, says: <d>[English] You would rather hate me than tell the truth.</d> Deliver the line evenly, with slight extra weight on “hate me” and a firm landing on “truth,” finishing by about 00:14.500. Cammy (S2) keeps eye contact, takes one audible inward breath as if to answer, parts her lips slightly, but says nothing. Chun-Li adds nothing. Rain continues along eyelashes, cheeks, and jaws; both breathing and tiny balance adjustments remain visible until 15.00 seconds. End on unresolved proximity with no freeze, fade, or reconciliation. Treat all six shots as one continuous confrontation, not separate rounds. No action-showcase fighting, no round reset, no routine reverse coverage, no arbitrary cuts, no showy orbiting, no uncontrolled shake, no blank expressions, no glassy eyes, no exaggerated lip sync, no weightless kicks, no unsupported aerial combos, no extra dialogue, no subtitles, no watermarks, no identity drift, no malformed bodies or hands.\n\noverall_soundscape: Two distinct female vocal identities only: Chun-Li mature and steady, Cammy younger and tighter, expressed only through breath and impact exhalations. Heavy rain, splashing steps, wet fabric friction, short solid body impacts, close breathing, and distant alley ambience remain continuous. Lower the environment slightly before the line while keeping rain audible, then retain Cammy’s unanswered inhale and nearby rainfall after it.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 270,
      "snapshotAt": "2026-09-06T02:44:35.372Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096176539792486835",
    "title": "战士之间的微妙对话",
    "titleEn": "A Subtle Exchange Between Warriors",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名写实奇幻战士在近景中交谈，停顿、表情变化与眼神反应共同推动情绪。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two photoreal fantasy warriors converse in close-up, with pauses, expressions, and eye movements carrying the scene. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CharaspowerAI/status/2096176539792486835",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-09-05T10:00:06.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096176539792486835.mp4",
    "posterUrl": "/posters/x/2096176539792486835.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Don’t forget how good Minimax H3 is at something people still don’t talk about enough:\nVoices + acting. 🎭🔥\n\nDialogue delivery, pauses, expressions, reactions…\nYou can push this model MUCH further than simple image animation.",
    "engagement": {
      "replies": 14,
      "reposts": 6,
      "likes": 88,
      "views": 4634,
      "snapshotAt": "2026-09-06T02:44:35.372Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096420254511751382",
    "title": "盛夏里的冷荞麦面",
    "titleEn": "Cold Soba on a Hot Summer Day",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发动画角色在餐桌前大口吃完冷荞麦面，端起碗喝尽汤汁，动作连贯自然。 原生视频规格为 10 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde anime character finishes a bowl of cold soba, then lifts the bowl to drink the remaining broth in one fluid action. Native-video output: 10s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kozakana51/status/2096420254511751382",
    "sourceLabel": "X 原帖 · @kozakana51",
    "author": "kozakana",
    "publishedAt": "2026-09-06T02:08:32.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096420254511751382.mp4",
    "posterUrl": "/posters/x/2096420254511751382.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "暑さのせいにしてお行儀悪く冷やしぶっかけそばを完食する九堂ミリアムさん\n#AI動画　#大食いガール　#MiniMaxH3練習",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 30,
      "snapshotAt": "2026-09-06T02:44:35.372Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096419150134731258",
    "title": "驶过郊外小站",
    "titleEn": "Passing Through a Rural Station",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "列车驾驶室视角沿轨道前进并穿过一座小型郊外车站，前景操控台与窗外景物保持稳定。 原生视频规格为 15 秒 · 736×992。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A driver-cab view advances along the rails and passes a compact rural station while the console and outside scenery remain coherent. Native-video output: 15s · 736×992. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mix_buchi_/status/2096419150134731258",
    "sourceLabel": "X 原帖 · @mix_buchi_",
    "author": "buchi_",
    "publishedAt": "2026-09-06T02:04:09.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096419150134731258.mp4",
    "posterUrl": "/posters/x/2096419150134731258.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "736×992",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "どこかにありそうな田舎駅。。でもない、隣の駅近っ！ｗ\n\n#MinimaxH3 T2V\n#Comfyui",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 21,
      "snapshotAt": "2026-09-06T02:44:35.372Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096412479307845838",
    "title": "公路热浪的视觉解释",
    "titleEn": "Explaining Heat Shimmer on the Road",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "复古拼贴动画用公路、汽车和分层气流图解太阳加热路面后产生的热浪折射。 原生视频规格为 30 秒 · 608×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A retro collage animation uses a highway, car, and layered airflow diagrams to explain heat shimmer above a sun-heated road. Native-video output: 30s · 608×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Jerryding123/status/2096412479307845838",
    "sourceLabel": "X 原帖 · @Jerryding123",
    "author": "Jerry Ding",
    "publishedAt": "2026-09-06T01:37:39.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096412479307845838.mp4",
    "posterUrl": "/posters/x/2096412479307845838.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "608×1056",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "motion-graphics",
      "retro"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Maybe it is now my life mission to bankrupt liberals at Vox.\n\nDone locally with Minimax H3 on 1xRTX Pro 6000, normal Codex harness no Astra or anything fancy. Transitions are absolutely incredible. Narrated with Chatterbox.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 93,
      "snapshotAt": "2026-09-06T02:44:35.372Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096408498821365816",
    "title": "水枪迎面喷射",
    "titleEn": "A Water Blaster Fired at the Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "粉发动画角色举起彩色水枪朝镜头喷射，水流与面部表情在近景中持续变化。 原生视频规格为 5 秒 · 832×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired anime character raises a colorful water blaster and fires toward the camera as the water stream and expression evolve in close-up. Native-video output: 5s · 832×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2096408498821365816",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-09-06T01:21:50.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096408498821365816.mp4",
    "posterUrl": "/posters/x/2096408498821365816.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 I2V\n\nここまでI2Vの出来がいいんだから、なんか出来るっぽい？フレームレート設定も検討したいなぁ。24fpsだとアニメっぽくはどうやってもならないし計算量も削減出来たらプラスでしかない",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 5,
      "views": 208,
      "snapshotAt": "2026-09-06T02:44:35.372Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096400072451748338",
    "title": "工地上的石匠寓言",
    "titleEn": "The Stonecutter Parable on a Construction Site",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏短片把石匠寓言搬到现代工地，通过工人、高楼与石匠的连续场景讲述职业羡慕与价值。 原生视频规格为 75 秒 · 704×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical short film retells the stonecutter parable on a modern construction site through workers, a rising tower, and the mason supporting it. Native-video output: 75s · 704×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NsitnovSitnov/status/2096400072451748338",
    "sourceLabel": "X 原帖 · @NsitnovSitnov",
    "author": "Nedelcho Sitnov",
    "publishedAt": "2026-09-06T00:48:21.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096400072451748338.mp4",
    "posterUrl": "/posters/x/2096400072451748338.jpg",
    "duration": 75,
    "aspectRatio": "portrait",
    "resolution": "704×1280",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "She envied every job on the site except the one that held the whole tower up. The stonecutter's parable, retold on a Sofia build site.\n\n▶ Watch the full film: https://www.youtube.com/watch?v=JLhYq8Kp1tY\nMade with MiniMax-H3\n#ShortFilm #Parable #AIFilm",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 30,
      "snapshotAt": "2026-09-06T02:44:35.372Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096398628348977187",
    "title": "居家完成一次引体向上",
    "titleEn": "Completing a Pull-Up at Home",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名女性在居家训练架上从悬垂姿态完成引体向上，身体与手臂动作清晰可见。 原生视频规格为 5 秒 · 768×1088。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman moves from a dead hang into a pull-up on a home training bar, with the arm and body motion clearly visible. Native-video output: 5s · 768×1088. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "懸垂して",
    "promptSourceUrl": "https://x.com/moon_himiko/status/2096398628348977187",
    "sourceUrl": "https://x.com/moon_himiko/status/2096398628348977187",
    "sourceLabel": "X 原帖 · @moon_himiko",
    "author": "Ai_Himiko",
    "publishedAt": "2026-09-06T00:42:36.000Z",
    "addedAt": "2026-09-06T02:44:35.372Z",
    "mediaUrl": "/media/x-2096398628348977187.mp4",
    "posterUrl": "/posters/x/2096398628348977187.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1088",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal",
      "vertical"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはよう (⋈◍＞◡＜◍)。✧♡\nみんなMiniMax H3でどんどんクオリティ高い作品作ってってすごい！！！\nそれでも、ひみこは今日も変わらず筋トレ \npromptは『懸垂して』 コストは8円\nこの料金も今日までかな？\n みんなも一緒に健康になりましょう！！！",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 80,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.372Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096391506651128264",
    "title": "沉船旁的骷髅潜水员",
    "titleEn": "A Skeleton Diver Beside a Shipwreck",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "骷髅穿着潜水装备游过布满海洋附着物的沉船残骸，蓝绿色水下环境保持连贯。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A skeleton in scuba gear swims past an encrusted shipwreck in a coherent blue-green underwater environment. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxCasu/status/2096391506651128264",
    "sourceLabel": "X 原帖 · @MaxCasu",
    "author": "max casu",
    "publishedAt": "2026-09-06T00:14:18.000Z",
    "addedAt": "2026-09-06T02:44:35.373Z",
    "mediaUrl": "/media/x-2096391506651128264.mp4",
    "posterUrl": "/posters/x/2096391506651128264.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "dark"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Underwater skeleton \nMade with @ViggleAI MiniMax H3 model\n#ViggleH3",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 6,
      "views": 75,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.373Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096390580763013389",
    "title": "哥布林一脚击倒冒险者",
    "titleEn": "A Goblin Kick Drops an Adventurer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "奇幻森林中，哥布林围住冒险者并发动踢击，随后镜头切到受伤角色与发光武器。 原生视频规格为 12 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a dark fantasy forest, goblins surround and kick an adventurer before the scene cuts to the injured hero and a glowing weapon. Native-video output: 12s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/6s9Mu3Y3ez4c6Jl/status/2096390580763013389",
    "sourceLabel": "X 原帖 · @6s9Mu3Y3ez4c6Jl",
    "author": "木沢　真流",
    "publishedAt": "2026-09-06T00:10:38.000Z",
    "addedAt": "2026-09-06T02:44:35.373Z",
    "mediaUrl": "/media/x-2096390580763013389.mp4",
    "posterUrl": "/posters/x/2096390580763013389.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "防御力-9999の冒険者にとってみれば、ゴブリンのキックが致命傷に！\n\n#anime #minimaxh3 #アニメ　#異世界漫画",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 29,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.373Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096388781876322701",
    "title": "耳勺直播的循环待机动画",
    "titleEn": "A Looping Ear-Cleaning Stream Intro",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发虚拟角色面向镜头缓慢移动耳勺，形成适合直播等待画面的短循环动画。 原生视频规格为 3 秒 · 1944×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired virtual character slowly moves an ear pick toward the camera in a short loop designed like a stream waiting screen. Native-video output: 3s · 1944×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hirosi1971/status/2096388781876322701",
    "sourceLabel": "X 原帖 · @hirosi1971",
    "author": "ばら＠α7iiiとAI修行中",
    "publishedAt": "2026-09-06T00:03:29.000Z",
    "addedAt": "2026-09-06T02:44:35.373Z",
    "mediaUrl": "/media/x-2096388781876322701.mp4",
    "posterUrl": "/posters/x/2096388781876322701.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "1944×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "耳かきASMR配信のりんねおねんねさん\n配信待機のループアニメーション風\n#おねんねあーと\n#MiniMaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 4,
      "views": 314,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.373Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096387910740144518",
    "title": "雨后独行的动画少女",
    "titleEn": "An Anime Girl Walking After the Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多镜头竖屏动画从积水中的鞋步切到云层与撑伞少女，组成一段安静的雨后行走片段。 原生视频规格为 30 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical multi-shot animation moves from shoes stepping through a puddle to clouds and a girl carrying an umbrella after the rain. Native-video output: 30s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/zyzel_zyz/status/2096387910740144518",
    "sourceLabel": "X 原帖 · @zyzel_zyz",
    "author": "ざいぜる",
    "publishedAt": "2026-09-06T00:00:01.000Z",
    "addedAt": "2026-09-06T02:44:35.373Z",
    "mediaUrl": "/media/x-2096387910740144518.mp4",
    "posterUrl": "/posters/x/2096387910740144518.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GPT-6 Astra＋ローカルAIで、30秒7カットのアニメを制作\n\n① コンセプトを元にラフ案をAstraが作成\n② AstraがComfyUIを操作\n・Animaでキャラと構図\n・ACE-StepでBGM\n・MiniMax H3でシーン別動画\n④ ボクは出力を見て「ここはこうしたい」とチェック＆ディレクション\n⑤ Astraが都度修正\n⑥ AstraがOpenCVやFFmpegでつなぎ・色・音を仕上げ\n\n指示はすべて自然言語\n人にお願いするのと同じようにAstraにお願いした",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 258,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.373Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096386367144284533",
    "title": "看台上睡着的小球迷",
    "titleEn": "The Littlest Fan Falls Asleep",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拟人猫咪一家穿着球队服装走过球场看台，孩子最终戴着头盔在父亲怀里睡着。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anthropomorphic cat family walks through a stadium concourse before the helmeted child falls asleep in a parent’s arms. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2096386367144284533",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-09-05T23:53:53.000Z",
    "addedAt": "2026-09-06T02:44:35.373Z",
    "mediaUrl": "/media/x-2096386367144284533.mp4",
    "posterUrl": "/posters/x/2096386367144284533.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "commercial"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "He cheered through every quarter…but bedtime won in overtime. 🥹🏈🐾\n\nGood night from the littlest football fan. 💙\n\nAnimated with MiniMax H3 on @PixVerse   #pixversecpp",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 9,
      "views": 145,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.373Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096379851779575911",
    "title": "赛博战士的扫描对决",
    "titleEn": "A Scan-Off Between Cyber Warriors",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名赛博装甲角色在工业场景中相互扫描、对峙并准备战斗，字幕与特写强化冲突。 原生视频规格为 38 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two cyber-armored characters scan and confront one another in an industrial setting as captions and close-ups build toward combat. Native-video output: 38s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JasminaAi/status/2096379851779575911",
    "sourceLabel": "X 原帖 · @JasminaAi",
    "author": "ジャスミーナ | AI Anime & Illustrations✨",
    "publishedAt": "2026-09-05T23:28:00.000Z",
    "addedAt": "2026-09-06T02:44:35.373Z",
    "mediaUrl": "/media/x-2096379851779575911.mp4",
    "posterUrl": "/posters/x/2096379851779575911.jpg",
    "duration": 38,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "互いをスキャン解析するジャスミーナとネフェレイラ\n\n「出力、速度、共に私より上……っ？！」\n\n上位スペックの強敵を相手に\nジャスミーナは危機を打破することができるのか……⁉️\n@Hailuo_AI\n#MiniMaxDesign #MiniMaxH3\nhttps://x.com/JasminaAi/status/2096012431395082445?s=20",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 24,
      "views": 885,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.373Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096372813699416474",
    "title": "猫耳角色的竖屏舞步",
    "titleEn": "A Cat-Eared Character’s Vertical Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发猫耳角色在明亮舞台上完成连续舞步与手势，镜头在全身和近景之间切换。 原生视频规格为 10 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired cat-eared character performs a continuous dance and hand gestures on a bright stage, alternating between full-body and close-up views. Native-video output: 10s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Chupico_nora/status/2096372813699416474",
    "sourceLabel": "X 原帖 · @Chupico_nora",
    "author": "Chupico",
    "publishedAt": "2026-09-05T23:00:02.000Z",
    "addedAt": "2026-09-06T02:44:35.373Z",
    "mediaUrl": "/media/x-2096372813699416474.mp4",
    "posterUrl": "/posters/x/2096372813699416474.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "cg-3d",
      "vertical"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはよーございます💜ノラダンスです✨CapCut のAIで作成しました！Astraはiphoneには来たけどcodexにはまだ来てくれてないのでソワソワしてます_(* ॑꒳ ॑*_ )\n@capcutapp_jp\n#capcutcpp\n#MiniMaxH3",
    "engagement": {
      "replies": 7,
      "reposts": 11,
      "likes": 44,
      "views": 252,
      "snapshotAt": "2026-09-06T02:44:35.373Z"
    },
    "approvedAt": "2026-09-06T02:44:35.373Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096378308753801232",
    "title": "静谧京都酒店广告",
    "titleEn": "A Quiet Kyoto Hotel Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "酒店外观、精致料理与留白排版组成一支克制的京都旅宿广告，镜头和文字衔接清晰。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A restrained Kyoto hotel commercial moves from the property exterior to refined dining and spacious typography with clean transitions. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2096378308753801232",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-09-05T23:21:52.000Z",
    "addedAt": "2026-09-06T02:48:51.413Z",
    "mediaUrl": "/media/x-2096378308753801232.mp4",
    "posterUrl": "/posters/x/2096378308753801232.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Codex (Astra) x MiniMax H3 x Remotion x Irodori\n\n前回は文字が出なくて、ナレーションも微妙だったので、\n文字はRemotion\nナレーションはIrodori\nを使用\n\n画像の動画化は今回もMiniMax H3を使用\n\n全体的にまとまりが良くなりましたね\n\nw2v（WEB to Video）システムが構築されつつある\n\nTopviewのMCPでMiniMax H3の生成をしているので、システムはCodex内で完結\n\nBGMはTuneeAIで生成しました　#TuneeAI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 14,
      "views": 819,
      "snapshotAt": "2026-09-06T02:48:51.413Z"
    },
    "approvedAt": "2026-09-06T02:48:51.413Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096496980180500694",
    "title": "绿色时尚海报里的横向追逐",
    "titleEn": "A Side-Scrolling Chase Through a Green Fashion Poster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画女孩抱着粉色爱心穿过鲜绿色时尚海报，横向滚动、分栏排版与歌词式文字共同形成音乐广告节奏。 原生视频规格为 15 秒 · 1920×2224。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustrated girl carries a pink heart through a vivid green fashion poster as side-scrolling motion, panel layouts, and lyric-like type create a music-commercial rhythm. Native-video output: 15s · 1920×2224. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2096496980180500694",
    "sourceLabel": "X 原帖 · @ponzponz15",
    "author": "ぽんず | AI映像",
    "publishedAt": "2026-09-06T07:13:25.000Z",
    "addedAt": "2026-09-07T01:01:29.590Z",
    "mediaUrl": "/media/x-2096496980180500694.mp4",
    "posterUrl": "/posters/x/2096496980180500694.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1920×2224",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ファッションポスター\n× AIモーショングラフィックス ③\n\n今回は「横スクロール × アニメ風イラスト × 楽曲CM風」\n\n毎度、ミヤマさんのプロンプトをお借りしています😎\n\n① ミヤマさんのプロンプトで画像作成\n② MiniMax H3 768pで動画化(@Hailuo_AI )\n③ GoogleのLyria 3.5で楽曲を生成/編集で追加\n\nちょっとLI⚪︎MO感あるな...🐿️\n\n#MiniMaxdesign",
    "engagement": {
      "replies": 3,
      "reposts": 14,
      "likes": 166,
      "views": 21194,
      "snapshotAt": "2026-09-07T01:01:29.590Z"
    },
    "approvedAt": "2026-09-07T01:01:29.590Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096577196915900712",
    "title": "猫耳厨师完成一盘炒饭",
    "titleEn": "A Cat-Eared Chef Finishes a Plate of Fried Rice",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝发猫耳厨师在中餐厅快速翻炒米饭，经历火焰、擦汗与装盘，最后把成品递给顾客。 原生视频规格为 15 秒 · 480×800。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A blue-haired cat-eared chef stir-fries rice in a Chinese restaurant, works through flame and exertion, plates the dish, and serves it to a customer. Native-video output: 15s · 480×800. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] High-quality Japanese anime style, 2D illustration with cel-shading. (S1) is a character showen in the <Picture 1>. The camera cuts to a medium shot of the same character (S1), maintaining (S1)'s  exact appearance and outfit from <Picture 1>, now standing in a vibrant Chinese restaurant kitchen. (S1) is positioned before a large wok on a powerful burner, surrounded by rising steam, hanging menu boards with illegible characters, and warm red and gold lighting. S1 begins to cook fried rice with focused, energetic movements. [Shot 2] At 00:04.500, the camera cuts to a series of rapid, cinematic close-ups. First, a tight shot of S1's determined eyes, then a shot of (S1)'s  hands skillfully tossing the wok. The camera pushes in with small amplitude at fast speed as (S1) exerts force to flip the rice, causing a few grains to dance in the air. A sudden burst of orange flame from the burner illuminates (S1)'s  face. Tiny beads of sweat form on (S1)'s  cheeks; (S1) briefly wipes (S1)'s  brow with (S1)'s  forearm without stopping (S1)'s  rhythmic stirring. [Shot 3] From 00:10.500 to 00:13.000, show the complete transfer of fried rice FROM THE WOK INTO AN EMPTY CERAMIC PLATE in one uninterrupted close medium shot. Both the wok and the empty plate must be clearly visible together. S1 holds the wok above the plate, visibly tilts its rim down toward the plate, and uses the spatula to slide and scrape the cooked fried rice out of the wok. Show the rice physically falling from the rim of the wok and accumulating into a mound on the plate. The plate starts empty and becomes filled on camera. Do not cut away, skip this action, or show an already-filled plate in its place. Only AFTER the rice has been transferred completely, from 00:13.000 to 00:15.000, S1 sets the wok down, lifts the filled plate, moves to the counter, and extends it toward a customer. The camera pans right with small amplitude at slow speed to follow the serving action. S1's expression shifts from intense concentration to a sincere, tired but happy smile, (S1)'s  chest heaving slightly from the effort. The customer's hand enters the frame to accept the plate. S1 gives a small, proud nod of satisfaction, (S1)'s  eyes shimmering with relief and accomplishment under the warm glow of the kitchen lights.\n\noverall_soundscape: The environment is filled with the rhythmic, metallic clanking of a spatula against a wok. Constant sizzling of rice and oil blends with the sudden, low-frequency \"whoosh\" of the burner's flames. The soft hiss of rising steam and the final, clean \"clink\" of a ceramic plate being placed on a wooden counter complete the atmosphere.\n\nnon_diegetic_music: An upbeat, heartwarming orchestral track with a fast tempo, featuring playful pizzicato strings and a bright flute melody that swells emotionally toward the end to emphasize a sense of accomplishment.",
    "promptSourceUrl": "https://x.com/IsekaiBardGuild/status/2096577421277335827",
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2096577196915900712",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-09-06T12:32:10.000Z",
    "addedAt": "2026-09-07T01:01:29.590Z",
    "mediaUrl": "/media/x-2096577196915900712.mp4",
    "posterUrl": "/posters/x/2096577196915900712.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×800",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\nチャーハン作り、参照画像を変えたら\n誰でも作れるようにしてみました\n\nできるだけ元の映像の流れを保存してみた\nつもりです（勝手に使ってすみません）\n\nRef2Va 用ですが、I2V でもできます\n（最初の数フレームが参照画像になります）\n\nプロンプトはリプ欄に",
    "engagement": {
      "replies": 5,
      "reposts": 5,
      "likes": 26,
      "views": 1593,
      "snapshotAt": "2026-09-07T01:01:29.590Z"
    },
    "approvedAt": "2026-09-07T01:01:29.590Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096390421626880069",
    "title": "同一张脸演绎多种细微情绪",
    "titleEn": "One Face Moves Through Subtle Emotions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一位写实女性在固定近景中连续呈现不安、惊讶、克制微笑与放松等表情，集中检验细微情绪控制。 原生视频规格为 65 秒 · 720×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same photoreal woman cycles through anxiety, surprise, restrained smiles, and relief in a fixed close-up, testing fine-grained emotional control. Native-video output: 65s · 720×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIPixLab/status/2096390421626880069",
    "sourceLabel": "X 原帖 · @AIPixLab",
    "author": "AI写真ラボ / AIバーチャルカメラマン",
    "publishedAt": "2026-09-06T00:10:00.000Z",
    "addedAt": "2026-09-07T01:01:29.590Z",
    "mediaUrl": "/media/x-2096390421626880069.mp4",
    "posterUrl": "/posters/x/2096390421626880069.jpg",
    "duration": 65,
    "aspectRatio": "portrait",
    "resolution": "720×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3はどこまで表情指示に追従できるのか？実用的プロンプト[prompt share]\n#MiniMaxH3 \n\n表情をジャンル分けナレッジ化\nDrama 感情表現 の作例",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 64,
      "views": 4983,
      "snapshotAt": "2026-09-07T01:01:29.590Z"
    },
    "approvedAt": "2026-09-07T01:01:29.590Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096749565412331841",
    "title": "巧克力蛋里长出一座夏威夷",
    "titleEn": "Hawaii Grows From a Chocolate Egg",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "一枚写实巧克力蛋沿裂纹打开，蛋壳与夹心连续变形成迷你夏威夷景观，最后以旅游广告式标题收束。 原生视频规格为 10 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A photoreal chocolate egg cracks open as its shell and filling continuously become a miniature Hawaii landscape, ending on a tourism-ad title card. Native-video output: 10s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[CITY] = ハワイ\n[CITY_NAME_DISPLAY] = ハワイ\n[FINAL_TAGLINE] = ひらいて、旅がはじまる。\n\nintegrated_multimodal_description:\n\n[Shot 1] A single continuous unbroken cinematic macro shot. No cuts, no scene changes, no teleporting, and no sudden replacements.\n\nThe scene begins with one realistic chocolate surprise egg placed upright on a simple elegant surface. The egg is the clear hero of the shot. It has a glossy chocolate shell with subtle handmade imperfections, soft highlights, delicate texture, and a premium confectionery feel. The lighting is warm, inviting, and slightly magical, with a shallow depth of field and a premium commercial look.\n\nThe egg must feel edible and real. The shell should look like real chocolate, not plastic, not ceramic, and not a toy. Fine surface texture, gentle gloss, tiny irregularities, and believable thickness should be visible.\n\nAt first, the chocolate egg is completely intact.\n\nGradually, subtle tension appears in the shell. Fine cracks begin to form naturally along the surface. The cracks spread slowly and elegantly, as if something beautiful is growing from within.\n\nThe transformation into [CITY] begins directly from this physical process.\n\nWithout interrupting the shot, the egg opens and transforms continuously.\n\nThe chocolate shell cracks, separates, curls, lifts, and unfolds in a graceful way. Pieces of the shell do not simply disappear. They remain physically connected to the transformation and become part of the emerging miniature city.\n\nThe inside of the egg reveals a warm, surprising inner world. The inner material may include soft chocolate layers, creamy confection-like texture, and delicate edible structural forms. Everything that appears must feel derived from the original egg.\n\nAs the egg opens, the miniature city of [CITY] begins to grow from inside it.\n\nThe transformation must feel like one continuous physical process:\nsealed chocolate egg → fine cracks → shell opening → interior revealed → miniature city structures rising from within → the completed city unfolding from the egg.\n\nNothing pops into existence.\nNothing instantly swaps shape.\nNothing should feel like a hidden cut.\n\nBefore generation, infer a small number of visually iconic elements strongly associated with [CITY], then reinterpret only the most recognizable ones as part of this miniature edible city.\n\nDo not turn the result into a crowded landmark collage.\nDo not overfill the frame.\nKeep the city readable, elegant, and beautiful.\n\nThe city should include a few recognizable urban and local elements such as streets, rooftops, bridges, towers, rail lines, rivers, trees, and culturally resonant architecture, but all reimagined as if they were born from the chocolate egg itself.\n\nThe shell edges may become ridges, terraces, or framing elements around the city.\nChocolate layers may become terrain and building mass.\nInterior creamy or confection-like details may become roads, rooftops, pathways, and decorative city features.\nThe material continuity must always remain believable.\n\nThe city should look like an edible miniature diorama: intricate, tactile, warm, and slightly surreal, but still physically plausible.\n\nCamera behavior:\nThe camera starts in tight macro close-up on the intact egg.\nAs the shell begins to crack and open, the camera remains intimate and steady, emphasizing material detail and the elegance of the transformation.\nAs the city becomes recognizable, the camera performs one extremely slow and smooth dolly backward with a subtle upward reveal.\nNo fast zoom.\nNo orbit.\nNo handheld movement.\n\nIn the final phase of the shot, reveal the completed miniature city of [CITY] emerging beautifully from the opened chocolate egg.\n\nThe final image should feel like a premium tourism commercial crossed with high-end confectionery advertising:\nwarm,\nwonder-filled,\nappetizing,\nsurprising,\nand emotionally memorable.\n\nSoft steam, glow, or delicate atmospheric motion may be present if appropriate, but keep it subtle and elegant.\n\nNear the end, minimal clean typography appears at the lower portion of the frame:\n\n\"[CITY_NAME_DISPLAY]\"\n\nDirectly beneath it, smaller:\n\n\"[FINAL_TAGLINE]\"\n\nThe text must be simple, elegant, readable, correctly spelled, and visually integrated into the final composition.\nNo other text appears.\n\nHold briefly on the final hero composition:\nthe opened chocolate egg acting like a cradle for the miniature city,\nbeautiful lighting,\nshallow depth of field,\nrich chocolate texture,\nand a visually irresistible tourism-commercial finish that makes the viewer want to visit [CITY].\n\nVisual style:\nphotorealistic macro commercial cinematography, premium confectionery advertisement, edible miniature city diorama, warm luxury food lighting, tactile chocolate texture, elegant surreal transformation, cinematic shallow depth of field, highly detailed but restrained composition.\n\nCritical continuity requirements:\nThe entire video is ONE continuous shot.\nThe entire city grows from ONE chocolate egg.\nThe egg remains the physical source of the entire transformation.\nAll changes are gradual, visible, and materially connected.\nThe shell must crack and open naturally.\nThe city must emerge progressively from within the egg.\nNo jump cuts.\nNo hidden transitions.\nNo sudden completed city.\nNo random fantasy city unrelated to [CITY].\n\nAvoid:\nplastic toy look,\ncartoon toy appearance,\nrubber texture,\nceramic egg appearance,\ngeneric fantasy castle city,\ninstant morphing,\nfast transformation,\njump cuts,\nscene changes,\nobjects appearing from nowhere,\novercrowded landmarks,\nunreadable text,\ngarbled Japanese characters,\nmisspelled city name,\nextra captions,\nextra logos,\ncheap CGI feeling,\nmessy composition,\nand any transformation that breaks physical continuity.\n\noverall_soundscape:\nSoft premium commercial ambience. Quiet room tone with subtle confectionery handling atmosphere. Gentle cracking sounds as the chocolate shell fractures. Delicate crisp chocolate break sounds, soft interior movement, faint textural shifting, and subtle reveal ambience as the city emerges. No dialogue.\n\nnon_diegetic_music:\nA warm, elegant, slightly magical commercial score. Soft piano and delicate orchestral textures, building gently as the egg opens and the city emerges. Emotional but restrained, never overwhelming the tactile chocolate sounds.",
    "promptSourceUrl": "https://x.com/loglogrog/status/2096749570227409041",
    "sourceUrl": "https://x.com/loglogrog/status/2096749565412331841",
    "sourceLabel": "X 原帖 · @loglogrog",
    "author": "マグマグ",
    "publishedAt": "2026-09-06T23:57:06.000Z",
    "addedAt": "2026-09-07T01:06:35.085Z",
    "mediaUrl": "/media/x-2096749565412331841.mp4",
    "posterUrl": "/posters/x/2096749565412331841.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "チョコエッグを割ったら、\n中からハワイと京都が出てきた。\n\nMiniMax H3 Max Turboで、\n同じプロンプトを使って、都市名だけ変えて2本作ってみました。\n\nハワイ → 京都。\n好きな街を1つ入れて、あなただけの10秒動画を作ってみてください。\n使ったプロンプトはリプ欄へ👇\n\n他の動画生成で試した結果なども共有して欲しいです。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 354,
      "snapshotAt": "2026-09-07T01:06:35.085Z"
    },
    "approvedAt": "2026-09-07T01:06:35.085Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096761524698169807",
    "title": "街头海报中破框而出的时尚舞者",
    "titleEn": "A Streetwear Performer Breaks Through the Poster Grid",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑衣表演者穿过海报墙、复印室与屋顶，以跳跃、转身和定格动作串起高对比排版的街头服饰广告。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black-clad performer crosses a poster wall, copy room, and rooftop, linking jumps, turns, and freezes into a high-contrast streetwear commercial. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create one coherent 15-second 16:9 MV with synchronized picture, original song, vocals, sound design, and cuts. Anime fused with live-action photography: a single original avant-garde street-fashion anime performer breaks through a street poster wall and travels through one connected nocturnal print route: poster corridor, copy room, tiled underpass/stairs, rooftop billboard. Gritty manga energy inspired by Gachiakuta, but no copied character, panel, logo, or artwork. Keep one face, hair silhouette, body proportions, cropped utility jacket, layered shirt, loose cargo trousers, high-tops, chain, and long strap in every shot. English-only on-screen text.\n\nVISUAL: strict black #111111, paper white #F1EBDD, and aquamarine #49D6C5; grayscale photographic spaces and skin. Retro copy-zine, pasted street posters, halftone, torn paper, tape, scanlines, restrained VHS gate weave and fringe. Main type is ultra-black condensed grotesk, ALL CAPS, tight tracking, squared counters. Every text event has Layer A readable hero base, Layer B offset aquamarine/grayscale print plate, Layer C foreground paper/type occlusion; read clearly before transforming and recover within one beat. Use photocopy burst on snare, aquamarine misregistration on hi-hat, torn occlusion on vocal words, irregular split panels on 808, delayed double portrait on hi-hat, and a short black type field at the ending. No random print, fake logos, gibberish, or ordinary subtitles.\n\nMUSIC/PERFORMANCE: original K-pop hip-hop/trap-rap, 128 BPM 4/4, double-time hats with half-time body feel, dry kick, snare on beat three, clipped clap, sparse hats and one roll, moving 3-note 808 with a slide and octave answer, metallic pluck and distorted electric-piano stab. One confident neutral-masculine voice: crisp aggressive rap, then chant-sung hook; final tag adds low-octave double. Structure: 0–1.875 intro tag “CUT IN”; 1.875–5.625 rap; 5.625–7.5 rising pre-hook and breath; 7.5–11.25 call-and-response hook; 11.25–13.125 kick-out beat switch; 13.125–15 final tag, bass hit, tiny tape-stop, clean hard stop, no fade. Lyrics: Rap A1 “Cut through the paper, make the dead wall shake”; Rap A2 “Ink in my jacket, every fake frame breaks”; Pre-hook “Hold that breath, let it rise”; Hook 1 “BREAK THE GRID”; Hook 2 “MAKE IT LIVE”; Response “Again!”; Final tag “BREAK THE GRID”. A1/A2 have connected meaning and rhyme, with straight-sixteenth then triplet cadence. Do not subtitle the lyrics; visual words are CUT IN, BREAK, THE GRID, HOLD, MAKE IT LIVE, AGAIN. Actions include diagonal power walk, cross-step, low drop/rebound, pivot freeze, jacket snap/strap swing, push-wall rebound, paper passage, and controlled stair step-down. Include three large actions, two space contacts, one clothing-inertia action, and two design-led still points. Align mouth, jaw, eyes, shoulders, footfalls, breath, type, snare, and 808.\n\nSHOT 1 — 0.0–1.6s, wide low push-in, poster-wall entrance. Performer steps from deep background through torn seam, shoulder first; front foot transfers weight, jacket catches paper, strap trails, camera settles on face. Action: paper passage. Text “CUT IN”, 110% hero. Layer A behind shoulder; B aquamarine jitter; C taped strip crosses torso then peels. Rhythm/Cut: inhale tears paper, kick hard-cuts.\nSHOT 2 — 1.6–3.4s, eye-level side track, poster corridor. Diagonal power walk left-to-right, foot on kick, body crosses seam, strap and hem trail, camera jolts. Action: power walk. Text “BREAK”, 70% split by poster columns. A clean by face; B slips on hats; C foreground strip passes arm. Rhythm/Cut: snare halftone and hard cut.\nSHOT 3 — 3.4–5.2s, low close-medium backward dolly, copy doorway. Cross-step over floor strip, shoulder-led 90-degree pivot, elbow brushes copier, torso rebounds. Action: cross-step/half-turn. Text “THE GRID”, 90% across copier/body in two irregular panels. A reads; B scans on rhyme; C paper flap slices lower frame and recovers. Rhythm/Cut: triplet jumps, cut on 808.\n\nSHOT 4 — 5.2–7.0s, locked eye-level close-up, copy-room still point. Pivot ends; one collar pull, shoulders freeze, eyes track, jaw releases on breath, strap settles. Action: jacket pull/freeze. Text “HOLD”, 125% behind head with face pocket. A readable; B returns to registration; C tiny monospaced “BREATHE” seam note vanishes. Rhythm/Cut: low end drops, hold, flash-cut on pre-hook end.\nSHOT 5 — 7.0–8.9s, wide low orbit, tiled underpass. Hook hit: performer bursts through threshold, forearm contacts pillar, center drops, leg rebounds, jacket opens, camera orbits then snaps back. Action: push-wall rebound. Text “BREAK THE GRID”, 145% in three unequal panels. A reads; B jumps aquamarine on 808; C torn diagonal crosses pillar but reveals face. Rhythm/Cut: hook consonant and kick cut on rebound.\nSHOT 6 — 8.9–10.8s, overhead-to-front whip tilt, stairs. Two steps up, axis turns down, controlled short step-down, knees absorb, upright hit. Action: high-low step. Text “MAKE”, 80% vertical along stair diagonal. A reads on rise; B duplicates one character per hat; C railing tape flicks away. Rhythm/Cut: response lands on step-down, photocopy flash restores grayscale.\nSHOT 7 — 10.8–13.2s, lateral track into frontal, rooftop billboard. Shoulder parts hanging pasted sheet, torso follows, strap swings once, three-quarter turn, feet lock. Action: paper passage/strap swing/pivot freeze. Text “MAKE IT LIVE”, 115% two-line fold around body. A reads before turn; B aquamarine misregisters on hat roll; C strap briefly occludes lower words, then clear. Rhythm/Cut: beat switch removes kick half-bar, one frame jump at strap apex.\nSHOT 8 — 13.2–15.0s, frontal close-up to black type field, rooftop billboard. Feet finish turn, shoulders square, chin lifts, jacket and strap settle late, camera holds, then hard-cuts to black. Action: final stance/release. Text “BREAK THE GRID”, 150%, same type family, paper-white on black with aquamarine offset. A clear; B low-octave visual duplicate; C compresses like tape stop but remains briefly readable. Rhythm/Cut: final bass hit and tag land together, two-frame tape tail, hard cut.\n\nGLOBAL LOCKS: one protagonist only, stable anime identity and outfit integrated into real photographic spaces; connected route, strict three-color system, accurate English, clear face and mouth. Hard cuts, flash frames, paper occlusion, panel splits, misregistration, restrained VHS only. No soft dissolve, fade, generic dance loop, continuous shake, extra characters, face swap, outfit change, duplicate limbs, broken hands, covered eyes/mouth, watermark, logo, fake brand, gibberish, or unreadable text. End decisively.",
    "promptSourceUrl": "https://x.com/RickTripstar/status/2096766667233599557",
    "sourceUrl": "https://x.com/RickTripstar/status/2096761524698169807",
    "sourceLabel": "X 原帖 · @RickTripstar",
    "author": "Riccardo",
    "publishedAt": "2026-09-07T00:44:38.000Z",
    "addedAt": "2026-09-07T01:11:00.236Z",
    "mediaUrl": "/media/x-2096761524698169807.mp4",
    "posterUrl": "/posters/x/2096761524698169807.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "stylized",
      "commercial"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Fashion advertisers are screwed.\n\nI made a streetwear commercial in just 20 minutes with Minimax H3.\n\nFiction Brand Name: “Break the Grid”",
    "engagement": {
      "replies": 3,
      "reposts": 3,
      "likes": 3,
      "views": 38,
      "snapshotAt": "2026-09-07T01:11:00.236Z"
    },
    "approvedAt": "2026-09-07T01:11:00.236Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096754759441260736",
    "title": "小蜘蛛把月光织成一双翅膀",
    "titleEn": "A Little Spider Weaves Moonlight Into Wings",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "阁楼蛛网上的小蜘蛛把金色光线织成发亮的飞蛾，放飞之后回望窗边的女人，形成温柔的夜间童话。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A little attic spider weaves golden light into a glowing moth, releases it, and looks back toward a woman by the window in a gentle nighttime fable. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2096754759441260736",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-09-07T00:17:45.000Z",
    "addedAt": "2026-09-07T01:13:38.628Z",
    "mediaUrl": "/media/x-2096754759441260736.mp4",
    "posterUrl": "/posters/x/2096754759441260736.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "creatures",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Some dreams arrive tangled and need a little time before they can take flight.\n\nKeep weaving. Keep believing. You never know which wish tomorrow may carry into the world. ✨\n\nGood night, everyone. May one of your dreams find its wings. 🩵\n\nMiniMax H3 @PixVerse   #pixversecpp",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 6,
      "views": 108,
      "snapshotAt": "2026-09-07T01:13:38.628Z"
    },
    "approvedAt": "2026-09-07T01:13:38.628Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096741221293265229",
    "title": "空教室里的五秒仰卧起坐",
    "titleEn": "Five Seconds of Sit-Ups in an Empty Classroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名女子在安静教室的地板上连续完成仰卧起坐，固定竖屏镜头突出短提示词驱动的单一动作测试。 原生视频规格为 5 秒 · 768×1088。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman performs sit-ups on the floor of a quiet classroom as a fixed vertical shot focuses on a simple action driven by a short prompt. Native-video output: 5s · 768×1088. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "教室で腹筋して",
    "promptSourceUrl": "https://x.com/moon_himiko/status/2096741221293265229",
    "sourceUrl": "https://x.com/moon_himiko/status/2096741221293265229",
    "sourceLabel": "X 原帖 · @moon_himiko",
    "author": "Ai_Himiko",
    "publishedAt": "2026-09-06T23:23:57.000Z",
    "addedAt": "2026-09-07T01:16:06.671Z",
    "mediaUrl": "/media/x-2096741221293265229.mp4",
    "posterUrl": "/posters/x/2096741221293265229.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1088",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございまーす\npromptは『教室で腹筋して』MiniMax H3\n元画像は最初に入っちゃうんだね。消すの面倒だね。\n一言でここまでできるのすごいな\n一つ間違いがあるとしたら、ロッカーの位置\nそこは先生の席だよ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 60,
      "snapshotAt": "2026-09-07T01:16:06.671Z"
    },
    "approvedAt": "2026-09-07T01:16:06.671Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096736067139563909",
    "title": "树影下的竖屏自拍推进",
    "titleEn": "A Vertical Selfie Push-In Under Dappled Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从树影中的半身自拍平滑推向面部特写，再轻微侧移并退回，持续检验自然表情、皮肤纹理与人物一致性。 原生视频规格为 7 秒 · 2160×3840。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The camera moves smoothly from a dappled-light selfie into a facial close-up, shifts gently to the side, and pulls back while testing natural expression, skin texture, and identity consistency. Native-video output: 7s · 2160×3840. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "竖屏9:16自拍角度，支持实时视频拍摄，镜头全程如陀螺仪般稳定平移俯仰，手持拍摄无任何晃动。\n\n禁止裁剪或分割镜像来截取画面。参考图中的黑发女孩拥有自然垂落的刘海、清澈灵动的大眼睛、淡妆、白皙肌肤，身穿白色上衣，需保持与参考图中人物一致的身份、妆容、发型和表情基调。\n\n开场角色眼神活泼可爱，眼动柔和自然，伴随自然眨眼；动作轻盈俏皮，嘴角带着淡淡的自然微笑，以细微的眼神变化为主焦点，避免夸张表情。镜头从开始即有轻微动态，缓慢而均匀地向前推进，同时伴有轻微的自然横向滑动和极小的高度变化，三者融合为一条连贯流畅的曲线，而非机械式的直线运动。\n\n当镜头拉近至脸颊特写并贴近面部时，并未停顿或后退，而是自然地向侧面偏移并轻微环绕，轻柔地控制环绕幅度以保持与面部的距离。镜头缓缓从前侧移动至微侧角度，随后回到正确位置，在此过程中清晰呈现出肌肤的真实质感：毛孔纹理、明显的面部绒毛根部、自然的粉状痕迹以及细微的皮肤瑕疵，这些缺陷真实自然，毫无刻意堆砌之感。周围人物呼吸自然，发丝微微晃动，眼神也随镜头角度的变化而轻轻转动。\n\n随后，镜头缓慢而稳定地拉回至正常的自拍距离，同时伴随轻微的横向移动和构图调整。当镜头后退时，人物微微低头，自然地耸肩微笑，神情轻松优雅，仿佛在躲避微风，又似在与之嬉戏。他们的动作略显克制，不夸张也不生硬，双手自然下垂，全程未抬手或摸头发。风渐渐平息，角色们缓缓抬头，眼神灵动，笑容自然，目光望向镜头末端。\n\n整个过程采用自然光线效果，阳光透过树叶洒在脸上，光影随着人物动作和树影的移动而自然流动变化，增强了真实光影层次感。\n\n画面明亮通透，温暖富有电影质感，搭配自然的光影效果。皮肤呈现真实、自然、高清的纹理，不粗糙、不过于光滑，也无塑料感或过度美化。摄像机流畅无缝的运动增强了画面的稳定性和质感，营造出一种非AI风格的高清晰度真实感。",
    "promptSourceUrl": "https://x.com/DDJCXX/status/2096736067139563909",
    "sourceUrl": "https://x.com/DDJCXX/status/2096736067139563909",
    "sourceLabel": "X 原帖 · @DDJCXX",
    "author": "硅基废话",
    "publishedAt": "2026-09-06T23:03:28.000Z",
    "addedAt": "2026-09-07T01:18:47.983Z",
    "mediaUrl": "/media/x-2096736067139563909.mp4",
    "posterUrl": "/posters/x/2096736067139563909.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "vertical"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Rtx5070 12g+32g内存，2小时用本地部署MiniMaxH3渲染出来的效果。\n\nPrompt:竖屏9:16自拍角度，支持实时视频拍摄，镜头全程如陀螺仪般稳定平移俯仰，手持拍摄无任何晃动。\n\n禁止裁剪或分割镜像来截取画面。参考图中的黑发女孩拥有自然垂落的刘海、清澈灵动的大眼睛、淡妆、白皙肌肤，身穿白色上衣，需保持与参考图中人物一致的身份、妆容、发型和表情基调。\n\n开场角色眼神活泼可爱，眼动柔和自然，伴随自然眨眼；动作轻盈俏皮，嘴角带着淡淡的自然微笑，以细微的眼神变化为主焦点，避免夸张表情。镜头从开始即有轻微动态，缓慢而均匀地向前推进，同时伴有轻微的自然横向滑动和极小的高度变化，三者融合为一条连贯流畅的曲线，而非机械式的直线运动。\n\n当镜头拉近至脸颊特写并贴近面部时，并未停顿或后退，而是自然地向侧面偏移并轻微环绕，轻柔地控制环绕幅度以保持与面部的距离。镜头缓缓从前侧移动至微侧角度，随后回到正确位置，在此过程中清晰呈现出肌肤的真实质感：毛孔纹理、明显的面部绒毛根部、自然的粉状痕迹以及细微的皮肤瑕疵，这些缺陷真实自然，毫无刻意堆砌之感。周围人物呼吸自然，发丝微微晃动，眼神也随镜头角度的变化而轻轻转动。\n\n随后，镜头缓慢而稳定地拉回至正常的自拍距离，同时伴随轻微的横向移动和构图调整。当镜头后退时，人物微微低头，自然地耸肩微笑，神情轻松优雅，仿佛在躲避微风，又似在与之嬉戏。他们的动作略显克制，不夸张也不生硬，双手自然下垂，全程未抬手或摸头发。风渐渐平息，角色们缓缓抬头，眼神灵动，笑容自然，目光望向镜头末端。\n\n整个过程采用自然光线效果，阳光透过树叶洒在脸上，光影随着人物动作和树影的移动而自然流动变化，增强了真实光影层次感。\n\n画面明亮通透，温暖富有电影质感，搭配自然的光影效果。皮肤呈现真实、自然、高清的纹理，不粗糙、不过于光滑，也无塑料感或过度美化。摄像机流畅无缝的运动增强了画面的稳定性和质感，营造出一种非AI风格的高清晰度真实感。",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 6,
      "views": 786,
      "snapshotAt": "2026-09-07T01:18:47.983Z"
    },
    "approvedAt": "2026-09-07T01:18:47.983Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096715442307883140",
    "title": "玻璃球里的一场微型暴雨",
    "titleEn": "A Miniature Storm Inside a Glass Globe",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "窗台上的玻璃球包住一栋微型房屋，内部乌云、闪电与彩虹依次变化，外壁凝结水珠强化了冷热边界。 原生视频规格为 15 秒 · 4096×2340。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A glass globe on a windowsill encloses a miniature house as storm clouds, lightning, and a rainbow pass inside while condensation reinforces the boundary between climates. Native-video output: 15s · 4096×2340. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "a few drops of condensation running down the outside of the glass from the cold inside",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2096715442307883140",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2096715442307883140",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-06T21:41:31.000Z",
    "addedAt": "2026-09-07T01:22:53.471Z",
    "mediaUrl": "/media/x-2096715442307883140.mp4",
    "posterUrl": "/posters/x/2096715442307883140.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "4096×2340",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3. Sunny out. Terrible in the globe.\n\nPrompt: \"a few drops of condensation running down the outside of the glass from the cold inside\"\n\nDuration: 15s. Camera: locked, sill height.\n\nThe condensation on the outside is the detail that makes the inside real. Physics leaking across the boundary sells the boundary.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 144,
      "snapshotAt": "2026-09-07T01:22:53.471Z"
    },
    "approvedAt": "2026-09-07T01:22:53.471Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096728398995320924",
    "title": "她走开后影子还留在墙上",
    "titleEn": "Her Shadow Stays After She Walks Away",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名女子在旧墙与木门前缓慢移动，她的身体离开原位后，墙上影子仍保持先前姿势，制造出克制的悬疑感。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman moves slowly beside an old wall and wooden door, but her shadow holds its former pose after she leaves, creating restrained supernatural suspense. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bmx_ai13/status/2096728398995320924",
    "sourceLabel": "X 原帖 · @bmx_ai13",
    "author": "BMX",
    "publishedAt": "2026-09-06T22:33:00.000Z",
    "addedAt": "2026-09-07T01:25:37.271Z",
    "mediaUrl": "/media/x-2096728398995320924.mp4",
    "posterUrl": "/posters/x/2096728398995320924.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "She moved. Her shadow didn’t.\n\nMinimax h3 on @creativeflyai",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 6,
      "views": 178,
      "snapshotAt": "2026-09-07T01:25:37.271Z"
    },
    "approvedAt": "2026-09-07T01:25:37.271Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096708322204475514",
    "title": "城市尽头撑开的能量护盾",
    "titleEn": "An Energy Shield Beneath a City-Sized Blast",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少女站在空旷街心迎向不断扩张的爆炸云，绿色能量罩从她身前升起并包裹整条街道。 原生视频规格为 7 秒 · 640×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl faces an expanding blast cloud in an empty avenue as a green energy shield rises around her and spans the street. Native-video output: 7s · 640×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/maestro_mu/status/2096708322204475514",
    "sourceLabel": "X 原帖 · @maestro_mu",
    "author": "アニ廃人",
    "publishedAt": "2026-09-06T21:13:13.000Z",
    "addedAt": "2026-09-08T01:43:38.865Z",
    "mediaUrl": "/media/x-2096708322204475514.mp4",
    "posterUrl": "/posters/x/2096708322204475514.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "640×1152",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "推しキャラ、ミント再び。\nAI動画MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 157,
      "snapshotAt": "2026-09-08T01:43:38.865Z"
    },
    "approvedAt": "2026-09-08T01:43:38.865Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096702671667270121",
    "title": "阳光广场里的挥手问候",
    "titleEn": "A Sunlit Hello in the Town Square",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫女孩在明亮的欧式广场向镜头挥手，发丝、衣摆和手臂动作保持轻快而连贯。 原生视频规格为 10 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime girl waves to the camera in a bright European-style square, with buoyant, consistent motion in her hair, skirt, and arms. Native-video output: 10s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KwevoLoFi/status/2096702671667270121",
    "sourceLabel": "X 原帖 · @KwevoLoFi",
    "author": "Kwevo LoFi 🌻 Nostalgic Beats",
    "publishedAt": "2026-09-06T20:50:46.000Z",
    "addedAt": "2026-09-08T01:43:38.865Z",
    "mediaUrl": "/media/x-2096702671667270121.mp4",
    "posterUrl": "/posters/x/2096702671667270121.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A little sunlit hello from Kwevo-san ☀️🌻\n\nTried MiniMax H3 for this one and I’m really impressed with how naturally it brought the scene to life.\n\nOriginal music: Sunlit Hello 🎧",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 120,
      "snapshotAt": "2026-09-08T01:43:38.865Z"
    },
    "approvedAt": "2026-09-08T01:43:38.865Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096689975786086537",
    "title": "伤后休养的宝可梦小家",
    "titleEn": "A Pokémon Family Tends to an Injury",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绘本质感的连续画面里，受伤的小生物被家人轻轻盖上毛毯，最后在床边安静睡去。 原生视频规格为 10 秒 · 1152×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a storybook-like sequence, an injured creature is gently tucked beneath a blanket by its family and settles to sleep. Native-video output: 10s · 1152×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelrhythms/status/2096689975786086537",
    "sourceLabel": "X 原帖 · @pixelrhythms",
    "author": "Keshav",
    "publishedAt": "2026-09-06T20:00:19.000Z",
    "addedAt": "2026-09-08T01:43:38.865Z",
    "mediaUrl": "/media/x-2096689975786086537.mp4",
    "posterUrl": "/posters/x/2096689975786086537.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1152×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Shelby had to rest for a few days after 🥺. \n Her family stayed with her while she got better 🥹.\n\nCoAuthored with ChatGPT6-Astra\nShot with Krea + MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 1152,
      "snapshotAt": "2026-09-08T01:43:38.865Z"
    },
    "approvedAt": "2026-09-08T01:43:38.865Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096689395831378359",
    "title": "森林洞穴里的败退",
    "titleEn": "A Defeat in the Forest Cave",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉色小生物在阴湿洞穴里遭遇体型更大的对手，被击倒后挣扎起身并独自离开。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A small pink creature confronts a much larger opponent in a damp forest cave, falls, recovers, and walks away alone. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelrhythms/status/2096689395831378359",
    "sourceLabel": "X 原帖 · @pixelrhythms",
    "author": "Keshav",
    "publishedAt": "2026-09-06T19:58:01.000Z",
    "addedAt": "2026-09-08T01:43:38.865Z",
    "mediaUrl": "/media/x-2096689395831378359.mp4",
    "posterUrl": "/posters/x/2096689395831378359.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Continuing to make the little film about Shelby 🥹 \nShe got beat up pretty badly by Purugly 😭 \nShe tried her best 😪\n\nScript coauthored with ChatGPT-6 Astra\nShot with Krea + MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 1360,
      "snapshotAt": "2026-09-08T01:43:38.865Z"
    },
    "approvedAt": "2026-09-08T01:43:38.865Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096689087553950181",
    "title": "夏日海岸的无尽平移",
    "titleEn": "An Endless Pan Across a Summer Shore",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "极简蓝色海天与白色浪线持续横向流动，画面用克制的色块和缓慢节奏表现夏日海岸。 原生视频规格为 13 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Minimal blue sea and sky bands drift horizontally while a thin white surf line gives the summer shoreline a restrained, slow rhythm. Native-video output: 13s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ramblingpix/status/2096689087553950181",
    "sourceLabel": "X 原帖 · @Ramblingpix",
    "author": "Ramblingpix",
    "publishedAt": "2026-09-06T19:56:47.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096689087553950181.mp4",
    "posterUrl": "/posters/x/2096689087553950181.jpg",
    "duration": 13,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Summer\n\nDigital painting/MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 110,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096688877339906480",
    "title": "蜘蛛主厨的八道菜",
    "titleEn": "An Eight-Course Dinner for a Spider Critic",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拟人蜘蛛主厨在灯光温暖的餐桌前摆盘，盛装瓢虫评审品尝后给出满意反应。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anthropomorphic spider chef plates a dish at a warmly lit table, where a formally dressed ladybug critic tastes it and approves. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2096688877339906480",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-09-06T19:55:57.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096688877339906480.mp4",
    "posterUrl": "/posters/x/2096688877339906480.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "food",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Eight legs. Eight courses. One very hard-to-impress critic. 🕷️👨‍🍳\n\nChef spun her finest creation—and the verdict is in. ⭐⭐⭐⭐⭐\n\nWhat’s happening in your corner of the web today?\n\nHappy Spider Sunday! 🕸️\n\nMiniMax H3 @PixVerse   #pixversecpp\n#SpiderSunday",
    "engagement": {
      "replies": 6,
      "reposts": 3,
      "likes": 17,
      "views": 271,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096682171712885083",
    "title": "猫耳少女骑过油菜花田",
    "titleEn": "A Cat-Eared Rider Crosses the Flower Fields",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "猫耳少女驾驶黄色踏板车穿过起伏的乡间道路，镜头从近景跟拍逐渐拉远到开阔田野。 原生视频规格为 7 秒 · 544×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared girl rides a yellow scooter along rolling country roads as the camera pulls from a close follow shot into broad fields. Native-video output: 7s · 544×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Natsuexp/status/2096682171712885083",
    "sourceLabel": "X 原帖 · @Natsuexp",
    "author": "Natsuexp (奇策士) 🇦🇷",
    "publishedAt": "2026-09-06T19:29:18.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096682171712885083.mp4",
    "posterUrl": "/posters/x/2096682171712885083.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "544×960",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます☀️\n今日は9月7日、月曜日！\n\n昨日は昔作った画像を使って、新しい動画生成モデル「MiniMax H3」をいろいろ試していました🎬✨\n手や顔が少し崩れることもありますが、思った以上にいい感じに動いてくれました🐱\n\n今週もゆるっとよろしくお願いします🌸",
    "engagement": {
      "replies": 21,
      "reposts": 101,
      "likes": 445,
      "views": 3014,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video.",
    "featured": {
      "at": "2026-09-13T00:21:49.933Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2096674510392918507",
    "title": "竹林里的剑舞练习",
    "titleEn": "Sword-Dance Practice in a Bamboo Grove",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "黑衣剑士在雾气与逆光笼罩的竹林完成连续挥砍、转身与架势变化，衣袖和刀光同步展开。 原生视频规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-clad swordswoman performs connected slashes, turns, and stances in a misty backlit bamboo grove, with sleeves and blade motion in sync. Native-video output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/one_good_prompt/status/2096674510392918507",
    "sourceLabel": "X 原帖 · @one_good_prompt",
    "author": "One Good Prompt Production",
    "publishedAt": "2026-09-06T18:58:52.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096674510392918507.mp4",
    "posterUrl": "/posters/x/2096674510392918507.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A short of Japanese girl practicing her sword dancing skill. Animation was done locally with RTX 5070 in @ComfyUI with @Hailuo_AI miniMAX H3 r2v. Super @Grok helped with designing the character and environment. Stitched and synced all the stuff in Adobe Premiere.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 112,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096672719928730089",
    "title": "红黑时装片的连环转场",
    "titleEn": "Rapid Transitions in a Red-and-Black Fashion Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红、黑、白三色人物剪影在酒杯、锁链、钟面和几何舞台之间连续切换，形成高密度时装视觉片。 原生视频规格为 28 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Red, black, and white figures cut rapidly between wine glasses, chains, clock faces, and geometric stages in a dense fashion-film montage. Native-video output: 28s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/getlower12/status/2096672719928730089",
    "sourceLabel": "X 原帖 · @getlower12",
    "author": "Beachwagon",
    "publishedAt": "2026-09-06T18:51:45.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096672719928730089.mp4",
    "posterUrl": "/posters/x/2096672719928730089.jpg",
    "duration": 28,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "最近很火的minimax h3的红黑风格尝试 欢迎交流",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 117,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096664712704647500",
    "title": "史诗壁画里的火焰神话",
    "titleEn": "A Fire Myth Unfolds Across Epic Murals",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "神话人物、骰子、海浪与火焰在壁画式构图中接力出现，组成一段带手绘纹理的史诗预告。 原生视频规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Mythic figures, dice, waves, and fire pass through mural-like compositions to form an epic trailer with hand-painted texture. Native-video output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Rishavchhajer/status/2096664712704647500",
    "sourceLabel": "X 原帖 · @Rishavchhajer",
    "author": "rishabh chhajer 🔆",
    "publishedAt": "2026-09-06T18:19:56.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096664712704647500.mp4",
    "posterUrl": "/posters/x/2096664712704647500.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Inspired by the legendary Indian mythology Mahabharata, and also by a recent video from\n@NVTDanh that really got me to think and create this 30 second trailer.\nHope you enjoy this little experiment 🙏✨\nModel : Minimax H3 Max (@Hailuo_AI )\nSound track from @Gemini Lyria 3 Pro\nCreated By  @AtlabsAI\n#lyria3pro #minimaxh3 #aiart #mahabharta",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 170,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096661667383808474",
    "title": "荒原赛车的贴地追逐",
    "titleEn": "A Low-Slung Chase Across the Wasteland",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "改装赛车在沙尘荒原高速追逐，驾驶舱近景与贴地外拍交替强调速度、颠簸和压迫感。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Modified race cars tear across a dusty wasteland as cockpit close-ups and low exterior angles trade off to emphasize speed and impact. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HexxRL/status/2096661667383808474",
    "sourceLabel": "X 原帖 · @HexxRL",
    "author": "Hexx ./",
    "publishedAt": "2026-09-06T18:07:50.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096661667383808474.mp4",
    "posterUrl": "/posters/x/2096661667383808474.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Used GPT 6 Astra to create a concept outline for a Race in the Wastelands. 🌵\n\n🏎️ Punks speeding throughout the terrain in fierce race\n\n🎥 shot on @vidmuse_ai freeform with MiniMax H3. how do you think GPT 6 did with the outline and prompt structure?",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 22,
      "views": 1014,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096659449226375484",
    "title": "禁书让黑龙膨胀失控",
    "titleEn": "A Forbidden Grimoire Makes a Black Dragon Swell",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑龙翻开禁书后身体不断膨胀，先挤满木屋，随后撑破屋顶并俯瞰整座营地。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black dragon opens a forbidden grimoire and keeps swelling, first filling a wooden room and then bursting through the roof above the camp. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chasinglightX64/status/2096659449226375484",
    "sourceLabel": "X 原帖 · @chasinglightX64",
    "author": "chasinglight",
    "publishedAt": "2026-09-06T17:59:01.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096659449226375484.mp4",
    "posterUrl": "/posters/x/2096659449226375484.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Reading the contents of a forbidden grimoire can lead to unexpected consequences, even if you do not possess magical abilities - Black dragon inflation made by MiniMax h3",
    "engagement": {
      "replies": 1,
      "reposts": 17,
      "likes": 120,
      "views": 1930,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096656851727442350",
    "title": "暗黑糖果舞台上的偶像",
    "titleEn": "An Idol on a Dark Candy Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "粉蓝发动漫偶像在蝙蝠、碎心与霓虹几何体包围的舞台变换姿态，甜美配色与暗黑符号相互碰撞。 原生视频规格为 15 秒 · 768×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-and-blue-haired anime idol changes poses among bats, broken hearts, and neon geometry, mixing candy colors with dark iconography. Native-video output: 15s · 768×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2096656851727442350",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-09-06T17:48:42.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096656851727442350.mp4",
    "posterUrl": "/posters/x/2096656851727442350.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1152",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIart #H3MaxTurbo #aivideo \n異常な安さのセールが9/7までなので作りまくっており、いいのができるので連投すみません。\nプロンプトを「ダークホラーカワイイ」に書き換え、原画指定、Minimax H3 Max Turboポン出し。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 9,
      "views": 467,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096654993390719232",
    "title": "把虚幻校园染成水彩",
    "titleEn": "Turning a Virtual Campus into Watercolor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "校园教室与走廊从三维场景转化为柔和水彩质感，人物行走和空间透视在风格迁移中保持连续。 原生视频规格为 30 秒 · 854×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A virtual classroom and corridor shift into soft watercolor while a walking character and the spatial perspective remain continuous through the transformation. Native-video output: 30s · 854×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kiyoshi_shin/status/2096654993390719232",
    "sourceLabel": "X 原帖 · @kiyoshi_shin",
    "author": "新清士@AIコンテンツ開発者",
    "publishedAt": "2026-09-06T17:41:19.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096654993390719232.mp4",
    "posterUrl": "/posters/x/2096654993390719232.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "854×480",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "UE5で作った動画を、さらにMiniMax H3（ローカル）に入れて水彩画調にするテスト。動画は元のままだとCGが強く残るので、一度Depthに変換して、参照画像を１枚スクショで水彩画に変換している（制御はGPT-6 Astraにやらせている）。これを、GPTで使わないで自動化するワークフローを組みたいところ。",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 41,
      "views": 4211,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096636966913896469",
    "title": "霓虹动态图形里的动漫舞者",
    "titleEn": "An Anime Dancer Inside Neon Motion Graphics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "蓝发动漫舞者在爱心、气泡、星形和巨大排版之间切换景别，以流行偶像姿态贯穿整段竖屏表演。 原生视频规格为 15 秒 · 1080×1888。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A blue-haired anime dancer moves through hearts, bubbles, stars, and oversized typography, maintaining pop-idol poses across the vertical performance. Native-video output: 15s · 1080×1888. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "9:16 vertical.\n\nA spectacular high-end 2D anime music-video performance featuring a cute silver-haired girl in a stylish kawaii casual street-fashion outfit, wearing a simple coordinated top and opaque shorts with ankle socks and fashionable sneakers, dancing alone in an extraordinary kawaii motion-graphics world.\n\nShe has a beautiful delicate anime face, large highly expressive eyes, soft silver hair with individually rendered strands, a charming youthful idol-like appearance, and an elegant stylized character design. Her face, hairstyle, proportions, clothing, and overall identity remain highly consistent throughout the entire video.\n\nHer outfit is tasteful, practical, and natural for a pop performer. The clothing uses high-quality fabric with a natural soft drape and maintains an elegant, clean silhouette throughout the dance.\n\nShe performs a stylish kawaii dance with playful hand gestures, cute rhythmic steps, light bouncing movements, expressive arm movements, slow controlled turns, charming poses, and confident idol-like expressions. Her choreography feels musical, fluid, precise, and full of personality. Fast spins and rapid turns are avoided; slow controlled turns are allowed.\n\nThe entire world behaves like a sophisticated animated music-video graphic system synchronized with her dancing.\n\nHundreds of colorful graphic elements dynamically respond to her movements: hearts, stars, sparkles, ribbons, circles, diamonds, flowers, bubbles, glossy candy-like shapes, translucent glass objects, glowing particles, geometric fragments, floating panels, and luminous 2D symbols.\n\nKawaii pastel colors collide with vivid cyan, hot pink, lavender, electric blue, white, and occasional deep black accents.\n\nCreate an extremely sophisticated mixture of 2D anime animation, graphic design, cel animation, glossy 3D objects, translucent materials, neon light, and abstract motion graphics.\n\nThe camera constantly evolves with the performance: dramatic push-ins, rapid pull-backs, smooth orbital movement, natural eye-level perspectives, medium shots, full-body compositions, expressive face close-ups, and hand close-ups. Camera movement remains intentional and rhythmically connected to the dance, while keeping the character and her performance clearly visible.\n\nAvoid voyeuristic or suggestive camera angles. Do not intentionally emphasize the chest, hips, buttocks, or areas normally covered by clothing. Keep the camera at tasteful, performance-oriented angles.\n\nUse dynamic music-video editing language: rhythmic jump cuts, graphic wipes, rapid zoom transitions, match cuts, freeze-frame accents, layered compositions, brief strobing flashes, split screens, mirrored compositions, diagonal divisions, rotating frames, and multiple simultaneous versions of the same scene appearing as graphic layers.\n\nAt certain moments, the screen becomes a sophisticated kinetic typography composition. Large fictional graphic lettering such as “KAWAII”, “MIRAI”, “POP STAR”, “LOVE MODE”, and abstract decorative characters appear as visual design elements, transforming, stretching, rotating, scattering, and snapping into geometric formations in rhythm with the performance. The typography is clearly rendered and remains visually coherent, functioning as part of the motion graphics rather than subtitles.\n\nUse controlled glitch aesthetics: RGB channel separation, digital displacement, scanline textures, pixel fragments, chromatic aberration, brief frame offsets, digital tearing, geometric corruption, and rapid graphic reconstruction. Glitches are highly stylized and intentional, never damaging the character's face or anatomy.\n\nCreate moments where the environment dramatically expands beyond normal space: enormous floating hearts and geometric structures surround the dancer, gigantic glossy objects pass behind her, luminous rings expand from her body, and hundreds of tiny graphic elements burst outward from her gestures.\n\nOccasionally transform the background into an enormous abstract graphic composition resembling a futuristic kawaii album cover, then seamlessly reconstruct it around the dancing character.\n\nUse strong foreground-background layering and parallax. Some graphic elements pass close to the camera while others remain far behind the character, creating a deep three-dimensional stage.\n\nThe dancer always remains the visual center of the composition. Effects enhance her performance rather than obscuring her.\n\nPremium Japanese anime music-video aesthetic, exceptionally crisp anime linework, detailed cel shading, beautiful hair rendering, expressive eyes, sophisticated character animation, polished compositing, precise graphic design, cinematic lighting, subtle bloom, luminous highlights, controlled film grain, high visual density, professional music-video production quality.\n\nThe result should feel like an impossibly polished fusion of kawaii idol culture, experimental motion graphics, futuristic pop design, anime opening animation, and a high-budget electronic music video.\n\nCute, beautiful, energetic, stylish, surreal, colorful, sophisticated, visually overwhelming, but always clean and intentional.\n\nNo NSFW content, no sexualized presentation, no voyeuristic framing, no extra characters, no singing, no speaking, no watermark, no subtitles, no morphing face, no identity changes, no distorted anatomy, no duplicated limbs, no extra fingers, no malformed hands, no missing clothing, no transparent clothing, no sheer fabric, no sudden costume changes, no photorealism, no sudden art-style changes.",
    "promptSourceUrl": "https://x.com/sorastg524/status/2096640607708188717",
    "sourceUrl": "https://x.com/sorastg524/status/2096636966913896469",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-09-06T16:29:41.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096636966913896469.mp4",
    "posterUrl": "/posters/x/2096636966913896469.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1888",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIart #H3MaxTurbo #aivideo #ai動画\nさっきの動画と同じプロンプトでの別出力例。\n同じプロンプトで同じくらいのクオリティの演出の違う出力結果が出る。\n（MiniMax H3 Max Turbo 出力動画をCapCutの画質向上（AIHD）+Video2Xでフレーム補間）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 12,
      "views": 481,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2097112285827268887",
    "title": "黑花园里的石像预告片",
    "titleEn": "A Stone-Garden Trailer in Blue and Red",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝黑花园、红花、乌鸦与古典石像在深色调镜头中接力出现，组成一段带悬疑气息的短预告。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-black garden, red flowers, crows, and classical statues pass through a darkly graded sequence that plays like a mystery trailer. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2097112285827268887",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-09-07T23:58:26.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2097112285827268887.mp4",
    "posterUrl": "/posters/x/2097112285827268887.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized",
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 + Midjourney v8.2\n\nCorrect me if I’m wrong, but if you give H3 high-quality references, you get high-quality results. This trailer uses six Midjourney images as references. I’m adding a few of the references and the video prompt in the replies.\n\nThe Stone Garden",
    "engagement": {
      "replies": 7,
      "reposts": 1,
      "likes": 34,
      "views": 1231,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2097001707683119410",
    "title": "柠檬黄影棚里的粉蓝残影舞",
    "titleEn": "Pink-and-Cyan Afterimages on a Lemon Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "红发动漫舞者在纯黄色影棚完成快速舞步，粉色与青色残影沿四肢动作展开并重新合拢。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A red-haired anime dancer performs fast choreography on a pure yellow stage as pink and cyan afterimages peel from her limbs and snap back together. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15-second experimental anime fashion film. Use @[character reference] for the character’s exact identity, outfit and rendering style. A pristine, uniform lemon-yellow cyclorama with a seamless matte floor. Flat, clean yellow throughout, free of grain, gradients and set dressing. 16:9. No text or logos.\n\nThe character performs one fast, unconventional dance phrase: ankles weave while the torso stays suspended, a wrist flick ripples through both elbows into a chest pop, knees fold into a low corkscrew pivot, then a heel swivel unwinds the body into an off-balance editorial pose. Continue through asymmetric shoulder rolls, angular arm threading around the face and a sideways glide with the head counter-rotating toward the lens. Every sharp lock immediately releases into another surprising movement. Convincing weight, precise foot contact and expressive follow-through.\n\nElectric cyan and hot pink dominate the motion effects. Bold, opaque silhouette afterimages peel away from opposite sides of the body, replay the previous gesture a fraction late and slam back into alignment on each pose. Arm sweeps stretch them into broad curved ribbons; crossing steps weave pink and cyan trails around the ankles; rapid turns fan out overlapping screen-print impressions. At the strongest accents, oversized cyan and pink cutouts of the current pose slide across the foreground, briefly masking the lens. Keep the real character crisp and recognizable, with clean yellow visible between effect bursts.\n\nThe camera dances in counterpoint: skim backward at ankle height through advancing footwork, whip upward around an outstretched wrist into an extreme face close-up, then use the passing forearm to cut into a rotating overhead view of the low pivot. Dive to a tilted low-angle full-body composition as the character rises. Orbit opposite the sideways glide, exaggerating hands and feet near the wide lens. Accelerate between compositions and brake hard on pose locks.\n\nFinish with a sudden forward lean toward the lens while enormous pink and cyan echoes continue leaning farther, sweeping past both sides of the camera. Snap wide: the character lands a final asymmetric pose as the colored echoes collapse precisely into the silhouette. Hard cut to black.\n\n170 BPM broken beat, syncopated accents, fluid animation and razor-fast camera transitions. No slow motion or prolonged holds.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2097001707683119410",
    "sourceUrl": "https://x.com/aimikoda/status/2097001707683119410",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-09-07T16:39:02.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2097001707683119410.mp4",
    "posterUrl": "/posters/x/2097001707683119410.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Prompt Share\n\nAnime fashion film with dance moves.\nYou can use this for your anime music videos, openings or outros. Just add your character.  Also, you can tweak the prompt however you like.\n\nPrompt:\n15-second experimental anime fashion film. Use @[character reference] for the character’s exact identity, outfit and rendering style. A pristine, uniform lemon-yellow cyclorama with a seamless matte floor. Flat, clean yellow throughout, free of grain, gradients and set dressing. 16:9. No text or logos.\n\nThe character performs one fast, unconventional dance phrase: ankles weave while the torso stays suspended, a wrist flick ripples through both elbows into a chest pop, knees fold into a low corkscrew pivot, then a heel swivel unwinds the body into an off-balance editorial pose. Continue through asymmetric shoulder rolls, angular arm threading around the face and a sideways glide with the head counter-rotating toward the lens. Every sharp lock immediately releases into another surprising movement. Convincing weight, precise foot contact and expressive follow-through.\n\nElectric cyan and hot pink dominate the motion effects. Bold, opaque silhouette afterimages peel away from opposite sides of the body, replay the previous gesture a fraction late and slam back into alignment on each pose. Arm sweeps stretch them into broad curved ribbons; crossing steps weave pink and cyan trails around the ankles; rapid turns fan out overlapping screen-print impressions. At the strongest accents, oversized cyan and pink cutouts of the current pose slide across the foreground, briefly masking the lens. Keep the real character crisp and recognizable, with clean yellow visible between effect bursts.\n\nThe camera dances in counterpoint: skim backward at ankle height through advancing footwork, whip upward around an outstretched wrist into an extreme face close-up, then use the passing forearm to cut into a rotating overhead view of the low pivot. Dive to a tilted low-angle full-body composition as the character rises. Orbit opposite the sideways glide, exaggerating hands and feet near the wide lens. Accelerate between compositions and brake hard on pose locks.\n\nFinish with a sudden forward lean toward the lens while enormous pink and cyan echoes continue leaning farther, sweeping past both sides of the camera. Snap wide: the character lands a final asymmetric pose as the colored echoes collapse precisely into the silhouette. Hard cut to black.\n\n170 BPM broken beat, syncopated accents, fluid animation and razor-fast camera transitions. No slow motion or prolonged holds.",
    "engagement": {
      "replies": 9,
      "reposts": 22,
      "likes": 271,
      "views": 7590,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096980289184805308",
    "title": "暴风雪悬崖上的最后抓点",
    "titleEn": "A Last Handhold on a Blizzard Cliff",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "攀登者在雪山绝壁上打滑、坠落并由绳索拉停，最后抓住崖缘看见埋在雪中的入口。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A climber slips on a snowy cliff, drops until her rope catches, and finally secures the ledge below a buried mountain entrance. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description:\n[Shot 1] Photorealistic cinematic survival action, 15 seconds. Lara climbs a steep cliff in a blizzard toward a snow-buried ancient entrance. Lock her identity, hair, clothing, injuries, and equipment. Keep one climbing-tool type, secured when released; maintain the same rope-to-harness connection and upper anchor throughout.\n\nExtreme close tracking begins on her left hand gripping a snowy crack, then tilts rapidly along her arm to her tense profile. Her right hand pulls the rope, knee braces against rock, and she pushes upward half a step. Eyes target the next hold; jaw tightens during effort. Lara (S1), quietly breathless, says, <d>[English] Almost there…</d> Wind drives snow sideways; loosened powder falls and the rope trembles.\n\n[Shot 2] At 00:02.000, cut over her shoulder into a slightly wide subjective view. Advance with her head lift, then tilt upward toward the ancient doorway emerging through snow. She judges the route for half a beat and points her climbing tool toward an upper-right hold. No frontal insert or dialogue. A ruined pinnacle appears briefly, valley clouds churn below, and wind sharpens.\n\n[Shot 3] At 00:04.000, cut to a front-side medium close-up, tracking her lateral traverse. Her right foot loads a thin ice edge and slips. Only then do her eyes widen and breath catch; she checks her foot, tightens her jaw, and searches for another grip. Lara (S1) produces <d>[English] Tsk.</d> as one clipped tongue click, not spoken letters. Snow and stone fragments fall into the ravine.\n\n[Shot 4] At 00:06.000, track her sudden drop, then arrest sharply as the rope loads through her harness. She swings outward beneath the anchor, left hand gripping rope, right hand reaching for projecting rock; the tool remains secured. Her breath stops, then her eyes lock onto the target. Lara (S1), through clenched teeth, says, <d>[English] Don’t fall now.</d> Finish on one short exhalation, slightly stressing \"now.\" Rope vibration, crumbling rock, and falling snow follow the load.\n\n[Shot 5] At 00:08.000, cut to a diving angle following her swing toward the new hold, then move rapidly back toward her face. She catches the projection and uses her core and momentum to return against the cliff. Her chest strikes snowy rock; eyes briefly close and breath is knocked out. She immediately looks upward and resumes short breaths without crying out. Impact throws snow outward; wind whips her hair and briefly reveals the ruins.\n\n[Shot 6] At 00:10.000, frame her gripping hand, push toward her eyes, then tilt slightly toward the cliff lip. Two quick climbing movements bring her hand onto a projecting stone edge. As she pulls, it breaks. Alarm follows the fracture: eyes widen, inhalation catches, and her gaze tracks the failing hold. Lara (S1) says, <d>[English] Wait!</d> sharply, once. Rock tumbles, snow dust strikes her face, and wind curls over the lip.\n\n[Shot 7] At 00:12.000, follow a half-torso drop, then stop on her hand catching the solid cliff edge. She brings the other hand up and stabilizes her weight below the lip; the rope remains connected and physically loaded. Only after securing both hands does she exhale. Slowly tilt upward with her gaze to the enormous snow-buried doorway. Her eyelids and jaw ease slightly, retaining exhaustion and vigilance. Lara (S1), low and breathless, says, <d>[English] Found you.</d> A gust clears the entrance, revealing faint cold light through its seam. Leave the doorway visible after the line. End at 15.00 seconds with Lara still braced at the edge, without standing, walking, or opening the door.\n\nPreserve all seven intervals, camera movements, and action causality. Lara is the only speaker; each exact utterance finishes within its assigned interval. Keep one voice, natural lip synchronization, and physically motivated breathing. No extra dialogue, screams, narration, reaction inserts, or performance pauses. Fear follows danger; relief follows a secure grip. Maintain realistic gripping, rope tension, swing direction, impact, and falling debris. No identity drift, distorted anatomy, extra limbs, intersecting hands or ropes, drifting holds, disappearing equipment, unsupported hovering, exaggerated expressions, subtitles, text, or watermarks.\n\noverall_soundscape:\nContinuous blizzard wind, clothing friction, exertion breaths, rope loading, cracking rock, and falling debris. Synchronize the rope arrest and muffled chest impact precisely; keep dialogue intelligible beneath the storm.\n\nnon_diegetic_music:\nN/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2096980289184805308",
    "sourceUrl": "https://x.com/ou_zhen599/status/2096980289184805308",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-07T15:13:55.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096980289184805308.mp4",
    "posterUrl": "/posters/x/2096980289184805308.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI\n\nImage-to-Video：🧗“I was looking at my feet and almost didn't make it up.”\n😶The H3's high dynamic range always seems to be a cause for concern.\n\nMiniMax-H3 prompt👇\nintegrated_multimodal_description:\n[Shot 1] Photorealistic cinematic survival action, 15 seconds. Lara climbs a steep cliff in a blizzard toward a snow-buried ancient entrance. Lock her identity, hair, clothing, injuries, and equipment. Keep one climbing-tool type, secured when released; maintain the same rope-to-harness connection and upper anchor throughout.\n\nExtreme close tracking begins on her left hand gripping a snowy crack, then tilts rapidly along her arm to her tense profile. Her right hand pulls the rope, knee braces against rock, and she pushes upward half a step. Eyes target the next hold; jaw tightens during effort. Lara (S1), quietly breathless, says, <d>[English] Almost there…</d> Wind drives snow sideways; loosened powder falls and the rope trembles.\n\n[Shot 2] At 00:02.000, cut over her shoulder into a slightly wide subjective view. Advance with her head lift, then tilt upward toward the ancient doorway emerging through snow. She judges the route for half a beat and points her climbing tool toward an upper-right hold. No frontal insert or dialogue. A ruined pinnacle appears briefly, valley clouds churn below, and wind sharpens.\n\n[Shot 3] At 00:04.000, cut to a front-side medium close-up, tracking her lateral traverse. Her right foot loads a thin ice edge and slips. Only then do her eyes widen and breath catch; she checks her foot, tightens her jaw, and searches for another grip. Lara (S1) produces <d>[English] Tsk.</d> as one clipped tongue click, not spoken letters. Snow and stone fragments fall into the ravine.\n\n[Shot 4] At 00:06.000, track her sudden drop, then arrest sharply as the rope loads through her harness. She swings outward beneath the anchor, left hand gripping rope, right hand reaching for projecting rock; the tool remains secured. Her breath stops, then her eyes lock onto the target. Lara (S1), through clenched teeth, says, <d>[English] Don’t fall now.</d> Finish on one short exhalation, slightly stressing \"now.\" Rope vibration, crumbling rock, and falling snow follow the load.\n\n[Shot 5] At 00:08.000, cut to a diving angle following her swing toward the new hold, then move rapidly back toward her face. She catches the projection and uses her core and momentum to return against the cliff. Her chest strikes snowy rock; eyes briefly close and breath is knocked out. She immediately looks upward and resumes short breaths without crying out. Impact throws snow outward; wind whips her hair and briefly reveals the ruins.\n\n[Shot 6] At 00:10.000, frame her gripping hand, push toward her eyes, then tilt slightly toward the cliff lip. Two quick climbing movements bring her hand onto a projecting stone edge. As she pulls, it breaks. Alarm follows the fracture: eyes widen, inhalation catches, and her gaze tracks the failing hold. Lara (S1) says, <d>[English] Wait!</d> sharply, once. Rock tumbles, snow dust strikes her face, and wind curls over the lip.\n\n[Shot 7] At 00:12.000, follow a half-torso drop, then stop on her hand catching the solid cliff edge. She brings the other hand up and stabilizes her weight below the lip; the rope remains connected and physically loaded. Only after securing both hands does she exhale. Slowly tilt upward with her gaze to the enormous snow-buried doorway. Her eyelids and jaw ease slightly, retaining exhaustion and vigilance. Lara (S1), low and breathless, says, <d>[English] Found you.</d> A gust clears the entrance, revealing faint cold light through its seam. Leave the doorway visible after the line. End at 15.00 seconds with Lara still braced at the edge, without standing, walking, or opening the door.\n\nPreserve all seven intervals, camera movements, and action causality. Lara is the only speaker; each exact utterance finishes within its assigned interval. Keep one voice, natural lip synchronization, and physically motivated breathing. No extra dialogue, screams, narration, reaction inserts, or performance pauses. Fear follows danger; relief follows a secure grip. Maintain realistic gripping, rope tension, swing direction, impact, and falling debris. No identity drift, distorted anatomy, extra limbs, intersecting hands or ropes, drifting holds, disappearing equipment, unsupported hovering, exaggerated expressions, subtitles, text, or watermarks.\n\noverall_soundscape:\nContinuous blizzard wind, clothing friction, exertion breaths, rope loading, cracking rock, and falling debris. Synchronize the rope arrest and muffled chest impact precisely; keep dialogue intelligible beneath the storm.\n\nnon_diegetic_music:\nN/A",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 7,
      "views": 373,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2096921323624972600",
    "title": "舷窗星海旁的暧昧对话",
    "titleEn": "A Teasing Conversation by the Starship Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "两名女子和一只猫蜷在温暖的飞船舱室里，镜头在杯子、神情和脚部动作间切换，完成一段轻声调侃。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two women and a cat share a warm spacecraft cabin as the camera moves between a mug, subtle expressions, and small gestures during a teasing exchange. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, ultra-realistic. The video starts exactly from <Picture 1>, preserving the warm spacecraft cabin, sofa area, porthole windows, pizza box, old astronaut helmet, candlelight, exposed wall wiring, and the full spatial continuity. The left girl is always A, the right girl is always B. Their identity, face, hairstyle, outfit, voice, and left-right relationship remain unchanged. A stays wrapped in a bathrobe and keeps holding the hot mug throughout. B stays side-lying close to her with both feet lifted behind her. The camera glides low across the pizza box, old helmet, candles, and wall wiring, then settles on both girls and the cat together. A looks down at the steam above the mug, thoughtful and avoiding eye contact; her eyelids are lowered, lips closed, shoulders slightly tucked, breathing shallow and even. B watches A with a restrained smile and softly swaying feet. Stars drift slowly outside, with a faint cool-blue reflection on the planetary curve. Thin steam rises from the mug; candle flames flicker gently without pulsing. [Shot 2] At 00:02.000, the camera gently pushes toward B while keeping her swaying calves and feet visible. B tilts her head slightly, hides a smile, then lets it show as she speaks and keeps watching A afterward. B, in a light teasing voice (S1), says, <d>[English] Daydreaming again? Who about?</d> “again” lifts slightly; there is a short pause before “Who about?”, which lands softer and more curious than pressing. [Shot 3] At 00:04.000, the shot cuts to A in close-up and sinks slightly with her movement. A freezes for a beat, blinks once, raises the mug to shield herself a little, lowers her head, and takes only a small sip. She swallows before speaking, keeping her gaze near the cup; the corner of her mouth almost lifts, then settles. A, in a soft inward voice (S2), says, <d>[English] No one… Don’t guess.</d> “No one” is short and quiet, followed by a brief hesitation; “Don’t guess” is soft, not angry. Drinking and speaking do not overlap. [Shot 4] At 00:06.000, the cat rises between them and the camera follows it naturally back into a medium close two-shot. A lightly strokes the cat’s ear to avoid B’s gaze. B inches closer, her feet still swaying, glances at A’s hands around the mug, then back to A’s face. Her smile deepens gently without exaggeration. B (S1), lowering her voice with playful warmth, says, <d>[English] Why so tight? You’re making the cup blush.</d> The first question is light and quick; “blush” lands with a smile. A begins to lose control of her smile by the end. [Shot 5] At 00:09.000, the camera returns to a slightly front-facing close two-shot favoring A. A lets out a small breathy laugh, shoulders relaxing. Keeping the mug upright in one hand, she lightly taps the back of B’s wrist with the other hand, then gently nudges B with one knee. Her eyes flick briefly toward B’s swinging feet, then toward B’s wrist, still not fully locking eye contact. B’s smile opens a little but she does not overreact. A (S2), smiling now, says, <d>[English] Keep your feet still. You’re flustering me even more.</d> The first sentence is playfully corrective; “even more” softens at the end. [Shot 6] At 00:12.000, the camera follows B in a tiny forward move. B shows a brief victorious smile, then softens; her feet gradually slow instead of stopping abruptly. She strokes the cat without pulling away. A finally raises her eyes and truly looks at her, keeping the faint smile from before and no longer looking away. B (S1) says, <d>[English] Then I’ll stop swinging my feet—and sway you.</d> The first half is light and playful; “sway you” becomes softer and more intimate, expressed only through voice and eyes, with no added pushing or body movement. After the line, the camera slowly pulls back to hold both girls, the cat between them, and the full warm cabin while they maintain eye contact. Dialogue order stays strictly B → A → B → A → B. No extra lines, no swapped roles, no axis crossing, no left-right swap, no cup hand change, no kiss, no hug, no subtitles, no text, no watermark.\n\noverall_soundscape: Low steady spacecraft room tone throughout, with soft fabric movement, subtle sofa friction, faint mug handling, quiet breathing, light cat contact sounds, and gentle foot-sway rustle. Ambient sound stays restrained and never covers the dialogue.\n\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2096921323624972600",
    "sourceUrl": "https://x.com/ou_zhen599/status/2096921323624972600",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-07T11:19:37.000Z",
    "addedAt": "2026-09-08T01:43:38.867Z",
    "mediaUrl": "/media/x-2096921323624972600.mp4",
    "posterUrl": "/posters/x/2096921323624972600.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "sci-fi"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI\n\n🇹🇻🌌Image-to-Video：“Chat History in Space”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, ultra-realistic. The video starts exactly from <Picture 1>, preserving the warm spacecraft cabin, sofa area, porthole windows, pizza box, old astronaut helmet, candlelight, exposed wall wiring, and the full spatial continuity. The left girl is always A, the right girl is always B. Their identity, face, hairstyle, outfit, voice, and left-right relationship remain unchanged. A stays wrapped in a bathrobe and keeps holding the hot mug throughout. B stays side-lying close to her with both feet lifted behind her. The camera glides low across the pizza box, old helmet, candles, and wall wiring, then settles on both girls and the cat together. A looks down at the steam above the mug, thoughtful and avoiding eye contact; her eyelids are lowered, lips closed, shoulders slightly tucked, breathing shallow and even. B watches A with a restrained smile and softly swaying feet. Stars drift slowly outside, with a faint cool-blue reflection on the planetary curve. Thin steam rises from the mug; candle flames flicker gently without pulsing. [Shot 2] At 00:02.000, the camera gently pushes toward B while keeping her swaying calves and feet visible. B tilts her head slightly, hides a smile, then lets it show as she speaks and keeps watching A afterward. B, in a light teasing voice (S1), says, <d>[English] Daydreaming again? Who about?</d> “again” lifts slightly; there is a short pause before “Who about?”, which lands softer and more curious than pressing. [Shot 3] At 00:04.000, the shot cuts to A in close-up and sinks slightly with her movement. A freezes for a beat, blinks once, raises the mug to shield herself a little, lowers her head, and takes only a small sip. She swallows before speaking, keeping her gaze near the cup; the corner of her mouth almost lifts, then settles. A, in a soft inward voice (S2), says, <d>[English] No one… Don’t guess.</d> “No one” is short and quiet, followed by a brief hesitation; “Don’t guess” is soft, not angry. Drinking and speaking do not overlap. [Shot 4] At 00:06.000, the cat rises between them and the camera follows it naturally back into a medium close two-shot. A lightly strokes the cat’s ear to avoid B’s gaze. B inches closer, her feet still swaying, glances at A’s hands around the mug, then back to A’s face. Her smile deepens gently without exaggeration. B (S1), lowering her voice with playful warmth, says, <d>[English] Why so tight? You’re making the cup blush.</d> The first question is light and quick; “blush” lands with a smile. A begins to lose control of her smile by the end. [Shot 5] At 00:09.000, the camera returns to a slightly front-facing close two-shot favoring A. A lets out a small breathy laugh, shoulders relaxing. Keeping the mug upright in one hand, she lightly taps the back of B’s wrist with the other hand, then gently nudges B with one knee. Her eyes flick briefly toward B’s swinging feet, then toward B’s wrist, still not fully locking eye contact. B’s smile opens a little but she does not overreact. A (S2), smiling now, says, <d>[English] Keep your feet still. You’re flustering me even more.</d> The first sentence is playfully corrective; “even more” softens at the end. [Shot 6] At 00:12.000, the camera follows B in a tiny forward move. B shows a brief victorious smile, then softens; her feet gradually slow instead of stopping abruptly. She strokes the cat without pulling away. A finally raises her eyes and truly looks at her, keeping the faint smile from before and no longer looking away. B (S1) says, <d>[English] Then I’ll stop swinging my feet—and sway you.</d> The first half is light and playful; “sway you” becomes softer and more intimate, expressed only through voice and eyes, with no added pushing or body movement. After the line, the camera slowly pulls back to hold both girls, the cat between them, and the full warm cabin while they maintain eye contact. Dialogue order stays strictly B → A → B → A → B. No extra lines, no swapped roles, no axis crossing, no left-right swap, no cup hand change, no kiss, no hug, no subtitles, no text, no watermark.\n\noverall_soundscape: Low steady spacecraft room tone throughout, with soft fabric movement, subtle sofa friction, faint mug handling, quiet breathing, light cat contact sounds, and gentle foot-sway rustle. Ambient sound stays restrained and never covers the dialogue.\n\nnon_diegetic_music: N/A",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 7,
      "views": 1185,
      "snapshotAt": "2026-09-08T01:43:38.867Z"
    },
    "approvedAt": "2026-09-08T01:43:38.867Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2097090946555052294",
    "title": "健身房里的硬拉练习",
    "titleEn": "Deadlift Practice at the Gym",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名女子在明亮健身房反复完成杠铃硬拉，固定竖屏镜头突出动作起落与身体姿态。 原生视频规格为 5 秒 · 768×1088。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman repeats barbell deadlifts in a bright gym while a fixed vertical shot emphasizes the lift cycle and body posture. Native-video output: 5s · 768×1088. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "ジムで筋トレして",
    "promptSourceUrl": "https://x.com/moon_himiko/status/2097090946555052294",
    "sourceUrl": "https://x.com/moon_himiko/status/2097090946555052294",
    "sourceLabel": "X 原帖 · @moon_himiko",
    "author": "Ai_Himiko",
    "publishedAt": "2026-09-07T22:33:38.000Z",
    "addedAt": "2026-09-08T01:43:38.868Z",
    "mediaUrl": "/media/x-2097090946555052294.mp4",
    "posterUrl": "/posters/x/2097090946555052294.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1088",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます\nMiniMax H3 で今日もひみこは筋トレ\npromptは『ジムで筋トレして』コストは8円\n寝る前に”私は戦国武将の誰に似てるのか”と思い診断作ってみた\n毛利元就でした\n戦国武将診断\nhttps://shindan.popopopo.net/sengoku",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 3,
      "views": 73,
      "snapshotAt": "2026-09-08T01:43:38.868Z"
    },
    "approvedAt": "2026-09-08T01:43:38.868Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video."
  },
  {
    "id": "x-2094490203335352524",
    "title": "油彩长成爱丁堡城市",
    "titleEn": "A City Painted into Existence",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "油彩纹理逐步变成街道、建筑与城市海报，过程与最终画面都有记忆点；适合城市宣传、旅行和品牌概念片。",
    "summaryEn": "Oil-paint ridges turn into streets and landmarks before resolving into an Edinburgh poster. The published prompt includes city and quote variables.",
    "prompt": "[CITY_NAME] = EDINBURGH\n[QUOTE] = A city carved from stone and story.\n\nCreate a fast, dense, cinematic 16:9 macro travel film where the entire identity of [CITY_NAME] is continuously painted into existence by living oil paint.\n\nUse a low grazing macro camera with extremely shallow depth of field. The world is a vast canvas covered in thick wet impasto paint, sculpted brushstroke ridges, carved grooves, glossy reflections and dense gold glitter. The paint must behave like an active force at all times. It flows like rivers, sweeps like brushwork, gathers like tides, curls at the edges, spills into new paths and physically constructs the city in real time.\n\nThe film must feel fast, rich and uninterrupted. Use one continuous flowing camera move with no cuts. The camera races across the canvas, skimming over wet paint valleys, weaving between rising structures, accelerating through the city as new landmarks constantly appear ahead. Avoid slow drifting. Keep the visual progression energetic and tightly packed.\n\nStart with abstract moving pigment and rapidly transition into recognizable city formation. Show thick paint strokes being laid down, dragged, folded and pulled into shape by an invisible artistic force. Let streams of pigment split and merge like waterways, while dimensional brushstroke masses rise into streets, bridges, domes, towers, rooftops, plazas, canals and layered miniature architecture. Make the city feel as if it is being painted and built at the same time.\n\nReveal many landmarks across the journey, not just one. Let the camera travel seamlessly through the whole city, with one landmark flowing directly into the next. As the camera advances, fresh paint forms arches, facades, bell towers, canal edges, stairways, statues, porticoes and skyline silhouettes associated with [CITY_NAME]. The transitions between landmarks should feel fluid and continuous, with brushstrokes extending forward to pull the viewer deeper into the city.\n\nMake the paint motion highly visible and expressive. Show pigment actively spreading across canvas, pouring into channels, stacking into forms, curving into structures and leaving wet textured trails behind. Gold glitter should move with the paint and collect along ridges and valleys, creating sparkling highlights that emphasize speed, direction and shape.\n\nUse warm golden key light from the lower left to ignite glossy paint and glitter. Add cool purple-blue ambient fill from the upper right to maintain a dreamy dusk atmosphere. Keep the distant background soft with haze and circular bokeh, but preserve strong clarity on the active foreground paint and newly formed landmarks.\n\nToward the end, the camera rises and pulls back just enough to reveal a broader final view of the fully formed painted city, making it clear that the viewer has traveled through an entire living city built from moving paint.\n\nEnd on a striking travel-poster composition. Place elegant typography in the top left with generous breathing room. Show [CITY_NAME] in refined serif uppercase, with [QUOTE] beneath it in smaller delicate type.\n\nThe overall feeling should be luxurious, poetic and visually intense, with continuous motion, dense landmark reveals and the clear impression that the whole city is being painted alive in real time.",
    "sourceUrl": "https://x.com/aimikoda/status/2094490203335352524",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2094490203335352524.mp4",
    "posterUrl": "/posters/x/2094490203335352524.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "1440p",
    "tags": [
      "油彩长成爱丁堡城市"
    ],
    "category": "advertising",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptSourceUrl": "https://x.com/aimikoda/status/2094490203335352524",
    "promptCompleteness": "complete",
    "featured": {
      "at": "2026-09-08T11:10:35.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2095141166962311555",
    "title": "从骨架到成品的角色手办制作",
    "titleEn": "Building a Character Bust from Scratch",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "从支架、塑形到上色与配饰，能看懂制作过程，适合手工、角色周边与产品展示。",
    "summaryEn": "Hands assemble an armature, sculpt a face and add paint, hair and accessories to a finished character bust. The prompt requires a character reference image.",
    "prompt": "A cinematic creation film follows one maker reconstructing the main subject shown in the referenced image completely from scratch. The referenced image defines the finished subject’s complete visible appearance, proportions, structure, materials, colors, clothing or surface details, and distinctive features; ignore its background, framing, lighting, and unrelated elements.  Begin directly on an empty, clean virtual workbench. Use a stable front three-quarter overhead view that keeps the developing subject readable. There is one maker throughout, represented by the same consistent left and right hands and forearms. Show no more than two hands at once.  From 0 to 14 seconds, the entire creation unfolds as a clearly accelerated timelapse with rapid, purposeful hand movement and restrained motion blur. Short jump cuts compress repetitive manual work only after each action has visibly completed. Every cut inherits the exact form and progress left by the previous action.  0-2 seconds: One hand enters already holding the first foundation material, armature, or base element appropriate to the referenced subject and places it at the center. The second hand steadies it as the maker establishes the initial supporting form.  2-8 seconds: The maker rapidly develops the subject’s major structure and volumes using one coherent creation method appropriate to what the referenced image depicts. A living subject is sculpted as one continuous, non-gory digital form from armature to anatomy; a vehicle or machine is built from chassis to functional structure; an object is formed or assembled from its supporting body outward. Each additional material or component enters from outside the frame while firmly held by one of the maker’s hands, is carried to its destination, and remains under hand control until attached or shaped.  8-12 seconds: The same hands develop the recognizable outer form and reference-specific features. The maker sculpts, fits, wraps, stitches, fastens, carves, or polishes only where appropriate to the subject. Facial features, hair, clothing, body panels, wheels, glass, surfaces, accessories, or equivalent defining elements emerge through visible hand and tool contact, never through spontaneous transformation.  12-14 seconds: The maker refines proportions, edges, joints, surface transitions, textures, colors, and distinctive details until the developing subject closely matches the referenced image. One hand stabilizes the form while the other performs each final adjustment with a hand-held tool.  14-15 seconds: The maker removes the last tool by hand and withdraws both hands. The timelapse returns to normal speed as the camera makes a restrained push toward the completed subject and holds on a clean final view.  Materials and components do not need to be visible before use, but anything newly introduced must enter the frame already held by one of the maker’s hands. Nothing moves, assembles, appears, disappears, or changes material independently. Maintain one maker, one continuous subject, one creation position, and one category-appropriate construction method. No assistants, extra hands, detached anatomy, duplicated elements, magical morphing, drawing phase, software interface, cursor, menus, annotations, or text overlays.",
    "sourceUrl": "https://x.com/aimikoda/status/2095141166962311555",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2095141166962311555.mp4",
    "posterUrl": "/posters/x/2095141166962311555.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "1440p",
    "tags": [
      "从骨架到成品的角色手办制作"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptSourceUrl": "https://x.com/aimikoda/status/2095141166962311555",
    "promptCompleteness": "complete",
    "featured": {
      "at": "2026-09-08T11:10:34.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2096554023809695817",
    "title": "玻璃爆裂并重组成 SHATTER",
    "titleEn": "Glass Shards Reassemble into SHATTER",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "破裂、穿越碎片、重组文字形成完整视觉过程，最终标题清楚；适合片头与品牌字效。",
    "summaryEn": "A glass monolith fractures into flying shards that assemble into a readable title, followed by another burst of fragments.",
    "prompt": "Premium action-film title sequence beginning with a gigantic transparent glass monolith floating against complete blackness. A tiny projectile impacts the exact center and an intricate network of cracks instantly races across the entire surface. The camera pushes directly into one expanding fracture, traveling through the microscopic crystalline structure as tension builds everywhere. Suddenly the monolith violently detonates into millions of glass fragments. The camera performs a fast 360-degree orbital move through the suspended crystal storm as every shard reflects sharp beams of cinematic light. An invisible force begins pulling selected fragments together, constructing the enormous word \"SHATTER\" from overlapping razor-thin glass shards. The completed transparent typography refracts the entire environment like a giant prism before another pressure wave passes through it, producing thousands of secondary fractures. The letters explode outward directly toward the lens in spectacular ultra slow motion. Extremely detailed glass physics, realistic refraction and caustics, elegant brutality, premium Hollywood VFX.",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2096554023809695817",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-09-06",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2096554023809695817.mp4",
    "posterUrl": "/posters/x/2096554023809695817.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "玻璃爆裂并重组成 SHATTER"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2096554023809695817",
    "promptCompleteness": "complete",
    "featured": {
      "at": "2026-09-08T11:10:33.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2094985989404205107",
    "title": "漫画人物走出书页，却遇见另一个自己",
    "titleEn": "Escaping a Manga Book, Meeting Another Self",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "二维漫画到真人的转换有明确叙事，结尾反转补足趣味，适合学习跨画风互动。",
    "summaryEn": "A manga character emerges from the pages into live action and discovers another version still inside. The prompt references multiple images; some panel and speech-bubble exclusions are not followed.",
    "prompt": "【REFERENCE】\nImage1 `01 / SCENE_TRAPPED`：本の中に閉じ込められた第一場面、巨大本、室内、人物の位置とポーズ。\nImage2 `02 / SCENE_BREAKTHROUGH`：ページを押し破る第二場面、動作、紙の亀裂。\nImage3 `03 / SCENE_EMERGENCE`：現実へ出た第三場面、破れ穴、着地位置。\nImage4 `04 / SCENE_ENCOUNTER`：本の中の自分と対面する第四場面の左右、指差し、巨大本。\nImage5 `05 / WOMAN_MASTER`：利用者が差し替える女性全身アンカー。顔、髪、年齢感、肌、体型、全身比率、衣装、素材、配色、靴の唯一の正本。\n`06 / MANGA_SELF`：05を同じ人物・衣装のままモノクロ漫画化した状態。\n`07 / REAL_SELF`：05を正確に維持した実写状態。\n`08 / BOOK_ROOM`：01～04共通の巨大な見開き本、木製床、寝具、カーテン、植物、暖かな昼光の寝室。\n\n各ショットはIDだけで参照する。01～04の女性からは位置、ポーズ、動作だけを使い、外見は使わない。人物同一性は05を最優先し、06と07を同一人物の二状態とする。女性アンカーの差し替えごとに05から読み直す。場面アンカーの吹き出し、印刷文字、コマ枠、白い分割余白は使わない。\n\n【CONDITION DEFINITION】\n15秒、1:1。シネマティック実写と精密なモノクロ漫画の融合。実写は自然な肌、布、髪、暖かな立体光、漫画は紙目、網点、インク線を明瞭に分ける。4パネルは物語アンカーで、カット数は4つに限定しない。前半は真剣な脱出劇、最後は「自分が本の中に残る」不条理を驚きとパロディーへ転換。表情は困惑→恐怖→決意→安堵→違和感→強い驚き。顔を変形させず眉、瞳、口、顎、肩、呼吸を連動する。0.0～4.6秒は全身を本の印刷面内に閉じ込めた平面モノクロ漫画とし、紙の内側から現実側へ押す。肌色、実写衣装、立体化、身体突出、床への接地は4.6秒から。\n\n【SHOT / FLOW】\n0.0～1.0秒【参照：01、06、08】巨大本を低い広角で見せ、本の中へ高速プッシュイン。MANGA_SELFは困惑して見回し、眉が寄り呼吸が浅くなる。\n1.0～3.3秒【参照：01、06】顔、境界を探る手、紙を押す横顔をハードカット。目が泳ぎ口元が震える。紙越しに「ここは……どこ？　出られない……」。継ぎ目を見つけ、恐怖から決意へ。\n3.3～4.6秒【参照：02、06、08】低い斜め前方。踏ん張る靴、紙へ食い込む指、歯を食いしばる顔を三連続で切る。「あいて……！」。集中線と黒インクの圧力リングが手から広がる。\n4.6～6.8秒【参照：02、05、06、07、08】紙が現実側へ膨らみ、手、顔、上半身、脚の順に平面のインクと紙繊維が07の肌、髪、衣装へ連続変換。境界で網点が色粒子へ変わる。カメラは裂け目を半周し、立体化に合わせ被写界深度を浅くする。破裂時、手描きの`バリッ！`が手前へ飛び、黒インク片へ砕け0.35秒以内に消える。女性は一人のまま05を維持する。\n6.8～8.4秒【参照：03、05、07、08】REAL_SELFが紙片と床へ着地し少しよろける。低い追従から顔へ寄る。両手と参照衣装を確認し、驚きが安堵と小さな笑みへ。「出られた……？」。\n8.4～10.3秒【参照：05、07、08】乾いたページ音。BGMに合わせ笑みが止まり、口角が落ち、眉が上がり、瞳だけ本へ動く。つばを飲み恐る恐る振り向く。顔から本へ高速ホイップパンし、肩越しのまま10.3秒の04へ接続。安堵が違和感へ変わる。\n10.3～11.2秒【参照：04、05、06、07、08】左右が読める中広角。左のREAL_SELFと本の中央のMANGA_SELFが同時に指を差し、終端で短くヒットストップ。BGMはレコードスクラッチで急停止。\n11.2～12.8秒【参照：04、05、06】MANGA_SELFへスマッシュズーム。目と口を大きく開き、頬のハッチングが増える。紙越しに「まさか……本当に出てきた！？」。背後に放射状の漫画線。\n12.8～14.0秒【参照：04、05、07】85mm相当でREAL_SELFの顔へスマッシュズーム。05の顔比率を保ち、眉が跳ね、瞳孔が縮み、口が開き、肩が引ける。短い間の後、室内音声で「あなた……誰？」。\n14.0～15.0秒【参照：04、05、07】85mmのまま一段近い驚き顔。顔と衣装はカラー、背景だけ白黒ハーフトーンと集中線へ変わり、巨大な`！？`が頭の後ろから出て一度弾む。本、カメラ、本の順に二度見し、コミカルな一音で終了。\n\n【CAMERA / EDITING】\n広角、顔、手元、低角度、短い回り込み、ホイップパン、交互スマッシュズームを使い、同じ中広角を続けない。動作か表情変化で切り、長い静止、スロー、無目的な360度回転、フェードなし。最終4.7秒は対面、交互リアクション、驚き顔アップの順に加速。\n\n【MOTION GRAPHICS / TYPOGRAPHY】\n集中線、ハーフトーン、インクの圧力リング、短いヒットストップ、紙片の奥行きを使う。`バリッ！`は黒インク、`！？`は白文字に赤いずれ影。わずかにオーバーシュートし、顔を隠さない。表示文字はこの二つだけで字幕や吹き出しにしない。\n\n【SOUND】\noverall_soundscape:\n紙越しと室内の声は同じ基礎声質で距離感だけ変える。紙の張り、破裂、紙片、着地、衣擦れ、ページ音を画面へステレオ同期。台詞を重ねず、追加の声や笑い声なし。\nnon_diegetic_music:\n0.2秒の演出無音を除き全編に約90 BPMのミステリーコメディーBGMを連続使用。ピチカート弦、バスクラリネット、チェレスタ、軽い打楽器。脱出まで上昇し、破裂で低音と打楽器を同期、着地後は一瞬明るくする。10.3秒でレコードスクラッチ後0.2秒だけ無音、以後は低いピチカート、最後は乾いた一音。台詞中はダッキング。\n\n【NEGATIVE】\n場面アンカーの吹き出し、縦書き文字、複数コマ、白い分割余白を出さない。許可した`バリッ！`と`！？`以外の文字、字幕、ロゴ、透かしなし。第三の女性、同一状態の重複、01～04の人物外見流用、別人化、05の人物・衣装・配色の逸脱、広角による顔変形、4.6秒前の肌色・実写化・立体化・身体突出・床接地、解剖学的変形、無表情、10.3秒以降の長い同一構図、指定0.2秒以外のBGM欠落、イベント順の変更を避ける。",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2094985989404205107",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2094985989404205107.mp4",
    "posterUrl": "/posters/x/2094985989404205107.jpg",
    "duration": 15.139,
    "aspectRatio": "1:1",
    "resolution": "1440p",
    "tags": [
      "漫画人物走出书页，却遇见另一个自己"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2094985989404205107",
    "promptCompleteness": "complete",
    "featured": {
      "at": "2026-09-08T11:10:32.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2095357375549182080",
    "title": "真人与二维小精灵同框互动",
    "titleEn": "A Live-Action Muse Meets a Tiny 2D Fairy",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "大小比例和二维、真人的视觉差异明确，手掌接触与模仿动作使互动成立。",
    "summaryEn": "A tiny drawn fairy interacts with a live-action woman through palm contact and mirrored poses. Character reference images are required by the published prompt.",
    "prompt": "Create a 15-second, 16:9, 24fps ultra-photorealistic live-action Y2K fashion film blended with one pure 2D manga fairy.\nHIGHEST-PRIORITY RULES\nExactly TWO characters: ONE fictional 22-year-old Korean female idol + ONE tiny 2D manga fairy. No extras, duplicates, clones, motion-trail copies, or identifiable reflections.\nZERO readable text or numbers anywhere: no captions, logos, labels, signs, UI, watermarks, license plates, posters, or environmental writing.\nCHARACTER LOCK\nWoman: elegant oval face, cool fair skin with realistic texture, almond eyes, glossy pink lips, extremely long silky straight black hair to the waist with thin bangs. Face, hair, proportions, makeup and wardrobe remain identical.\nWardrobe: black satin cropped camisole with lace detail, cropped glossy motorcycle jacket loosely around her arms, black low-rise pleated micro skirt, silver waist chain/jewelry, black over-the-knee boots. Premium, sensual, sophisticated K-pop editorial.\nFairy: tiny PURE 2D cel-shaded manga character with clean outlines, short black hair, elf ears, large pink eyes, pink-and-white cropped top, silver pleated skirt, translucent pink wings. Always flat 2D—never 3D, photorealistic, or humanized.\nENVIRONMENT\nOne continuous bright pink-and-ice-blue neon fashion studio. Black vintage convertible in the center, silver vanity mirror, transparent acrylic chair, chrome spheres, silver props, glossy floor. Style: K-pop comeback film × glossy Y2K campaign × cute high-energy MV.\nPHYSICAL CONTINUITY\nNo teleportation, clipping, spawning, or passing through objects. Fairy must physically fly above, below, around, or beside the windshield, car, mirror, chair, props, and woman. Glossy surfaces show only abstract light, never recognizable reflections.\n\nSHOT FLOW\n0:00–0:01.5 — Low-angle wide shot. Only fairy is visible behind the wheel. She waves, opens wings on beat, flies vertically until fully above the windshield, exits through the open roof, half-spins, then flies toward the woman. Driver seat becomes empty.\n0:01.5–0:03.5 — Reveal woman leaning outside the car door, looking down while adjusting a ring. Fairy curves toward her shoulder and gives a TA-DA pose. The same woman raises her head, looks at fairy, then camera. Diagonal push-in from headlight.\n0:03.5–0:05.5 — Woman flicks one finger toward fairy on a sharp beat. Fairy reacts with one continuous half-spin to camera-right. Woman shifts into a mischievous smile. Fast side arc; car door briefly wipes foreground.\n0:05.5–0:07.5 — Fairy flies around the OUTER edge of the vanity mirror, never through it. Camera follows and finds the same woman seated on the acrylic chair. She crosses her legs and makes a tiny finger heart. Fairy hovers beside her face and imitates it. Woman smiles.\n0:07.5–0:09.5 — Woman opens her palm. Fairy descends continuously and visibly lands, feet making contact. Fairy poses proudly and copies woman’s head tilt. Woman pauses, then smiles warmly. Add a cute electronic bling; push from hand toward faces.\n0:09.5–0:11.5 — Fairy physically takes off from palm first. Woman stands and walks 2–3 confident steps toward center studio. Camera tracks backward with a lateral arc while fairy flies backward in front of her.\n0:11.5–0:13.0 — Woman strikes a sophisticated K-pop pose with convertible diagonally behind. She touches her long hair; fairy imitates fixing her own short hair. Woman notices and breaks into a smile. Slight low-angle push-in.\n0:13.0–0:14.5 — Medium close-up. Woman makes a finger heart near her face while holding her hair. Fairy hovers beside her, wings open, cheerful double-V pose. Both look into camera. Final beat triggers ONE camera flash without washing out the studio.\n0:14.5–0:15.0 — Freeze the ACTUAL previous frame. No new photo, card, border, or white background. Keep neon studio, car, silver props and depth visible. Add only 4–6 small clean pink hearts/stars/sparkles near the corners, inside frame, never covering either character.\nCAMERA & PERFORMANCE\nUse low angles, diagonal views, subtle ultra-wide shots, foreground occlusion, mirror framing, curved push-ins, short tracking, gentle rises/drops. Avoid repetitive static frontal framing.\nEmotion: cool fashion pose → curiosity → playful interaction → amused smile → joyful final pose. Natural, confident, star-like, never childish.\nMUSIC\nBubblegum Pop × Kawaii Pop × Y2K Dance Pop, 125–135 BPM, punchy drums, bouncing bass, crisp claps, sparkling synths. Sync wing opening, finger flick, fairy spin, finger-heart, palm landing, takeoff, hair imitation, final pose and camera flash to major beats.\nNEGATIVE\nNo second human/fairy, duplicates, face drift, hairstyle/color/costume changes, malformed hands, clipping, teleportation, 3D or photorealistic fairy, white ending, Polaroid/frame/PIP, chaotic doodles, readable text, letters, numbers, logos, subtitles, labels, UI or watermarks.",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2095357375549182080",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-09-03",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2095357375549182080.mp4",
    "posterUrl": "/posters/x/2095357375549182080.jpg",
    "duration": 15.402,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "真人与二维小精灵同框互动"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "performance",
      "fantasy"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2095357375549182080",
    "promptCompleteness": "complete",
    "featured": {
      "at": "2026-09-08T11:10:31.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2095096624192352458",
    "title": "REVEAL：黑橙杂志风动态时装片",
    "titleEn": "REVEAL: A Graphic Fashion Editorial",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物、配色和版式风格统一，镜头与动态图形安排具体；可作为时装、编辑设计的案例。",
    "summaryEn": "A restrained fashion portrait is surrounded by oversized typography, rings and distressed layouts in black, orange and cream. The creator publishes a detailed shot plan.",
    "prompt": "Create a 15-second, 16:9, 24fps Korean high-fashion motion graphics film with EXACTLY 13 CUTS.\nSTYLE:\nPremium AAA motion graphics × Korean fashion editorial × experimental graphic design × distressed print campaign.\nCore concept: CALM CHARACTER + EXPLOSIVE GRAPHIC WORLD.\nThe woman stays elegant, restrained and almost still while giant typography, rings, ink strokes, graphic blocks, film strips, scan lines and speed trails move aggressively around her.\nCHARACTER LOCK:\nONE young Korean woman only, age 20–23. Narrow oval face, dark-brown eyes, long wavy black hair, wispy bangs, realistic skin texture, subtle smoky makeup, nude brown-red lips.\nKeep EXACTLY the same face, hair, makeup, body proportions, outfit and accessories in all cuts.\nOUTFIT:\nBlack asymmetric cropped blazer, charcoal thin-strap top, loose black tailored trousers, black belt, thin choker, minimal silver jewelry. High-fashion, sophisticated, NOT gothic or nightclub.\nPALETTE:\nDeep black, charcoal, burnt orange, rust red, dark olive, aged cream. Warm orange rim light + soft cream key light.\nNO blue, purple, pink or cyberpunk neon.\nCUTS:\n01 Graphic opening: giant “RE”, ink strokes, expanding olive ring.\n02 Extreme eye macro with orange scanning ring.\n03 Woman emerges behind giant vertical “REVEAL”.\n04 Four asymmetric editorial panels showing face, profile, outfit and eye details.\n05 Rotating rings around her, giant “FOCUS” behind, snap zoom.\n06 Aged-cream background, huge “BURN”, black ink slashes.\n07 Giant “MOVE”; subtle transparent motion trails from her head turn.\n08 Fast typography barrage: WILD / SOFT / DEEP / REAL.\n09 Seated portrait with editorial targeting circles and graphic marks.\n10 Extreme face close-up through transparent printed-film layers.\n11 Fast horizontal film-strip montage with four restrained poses.\n12 CLIMAX: graphics disappear briefly, she stands perfectly still. On the beat drop, huge 2D shockwave of cream brush strokes, rust-red ink, olive blocks, orange rings, typography fragments and speed lines explodes behind her. NO real explosion or magic.\n13 Final hero frame: giant “REVEAL” typography, woman seated in right third, olive ring and orange lines behind. She slowly looks into camera. Final light sweep, one impact vibration, hard cut to black.\nCAMERA / EDIT:\nHard cuts, graphic match cuts, snap zooms, foreground wipes, typography stretch, film strips, speed ramps and restrained impact shake.\nHer movement stays minimal and controlled. She does NOT dance. The graphics carry the energy.\nMUSIC:\nHard electronic fashion beat, industrial pop, heavy percussion, sub-bass, metallic hits, typography slams and paper snaps.\nCUT12 is the biggest visual and musical climax.\nNEGATIVE:\nNo character changes, duplicate people, outfit changes, blonde hair, Western facial features, cute idol styling, lingerie, gothic cosplay, pink/purple/blue neon, cyberpunk, weapons, fire, real explosions, fantasy magic, unreadable typography or excessive beauty smoothing.\nFINAL LOOK:\nKorean high-fashion model × black designer fashion × burnt orange / rust red / dark olive / aged cream × giant typography × aggressive editorial motion graphics.",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2095096624192352458",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2095096624192352458.mp4",
    "posterUrl": "/posters/x/2095096624192352458.jpg",
    "duration": 14.86,
    "aspectRatio": "16:9",
    "resolution": "2160p",
    "tags": [
      "REVEAL：黑橙杂志风动态时装片"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2095096624192352458",
    "promptCompleteness": "complete",
    "featured": {
      "at": "2026-09-08T11:10:30.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2094305294117429711",
    "title": "黄绿撞色的动漫角色登场短片",
    "titleEn": "Jade Panic: A Yellow-and-Green Character Entrance",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "角色造型、撞色和图形语言统一，动作画面有张力，适合角色宣传与动画视觉设计。",
    "summaryEn": "A stylized anime character moves through bold yellow-and-green graphic compositions, action poses and impact typography. The prompt requires a character reference.",
    "prompt": "15s / 16:9 / high-energy 2D anime pop-punk MV — character-locked\nCreate a 15-second horizontal 16:9 high-energy 2D anime pop-punk music video featuring the exact heroine from the provided character reference. The entire sequence must feel like one continuous, tightly choreographed MV passage; each movement must motivate the next transition.\nCHARACTER LOCK (absolute authority = the reference still)\nTwin high black space-buns with messy volume and warm brown inner strands. Green jade disc clusters pinned into both buns. Straight blunt bangs with a small center split. Huge sharp green eyes, thick black outlines, strong lashes. Small mouth, warm cheek blush, pale skin. Green circular dangling jade earrings. Sleeveless lime-to-jade floral qipao with black piping, black frog-button placket, high collar, and black line-art lotuses on the fabric. Bare shoulders and arms. Stylized proportions: oversized head, compact torso, long simplified limbs, chunky hands. Bold black outlines, flat saturated colors, minimal cel shading, early-2000s TV-anime / handmade cartoon finish. Personality: loud, cheeky, rebellious, playful, hyperactive — the still’s fierce stare becomes motion, not a different face.\n\nDo not change hair color, bun structure, eye color, earrings, qipao cut, floral linework, or outline weight. No pink hair, no red plaid uniform, no extra characters, no 3D, no photorealism, no glossy modern anime redesign.\nGRAPHIC LANGUAGE (from the still)\nPrimary palette: lemon yellow, jade / lime green, ink black, paper white. Motifs: spinning jade discs, qipao lotus line-art, frog-button glyphs, circular earrings as motion graphics, torn yellow paper, black brush X’s, photocopy grain, tape strips, hand-drawn stars and crosses, rough ink splatters. Kinetic type must look stamped, ripped, or screen-printed — never clean corporate sans.\n0.0–2.5s\nStart in motion. She slides in from frame-left, one foot skidding across a clean yellow graphic floor. Body leans forward; both buns, earrings, and qipao hem trail behind. She stomps the second foot down, snaps her chin up, and flashes a mischievous grin over the reference’s fierce eye shape. Impact instantly tiles the yellow field into huge jade-and-black floral blocks. Bold kinetic type “JADE PANIC!” slams in, letters made from ripped yellow paper + green lotus outlines.\n\n2.5–5.0s\nNo pause. Rebound into a small hop and sharp half-turn. Camera swings from front ¾ into a fast side-track as she takes two exaggerated cartoon steps. Feet hit on alternating beats, leaving jade discs, lotus scribbles, tape, crosses, and comic impact marks. Buns and earrings bounce hard; qipao hem flares.\n5.0–7.5s\nSide kick on the beat. Kick momentum spins her. The floral pattern on the qipao stretches into a giant rotating graphic plane and pulls the frame into an abstract 2D punk world: black brush strokes, green lotus line-art, yellow paper shards, jade lightning (mint/cyan accents only as small bolts, never as hair), photocopy texture. She flows straight into a bouncy phrase: head whip → shoulder hit → side step → small kick → fast turn. Poses stay cartoon-exaggerated.\n\n7.5–10.0s\nTurn accelerates; camera orbits close. Environment recomposes to the choreography: checker yellow/green slides, hand-drawn crosses rotate, torn-paper strips snap, oversized comic type hits on drums (BAM, 叩, 炸). She lands hard, feet apart. A huge hand-drawn “BAM!” bursts under her feet and shakes every graphic layer. She rebounds upward immediately with a cocky smile — no held fashion pose.\n10.0–12.3s\nSame continuous body motion, camera racing along it: whip of the head → bright green eye → jade bun ornament → swinging black bun → bouncing earring → frog-button collar → floral qipao panel → stomping foot. Not a beauty-cut montage; one traveling shot. The footfall fires a jade-floral shockwave across the frame.\n\n12.3–15.0s\nRide the shockwave back to full body. Two confident bouncing steps, abrupt pivot, finish facing camera: weight on one leg, shoulders slightly forward, cheeky rebellious grin. Buns and earrings still swinging. Background erupts into layered yellow paper, black brush X’s, white stars, green lotus line-art, spinning jade discs, rough ink. Elements snap into a 2000s punk magazine-cover lockup. Large “JADE PANIC!” type stacks behind her silhouette on the final beat. Hold her recognizable twin-bun / qipao silhouette. No design drift.\nMOTION\nFast, bouncy, limited-animation timing: strong keys, overshoot, smear frames, anticipation, cartoon impacts. Hair buns, earrings, and qipao hem are the secondary rhythm; feet are the downbeat. Transitions grow only from action: skid → pattern flood → kick → rotating floral plane → spin → orbit → stomp → shockwave. No random cuts, no palm-to-lens, no pointing at camera, no outfit/hair morph.\n\nAUDIO\nTight sync to fast Japanese pop-punk vocals, punchy electronic drums, distorted guitar. SFX: boot/foot skids, stomps, paper snaps, cartoon hits, guitar chops, short graphic whooshes. Type slams on snare/crash.\nNEGATIVES\nNo hand reaching toward camera, no palm covering the lens, no pointing into the lens, no outfit transformation, no hairstyle transformation, no extra characters, no realistic rooms, no photorealism, no 3D CGI, no modern glossy anime, no cyberpunk holograms, no slow fashion posing, no facial drift, no random accessories, no excessive jump cuts, no disconnected moves, no uncontrolled morphing, no unreadable type, no pink hair, no red plaid school uniform.",
    "sourceUrl": "https://x.com/Just_sharon7/status/2094305294117429711",
    "sourceLabel": "X 原帖",
    "author": "Just_sharon7",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2094305294117429711.mp4",
    "posterUrl": "/posters/x/2094305294117429711.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "1440p",
    "tags": [
      "黄绿撞色的动漫角色登场短片"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2094305294117429711",
    "promptCompleteness": "complete",
    "featured": {
      "at": "2026-09-08T11:10:29.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2095573685671469310",
    "title": "小怪兽看恐怖片：受惊后还要装镇定",
    "titleEn": "A Movie Scare and a Dignified Recovery",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "小怪兽看电影时突然受惊、躲进毯子，再逐渐恢复镇定。原作者只公开了 Prompt 片段，未公开完整文本。",
    "summaryEn": "A furry one-eyed creature jumps during a movie, hides under a blanket and regains composure. The creator shares only a prompt excerpt, not the complete prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2095573685671469310",
    "sourceLabel": "X 原帖",
    "author": "GlennHasABeard",
    "publishedAt": "2026-09-03",
    "addedAt": "2026-09-08T11:06:31.003Z",
    "approvedAt": "2026-09-08T11:06:31.003Z",
    "mediaUrl": "/media/x-2095573685671469310.mp4",
    "posterUrl": "/posters/x/2095573685671469310.jpg",
    "duration": 15.175,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "小怪兽看恐怖片：受惊后还要装镇定"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "featured": {
      "at": "2026-09-08T11:10:28.004Z",
      "basis": "manual"
    }
  },
  {
    "id": "x-2096833346555023816",
    "title": "海风里的针织时装漫步",
    "titleEn": "A Coastal Walk in Knitwear",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子沿海岸步道散步，镜头穿插山海远景与人物近景。",
    "summaryEn": "A coastal walk alternates mountain and sea views with close portraits.",
    "prompt": "A cinematic fashion film set on a breathtaking coastal landscape. A stylish young woman walks naturally along a rugged seaside path, surrounded by wild greenery and delicate flowers. She wears a relaxed oversized brown knit sweater with elegant ivory trousers. Behind her, crystal-blue ocean waters stretch toward a towering lush green mountain rising from the sea.\n\nSoft golden-hour sunlight creates a warm, dreamy atmosphere. A gentle ocean breeze naturally moves her hair and sweater. Start with a wide cinematic establishing shot of the coastline, then smoothly track alongside her as she walks, transitioning into elegant close-ups of her outfit, hair, and subtle facial expressions.\n\nSpoken Voice / Dialogue: The woman speaks naturally while walking, in a soft, calm, confident female voice. Her delivery should feel spontaneous and conversational, with realistic breathing and subtle pauses. She says:\n\n“Sometimes, all you need is a quiet place, a little fresh air, and a moment to simply slow down and enjoy where you are.”\n\nHer lip movements must accurately synchronize with every spoken word. Natural facial expressions, realistic mouth movement, and subtle head gestures while speaking. No exaggerated acting.\n\nAmbient Audio: Gentle ocean waves crashing against the rocks, soft coastal wind, subtle rustling of flowers and leaves, quiet footsteps, and distant seabirds. Keep the environmental sounds natural and cinematic underneath the dialogue. No background music and no additional voices.\n\nPhotorealistic, natural skin and hair, realistic fabric movement, cinematic lighting, smooth camera movement, shallow depth of field, soft bokeh, subtle film grain, premium fashion-commercial aesthetic, atmospheric color grading, 4K cinematic quality. No text, no logos, no watermark.",
    "sourceUrl": "https://x.com/CaliraVal/status/2096833346555023816",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-09-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2096833346555023816.mp4",
    "posterUrl": "/posters/x/2096833346555023816.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "时尚",
      "自然风景"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "fashion",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2096833346555023816"
  },
  {
    "id": "x-2096817125864484954",
    "title": "从圣索菲亚到博斯普鲁斯的旅行",
    "titleEn": "From Hagia Sophia to the Bosphorus",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "旅行者从城市地标走到海峡渡船，最后以日落天际线收尾。",
    "summaryEn": "A traveler moves from city landmarks to a ferry before a sunset skyline closes the sequence.",
    "prompt": "A continuous 4K travel vlog video featuring a young East Asian woman with long dark hair, wearing a white crop top, open white shirt, and beige shorts, exploring Istanbul. The video seamlessly sequences through dynamic travel moments: starting with a close-up ASMR intro whispering into a mic near Hagia Sophia, transitioning to a handheld selfie camera spinning through a sunny square filled with tourists. Cut to a low-angle shot inside a historic mosque gazing up at grand domed ceilings and Ottoman arches, followed by a golden-hour rooftop moment near Galata Tower with her hair blowing softly in the breeze. She then stands on a Bosphorus ferry railing smiling at the camera, walks down a narrow cobblestone alley lined with outdoor cafes, offers a cup of Turkish coffee close to the lens in a cozy cafe, and ends with a wide cinematic twilight shot of the illuminated Istanbul skyline across the water. Hyper-detailed, photorealistic, 8k resolution, cinematic lighting, vibrant natural colors, fluid movement, photorealistic face and physics.",
    "sourceUrl": "https://x.com/Zyrellix/status/2096817125864484954",
    "sourceLabel": "X 原帖",
    "author": "Zyrellix",
    "publishedAt": "2026-09-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2096817125864484954.mp4",
    "posterUrl": "/posters/x/2096817125864484954.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Zyrellix/status/2096817125864484954"
  },
  {
    "id": "x-2096807169945731290",
    "title": "粉色首尔时装片的图形切换",
    "titleEn": "Pink Seoul Fashion in Graphic Cuts",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "粉色街头造型与大字版式交替出现，人物穿行于城市和纯色背景之间。",
    "summaryEn": "Pink streetwear and oversized type alternate as the model moves between city scenes and solid backdrops.",
    "prompt": "Create a 15s, 9:16, 24fps high-energy anime fashion commercial for “SOLVÉRA — PINK SEOUL MODE.”\nStyle: K-pop girl-group MV × Korean Y2K × Pink Cyber × Seoul street fashion × photo-booth aesthetics × cute luxury. Pink, glossy, playful, sexy, stylish, rebellious and premium. Adult fashion only, never childish or schoolgirl.\nCharacter: One fictional adult Korean anime woman in her 20s with refined Korean features, fair luminous skin, long dark-brown hair, glossy lips, slim feminine figure, long legs and confident K-pop idol energy. Keep exactly the same face, hairstyle, age and body proportions in every shot. Outfits may change.\nPalette: hot pink, baby pink, blush pink, glossy white, chrome silver, small black accents.\nEvery shot combines character movement + dynamic camera movement + foreground graphics. Use pink bows, chrome hearts, stickers, rhinestones, plush charms, photo-booth cards, holographic stars and bold Korean typography.\n0–1s — HOOK\nExtreme close-up. Camera rushes through pink bows and heart charms. She lowers pink sunglasses and looks directly at camera. Text: “PINK!” / “오늘도 핑크.” Giant heart wipes lens.\n1–3s — SEOUL STREET\nShe confidently walks through a neon pink Hongdae street in a fitted pink mini dress, white jacket and silver boots. Low-angle tracking, strong parallax, flying stickers and hearts. Text: “SO CUTE.”\n3–5s — Y2K SWITCH\nPink handbag wipes lens. New baby-pink Y2K look. She spins and looks back while camera orbits. Korean photo-booth frames appear around her. Text: “PINK MODE.”\n5–7s — K-POP / CYBER\nFast cut to glossy K-pop stage, then futuristic pink Seoul subway. She performs a short dance move and reaches toward lens. Chrome hearts and neon Hangul orbit her. Text: “NO RULES.”\n7–9s — GRAPHIC EXPLOSION\nHuge pink Korean typography: “핑크는 자유야.” Camera flies through letters, torn paper, glitter, stickers and chrome graphics. Cut to her stepping from a pink elevator in a deep-pink mini dress. Text: “TOO CUTE?”\n9–10.5s — MAXIMUM PINK\nRapid pull-back reveals a surreal pink world filled with bows, plush charms, handbags, photo cards, headphones and floating hearts. She raises one eyebrow. Text: “아직 부족해.” Everything suddenly stops.\n10.5–12s — SOFT BREAK\nPure blush-pink background. Pale-pink mini dress and oversized white blazer. She walks forward slowly for two steps. Minimal graphics. Only bell and footsteps.\n12–15s — FINAL EXPLOSION\nBeat returns. Photo-booth cards of previous looks fly around her in a 3D collage. End on hot-pink hero background. She gives a confident cute-sexy pose and small smile.\nHuge text hits with beat:\n“CUTE”\n“IS POWER”\n“귀여움은 힘이야.”\n“SOLVÉRA — PINK SEOUL — OUT NOW”\nMusic: K-pop hyperpop × bubblegum bass, ~150 BPM, punchy kick, bright synths, bells, cute vocal chops and camera shutters.\nAvoid: face drift, body changes, inconsistent character, chibi, schoolgirl styling, childish proportions, static posing, weak camera motion, unreadable text, generic cute visuals.\nFinal feel: a premium pink Korean K-pop fashion campaign — cute, sexy, Y2K, energetic, glossy and unforgettable.",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2096807169945731290",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-09-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2096807169945731290.mp4",
    "posterUrl": "/posters/x/2096807169945731290.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "商业广告感",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2096807169945731290"
  },
  {
    "id": "x-2096555195618689457",
    "title": "草莓酱浇上芝士冰淇淋",
    "titleEn": "Strawberry Sauce over Cheesecake Ice Cream",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "冰淇淋、流动果酱和包装特写组成甜品广告；原帖附的是分镜图提示词。",
    "summaryEn": "Ice cream, flowing sauce and packaging close-ups form a dessert ad; the post supplies a storyboard-image prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/itxabdullaa/status/2096555195618689457",
    "sourceLabel": "X 原帖",
    "author": "itxabdullaa",
    "publishedAt": "2026-09-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2096555195618689457.mp4",
    "posterUrl": "/posters/x/2096555195618689457.jpg",
    "duration": 10.208,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2096297619391975646",
    "title": "皇帝新衣的金色时装演绎",
    "titleEn": "The Emperor's New Clothes in Gold",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "金色披风人物与首饰细节、大幅字体共同构成角色时装展示。",
    "summaryEn": "A gold-caped character appears alongside jewelry details and oversized typography.",
    "prompt": null,
    "sourceUrl": "https://x.com/airina_xyz/status/2096297619391975646",
    "sourceLabel": "X 原帖",
    "author": "airina_xyz",
    "publishedAt": "2026-09-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2096297619391975646.mp4",
    "posterUrl": "/posters/x/2096297619391975646.jpg",
    "duration": 15.146667,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "动态图形",
      "时尚",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2095875210562142588",
    "title": "像素游戏风的动漫角色开场",
    "titleEn": "A Pixel-Game Anime Character Intro",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "荧光色动漫角色穿插游戏机、星星与像素字体，形成角色开场片。",
    "summaryEn": "A neon-colored anime character appears among game consoles, stars and pixel typography.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2095875210562142588",
    "sourceLabel": "X 原帖",
    "author": "Mayz1169",
    "publishedAt": "2026-09-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095875210562142588.mp4",
    "posterUrl": "/posters/x/2095875210562142588.jpg",
    "duration": 14.117732,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2095872189337870433",
    "title": "早餐角色冲向周末终点",
    "titleEn": "Breakfast Characters Race into the Weekend",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "咖啡杯与拟人早餐角色在赛道上冲向写有周末的终点横幅。",
    "summaryEn": "A coffee cup and animated breakfast characters race toward a weekend finish banner.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2095872189337870433",
    "sourceLabel": "X 原帖",
    "author": "altfortomorrow",
    "publishedAt": "2026-09-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095872189337870433.mp4",
    "posterUrl": "/posters/x/2095872189337870433.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "美食",
      "动物与生物"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "food",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2095867010919845981",
    "title": "丝绸背景上的钻石戒指特写",
    "titleEn": "Diamond Ring Details against Silk",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "戒指在深色丝绸背景上展示金属结构和宝石细节。",
    "summaryEn": "A ring reveals its metalwork and gemstone details against dark silk.",
    "prompt": null,
    "sourceUrl": "https://x.com/Kiber_Alla/status/2095867010919845981",
    "sourceLabel": "X 原帖",
    "author": "Kiber_Alla",
    "publishedAt": "2026-09-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095867010919845981.mp4",
    "posterUrl": "/posters/x/2095867010919845981.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2095857062987120666",
    "title": "湖畔蓝裙与花束",
    "titleEn": "A Blue Dress and Flowers by the Lake",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子手持花束站在明亮湖岸，蓝色裙装与远山湖水呼应。",
    "summaryEn": "A woman holds flowers on a bright lakeshore, with a blue dress echoing the water and distant mountains.",
    "prompt": "A cinematic medium shot of a gentle East Asian young woman with dark hair styled in a bun adorned with a blue-and-white checkered bow, standing on a sunny pebbled lakefront. She wears a sleeveless light blue polka-dot tulle dress with long, translucent white ribbons flowing from her shoulder that flutter gracefully in a gentle breeze. Holding a lush bouquet of pastel pink, baby blue, and white flowers, she smiles warmly at the camera, softly touches the flower petals, and tilts her head back slightly to close her eyes and feel the breeze. In the background, glistening blue lake water sparkles under the sun, bordered by tall green reeds along the shore and serene blue mountains under a bright, clear sky, captured in soft natural daylight with a shallow depth of field.",
    "sourceUrl": "https://x.com/TaliaAariz/status/2095857062987120666",
    "sourceLabel": "X 原帖",
    "author": "TaliaAariz",
    "publishedAt": "2026-09-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095857062987120666.mp4",
    "posterUrl": "/posters/x/2095857062987120666.jpg",
    "duration": 15.168,
    "aspectRatio": "4:7",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "时尚",
      "自然风景"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TaliaAariz/status/2095857062987120666"
  },
  {
    "id": "x-2095846567853093063",
    "title": "快递送进动物医院手术室",
    "titleEn": "A Delivery Enters the Veterinary Theater",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "抱着纸箱的男子进入动物医院，手术台周围的医护人员转身回应。",
    "summaryEn": "A man carrying a parcel enters a veterinary theater as the staff around the operating table turn toward him.",
    "prompt": "\"integrated_multimodal_description: [Shot 1] Stylized 3D Pixar-animated look, soft rounded\ncharacter design, expressive exaggerated eyes, glossy cartoon textures, warm cinematic lighting,\n16:9, fast editorial comedic energy with snap reframes. A veterinary hospital back corridor with\nscuffed skirting, laminated posters and strip lighting. A medium tracking shot follows OSKAR, a\nstylized man in his forties in a courier's fleece with a lanyard swinging, arms wrapped around a\nlarge cardboard box with airholes punched in the lid. He hurries along scanning door numbers.\nOskar (S1) mutters to himself in a fast anxious voice, <d>[English] Reception. Just needs a\nsignature. Reception.</d>\n[Shot 2] At 00:02.500, the camera cuts to a wide shot as he backs through the wrong door and\nturns around into a bright operating theatre — four masked surgical staff in scrubs standing\naround a table under a huge overhead lamp, every pair of eyes above the masks turned toward him\nin complete silence. A monitor beeps steadily. The box slips an inch in his arms.\nOskar (S1) says in a very small voice, <d>[English] Is this not reception.</d>\n[Shot 3] At 00:05.500, the camera cuts to a medium shot of a formidable surgeon, a woman in her\nsixties, gloved hands held up and clear, mask pulled tight. She looks him up and down once, then\njerks her chin firmly at the empty space beside the table. The surgeon (S2) says in a brisk,\ncompletely certain voice, <d>[English] Finally. Scrub in, we're waiting.</d> Trays rattle.\nEveryone shuffles round to make room for him.\n[Shot 4] At 00:08.500, the camera cuts to a fast montage of Oskar bluffing at the table, box now\nabandoned on a stool behind him — snapping on gloves far too enthusiastically, holding out a flat\npalm with enormous authority, then nodding gravely at a monitor he clearly cannot read. A nurse\nslaps an instrument into his hand. He holds it up to the light and turns it round twice. Every\nstaff member responds instantly and correctly to each gesture, working faster and more smoothly\nthan before he arrived. Oskar's face cycles through panic, disbelief and delight.\n[Shot 5] At 00:12.000, the camera cuts to a wide shot as the door swings open and a young vet in\nscrubs hurries in tying her mask, sees the operation already running, and stops. The whole room\nturns between the two of them. Oskar is already moving — he peels the gloves off, scoops up the\ncardboard box, and backs toward the door giving the room a double thumbs up. Oskar (S1) says in a\nbright rushed voice, <d>[English] Great work everyone, sign here next time.</d> He is gone. The\nsurgeon watches him go with visible approval. The camera holds still on the room.\n\noverall_soundscape: A theatre room tone with a steady monitor beep and the hum of overhead lamps.\nFootsteps and cardboard shifting in a corridor. A heavy door swinging on its hinge. Complete\nsilence with one instrument tray tick. Wheels rolling on hard flooring. Latex gloves snapping.\nMetal instruments slapped into a palm. Muffled voices behind masks. Footsteps leaving fast and a\ndoor swinging shut.\n\nnon_diegetic_music: A light plucked pizzicato strings figure at a brisk tempo opens the video and\ncuts out completely at 00:02.500 into one beat of total silence. A single low bassoon note\nfollows. From 00:08.500 a tense, mock-serious strings and low percussion figure builds through the\nmontage, playing the situation completely straight while getting more absurd, and cuts off on one\nhard comedic sting as he reaches the door at 00:14.000.\"",
    "sourceUrl": "https://x.com/ManuAGI01/status/2095846567853093063",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-09-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095846567853093063.mp4",
    "posterUrl": "/posters/x/2095846567853093063.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2095846567853093063"
  },
  {
    "id": "x-2095732264889884918",
    "title": "街头边看手机边演唱",
    "titleEn": "A Street Performance with a Phone",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子站在城市街边，拿着手机对镜头做连续表演。",
    "summaryEn": "A woman holds a phone and performs toward the camera on a city sidewalk.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI_VideoLab/status/2095732264889884918",
    "sourceLabel": "X 原帖",
    "author": "AI_VideoLab",
    "publishedAt": "2026-09-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095732264889884918.mp4",
    "posterUrl": "/posters/x/2095732264889884918.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "生活实拍感",
      "人物表演",
      "城市街景"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2095362627677438174",
    "title": "小狐狸仰望湖上的满月",
    "titleEn": "A Fox Cub Gazes at the Full Moon",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "小狐狸坐在树下湖边，在萤火虫与月光之间仰头看向夜空。",
    "summaryEn": "A fox cub looks up from a lakeside tree amid fireflies and moonlight.",
    "prompt": "A tiny adorable orange fox cub named Milo lives in a cozy little forest beneath a giant ancient oak tree. One magical night, Milo looks up at the enormous glowing full moon reflected in a quiet lake. He reaches his tiny paw toward the moon, wishing he could touch it.\n\nA gentle breeze moves through the grass and fireflies sparkle around him. Milo suddenly notices a beautiful glowing feather floating down from the sky. His eyes widen with wonder.\n\nCamera: Slow cinematic push-in toward Milo, then tilt upward toward the glowing moon.\n\nMood: Magical, peaceful, emotional.",
    "sourceUrl": "https://x.com/aniyaintel/status/2095362627677438174",
    "sourceLabel": "X 原帖",
    "author": "aniyaintel",
    "publishedAt": "2026-09-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095362627677438174.mp4",
    "posterUrl": "/posters/x/2095362627677438174.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动物与生物",
      "自然风景"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/aniyaintel/status/2095362627677438174"
  },
  {
    "id": "x-2095362596962795795",
    "title": "被重力支配的纯白未来城市",
    "titleEn": "A White Future City Ruled by Gravity",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "白色巨构建筑、密集人流与弧形空间交替呈现城市的尺度。",
    "summaryEn": "White megastructures, crowds and curved spaces convey the scale of a futuristic city.",
    "prompt": "integrated_multimodal_description:\n15秒、1:1スクエア、参照画像なし。高品質な日本の2D劇場アニメーションによるオリジナルゲーム世界観PV。主役は人間ではなく、重力によって秩序を強制する巨大アーキテクチャと空間そのもの。人物は多数の匿名市民のみ。民衆は都市の規模、圧力、強制された動線を可視化するために存在し、特定人物を追わない。\n世界は白〜淡灰色を基調とした明るい巨大都市。黒は深い影、細い構造線、タイポやMGの一部に限定し、黒背景や黒い無限空間にしない。各区間で完成建築の形、奥行き、画角を明確に変え、同じ階段・回廊・橋の反復にしない。\nこの世界では重力方向そのものが統制装置である。重力変化によって上下関係、床と壁、建築層、群衆の進行方向が再定義される。ただし変形途中を長く滑らかに見せない。連続モーフィング、流体的な補間、長い回転運動を避ける。強い完成画を短く成立させた直後、瞬間断裁、スナップ移動、ハードカット、急激な画角変更によって次の完成状態へ飛ぶ。静止に近い強い構図と瞬間的な高速変化を交互に置き、緩急を作る。\nタイポグラフィとエディトリアルMGは積極的に使用するが主役にはしない。特定の単語を各SHOTの発動スイッチとして割り当てない。短い英単語、部分文字、巨大なクロップ文字、画面端で切れた文字形、細いライン、面、グリッド断片が建築と同時に現れ、断裁、圧縮、レイヤーずれ、構図変更へ参加する。文字は一瞬だけ可読でもよく、常に全文を読ませる必要はない。タイポ単独の画面、中央固定タイトル、説明UI、カード、HUDにはしない。\nカメラは超広角、強い正面構図、垂直構図、高俯瞰、極端な建築ディテールへの瞬間的な寄り、スナップズームを交差させる。長いパン、長い回転、滑らかな連続追従を主体にしない。画面が常に動き続けるのではなく、「完成画→瞬間変化→完成画」という断続的なリズムを15秒全体で維持する。\n人物中心の芝居、単一の可動建築デモ、大量の可読文字、タイポだけの抽象映像、同じ転換の反復、爆発や瓦礫による通常の都市破壊、ネオン主体のサイバーパンク、実写、写実的3DCGは避ける。\n\n[Shot 1]\n巨大な白い都市の完成画から即開始。多層公共空間、遠景まで続く建築群、複数階に流れる民衆を超広角で強く見せる。約1秒は完成された都市として成立させる。突然、遠景の建築層だけが斜め下へ瞬間的に落ち、前景との上下関係が破断する。巨大な部分文字と建築エッジが同時に画面を切断し、一度強くスナップして、全く異なる垂直都市の完成画へ切り替わる。\n\n[Shot 2] At 00:02.100,\n極端に深い垂直都市峡谷の完成画。巨大壁面、細い空中通路、複数高さの民衆を短く見せる。カメラが急上昇した直後に一瞬止まり、次の瞬間だけ重力方向が90度変化。壁面と床の関係が一発で入れ替わる。タイポ断片、窓列、構造線が一瞬だけ3層へずれ、ハードカットで巨大内部空間へ飛ぶ。\n\n[Shot 3] At 00:04.100,\n巨大な明るい内部空間。円環状の床、巨大梁、複数高さを横切る民衆を強い対称構図で見せる。約0.7秒後、建築ディテールへ極端なスナップズーム。巨大梁と部分文字が画面いっぱいを横切る瞬間にハードカットし、その梁の形状を次の外部都市の地平線へマッチさせる。直後に円環構造が一瞬だけ縦位置へ跳び、新しい巨大外壁として完成する。変形過程ではなく、状態Aと状態Bの差を強く見せる。\n\n[Shot 4] At 00:06.200,\nここから編集密度を上げる。高層テラス都市の完成画を短く見せる→瞬間断裁→巨大アーチ都市の完成画→ハードカット→垂直回廊都市の完成画。各完成画は0.6〜1.0秒程度でも一度明確に成立させる。前画面の主要な建築形状を次画面へ引き継ぎ、同じ構造が別用途・別方向の建築へ瞬時に再構築される。高層テラスの水平層は次画面の巨大アーチへ、アーチの反復形状は次の垂直回廊群へ接続する。タイポ、MG、建築面の線がカット境界と同期し、前画面の形状を利用して次の構図を瞬間的に作る。ワイド、極端な寄り、俯瞰を高速で切り替え、同じカメラ運動を続けない。\n\n[Shot 5] At 00:09.100,\n最大密度。巨大都市の超広角完成画→一瞬の高俯瞰→建築ディテールの極端な寄り→再び超広角を短いハードカットで交差させる。その間に都市全域の重力軸が破断し、巨大な建築層が一斉に異なる方向へ落下する。ただし落下を長く追わず、落下開始の瞬間、途中の強い断片、再構築後の完成状態をスナップするように切り替える。巨大な部分文字とMGも建築層と同時に断裂・圧縮され、次の構図に断片だけ残る。複数の完成都市画が短く連打された後、全建築層が一瞬で別の上下関係へロックされ、新しい巨大都市の完成画になる。\n\n[Shot 6] At 00:13.100,\n再構築された巨大都市をワイドで強く見せる。旧壁面を民衆が床として歩き、旧階段や橋は全く別の方向へ接続されている。終盤は流れるような大変形を続けず、完成した新秩序を見せる。遠景の一つの巨大建築層だけが最後に短くスナップして最終位置へロック。タイポとMGはほぼ消え、最後の0.5秒は建築と群衆の小さな動きだけを残し、世界そのものの強いキービジュアルで終了する。\n\noverall_soundscape:\n台詞なし、ナレーションなし。遠い群衆音、巨大空間の反響、低い重力圧、建築層の摩擦、乾いた断裁音、短く重いロック音を使用する。連続した流体音ではなく、短い静けさと瞬間的な衝撃音を交互に置く。ハードカット、スナップズーム、瞬間断裁、再構築のタイミングへ音を鋭く同期させる。\n\nnon_diegetic_music:\nBGMあり。低い弦、乾いた低音打楽器、短い金属打楽器を主体とした暗く緊張感のある劇伴。一定に流れ続ける音楽にせず、短い間、アクセント、急激な密度変化を使って編集の断続性を強調する。00:06.200以降で密度を上げ、00:09.100〜13.100で最大密度。終端は急速に収束し、低い弦の余韻だけを残す。電子シンセ主体のダンスビートにはしない。",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2095362596962795795",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-09-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095362596962795795.mp4",
    "posterUrl": "/posters/x/2095362596962795795.jpg",
    "duration": 15.13941,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "风格化与实验",
      "城市街景",
      "科幻太空"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2095362596962795795"
  },
  {
    "id": "x-2095357869595975893",
    "title": "动漫女孩穿越真实城市",
    "titleEn": "An Anime Girl Across Live-Action Cities",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "二维女孩出现在夜间街道、现代高楼和城市广场中，以旋转运镜连接场景。",
    "summaryEn": "A drawn girl appears in night streets, modern towers and city squares connected by rotating camera moves.",
    "prompt": "## REFERENCE\nUse Reference Image 1 as the identity reference for the girl. Preserve her identity throughout.\nShe is a clean 2D anime-illustrated character composited into a cinematic live-action city environment.\n## CORE CONCEPT\n15-second sequence built from rotating whip-pan transitions between cinematic city street locations.\nEvery transition: the top of the frame swings right, the bottom swings left, with directional motion blur. The outgoing street exits this way; the next street enters the same way and settles level.\nThe girl stays anime-style, stable, grounded — only the camera/environment rotates.\nEach location carries a different outfit suited to its setting, time of day, and mood. Clothing changes happen only between locations, never through visible morphing.\n## LOCATION & OUTFIT PROGRESSION\n1. Original outfit — neon-lit downtown crossing at night, wet asphalt reflections\n2. Elegant black dress — upscale glass-tower/hotel street at night, warm lighting\n3. Red stage costume — live-house/club street, colorful signage, blurred crowd\n4. White blouse, dark skirt — daytime business district, glass office buildings\n5. Casual modern outfit — sunny shopping/residential street, convenience store front\n6. Dark purple fantasy outfit — dusk alley with izakaya lanterns and vending machines\n7. Return to the original outfit — back to the original neon crossing\nMaintain the same identity and outfit-appropriate look across all seven locations.\n## TIMELINE\n0.0–2.0s: centered at the neon crossing. Slow push-in. Whip-pan begins at the end.\n2.0–3.5s: crossing exits via whip-pan; hotel street enters, settles level, revealing the black dress. Brief push-in.\n3.5–5.0s: black dress/hotel street, medium shot closing in. On the beat, whip-pan reveals the club street/red stage costume.\n5.0–6.5s: club street, upper-body framing, short zoom. Fast whip-pan reveals the business street/white blouse-dark skirt.\n6.5–8.0s: business street, full-body, subtle hair/clothing movement. Whip-pan reveals the shopping street/casual outfit.\n8.0–9.5s: shopping street, slow push-in to upper body. Fast whip-pan reveals the izakaya alley/dark purple outfit.\n9.5–11.0s: izakaya alley, medium → upper-body framing, short zoom and lateral movement. Strong whip-pan with motion blur reveals another clearly different street/outfit.\n11.0–12.5s: new location/outfit. Whip-pan settles level; girl stays stable in frame.\n12.5–14.0s: three rapid location+outfit changes on beat, same whip-pan each time. Full body → upper body → chest-up. Each transition reveals a clearly different street/outfit.\n14.0–15.0s: return to the original neon crossing, centered. Gentle whip-pan into level alignment. Slow push-in. End on a clean centered frame.\n## ROTATING WHIP TRANSITION\nEvery transition: top of frame swings right, bottom swings left, with directional motion blur, simultaneous with translation. Roughly 10–30° of apparent rotation per transition. Smooth momentum, controlled easing. Reads as a camera whip-pan/spin cut through real locations — not an illustrated panel sliding.\n## CRITICAL ROTATION RULE\nRotation always moves the same way: top of frame tilts right, bottom tilts left. Never reverse mid-sequence. Never make the girl herself spin — only the camera/environment rotates. Her body, head, arms, and legs stay stable and grounded in each shot.\n## CAMERA / EFFECTS\nPush-in/pull-back, rotating whip-pan, directional motion blur, slight scale changes, brief impact shake, beat-synced hard cuts, short freeze moments. Keep effects clean and controlled.\n## VISUAL STYLE\nstyle_separation:\n  character: clean 2D cel-shaded anime illustration, faithful to Reference Image 1, non-sexualized design\n  environment: cinematic live-action city street style — accurate architecture, signage, blurred/anonymous pedestrians, realistic lighting/reflections/materials\nretained_elements: the girl's anime style never shifts toward realism; the streets never shift toward illustration.\nVisual variety comes from location, lighting, outfit, and framing while identity and rendering style stay identical.\n## CHARACTER CONSISTENCY\nPreserve identity from Reference Image 1. Only outfit, pose, framing, location, and lighting may change. No additional characters or duplicates. Background pedestrians stay anonymous/blurred and never resemble the girl.\n## RESTRICTIONS\nGirl stays upright and grounded — body never spins. Clothing changes only between locations — no morphing. Anatomy stays correct — no extra limbs or uncontrolled transformations. No readable text, logos, watermarks, or commercial branding. Camera shake stays brief and controlled — no excessive shake or random unmotivated scene changes.\nPriority: stable character identity, coherent cinematic city environments, clear outfit changes matched to each location, smooth whip-pan transitions with consistent direction, precise beat synchronization.",
    "sourceUrl": "https://x.com/yudotanaka/status/2095357869595975893",
    "sourceLabel": "X 原帖",
    "author": "yudotanaka",
    "publishedAt": "2026-09-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095357869595975893.mp4",
    "posterUrl": "/posters/x/2095357869595975893.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "照片级写实",
      "城市街景",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/yudotanaka/status/2095357869595975893"
  },
  {
    "id": "x-2095240665588486165",
    "title": "阿拉丁逃出巴格达街市",
    "titleEn": "Aladdin Escapes through Baghdad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "阿拉丁从囚室逃入拥挤街市，接连穿过摊位和追逐场面。",
    "summaryEn": "Aladdin escapes a cell into a crowded market, moving through stalls and chase sequences.",
    "prompt": null,
    "sourceUrl": "https://x.com/AITalesNBH/status/2095240665588486165",
    "sourceLabel": "X 原帖",
    "author": "AITalesNBH",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095240665588486165.mp4",
    "posterUrl": "/posters/x/2095240665588486165.jpg",
    "duration": 22.546576,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "城市街景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2095100013315776751",
    "title": "地下车库里的霓虹斩击",
    "titleEn": "Neon Sword Strikes in a Parking Garage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "紫色运动服剑客在地下车库挥出明亮的弧形剑光。",
    "summaryEn": "A swordsman in a purple tracksuit swings bright arcs of light through a parking garage.",
    "prompt": null,
    "sourceUrl": "https://x.com/akemiwork/status/2095100013315776751",
    "sourceLabel": "X 原帖",
    "author": "akemiwork",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095100013315776751.mp4",
    "posterUrl": "/posters/x/2095100013315776751.jpg",
    "duration": 15.13941,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2095091133269647852",
    "title": "霓虹屋顶上的西装舞步",
    "titleEn": "Suited Dance Steps on a Neon Rooftop",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "戴墨镜的西装男子在城市屋顶起舞，镜头切换人物和脚步细节。",
    "summaryEn": "A suited man in sunglasses dances on a city rooftop, with cuts between his portrait and footwork.",
    "prompt": "（高品質なアニメーション・ミュージックビデオ、洗練された大人の男性が主役、スタイリッシュで色気のあるダンスMV）\n\n（キャラクターデザイン、顔、髪型、サングラス、ひげ、服装を全カットで一貫して維持）\n\n夜の都会にある巨大な高級ホテルの屋上。\n背景には無数の都会の光、遠くの高層ビル、ネオン、淡い霧。\nブラック、ディープブルー、アンバー、ゴールドを中心としたシネマティックな色彩。\n床には薄く水が張られており、ネオンと照明が反射している。\n軽い風でジャケットの裾とシャツが自然になびく。\n\n男性が音楽のビートに合わせて、一人でスタイリッシュに踊る。\n激しく飛び跳ねるダンスではなく、大人の余裕を感じるクールなダンス。\nゆっくりした肩のロール、身体を斜めに入れるステップ、滑るようなフットワーク、軽いボディウェーブ、腕を大きく使った流れるような動き。\n時折ジャケットの襟を指でつかみ、ジャケットを軽く翻しながらターンする。\nサングラスに指を添えながらカメラを見る仕草。\n一瞬静止してから強いビートに合わせて鋭くステップを踏み、その後再び滑らかな動きへ移行する。\nダンスには「余裕、色気、自信、危険すぎないワイルドさ」がある。\n\nカメラは本格的なミュージックビデオ撮影。\n最初はローアングルの全身ショット。\n男性がカメラに向かって踊りながら近づいてくる。\nカメラも後退しながら追従する。\nその後、身体の周囲を半円状に回り込むダイナミックなオービットショット。\n足元のステップをローアングルで撮影。\nジャケットが翻る瞬間にスローモーション。\nサングラス越しの顔へ急速に寄るクローズアップ。\n最後は男性が軽く身体を横に向け、片手をポケットに入れ、もう片方の手でサングラスに触れながらカメラを見る。\n\n（プロフェッショナルなMV演出、ダイナミックなカメラワーク、リズムと完全に同期した編集、映画的なライティング、浅い被写界深度、ボリュームライト、リムライト、ネオンの反射、繊細な布の動き）\n\n（重要：実写・フォトリアルにはしない。最初から最後まで高品質な現代アニメ風イラストレーションとして描写する。セル塗りと厚塗りの中間、高精細なアニメ塗り、明確な輪郭、美しい有彩色の陰影、繊細なハイライト、アニメMVのキービジュアルがそのまま動き出したような質感。顔を実写の人間に変化させない。）\n\n（文字、字幕、ロゴ、タイトル、名前は表示しない）\n\n・ネガティブプロンプト\nphotorealistic, live action, realistic human skin, 3D CG, western comic style, chibi, cartoon comedy style, low quality, blurry, distorted face, changing face, changing hairstyle, changing clothes, missing sunglasses, transparent sunglasses, visible eyes through sunglasses, different beard, extra fingers, deformed hands, broken anatomy, duplicated body, multiple men, background dancers, exaggerated dance, childish dance, goofy dance, smiling excessively, text, logo, subtitles, title, watermark, camera shake, flicker, inconsistent character design",
    "sourceUrl": "https://x.com/Neko_ai_p/status/2095091133269647852",
    "sourceLabel": "X 原帖",
    "author": "Neko_ai_p",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095091133269647852.mp4",
    "posterUrl": "/posters/x/2095091133269647852.jpg",
    "duration": 10.336,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "舞蹈",
      "写实电影感",
      "城市街景"
    ],
    "category": "dance",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dance",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Neko_ai_p/status/2095091133269647852"
  },
  {
    "id": "x-2095088358129311893",
    "title": "推门走进高层酒店套房",
    "titleEn": "Stepping into a High-Rise Hotel Suite",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头跟随入住者进入套房，展示起居空间并用手机记录室内。",
    "summaryEn": "The camera follows a guest into a suite, showing its living space and a phone recording the room.",
    "prompt": "A smooth, continuous 15-second first-person / over-the-shoulder cinematic mobile video of a young woman with long dark hair, wearing a beige sweater and backpack, exploring a luxurious modern hotel room. Shot on smartphone in vertical/mobile format, natural soft daylight, elegant and calm atmosphere \n\n0:00–0:01 – Close-up of a hand holding a gold keycard tapping the electronic door lock and opening the wooden door, revealing a bright modern hotel room.\n0:01–0:02 – She steps inside from behind, holding her phone up as if filming.\n0:02–0:03 – Wide shot of the entire elegant room: large bed with white linens, seating area, big windows with city skyline view, natural light flooding in.\n0:03–0:04 – Slow camera pan right across the cozy king-size bed with white duvet and modern headboard art.\n0:04–0:05 – Camera moves toward the seating area by the large window: two grey armchairs and a round wooden coffee table.\n0:05–0:06 – She stands at the window filming the panoramic city view with high-rises and a stadium outside.\n0:06–0:07 – Camera pans left to the modern wooden work desk with chair, lamp and stationery.\n0:07–0:08 – Camera glides toward the flat-screen TV on a light wood media console with mini cabinet.\n0:08–0:09 – Close-up of the mini-bar / coffee & tea station: electric kettle, cups, sachets, neatly arranged under warm lighting.\n0:09–0:10 – Camera turns toward the open bathroom doorway.\n0:10–0:11 – Smooth move inside the clean modern bathroom: marble sink, large mirror, glass shower, toilet, bright lighting.\n0:11–0:12 – She turns back into the main room, smiling happily while holding her phone.\n0:12–0:13 – She gestures happily toward the room with one hand while filming, showing the space again.\n0:13–0:14 – Extreme close-up of the smartphone screen showing the live recording of the hotel room with red record button.\n0:14–0:15 – Final shot of her standing at the large window, enjoying the city view, holding the phone, soft natural light, peaceful ending.\n\nStyle & technical details:\nCinematic mobile footage, smooth handheld + subtle gimbal movement, soft natural window light, warm and inviting color grade, high detail, realistic textures, calm and luxurious mood. No text overlays, no logos. Vertical 9:16 aspect ratio, 15 seconds total. Soft ambient hotel atmosphere with gentle feel-good background music feel.",
    "sourceUrl": "https://x.com/Lianaalane/status/2095088358129311893",
    "sourceLabel": "X 原帖",
    "author": "Lianaalane",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095088358129311893.mp4",
    "posterUrl": "/posters/x/2095088358129311893.jpg",
    "duration": 15.627029,
    "aspectRatio": "68:45",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "生活实拍感",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Lianaalane/status/2095088358129311893"
  },
  {
    "id": "x-2095086312319746202",
    "title": "花影与字体交织的时装片",
    "titleEn": "Fashion Portraits through Flowers and Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物面部、花卉和细线字体层叠交替，组成杂志式时装画面。",
    "summaryEn": "Faces, flowers and fine-lined typography overlap in a magazine-style fashion sequence.",
    "prompt": "A 15-second editorial fashion film / fashion magazine commercial. Fast, stylish, contemporary, and visually bold. The film should feel like a rush of sharp fragments rather than a quiet portrait montage. Prioritize visual quality, movement, and editorial design above everything.\n\nUse the reference images as live visual material for face presence, fashion language, silhouette, surface texture, and mood, but do not recreate them as frozen portraits. Transform them into a living fashion film.\n\nCore rules:\n- The people must feel real first: breathing, blinking, gaze shifts, weight shifts, shoulders, fingertips, and mouth corners should always be alive.\n- No doll-like stillness, no paper-doll motion.\n- Every performer should have a distinct personality readable through movement, gaze, breathing, reaction speed, hand habits, posture, and expression.\n- Poses must be strong and clearly modeled, like fashion-model poses.\n- Cuts must be fast.\n- No shot should last longer than 2 seconds.\n- Scenes should change constantly, like a rush of sharp momentary fragments.\n- Use split-screen often enough to matter, with left and right sides acting on different intentions and never fully synchronized.\n- Keep the camera constantly moving: push-in, pull-out, orbiting, oblique travel, occlusion, depth shifts, and rapid reframing.\n- Use expressive transitions: flash cuts, match cuts, reverse flips, shutter-like wipes, split-screen interruptions, and speed-based continuity.\n- Avoid boring fades, boring slides, and quiet portrait sequences.\n\nPerformer directions:\n- Performer 1: cold observer, restrained, gaze leads the motion.\n- Performer 2: ritual-like, hand-driven, meaningful interaction with clothing or props.\n- Performer 3: model-like, visual anchor, confident center presence.\n- Performer 4: trembling, breathable, delayed reactions, human softness.\n\nSubtle fashion gestures:\nEach performer should include subtle expressive fashion gestures as accents:\n- licking a finger\n- touching the cheek or chest\n- brushing hair back\n- blowing a soft breath toward the camera\n- winking\n- kiss-like motion\n- gaze-play\n\nKeep the tone chic, suggestive, and editorial rather than explicit.\n\nTypography:\n- English typography only.\n- The typography should feel like a designer logo system or a world-building mark, not subtitles.\n- It should behave like a continuous organism across the entire video: stretching, wrapping, splitting, colliding, trailing, slipping in front of and behind people, crossing cuts, and remaining present as an integrated graphic force.\n- Prioritize aesthetic motion and graphic presence over perfect readability.\n- The text may be abstracted, symbolic, geometric, and design-forward.\n\nUse words related to the project theme, such as:\n[WORD 1], [WORD 2], [WORD 3], [WORD 4], [WORD 5]\n\nOverall feel:\nA chic fashion magazine commercial with runway energy, crossed with an editorial visual language.\nThe final video should feel like distinct personalities moving through one editorial world, with living typography cutting through the frame like a second organism.\nStrong, modern, editorial, with a little strange beauty and controlled unease.",
    "sourceUrl": "https://x.com/TaoRInne/status/2095086312319746202",
    "sourceLabel": "X 原帖",
    "author": "TaoRInne",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095086312319746202.mp4",
    "posterUrl": "/posters/x/2095086312319746202.jpg",
    "duration": 34.458413,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "动态图形",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TaoRInne/status/2095086312319746202"
  },
  {
    "id": "x-2095066419192909875",
    "title": "从切洋葱到擦亮餐盘",
    "titleEn": "From Chopping Onions to Polishing a Plate",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "厨房中的女子连续切菜、摆盘和清洁餐具，最后举起白色餐盘。",
    "summaryEn": "A woman chops, plates food and cleans up in a kitchen, finishing with a polished white plate.",
    "prompt": "A stylish blonde woman wearing a light blue bow blouse, pearl necklace, and white apron stands in a commercial kitchen before rapidly chopping an onion on a wooden cutting board. Seen from a high overhead camera angle, tears pool in her eyes with shimmering crystal gem accents on her cheeks, before the scene transitions to her aggressively scrubbing a red-sauce-stained plate with a sponge amidst dynamic flying sauce splatters. Red comic-style speed lines frame her as she strikes a confident pose, ending with a playful wink while presenting a squeaky-clean white plate directly to the camera.",
    "sourceUrl": "https://x.com/TaliaAariz/status/2095066419192909875",
    "sourceLabel": "X 原帖",
    "author": "TaliaAariz",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095066419192909875.mp4",
    "posterUrl": "/posters/x/2095066419192909875.jpg",
    "duration": 15.168,
    "aspectRatio": "4:7",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "美食",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TaliaAariz/status/2095066419192909875"
  },
  {
    "id": "x-2095023536410911056",
    "title": "暖色现代住宅的室内外导览",
    "titleEn": "A Warm Modern Home, Inside and Out",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头依次展示现代住宅外观、厨房起居区和日落庭院。",
    "summaryEn": "The sequence tours a modern home's exterior, kitchen and living areas, and sunset terrace.",
    "prompt": "Create a 15-second ultra-photorealistic cinematic luxury real-estate house tour, using the provided reference image as the visual style reference.\n\nIMPORTANT:\nMaintain a consistent architectural design, materials, furniture style, warm color palette, and premium luxury aesthetic throughout the entire video. The house must feel like one real, connected property. No random architectural changes.\n\nVIDEO FORMAT:\nVertical 9:16\nDuration: exactly 15 seconds\nUltra-realistic 4K\nProfessional luxury real-estate cinematography\nSmooth stabilized camera movement\nNatural lighting\nRealistic reflections and shadows\nPremium editorial look\nNo people\nNo text\nNo logos\nNo watermark\n\nSCENE 1 — 0-2 SECONDS — EXTERIOR REVEAL\nStart with a cinematic wide establishing shot of the complete luxury modern house during golden hour.\nShow the elegant multi-level architecture, large floor-to-ceiling windows, stone exterior walls, warm interior lights, landscaped garden, clean driveway and premium entrance.\nCamera slowly pushes forward toward the house like a professional real-estate drone/gimbal shot.\n\nTRANSITION:\nSmooth forward movement transitioning naturally toward the main entrance.\n\nSCENE 2 — 2-4 SECONDS — MAIN ENTRANCE & FOYER\nCamera smoothly enters through the large front door into a sophisticated luxury foyer.\nShow polished flooring, warm wooden staircase, elegant wall decoration, large round mirror, designer console table and soft ambient lighting.\nCamera moves forward slowly and naturally, revealing the open-plan interior.\n\nSCENE 3 — 4-6 SECONDS — LUXURY LIVING ROOM\nTransition into the main living room shown in the reference image.\nShow the large caramel-brown leather sectional sofa, cream upholstered lounge chairs, round wooden coffee table, patterned rug, decorative plants, wooden furniture and the central modern fireplace with gray stone/tile surround.\nLarge windows bring in beautiful natural daylight.\nCamera makes a smooth cinematic side-to-side movement, showcasing the entire living space.\n\nSCENE 4 — 6-8 SECONDS — OPEN KITCHEN & DINING\nCamera smoothly pans from the living room into the open-plan luxury kitchen and dining area.\nShow a large premium kitchen island, elegant cabinetry, modern appliances, sophisticated pendant lights, marble/stone surfaces and a stylish dining area.\nKeep the same warm neutral color palette and architectural style.\nCamera slowly glides past the island for a high-end property-tour feeling.\n\nSCENE 5 — 8-10 SECONDS — MASTER BEDROOM\nSmooth transition into a spacious luxury master bedroom.\nShow a large elegant bed with premium neutral bedding, soft textured fabrics, wooden accents, designer bedside tables and large windows overlooking greenery.\nWarm natural sunlight enters through the windows.\nCamera slowly moves toward the bed and then gently pans across the room.\n\nSCENE 6 — 10-12 SECONDS — LUXURY MARBLE BATHROOM\nCamera transitions into a high-end master bathroom.\nShow large marble surfaces, a freestanding modern bathtub, spacious glass shower, elegant vanity, large mirror and warm hidden LED lighting.\nUse a slow cinematic camera sweep to reveal the luxurious finishes.\nEverything should look realistic and physically accurate.\n\nSCENE 7 — 12-15 SECONDS — BACKYARD & FINAL HERO SHOT\nCamera smoothly exits toward the backyard/patio.\nReveal a beautiful landscaped outdoor lounge area with comfortable designer seating, greenery, warm architectural lighting and a premium swimming pool.\nEnd with a cinematic wide hero shot of the entire property glowing during sunset.\nCamera slowly pulls back and rises slightly, creating a premium luxury-real-estate commercial ending.",
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2095023536410911056",
    "sourceLabel": "X 原帖",
    "author": "ayzalnooor24521",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095023536410911056.mp4",
    "posterUrl": "/posters/x/2095023536410911056.jpg",
    "duration": 15.859229,
    "aspectRatio": "56:45",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ayzalnooor24521/status/2095023536410911056"
  },
  {
    "id": "x-2095014239425716588",
    "title": "樱花树下的旅行自拍",
    "titleEn": "A Travel Selfie beneath Cherry Blossoms",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子在樱花树下自拍和仰望落花，画面转向草地上的赏花人群。",
    "summaryEn": "A woman films herself beneath cherry blossoms before the view turns to picnickers on the grass.",
    "prompt": "Create a 15-second cinematic girl travel vlog in 16:9 landscape. A young woman walks through a beautiful Japanese park during peak sakura season while filming herself with a compact vlog camera. She laughs softly, spins slightly beneath the cherry blossom canopy and looks upward as petals fall around her. She then points toward a peaceful hanami picnic area filled with people enjoying spring. Her hair moves naturally in the breeze. Warm golden sunlight, authentic candid emotion, realistic people and environment, handheld influencer vlog aesthetic, photorealistic 4K, 16:9.",
    "sourceUrl": "https://x.com/iamrealsnow/status/2095014239425716588",
    "sourceLabel": "X 原帖",
    "author": "iamrealsnow",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095014239425716588.mp4",
    "posterUrl": "/posters/x/2095014239425716588.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "自然风景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/iamrealsnow/status/2095014239425716588"
  },
  {
    "id": "x-2095008773245211050",
    "title": "蓝色球场上的网球广告",
    "titleEn": "A Tennis Ad on a Blue Court",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "网球选手的挥拍、脚步和球体特写交替突出训练动作。",
    "summaryEn": "Racket swings, footwork and ball close-ups emphasize a tennis player's movement.",
    "prompt": "Create a fast-paced, high-energy cinematic sports commercial featuring a professional female tennis player competing on a blue hard court.\n\nScene & Camera:\n\n- Begin with an overhead bird’s-eye shot of her powerful serve.\n- Transition into low-angle tracking shots capturing her fast footwork and movement across the court.\n- Use intense close-ups of her focused face, hands, racket, and ball impact.\n- End with a dramatic golden-hour silhouette as she strikes the ball.\n\nVisual Style:\nPhotorealistic, premium sports-commercial aesthetic, bright natural sunlight, subtle lens flares, crisp focus, realistic motion blur, dynamic camera movement, cinematic composition, 8K detail, 35mm film look, subtle film grain.\n\nPerformance:\nNatural athletic movement, realistic tennis technique, accurate racket and ball physics, powerful swings, quick footwork, confident expressions, and believable body mechanics.\n\nAudio:\nRealistic tennis-court ambience, racket impacts, ball bounces, quick footsteps, and subtle cinematic sports sound design. No dialogue.\n\nNegative Prompt:\nDistorted hands, extra fingers, unnatural anatomy, incorrect tennis technique, floating racket, unrealistic ball movement, duplicated objects, face distortion, flickering, jitter, slow motion unless intentional, CGI appearance, low detail, watermark, text, or logos.",
    "sourceUrl": "https://x.com/DaniaSafvi/status/2095008773245211050",
    "sourceLabel": "X 原帖",
    "author": "DaniaSafvi",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2095008773245211050.mp4",
    "posterUrl": "/posters/x/2095008773245211050.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/DaniaSafvi/status/2095008773245211050"
  },
  {
    "id": "x-2094983662253261308",
    "title": "海岸骑行途中的偶遇",
    "titleEn": "An Encounter on a Coastal Bike Ride",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子沿海骑车后停在护栏旁，与另一名旅人一起看向海面。",
    "summaryEn": "A cyclist stops beside a coastal guardrail and looks out to sea with another traveler.",
    "prompt": "[0–3s] — THE RIDE\nLow cinematic tracking shot from behind. The girl rides her red bicycle along a quiet coastal road, guardrails beside her and the endless blue ocean beyond. Her skirt and loose strands of hair move naturally in the wind. Sunlight flickers across the road as she passes.\n\n[3–5s] — SOMETHING BEAUTIFUL\nSide shot as she suddenly notices a flock of white birds sweeping dramatically across the ocean. She slows down, smiles softly, and reaches into her backpack for a small instant camera.\n\n[5–8s] — THE PHOTOGRAPH\nShe stops beside the overlook and raises the camera toward the ocean. As she takes the picture, the male character quietly walks into the far edge of the frame, completely unaware that he has just been captured in her photograph.\n\n[8–10s] — THE SURPRISE\nExtreme close-up of the photograph developing in her hands. The ocean appears first… then she notices the mysterious guy standing in the background of the photo. Her expression changes from peaceful curiosity to surprised amusement.\n\n[10–12s] — LOOK UP\nShe slowly looks up from the photograph. Cut to the man a few meters away, standing by the guardrail with his backpack, looking directly at her. Wind moves his hair. He gives a subtle, almost playful smile.\n\n[12–15s] — THE LITTLE TWIST\nHe walks toward her and gently points at the photograph, realizing he's in it. She laughs quietly. He gestures toward the road ahead as if asking, “Going this way?” She smiles, gets back on her bicycle, and he starts walking beside her.\n\nFinal wide shot: They move together along the coastal road while the ocean stretches endlessly behind them. The little photograph remains visible in her hand as the camera slowly pulls back into the glowing blue-and-gold horizon.\n\nMOOD: spontaneous, beautiful, slightly mysterious, innocent romance, cinematic fate, natural chemistry.\n IMPORTANT: No exaggerated acting, no cheesy romance, no text, no logos, realistic physics and natural human movement.",
    "sourceUrl": "https://x.com/codewithhajra/status/2094983662253261308",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094983662253261308.mp4",
    "posterUrl": "/posters/x/2094983662253261308.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "自然风景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2094983662253261308"
  },
  {
    "id": "x-2094951518839652411",
    "title": "齿轮机关中的蒸汽朋克角色",
    "titleEn": "Steampunk Characters inside Clockwork",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "金色机械门和齿轮轮盘依次揭示多位动漫角色。",
    "summaryEn": "Golden mechanical doors and a gear wheel reveal a cast of anime characters.",
    "prompt": null,
    "sourceUrl": "https://x.com/challenger_ND/status/2094951518839652411",
    "sourceLabel": "X 原帖",
    "author": "challenger_ND",
    "publishedAt": "2026-09-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094951518839652411.mp4",
    "posterUrl": "/posters/x/2094951518839652411.jpg",
    "duration": 15.146667,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "奇幻",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094788953186689068",
    "title": "浴室门口的回眸与走近",
    "titleEn": "A Glance and a Step from the Bathroom Door",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "湿发女子在门边转头看向镜头，随后向前走近。",
    "summaryEn": "A woman with wet hair turns toward the camera at a doorway and steps closer.",
    "prompt": "5 秒短视频提示词：\n基于参考图中的年轻成年东方女性、湿发、浅灰紫色吊带修身短裙、肩上搭着浅色毛巾、浴室门口与卧室走廊环境，保持人物长相、身材比例、服装、湿发状态和室内布局高度一致。\n\n真实 iPhone 手机随手拍摄质感，像洗完澡后被身边的人随手记录下来的生活瞬间。\n\n0–2 秒：\n她刚从浴室走出来，停在门框旁，头发湿润凌乱，发梢仍有少量水珠。她微微侧头看向画面左侧，一只手自然抬起，从耳后轻轻拨开湿发。肩上的毛巾随着动作轻微滑动，胸口和锁骨附近残留的细小水珠在室内暖光下自然反光。\n\n2–4 秒：\n她像听见镜头这边有人叫她，眼神缓慢从侧面移向镜头，动作有非常轻微的停顿。嘴角出现一点若有若无的笑意，不是摆拍式微笑，而是“发现有人在拍她”后的自然反应。另一只手保持放松垂在身侧。\n\n4–5 秒：\n她轻轻把湿发往后拨了一下，同时身体自然向镜头方向迈出半步，毛巾跟随动作轻微晃动。就在她快要靠近镜头时结束画面。\n\n人物状态：\n年轻成年东方女性，洗澡后微微疲惫但放松，神态自然、松弛、带一点亲近感。湿发贴在脸侧和颈部，真实皮肤纹理，轻微肤色不均和水珠，不磨皮，不塑料感。不要刻意撩拨，不要模特摆拍。\n\n镜头：\niPhone 主摄普通视频模式，拍摄者站在距离人物约 1.5–2 米的位置，轻微手持晃动。镜头基本固定，只在人靠近时产生一点自然跟随。不要运镜炫技，不要慢动作，不要电影级推轨。\n\n光线：\n浴室内部略偏冷白光，走廊和卧室是柔和暖光，形成真实住宅里的轻微冷暖反差。允许轻微自动曝光变化、暗部噪点、湿皮肤局部高光，不做 HDR，不做商业打光。\n\n声音：\n很轻的浴室水声余音、赤脚或拖鞋轻微摩擦地面的声音、毛巾和湿发摩擦声，环境安静，不需要对白。\n\n限制：\n不要突然大幅转身，不要夸张甩头发，不要媚眼，不要刻意挺胸，不要身体比例变化，不要衣服变形，不要毛巾消失，不要湿发突然变干，不要面部漂移，不要手指畸形，不要大幅镜头运动，不要商业写真感。",
    "sourceUrl": "https://x.com/nopinduoduo/status/2094788953186689068",
    "sourceLabel": "X 原帖",
    "author": "nopinduoduo",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094788953186689068.mp4",
    "posterUrl": "/posters/x/2094788953186689068.jpg",
    "duration": 5.248,
    "aspectRatio": "15:26",
    "resolution": "480p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "日常生活",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/nopinduoduo/status/2094788953186689068"
  },
  {
    "id": "x-2094782705577996727",
    "title": "绿色双马尾角色的游戏式登场",
    "titleEn": "A Green-Haired Character's Game-Style Entrance",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "绿色双马尾角色与像素字体、复古游戏画面一起登场。",
    "summaryEn": "A green-haired character appears among pixel typography and retro game visuals.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2094782705577996727",
    "sourceLabel": "X 原帖",
    "author": "Mayz1169",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094782705577996727.mp4",
    "posterUrl": "/posters/x/2094782705577996727.jpg",
    "duration": 14.814331,
    "aspectRatio": "60:67",
    "resolution": "1080p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094780772318339508",
    "title": "平地拍爬楼的超级英雄片场",
    "titleEn": "A Superhero Set Built on Forced Perspective",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "片场镜头利用平铺建筑与视角转换，制造超级英雄攀楼的错觉。",
    "summaryEn": "A film-set scene uses a flat building facade and changing perspective to create a superhero climbing illusion.",
    "prompt": "Create a highly realistic viral behind-the-scenes superhero filmmaking video inspired by the kind of practical forced-perspective stunt videos that go viral on TikTok, Instagram, and X — but make the concept, staging, characters, and visual execution original.\n\nThe superhero is an original Superman-inspired flying superhero: a handsome adult male superhero wearing a premium blue-and-red superhero suit with a flowing red cape, but do NOT reproduce any exact movie costume, logo, emblem, actor likeness, or copyrighted Superman design.\n\nThe entire sequence is filmed from a very high overhead camera looking almost straight down at a huge outdoor film set.\n\nAt first glance, the scene appears to show the superhero climbing vertically up the side of an enormous skyscraper. The illusion is created by a gigantic forced-perspective city set laid completely flat on the ground. The “skyscraper” is actually a highly detailed printed/constructed city facade positioned on the road, with perspective lines making it look vertical from the camera angle.\n\nShow the complete filmmaking environment:\n\n- the superhero lying/positioned on the fake building surface\n- multiple film cameras\n- camera operators\n- director\n- assistant directors\n- lighting equipment\n- cables\n- crew members\n- stunt coordinators\n- production assistants\n- spectators watching the unusual shoot\n- equipment cases and production vehicles around the perimeter\n\nThe director is actively directing the superhero, gesturing toward different positions and calling out instructions. Crew members move around the set naturally, making the scene feel like genuine behind-the-scenes footage from a massive movie production.\n\nThe superhero performs a dramatic “vertical climb” across the fake skyscraper. From the overhead perspective, the illusion looks incredibly convincing.\n\nThen create the viral reveal:\n\nThe camera slowly pulls even farther upward, revealing the full scale of the setup and making it obvious that the entire skyscraper is actually lying flat on the ground.\n\nAfter the reveal, transition seamlessly from the overhead BTS shot into the “finished movie shot.” Suddenly the perspective changes and the same setup looks like an enormous vertical skyscraper, with the superhero appearing to climb the side of the building hundreds of feet above the city.\n\nMake the final cinematic shot look completely believable: dramatic sunlight, realistic atmospheric haze, detailed glass skyscrapers, physically accurate shadows, realistic cape movement, cinematic depth of field, subtle camera shake, premium blockbuster cinematography.\n\nThe contrast between the ridiculous real-world setup and the spectacular final shot should be the main comedic hook.\n\nStyle: ultra-realistic live-action, documentary BTS footage mixed with blockbuster superhero cinematography, natural human movement, realistic production equipment, physically believable lighting, detailed environments, authentic camera operators and crew.\n\nCamera progression:\n\n1. Extreme overhead establishing shot.\n2. Slow controlled aerial push-in.\n3. Medium overhead shot showing the superhero and crew.\n4. Wider reveal exposing the entire fake skyscraper.\n5. Dramatic transition into the finished cinematic superhero shot.\n6. Final heroic shot of the superhero climbing/flying beside the enormous skyscraper.\n\nNo text, no subtitles, no logos, no watermarks, no recognizable actors, no exact recreation of an existing movie scene.",
    "sourceUrl": "https://x.com/john_my07/status/2094780772318339508",
    "sourceLabel": "X 原帖",
    "author": "john_my07",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094780772318339508.mp4",
    "posterUrl": "/posters/x/2094780772318339508.jpg",
    "duration": 15.186009,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "人物表演",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/john_my07/status/2094780772318339508"
  },
  {
    "id": "x-2094777057695936901",
    "title": "粉色图形里的双人角色揭晓",
    "titleEn": "A Duo Reveal in Pink Graphics",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "卡通情侣在粉色圆环、心形和大字排版之间共同登场。",
    "summaryEn": "A cartoon couple appears among pink rings, hearts and oversized typography.",
    "prompt": "Create an elegant motion-graphics-driven romantic duo reveal trailer in 16:9, exactly 13 distinct cuts, 24fps.  \nSole visual reference is the attached image. Preserve both characters exactly: the boy with soft messy chestnut-brown hair, large warm brown eyes, gentle smile, white button-up shirt with sleeves rolled, dark trousers, white sneakers, silver necklace and wristwatch; the girl with long wavy chestnut-brown hair, large pink satin bow, tiny heart clips, huge glossy brown eyes, rosy cheeks, soft smile, pink off-shoulder ruffled mini dress, small pink heart-clasp bag, white socks with pink bows, pink-and-white sneakers. Never redesign faces, proportions, outfits, materials or colors.  \n\nFilm is 80% bold yet soft graphic design in motion and 20% gentle character performance. Giant kinetic typography, hard-edged pastel shapes, split-screens, flat color fields, floating heart motifs, petal bursts, soft shutter flashes and elegant speed lines.  \nPalette: soft pink, lavender mist, blush rose, pure white, warm cream, light gold.  \nStyle: premium AAA motion-graphics title sequence × dreamy lifestyle campaign. Every graphic moves with graceful momentum and snaps cleanly on the beat. Cozy elements (pink roses, heart balloons, wooden railing, string lights, floating petals, teddy silhouette, gift box) appear only as abstract graphic layers or soft atmosphere.\n\nCUT 01 | 0.00-1.00s  \nPure graphics: massive soft-pink circle blooms on white field, two translucent lavender bars wipe across, floating heart particles sparkle; elegant letters punch in one after another with gentle impact shake.\n\nCUT 02 | 1.00-2.00s  \nFinal letter becomes soft circular frame: extreme close-up of the girl’s warm brown eye inside it, lashes fluttering; soft chromatic flash, circle dissolves into drifting pink and lavender petals.\n\nCUT 03 | 2.00-3.10s  \nWhite frame with enormous soft-pink typography; the couple steps lightly in from frame left and glides across the baseline of the letters, delicate speed lines and rose-petal trails following, letter fragments lifting like confetti. Soft whip-pan out.\n\nCUT 04 | 3.10-4.00s  \nSplit-screen triptych of flat soft-pink / white / lavender panels; in each panel their silhouettes perform different gentle poses in soft stroboscopic freeze: soft turn, hair flip, shared smile. Huge vertical text DREAMY scrolls slowly behind.\n\nCUT 05 | 4.00-5.10s  \nThey rise together through a giant slowly rotating ring of typography reading SWEET MOMENTS; camera tracks their mid-air twirl in gentle slow motion as letters scatter like petals, then soft snap-zoom on their confident smiling faces.\n\nCUT 06 | 5.10-6.00s  \nHard cut: white editorial card, huge black HEART with soft-pink slash; the girl floats over the word, fingertips touching the letter, dress and hair flowing while the boy supports her; the word gently compresses and rebounds with a soft bounce.\n\nCUT 07 | 6.00-7.00s  \nLavender field with soft diagonal rose stripe; they perform a light graceful spin along the stripe in three soft ghost frames, each tinted a different pastel; giant outlined text TWIRL rotates 180 degrees in perfect sync.\n\nCUT 08 | 7.00-8.00s  \nKinetic type barrage: the words SOFT / BRIGHT / WARM / LOVE appear one per beat with soft shutter flashes and light camera sway, while they glide across the foreground, clothes fluttering, tiny petal particles bursting from their sneakers.\n\nCUT 09 | 8.00-9.00s  \nSoft black frame, giant white wireframe heart grid tilting gently in 3D; they step lightly up the grid, push off and hang suspended as everything freezes; a soft-pink circle stamps around their frozen pose with delicate bracket accents and floating heart ticks.\n\nCUT 10 | 9.00-10.10s  \nFreeze releases into a soft burst: they drift toward camera through layered flat-color panes that dissolve one by one like silk, each pane revealing a bigger letter of a romantic motif; foreground wipe with sneakers and a swirl of petals.\n\nCUT 11 | 10.10-11.10s  \nRapid-fire poster montage: four full-screen graphic posters of the couple in different elegant poses (mid-twirl, soft slide, landing smile, gentle embrace) snap past with clean cuts, each with bold yet soft layout, oversized numbers 01–04, rose motifs and gentle diagonal lines.\n\nCUT 12 | 11.10-13.00s  \nHero moment on clean white cyclorama: they land a final soft spin dead center in slow motion, straighten into a close embrace with gentle smiles, and a shockwave of concentric soft-pink rings, wind streaks and shattered pastel typography blooms outward; brief iconic freeze, then soft overexpose to white.\n\nCUT 13 | 13.00-15.00s  \nFinal identity card: enormous soft-pink word dominates pale white field with translucent lavender rings, delicate arcs, soft halftone and a rough circular emblem containing a ghosted rose/heart motif; the couple stands relaxed overlapping the letters, soft breeze rippling their clothes and hair; one last soft-pink pulse sweeps through the typography and a delicate petal-ring flash ends the film.\n\nEditing: extremely rhythmic yet elegant — clean hard cuts on every beat, graphic matches, soft whip pans, gentle snap zooms, soft stroboscopic freezes, foreground petal wipes, light chromatic flashes, impact shakes and subtle speed ramps. Every cut compositionally different; typography always fully readable before characters overlap it.  \nNo weapons, combat or harsh elements; energy comes from motion design, soft wind, petals, light dissolves and their graceful lifestyle movement.\n\nBGM: catchy upbeat electronic / future-pop with bright melodic hooks, punchy drums, attractive drops, risers and sparkling fills locked to every cut. Sneaker steps, soft whooshes, petal bursts and typography impacts are rhythmic sound-design elements. Peak energy at CUT 12, close with a bright memorable electronic logo stinger.  \n\nPremium AAA quality, anime-inspired cinematic rendering, stylish and luminous, strong graphic-design identity, consistent character design, exactly 13 cuts.",
    "sourceUrl": "https://x.com/Chaemate_/status/2094777057695936901",
    "sourceLabel": "X 原帖",
    "author": "Chaemate_",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094777057695936901.mp4",
    "posterUrl": "/posters/x/2094777057695936901.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Chaemate_/status/2094777057695936901"
  },
  {
    "id": "x-2094765380249997695",
    "title": "酒红长裙的港风日常片段",
    "titleEn": "A Hong Kong-Inspired Look in Burgundy",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子穿酒红长裙，在室内窗边和走廊间转换姿态。",
    "summaryEn": "A woman in a burgundy dress changes poses by an indoor window and along a corridor.",
    "prompt": "时长：15秒\n比例：9:16\n人物参考：@图片1\n语言：无对白\n字幕：无字幕、无旁白、无文字\n\n人物参考规则：\n仅参考@图片1锁定成年女主的五官、脸型、肤色和人物身份，不参考图片中的服装、发型与背景。全程同一张脸，妆容、发型和身材比例保持一致。\n\n女主造型：\n复古港风长卷黑发，偏分发型，发丝蓬松自然；细长眉、黑色眼线、哑光酒红唇妆。佩戴一对华丽复古金色雕花耳坠。\n\n穿深酒红色丝绒修身吊带长裙，心形领口、贴合腰线、长度至小腿，裙摆后方小开衩；搭配黑色尖头细高跟鞋。裙子颜色全程保持深酒红，不变成鲜红或紫色。\n\n场景：\n1990年代香港豪华酒店套房。深色胡桃木墙板、米白墙面、黄铜壁灯、黑色木质百叶窗、墨绿色丝绒沙发、复古电话、暖色大理石洗手台和大型浴室镜。落日从百叶窗射入，在墙面与人物脸上形成横向光影。\n\n整体风格：\n真实港风酒店随拍Vlog，复古电影质感，暖金高光、墨绿阴影、轻微胶片颗粒、柔焦辉光、自然皮肤纹理。人物冷艳、自信、松弛，不持续微笑，不做网红式夸张动作。\n\nSHOT 1（00:00-00:03.50）床头怼脸\n\nSubject：\n女主侧身倚靠在深色软包床头，镜头距离面部很近。\n\nAction：\n00:00-00:01.20，女主闭着眼睛，头轻轻靠向右侧，右手托住脸颊，像刚从午睡中醒来。\n\n00:01.20-00:02.40，她缓慢睁眼，先看向窗外，再把视线移回镜头；指尖沿着耳侧整理一缕长发，金色耳坠轻轻摆动。\n\n00:02.40-00:03.50，她嘴角出现很轻的笑意，手掌继续托着下巴，保持眼神交流。\n\nCamera：\n面部近景，50mm质感，轻微手持呼吸感；缓慢向前推进约10厘米，焦点锁定双眼。\n\nLighting：\n柔和暖光照亮一侧面部，另一侧保留自然阴影。\n\nSHOT 2（00:03.50-00:07.20）浴室门口走出\n\nSubject：\n女主站在酒店浴室门口，背后是大理石洗手台、复古镜框和两盏黄铜壁灯。\n\nAction：\n00:03.50-00:04.60，动作匹配切到全身。女主双臂轻轻交叉，背靠门框，冷静看向镜头。\n\n00:04.60-00:06.00，她放下双手，从浴室向镜头缓慢走出两步。酒红裙摆随步伐轻微摆动，高跟鞋落地稳定。\n\n00:06.00-00:07.20，她停下，一只手扶腰，身体转向侧面约30度，展示裙子的腰线；随后回头看向镜头。\n\nCamera：\n固定全身景别，略低机位。人物靠近时镜头轻微后退，保持头顶和高跟鞋完整入画。\n\nPerformance：\n步伐从容，不扭腰过度；表情冷艳，回眸时眉眼稍微放松。\n\nSHOT 3（00:07.20-00:11.20）百叶窗光影\n\nSubject：\n女主站在黑色木质百叶窗旁，落日光线横向划过面部和肩颈。\n\nAction：\n00:07.20-00:08.50，胸像侧面特写。女主抬头看向窗外，阳光从鼻梁缓慢移动到眼睛，发丝边缘出现金色轮廓光。\n\n00:08.50-00:09.80，她抬起右手触碰百叶窗，轻轻推开一片叶板，更亮的光线落在脸上；她微微眯眼，保持自然呼吸。\n\n00:09.80-00:11.20，她放下手，缓慢转头直视镜头。神情由疏离转为自信，嘴角仅轻微上扬。\n\nCamera：\n85mm侧脸特写，缓慢横移，从纯侧脸移动到四分之三正脸。背景虚化，耳坠和眼睛保持清晰。\n\nSHOT 4（00:11.20-00:15.00）套房收尾\n\nAction：\n00:11.20-00:12.40，女主走到墨绿色丝绒沙发旁，一只手轻轻扶住沙发靠背，身体向后倚靠，长发自然落在肩后。\n\n00:12.40-00:13.50，她坐到沙发扶手边缘，双腿自然交叠，侧头笑一下；不是大笑，只是松弛、满足的微笑。\n\n00:13.50-00:15.00，切回浴室门口全身构图。女主双手整理耳侧长发，然后一手叉腰，另一只手自然垂下。她正面看向镜头，保持冷艳自信的定格姿态。\n\nCamera：\n先以中景跟随她走向沙发，再用动作匹配切回门口全身镜头。最后缓慢后退，展示完整酒店套房。\n\nAudio：\n复古粤语流行音乐氛围的纯器乐，低音鼓、柔和萨克斯与电钢琴；保留高跟鞋脚步、裙料摩擦、百叶窗轻响和室内环境声。无人物说话。\n\n连续性要求：\n同一酒店套房、同一酒红裙、同一发型与耳环。镜头之间通过整理头发、转头和扶腰动作进行匹配剪辑。所有动作自然缓慢，保留真实Vlog随拍感。\n\nNegative：\n现代网红公寓，廉价酒店，白色快捷酒店，背景杂乱，霓虹夜店，人物换脸，五官漂移，年龄变化，发型变化，短发，服装变化，红裙变色，裙子忽长忽短，耳环消失，夸张扭腰，持续媚笑，嘟嘴，过度眨眼，机械走路，脚底滑动，高跟鞋变形，手指粘连，多余手臂，面部扭曲，塑料皮肤，过度磨皮，动漫感，强烈美颜滤镜，镜头乱晃，快速闪切，曝光过度，字幕，旁白，文字，水印，Logo",
    "sourceUrl": "https://x.com/johnAGI168/status/2094765380249997695",
    "sourceLabel": "X 原帖",
    "author": "johnAGI168",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094765380249997695.mp4",
    "posterUrl": "/posters/x/2094765380249997695.jpg",
    "duration": 15.16263,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "复古胶片",
      "时尚",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "retro"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/johnAGI168/status/2094765380249997695"
  },
  {
    "id": "x-2094755037826130121",
    "title": "木屑逐字聚成 RENDER",
    "titleEn": "Wood Shavings Form RENDER",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "木屑从工作台聚起，形成六个直立字母并逐渐变为实木字形；原帖仅提供 Prompt 片段。",
    "summaryEn": "Wood shavings rise from a workbench into six upright letters and solid wooden forms; only a prompt excerpt is published.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2094755037826130121",
    "sourceLabel": "X 原帖",
    "author": "GlennHasABeard",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094755037826130121.mp4",
    "posterUrl": "/posters/x/2094755037826130121.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094744778286207061",
    "title": "蜘蛛纹连体服的室内表演",
    "titleEn": "An Indoor Performance in a Spider-Pattern Suit",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "短发女子穿蜘蛛纹连体服，在固定机位前轻轻摆动身体。",
    "summaryEn": "A short-haired woman in a spider-pattern suit moves gently before a fixed camera.",
    "prompt": "[镜头 1] 写实数字人时尚表演风格，竖屏 9:16，时长 8 秒。一名成年东亚女性位于简约客厅中央。她拥有瓷白色肌肤、黑色齐耳直短发、深红色眼妆和柔和的红唇，身穿黑色高光蜘蛛主题连体紧身衣。服装带有精细的银色蛛网纹理、一体式手套、反光侧面拼接和胸前白色蜘蛛图案。摄影机采用与视线平齐的中全景构图，取景范围从大腿上部至头顶。女子将双手放在脑后，双肘对称地向两侧展开，直视镜头，同时配合音乐做轻微口型，嘴巴微张，并缓慢左右摆动髋部和上半身。在 00:04.500，左右摆动髋部和上半身的时候，她以小幅度、慢速抬起下巴，双眼向上转动，形成一次鲜明的表情节拍。在 00:05.800，她眨眼后重新直视镜头，最后回到开场姿势。摄影机使用 Static Shot（固定镜头），零幅度、零移动速度，无摇摄、俯仰、推拉或变焦。保持角色面部身份、短发造型、身体比例、服装纹理、蜘蛛图案、手臂姿势、灯光及背景结构一致。身体结构自然，眼睛稳定，口型自然，反光布料物理效果真实。无字幕、无水印、无账号文字、无品牌标志、无额外人物、无多余肢体、无手部畸形、无服装纹理漂移、无面部变形、无背景扭曲、无画面闪烁、无镜头抖动。\n\noverall_soundscape: 极其轻微的室内底噪。女子摆动身体时，紧身合成面料发出细微的拉伸与摩擦声，同时保留克制而自然的呼吸声。没有对白，也没有突出的环境声。所有物理动作声保持轻微，并位于音乐声层之下。\n\nnon_diegetic_music: 约 116 BPM 的暗黑性感电子流行音乐循环。使用稳定的底鼓、清脆的军鼓、脉冲式次低音、短促的合成器重音，以及经过处理的气声女性人声纹理，不出现可辨识歌词。在 00:04.500，将最强音乐重音与女子向上转动眼睛的动作严格同步。整体采用具有社交媒体短视频质感的高压缩混音，但避免数字削波和爆音。在 00:08.000 干净利落地硬切结束，不保留混响尾音。",
    "sourceUrl": "https://x.com/VoxcatAI/status/2094744778286207061",
    "sourceLabel": "X 原帖",
    "author": "VoxcatAI",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094744778286207061.mp4",
    "posterUrl": "/posters/x/2094744778286207061.jpg",
    "duration": 8.064,
    "aspectRatio": "19:33",
    "resolution": "608p",
    "tags": [
      "舞蹈",
      "照片级写实",
      "人物表演"
    ],
    "category": "dance",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/VoxcatAI/status/2094744778286207061"
  },
  {
    "id": "x-2094736167681826950",
    "title": "蓝发动漫人物的夜间近景",
    "titleEn": "A Blue-Haired Character at Night",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝发动漫人物在室内灯光下转头，画面切换手部与面部近景。",
    "summaryEn": "A blue-haired anime character turns under indoor lighting, with cuts between hands and face.",
    "prompt": "夜の部屋を舞台にした、センシュアルなムードフィルム。\n日本のアニメ作画。手描きの繊細な線、平坦なセル塗り、フルカラー。\n浅い被写界深度、フルサイズ換算24〜85mmのズーム感。\n60 BPM。BGMあり、セリフなし、環境音のみ。\n\n# 制約\n@[char ref] は彼女の顔立ち・髪型・体型・衣装・アクセサリー・アニメの作画様式を定義する。\n参照画像の背景と構図は使わない。アクセサリーを新たに足さない。\n出力には常に彼女ひとりだけを映す。他の人物・手・撮影者・カメラ機材は一切映さない。\n彼女の同一性、解剖、髪型、衣装、アクセサリー、アニメの作画様式は全カットで完全に同一に保つ。\n場所の固定：全編が同じひとつの部屋の中である。家具はソファとベッドの二つだけで、\nどちらも同じ部屋にある。窓は常に画面奥にある。部屋の外へは一度も出ない。\n煙草の定義：火のついた煙草は [A] と [B] にだけ存在し、常に一本だけである。\n[A] の終わりに彼女が灰皿の縁へ置き、[B] ではその一本が灰皿で燃え続けている。\n[C] 以降、彼女の手に煙草はなく、口へ運ぶ動作も、新しく火をつける動作も一切映さない。\n[C] 以降、煙は画面のどこにも漂っていない。\n靴の定義：彼女は [A] と [C] ではヒールを履いている。[F] で片方が脱げ、\n[E] では両足とも素足である。ヒールは一足だけであり、増えない。\n小物の定義：室内にあるのは、吸い殻が二本入った灰皿、火のついた煙草、サテンのシーツ、\n薄いカーテン、彼女のヒール、これだけである。\nロゴ・文字・数字・ラベルは画面のどこにも存在しない。\n編集のリズム：長く回し続けるロングテイクが主体であり、そのあいだに二度だけ、\n極端に短いカットの連打が割り込む。均等な尺のカット割りにしない。\nロングテイクの中でカメラは一度も止まらず動き続ける。\n\n# 情景\n夜の部屋。天井灯は消えており、光源は低い位置の暖かいテーブルランプがひとつだけ。\n窓には薄いカーテンが引かれ、その向こうから街灯の弱い光が拡散して入る。\n床とサテンのシーツだけが艶を持ち、暖かい光を鈍く映す。\n誰にも見られていない、静かで親密な深夜。\n光と影が肌の輪郭をやわらかく浮き上がらせる。\n奥行きは前景・中景・遠景を明確に分離する。中景に彼女、遠景にカーテンと窓。\n\n# シーン\n\n[A] 0:00-0:03.6 ロングテイク。カメラは止まらずに動き続ける。\n画面は前景に漂う煙草の煙で埋まっている。カメラが抑えたプッシュインで煙を抜けると、\nソファに深く沈んで座る彼女の正面ミディアムクローズアップが現れる。\n彼女は指のあいだに火のついた煙草を持っており、それを唇へ運び、ゆっくり吸い込む。\n先端の火が暗がりで赤く強くなり、彼女の顔の下半分だけを一瞬照らす。\n下唇が煙草から離れるとき、唇が歯に一度引っかかってから滑り落ちる。\nカメラはそのまま速い弧を描いて下降し、顎が持ち上がって喉の線がぴんと張るところで、\n喉が一度だけゆっくり動いて嚥下する。\n彼女の指が煙草をわずかに傾け、灰が音もなく落ちる。\nそのまま手が伸びて、煙草を灰皿の縁へ静かに置く。\n指の力が抜け、手首がソファの肘掛けの上へゆっくり倒れていく。\n\n[B] 0:03.6-0:04.6 短いカットの連打。2カット、各0.5秒。ハードカット。\nb1: 灰皿の極寄り。縁に置かれた一本の煙草の先が赤く燃え、細い煙が立ち上っている。\nその隣に吸い殻が二本あり、片方の先に口紅がついている。\nb2: 唇の極寄り。細く長い煙が唇のあいだからゆっくり押し出される。\n\n[C] 0:04.6-0:08.4 ロングテイク。カメラは止まらずに動き続ける。\n同じソファ。彼女は浅く腰かけ直して背筋を立て、上体をまっすぐ起こしている。\n空気から煙は完全に消えている。左の横顔。両手は何も持っていない。\n舌先が口角をゆっくり湿らせる。\nカメラは頬のすぐ横をホイップして通過し、鎖骨の高さまで寄って回り込む。\n彼女の指先が喉のくぼみに触れ、そこから鎖骨のラインを辿り、\n胸元の布の縁までゆっくりと滑り降りて、そこで止まる。\n指先が布の縁の上で一度だけ押し当てられ、生地がわずかに沈む。\n指が通ったあとを、横からの鋭いリムライトが追う。\nカメラは胸元のクローズアップまで寄り、鎖骨から胸にかけてのラインが\n呼吸で浅く上下しているところを捉える。\n続けて彼女が脚を組み替え、生地が擦れて張り、\nカメラは腰のラインを横へ高速でスライドしながら、その動きを追い越す。\n\n[F] 0:08.4-0:09.9 短いカットの連打。3カット、各0.5秒。ハードカット。\nf1: 真後ろからのクローズアップ。薄い生地の下で肩甲骨が動く。\nf2: 睫毛が触れ合ったまま留まるスローブリンクの極寄り。\nf3: 足元のクローズアップ。爪先が床を押さえ、踵がヒールから抜けていき、\n素足が完全に離れて、空になったヒールが床の上に横倒しに残る。\n\n[E] 0:09.9-0:15.0 ロングテイク。カメラは止まらずに動き続ける。\nベッドの上。カメラは床スレスレの高さからサテンのシーツの上を前進していく。\n彼女は素足でシーツへ身を沈め、片膝を立てる。\nサテンがふくらはぎの下で滑って皺を作る。\n片手がシーツを深く握り込み、布が指のあいだで束になって引き寄せられ、\n指の関節が白くなってから、ゆっくり力が抜けて開く。\nカメラは胴体の高さへダイブし、そのまま上昇して顔の近くを通過し、\n彼女が口をレンズのほうへ向けて肩越しに振り返るところで速度を落とす。\nそのままカメラは彼女の周囲をタイトに周回する。\n彼女は背中を反らせて長く上へ伸び、布が肋骨の上で張り、\n両手が髪を下から掬い上げてうなじから離す。\nカメラは肩から顔へ連続してスキムし、前景のカーテンの襞と背景の窓に強い視差が生まれる。\n彼女は顎の線とうなじが同じランプの光を受ける角度まで、わずかに振り返る。\n指先が自分の首筋の脈に当てられる。\n最後に彼女が細く長い息をゆっくり吐き、唇がわずかに開いたまま留まり、\nレンズを見下ろす姿勢で静止する。\nカメラは外へ弧を描きながら下降して引き、部屋の全体が暖かい光の中に開く。\nその引きの構図のまま完全に静止して終わる。\n\n# 音\n部屋の低い静寂、遠い車の走行音がかすかに続く。\n冒頭では煙草の紙が燃える微かな音、息を吸い込む音、煙を吐く音が入る。\n全編を通して、唇が離れる小さな音、髪が動く音、布が擦れる音がある。\nヒールが脱げる場面では、革が踵を離れる短い音と、\nヒールが床に落ちて一度だけ鳴る乾いた音が入る。\nベッドではサテンが滑る音、シーツが握り込まれる音、静かな呼吸だけになる。\n台詞と歌は一切ない。",
    "sourceUrl": "https://x.com/studio_oneroom/status/2094736167681826950",
    "sourceLabel": "X 原帖",
    "author": "studio_oneroom",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094736167681826950.mp4",
    "posterUrl": "/posters/x/2094736167681826950.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "人物表演",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/studio_oneroom/status/2094736167681826950"
  },
  {
    "id": "x-2094709258562764940",
    "title": "四叶草里的幸运角色登场",
    "titleEn": "A Lucky Character among Clover Leaves",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "四叶草遮挡揭开小女孩角色，绿色大字和几何图形连续切换。",
    "summaryEn": "Clover leaves reveal a small cartoon character amid green lettering and shifting geometric layouts.",
    "prompt": "Locked character (do not redesign): chibi proportions, black wavy hair + messy bangs + twin tails, white lace bonnet, huge glossy dark eyes, rosy cheeks, cream puff-sleeve dress with lace hem, chunky yellow rain boots, giant four-leaf clover on a green stem. Palette: leaf green, gold, cream, white, charcoal, yellow.\nRejected on purpose: cobalt slam-circle, eye-in-the-O porthole, cobalt/white/ice triptych, wall-run on a 3D grid, four action posters 01–04.\n\n13-cut board (16:9, 24fps, ~15s)\nCUT 01 | 0.00–1.00\n\nCream field. Four heart-leaflets slam in one-by-one and lock into a giant four-leaf mark. Gold micro-ticks + a luck “barcode” flicker. Letters L · U · C · K punch in with bounce-shake.\nCUT 02 | 1.00–2.00\n\nThe clover becomes a spinning charm-window: her face peeks through the center gap, smiling, stem in both hands. Leaflets peel off as flat green shards. Gold RGB tick, no eye-macro.\nCUT 03 | 2.00–3.10\n\nHuge green word FOUR on cream. She skip-hops in from the right, swinging the clover like a parasol; leaf confetti + gold pollen kick up from the yellow boots. Whip-pan off her kick.\nCUT 04 | 3.10–4.00\n\nOne giant clover used as a 4-leaflet layout (not a split-screen triptych). Each leaflet holds a tiny different beat: hop / peek / twirl / boot-kick. A ribbon of type LUCKY threads down the stem.\nCUT 05 | 4.00–5.10\n\nShe jumps through a slow-rotating type ring FIND YOUR LUCK. Mid-air she spins the clover overhead like a pinwheel. Letters scatter as leaves. Snap-zoom on the smile, not a parkour face-lock.\nCUT 06 | 5.10–6.00\n\nWhite editorial card. Giant black CHARM with a single green slash. She pole-vaults using the clover stem, palm on the C; the word compresses like a spring and rebounds. No “JUMP” vault.\nCUT 07 | 6.00–7.00\n\nSoft green field, gold diagonal path. She cartwheels along it in three ghost frames (cream / leaf / gold). Outlined type SPIN rotates 180° with her. No backflip-on-stripe streetwear read.\nCUT 08 | 7.00–8.00\n\nType barrage on beat: SOFT / WILD / LUCKY / GREEN. She skip-slides on the yellow boots in the foreground, dress flaring, clover leaves fluttering, pollen burst from the soles. Jacket-flare language replaced by dress + bonnet flutter.\nCUT 09 | 8.00–9.00\n\nCream void. A flat paper clover unfolds in 3D (not a wireframe city grid). She balances on one leaflet like a tiny raft, pose freezes. A gold seal stamps around her with luck-% ticks and a 4-leaf bracket. No wall-run.\nCUT 10 | 9.00–10.10\n\nFreeze breaks. She leans toward camera through layered translucent green panes that shatter like pressed leaves, each pane revealing a bigger letter L-U-C-K. Foreground wipe is the yellow boot + clover tip.\nCUT 11 | 10.10–11.10\n\nNot posters. A slot-machine / stamp barrage: giant clover icons spin and lock 4-up, overlaid numbers 4 · 7 · 4 · 7, barcodes, gold slashes. She is only a repeating silhouette spinning the stem inside the lock-up. Hard cuts, no 01–04 pose posters.\nCUT 12 | 11.10–13.00\n\nWhite cyclorama. She lands a playful spin, plants the clover like a standard, and a shockwave of concentric green rings + leaf shards + gold dust blows out. Brief freeze, faint smile, overexpose to cream-white.\nCUT 13 | 13.00–15.00\n\nIdentity card: enormous charcoal 4 (or LUCK) on pale cream, translucent green rings, halftone pollen, rough circular emblem with ghosted four-leaf + barcode. She stands relaxed overlapping the letter, wind in dress and bonnet. One green pulse through the type, frost-ring replaced by a gold pollen-ring flash.\nEdit language: hard cuts on the beat, graphic matches (leaflet → window → type), whip pans, snap zooms, 3-frame ghosts, foreground leaf/boot wipes, shutter flashes, impact shakes. Type always readable before she overlaps it. Energy = luck, wind, leaves, glass-leaf shatter, spring — no weapons, no combat, no fire.\nBGM: bright future-bass with mallet hits and clover-leaf whooshes instead of industrial subs; still lock slams to every cut; peak at CUT 12; cold lucky-charm stinger on CUT 13.",
    "sourceUrl": "https://x.com/Just_sharon7/status/2094709258562764940",
    "sourceLabel": "X 原帖",
    "author": "Just_sharon7",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094709258562764940.mp4",
    "posterUrl": "/posters/x/2094709258562764940.jpg",
    "duration": 15.168,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2094709258562764940"
  },
  {
    "id": "x-2094705572427993463",
    "title": "奶油旋转成甜品字效",
    "titleEn": "Cream Swirls into Dessert Typography",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "粉紫奶油堆叠后变为甜品标题与装饰图形。",
    "summaryEn": "Pink and purple cream stacks transform into dessert lettering and decorative graphics.",
    "prompt": "A 15-second 2D flat motion-graphics commercial in a clean, minimal kawaii dessert style. Soft warm beige studio background with a faint paper grain. Limited palette: milk-chocolate brown, blush pink, lavender purple, cream white, cherry-blossom pink, pale yellow-green leaves, and glossy black. Smooth, bouncy frosting animation, soft squash-and-stretch, and gentle sparkle pops. No photorealism — bold graphic cake shapes with clean piped-line texture.\n0.0–1.0s\n\nTiny pastel frosting flowers (pink, lavender, white) and small chocolate-brown piping dots drift and spin slowly across the empty beige background, each with a tiny white highlight.\n1.0–2.0s\n\nThe pieces gather and stack downward, forming a round cake on a simple white scalloped glass stand. Soft lavender-pink frosting swirls around the sides in even horizontal ridges. A twisted lavender rope of frosting coils around the base.\n2.0–3.5s\n\nA smooth cream-white disc drops onto the top as the face. Chocolate-brown frosting combs itself into neat bangs, then two long braids grow down the sides with a soft bounce. Small fondant flowers and green leaves tuck themselves into the hair.\n3.5–5.5s\n\nThe finished cake: round kawaii girl face with big glossy black eyes, tiny white sparkles, short eyelashes, round pink blush circles, and a simple curved smile. Purple heart “earrings” sit at the ends of the braids. A crown of piped lavender stars lines the top of the hair. Soft yellow sparkles and tiny frosting petals pop around the cake. The camera eases slightly closer; the braids sway once.\n5.5–7.5s\n\nThe cake dissolves into large, bold, rounded white typography that reads CUTIE GIRL CAKE. The letters first appear as if they are piped from the same chocolate and pink frosting, then the frosting settles away leaving solid cream-white lettering sitting on a wide, organic lavender splash that looks like overflowing buttercream.\n7.5–15s\n\nThe title stays locked on screen. Small flowers and chocolate droplets continue to bounce on and around the letters. A short tagline types on underneath in a friendly rounded pastel font: first “sweet & smiling,” then expanding to “sweet & smiling, a little celebration.”\n\nA tiny logo appears at the bottom center: a miniature version of the same girl cake plus the words Aaliya Motion Graphics. Subtle sparkle and frosting-dot animation continues until the end.\nMood: cute, gentle, birthday-party kawaii. Timing should feel snappy and satisfying — frosting has weight, braids bounce, flowers land softly, text has a creamy “pipe-on” reveal. Keep everything graphic, flat, and highly readable.",
    "sourceUrl": "https://x.com/AIwithAliya/status/2094705572427993463",
    "sourceLabel": "X 原帖",
    "author": "AIwithAliya",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094705572427993463.mp4",
    "posterUrl": "/posters/x/2094705572427993463.jpg",
    "duration": 15.13941,
    "aspectRatio": "26:15",
    "resolution": "480p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AIwithAliya/status/2094705572427993463"
  },
  {
    "id": "x-2094701248045789443",
    "title": "超市里的橙色零食挑战",
    "titleEn": "An Orange Snack Challenge at the Supermarket",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子推购物车穿过货架，最后抱起一堆橙色包装零食。",
    "summaryEn": "A woman pushes a cart through supermarket aisles and gathers orange-packaged snacks.",
    "prompt": null,
    "sourceUrl": "https://x.com/sophiaparkerr_/status/2094701248045789443",
    "sourceLabel": "X 原帖",
    "author": "sophiaparkerr_",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094701248045789443.mp4",
    "posterUrl": "/posters/x/2094701248045789443.jpg",
    "duration": 15.16263,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "生活实拍感",
      "照片级写实",
      "美食",
      "日常生活"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094686415585972658",
    "title": "点击服装图标切换穿搭",
    "titleEn": "Changing Outfits with Clothing Icons",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物站在服装图标之间，点击动作对应不同的裙装与搭配。",
    "summaryEn": "A model stands among clothing icons as taps trigger changes in dresses and outfits.",
    "prompt": "互动点击变装提示词\n\n时长：15秒\n比例：9:16\n人物参考：图片1\n语言：无对白\n音频：轻快时尚音乐、鼠标点击声、短促换装音效\n屏幕文字：仅允许标题“你选哪一套”\n\n人物参考规则：\n仅参考 📷图片1 图片1锁定成年女主的五官、脸型、肤色、发型和人物身份。全程必须是同一个人，服装变化时脸部、身材、发型与妆容保持一致。\n\n场景：\n\n现代高档公寓客厅，奶油色大理石地面、深色胡桃木墙板、暖白隐藏灯带、落地窗、远处城市夜景、浅色设计师沙发和金属落地灯。空间整洁、宽敞、高级，暖色电影灯光，真实手机时尚穿搭视频质感。\nCamera：\n\n竖屏固定机位，全身构图，镜头略低于胸口，女主站在画面中央。全程不推拉、不摇移、不切换背景。\n标题设计：\n\n画面顶部中央固定显示准确中文：\n“你选哪一套”\n\n高级香槟金中文字体，轻微立体浮雕与柔和发光，下方有一条纤细金色手绘弧线。标题从开头淡入后保持不动，不能变字、错字或重复出现。\n贴纸菜单：\n\n画面左右边缘固定排列六枚穿搭贴纸，每枚贴纸都是女主对应服装的全身剪影，带白色模切描边和轻微阴影，像时尚手账贴纸。贴纸不能遮挡女主面部和身体。\n\n一枚清晰的白色鼠标箭头带深灰描边，在贴纸之间移动。每次鼠标按下时：\n\n1. 鼠标尖端必须准确落在对应贴纸上；\n2. 出现一圈小型点击波纹；\n\n3. 被点击贴纸快速放大至108%，再弹回原位；\n\n4. 同一瞬间主角服装匹配剪辑切换；\n5. 人物身体、手势、站位和背景完全不动；\n6. 伴随清脆“咔哒”声和短促“唰”声。\nSHOT 1（00:00-00:02.00）初始造型\n女主穿第一套黑色时尚造型站在公寓中央：黑色修身吊带短裙，外搭黑色透明薄纱长袖上衣，银色细项链，黑色高跟鞋。\n她双手在腰前轻轻交叠，微笑看向镜头。六枚穿搭贴纸从画面边缘依次弹入，顶部标题“你选哪一套”淡入。\n\n00:01.50，鼠标移动至左上方浅绿色贴纸并点击。\n\nSHOT 2（00:02.00-00:04.00）浅绿色礼裙\n\n点击瞬间，服装切换为浅鼠尾草绿色单肩修身短裙：立体花朵肩饰、不对称斜领、褶皱束腰、雕花蕾丝拼接裙身、同色细高跟鞋。\n\n女主保持原站位，右手轻轻托起裙侧，身体向左转约20度，展示腰线后重新面对镜头。\n\n00:03.50，鼠标移动至左侧中部灰色贴纸并点击。\n\nSHOT 3（00:04.00-00:06.00）银灰色修身裙\n\n服装瞬间切换为高级银灰色挂脖修身短裙：不对称胸前剪裁、腰侧小面积镂空、侧边褶皱、银灰细高跟鞋。\n\n女主左手抱在腰前，右手指尖轻触下巴，神情冷艳自信。人物切换前后头部角度和双腿位置必须一致。\n\n00:05.50，鼠标点击右侧中部牛仔贴纸。\n\nSHOT 4（00:06.00-00:08.00）高级牛仔套装\n\n服装切换为深蓝短款牛仔夹克、象牙白修身背心、高腰深蓝牛仔短裤、金属腰链和白色高跟短靴。\n\n女主一手叉腰，另一只手做轻微展示动作，肩膀随音乐小幅摆动，表情自然俏皮。\n\n00:07.50，鼠标点击右上方紫色贴纸。\n\nSHOT 5（00:08.00-00:10.00）淡紫色层叠套装\n\n服装切换为淡紫色结构感挂脖上衣：立体收腰、不对称尖角下摆；搭配白色双层荷叶边短裙、淡紫色细高跟鞋。\n\n女主双手在脸侧短暂张开，随后一只手托在腰侧，裙摆因转身轻轻摆动。\n\n00:09.50，鼠标点击右侧下方黑白贴纸。\n\nSHOT 6（00:10.00-00:12.00）黑白名媛套装\n\n服装切换为黑色修身针织短开衫、白色包边与珍珠纽扣、白色内搭、黑白蕾丝层叠短裙、黑色高跟鞋。\n\n女主轻轻整理开衫下摆，侧身展示裙子的黑白蕾丝层次，再回头看向镜头微笑。\n\n00:11.50，鼠标移动至左下方紫色长裙贴纸并点击。\n\nSHOT 7（00:12.00-00:15.00）紫色压轴造型\n\n服装切换为白底紫色花卉露肩束身上衣，搭配高级深紫色高腰修身长裙和银色高跟鞋。长裙线条利落，面料带轻微丝绒光泽。\n\n00:12.00-00:13.50，女主双手沿上衣轮廓做简洁展示，然后右手叉腰，身体转向侧面约30度。\n\n00:13.50-00:15.00，鼠标箭头快速划过六枚贴纸，每枚贴纸依次轻微发光。女主回到正面，左手摊开指向贴纸菜单，右手叉腰，歪头看向镜头，露出询问式微笑。\n\n顶部“你选哪一套”微微增亮，所有贴纸保持清楚，画面定格结束。\n\n变装约束：\n\n每次换装只能由鼠标点击贴纸触发。不能在鼠标到达前换装；不能点击错误贴纸；主角穿上的服装必须与被点击贴纸完全对应。转场使用1—2帧匹配硬切加轻微白色轮廓闪光，不使用烟雾、爆炸、遮挡或身体融化。\n\nNegative：\n\n鼠标消失，多个鼠标，鼠标点空白处，点击位置错误，贴纸与服装不对应，未点击提前换装，自动连续换装，人物姿势跳变，人物位移，背景变化，镜头切换，服装融化，身体变形，脸部变化，发型变化，贴纸位置漂移，贴纸遮挡面部，重复贴纸，乱码贴纸，错误中文，标题变成其他文字，额外字幕，英文文字，服装串套，裙装变裤装，衣料穿模，多余手臂，多余手指，手指粘连，脚底滑动，塑料皮肤，动漫感，过度磨皮，低档出租屋，杂乱背景，水印，Logo",
    "sourceUrl": "https://x.com/johnAGI168/status/2094686415585972658",
    "sourceLabel": "X 原帖",
    "author": "johnAGI168",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094686415585972658.mp4",
    "posterUrl": "/posters/x/2094686415585972658.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/johnAGI168/status/2094686415585972658"
  },
  {
    "id": "x-2094667719497187408",
    "title": "巧克力液流组成饮品广告",
    "titleEn": "Chocolate Pours into a Drink Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "巧克力液流和杯形线条变成大标题与饮品标识，末段部分小字有变形。",
    "summaryEn": "Chocolate flow and cup outlines become a headline and drink branding, with some malformed small text near the end.",
    "prompt": "A 15-second 2D flat motion-graphics commercial in a clean, rich chocolate-drink style. Soft warm beige / pale cream paper-textured background. Limited palette: deep cocoa brown, creamy off-white, milk-chocolate amber, gold sparkle accents, and thin warm-cream highlight lines. Smooth, weighty liquid animation, glossy pours, and soft squash-and-stretch. No photorealism — bold graphic shapes with slight paper grain.\n0.0–1.2s\nScattered glossy dark-brown chocolate droplets of different sizes float and drift slowly across the empty cream background, each with a small cream highlight.\n\n1.2–2.5s\nThe droplets gather and pour downward in a thick, ribbon-like stream, forming a white-outlined short tumbler glass. Rich brown liquid fills it from the top and overflows in a rounded, velvety splash.\n2.5–4.0s\nA second graphic can tilts in from the top-right and pours a continuous chocolate stream into the glass. The liquid fills completely; cream-colored circular swirls rise inside. Small gold sparkles appear around the rim and splash.\n\n4.0–6.0s\nThe finished drink: short white-outlined glass of glossy chocolate, condensation dots as simple white circles, “Cokoa” lettering on the glass. A standing can appears beside it. Graphic chocolate chunks and a cocoa-powder pile slide in at the bottom-left. Blue-cream curved splash lines and gold four-pointed stars pop around the scene. The camera eases slightly closer; the pour feels thick, creamy, and alive.\n6.0–8.5s\nThe drink and splash dissolve into large, bold, condensed cream typography that reads THE PERFECT POUR. The letters first appear as if they are filled with the same flowing chocolate and cream swirls, then the liquid drains away leaving solid cream lettering sitting on a wide, organic dark-brown splash / wave shape that looks like overflowing chocolate.\n\n8.5–15s\nThe title stays locked on screen. Small brown droplets continue to bounce on and around the letters. Supporting text types on underneath in a clean condensed cream font: first CREAMY. then expanding to the short product line Indulge in the rich and creamy goodness of Cokoa.\nA tiny logo lockup appears at the bottom center: the word Cokoa plus a miniature standing can. Subtle swirl, droplet, and sparkle animation continues until the end.\nMood: rich, satisfying, slightly luxurious everyday chocolate treat. Timing should feel snappy and weighty — liquids have body, cream is soft, pours have momentum, text has a juicy “splash” reveal. Keep everything graphic, flat, and highly readable.",
    "sourceUrl": "https://x.com/Just_sharon7/status/2094667719497187408",
    "sourceLabel": "X 原帖",
    "author": "Just_sharon7",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094667719497187408.mp4",
    "posterUrl": "/posters/x/2094667719497187408.jpg",
    "duration": 15.168,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2094667719497187408"
  },
  {
    "id": "x-2094653824224940225",
    "title": "首尔街头的面包与花店",
    "titleEn": "Bread and Flowers on a Seoul Morning",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子穿过街区、走近花店，随后在街边品尝面包。",
    "summaryEn": "A woman walks through a neighborhood, visits a flower shop and tastes bread outdoors.",
    "prompt": "Ultra-photorealistic cinematic 15-second Korean lifestyle film following a young Korean woman with long dark hair in a low ponytail, wearing a fitted cream T-shirt, relaxed light-blue jeans, white sneakers, and carrying a natural canvas tote bag. Maintain consistent face, hairstyle, clothing, body proportions, and accessories throughout.\n\nShe leaves her apartment and walks through a quiet Korean residential alley in soft morning sunlight. She locks the door, adjusts her tote, and continues naturally toward a cozy neighborhood bakery.\n\nInside the bakery, she chooses a fresh loaf and small pastry. The baker places them into a paper bag, she pays, takes the bag, and carefully places it inside her tote. The added weight subtly changes the tote's position and swing.\n\nShe continues to a tiny flower shop, briefly touches a colorful bouquet as flowers move gently in the breeze, smiles at the shopkeeper, and walks on.\n\nShe reaches a small wooden bench, sits naturally, places her tote beside her, unwraps the pastry, and takes a bite. Two bicycles pass naturally in the background as she briefly watches them.\n\nThin clouds soften the sunlight while a gentle breeze moves her ponytail and nearby tree leaves. End on a warm close-up of her relaxed natural smile, with the Korean residential alley softly blurred behind her.\n\n100% photorealistic live-action, authentic Seoul neighborhood atmosphere, natural Korean lifestyle, realistic human biomechanics, believable gravity, inertia, friction, momentum, object weight, bicycle mechanics, paper movement, natural hair and fabric motion, consistent lighting and shadows, subtle handheld camera movement, realistic skin and textures, cinematic depth of field, soft natural color grading.\n\nNo CGI appearance, no artificial faces, no robotic movement, no teleportation, no floating objects, no distorted hands, no extra fingers or limbs, no changing face, clothing, or hairstyle, no duplicated people, no incorrect bicycle physics, no exaggerated wind, no excessive camera shake, no oversaturated colors, no unrealistic physics, no text, subtitles, logos, or watermark.",
    "sourceUrl": "https://x.com/nawalsehar/status/2094653824224940225",
    "sourceLabel": "X 原帖",
    "author": "nawalsehar",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094653824224940225.mp4",
    "posterUrl": "/posters/x/2094653824224940225.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "美食",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "food",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/nawalsehar/status/2094653824224940225"
  },
  {
    "id": "x-2094645801808969771",
    "title": "老胶片里的偷偷喂狗",
    "titleEn": "Secretly Feeding a Dog on Old Film",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子在庭院俯身照顾狗，随后看向拍摄者，保留家用胶片画面质感。",
    "summaryEn": "A woman bends to care for a dog in a courtyard and glances toward the camera in a home-movie film look.",
    "prompt": "ULTRA REALISTIC 15 SECOND HOME MOVIE. 1080p render of early 1950s 16mm Kodachrome film footage. 4:3 frame. Single continuous handheld take, no cuts. Title: WOH KUTTA.\n\nCONCEPT\n1953. A young woman is secretly feeding a street dog at the back door of the family house. She has been told not to. Her cousin is filming her doing it, which means there is now evidence. The dog behaves like it already lives there.\n\nSUBJECT LOCK\nIndian woman, early twenties, warm medium brown skin, realistic pores, no modern makeup, small bindi. Hair centre parted, pulled back into a low bun with strands loose at the temples. Cream hand knit cardigan with stretched cuffs over a plain cotton saree, thin gold bangles, simple leather sandals. Same face, hair and outfit for the full 15 seconds. Crouched on her heels, one hand shielding a steel bowl, glancing over her shoulder at the house.\n\nANIMAL LOCK\nMedium sized Indian street dog, short tan coat, one ear folded, ribs faintly visible, completely at ease. Sits before it is asked to. Eats fast, then leans its whole weight against her knee.\n\nSETTING\nBack doorstep of an older family house, early 1950s, late afternoon. Lime washed wall, a drain channel, a stack of firewood, a hand pump, wet patches on the stone, a neem tree throwing broken shade. Period accurate, no modern objects anywhere in frame. No visible brands, no logos, no readable signage.\n\nCAMERA\nEarly 1950s handheld 16mm spring wound movie camera at chest height, inexperienced operator. Framing too low, cutting off her head at the top of frame for two seconds before correcting. Fixed focus so the dog is sharp and her face is soft. Hand set exposure, sunlit wall blown to white, the shaded doorstep crushed to black. One jerk downward when the operator laughs. No zoom, no stabilization, no gimbal.\n\nFILM STOCK AND ARTEFACTS\nEarly Kodachrome, heavy warm saturation in reds and yellows, weak blues, green cast in shadow. Gate weave, frame breathing side to side. Dust and hair on the gate, one vertical scratch, one splice jump. Halation glow on highlights. Heavy grain. Undercranked 16 frames per second played at 24, so movement is slightly too fast and jerky. Rounded soft corners, mild vignette. No digital sharpness.\n\nBEATS\n0:00 to 0:04 She is crouched with the bowl, dog already eating, her attention entirely on the doorway behind her.\n0:04 to 0:08 She notices the camera, freezes for half a second, then keeps feeding it anyway and speaks the first line.\n0:08 to 0:12 Dead air. The dog finishes and leans into her leg. She scratches its folded ear and says nothing. Camera holds too long.\n0:12 to 0:15 A door sound from inside the house. She stands fast, hides the bowl behind her back, the dog does not move at all, and the spring runs down.\n\nDIALOGUE AND SOUND TREATMENT\nCamera records no sound. Audio is a thin separate wild track, mono, band limited, hissy, one to two frames out of sync with the lips.\n0:06 She says Kisi ko mat batana, samjhe\n0:13 She says Ye mera nahi hai, sach mein\nSpoken language locked to Hindi. Mouth movement roughly matches but never locks precisely. No other speech, no narration, no voiceover.\n\nAUDIO\nDiegetic only. Steel bowl scraping stone, the dog eating, crows, a hand pump handle, water on stone, a bicycle bell in the lane, a door latch inside the house, the mechanical whirr of the spring wound camera. Thin, mono, slightly hissy. No music, no score, no added ambience beds.\n\nSTYLE RULES\nNothing dramatic happens on purpose. No posing, no exaggerated reactions, no hero framing, no sentimental animal close ups. The dog is treated as a person she is covering for. Realism comes from the guilty glance, the low framing and the dead air nobody bothered to cut. If she performs for the lens the illusion breaks.\n\nNEGATIVE\nNo music. No subtitles. No captions. No watermark. No timestamp. No slow motion. No colour grading. No lens flare. No shallow depth of field. No modern clothing. No modern objects. No brand logos. No readable text.\n\nLAST FRAME\nHer standing with the bowl hidden behind her back, dog still sitting at her feet, frame weaving and underexposed, image fading as the spring winds down, then hard cut to black. Absolutely no on screen text, no title card, no lettering, no numbers, no logo, no signage of any kind in the final frame.",
    "sourceUrl": "https://x.com/bmx_ai13/status/2094645801808969771",
    "sourceLabel": "X 原帖",
    "author": "bmx_ai13",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094645801808969771.mp4",
    "posterUrl": "/posters/x/2094645801808969771.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "768p",
    "tags": [
      "电影叙事",
      "复古胶片",
      "生活实拍感",
      "动物与生物",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "ugc"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/bmx_ai13/status/2094645801808969771"
  },
  {
    "id": "x-2094643685338300751",
    "title": "从市集小吃走进雨巷",
    "titleEn": "From Market Snacks to a Rainy Alley",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子在市集购买食物，与摊主交流后走进湿润的窄巷。",
    "summaryEn": "A woman buys market food, interacts with a vendor and walks into a wet narrow alley.",
    "prompt": "Ultra-photorealistic cinematic 15-second Korean lifestyle film following a young Korean woman with long dark hair in a low ponytail, wearing a cream T-shirt, loose light-blue jeans, white sneakers, and carrying a beige reusable shopping bag. Maintain consistent face, hairstyle, clothing, proportions, and accessories throughout.\n\nShe enters a lively traditional Korean market, walking naturally between crowded stalls while vendors call to customers and shoppers move around her. She stops at a produce stall, gently inspects peaches and tomatoes, selects fresh produce, and places them into her reusable bag. The bag swings naturally with her movement and becomes slightly heavier.\n\nShe approaches a hot-food stall where a vendor stirs sizzling food in a metal pan. Steam rises naturally from the heat. The vendor offers her a small sample on a wooden pick. She tastes it and reacts with a subtle, genuine smile.\n\nShe then pays the vendor with several coins, creating a soft metallic sound, secures her shopping bag, and walks toward the market exit.\n\nTransition into a narrow traditional Korean residential alley. As she walks beneath tiled rooftops, light rain begins to fall. The pavement gradually becomes wet, small ripples form in shallow puddles, and subtle reflections appear on the ground. She looks upward, smiles, pulls the shopping bag closer, and continues walking as the rain gently increases.\n\nEnd with a cinematic rear tracking shot as she disappears deeper into the peaceful rainy alley.\n\n100% photorealistic live-action, authentic Korean market atmosphere, natural human acting, realistic food preparation, physically accurate object interaction, believable gravity, momentum, friction, weight, steam convection, rain dynamics, wet pavement, natural hair and clothing movement, documentary-style handheld camera, realistic skin texture, cinematic depth of field, natural lighting, restrained color grading, smooth transitions, authentic ambient market and street sounds.\n\nNo CGI look, no artificial faces, no robotic movement, no teleportation, no floating objects, no duplicated people, no changing face, clothing, or hairstyle, no unrealistic steam or rain, no weightless shopping bag, no fantasy physics, no excessive camera shake, no slow motion, no text, subtitles, logos, or watermark.",
    "sourceUrl": "https://x.com/aiwithaly/status/2094643685338300751",
    "sourceLabel": "X 原帖",
    "author": "aiwithaly",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094643685338300751.mp4",
    "posterUrl": "/posters/x/2094643685338300751.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "美食",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "food",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/aiwithaly/status/2094643685338300751"
  },
  {
    "id": "x-2094640598733234488",
    "title": "雪山上的单板滑行与腾空",
    "titleEn": "Snowboarding across Mountain Slopes",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝衣滑雪者调整护目镜后沿雪坡滑行，并完成一次腾空动作。",
    "summaryEn": "A snowboarder in blue adjusts her goggles, carves down a slope and goes airborne.",
    "prompt": "Create a photorealistic cinematic winter sports video of a young female snowboarder at a breathtaking high-altitude snowy mountain resort. She wears a bright blue ski jacket, black snow pants, black gloves, and a white helmet with reflective ski goggles.\n\nStart with a close-up as she adjusts her goggles and smiles confidently at the camera, surrounded by vast snow-covered mountains under a crystal-clear blue sky. Transition into dynamic action shots of her snowboarding rapidly down a pristine slope, carving smoothly through fresh powder and leaving a dramatic trail of snow behind her. Follow her from behind and from low-angle tracking shots as she approaches a large snow ramp, launches into the air, performs a stylish snowboard trick, and lands smoothly.\n\nUse realistic snow physics, natural body movement, detailed mountain textures, crisp sunlight, cinematic camera movement, slight motion blur, dramatic wide-angle aerial shots, and an energetic luxury sports-commercial aesthetic. Vertical 9:16, ultra-realistic, high detail, smooth transitions, natural lighting, 4K cinematic quality.",
    "sourceUrl": "https://x.com/CaliraVal/status/2094640598733234488",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094640598733234488.mp4",
    "posterUrl": "/posters/x/2094640598733234488.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "自然风景",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2094640598733234488"
  },
  {
    "id": "x-2094640015833960801",
    "title": "黑红拼贴中的动漫群像",
    "titleEn": "An Anime Ensemble in Black-and-Red Collage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "多位动漫角色与黑红字体、棋盘格和装饰元素交替出现。",
    "summaryEn": "An anime ensemble appears among black-and-red lettering, checkerboards and decorative elements.",
    "prompt": "15秒、16:9。高品質2DアニメPV / グラフィックMV。\n\n1、 2、 3、 4、 5 の5人を主要キャラクターとして使用する。\n\n6 は終盤に使用する総集ビジュアル参照。\n\n参照画像そのものの構図、背景、ポーズ、カメラ位置をそのまま再現しない。\n\n各参照は人物同一性・顔・衣装・装備・世界観の基準として使用し、映像内では新しいポーズ、新しいアングル、新しい構図へ再構成する。\n\n4はキャラクターと装備のみ参照し、キャラクターシートのレイアウト、図表、文字、説明欄、白背景を映像へ持ち込まない。\n\n6は一枚絵としてそのまま貼り付けず、終盤の総集演出・最終コラージュ・エンドカード演出の参考として使う。\n\nセリフなし。字幕なし。説明文なし。\n\n画面内に存在する文字はすべて映像演出としてのタイポグラフィであり、字幕として配置しない。\n\n全体コンセプト\n\n退廃的で華やかなゴシック・アニメPV。\n\n基本色：\n\n黒 / 赤 / 白\n\nアクセントとして各キャラクター固有色を残す。\n\n画面外周には細密なゴシック装飾フレーム。\n\n背景には、\n\n黒い虚空\n\n赤く発光する床\n\n白黒格子\n\nゴシック建築の抽象シルエット\n\n万華鏡\n\n赤い円環\n\n巨大な手\n\n鎖\n\n幾何学模様\n\nを使用する。\n\n全体を一つの現実空間として描く必要はない。\n\nアニメPV、漫画PV、ファッション映像、タイトルシークエンス、コラージュMVを融合した映像空間として構成する。\n\nテンポは非常に速い。\n\n0.5秒ごとに、人物・構図・タイポグラフィ・背景・カメラ・ポーズのいずれかを必ず大きく変化させる。\n\nタイポグラフィ\n\n使用する単語：\n\nENVY\n\nPRIDE\n\nWRATH\n\nGREED\n\nLUST\n\nGLUTTONY\n\nSLOTH\n\n中心語は ENVY。\n\n文字は極太の縦長サンセリフ。\n\n文字を単純に読ませるのではなく、\n\n巨大な壁\n\n人物を遮る前景\n\n背景建築\n\n仮面\n\n額縁\n\nトンネル\n\n床\n\n万華鏡の断片\n\nシルエット内部\n\nキャラクター同士の境界\n\n画面を分断する板\n\n崩壊する空間構造\n\nとして扱う。\n\n文字が人物の前と後ろの両方を通る。\n\n文字の一部が顔、腕、身体を隠してよい。\n\n文字には必要に応じて、\n\n薄い水平グリッチ、走査線、色収差、赤い発光、短いフラッシュ、残像\n\nを使用する。\n\nただしサイバーパンクにはしない。\n\nあくまでゴシック × 退廃 × ファッション × タイポグラフィを維持する。\n\nタイムライン\n\n0.0–0.5秒\n\n5。\n\n黒背景。\n\n中央に静かに立つ 5を、左右から現れる巨大な赤い手が囲む。\n\n背後には巨大な白文字 ENVY。\n\n文字の背後に赤黒い万華鏡が高速回転。\n\nカメラがわずかに前進。\n\n0.5–1.0秒\n\n高速フラッシュ。\n\n3を左右反転した二重像として配置。\n\n同一人物を鏡像化した疑似双子構図。\n\n巨大な PRIDE が前景を横切り、二人の顔を交互に遮る。\n\n背景は白黒格子の万華鏡。\n\n1.0–1.5秒\n\n3の顔アップ。\n\n手を顔の前で合わせる。\n\n手前に巨大な手のシルエット。\n\nPRIDE の文字が左右へ開き、その隙間から瞳が現れる。\n\n急速なズームイン。\n\n1.5–2.0秒\n\n3が手を交差。\n\nカメラを少し傾ける。\n\n白黒幾何学背景。\n\n細長く引き伸ばされた LUST が人物の前後を蛇行する。\n\n2.0–2.5秒\n\n赤背景へ瞬間切替。\n\n2を完全な黒いシルエットとして後ろ姿から見せる。\n\nフードと衣装の輪郭だけ残す。\n\n巨大な SLOTH が背景の建築物のように垂直に立つ。\n\n髪と衣装だけがゆっくり揺れる。\n\n2.5–3.0秒\n\n1と 2。\n\n二人をわずかに重ねたコラージュ。\n\n片方は正面、片方は横顔。\n\nピンク、紫、黒の万華鏡。\n\n中央を巨大な ENVY が通過し、二人を分断する。\n\n3.0–3.5秒\n\n2。\n\nローアングル。\n\nカメラへ向かって手を伸ばす。\n\n巨大な GREED が奥からカメラ方向へ高速で迫る。\n\n文字と人物が同一遠近空間に存在する。\n\n3.5–4.0秒\n\n赤一色の背景。\n\n4と 5を黒いシルエット化。\n\n左右から向き合う。\n\n5が手を伸ばす。\n\n二人の間を WRATH が斜めに横断。\n\n文字中央に亀裂が走る。\n\n4.0–4.5秒\n\n3を二重化した鏡像構図。\n\n二人が背中合わせで低く屈む。\n\n黒背景。\n\n赤く光るひし形ステージ。\n\n床面に巨大な PRIDE が反射している。\n\n4.5–5.0秒\n\n鏡像 3が向き合う。\n\n中央で片手を合わせる。\n\n左右に分割された ENVY が中央で接続する。\n\n完全な左右対称構図。\n\n5.0–5.5秒\n\n鏡像 3。\n\n極端な顔寄り。\n\n二人が背中合わせ。\n\nカメラが高速で周囲を約30度回り込む。\n\nPRIDE / ENVY の巨大文字が多重露光のように前後へずれる。\n\n5.5–6.0秒\n\n1と 5。\n\n背中合わせで立つ。\n\n黒い虚空。\n\n足元だけ赤いひし形ステージ。\n\n背景奥へ GREED が何層も連続し、巨大な文字トンネルを作る。\n\n6.0–6.5秒\n\n1が前方へ回り込む。\n\n身体を低く落とす。\n\nカメラも同時に下降。\n\n巨大な GLUTTONY が横方向へ圧縮され、画面全体を押し潰す。\n\n6.5–7.0秒\n\n1が膝をつく。\n\nその背後に 5。\n\n5がカメラへ手を伸ばす。\n\n背後の巨大な ENVY が縦長の祭壇のように立つ。\n\n7.0–7.5秒\n\n3と 4。\n\n赤いひし形ステージ上。\n\n二人とも低い姿勢。\n\nカメラは低い位置から急接近。\n\n周囲を SLOTH の巨大文字が壁のように囲む。\n\n7.5–8.0秒\n\nさらに二人が沈み込む。\n\n背景が完全な黒へ。\n\n人物だけに細い赤と白のリムライト。\n\nSLOTH の輪郭文字が暗闇へ沈む。\n\n音楽も一瞬だけ音数を減らす。\n\n8.0–8.5秒\n\n高速横ワイプ。\n\n画面を複数の水平帯へ分割。\n\nそれぞれの帯に、\n\n1、 2、 3、 4、 5 の目元、手元、横顔、装備、シルエットを一瞬ずつ表示。\n\n同時に、\n\nENVY / WRATH / PRIDE / GREED の文字断片が万華鏡のように回転。\n\n8.5–9.0秒\n\n3。\n\n真正面。\n\n手を顔の前で合わせる。\n\n巨大な ENVY が顔の前に存在し、文字そのものが仮面になる。\n\n9.0–9.5秒\n\n@3が手を開く。\n\n同時に ENVY が左右へ裂ける。\n\n文字の隙間から顔が現れる。\n\nカメラへ高速ズーム。\n\n白黒格子も放射状に広がる。\n\n9.5–10.0秒\n\n3の手が下へスライド。\n\n巨大な黒い PRIDE が上から落下。\n\n人物の大部分を覆う。\n\n薄い赤グリッチ。\n\n完全暗転する直前で切替。\n\n10.0–10.5秒\n\n5。\n\n冒頭構図へ回帰。\n\n巨大な赤い手。\n\n黒背景。\n\n巨大な ENVY。\n\n今回は文字をより前景へ配置し、 5の身体の一部まで遮る。\n\n10.5–11.0秒\n\n1と 2。\n\n二人を重ねる。\n\n背景はピンク・赤・黒の万華鏡。\n\n細長い ENVY が二人の周囲をリボンのように旋回する。\n\n11.0–11.5秒\n\n2。\n\nカメラへ急接近。\n\n片手を前へ。\n\n巨大な GREED も同時にカメラへ迫り、文字が画面いっぱいになる。\n\n11.5–12.0秒\n\n赤背景。\n\n4のシルエット。\n\n衣装と装備の輪郭を強調。\n\n髪と布が横風になびく。\n\n一方の手を伸ばす。\n\n巨大な WRATH がその手の方向から裂ける。\n\n12.0–12.5秒\n\n3を再び鏡像二重化。\n\n暗い円形ステージで二人が膝をつく。\n\n周囲に巨大な円環状 SLOTH / PRIDE 文字がゴシックの祭壇のように回転。\n\n12.5–13.0秒\n\n鏡像 3が中央で手を合わせる。\n\n左右から ENVY が扉のように閉じる。\n\n閉じる直前、二人の顔だけが文字の隙間に残る。\n\n13.0–13.5秒\n\n1と 5。\n\n再び背中合わせ。\n\n赤い床が一気に点灯。\n\n背景奥へ GREED / GLUTTONY / GREED / GLUTTONY が交互に連続。\n\n文字トンネルが高速で奥へ後退する。\n\n13.5–14.0秒\n\n1が低く身をかがめる。\n\n5はその背後からカメラへ手を伸ばす。\n\n画面全体の文字、フレーム、格子が砕け始める。\n\n破片が万華鏡へ吸い込まれる。\n\n14.0–14.5秒\n\n高速モンタージュ。\n\n1、 2、 3、 4、 5 を0.1秒前後ずつ断続的に切り替える。\n\n顔アップ、手、目、横顔、全身シルエット。\n\n周囲に七つの単語が断片化して飛散。\n\n飛散した破片が一点へ収束し、 6の構図へ向かう予兆を作る。\n\n14.5–15.0秒\n\n6を基準にした総集コラージュ・エンドカード。\n\nここで初めて、5人すべてを一つの画面に集約する。\n\nただし@6をそのまま静止画として貼らず、動く最終ビジュアルとして再構成する。\n\n中央に 5、背後に巨大な ENVY。\n\n周囲に 1、 2、 3、 4 をバランス良く配置。\n\n左右から巨大な赤い手。\n\n外周にはゴシック装飾フレーム。\n\n周辺に PRIDE / WRATH / GREED / LUST / GLUTTONY / SLOTH をコラージュ配置。\n\n小さな揺れ、髪のなびき、文字の脈動、微細なズーム、赤いグリッチ、万華鏡の回転を入れて、完全静止にはしない。\n\n最終瞬間、\n\n中央の ENVY がわずかに明滅し、\n\n赤い水平グリッチを一発だけ走らせて、\n\n黒へ落として終了。\n\nカメラ・編集ルール\n\n0.5秒以上、同一構図を固定しない。\n\n静止画を順番に並べたスライドショーにしない。\n\nキャラクター自身も、手、髪、身体、視線、姿勢を継続的に動かす。\n\n顔アップ → 全身 → 横顔 → ローアングル → シルエット → 対称構図を激しく切り替える。\n\n同一人物でも毎回異なる構図を使用する。\n\n参照画像をそのまま画面へ貼らない。\n\n1〜 5を融合させない。\n\n別キャラクターの顔、髪、衣装、装備を混ぜない。\n\n鏡像演出では同じ人物を明確に複製し、別人物との融合にしない。\n\n巨大文字は必ず3D的な奥行き、遮蔽、前後関係を持たせる。\n\nゴシック装飾フレームは存在するが、映像を窮屈にしすぎない。\n\nグリッチは切替補助として短く使用し、映像全体を常時壊さない。\n\n6は終盤の総集演出専用。中盤以前では直接再現しない。\n\nラストの パートも「静止ポスター化」ではなく、PVとしての動きと奥行きを持たせる。音速いジャズ・スウィング。ブラス、ウッドベース、ブラシドラムを中心にする。電子EDMにはしない。文字の出現には短い打楽器アクセント。画面切替にはブラシ、シンバル、ブラスヒットを同期。8秒付近だけ一瞬音数を減らす。8.5秒から再加速。14秒以降は急激に密度を上げる。14.5秒以降、 6の総集エンドカードでは主旋律を最も強くまとめる。15.0秒で短く明確に終了する。",
    "sourceUrl": "https://x.com/haruuraeadss/status/2094640015833960801",
    "sourceLabel": "X 原帖",
    "author": "haruuraeadss",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094640015833960801.mp4",
    "posterUrl": "/posters/x/2094640015833960801.jpg",
    "duration": 15.296,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2094640015833960801"
  },
  {
    "id": "x-2094628597126558123",
    "title": "粉彩洗衣房里的流行表演",
    "titleEn": "A Pop Performance in a Pastel Laundromat",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "双马尾女子在粉彩洗衣房里对镜头表演，鱼眼近景强调人物与设备。",
    "summaryEn": "A woman with twin ponytails performs in a pastel laundromat, using close fisheye views of her and the machines.",
    "prompt": "Create a 15-second,16:9,24fps,ultra-realistic live-action high-energy Y2K cute-girl music video with an original English http://song.Style:Cute Bubblegum Pop+Kawaii Electro Pop+Light Jersey Club+Y2K Girly Pop.CUTE,KAWAII,SWEET,PLAYFUL,ENERGETIC,slightly chaotic but never childish,sexy,cool,or aggressive.ONLY ONE GIRL,around 20,young Western woman.Long creamy-blonde hair in high fluffy twin tails with soft curls,wispy bangs,baby-blue and pastel-pink hair clips.Small soft face,large natural brown eyes,long lashes,soft brown eyeliner,pink blush,rosy nose tip,glossy pink lips,realistic skin texture.Outfit:cream puff-sleeve cropped top with pale-blue bow,baby-blue pleated mini skirt,white slouch socks,cream platform sneakers,pale-blue beaded necklace,minimal silver jewelry,pastel manicure.SAME GIRL,SAME FACE,SAME HAIR,SAME OUTFIT,SAME MAKEUP throughout.Scene:bright dreamy retro pastel laundromat in cream,baby blue,pastel pink,and creamy yellow.Rows of large pastel front-loading washers with circular glass http://doors.Pink,blue,and white clothes rotate http://inside.Blue/pink laundry baskets,transparent blue detergent,colorful socks,light checkerboard floor.Bright,clean,high http://exposure.MUSIC:original English cute pop song,148–152BPM,bouncy drums,cute electronic bass,bright synths,bubble SFX,claps,digital chimes,sparkling percussion.Sweet energetic young female vocal.Sample lyrics:“Round and round,I’m feeling good,spin it faster like I knew you would,bubble pop,come dance with me,little bit of magic,one two three,everything feels right,round and round we go.”The girl MUST visibly sing whenever vocals play.ACCURATE ENGLISH LIP SYNC.VISIBLE MOUTH MOVEMENT.Clear vowel shapes,consonant stops,natural breathing.Never close her mouth while vocals http://continue.CAMERA:12–14mm fisheye,playful mobile-MV energy.Dynamic push-ins,pull-backs,lateral tracking,low angles,and small rotations.Hands,shoes,and props may become oversized foreground elements without severely distorting her face.0–1.7s:She sits sideways on a pastel-pink washer,legs gently swinging.She pokes one cheek with one index http://finger.As“Round and round…”begins,she looks into camera,sings,puffs her cheek,winks once,and lightly http://bounces.Camera quickly pushes closer from a low angle.Tiny blue/pink stars pop beside her face.1.7–3.3s:She hops down and leans toward the fisheye,extending ONE index finger toward lens while singing“I’m feeling good…”.The fingertip enlarges in foreground.When it nearly touches lens,a transparent ripple expands with tiny bubbles.She immediately retracts it.3.3–5s:She clearly picks up a transparent pale-blue detergent bottle from beside her and briefly uses it as a funny microphone while singing“Spin it faster…”.She rocks her shoulders,twin tails bouncing.On the beat,she moves the bottle away and sings directly to camera.Small bubbles appear around the bottle.5–6.7s:She crouches beside a rotating baby-blue washer,one hand on the circular rim.She watches the colorful clothes spin,then suddenly turns toward camera with wide delighted eyes and a cute head tilt while continuously http://singing.Camera lowers with her.6.7–8.3s:On the beat,she stands and makes a happy half-turn,twin tails flying naturally.She briefly holds both twin-tail ends,shakes left once→right once→http://releases.Blue stars and pink sparkles trace the hair for 0.3s.8.3–10s:She performs one clean finger-dance:right index points down→left index slides sideways→wrists cross→both hands open like a flower.Every movement hits the beat.She continuously lip-syncs.Short blue lines,pink stars,and bubbles follow the hands.10–11.7s:A baby-blue laundry basket fills the foreground.She pops her head and upper body from behind it,both hands on the http://rim.Camera pushes into a fisheye close-up.She keeps singing while changing expressions:cheek puff→wide eyes→tiny nose scrunch→wink→bright smile.Bubbles bounce around the basket.11.7–13.3s:She naturally takes one fluffy pink sock and one blue sock from the http://basket.Pink sock toward lens→retract→blue sock toward lens→retract.Her head playfully moves left→right behind them while singing.Tiny sparkles trace the socks.13.3–15s:She clearly puts both socks back,then quickly sits on the original pink washer.She playfully extends ONE cream sneaker toward the fisheye so it briefly becomes large in foreground,then http://retracts.Camera pulls back slightly.She sings the final word,opens both palms beside her face like a flower,tilts her head,and smiles with crescent http://eyes.Final beat:transparent bubbles,pink stars,blue sparkles,white flowers,candy-like light particles,and soft Y2K doodles burst outward.Freeze on her joyful smile.Effects must react to actions and beats only:touch=ripple,hair=stars,socks=sparkles,final beat=bubble burst.EVERY SECTION MUST HAVE A DIFFERENT http://ACTION.NO REPEATED http://POSES.NO http://V-SIGNS.NO FINGER http://HEARTS.NO CAT http://PAWS.NO SEXY http://POSES.NO DARK http://CYBERPUNK.NO http://NIGHTCLUB.NO AGGRESSIVE http://EXPRESSIONS.NO RANDOM SCENE http://CHANGES.NO CHARACTER http://SWITCHING.NO OUTFIT http://CHANGES.NO EXTRA http://PEOPLE.NO EXTRA http://ARMS.NO EXTRA http://FINGERS.NO DUPLICATED http://HANDS.NO OBJECTS APPEARING FROM http://NOWHERE.NO FROZEN MOUTH DURING http://VOCALS.NO http://SUBTITLES.NO LYRICS ON http://SCREEN.NO http://TEXT.NO http://LOGO.NO WATERMARK.",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2094628597126558123",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094628597126558123.mp4",
    "posterUrl": "/posters/x/2094628597126558123.jpg",
    "duration": 15.37161,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "商业广告感",
      "照片级写实",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2094628597126558123"
  },
  {
    "id": "x-2094580769905934790",
    "title": "柚子汽水罐的图形广告",
    "titleEn": "A Graphic Ad for Yuzu Sparkling Water",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝绿色汽水罐、拉环和柑橘图案在平面广告构图中连续切换。",
    "summaryEn": "A teal drink can, pull tab and citrus motifs cycle through flat graphic ad compositions.",
    "prompt": "Flat 2D vector motion graphics commercial advertisement, 15 seconds, 16:9 widescreen, high energy product-focused ad style, zero storytelling, zero narrative arc, pure fast paced product hero showcase. Inspired by Pinterest botanical sparkling-can packaging (clean fruit-soda cans, Twist-style fruit fizz, Petali botanical soda, tropical mocktail cans) but fully original. Tokyo modern editorial pop with risograph print sensibility, refined premium energy, non-stop motion in every single frame, hard cuts only, no dissolves, no static holds until the final end card. Subtle paper grain and risograph ink texture throughout. Cool porcelain background base color hex F4F1EA, generous negative space, grid-aligned composition. Locked palette of three colors only: deep ink teal hex 123C3A, luminous citron hex E8F06A, pure white, plus faint ice-mint highlights. No other colors anywhere in frame. The product is a fictional sparkling yuzu soda brand called SOLA, a sleek deep teal can with a minimal white sun-burst yuzu slice emblem and a thin citron ring accent, consistent in every scene.\nAUDIO LAYER, drives the cut timing: warm punchy French-house-flavored electronic commercial track, four on the floor kick starting at 0.0s, bass drop synced to the can launch at 4.0s, whoosh sound effect on every camera whip pan, sharp crack sound effect on the can opening, fizzy sustained sparkle sound during the fizz explosion, deep sub bass thud on every impact and squash landing, glass clink sound effect on every ice cube landing, citrus peel snap on fruit slice hits, rising riser sound effect building into the logo lock, satisfying snap sound effect on the logo lock, music hits a full stop exactly on the final frame.\nSCENE 1, 0.0s to 1.0s: bold editorial open, a sweeping citron light ray wipes diagonally across the clean porcelain background in one confident motion, revealing a rising deep teal silhouette taking shape behind it, sharp and deliberate, zero clutter, premium controlled energy, a thin hairline teal rule draws itself across the lower third, kick drum hits exactly on the wipe.\nSCENE 2, 1.0s to 2.0s: hard cut, the silhouette snaps into full color and detail as the SOLA soda can in a single confident beat, condensation already gleaming on its surface as flat white crescent shapes, soft upward white rays radiating behind it, camera holds a slow deliberate push in, poised and polished rather than chaotic.\nSCENE 3, 2.0s to 3.0s: hard cut, the circular white yuzu-slice sun emblem with a citron ring flashes onto the can face with a sharp lens flare burst and a satisfying mechanical snap, one clean ring of citron light pulses outward once and fades, tiny white small-caps microtype arcs around the emblem, confident and premium.\nSCENE 4, 3.0s to 4.0s: quick controlled zoom out, the can now fully formed hangs centered and commanding in frame over vast porcelain negative space, camera whip reframes to hero angle, tension builds for one beat, bass drop hits exactly as the can launches.\nSCENE 5, 4.0s to 5.2s: hard cut, the can launches downward from its hover like a meteor, trailing a white ice crystal comet tail and curved teal speed lines, slams into center frame with heavy squash and stretch, ground cracks outward in a starburst pattern, radial citron and teal shockwave ring bursts from the impact point, sub bass thud lands on impact.\nSCENE 6, 5.2s to 6.2s: hard cut, six chunky white ice cubes cannonball into frame from both left and right edges in a fast alternating rhythm, each landing throws flat white splash droplets outward in a fan burst, glass clink sound effect on every landing, camera whip pans following the rhythm left then right.\nSCENE 7, 6.2s to 7.4s: hard cut, extreme close push into the can top, the tab cracks open in one sharp motion, crack sound effect syncs exactly to the open, a vertical explosion of white fizz and faint ice-mint sparkling soda rockets straight up past the top of frame, camera shakes once on the burst.\nSCENE 8, 7.4s to 8.8s: hard cut, three flat white yuzu wheels slice in from the top edge and pin themselves around the can in a tight orbit, each slice throws a thin citron juice arc, citrus peel snap on every hit, camera orbits 20 degrees around the hero can, still locked to 16:9 center.\nSCENE 9, 8.8s to 10.2s: hard cut, the can tilts and pours a graphic stream of white sparkling liquid into a waiting teal-outlined glass at lower right, liquid fills in stepped beats, ice cubes already inside bounce once, fizz beads climb the glass wall as small white dots, whoosh on the pour start.\nSCENE 10, 10.2s to 11.6s: hard cut, three matching SOLA cans slam into a tight diagonal product stack from off-frame, each lands with squash and a teal shock ring, middle can stays hero-size, condensation crescents multiply, camera whip to a low 16:9 product-shelf angle.\nSCENE 11, 11.6s to 13.2s: hard cut, bold teal wordmark SOLA slams on from the left as oversized editorial type, citron underline draws itself in one stroke beneath it, micro line YUZU SPARKLING snaps under in tiny white small-caps, riser builds hard.\nSCENE 12, 13.2s to 15.0s: hard cut to end card, single centered SOLA can, emblem locked, vast porcelain negative space, teal wordmark and citron ring lock into final position with a snap, one last white fizz sparkle flickers once at the can rim then freezes. Hold only on the last frame. Music stops dead on 15.0s.",
    "sourceUrl": "https://x.com/AIwithJessica/status/2094580769905934790",
    "sourceLabel": "X 原帖",
    "author": "AIwithJessica",
    "publishedAt": "2026-09-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094580769905934790.mp4",
    "posterUrl": "/posters/x/2094580769905934790.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AIwithJessica/status/2094580769905934790"
  },
  {
    "id": "x-2094541985466216780",
    "title": "黄昏加油站旁的家庭录像",
    "titleEn": "A Home Video beside a Petrol Station",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子站在旧车与加油机旁对拍摄者说话，手持镜头留下倾斜和晃动。",
    "summaryEn": "A woman speaks beside an old car and petrol pumps while the handheld camera tilts and wobbles.",
    "prompt": "ULTRA REALISTIC 15 SECOND PERSONAL HOME VIDEO. 1080p 16:9 widescreen render of mid 1990s Hi8 camcorder tape footage. Single continuous handheld take, no cuts. Dialogue driven, Korean. Title: GASOLINE DREAMS.\n\nCONCEPT\n1997. A borrowed car at a rural highway petrol station outside Seoul, dusk in late autumn. Three hours still to drive. She is waiting for the tank to fill and her friend is filming instead of reading the map. Nothing happens. The tape is just running.\n\nSUBJECT LOCK\nKorean woman, early twenties. Realistic skin with visible pores, faint under eye shadow, minimal makeup, nose and ears pink from the cold, breath faintly visible. Dark brown hair loosely tied back, front strands loose, flattened on one side from sleeping against the car window. Oversized cream knit sweater with stretched cuffs pulled down over her hands, blue jeans, white sneakers scuffed at the toe. Same face, hair and outfit for the full 15 seconds. Slouched, stiff from sitting, stretching her back.\n\nSECOND VOICE LOCK\nThe friend holding the camcorder is never seen, not a hand, not a shadow, not a reflection in the car glass. Voice only, young female, close to the mic, sleepy and unbothered. Never addresses an audience. A station attendant in a padded work jacket is visible at a distance but never speaks.\n\nSETTING\nSmall rural petrol station on a two lane highway outside Seoul, late autumn dusk, 1997. Two pumps under a concrete canopy with one flickering fluorescent tube, a stack of oil cans, a plastic stool, a beige sedan with the fuel flap open and the passenger door hanging open. Beyond the canopy, bare rice paddies, a line of poplars, low grey hills, a truck passing, sky going orange to deep blue. Lived in and unstyled. No visible brands, no logos, no readable signage, no Hangul text anywhere in frame.\n\nFRAMING\n16:9 widescreen throughout, treated as a later widescreen transfer of the original tape. The extra width is used badly, not compositionally. She sits low and off to one side with dead empty space filling the other half of the frame. Horizon slightly tilted. Nothing centred on purpose.\n\nCAMERA\nRaw handheld Hi8 camcorder at chest height, operator leaning on the car. Autofocus hunting between the pump, the car glass and her face, repeatedly grabbing the wrong thing. Exposure pumping hard as the lens drifts from the bright sky to the shade under the canopy. One lazy unmotivated zoom that overshoots past her shoulder and corrects. She drifts out of frame once and the operator is late to follow. Analogue tape noise, chroma bleed on the orange sky, mild tracking wobble at the frame edges, one dropout glitch. No stabilization, no gimbal, no cinematic choreography.\n\nBEATS\n0:00 to 0:04 She climbs out of the passenger door mid stretch, arms up, yawning, and walks toward the pump. The camera swings after her too slowly and catches empty canopy for a beat.\n0:04 to 0:08 She stands at the open fuel flap watching the numbers on the pump. The exchange, overlapping, both half laughing.\n0:08 to 0:12 Dead air. She sits sideways on the car bonnet, feet not reaching the ground, a hot can held against her cheek, watching a truck pass. Nobody speaks for four full seconds. The fluorescent tube flickers.\n0:12 to 0:15 A horn from the road. She slides off the bonnet, says the last line without turning around, and the camera is still pointed at the empty bonnet when the tape cuts.\n\nDIALOGUE\nSpoken language locked to Korean, casual banmal between friends, flat and unprojected, thinned by the camcorder mic.\n0:05 She says Gadeuk chaewoya dwae, apeuro amugeotdo eopseo\n0:07 Friend off camera says Neo don isseo\n0:08 She says Dangyeonhi eopji\n0:13 She says Gaja, got eoduwojinda\nVisible mouth movement with bilabial closure on the m, b and p sounds. Friend line off camera only, closer to the mic. No other speech, no narration.\n\nAUDIO\nDiegetic only. Fuel pump motor humming, nozzle clicking off, fluorescent ballast buzz, a truck passing with a doppler horn, wind across open paddies, a can tab, her sneakers on grit, faint hand noise on the camcorder body. No music, no score, no radio, no added ambience beds.\n\nSTYLE RULES\nNothing dramatic happens on purpose. No posing, no road trip montage energy, no wind in hair, no golden hour beauty lighting, no hero framing. Realism comes from the cold, the flat delivery and the four seconds of dead air nobody bothered to cut. She is talking to her friend, not to viewers. If she performs for the lens the illusion breaks.\n\nNEGATIVE\nNo music. No subtitles. No captions. No watermark. No timestamp burn in. No slow motion. No colour grading. No lens flare. No shallow depth of field. No visible second person. No brand logos. No readable text.\n\nLAST FRAME\nWide crooked frame of the empty car bonnet under the flickering fluorescent tube with the dark highway filling the right half, image tears with one frame of tape distortion, then hard cut to black. Absolutely no on screen text, no title card, no lettering, no numbers, no logo, no signage of any kind in the final frame.",
    "sourceUrl": "https://x.com/bmx_ai13/status/2094541985466216780",
    "sourceLabel": "X 原帖",
    "author": "bmx_ai13",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094541985466216780.mp4",
    "posterUrl": "/posters/x/2094541985466216780.jpg",
    "duration": 15.16263,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "复古胶片",
      "生活实拍感",
      "人物表演",
      "日常生活"
    ],
    "category": "dialogue",
    "styles": [
      "retro",
      "ugc"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/bmx_ai13/status/2094541985466216780"
  },
  {
    "id": "x-2094472529939136738",
    "title": "小怪兽追逐激光红点",
    "titleEn": "A Little Monster Chases a Laser Dot",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "紫色独眼小怪兽扑向地板上的红点，最后躺倒在沙发上；原帖只给出 Prompt 片段。",
    "summaryEn": "A purple one-eyed creature pounces on a red dot and ends up on a sofa; the post gives only a prompt excerpt.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2094472529939136738",
    "sourceLabel": "X 原帖",
    "author": "GlennHasABeard",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094472529939136738.mp4",
    "posterUrl": "/posters/x/2094472529939136738.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动物与生物",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094385760514589091",
    "title": "URBAN BLACK 黑白红角色片",
    "titleEn": "URBAN BLACK in Black, White and Red",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "短发动漫人物与极简大字切换，黑白服装和红裤构成统一配色。",
    "summaryEn": "A short-haired anime character alternates with minimal oversized type in a black, white and red palette.",
    "prompt": "[REFERENCE]\nimage1 は今回のデザイン重視キャラクターシート1枚である。\nimage1 は顔、髪、年齢感、肌色、体型、身長感、頭身、全身比率、衣装、靴、アクセサリー、配色だけに適用する。\nimage1 内の URBAN BLACK、FRONT、SIDE、BACK、カラースウォッチ、区切り線、余白、シート固有の白背景デザイン、レイアウト枠は人物参照として使用しない。\nキャラクターデザインは全カットで固定する。別人化、髪型変更、衣装変更、配色変更をしない。\n人物は、黒の非対称ショートヘアと刈り上げ、シャープでワイルドな雰囲気、黒白のオーバーサイズジャケット、白いクロップトップ、鮮やかな赤の細身パンツ、白黒のハイカットスニーカー、シルバー系アクセサリーを維持する。\n\n[CONDITION]\n15秒ちょうど、1:1スクエア、24fps。高品質アニメ調シネマティックレンダリング。プレミアムAAA品質。全体の80％は大胆なモーショングラフィック、20％はキャラクターアクション。全体スタイルは、ストリートファッションの広告映像 × 高密度モーションタイトルシーケンス × 都市的で硬質なUIデザイン。グラフィック言語は、黒、白、赤、シルバー基調。金属フレーム、ウィンドウバー、ワイヤーグリッド、警告ライン、斜線、セグメント数字、円形スタンプ、パネルシャッター、ガラス片、速度線、CRT走査線を使用する。タイポグラフィ × アクションの関係性を強く維持し、文字やUIは背景ではなく、床、壁、障害物、通過ゲート、回転軸、破壊対象、着地点として使う。音楽プレイヤー中心ではなく、都市的で反骨的、少しワイルドな世界観にする。\n編集は極めて攻撃的。すべてのカットで構図を明確に変える。ハードカット、ウィップパン、スナップズーム、RGB split、シャッターフラッシュ、インパクトシェイク、ストロボ残像、スピードランプ、前景ワイプを強く使う。URBAN、REDLINE、BREAK、RISE、URBAN BLACK の主要タイポグラフィは、人物が重なる前に一度は明瞭に読み取れる状態を作る。その他の補助タイポグラフィは人物アクションとの同期を優先する。全体はちょうど13カット。カット数を増減しない。\n\n[SHOT FLOW]\n[Shot 1]\n0.00–1.00\n白に近いクリーンな背景へ、黒いタイトルバーと赤い警告ラインが勢いよく打ち込まれる。巨大タイポ URBAN が一文字ずつ強い衝撃で出現し、四角いUI枠、細い進行バー、短いセグメント線が点滅する。最後に赤い縦ラインがフレームを切断して次へ入る。\n\n[Shot 2]\n1.00–2.00\n彼女の目元、髪、刈り上げ側、シルバーアクセサリーを極端な寄りで見せる。黒と赤の矩形フレーム、硬質な円形ロックUIが顔の一部を一瞬ずつ切り取り、鋭い目線が上がった瞬間にフレームが赤黒のタイル片へ砕ける。\n\n[Shot 3]\n2.00–3.10\n巨大タイポ REDLINE が横長に出現し、文字列のベースライン自体が床になる。彼女が画面左から高速で走り込み、そのまま文字の上を低い姿勢でパワースライドし、文字片と赤い破片を蹴り上げる。最後は勢いを保ったままウィップパンで抜ける。\n\n[Shot 4]\n3.10–4.00\n黒背景に巨大な白いワイヤーグリッドが斜めに立ち上がり、壁のような構造になる。背景には縦長の SECTOR 01 が流れ続ける。彼女がグリッド壁を駆け上がり、中腹で強いキックオフを行い、同時に赤いブロックUIがビートに合わせて立ち上がって崩れる。\n\n[Shot 5]\n4.00–5.10\n荒い赤い円形タイポリング NO LIMIT が回転しながら画面前面に現れる。彼女がジャンプし、リング中央を突き抜けながら空中で身体をひねる。通過の衝撃でリングの文字が散り、カメラがその回転を追いながら一瞬だけ挑発的な目線を捉える。\n\n[Shot 6]\n5.10–6.00\n明るい背景に巨大な黒タイポ BREAK が現れる。彼女は B に片手をつき、身体を持ち上げてボルトで飛び越える。接触の瞬間だけ文字全体が圧縮され、バネのように反発して元へ戻る。\n\n[Shot 7]\n6.00–7.00\n白背景に黒い斜めメタルバーが一本走る。彼女がそのバーに沿ってバックフリップし、同時に巨大アウトライン文字 SHIFT が180度回転する。白、赤、黒のストロボ残像が短く3段で残り、回転の勢いを強調する。\n\n[Shot 8]\n7.00–8.00\nタイポグラフィの連打。WILD、SHARP、RUSH、EDGE の4語が一拍ごとに1語ずつ強く打ち込まれる。前景では彼女が膝スライドで横切り、ジャケットが大きく開いて翻り、足元から赤黒のタイポ片と細いライン粒子が爆発する。\n\n[Shot 9]\n8.00–9.00\n黒背景に巨大な3Dワイヤーグリッド面が手前へ傾いている。彼女がその面を走り上がり、途中で強く蹴り出して宙へ飛ぶ。空中で一瞬フリーズし、赤い円形スタンプ LOCK、小さなターゲットUI、短い目盛り、技術線が彼女の周囲にロックされる。\n\n[Shot 10]\n9.00–10.10\nフリーズが解放され、彼女がカメラへ向かって前進する。前方に4層の硬質なウィンドウパネルが並び、彼女は4枚を順番に1枚ずつ突き破る。各パネルが割れるたびに R、I、S、E が1文字ずつ露出し、4枚目を突破した瞬間に RISE が完成する。最後はスニーカーの前景ワイプで切る。\n\n[Shot 11]\n10.10–11.10\n1つのカット内で行う高速ポスターモンタージュ。4種類の全画面ポスターレイアウトが、追加のカットを作らず、シャッターフラッシュと瞬間的なグラフィック差し替えで高速に切り替わる。01 はウォールラン途中、02 はスライド、03 は着地直前、04 は鋭い立ち姿。各ポスターは大きな番号 01–04、赤い斜線、細いUI帯、短い警告記号を持ち、毎回構図を明確に変える。\n\n[Shot 12]\n11.10–13.00\n最大の見せ場。中央のクリーンな空間で、彼女が大きなバックフリップから真ん中へ着地する。着地の瞬間、赤い同心円、白い衝撃リング、黒いタイポ破片、風の筋、グリッド片が外側へ爆発し、画面全体を押し広げる。短い象徴的フリーズで鋭い決め顔を見せ、そのまま白飛び気味にオーバーエクスポーズして次へつなぐ。\n\n[Shot 13]\n13.00–15.00\n最終アイデンティティカード。淡い白背景に巨大な黒タイポ URBAN BLACK が主役として配置される。背後には薄い赤の円形スタンプ、シルバーの技術アーク、透明パネル、走査線が重なり、彼女はタイポに少し重なりながらリラックスした立ち姿で止まる。髪とジャケットだけがわずかに風で揺れ、最後に赤いパルスが文字全体を走り、ウィンドウクロームのフラッシュで映像を締める。\n\n[SOUND]\nBGMは、硬質で重い電子ビートを主体にした、ドラム主導のアグレッシブなエレクトロニックトラック。重いキック、短いグリッチ、サブヒット、ライザー、ドロップを各カットの切替に強く同期させる。\n効果音は、スニーカー着地、スライド、風切り、グリッド接触、ガラスやパネルの破壊、ウィンドウスラム、タイポ衝突、UIロック音をリズミカルに配置する。\nピークは Shot 12。最後は冷たい電子的ロゴスティンガーで終える。\n台詞、ナレーションなし。\n\n[NEGATIVE]\n武器、戦闘、炎、爆発炎上表現は入れない。",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2094385760514589091",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094385760514589091.mp4",
    "posterUrl": "/posters/x/2094385760514589091.jpg",
    "duration": 15.13941,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "时尚",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2094385760514589091"
  },
  {
    "id": "x-2094371536623092160",
    "title": "蓝皮肤街头角色的动态图形揭晓",
    "titleEn": "A Blue-Skinned Streetwear Character Reveal",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝皮肤角色、针织帽与格纹服装在大字和色块之间被逐步展示。",
    "summaryEn": "A blue-skinned character, knit cap and checkered outfit are revealed among oversized type and color blocks.",
    "prompt": "Create an explosive motion-graphics character reveal using MiniMax H3 on @Hailuo_AI.\n\nCHARACTER LOCK: Exact reference character: blue skin, messy black hair, orange beanie, cream hoodie with pink hood/polka dots, orange chest mark, black sling bag, CAF cup, teal/cream checker pants with BIG tag, blue socks, orange-black-white sneakers with pink laces. Preserve exact face, proportions, outfit, colors, materials and branding. No redesign.\n\nSTYLE: 80% kinetic graphics / 20% character action. Premium AAA title sequence × futuristic streetwear × NYC energy. Cream, teal, pink, orange, blue, black, white. Fast typography, portals, grids, stickers, checker patterns, halftone, UI marks and speed trails.\n\nCUT 01 | 0–1s — Black field. Orange dot expands into spinning teal/pink portal; character silhouette flashes.\n\nCUT 02 | 1–2s — Extreme eye close-up. Blink and sideways glance; scanning lines form around the face, then RGB flash.\n\nCUT 03 | 2–3.1s — Cream field, giant NO RUSH. Character walks through holding CAF cup; letters stretch with each step.\n\nCUT 04 | 3.1–4s — Six-panel character contact sheet rapidly rearranges into one portrait with pink arrows and graphic borders.\n\nCUT 05 | 4–5.1s — Teal city grid. Character walks along glowing route while orange circles appear beneath sneakers. Giant NEXT rotates behind.\n\nCUT 06 | 5.1–6s — Sneaker hits orange circle; checkerboard ripple expands and becomes STEP. Fast camera push.\n\nCUT 07 | 6–7s — Pink field. Character performs smooth 180° turn with three motion echoes. TURN rotates around them.\n\nCUT 08 | 7–8s — URBAN / SIMPLE / FRESH / BIG slam in rapidly as character crosses foreground, hoodie and bag moving naturally.\n\nCUT 09 | 8–9s — White field with moving black spotlight. Character raises CAF cup; steam becomes city lines. Orange UI lock-on.\n\nCUT 10 | 9–10.1s — Floating stickers surround character; each disappears on contact. Giant teal arrow becomes a full-screen wipe.\n\nCUT 11 | 10.1–11.1s — Rapid close-ups: eye, beanie, cup, sneakers, full character. Graphic frames and 01–04 flash between shots.\n\nCUT 12 | 11.1–13s — HERO: cream studio. Character walks center, stops with subtle smile. Portal explodes behind with teal rings, pink ribbons, orange shapes and checker tiles. Freeze → white flash.\n\nCUT 13 | 13–15s — Cream identity card. Huge black BIG, character relaxed beside it. Teal orbit, pink/orange marks and final checker flash. Logo pulses once → black.\n\nEDITING: Hard beat cuts, snap zooms, whip pans, speed ramps, freezes, RGB splits, wipes and impact shakes. Every cut visually distinct. Perfect character consistency.\n\nSOUND: Heavy electronic beat, sub-bass, glitches, whooshes, sneaker hits, cup clinks and UI sounds synced to cuts. Peak at CUT 12, cold logo stinger.\n\nQUALITY: Premium AAA cinematic × anime streetwear campaign, crisp character, natural motion, polished graphics. No weapons/combat/fire. Exactly 13 cuts, exactly 15 seconds.",
    "sourceUrl": "https://x.com/itxabdullaa/status/2094371536623092160",
    "sourceLabel": "X 原帖",
    "author": "itxabdullaa",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094371536623092160.mp4",
    "posterUrl": "/posters/x/2094371536623092160.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动态图形",
      "时尚",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/itxabdullaa/status/2094371536623092160"
  },
  {
    "id": "x-2094359148590969166",
    "title": "海岸列车微缩片场的揭晓",
    "titleEn": "Revealing a Miniature Coastal Train Set",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "海岸列车与浪花镜头转到拍摄现场，展示微缩布景的片场错觉。",
    "summaryEn": "A coastal train and waves give way to a filming setup, revealing a miniature-set illusion.",
    "prompt": null,
    "sourceUrl": "https://x.com/Rosey_watson/status/2094359148590969166",
    "sourceLabel": "X 原帖",
    "author": "Rosey_watson",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094359148590969166.mp4",
    "posterUrl": "/posters/x/2094359148590969166.jpg",
    "duration": 15.23229,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "城市街景",
      "产品与广告"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094333791904923946",
    "title": "格纹西装人物的复古图形片",
    "titleEn": "A Retro Graphic Portrait in Tartan",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "格纹西装人物出现在汽车和黑白大字之间，构成复古时装展示。",
    "summaryEn": "A tartan-clad character appears between car interiors and oversized black-and-white type.",
    "prompt": "Create an explosive, motion-graphics-driven character reveal trailer in 16:9, exactly 13 distinct cuts, 24fps. The character is the young woman from the reference still: pale freckled skin, rosy cheeks, soft green-blue eyes, long wavy champagne-blonde hair with blunt bangs and loose curls, gold floral drop earrings, a pink-and-white tartan blazer over a pale-yellow collared shirt with a matching blush-pink tie. She is seated in a cream vintage car interior with a white farmhouse and green field visible through the window in golden-hour light. Preserve her exact face, proportions, hairstyle, outfit, materials, accessories and colors throughout; never redesign her. This film is 80% bold graphic design in motion and 20% character action: giant kinetic typography, hard-edged shapes, split-screens, flat color fields, torn-paper edges, analog film frames, gold-foil stamps, plaid fragments, sunflare bursts and shutter flashes. Palette: blush pink, cream, pale yellow, champagne gold, sage, sky blue, black. Style: premium AAA motion-graphics title sequence x vintage-editorial campaign film. Every graphic element moves fast and snaps hard on the beat.\nCUT 01 | 0.00–1.00s - Pure graphics: a massive blush-pink circle slams onto a cream field, two gold bars wipe across, tiny film-sprocket ticks and a torn-plaid strip flicker; the letters G and H punch in one after another with heavy impact shake.\nCUT 02 | 1.00–2.00s - The H becomes a car-window frame: extreme close-up of her green-blue eye inside it, glancing toward camera; analog light-leak flash, then the frame shatters into shards of flat pastel color.\nCUT 03 | 2.00–3.10s - Cream frame with an enormous black GH; she leans in from frame left and slides across the baseline of the typography, gold dust and pink plaid ribbons trailing, shards of the letters kicked up like petals. Whip-pan out.\nCUT 04 | 3.10–4.00s - Split-screen triptych of flat blush / cream / sage panels; in each panel her silhouette performs a different pose in stroboscopic freeze: look-back, hair toss, settle. Huge vertical text GOLDEN HOUR scrolls behind the panels.\nCUT 05 | 4.00–5.10s - She rises through a giant rotating ring of typography reading SOFT LIGHT HARD STARE; camera tracks her as she turns toward us in slow motion as the letters scatter, then snap-zoom on her confident face.\nCUT 06 | 5.10–6.00s - Hard cut: cream editorial card, huge black BLOOM with a gold slash; she leans over the word itself, fingers grazing the B, hair whipping across frame; the word compresses like paper under her hand and rebounds.\nCUT 07 | 6.00–7.00s - Pale-yellow field with a black diagonal stripe; she turns along the stripe in three stroboscopic ghost frames, each ghost tinted a different palette color; giant outlined text TURN rotates 180 degrees in sync with her motion.\nCUT 08 | 7.00–8.00s - Kinetic type barrage: the words WARM / SHARP / PRETTY / WILD slam in one per beat with shutter flashes and camera shake, while she slides across the foreground in the car seat, blazer flaring, gold dust bursting from the window light.\nCUT 09 | 8.00–9.00s - Black frame, giant white wireframe farmhouse silhouette tilting in 3D; she sits up through the grid like a rising portrait, then hangs frozen; a blush circle stamps around her pose like a target lock with foil brackets and sprocket ticks.\nCUT 10 | 9.00–10.10s - Freeze releases into a burst: she leans toward camera through layered flat-color panes that shatter one by one like glass shutters, each shattering pane revealing a bigger letter of G-H; foreground wipe with her earring.\nCUT 11 | 10.10–11.10s - Rapid-fire poster montage: four full-screen graphic posters of her in different still poses (window gaze, three-quarter, chin down, hands on lap) snap past with hard cuts, each with different bold layout, oversized numbers 01–04, plaid slashes and gold stamps.\nCUT 12 | 11.10–13.00s - Hero moment on a clean cream cyclorama: she settles dead center in slow motion, lifts her chin, and a shockwave of concentric blush rings, sunflare streaks and shattered typography blasts outward; brief iconic freeze with her faint closed-mouth look, overexpose to white.\nCUT 13 | 13.00–15.00s - Final identity card: enormous black GH dominates a pale cream field with translucent gold rings, technical arcs, halftone and a rough black circular emblem containing a ghosted farmhouse/window motif; she sits relaxed overlapping the letters, wind rippling her hair and tie; one last blush pulse sweeps through the typography and a gold-ring flash punctuates the end.\nEditing: extremely aggressive rhythm — hard cuts on every beat, graphic matches, whip pans, snap zooms, stroboscopic freezes, foreground wipes, light-leak splits, shutter flashes, impact shakes and speed ramps; every cut must feel compositionally different; typography is always fully readable before the character overlaps it. No weapons, no combat, no fire; all energy comes from motion design, wind, gold dust, glass, paper and her quiet, precise presence. BGM: hard-hitting electronic / drum-heavy future bass with aggressive drops, risers, sub hits and glitch fills locked to every cut; fabric whooshes, glass shatters, foil stamps and typography slams are rhythmic sound-design elements. Peak at CUT 12 and end with a cold electronic logo stinger. Premium AAA quality, photoreal-to-editorial cinematic rendering, stylish and explosive, strong graphic-design identity, consistent character design from the reference image, exactly 13 cuts.",
    "sourceUrl": "https://x.com/Just_sharon7/status/2094333791904923946",
    "sourceLabel": "X 原帖",
    "author": "Just_sharon7",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094333791904923946.mp4",
    "posterUrl": "/posters/x/2094333791904923946.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "复古胶片",
      "动态图形",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "retro",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2094333791904923946"
  },
  {
    "id": "x-2094319440745677219",
    "title": "暴龙与棘龙正面交锋",
    "titleEn": "T-Rex and Spinosaurus Face Off",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "两只大型恐龙在岩地近距离对峙，随后冲撞和撕咬。",
    "summaryEn": "Two large dinosaurs confront each other on rocky ground and clash at close range.",
    "prompt": null,
    "sourceUrl": "https://x.com/CanAiStudios/status/2094319440745677219",
    "sourceLabel": "X 原帖",
    "author": "CanAiStudios",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094319440745677219.mp4",
    "posterUrl": "/posters/x/2094319440745677219.jpg",
    "duration": 15.168,
    "aspectRatio": "9:5",
    "resolution": "480p",
    "tags": [
      "动作特效",
      "3D / CG",
      "写实电影感",
      "动物与生物",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "creatures",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094319381647966415",
    "title": "教堂讲台上的意外着装",
    "titleEn": "An Unexpected Outfit at the Pulpit",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "教堂场景揭示讲台人物的短裤，随后转向人物准备周日工作的画面。",
    "summaryEn": "A church scene reveals the speaker's shorts before cutting to preparations for Sunday work.",
    "prompt": null,
    "sourceUrl": "https://x.com/jonnyboy/status/2094319381647966415",
    "sourceLabel": "X 原帖",
    "author": "jonnyboy",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094319381647966415.mp4",
    "posterUrl": "/posters/x/2094319381647966415.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "写实电影感",
      "人物表演",
      "日常生活"
    ],
    "category": "advertising",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094318362470101207",
    "title": "落叶与火光聚成凤凰",
    "titleEn": "Leaves and Fire Gather into a Phoenix",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "漂浮树叶与光点聚成燃烧的鸟形，随后飞过森林上空。",
    "summaryEn": "Floating leaves and light particles form a fiery bird that flies above a forest.",
    "prompt": null,
    "sourceUrl": "https://x.com/MrDasOnX/status/2094318362470101207",
    "sourceLabel": "X 原帖",
    "author": "MrDasOnX",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094318362470101207.mp4",
    "posterUrl": "/posters/x/2094318362470101207.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "风格化与实验",
      "动物与生物",
      "奇幻"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "creatures",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094299737839305097",
    "title": "云海之上的狐面创造者",
    "titleEn": "A Fox-Masked Creator above the Clouds",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "狐面人物站在云海之间，以水流和光带构成奇幻空间。",
    "summaryEn": "A fox-masked figure stands among clouds, shaping a fantasy space with water and ribbons of light.",
    "prompt": null,
    "sourceUrl": "https://x.com/ewotanosimo2310/status/2094299737839305097",
    "sourceLabel": "X 原帖",
    "author": "ewotanosimo2310",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094299737839305097.mp4",
    "posterUrl": "/posters/x/2094299737839305097.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "奇幻",
      "人物表演"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094299301497479222",
    "title": "午夜雨城的时间停顿",
    "titleEn": "Time Stalls in a Rainy Midnight City",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "雨夜街道上的男子穿过停顿的人群，随后与另一名兜帽人物相遇。",
    "summaryEn": "A man crosses a rainy street among motionless people and encounters a hooded stranger.",
    "prompt": "Create a cinematic supernatural thriller, 15 seconds, 16:9. 0–4s: a crowded city street at night, rain falling heavily. 4–6s: the clock strikes midnight and everything suddenly freezes—people, cars, rain and birds. 6–10s: the protagonist walks through the frozen world in disbelief. 10–13s: they notice another person moving in the distance. 13–15s: the stranger turns around and has the protagonist’s exact face. Dark cinematic lighting, realistic rain physics, dramatic close-ups, eerie atmosphere, photorealistic.",
    "sourceUrl": "https://x.com/iamrealsnow/status/2094299301497479222",
    "sourceLabel": "X 原帖",
    "author": "iamrealsnow",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094299301497479222.mp4",
    "posterUrl": "/posters/x/2094299301497479222.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "城市街景",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/iamrealsnow/status/2094299301497479222"
  },
  {
    "id": "x-2094284844792520881",
    "title": "未来城市列车顶上的旅人",
    "titleEn": "A Traveler atop a Futuristic City Train",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物站在列车顶部，镜头沿高楼之间的轨道追随前行。",
    "summaryEn": "A figure stands atop a train as the camera follows its route between futuristic towers.",
    "prompt": "Cinematic photorealistic video of a stylish young woman standing on the roof of a modern moving train in a futuristic metropolitan city. She has long dark wavy hair and wears a fitted black leather jacket over a simple white top, relaxed dark blue jeans, and white sneakers.\n\nThe video begins with a smooth cinematic tracking shot from behind as the train moves through the city, with tall modern skyscrapers, glowing windows, and elevated railway tracks surrounding her. The camera then moves to a low-angle front shot as she walks naturally across the train roof, her hair and jacket gently flowing in the wind.\n\nAdd a close-up as she looks toward the camera with a calm, confident expression, followed by a wide aerial shot showing the entire moving train and massive city skyline. Natural body movement, realistic wind and fabric physics, detailed urban environment, realistic reflections, cinematic lighting, shallow depth of field, smooth camera transitions, subtle motion blur, photorealistic textures, premium cinematic look, 4K, 16:9.",
    "sourceUrl": "https://x.com/CaliraVal/status/2094284844792520881",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094284844792520881.mp4",
    "posterUrl": "/posters/x/2094284844792520881.jpg",
    "duration": 15.168,
    "aspectRatio": "106:45",
    "resolution": "360p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "城市街景",
      "科幻太空"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2094284844792520881"
  },
  {
    "id": "x-2094282576542470337",
    "title": "纸张世界里的黑白红拼贴 MV",
    "titleEn": "A Black-and-Red MV in a Paper World",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在纸张与复印拼贴空间中表演，最后与大幅红黑字体同框。",
    "summaryEn": "A character performs in a paper-and-photocopy world before appearing beside bold red-and-black type.",
    "prompt": "生成一个 动漫角色风格的15 秒复古潮流拼贴 MV，街头海报墙、复印 zine、VHS、halftone、黑白红配色、英文大字动态字体，K-pop hip-hop 节奏",
    "sourceUrl": "https://x.com/ponyodong/status/2094282576542470337",
    "sourceLabel": "X 原帖",
    "author": "ponyodong",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094282576542470337.mp4",
    "posterUrl": "/posters/x/2094282576542470337.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ponyodong/status/2094282576542470337"
  },
  {
    "id": "x-2094279910034706924",
    "title": "红黑角色的写真人像切换",
    "titleEn": "Portrait Cuts of a Red-and-Black Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "同一红黑造型人物在室内外近景中切换，突出脸部和服装特征。",
    "summaryEn": "A red-and-black character appears in indoor and outdoor portraits emphasizing the face and costume.",
    "prompt": null,
    "sourceUrl": "https://x.com/GeekCatX/status/2094279910034706924",
    "sourceLabel": "X 原帖",
    "author": "GeekCatX",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094279910034706924.mp4",
    "posterUrl": "/posters/x/2094279910034706924.jpg",
    "duration": 14.464,
    "aspectRatio": "24:43",
    "resolution": "480p",
    "tags": [
      "综合展示",
      "照片级写实",
      "时尚",
      "人物表演"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094270762001514657",
    "title": "纽约、伊斯坦布尔与迪拜自拍",
    "titleEn": "Selfies across New York, Istanbul and Dubai",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子在三座城市的街景中连续自拍，构成旅行地点切换。",
    "summaryEn": "A woman films herself against street scenes in three cities, creating a travel-location montage.",
    "prompt": "Create a 15-second ultra-realistic cinematic travel vlog filmed entirely as natural handheld smartphone footage. Keep the exact same young woman throughout all three locations with identical face, facial proportions, skin tone, long dark-brown hair and natural expressions. No beauty filters, no subtitles, no music. Use realistic objects passing directly in front of the camera to create seamless transitions.\n\n0:00–0:02.5 — NEW YORK CITY, USA 🇺🇸\nEarly morning in Manhattan. The woman walks along a busy downtown sidewalk while filming herself in selfie mode. Yellow taxis, tall buildings, steam rising from a street grate and realistic pedestrians fill the background. She holds a paper coffee cup and says: “New York mornings!”\n\n0:02.5–0:05 — TRANSITION\nA yellow taxi passes extremely close in front of the camera, completely blocking the frame. As the taxi fills the screen, the scene transitions naturally.\n\n0:05–0:07.5 — ISTANBUL, TURKEY 🇹🇷\nThe taxi clears to reveal a lively Istanbul street near the Bosphorus. The woman is wearing casual modest travel clothing and holds a warm Turkish simit. Ferries and historic architecture appear behind her. She takes a bite and says: “So good!”\n\n0:07.5–0:10 — TRANSITION\nA large red-and-white Turkish flag carried by a passing pedestrian moves directly across the camera lens, completely covering the frame.\n\n0:10–0:12.5 — DUBAI, UAE 🇦🇪\nThe flag clears to reveal a modern Dubai evening scene with illuminated skyscrapers, luxury architecture and busy pedestrian areas. The woman walks toward the camera holding a small cup of Arabic coffee and says with a smile: “Dubai at night hits different.”\n\n0:12.5–0:15 — FINAL TRANSITION / ENDING\nShe turns the camera toward the illuminated skyline, then brings her hand toward the lens until her palm completely covers the screen. Cut to black.\n\nVisual style: photorealistic live-action, authentic smartphone travel vlog, realistic skin texture, natural facial expressions, documentary realism, subtle handheld movement, cinematic but believable lighting, realistic crowds and environmental details.\n\nAudio: natural location sounds only — New York traffic and footsteps, Istanbul street ambience and ferry sounds, Dubai city ambience and distant conversations. No music, no subtitles, no artificial sound effects.",
    "sourceUrl": "https://x.com/AIwithJames/status/2094270762001514657",
    "sourceLabel": "X 原帖",
    "author": "AIwithJames",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094270762001514657.mp4",
    "posterUrl": "/posters/x/2094270762001514657.jpg",
    "duration": 16.068209,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AIwithJames/status/2094270762001514657"
  },
  {
    "id": "x-2094263172261933528",
    "title": "露台吉他与即兴合唱",
    "titleEn": "Guitar and Casual Singing on a Terrace",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "男子坐在露台弹吉他，身旁女子随之演唱，画面保留家用录像质感。",
    "summaryEn": "A man plays guitar on a terrace while a woman sings beside him in a home-video style.",
    "prompt": "ULTRA REALISTIC 15 SECOND PERSONAL HOME VIDEO. 1080p render of 4:3 early 2000s MiniDV tape footage. Single continuous handheld take, no cuts. Sung Hindi vocal, diegetic performance. Title: TERI AADAT.\n\nCONCEPT\nDusk on a terrace. A young man plays an old acoustic guitar sitting on the parapet, and the young woman sings along with him, badly and happily, halfway through a verse. Their friend is filming on a camcorder from a few feet away. It is a love song but nobody is performing a love song, they are just two people killing an evening and the tape happened to be running.\n\nSUBJECT LOCK ONE\nIndian woman, early twenties, warm medium brown skin. Realistic skin with visible pores, faint under eye shadow, minimal makeup, slight shine on the forehead in the low sun. Dark brown hair loosely tied back with strands pulling loose in the wind. Oversized cream knit sweater with stretched cuffs, blue jeans, white sneakers scuffed at the toe. Same face, same hair, same outfit held identically for the full 15 seconds. Sitting cross legged on a folded mat, knees tucked, unposed and slightly slouched.\n\nSUBJECT LOCK TWO\nIndian man, early twenties, medium brown skin, short unstyled hair, faint stubble. Faded olive shirt with sleeves pushed to the elbows, dark jeans, worn sandals. Sitting on the low parapet with a scratched acoustic guitar across his lap, head down watching his own fingers, glancing up at her twice.\n\nSETTING\nOpen concrete terrace of an older residential building, autumn evening. Low parapet with peeling paint, a black water tank, a sagging clothesline with nothing on it, dry leaves in one corner. Beyond the parapet, a dense low rise neighbourhood of flat roofs, cables and water tanks, hazy warm sun almost down. Lived in and unstyled. No visible brands, no logos, no readable signage anywhere in frame.\n\nCAMERA\nRaw handheld MiniDV camcorder held by a friend sitting a few feet away, angle slightly low and off center. Autofocus hunting between the guitar strings and her face. Exposure pumping as the lens drifts toward the bright sky and back down. One slow unmotivated zoom in on her, overshooting so her forehead is cropped, then correcting. Framing keeps both of them in shot but never composes them nicely. Mild tape noise, interlace combing on fast movement, one brief tape dropout glitch. No stabilization, no gimbal, no cinematic choreography, no rack focus.\n\nBEATS\n0:00 to 0:04 He is already mid strum. She listens with her chin on her knee, mouthing the words before she commits to singing them.\n0:04 to 0:10 She sings the first two lines properly, off pitch on the second, and laughs at herself without stopping. He keeps playing and grins down at the guitar.\n0:10 to 0:13 She sings the last two lines quieter, watching him instead of the camera. He looks up at her once. Neither reacts, no kiss, no held gaze, they both look away.\n0:13 to 0:15 The guitar rings out unresolved, she pulls her sleeve over her hand, wind hits the mic, tape cuts.\n\nVOCAL AND LYRICS\nOriginal composition, not an existing song. Spoken and sung language locked to Hindi. Single female voice, untrained, close and unmixed, sung at conversational volume.\n0:04 Dheere dheere shaam dhalti hai\n0:07 Kuch bhi nahi kehna aaj\n0:10 Teri aadat si ho gayi hai\n0:12 Rehne de, bas yahi theek hai\nPrecise lipsync to the sung line, bilabial closure on the m, b and p sounds, jaw and throat movement consistent with singing rather than speaking. Slight breath before each line. No harmony, no second voice, no backing vocal.\n\nAUDIO\nDiegetic only. The acoustic guitar visible in frame is the only instrument, recorded on the camcorder mic so it is thin and boxy. Her voice, wind across the terrace clipping the mic, guitar body knocks, crows, a distant pressure horn, a television faint from a floor below, faint hand noise on the camcorder body. No studio track, no backing music, no reverb, no score, no added ambience beds.\n\nSTYLE RULES\nNothing dramatic happens on purpose. No lip biting, no slow motion hair, no romantic staging, no golden hour beauty lighting, no lingering eye contact, no music video choreography. The romance is only in the fact that neither of them looks at the camera. Realism comes from the off pitch line, the imperfect framing, and the unresolved chord at the end. If either of them performs for the lens the illusion breaks.\n\nNEGATIVE\nNo music track. No subtitles. No captions. No watermark. No timestamp burn in. No slow motion. No color grading. No lens flare. No shallow depth of field. No crowd. No third visible person. No brand logos. No readable text.\n\nLAST FRAME\nWide slightly crooked frame of both of them on the terrace with the sun gone and the sky flat blue, image tears with one frame of tape distortion, then hard cut to black. Absolutely no on screen text, no title card, no lettering, no numbers, no logo, no signage of any kind in the final frame.",
    "sourceUrl": "https://x.com/bmx_ai13/status/2094263172261933528",
    "sourceLabel": "X 原帖",
    "author": "bmx_ai13",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094263172261933528.mp4",
    "posterUrl": "/posters/x/2094263172261933528.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "768p",
    "tags": [
      "音乐视频",
      "复古胶片",
      "生活实拍感",
      "音乐演出",
      "日常生活"
    ],
    "category": "music",
    "styles": [
      "retro",
      "ugc"
    ],
    "scenes": [
      "music-performance",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/bmx_ai13/status/2094263172261933528"
  },
  {
    "id": "x-2094260209602424868",
    "title": "首尔、河内与马拉喀什的旅行切换",
    "titleEn": "Travel Cuts across Seoul, Hanoi and Marrakech",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "同一旅行者从冬日街道走到繁忙市集和灯具商铺。",
    "summaryEn": "The same traveler moves from a wintry street to busy markets and a lamp shop.",
    "prompt": "Create a 15-second ultra-realistic cinematic travel vlog filmed entirely as natural handheld smartphone footage. Use the same young woman throughout all three locations, maintaining identical face, facial proportions, skin tone, long dark-brown hair, hairstyle, and natural expressions. No beauty filters, no subtitles, no music. Fast but smooth transitions created by objects covering the camera lens.\n\n0:00–0:02.5 — SEOUL, SOUTH KOREA 🇰🇷\nCold winter morning with light snow falling. The woman walks through a busy Seoul street, filming herself in selfie mode while holding a takeaway coffee. Seoul Tower is visible in the background. She smiles naturally and says: “Seoul mornings hit different!” Realistic traffic, pedestrians, footsteps, wind and soft city ambience.\n0:02.5–0:05 — TRANSITION\nShe turns the phone away from herself, revealing the Seoul street and N Seoul Tower. She raises her sleeve toward the camera until the fabric completely covers the lens, creating a natural seamless transition.\n\n0:05–0:07.5 — HANOI, VIETNAM 🇻🇳\nThe sleeve reveals a completely new location: lively Hanoi street filled with motorbikes, Vietnamese flags, street vendors and old buildings. The woman is now wearing a traditional Vietnamese conical hat and casual light clothing. She holds a fresh bánh mì close to the camera, takes a bite and says naturally: “Best sandwich ever!”\n0:07.5–0:10 — TRANSITION\nShe quickly tilts the camera upward toward colorful old buildings, tangled electrical wires and Vietnamese flags. A passing cyclo moves directly across the lens, completely blocking the frame and creating the transition.\n\n0:10–0:12.5 — MARRAKECH, MOROCCO 🇲🇦\nThe cyclo clears to reveal a warm, atmospheric Marrakech souk. Golden lanterns, colorful textiles, narrow market alleys and traditional Moroccan architecture surround her. She holds a glass of mint tea toward the camera, smiles and says: “This place is magical.” Rich warm lighting and authentic market ambience.\n\n0:12.5–0:15 — FINAL TRANSITION / ENDING\nShe turns the camera around and begins walking deeper through the crowded souk. She looks back toward the camera, reaches her hand forward, and places her palm directly over the lens until the screen becomes completely covered. Cut to black.\n\nVisual style: photorealistic live-action, authentic travel vlog, natural smartphone camera movement, realistic skin texture, natural lighting, documentary realism, subtle handheld shake, genuine facial expressions, believable environments, fast-paced travel montage.\nAudio: natural location sounds only — Seoul footsteps, traffic, light snow and wind; Hanoi motorbikes, vendors and street voices; Marrakech market chatter, lantern ambience and distant calls. No music, no subtitles, no artificial sound effects.",
    "sourceUrl": "https://x.com/codewithhajra/status/2094260209602424868",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094260209602424868.mp4",
    "posterUrl": "/posters/x/2094260209602424868.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2094260209602424868"
  },
  {
    "id": "x-2094108309011980339",
    "title": "粉色背景上的刺客剪影片头",
    "titleEn": "An Assassin Silhouette on Pink",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑色动漫人物剪影与眼部特写、银色标题在粉色背景上切换。",
    "summaryEn": "A black anime silhouette, eye close-ups and a silver title alternate against pink.",
    "prompt": "Use the attached character reference (Image1) as the silhouette base. Create a pastel neo-noir teen-assassin anime title sequence, cut like a music video with hard, punchy edits (no crossfades). Style the character as a flat black silhouette against solid baby-pink gradients (blush pink, bubblegum, rose quartz, cotton-candy white), rendered in bold pop-art compositions with retro shōjo-manga panel breaks, chrome-and-neon accent lines, extreme close-ups on eyes/lips/weapon, and wide skyline or rooftop establishing shots.\n\nAlternate fast, kinetic action bursts (sprint, blade draw, spin-kick, hair whip) with hard freeze-frame beats — each freeze holding just long enough to read as a \"poster moment.\" Use motion graphics throughout: speed-lines, halftone dot shading, comic-style impact bursts (\"POW\"-style but in soft pastel palettes), chromatic glitch flickers, and light streaks trailing off silhouette edges on transitions.\n\nBold typography: chunky, rounded bubble-serif or Y2K-manga-style lettering (like sticker-text or magazine cutouts) slamming into frame on each hard cut, sometimes overlapping the silhouette, sometimes punched into negative space. Title cards should feel like collectible trading-card / vinyl-poster inserts — thick outlines, drop shadows, slight jitter/shake on impact.\n\nFinish with the character frozen in full silhouette against a solid baby-pink background as the title logo stamps in with a soft chrome shimmer and a light glitter-dust settle effect.\n\nApply a light VHS/Y2K digicam texture throughout — subtle scan lines, soft chromatic aberration, occasional analog glitch/tracking-line flicker, and faint sparkle/bloom — for a nostalgic early-2000s music-video finish (instead of film grain/dust, since there's no rugged western texture here). BGM: hard-hitting electronic / drum-heavy future bass with aggressive drops, risers, sub hits and glitch fills locked to every cut. Sneaker impacts, whooshes, glass shatters, window-slam hits and typography slams are rhythmic sound-design elements. Peak at CUT 12 and end with a cold electronic logo stinger.  Premium AAA quality, anime-inspired cinematic rendering, stylish and explosive, strong graphic-design identity, consistent character design, exactly 13 cuts.",
    "sourceUrl": "https://x.com/AIwithAliya/status/2094108309011980339",
    "sourceLabel": "X 原帖",
    "author": "AIwithAliya",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094108309011980339.mp4",
    "posterUrl": "/posters/x/2094108309011980339.jpg",
    "duration": 15.13941,
    "aspectRatio": "26:15",
    "resolution": "480p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AIwithAliya/status/2094108309011980339"
  },
  {
    "id": "x-2094104586793455656",
    "title": "仲夏与柴犬一起午睡",
    "titleEn": "A Midsummer Nap with a Shiba",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子侧卧在柴犬身旁，日光照进安静的木屋空间。",
    "summaryEn": "A woman rests beside a Shiba as daylight enters a quiet wooden room.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tomw852/status/2094104586793455656",
    "sourceLabel": "X 原帖",
    "author": "Tomw852",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094104586793455656.mp4",
    "posterUrl": "/posters/x/2094104586793455656.jpg",
    "duration": 8.057324,
    "aspectRatio": "5:9",
    "resolution": "480p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动物与生物",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094085208027103678",
    "title": "薄荷绿头发与橙色耳机的角色片",
    "titleEn": "Mint Hair and Orange Headphones in Motion",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "动漫角色在橙绿圆环、鞋部近景和大字构图之间切换。",
    "summaryEn": "An anime character moves between orange-and-green rings, shoe close-ups and bold type.",
    "prompt": "Create an explosive, motion-graphics-driven character reveal trailer in 16:9, exactly 13 distinct cuts, 24fps, total 15.00s.\n\nCHARACTER (lock this design; never redesign):\nAnime streetwear girl from the reference still. Twin high buns of vivid mint-teal hair with long flowing tails and warm orange/gold streaks. Messy side-swept bangs. Large orange over-ear headphones with mint accents and a small logo plate. Sharp amber-orange eyes, one eye winking. Playful open-mouth grin. Oversized color-block windbreaker: navy body, vivid orange sleeves, white ribbed cuffs, silver zippers, circular teal tech buttons, a teal utility pocket on the sleeve. Orange cropped turtleneck under the open jacket. Light-wash ripped denim shorts, thick orange belt with a silver buckle. Navy thigh-high socks with orange ribbed cuffs and an orange X stitch on the left shin. Chunky white sneakers with orange details. Preserve exact face, proportions, hairstyle, outfit, materials, accessories and colors in every frame.\n\nThis film is 80% bold graphic design in motion and 20% character action. Graphic language is **retro OS chrome + music-player UI**, not ice/frost/barcode-lock. Use slamming window frames, title bars, close/minimize widgets, equalizer bars, waveforms, progress ticks, folder tiles, cursor arrows, CRT scanlines, pixel shatter, vinyl-ring stamps, music-note particles, media-player transport icons. Palette: mint teal, vivid orange, navy, cream-white, silver. Style: premium AAA motion-graphics title sequence × streetwear campaign film × Windows-era media player. Every graphic element moves fast and snaps hard on the beat.\n\nCUT 01 | 0.00–1.00s — Pure graphics: a mint title-bar slams onto a cream field, an orange CLOSE widget punches the corner, two navy window borders wipe in. Tiny equalizer ticks and a progress strip flicker. The letters **P** then **LAY** punch in one after another with heavy impact shake.\n\nCUT 02 | 1.00–2.00s — The **A** becomes a headphone cup: extreme close-up of her amber eye inside the orange earcup, glancing up; RGB split flash, then the cup shatters into flat mint/orange tiles.\n\nCUT 03 | 2.00–3.10s — Navy frame with enormous cream **PLAY**. She sprints in from frame left and power-slides across the baseline of the type, speed lines and orange streaks trailing, shards of the letters kicked up like sparks. Whip-pan out.\n\nCUT 04 | 3.10–4.00s — A giant retro media-player waveform explodes across the frame as a thick mint-and-orange audio spectrum bends into a tunnel. She bursts through the center of the waveform at full speed, briefly breaking into three stroboscopic motion trails behind her. Each trail leaves behind chunky navy equalizer blocks that rise and collapse to the beat. The camera rapidly pushes through the waveform tunnel with her, while huge vertical text **TRACK 01** scrolls continuously in the background. The waveform suddenly compresses into a single horizontal line and snaps shut behind her on the final beat.\n\nCUT 05 | 4.00–5.10s — She leaps through a giant rotating ring of typography reading **MAX VOLUME**. Camera tracks her mid-air spin in slow motion as the letters scatter, then snap-zoom on her wink.\n\nCUT 06 | 5.10–6.00s — Hard cut: cream editorial card, huge navy **DROP** with an orange slash. She vaults over the word itself, palm planted on the D, legs whipping across frame; the word compresses like a spring under her hand and rebounds.\n\nCUT 07 | 6.00–7.00s — Mint field with a navy diagonal window-bar. She back-flips along the bar in three stroboscopic ghost frames, each ghost tinted mint, orange, or navy. Giant outlined text **LOOP** rotates 180 degrees in sync with her rotation.\n\nCUT 08 | 7.00–8.00s — Kinetic type barrage: the words **LOUD / WILD / TEAL / HEAT** slam in one per beat with shutter flashes and camera shake, while she slides on her knees across the foreground, jacket flaring, music-note particles bursting from her sneakers.\n\nCUT 09 | 8.00–9.00s — Navy frame, giant cream wireframe window-grid tilting in 3D. She runs up the grid like a wall, kicks off, and hangs in the air as everything freezes. An orange circular stamp locks around her pose like a media-player target, with transport icons and EQ ticks.\n\nCUT 10 | 9.00–10.10s — Freeze releases into a burst: she dives toward camera through layered flat-color window panes that shatter one by one like glass shutters, each pane revealing a bigger letter of **P-L-A-Y**. Foreground wipe with her sneaker.\n\nCUT 11 | 10.10–11.10s — Rapid-fire poster montage: four full-screen graphic posters of her in different action poses (mid-flip, sliding, landing, headphones-up wink) snap past with hard cuts. Each has a different bold layout, oversized numbers **01–04**, equalizer strips and slashes.\n\nCUT 12 | 11.10–13.00s — Hero moment on a clean cream cyclorama: she lands a final backflip dead center in slow motion, straightens with one hand on her headphones, and a shockwave of concentric mint rings, wind streaks and shattered typography blasts outward from her landing. Brief iconic freeze on her wink, overexpose to white.\n\nCUT 13 | 13.00–15.00s — Final identity card: enormous navy **PLAY** dominates a pale cream field with translucent mint rings, technical arcs, scanlines, and a rough orange circular emblem containing a ghosted headphone / waveform motif. She stands relaxed overlapping the letters, wind rippling her jacket. One last orange pulse sweeps through the typography and a window-chrome flash punctuates the end.\n\nEditing: extremely aggressive rhythm — hard cuts on every beat, graphic matches, whip pans, snap zooms, stroboscopic freezes, foreground wipes, RGB splits, shutter flashes, impact shakes and speed ramps. Every cut must feel compositionally different. Typography is always fully readable before the character overlaps it. No weapons, no combat, no fire. All energy comes from motion design, wind, glass, UI chrome and her parkour-style athleticism.\n\nBGM: hard-hitting electronic / drum-heavy future bass with aggressive drops, risers, sub hits and glitch fills locked to every cut. Sneaker impacts, whooshes, glass shatters, window-slam hits and typography slams are rhythmic sound-design elements. Peak at CUT 12 and end with a cold electronic logo stinger.\n\nPremium AAA quality, anime-inspired cinematic rendering, stylish and explosive, strong graphic-design identity, consistent character design, exactly 13 cuts.",
    "sourceUrl": "https://x.com/Just_sharon7/status/2094085208027103678",
    "sourceLabel": "X 原帖",
    "author": "Just_sharon7",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094085208027103678.mp4",
    "posterUrl": "/posters/x/2094085208027103678.jpg",
    "duration": 15.168,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "时尚",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Just_sharon7/status/2094085208027103678"
  },
  {
    "id": "x-2094068060059799762",
    "title": "蓝色街头人物的平面角色揭晓",
    "titleEn": "A Flat Graphic Reveal of a Blue Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝色人物与帽子、运动鞋和几何色块组合成角色宣传片。",
    "summaryEn": "A blue character, knit cap and sneakers combine with geometric blocks in a character promo.",
    "prompt": "CHARACTER LOCK (do not redesign): Use the exact figure from the reference still. Cool, composed cartoon streetwear character with vivid blue skin, short black hair with messy bangs, orange knit beanie, oversized cream hoodie with a pink inner hood and large pink polka dots on the sleeves, small orange square chest mark, black crossbody sling bag with a white logo, holding a paper coffee cup marked CAF, teal-and-cream checkerboard baggy pants with a small black BIG tag on the lower left leg, light-blue socks, chunky orange-black-white sneakers with pink laces. Preserve exact face, proportions, hairstyle, outfit, materials, accessories, and colors in every frame. Never change gender, silhouette, palette, or branding.\nThis film is 80% bold graphic design in motion and 20% character action: giant kinetic typography, hard-edged shapes, split-screens, flat color fields, checker tiles, coffee-ring stamps, city-grid overlays, UI ticks, halftone, speed lines, and shutter flashes. Palette: cream, hot pink, teal, orange, cobalt-blue skin, black, white. Style: premium AAA motion-graphics title sequence × streetwear campaign film × NYC block energy. Every graphic element moves fast and snaps hard on the beat.\nCUT 01 | 0.00–1.00s — Pure graphics: a massive teal checker square slams onto a cream field, two hot-pink bars wipe across, tiny UI ticks and a coffee-ring stamp flicker; the letters B and I and G punch in one after another with heavy impact shake.\nCUT 02 | 1.00–2.00s — The G becomes a circular stamp: extreme close-up of the character’s eye inside it, glancing sideways; RGB split flash, then the stamp shatters into flat pink and teal shards.\nCUT 03 | 2.00–3.10s — Black frame with an enormous cream BIG; the character walks in from frame left holding the CAF cup, then power-slides across the baseline of the typography, speed lines and pink-dot trails kicking letter shards like sparks. Whip-pan out.\nCUT 04 | 3.10–4.00s — Split-screen triptych of flat cream / teal / hot-pink panels; in each panel the silhouette performs a different freeze: stride, cup-raise, look-back. Huge vertical text BLOCK MODE scrolls behind the panels.\nCUT 05 | 4.00–5.10s — The character steps through a giant rotating ring of typography reading CITY WALK NO CLOCK; camera tracks a mid-stride turn in slow motion as the letters scatter, then snap-zoom on the composed face.\nCUT 06 | 5.10–6.00s — Hard cut: cream editorial card, huge black SIP with an orange slash; the character vaults the word, cup planted on the S, legs whipping across frame; the word compresses like a spring under the cup and rebounds.\nCUT 07 | 6.00–7.00s — Hot-pink field with a teal diagonal checker stripe; the character spin-turns along the stripe in three stroboscopic ghost frames, each ghost tinted cream / teal / orange; giant outlined text SPIN rotates 180 degrees in sync.\nCUT 08 | 7.00–8.00s — Kinetic type barrage: the words FRESH / LOUD / SOFT / BIG slam in one per beat with shutter flashes and camera shake, while the character slides on sneakers across the foreground, hoodie flaring, pink-dot particles bursting from the laces.\nCUT 09 | 8.00–9.00s — Black frame, giant white wireframe city grid tilting in 3D; the character walks up the grid like a crosswalk, kicks off, and hangs mid-step as everything freezes; an orange circle stamps around the frozen pose like a target lock with UI brackets and data ticks.\nCUT 10 | 9.00–10.10s — Freeze releases into a burst: the character leans toward camera through layered flat-color panes that shatter one by one like glass shutters, each pane revealing a bigger letter of B-I-G; foreground wipe with the orange sneaker and CAF cup.\nCUT 11 | 10.10–11.10s — Rapid-fire poster montage: four full-screen graphic posters of the character in different street poses (mid-stride, cup up, look-back, hands-in-pocket) snap past with hard cuts, each with a different bold layout, oversized numbers 01–04, barcodes, slashes, and checker bands.\nCUT 12 | 11.10–13.00s — Hero moment on a clean cream cyclorama: the character lands a final easy stride dead center in slow motion, straightens with the cup at the hip, and a shockwave of concentric teal-and-pink rings, wind streaks, and shattered typography blasts outward; brief iconic freeze with a faint confident smile, overexpose to white.\nCUT 13 | 13.00–15.00s — Final identity card: enormous black BIG dominates a pale cream field with translucent pink rings, technical arcs, halftone, and a rough black circular emblem containing a ghosted coffee-ring / barcode motif; the character stands relaxed overlapping the letters, wind rippling the hoodie; one last teal pulse sweeps through the typography and a checker-flash punctuates the end.\nEditing: extremely aggressive rhythm — hard cuts on every beat, graphic matches, whip pans, snap zooms, stroboscopic freezes, foreground wipes, RGB splits, shutter flashes, impact shakes, and speed ramps; every cut must feel compositionally different; typography is always fully readable before the character overlaps it. No weapons, no combat, no fire; all energy comes from motion design, wind, frost-free city air, glass, coffee steam, and easy athletic street walking. BGM: hard-hitting electronic / drum-heavy future bass with aggressive drops, risers, sub hits, and glitch fills locked to every cut; sneaker impacts, whooshes, glass shatters, cup clinks, and typography slams are rhythmic sound-design elements. Peak at CUT 12 and end with a cold electronic logo stinger. Premium AAA quality, anime-inspired cinematic rendering composited over / matching the NYC street still energy, stylish and explosive, strong graphic-design identity, consistent character design, exactly 13 cuts.",
    "sourceUrl": "https://x.com/AIwithAliya/status/2094068060059799762",
    "sourceLabel": "X 原帖",
    "author": "AIwithAliya",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094068060059799762.mp4",
    "posterUrl": "/posters/x/2094068060059799762.jpg",
    "duration": 15.168,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "动态图形",
      "时尚",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AIwithAliya/status/2094068060059799762"
  },
  {
    "id": "x-2094059927023104206",
    "title": "红色亮面裙的楼梯时装镜头",
    "titleEn": "A Stairway Fashion Shot in Glossy Red",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物穿红色亮面裙沿楼梯行走；原帖附有参考人像提示词。",
    "summaryEn": "A figure in a glossy red dress walks along a stairway; the post includes a reference-portrait prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/KeorUnreal/status/2094059927023104206",
    "sourceLabel": "X 原帖",
    "author": "KeorUnreal",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094059927023104206.mp4",
    "posterUrl": "/posters/x/2094059927023104206.jpg",
    "duration": 10.208333,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2094002244853059848",
    "title": "黑色巨石坠落后的 APEX 字效",
    "titleEn": "APEX Revealed by a Falling Monolith",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "巨石从雷云坠落并击碎地面，碎片散开后显出标题。",
    "summaryEn": "A monolith falls from storm clouds and shatters the ground before fragments reveal a title.",
    "prompt": "Epic superhero-scale title reveal: a gigantic black monolith falls from the storm-filled sky at supersonic speed, surrounded by plasma, lightning and atmospheric shockwaves. It slams into the ground with devastating force, sending concentric waves through the landscape while millions of rocks and metallic fragments explode upward. The camera charges directly toward the impact zone, diving beneath airborne debris before violently rotating around the crater. Suddenly every fragment freezes in midair, then accelerates toward the center like ammunition pulled by an immense magnetic force. Pieces collide, weld and lock together at tremendous speed, constructing the enormous word \"APEX\" from obsidian-black metal with razor-sharp edges and glowing white-hot seams. The camera pushes extremely close across the letters as electricity races through their surfaces, then snaps backward for the final hero reveal just as a gigantic lightning strike hits the title and sends an explosive circular shockwave across the entire environment. Monumental, brutal, ultra-premium theatrical VFX.",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2094002244853059848",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2094002244853059848.mp4",
    "posterUrl": "/posters/x/2094002244853059848.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动态图形",
      "奇幻"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2094002244853059848"
  },
  {
    "id": "x-2093956013338849386",
    "title": "洗衣店变成发光涂鸦夜空",
    "titleEn": "A Laundromat Becomes a Glowing Doodle Sky",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "指尖的发光涂鸦进入洗衣机，逐渐覆盖整间洗衣店。",
    "summaryEn": "A glowing fingertip doodle enters a washing machine and spreads through the laundromat.",
    "prompt": "15秒16:9横屏短片，深夜自助洗衣店，实拍叠加蜡笔质感抖动线条的手绘发光动画，手机手持拍摄。指尖一缕紫色丝线→紫色蜡笔水母→逃窜到洗衣栏里，变成手绘袜子→手绘袜子一蹦一跳撞上真实袜子，变成紫蓝色流星→滑进洗衣机，变成浅蓝色小鱼→变成浅蓝色小火车，触碰售货机按键→变成夹子机器人，紫色与淡蓝色手绘纹路铺满开。整间洗衣店变成夜空涂鸦，洗衣机圆形视窗化作闪烁的星座窗口。最后一只手绘袜子坠落到洗衣篮内。",
    "sourceUrl": "https://x.com/ponyodong/status/2093956013338849386",
    "sourceLabel": "X 原帖",
    "author": "ponyodong",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093956013338849386.mp4",
    "posterUrl": "/posters/x/2093956013338849386.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "照片级写实",
      "插画与手绘",
      "奇幻",
      "日常生活"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal",
      "illustration"
    ],
    "scenes": [
      "fantasy",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ponyodong/status/2093956013338849386"
  },
  {
    "id": "x-2093951338367599004",
    "title": "蓝发角色的复古动漫预告",
    "titleEn": "A Retro Anime Teaser for a Blue-Haired Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝发人物以挥拳和近景登场，最后切至预告标题版式。",
    "summaryEn": "A blue-haired character appears in action poses and close-ups before a teaser title card.",
    "prompt": null,
    "sourceUrl": "https://x.com/onofumi_AI/status/2093951338367599004",
    "sourceLabel": "X 原帖",
    "author": "onofumi_AI",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093951338367599004.mp4",
    "posterUrl": "/posters/x/2093951338367599004.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093926202172223994",
    "title": "漫画分格与三人角色拼贴",
    "titleEn": "Manga Panels and a Three-Character Collage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物近景被重排为漫画分格，画面在黑白与红色拼贴之间展开。",
    "summaryEn": "Character close-ups rearrange into manga panels across black-and-white and red collages.",
    "prompt": "高品質な15秒の二次元アニメMV / PV風コラージュ映像。16:9。  @Image11、@Image22、@Image33 の3人を主要キャラクターとして使用する。  参照画像そのものの構図、ポーズ、背景、カメラ位置を再現しない。 参照画像を一枚絵として画面に貼り付けない。 @Image11、@Image22、@Image33 を素材として再構成し、漫画コマ、顔アップ、目元アップ、手元、全身、バストアップ、横顔、斜めアングル、デフォルメキャラクターなど、多様な新規構図として展開する。 同じ参照画像の静止した立ち絵を繰り返し表示するのではなく、画面ごとに構図、トリミング、ポーズ、アングル、サイズを大胆に変化させる。  全体は、アニメPV、漫画PV、グラフィックMV、タイトルシークエンスを融合した高速コミックコラージュ。  赤・白・黒を主軸とした強いコントラスト。 必要に応じて @Image11、@Image22、@Image33 由来の色をアクセントとして残す。  演出の中心はキャラクターの芝居ではなく、  漫画コマの分割 レイヤーの増殖 スライド クロップ 回転 フラッシュ 急激なズーム 平面から3D空間への変形 巨大タイポグラフィ 画面密度の段階的上昇  によるグラフィックモーション。  映像は最初から最後まで絶えず画面構成が変化する。 同じレイアウトを長時間維持しない。  ────────────────  【0.0〜2.5秒：MONOCHROME COMIC】  真っ白な画面に黒い線が高速で走り、複数の漫画コマが次々に形成される。  @Image11、@Image22、@Image33 が、それぞれ異なる漫画コマとして出現。  一枚絵をそのまま表示するのではなく、  顔のクローズアップ 瞳だけの極端なアップ 横顔 手元 衣装や小物の部分アップ 斜め構図 小さな全身シルエット 勢いのあるバストアップ  などへ再構成する。  画面全体は白黒漫画。  0.3〜0.5秒単位でコマがスライド、拡大、縮小、入れ替わる。  小さなカメラシェイク。 短い白フラッシュ。 黒ベタ。 印刷ずれ。 紙の質感。 ハーフトーン。 漫画的なスピード線。  一部のコマだけ瞬間的にカラー化する。  巨大な英字タイポグラフィが漫画コマの背後を横切る。  “I WON’T LOSE”  文字が画面いっぱいに現れた直後、別の漫画コマによって部分的に隠される。  ────────────────  【2.5〜4.8秒：COLOR INVASION】  白黒漫画の中へカラーが侵入する。  @Image11 の新しいバストアップ構図が漫画コマを突き破るように出現。  FLASH。  別方向から @Image22。  さらにコマ全体が斜めに回転しながら @Image33 が出現。  3人を単純に横並びにはしない。  前景、中景、背景へ分離し、 巨大な顔アップと小さな全身、 部分的なクロップ、 漫画コマ、 文字を重ねて奥行きを作る。  “I’LL BEAT YOU”  太いゴシック文字が画面中央を横断。  すぐに漫画コマがその文字を切断する。  次に、  “EVEN HYPOCRISY WILL DO”  文字を複数行へ分割し、 異なるサイズで高速に配置。  EVEN  HYPOCRISY  WILL DO  というように、文章そのものをグラフィック要素として扱う。  キャラクター画像の上に文字を載せるだけではなく、 文字の背後からキャラクターが覗く、 文字がコマを分割する、 巨大文字の隙間に顔が現れる、 という構成にする。  ────────────────  【4.8〜7.2秒：2D → 3D COMIC ROOM】  画面中央の一本の黒い縦線が急速に広がる。  漫画ページそのものが折れ曲がり始める。  平面的だった漫画コマが、  左壁 右壁 奥壁 床 天井  へ展開。  巨大な漫画ページが奥行きのある3Dの箱型空間へ変形する。  これは普通の部屋ではない。  漫画ページ、印刷物、吹き出し、キャラクターコマそのものから構成されたグラフィック空間。  @Image11、@Image22、@Image33 の異なる新規構図が壁面や床面へ配置される。  一枚絵を壁に貼るのではなく、 複数の異なる漫画コマとして分解して配置する。  手前には @Image11、@Image22、@Image33 を元にした小さなデフォルメ全身キャラクター。  デフォルメキャラクターは静止立ちではなく、 それぞれ異なる軽いポーズ。  カメラが箱の内部へ一瞬飛び込んだ後、 急激に後退する。  ────────────────  【7.2〜8.5秒：COLLAGE EXPLOSION】  猛烈なズームアウト。  先ほどの3D箱が画面中央の小さな一要素だったことが判明。  その周囲に大量の漫画コマが生成され始める。  @Image11 @Image22 @Image33  の、  顔 目 横顔 手 全身 デフォルメ シルエット 極端なクロップ  が異なるサイズで次々に出現。  赤い矩形。 黒い筆文字。 白い吹き出し。 漫画のスピード線。 巨大な英字。 斜めのクレジット文字。 ハーフトーン。 紙片。 インク。 グリッド。  画面密度が急激に上昇する。  “I’LL WAIT IN HELL”  文字が巨大に出現。  HELL のみ画面いっぱいまで拡大。  その巨大文字の内部にも漫画コマが存在する。  続いて、  “I’M WAITING FOR YOU”  が細長い帯状文字として画面を斜めに横断。  ────────────────  【8.5〜10.0秒：COUNTDOWN】  ビートが一段強くなる。  背景のコラージュが瞬間的に停止。  巨大な赤い文字。  “UN”  FLASH。  別の漫画コラージュへ瞬間切り替え。  “DEUX”  さらに強いズーム。  FLASH。  “TROIS”  TROIS は画面いっぱいまで巨大化。  文字の前後に @Image11、@Image22、@Image33 の顔や漫画コマが高速で差し込まれる。  巨大文字の内部にもキャラクター断片をマスク表示する。  カウントダウン中も同じ画像を使い回さず、 毎回異なる構図とクロップに変える。  TROIS の直後、 完全な白フラッシュ。  ────────────────  【10.0〜11.5秒：TITLE CALL】  完全な白背景。  それまでの情報量を突然ゼロにする。  中央に黒い荒々しい手書き文字で、  FALSE SAINT  を大きく表示。  タイトル以外はほぼ何も置かない。  一瞬遅れて、 赤いインクの小さな飛沫。 黒い筆跡。 細い漫画線。  タイトルが一瞬だけ軽く振動。  強いタイトルコール。  ────────────────  【11.5〜15.0秒：FINAL COLLAGE】  タイトルを黒い漫画コマが高速で覆い隠す。  再び高密度コラージュ。  今度は序盤より圧倒的に情報量が多い。  @Image11、@Image22、@Image33 の異なる構図が大量に高速出現。  顔アップ。 横顔。 目。 手。 全身。 バストアップ。 デフォルメ。 極端な斜めクロップ。 シルエット。  鎖のグラフィック。 黒白漫画コマ。 赤い巨大文字。 吹き出し。 細かなクレジット風文字。 英字タイポグラフィ。 ハーフトーン。 紙の断片。 スピード線。  コマそのものが前後へ移動し、 巨大な二次元コラージュに奥行きが生まれる。  カメラが高速でズームアウトし続ける。  漫画コマがさらに漫画コマの中へ入り、 画面の外側にも無数のページが存在するようなスケール感。  最後の0.5秒。  大量のレイヤーが一瞬だけ中央へ収束。  強烈な白フラッシュ。  BLACK。  ────────────────  【最重要】  @Image11、@Image22、@Image33 の参照画像そのものを画面に貼り付けたような表現にしない。  参照画像の元ポーズ、元構図、元背景を固定しない。  各キャラクターを、 新規ポーズ、 新規アングル、 新規クロップ、 新規漫画構図、 新規デフォルメ表現 へ積極的に再構成する。  キャラクターの完全固定よりも、 MVとしての画面変化、構図変化、グラフィックモーションを優先する。  同じキャラクター画像を長時間維持しない。 静止立ち絵を中央に置き続けない。 3人を常に横並びにしない。  「キャラクター画像＋背景」という構造ではなく、 キャラクター、文字、漫画コマ、図形、余白そのものが一体化したモーショングラフィックスとして構成する。  映像全体の変化骨格を明確にする。  MONOCHROME COMIC → COLOR INVASION → 2D TO 3D → COLLAGE EXPLOSION → UN / DEUX / TROIS → FALSE SAINT → FINAL COLLAGE  情報密度は時間とともに上昇する。  キャラクターアニメーションより、 編集、画面設計、コラージュ、タイポグラフィ、カメラ、レイヤー変形を主役にする。",
    "sourceUrl": "https://x.com/haruuraeadss/status/2093926202172223994",
    "sourceLabel": "X 原帖",
    "author": "haruuraeadss",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093926202172223994.mp4",
    "posterUrl": "/posters/x/2093926202172223994.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2093926202172223994"
  },
  {
    "id": "x-2093923736483504439",
    "title": "蓝眼机器人的科技演示片",
    "titleEn": "A Technology Demo with a Blue-Eyed Robot",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "金属机器人与图表、剪辑时间线和发光界面交替出现。",
    "summaryEn": "A metallic robot alternates with charts, an editing timeline and glowing interfaces.",
    "prompt": "Create a 15-second cinematic futuristic AI technology video. Show a sleek humanoid AI robot with a metallic black and silver body and glowing electric-blue eyes standing in a dark high-tech digital environment. Surround the robot with floating holographic dashboards, rising blue bar charts, data visualizations, research panels, and futuristic interface elements. Include a large digital screen displaying “AI Robots Are Building Tomorrow”. Use smooth cinematic camera movements, subtle reflections on a glossy black floor, volumetric lighting, realistic 3D rendering, high contrast, premium sci-fi aesthetic, detailed robotics, seamless transitions, and an epic futuristic atmosphere. Vertical 9:16, photorealistic, ultra-detailed, cinematic quality.",
    "sourceUrl": "https://x.com/CaliraVal/status/2093923736483504439",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093923736483504439.mp4",
    "posterUrl": "/posters/x/2093923736483504439.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "科幻太空",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "sci-fi",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2093923736483504439"
  },
  {
    "id": "x-2093905392787165663",
    "title": "漫画页面逐格变成动态图解",
    "titleEn": "A Manga Page Becomes an Animated Explanation",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "原始漫画页面与局部放大动画并列，依次突出人物、文字和操作步骤。",
    "summaryEn": "A manga page sits beside enlarged animated panels emphasizing characters, text and steps.",
    "prompt": "フレームごとに展開されるタイポグラフィーとイラストの漫画解説アニメ。",
    "sourceUrl": "https://x.com/seiiiiiiiiiiru/status/2093905392787165663",
    "sourceLabel": "X 原帖",
    "author": "seiiiiiiiiiiru",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093905392787165663.mp4",
    "posterUrl": "/posters/x/2093905392787165663.jpg",
    "duration": 15.146667,
    "aspectRatio": "499:360",
    "resolution": "720p",
    "tags": [
      "本地生成与工作流",
      "插画与手绘",
      "动态图形",
      "产品与广告"
    ],
    "category": "workflow",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/seiiiiiiiiiiru/status/2093905392787165663"
  },
  {
    "id": "x-2093896492000166116",
    "title": "东京夜间赶车的旅行自拍",
    "titleEn": "A Nighttime Tokyo Vlog on the Way to a Train",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子从霓虹街道走进车站，自拍镜头最后切向列车。",
    "summaryEn": "A woman moves from neon streets into a station, ending her selfie sequence with a train.",
    "prompt": "FORMAT: 15 seconds | 9:16 vertical | Ultra-photorealistic smartphone vlog | Real-world Tokyo at night\n\nCHARACTER LOCK\nUse the attached girl as the exact character and identity reference throughout the entire video. Preserve her recognizable facial features, face shape, warm skin tone, brown eyes, natural makeup, soft features, and dark brown hair styled in a messy high bun with loose strands framing her face.\nShe is an adult woman with a natural, approachable travel-vlogger appearance.\n\nOUTFIT LOCK\n\nMatch the outfit shown in the storyboard:\n\nCream-colored knit top\n\nLight beige lightweight jacket, worn open\n\nCharcoal/dark gray straight-leg trousers\n\nClean white sneakers\nSmall silver hoop earrings\nSimple silver wristwatch\nMinimal accessories\nNatural, understated makeup\nThe hairstyle, face, outfit, accessories and overall appearance must remain identical across all six storyboard scenes.\n00–04s — RUNNING FOR IT\nHandheld selfie footage on a crowded Tokyo street at night.\n\nThe girl from the attached reference walks quickly toward the camera, slightly out of breath. Her beige jacket moves naturally as she walks.\n\nBright Tokyo signs, storefronts, pedestrians and traffic create an authentic nighttime city background.\n\nShe looks directly into the phone camera and laughs:\n\n“Wait—my train leaves in two minutes!”\n\nShe quickly turns the phone forward and starts jogging toward the station.\n\n04–08s — THE STATION\n\nHandheld camera follows her from behind and slightly beside her as she rushes down the station stairs.\n\nHer messy brown bun and beige jacket remain clearly recognizable.\n\nShe reaches the ticket gate, taps her transit card and quickly passes through.\n\nThe phone camera shakes naturally from her movement and briefly loses autofocus.\n\n08–12s — MADE IT\n\nShe reaches the platform just as the train arrives.\n\nThe camera captures the approaching train headlights before returning to her face.\n\nShe steps inside, turns the phone toward herself and smiles while catching her breath.\n\nShe says:\n\n“I actually made it!”\n\nThe train doors close behind her.\n\n12–15s — TOKYO THROUGH THE WINDOW\n\nInside the train, she turns the phone toward the window.\n\nTokyo's illuminated buildings and streets naturally blur past as the train accelerates.\n\nHer face appears softly reflected in the glass.\n\nShe looks toward the city and quietly says:\n\n“Okay… Tokyo at night hits different.”\n\nNatural cut.\n\nREALISM & CHARACTER CONSISTENCY\n\nSame attached girl in every shot. No character changes.\n\nMaintain the exact:\n\nFacial identity\n\nBrown eyes\n\nDark brown messy high bun\n\nLoose face-framing hair strands\nSkin tone\nFacial proportions\nCream knit top\nBeige jacket\nCharcoal trousers\nWhite sneakers\nSilver earrings\nWristwatch\nNatural skin texture, realistic hair strands, authentic blinking, subtle facial expressions and believable body movement.\nNo outfit changes, no hairstyle changes, no different woman, no face morphing, no beauty-filter effect.\n\nVISUAL STYLE\nAuthentic handheld smartphone footage rather than a polished movie.\n\nNatural motion blur, autofocus hunting, slight exposure shifts, realistic low-light phone noise, imperfect framing, natural Tokyo crowds and physically accurate movement.\n\nNo cinematic transitions, no fantasy effects, no exaggerated neon, no artificial-looking skin, no duplicated people, no impossible camera movement.\n\nAUDIO\nNatural Tokyo ambience, footsteps, station announcements, crowd noise, train brakes, door chime and her real voice.\n\nNo background music, no subtitles, no watermark.",
    "sourceUrl": "https://x.com/codewithhajra/status/2093896492000166116",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093896492000166116.mp4",
    "posterUrl": "/posters/x/2093896492000166116.jpg",
    "duration": 16.555828,
    "aspectRatio": "45:56",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2093896492000166116"
  },
  {
    "id": "x-2093755381050929623",
    "title": "金色宫殿里的孔雀礼服",
    "titleEn": "A Peacock Gown in a Golden Palace",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子穿蓝绿礼服走过宫殿大厅，身后的羽饰逐渐展开。",
    "summaryEn": "A woman walks through a palace in a blue-green gown as feather-like ornaments spread behind her.",
    "prompt": "A full-length cinematic video of a regal woman gracefully walking forward through an opulent, golden palace ballroom with crystal chandeliers and a polished black-and-white checkered marble floor. She wears an exquisite, high-fashion peacock-themed gown crafted from intricate golden filigree, iridescent emerald green and royal blue peacock feathers, and sparkling jewel accents, complemented by blue feather shoulder pauldrons and a matching ornate tiara. Glowing cyan sparkles and magical light emanate from the patterns of her dress as she walks toward the camera with a confident stride. As she smoothly turns to her side, a spectacular, fan-like display of grand peacock feathers opens dramatically behind her, capturing warm ambient lighting in a ultra-detailed, photorealistic fantasy fashion aesthetic.",
    "sourceUrl": "https://x.com/TaliaAariz/status/2093755381050929623",
    "sourceLabel": "X 原帖",
    "author": "TaliaAariz",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093755381050929623.mp4",
    "posterUrl": "/posters/x/2093755381050929623.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "时尚",
      "奇幻"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "fashion",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TaliaAariz/status/2093755381050929623"
  },
  {
    "id": "x-2093747276942606456",
    "title": "创业者穿行于不断变换的世界",
    "titleEn": "A Founder Walks through Changing Worlds",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "西装男子持续走向镜头，背景从办公室切换为机器人街道和积木世界。",
    "summaryEn": "A suited speaker walks toward the camera as an office becomes a robot-filled street and a block-built world.",
    "prompt": "Hyper-real startup launch video, one continuous forward-tracking shot, a young founder walks confidently toward the camera through a sleek San Francisco office while employees work behind him, speaking with polished launch-video energy: \"For the last two years, we've been asking one simple question: what if everything could be better? Today, we're finally ready to show you what we've been building.\" The office gradually becomes an underground parking garage without a cut, absolutely no music or soundtrack.\nHe keeps walking at exactly the same pace as the parking garage seamlessly transforms into a crowded supermarket and then slowly into a moving subway train, shoppers push carts around him, a forklift knocks over a display, then subway passengers sway beside him as tunnels flash past the windows, and he says: \"We started from first principles, threw away every assumption, and rebuilt the experience from scratch. The result isn't just faster or smarter. It fundamentally changes what's possible.\". no music.\nHe walks through the end of the subway carriage and emerges seamlessly onto an enormous medieval battlefield, armored soldiers charge around him, horses thunder past and flaming arrows cross the sky, then the battlefield gradually transforms into a gigantic futuristic neon metropolis with flying vehicles and giant robots fighting between skyscrapers while he never reacts and continues: \"And the crazy thing is, we’re only getting started. Because once you remove the old constraints, you realize the entire category was designed wrong.\" zero background music.\nThe neon city begins flooding while he keeps walking, water rises around him until the entire environment becomes a vast underwater civilization filled with glowing towers, fish and whales, then the water abruptly drains away to reveal that he is now walking across the surface of Mars as astronauts flee from a damaged colony behind him, and he says: \"So instead of asking how we could improve the existing product, we asked why it had to exist at all. That led us to something completely different, something we think will become obvious in hindsight.\" no music.\nMars slowly grows vegetation around his feet until it becomes a dense prehistoric jungle, enormous dinosaurs crash through the trees and a T-rex charges behind him, then the jungle begins collapsing into a modern city undergoing an impossible combination of earthquakes, tornadoes, meteor strikes and collapsing skyscrapers while crowds sprint in every direction, yet he calmly says: \"The best technology disappears. You don't think about it. It just becomes part of your life. That’s why we believe this isn't a feature, or even a product. It’s a new primitive.\" there is absolutely no music score.\nReality itself starts malfunctioning around him while he continues walking, people duplicate, buildings fragment into floating geometry, gravity reverses, objects become half-generated and the entire world dissolves into shimmering visual noise as he finishes: \"And once you have the primitive, you can build things nobody has even imagined yet.\" Everything suddenly snaps back to an utterly ordinary silent office conference room, he finally stops beside a small beige plastic doorstop displayed dramatically on a pedestal, gestures toward it and says with complete sincerity: \"Introducing Doorstop. It keeps your door open. Available today for ninety-nine dollars.\" All apocalyptic sound instantly cuts to mundane room tone, fluorescent hum, air conditioning and distant keyboard typing; after two seconds of silence, an employee in the background slowly closes the door.",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2093747276942606456",
    "sourceLabel": "X 原帖",
    "author": "cocktailpeanut",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093747276942606456.mp4",
    "posterUrl": "/posters/x/2093747276942606456.jpg",
    "duration": 55.648,
    "aspectRatio": "20:11",
    "resolution": "704p",
    "tags": [
      "广告与产品",
      "写实电影感",
      "人物表演",
      "科幻太空"
    ],
    "category": "advertising",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2093747276942606456"
  },
  {
    "id": "x-2093697204384887023",
    "title": "黑发动漫人物的时装近景",
    "titleEn": "Fashion Close-Ups of a Dark-Haired Anime Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑发动漫人物在简洁背景中变换姿态，镜头切换身体与服装细节。",
    "summaryEn": "A dark-haired anime character changes poses against a minimal background with cuts to costume details.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2093697204384887023",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093697204384887023.mp4",
    "posterUrl": "/posters/x/2093697204384887023.jpg",
    "duration": 15.168,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动漫",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093685105474568202",
    "title": "日落街区里的轻松散步",
    "titleEn": "An Easy Neighborhood Walk at Sunset",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子沿安静街道散步，镜头从衣着细节转向远处夕阳。",
    "summaryEn": "A woman strolls through quiet streets as the camera moves from clothing details to the sunset.",
    "prompt": "Style: Ultra-realistic cinematic handheld footage, natural daylight fading into golden-hour sunset. Soft realistic film grain, gentle handheld sway, shallow-to-medium depth of field, warm color grading. Real Seoul residential neighborhood (narrow alleys, brick low-rise houses, hanok-style tiled roofs, utility poles, parked cars). No CGI, no beauty filter, no overly polished commercial look — believable skin texture, natural flyaway hairs, authentic wind movement.\n\nReference: Use exactly one reference image for the woman. Preserve her identity, face, skin tone, hairstyle, body proportions, and wardrobe consistently across every shot.\n\nCharacter: Young Korean woman, early-to-mid 20s, cream cardigan over sage-green button-up blouse, blue straight-leg jeans, white sneakers, brown crossbody/shoulder bag, small silver hoop/pearl earrings. Natural makeup with soft red-toned lips, glossy dark hair worn down, warm authentic smile throughout.\n\n00:00–00:03 — Approach Walk: She walks toward camera down a quiet daytime alley between low white and brick walls, relaxed casual pace, soft ambient smile, mid-shot tracking backward in front of her.\n\n00:03–00:07 — Crouch & Cat: She crouches down at a wall corner where a small tabby street cat sits, leaning in with a delighted laugh, one hand resting on her knee, warm candid side-profile framing.\n\n00:07–00:10 — Golden Turn: Cut to golden-hour light — she walks down a sloped residential street, sun flare grazing her hair and shoulders, glancing upward and around with a soft content smile, hair lifting gently in the breeze.\n\n00:10–00:13 — Wide Street Walk: Wider tracking shot past hanok rooftops and brick houses, city skyline visible in the distance, she strolls confidently, glancing off to the side, golden backlight rimming her hair.\n\n00:13–00:16 — Close Glance Back: Close-up as she turns her head over her shoulder toward camera, playful knowing smile, hair swept by wind, warm rim-light on her cheek and earring.\n\n00:16–00:19 — Walking Away: Camera holds behind her as she continues down the alley past a weathered painted wall, sunset light raking across the brick, bag swaying gently on her shoulder.\n\n00:19–00:20 — Final Look: She glances back over her shoulder once more with a soft laugh/smile as the shot settles, warm dusk light silhouetting her hair — natural cut-off mid-motion.\n\nRealism: Accurate hands and natural gait throughout, believable cat behavior and interaction, consistent wardrobe and hairstyle across all cuts, natural long shadows and warm sun flare in the golden-hour segments, soft handheld micro-shake, no teleporting, duplication, warped hands, extra fingers, or wardrobe changes between cuts.\n\nAudio: Natural location sound only — footsteps on pavement, soft ambient neighborhood sound (birds, distant traffic, faint breeze), fabric rustling, a soft cat meow/purr during the crouch beat, quiet natural breathing and laughter. No music, narration, subtitles, captions, logos, watermark, or on-screen text.",
    "sourceUrl": "https://x.com/afrinxai/status/2093685105474568202",
    "sourceLabel": "X 原帖",
    "author": "afrinxai",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093685105474568202.mp4",
    "posterUrl": "/posters/x/2093685105474568202.jpg",
    "duration": 20.8,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "城市街景",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/afrinxai/status/2093685105474568202"
  },
  {
    "id": "x-2093680788990476668",
    "title": "酒店房间里的浴袍变装",
    "titleEn": "A Bathrobe-to-Eveningwear Change in a Hotel",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在同一酒店房间里从浴袍切换为黑色裙装，并坐入扶手椅。",
    "summaryEn": "A figure changes from a bathrobe to a black dress in the same hotel room and sits in an armchair.",
    "prompt": "【风格】酒店浴袍高跟鞋卡点变装（Bathrobe-to-Glam Heel Transformation），电影感时尚短片（Cinematic Fashion Film），真实手机竖屏实拍质感（Photorealistic Handheld），8K超清，暖调低光室内摄影，皮肤有真实毛孔和光泽、布料可见织纹和垂坠感，无AI塑料感，竖屏9:16\n【时长】10秒\n【场景】高级精品酒店房间，始终同一个机位、同一把椅子\n【角色】Anna @图片1（全片唯一角色）。五官、脸型、身形严格锁定参考图，变装前后不漂移。\n\n【景别与机位基准】固定机位，略低角度（相机高度约在坐姿时膝盖位置、微微仰拍），椅子居画面中央偏右，左侧大面积逆光窗帘过曝形成高光留白，右侧深色木质墙面和暗色窗帘压暗，形成强烈的明暗对比（Chiaroscuro Lighting）。全片机位不移动、不推拉摇移。\n\n【核心变装机制：抛鞋→踢腿】\n道具触发变装——Anna 坐在椅子上把一只黑色高跟凉鞋向上一抛，鞋子离手还在空中上升的那一帧硬切（Hard Cut）→ 下一帧她已经穿好全套造型，右腿猛地向上一甩踢向镜头、鞋底正对镜头（鞋子已经穿在脚上），踢腿和抛鞋形成一对跨越剪辑的镜像动作——\"抛出去的鞋，下一帧就穿在了脚上\"。踢腿后右腿缓缓收回，跷成二郎腿坐定。变装前后椅子位置、机位、光线完全一致，只有人物的穿着和姿态整个换掉。\n\n[00:00-00:01.5] 镜头1：空椅子·悬念开场（Establishing Shot）\n画面：酒店房间内，左侧落地窗透过白色薄纱窗帘漫进大片柔和白光（窗帘过曝），右侧深色木质竖条纹护墙板和深灰色厚窗帘形成暗部。画面中央偏右一把黑色哑光金属框架扶手椅（黑色皮质坐垫），椅子空着，右上角墙上隐约可见一幅装裱画的边角。深灰色哑光地板。\n画面里没有人——空椅子制造悬念。\n\n[00:01.5-00:02.5] 镜头2：入场·浴袍登场（Character Entrance）\nAnna 从画面右侧走入镜头，赤脚踩在深灰色地板上。她穿着白色酒店浴袍（厚实毛巾布料质感、腰间系带打结、左胸口绣有酒店首字母），头上裹着白色浴巾扎成高高的包头巾（turban），脸上已经画好全妆——正红色哑光唇膏（红唇是全画面唯一的高饱和色彩点）、精致眉眼。双手捧着一双黑色绑带高跟凉鞋（细带交叉款，鞋跟约8厘米）放在腰间。\n她面对镜头走到椅子前方站定，表情冷淡从容，嘴唇微微抿起，下巴微微抬高，眼神直视镜头——\"给我几秒钟\"的自信态度。\n\n[00:02.5-00:04.5] 镜头3：落座·把玩高跟鞋（Seated Tease）\nAnna 转身坐到黑色扶手椅上，浴袍下摆散开，赤足的双腿自然并拢垂在椅子前方。她把两只黑色高跟鞋放在膝盖上，双手握住鞋子，低头审视。\n表情：微微垂眼看着手中的鞋，嘴角不笑、红唇轻抿，带着一种\"我在挑选武器\"的挑剔和笃定。\n动作细节：她把两只鞋分开，左手一只右手一只，轻轻掂了掂重量，手指拨弄鞋面的黑色细带。\n\n[00:04.0-00:04.5] 镜头4：抛鞋·变装触发（The Throw）\nAnna 右手握住一只黑色高跟鞋，手腕一翻、向上一抛——鞋子脱手飞向空中，黑色细带在空中散开甩动。左手同时还握着另一只鞋。\n表情：抛鞋瞬间眼睛微微闭上，红唇轻抿，下巴微扬，一副\"就这样吧\"的随意。\n鞋子离手、还在空中上升的那一帧——**硬切（Hard Cut）**\n\n[00:04.5-00:05.5] 镜头5：踢腿亮鞋·变装完成（The Kick Reveal）\n硬切之后，Anna 依然坐在同一把黑色扶手椅上，但整个人焕然一新：\n穿着：白色浴袍和包头巾消失，头发放下来自然散落（hair down）。妆容比变装前更精致——眉形利落、眼妆深邃带微光、正红色哑光唇膏依旧，整体是高端时尚大片级别的妆面质感。身穿黑色无袖高领修身连衣裙（halter-neck mini dress，面料紧贴身体轮廓、裙摆到大腿中段），双腿穿着黑色薄透丝袜（sheer black tights，织纹细腻可见），脚上穿着刚才手里那双黑色绑带高跟凉鞋（细带交叉绑到脚踝上方）。\n关键动作：硬切的第一帧，右腿猛地向上甩踢，整条腿斜着踢向镜头方向——高跟鞋鞋底正对镜头、占据画面中央，鞋底的皮质纹理清晰可见。这一踢和上一帧的抛鞋动作形成跨越剪辑的因果呼应：\"刚才抛出去的鞋，这一帧已经穿在脚上了。\"\n表情：踢腿瞬间目光斜睨镜头，眉梢微挑，红唇微启——冷飒的\"准备好了\"。\n随后右腿缓缓从高处收回，优雅地跷在左腿上（二郎腿），身体微微向左后方靠进椅背，左手手指搭在左侧扶手边缘——慵懒、松弛、掌控一切的坐姿。\n\n[00:05.5-00:10.0] 镜头6：慵懒女王·时尚大片（Slow Fashion Pose）\n接下来4.5秒内，Anna 在椅子上做一组极慢的时尚大片姿态变换（Fashion Editorial Posing），每个动作之间自然衔接、一镜到底没有剪辑：\n① 二郎腿跷腿姿势稳定，目光从镜头缓缓移向左侧窗帘方向，侧脸轮廓被逆光勾勒出金色边缘光；\n② 右手从扶手上缓缓抬起，五指微张在空中做了一个轻柔的\"弹指/拨弄\"手势，手指骨节分明；\n③ 右手继续向上，指尖触到右侧肩旁的一缕发丝，轻轻捻起发梢向外拉出一小段，发丝在指间缠绕；\n④ 头微微向右后方仰，闭上眼睛约0.5秒，红唇微启，发丝从指间滑落；\n⑤ 缓缓睁眼，目光重新对准镜头，嘴角带出一丝极淡的微笑——但只有嘴角，眼神依然是冷的。右手回落到扶手上。\n画面在这个\"冷眼微笑\"的定格中自然结束。\n\n【音效】\n前半段（00:00-05:30）：安静的酒店房间环境音底噪——远处隐约的城市嗡鸣、空调低频运转。叠加：赤脚踩地板的轻柔脚步声、浴袍布料摩擦的窸窣声、鞋子被拿起和翻转时皮革与金属扣的轻微碰击声。\n硬切瞬间（05:30）：一声干脆利落的低频鼓点\"咚\"（Kick Drum Hit），标记变装切点。\n后半段（05:30-10:00）：慵懒的电子氛围音乐缓缓铺开（Lo-fi Chill / Downtempo），节奏极慢，配合Anna 的慢动作姿态。叠加：丝袜面料在交叉腿时的轻微摩擦声、高跟鞋跟在换姿势时轻轻碰到椅腿的\"嗒\"声、手指拨弄发丝的细微声响。\n\n【一致性锁定】\n1. 全片固定机位，不推拉摇移，不换角度。\n2. 同一把黑色扶手椅，位置不变。\n3. 变装前后光线一致：左侧窗帘逆光过曝、右侧暗部压低。\n4. 脸部严格锁定 @图片1，变装前后不出现任何面部漂移。\n5. 变装前的高跟鞋和变装后脚上穿的是同一双鞋（黑色细带交叉绑带凉鞋），细节一致。\n6. 红唇是全片唯一的高饱和色彩，变装前后保持一致。\n7. 变装只用一次硬切，不许出现任何转场特效、模糊、叠化、黑场、闪白、魔法粒子。\n8. 画面内不出现任何品牌logo、文字、字幕、水印。\n9. 不出现镜面反射或水面倒影中的人像。\n10. 后半段pose变换是连续拍摄，不再有任何剪辑切换，一镜到底。",
    "sourceUrl": "https://x.com/johnAGI168/status/2093680788990476668",
    "sourceLabel": "X 原帖",
    "author": "johnAGI168",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093680788990476668.mp4",
    "posterUrl": "/posters/x/2093680788990476668.jpg",
    "duration": 10.208,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/johnAGI168/status/2093680788990476668"
  },
  {
    "id": "x-2093670361879462037",
    "title": "饼干吉祥物的 Y2K 包装广告",
    "titleEn": "A Y2K Cookie Mascot and Packaging Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "拟人饼干、包装和大字在明亮的粉蓝广告场景中切换。",
    "summaryEn": "A cookie mascot, packaging and bold type alternate in bright pink-and-blue ad scenes.",
    "prompt": "Make a short advertisement for cookie product with the original COOKIE CLUB cookie brand as the protagonist. Keep the packaging, mascot and English text clear, and use Y2K street youth, sticker animation and colorful collage style to show the brand world, product unboxing, nibbling lifestyle and final version.",
    "sourceUrl": "https://x.com/yourPlugAI/status/2093670361879462037",
    "sourceLabel": "X 原帖",
    "author": "yourPlugAI",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093670361879462037.mp4",
    "posterUrl": "/posters/x/2093670361879462037.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/yourPlugAI/status/2093670361879462037"
  },
  {
    "id": "x-2093563293306929579",
    "title": "复古室内的双人对白近景",
    "titleEn": "Period-Style Dialogue in a Quiet Room",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "穿复古服装的男女在窗边对话，镜头停留于女方面部反应。",
    "summaryEn": "A pair in period clothing speak beside a window, with close-ups of the woman's reactions.",
    "prompt": null,
    "sourceUrl": "https://x.com/PixelAigc/status/2093563293306929579",
    "sourceLabel": "X 原帖",
    "author": "PixelAigc",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093563293306929579.mp4",
    "posterUrl": "/posters/x/2093563293306929579.jpg",
    "duration": 30.784,
    "aspectRatio": "9:5",
    "resolution": "360p",
    "tags": [
      "对白与口型",
      "写实电影感",
      "复古胶片",
      "人物表演"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "retro"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093550413144584656",
    "title": "摩托穿越烈焰中的恐龙森林",
    "titleEn": "A Motorcycle through a Burning Dinosaur Forest",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "骑手在燃烧的森林中疾驰，巨型恐龙出现在追逐路线旁。",
    "summaryEn": "A rider speeds through a burning forest with giant dinosaurs beside the chase route.",
    "prompt": null,
    "sourceUrl": "https://x.com/WuxiaRocks/status/2093550413144584656",
    "sourceLabel": "X 原帖",
    "author": "WuxiaRocks",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093550413144584656.mp4",
    "posterUrl": "/posters/x/2093550413144584656.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "3D / CG",
      "动作打斗",
      "动物与生物"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093545479338373323",
    "title": "云崖之上凝成金色法相",
    "titleEn": "A Golden Avatar Forms above the Cliffs",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "光尘聚成白衣人物，随后显现包围身体的巨大金色轮廓。",
    "summaryEn": "Particles form a white-robed figure before a vast golden outline appears around the body.",
    "prompt": "00：00-00：05（5秒）｜航拍镜头自高空俯冲向云雾缭绕的荒古悬崖，极缓慢下沉式推近，转为低角度仰拍。虚无空间中骤然迸发万千银白色与淡金色灵光粒子，如星河倒卷、流萤归巢，层层旋绕缓慢汇聚。光点从混沌模糊中逐步勾勒出一位女子修长挺拔的素白仙裙轮廓，广袖与裙摆在光尘中若隐若现。脚下金边白莲道印缓缓旋转凝形。粒子明暗跟随光影如呼吸般变化。史诗级开场氛围，浅景深，冷史诗色调。\n00：05-00：08（3秒）｜平滑弧形环绕跟拍，女子真身完整显现。她是一位二十岁女子，肤若凝脂覆淡银柔光，银白长发高束如月华垂落，头戴金丝五瓣星辰花步摇，珍珠流苏轻颤。远山弯月细眉，深邃沉静眼眸缓缓睁开，淡粉薄唇微抿，神情清冷疏离，睥睨天下。身着三重素白渐变鎏金高领广袖仙裙：内层珠光缎温润如脂，中层银线流光纱星辉隐现，外层蝉翼透纱如烟如雾。全身满绣重工金线天道云纹缠枝。三重叠加金色璎珞锁骨链，镂空天道铭文金丝臂钏，日月同辉灵珠带扣束腰。周身淡白护体灵光微微流转。侧逆光轮廓光，极致布料与珠宝细节，神性压迫氛围。\n00：08-00：10（2秒）｜镜头缓慢后拉仰拍，女子缓缓抬起右手，纤长玉指于胸前优雅结印。银白长发被渐起的气流拂动，三重广袖轻扬。结印完成瞬间，指间金芒迸现，淡金鸿蒙道纹自指尖向全身蔓延流转，如金色血脉蔓延。脚下白莲道印加速旋转光芒愈盛，周身灵光剧烈涌动，气流渐成旋涡。广袖与长发被愈演愈烈的能量之风猛烈扬起，眼眸中倒映金色光芒。此刻能量蓄满，一触即发，但法相尚未显现。真实布料动力学，动态能量气场，极致张力。\n00：10-00：15（5秒）｜女子眼中金芒大盛，周身淡金鸿蒙道纹轰然向外扩散，脚下金边白莲道印完全绽放，万丈光芒直冲云霄。身后空间骤然显现一尊半透明淡金万丈法相虚影，法相面容与她一致，双眼微阖后缓缓睁开，庄严宏大，神威如岳。狂暴灵光粒子呈环形向外层层扩散，虚空震鸣。三重素白纱袍广袖被法术劲风剧烈翻飞，蝉翼外层如云海狂舞，高开衩处腿部与金丝月白流苏足链完全显现。银色长发肆意狂舞，全身璎珞步摇腰链臂钏随真实物理运动震颤不止。极致布料动力学，能量大爆发，神性压迫氛围。",
    "sourceUrl": "https://x.com/johnAGI168/status/2093545479338373323",
    "sourceLabel": "X 原帖",
    "author": "johnAGI168",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093545479338373323.mp4",
    "posterUrl": "/posters/x/2093545479338373323.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "奇幻",
      "人物表演"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/johnAGI168/status/2093545479338373323"
  },
  {
    "id": "x-2093544965485859087",
    "title": "核动力货船的荒诞事故片段",
    "titleEn": "An Absurd Nuclear Cargo-Ship Incident",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "货船与集装箱在海港制造混乱，镜头切向驾驶室和岸边工作人员。",
    "summaryEn": "A cargo ship and containers create harbor chaos as the view cuts to the bridge and dock workers.",
    "prompt": null,
    "sourceUrl": "https://x.com/floopers966/status/2093544965485859087",
    "sourceLabel": "X 原帖",
    "author": "floopers966",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093544965485859087.mp4",
    "posterUrl": "/posters/x/2093544965485859087.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "城市街景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093532752541024373",
    "title": "纸飞机飞出列车窗口",
    "titleEn": "A Paper Plane Leaves a Train Window",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "纸飞机从车窗旁飞向田野上空，片尾附有制作平台展示。",
    "summaryEn": "A paper plane leaves a train window and flies over fields; the ending includes a platform showcase.",
    "prompt": null,
    "sourceUrl": "https://x.com/prompt_soru/status/2093532752541024373",
    "sourceLabel": "X 原帖",
    "author": "prompt_soru",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093532752541024373.mp4",
    "posterUrl": "/posters/x/2093532752541024373.jpg",
    "duration": 26.470748,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "自然风景",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093523920045699263",
    "title": "海滨爱情片的片头字幕",
    "titleEn": "Opening Credits for a Seaside Romance",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "列车、自行车和海边人物镜头与演员名字、片名交替出现。",
    "summaryEn": "Train, bicycle and seaside portraits alternate with cast names and a film title.",
    "prompt": "Opening Title｜15秒、2.39:1、24fps真人爱情卡司PV《SUMMER, WITH YOU》。原创成年演员MIRA LIN、JULIAN PARK，海边小城35mm胶片感，奶油白/海蓝/淡黄。四镜：海岸列车无字；书店女主配“STARRING/MIRA LIN”；港口男主配“JULIAN PARK”；街角相遇后夕阳步道揭示片名。文字衬线、留白充足、遮挡转场，不接吻摆拍、无额外文字/Logo/黑屏。",
    "sourceUrl": "https://x.com/ponyodong/status/2093523920045699263",
    "sourceLabel": "X 原帖",
    "author": "ponyodong",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093523920045699263.mp4",
    "posterUrl": "/posters/x/2093523920045699263.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "复古胶片",
      "写实电影感",
      "城市街景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ponyodong/status/2093523920045699263"
  },
  {
    "id": "x-2093516077674246461",
    "title": "黑白红字体组成的鼓励短片",
    "titleEn": "An Encouraging Film in Black, White and Red Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "简短英文句子通过字号、颜色和版式变化逐段出现。",
    "summaryEn": "Short English sentences appear through changes in size, color and layout.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2093516077674246461",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093516077674246461.mp4",
    "posterUrl": "/posters/x/2093516077674246461.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093514079180894701",
    "title": "狭窄巷道里的僵尸逃生",
    "titleEn": "Escaping Zombies through Narrow Alleys",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子挥动棍棒抵挡袭击，随后沿狭窄通道奔跑。",
    "summaryEn": "A woman uses a staff to fend off attackers and runs through narrow passages.",
    "prompt": null,
    "sourceUrl": "https://x.com/kan8888kawa/status/2093514079180894701",
    "sourceLabel": "X 原帖",
    "author": "kan8888kawa",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093514079180894701.mp4",
    "posterUrl": "/posters/x/2093514079180894701.jpg",
    "duration": 13.024,
    "aspectRatio": "17:23",
    "resolution": "544p",
    "tags": [
      "动作特效",
      "写实电影感",
      "暗黑恐怖",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093512228242931917",
    "title": "驾驶舱内的耳麦通话",
    "titleEn": "A Headset Conversation in the Cockpit",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "戴耳麦的飞行员坐在仪表台前，转头向旁人说话。",
    "summaryEn": "A headset-wearing pilot sits before the instruments and turns to speak to someone nearby.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2093512228242931917",
    "sourceLabel": "X 原帖",
    "author": "HBCoop_",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093512228242931917.mp4",
    "posterUrl": "/posters/x/2093512228242931917.jpg",
    "duration": 10.20043,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "写实电影感",
      "人物表演"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093497840358264932",
    "title": "书房里的蘸墨与书写",
    "titleEn": "Dipping Ink and Writing in a Study",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "长发动漫人物坐在书桌前，在蘸墨与落笔之间切换。",
    "summaryEn": "A long-haired anime character alternates dipping a pen and writing at a study desk.",
    "prompt": null,
    "sourceUrl": "https://x.com/imaginaldomain/status/2093497840358264932",
    "sourceLabel": "X 原帖",
    "author": "imaginaldomain",
    "publishedAt": "2026-08-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093497840358264932.mp4",
    "posterUrl": "/posters/x/2093497840358264932.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "576p",
    "tags": [
      "动画",
      "动漫",
      "日常生活",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093372840598974696",
    "title": "修好自行车后的重新出发",
    "titleEn": "Setting Off after Repairing a Bicycle",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "卡通老人从修车时的沮丧转向重新上路，图形文字穿插其间。",
    "summaryEn": "A cartoon elder moves from frustration with a bicycle repair to setting off again, intercut with graphic text.",
    "prompt": null,
    "sourceUrl": "https://x.com/airina_xyz/status/2093372840598974696",
    "sourceLabel": "X 原帖",
    "author": "airina_xyz",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093372840598974696.mp4",
    "posterUrl": "/posters/x/2093372840598974696.jpg",
    "duration": 15.146667,
    "aspectRatio": "4:7",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动态图形",
      "日常生活",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093371555350716920",
    "title": "黑色手套与极简时装章节",
    "titleEn": "Black Gloves and Minimal Fashion Chapters",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑衣模特、手套细节与留白背景组成章节式时装画面。",
    "summaryEn": "A black-clad model, glove details and spacious backdrops form a chapter-based fashion sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/sahilvermaai/status/2093371555350716920",
    "sourceLabel": "X 原帖",
    "author": "sahilvermaai",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093371555350716920.mp4",
    "posterUrl": "/posters/x/2093371555350716920.jpg",
    "duration": 15.232,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093371464682799229",
    "title": "灯光下熟睡的银发兽耳角色",
    "titleEn": "A Silver-Haired Animal-Eared Character Asleep",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "银发兽耳角色躺在枕边熟睡，床头灯照亮安静房间。",
    "summaryEn": "A silver-haired animal-eared character sleeps beside a pillow under a bedside lamp.",
    "prompt": null,
    "sourceUrl": "https://x.com/Venom_Creations/status/2093371464682799229",
    "sourceLabel": "X 原帖",
    "author": "Venom_Creations",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093371464682799229.mp4",
    "posterUrl": "/posters/x/2093371464682799229.jpg",
    "duration": 15.072,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093345360710344839",
    "title": "水墨武者的兵器连招",
    "titleEn": "Ink Warriors and Flowing Weapon Moves",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "弓箭、长兵器与旋转身形由黑色墨迹连接成连续武术画面。",
    "summaryEn": "Black ink connects archery, polearm moves and spinning figures in a martial-arts sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/vadooai/status/2093345360710344839",
    "sourceLabel": "X 原帖",
    "author": "vadooai",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093345360710344839.mp4",
    "posterUrl": "/posters/x/2093345360710344839.jpg",
    "duration": 14.272,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "插画与手绘",
      "风格化与实验",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093343507872383069",
    "title": "火龙果变成果汁的广告转场",
    "titleEn": "Dragon Fruit Becomes a Juice Commercial",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物展示火龙果，镜头切向搅拌杯与明亮红色果汁。",
    "summaryEn": "A presenter displays dragon fruit before cuts to a blender and bright red juice.",
    "prompt": null,
    "sourceUrl": "https://x.com/ZoeyBennettAi/status/2093343507872383069",
    "sourceLabel": "X 原帖",
    "author": "ZoeyBennettAi",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093343507872383069.mp4",
    "posterUrl": "/posters/x/2093343507872383069.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093340157621760470",
    "title": "粉色摩托与未来城市时装片",
    "titleEn": "A Pink Motorcycle in a Future-City Fashion Film",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "骑手与粉色摩托在棚拍近景、大字和未来街道之间切换。",
    "summaryEn": "A rider and pink motorcycle alternate between studio close-ups, oversized type and futuristic streets.",
    "prompt": null,
    "sourceUrl": "https://x.com/applete77191758/status/2093340157621760470",
    "sourceLabel": "X 原帖",
    "author": "applete77191758",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093340157621760470.mp4",
    "posterUrl": "/posters/x/2093340157621760470.jpg",
    "duration": 17.386667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "动态图形",
      "时尚",
      "科幻太空"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093322991602176374",
    "title": "风暴海岸上的幸存者",
    "titleEn": "Survivors on a Stormy Shore",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "男女幸存者坐在浪涛与火光之间，镜头最后推向人物面部。",
    "summaryEn": "Two survivors sit between surf and scattered fires before a close-up of a man's face.",
    "prompt": null,
    "sourceUrl": "https://x.com/Kiril_AI_art/status/2093322991602176374",
    "sourceLabel": "X 原帖",
    "author": "Kiril_AI_art",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093322991602176374.mp4",
    "posterUrl": "/posters/x/2093322991602176374.jpg",
    "duration": 14.464,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "自然风景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093309199027757134",
    "title": "舞台灯光下的哥特摇滚乐队",
    "titleEn": "A Gothic Rock Band under Stage Lights",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑衣乐队在背光舞台演奏，镜头穿插鼓手与贝斯手近景。",
    "summaryEn": "A black-clad band performs on a backlit stage, intercut with drummer and bassist close-ups.",
    "prompt": null,
    "sourceUrl": "https://x.com/tsukuyomi2021/status/2093309199027757134",
    "sourceLabel": "X 原帖",
    "author": "tsukuyomi2021",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093309199027757134.mp4",
    "posterUrl": "/posters/x/2093309199027757134.jpg",
    "duration": 15.046531,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "暗黑恐怖",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "dark"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093291618673578434",
    "title": "蓝色笔绘风的云端战斗",
    "titleEn": "A Blue-Ink Battle above the Clouds",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "兜帽武者在破碎石柱与云海之间交锋，保留蓝色线描纹理。",
    "summaryEn": "Hooded warriors clash among broken pillars and clouds in a blue line-drawn style.",
    "prompt": null,
    "sourceUrl": "https://x.com/VigoCreativeAI/status/2093291618673578434",
    "sourceLabel": "X 原帖",
    "author": "VigoCreativeAI",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093291618673578434.mp4",
    "posterUrl": "/posters/x/2093291618673578434.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "插画与手绘",
      "风格化与实验",
      "动作打斗",
      "奇幻"
    ],
    "category": "action-vfx",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093286493141823991",
    "title": "从眼镜日常装到精致穿搭",
    "titleEn": "From Everyday Glasses to a Styled Look",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物摘下眼镜并改变妆容服装，最后走进红色走廊。",
    "summaryEn": "A character removes her glasses, changes styling and clothing, and appears in a red corridor.",
    "prompt": "摘眼镜性感变装反差\n\n【时长】12秒\n【画幅】9:16竖屏\n【风格】真人短视频变装，日常生活感与精致性感造型形成强烈反差，手机纪实镜头，快速匹配剪辑，轻微梦幻过曝转场，真实光影与皮肤质感，4K高清\n【语言】无对白，无字幕\n【人物数量】画面中始终只有一名成年女性\n\n【人物角色卡】\n@图片1 命名“星野”。参考图只用于锁定人物长相、脸部身份和整体人物一致性，不继承参考图中的服装、眼镜、妆感和眼瞳状态。人物发型不需要文字描述，保持同一人物即可。\n\n【转场前状态】\n星野穿宽松柔软的家居服和居家长裤，赤脚，整体普通、舒适、略显松垮。佩戴一副日常框架眼镜，眼睛保持原本自然状态，没有彩色美瞳。\n\n妆感非常清淡，姿势随意，肩膀放松，神情慵懒冷淡，背景是带有生活杂物的普通室内空间。\n\n【转场后状态】\n仍然是星野本人，脸部身份完全一致，但已经完成明显升级变装：\n\n佩戴自然通透的冷色彩色美瞳，瞳色清晰可见但不能发光或变成异形眼睛；\n妆感变得精致立体，眼神更深、更有表现力；\n换成深色修身性感造型，以贴身剪裁、适度露肩或露腰设计、短款百褶裙和精致配饰构成；\n整体为高级、轻熟、时尚的性感，不是内衣造型，不色情化；\n重新戴好同一副框架眼镜，彩色美瞳可以透过镜片清楚看见。\n\n变装后身体姿态明显改变：肩背打开、腰线明确、重心集中、动作从慵懒随意变成自信、有控制力。\n\n【核心反差】\n转场前：宽松家居服、普通眼睛、淡妆、赤脚、生活感、慵懒冷淡。\n转场后：彩色美瞳、精致妆感、修身性感造型、成熟自信、镜头表现力明显增强。\n\n不能只更换背景，也不能只改变眼镜。服装轮廓、眼瞳颜色、妆感、姿态和人物气场必须同时发生明显变化。\n\n[00:00-00:01.20] 镜头1｜家居状态摘下眼镜\n\n普通室内生活空间，背景带有沙发、置物架和自然生活杂物。\n\n竖屏正面中近景，摄影机接近人物眼睛高度，基本固定。\n\n星野穿着宽松家居服，正面面对镜头。肩膀放松，姿势略显慵懒，嘴唇闭合，眼睑稍低，表情平静冷淡。\n\n00:00-00:00.35\n她缓慢抬起画面左侧的手，用拇指与食指夹住同侧镜框外缘与镜腿连接处。\n\n00:00-00:00.65\n她沿鼻梁缓慢把眼镜向下拉。\n\n双眼从镜框上缘露出。此时是普通自然眼睛，不能提前出现彩色美瞳。\n\n她从眼镜上方直视镜头，眼神慵懒、冷淡，带一点漫不经心。\n\n00:00-00:01.00\n她将眼镜完整取下，镜腿保持展开。\n\n眼镜经过鼻尖和嘴部前方，随后被带到胸前下方。眼镜不能贴到摄影机镜头，也不能遮满整个画面。\n\n00:01-00-01.20\n眼镜和手退出画面下沿。\n\n星野轻轻歪头，继续看着镜头，保持普通居家状态和冷淡表情。\n\n[00:01.20-00:03.70] 镜头2｜镜头下摇展示性感造型\n\n00:01.20-00:01.75\n星野保持原位，身体几乎不动。宽松家居服的生活感必须清楚可见。\n\n00:01.75-00:02.70\n摄影机突然快速向下俯摇。\n\n镜头依次扫过宽松家居服、居家长裤和赤脚，带有自然运动模糊。\n\n00:02.70-00:03.20\n镜头到达地面俯拍角度。\n\n地面整齐铺放着转场后要穿的完整性感造型：深色修身上装、短款百褶裙和对应配饰。服装摆放完整，轮廓清晰，不能堆成无法辨认的一团。\n\n星野向前移动双脚，站到造型旁边。\n\n00:03.20-00:03.70\n摄影机稳定停留在地面造型上。\n\n转场前宽松居家服的裤腿和赤脚仍位于画面边缘，与地面精致性感造型形成第一层视觉反差。\n\n[00:03.70-00:04.20] 镜头3｜过曝叠化完成变装\n\n白色高亮从地面性感造型周围开始扩散。\n\n高亮逐渐覆盖宽松家居裤、赤脚和地面，但仍短暂保留服装轮廓。\n\n随后发生约0.2秒交叉叠化：\n\n上一镜铺在地面的性感造型保持半透明；\n下一镜中，星野已经穿好这套造型；\n地面服装轮廓与穿着后的人物身体轮廓短暂对齐重合；\n普通自然眼睛同步变成清晰的彩色美瞳；\n淡妆同步升级为精致妆感；\n宽松家居状态同步变成挺拔、自信的性感状态。\n\n所有变化必须集中在同一个转场内完成，不能服装先变、美瞳几秒后才出现。\n\n[00:04.20-00:07.00] 镜头4｜性感造型正式揭晓\n\n硬切到明亮通透的室内空间，背景有玻璃门窗和圆桌。\n\n镜头为正面胸腰以上中近景。\n\n星野已经完成变装，穿着深色修身性感造型，并重新戴好框架眼镜。彩色美瞳透过镜片清晰可见。\n\n00:04.20-00:04.60\n她坐在圆桌后方，头略微低下，双手位于画面下沿。\n\n虽然已经完成性感变装，但不能第一帧就夸张摆拍。先让观众看清楚服装轮廓、眼镜、美瞳和妆感变化。\n\n00:04.60-00:05.10\n星野缓慢抬眼直视镜头。\n\n彩色美瞳成为视觉焦点。她的眼神从之前的慵懒冷淡，变成自信、锐利、有吸引力。\n\n肩膀自然打开，腰背挺直，整个人的气场明显改变。\n\n00:05.10-00:05.90\n她抬起一只手，指尖轻轻触碰一侧镜框和眼下脸颊。\n\n动作缓慢、精确，像是在刻意展示透过镜片可见的彩色美瞳。\n\n眼镜不能被再次摘下，也不能遮挡眼睛。\n\n00:05.90-00:06.30\n她把手放下，身体略微侧转，形成自然肩腰曲线。\n\n00:06.30-00:07.00\n星野完成一次从容的单眼眨眼。\n\n一只眼睛缓慢闭合，另一只佩戴彩色美瞳的眼睛继续盯着镜头。嘴角出现极浅的自信弧度，表现成熟性感，不故意卖萌。\n\n[00:07.00-00:08.00] 镜头5｜性感气场突然变得俏皮\n\n在单眼眨眼结束前直接跳切。\n\n星野突然双手举起粉色长耳玩偶，把玩偶竖直抱在脸前。\n\n双手分别握住玩偶身体两侧，一只玩偶长耳靠近她的嘴边。\n\n她张嘴轻轻含住玩偶耳朵，再次眨一只眼睛。\n\n这一瞬间形成第二层反差：\n\n人物依然穿着成熟性感造型并佩戴彩色美瞳；\n行为却突然变得俏皮、可爱；\n不能把人物重新变回家居服。\n\n本段为动态视频，手指、眼睑、嘴角和玩偶保持轻微自然运动。\n\n[00:08.00-00:09.00] 镜头6｜保持玩偶动作匹配换到室外\n\n在玩偶保持居中、双手抓握位置以及嘴贴玩偶耳朵的关系基本一致时，硬切到户外道路。\n\n镜头从室内坐姿中景切成室外站姿大腿以上中景。\n\n星野仍穿着相同性感造型，框架眼镜和彩色美瞳保持不变。\n\n她站在户外，继续双手抱住玩偶，玩偶长耳仍贴在嘴边，一只眼睛微微眯起，另一只眼睛直视镜头。\n\n转场前后道具位置和动作匹配，只有场景、光线、景别和身体重心突然改变。\n\n[00:09.00-00:10.30] 镜头7｜第二次户外匹配转场\n\n保持玩偶、双手和嘴部动作相似，再次硬切到阳光更强的另一处街边。\n\n00:09.00-00:09.50\n星野继续轻咬玩偶耳朵，双眼透过眼镜看向镜头，彩色美瞳保持清晰。\n\n00:09.50-00:10.00\n她缓慢松开嘴唇，把脸颊贴近玩偶，头部轻轻侧偏。\n\n00:10.00-00:10.30\n她抱紧玩偶，眼睛逐渐弯起，露出甜美笑容。\n\n性感造型、彩色美瞳与抱玩偶甜笑形成第三层反差：外形成熟性感，动作和表情却突然可爱柔软。\n\n[00:10.30-00:11.75] 镜头8｜甜美近景硬切冷艳全身\n\n在甜笑达到最明显的瞬间直接硬切。\n\n画面从户外阳光下的贴脸中近景，突然切到饱和红色墙面与红色立柱组成的室内走廊。\n\n镜头同时从中近景跳成正面全身远景。\n\n星野仍穿着同一套性感造型，彩色美瞳、妆感、眼镜和配饰全部保持连续。\n\n玩偶从脸旁降到胸腹前，她用双臂稳定环抱玩偶，双腿靠拢，一条腿略微交叠，身体重心落在另一条腿上。\n\n00:10.30-00:10.80\n切换后的第一帧，上一镜的甜笑立即消失。\n\n她的眼睛由半闭逐渐睁开，嘴唇从微张变成闭合。\n\n00:10.80-00:11.30\n她平静直视镜头，肩膀打开，腰背挺直，姿态成熟自信。\n\n00:11.30-00:11.75\n她抱着玩偶保持冷艳全身站姿。\n\n彩色美瞳透过眼镜看向镜头，眼神重新变得锐利克制。甜萌近景与冷艳性感全身镜头形成最后一次强反差。\n\n[00:11.75-00:12.00] 镜头9｜玫瑰遮罩收黑\n\n黑色遮罩从画面四周快速向中央收拢。\n\n中央暂时保留一朵玫瑰花形状的透明窗口。星野穿着性感造型、抱着玩偶的全身轮廓位于玫瑰花瓣内部，并产生轻微柔焦。\n\n玫瑰窗口继续缩小，最后一帧完全黑场。\n\n【声音设计】\n全片无对白，只使用节奏清晰、带梦幻变装感的背景音乐。\n\n摘眼镜配合细小镜框摩擦声；\n镜头快速下摇配合短促风切声；\n白色叠化变装配合明显上升音效和音乐重拍；\n性感造型出现时加入低沉有力的节拍落点；\n玩偶匹配换场使用连续快速鼓点；\n10.30秒甜美近景切冷艳全身时使用全片最强重拍；\n玫瑰遮罩随收束音效进入黑场。\n\n【主要约束】\n参考图只锁定人物长相，不能继承参考图服装限制变装；\n转场前必须是宽松家居服、普通眼睛和生活状态；\n转场后必须同时出现彩色美瞳、精致妆感和成熟性感造型；\n彩色美瞳自然通透，不发光、不变成异形瞳孔；\n性感造型保持高级时尚，不色情化、不变成内衣；\n禁止换脸、人物身份漂移、服装随机跳变、道具变形、额外人物、字幕、Logo和水印。",
    "sourceUrl": "https://x.com/johnAGI168/status/2093286493141823991",
    "sourceLabel": "X 原帖",
    "author": "johnAGI168",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093286493141823991.mp4",
    "posterUrl": "/posters/x/2093286493141823991.jpg",
    "duration": 12.32,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/johnAGI168/status/2093286493141823991"
  },
  {
    "id": "x-2093263619677212894",
    "title": "樱桃粉色的卡通 MV",
    "titleEn": "A Cherry-Pink Cartoon MV",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "粉发卡通人物在樱桃图案和明亮粉色版式中表演。",
    "summaryEn": "A pink-haired cartoon character performs among cherry motifs and bright pink layouts.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2093263619677212894",
    "sourceLabel": "X 原帖",
    "author": "Mayz1169",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093263619677212894.mp4",
    "posterUrl": "/posters/x/2093263619677212894.jpg",
    "duration": 14.628571,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093251869120565667",
    "title": "黑白烟雾中的 SMOKE 字效",
    "titleEn": "SMOKE Typography in Black and White",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑白字体在烟雾、颗粒和几何版式之间变形切换。",
    "summaryEn": "Monochrome type shifts through smoke, grain and geometric layouts.",
    "prompt": "Use the provided “SMOKE” image as the exact visual reference and opening frame.  Create a 15-second experimental 2D kinetic-typography video in a strict 16:9 format.  Preserve the original monochrome photocopy aesthetic: dirty white background, distressed black borders, analog film grain, dust, scratches, registration marks, microtext and raw printed textures.  Transform the central typography continuously through the \n\nfollowing sequence:  SMOKE → HAZE → VAPOR → CLOUD → AIR → DUST → TRACE → VOID → SMOKE \n\nThe words must change exclusively through the continuous deformation and reconstruction of the existing letterforms.  The current letters stretch, bend, split, compress, rotate as flat graphic shapes, change width, change weight, rearrange their internal components and reconnect to construct the next word.  Every new word must evolve directly from the previous word. Never remove one word and introduce another separate word.  Create all letter transformations exclusively through flat 2D motion-graphics animation. Use typographic shape morphing, path deformation, modular letter construction, animated masks, fragmented letter components and geometric restructuring.  The camera and overall framing remain completely fixed throughout the video.  Do not use zoom-in, zoom-out, camera movement, reframing or simulated camera scaling.  Do not use scene transitions, cuts, wipes, fades, flashes, sliding transitions or full-screen transition effects.  The transformation must always remain visible as one continuous motion inside the same composition.  Constantly vary the letterforms. Typography changes between condensed, extended, bold, thin, fragmented, angular, rounded and highly abstract geometric structures before becoming readable again.  Animate small flat geometric elements around and inside the changing typography: thin lines, circles, rectangles, grids, crosshairs, measurement marks, coordinates and abstract technical symbols.  Include many tiny and intricate HUD elements that move, flicker, update and restructure rapidly in synchronization with the transforming letters.  The HUD and geometric graphics should actively participate in the typography transformation. Lines connect letter fragments, grids guide their movement, circles reshape counters, and measurement marks reposition letter components.  Use the entire frame as one coordinated flat graphic system. The distressed border, background grain, scratches, registration marks, microtype and geometric elements should continuously react to the central letter deformation.  Around the middle of the video, briefly enter extreme super-slow motion. The typography continues deforming very slowly, allowing individual letter components, paths, microtext and thin geometric lines to become clearly visible.  After this short slow-motion moment, smoothly return to extremely fast typographic deformation without using a transition or cut.  End with all letter components continuously restructuring back into the original centered word:  SMOKE  Finish with the same composition, scale and visual balance as the supplied reference image.  STRICT CONSTRAINTS:  Purely flat 2D motion graphics only.  Letter changes must be created only through continuous typographic deformation.  Fixed camera and fixed framing.  No zoom-in or zoom-out.  No camera movement.  No cuts.  No scene transitions.  No wipes.  No fades.  No sliding words entering from outside the frame.  No separate titles appearing or disappearing.  No realistic smoke.  No 3D typography.  No extrusion.  No depth.  No perspective.  No dimensional rotation.  No glossy surfaces.  No shadows suggesting depth.  No color.  No logos or watermarks.",
    "sourceUrl": "https://x.com/totaku_original/status/2093251869120565667",
    "sourceLabel": "X 原帖",
    "author": "totaku_original",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093251869120565667.mp4",
    "posterUrl": "/posters/x/2093251869120565667.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "复古胶片",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "retro"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/totaku_original/status/2093251869120565667"
  },
  {
    "id": "x-2093222355665322439",
    "title": "从云层俯冲到城市人像",
    "titleEn": "From Clouds to a City Portrait",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头从高空云层切向密集城市，最后落到戴宽檐帽的人物近景。",
    "summaryEn": "The view moves from high clouds to a dense city and ends on a portrait with a wide-brimmed hat.",
    "prompt": null,
    "sourceUrl": "https://x.com/Nora_GMI/status/2093222355665322439",
    "sourceLabel": "X 原帖",
    "author": "Nora_GMI",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093222355665322439.mp4",
    "posterUrl": "/posters/x/2093222355665322439.jpg",
    "duration": 15.18585,
    "aspectRatio": "3:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "城市街景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093208454781423755",
    "title": "会场演讲与家中电脑的反差",
    "titleEn": "A Conference Speech and a Home Computer",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "演讲者与观众表情切换后，镜头转向家中的桌面电脑。",
    "summaryEn": "A speaker and audience reactions give way to a desktop computer at home.",
    "prompt": null,
    "sourceUrl": "https://x.com/GamerHashCom/status/2093208454781423755",
    "sourceLabel": "X 原帖",
    "author": "GamerHashCom",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093208454781423755.mp4",
    "posterUrl": "/posters/x/2093208454781423755.jpg",
    "duration": 10.24,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "写实电影感",
      "人物表演",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093200521360499158",
    "title": "概念手机的多配色网页广告",
    "titleEn": "A Multi-Color Concept Phone Landing Page",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "概念手机在绿色、银色和蓝色界面中切换，配合大字产品版式。",
    "summaryEn": "A concept phone cycles through green, silver and blue interfaces with oversized product typography.",
    "prompt": null,
    "sourceUrl": "https://x.com/mrdejie/status/2093200521360499158",
    "sourceLabel": "X 原帖",
    "author": "mrdejie",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093200521360499158.mp4",
    "posterUrl": "/posters/x/2093200521360499158.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "商业广告感",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093196455351902636",
    "title": "煎蛋从打散到上桌",
    "titleEn": "An Omelette from Mixing to Serving",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子打散鸡蛋、倒入平底锅，最后品尝完成的煎蛋。",
    "summaryEn": "A woman mixes eggs, pours them into a pan and tastes the finished omelette.",
    "prompt": "Create a 15-second ultra-photorealistic cinematic cooking video featuring the same young woman in a cozy modern kitchen. Keep her face, hairstyle, outfit, kitchen layout, lighting, and visual style consistent throughout.\n\nScene 1: The woman enters the kitchen carrying fresh eggs and smiles naturally at the camera.\n\nScene 2: Close-up of her cracking several eggs into a bowl. She whisks them smoothly with a fork, adding a little salt and pepper.\n\nScene 3: She pours the beaten eggs into a hot non-stick pan with a small amount of butter. Show the eggs spreading naturally and beginning to cook.\n\nScene 4: She gently folds and flips the eggs, creating a soft, fluffy golden omelette. Capture detailed textures, realistic steam, and sizzling sounds.\n\nScene 5: She places the freshly cooked omelette on a plate, sits at the kitchen table, takes a bite, and smiles naturally while enjoying the food.\n\nUse cinematic camera movements, realistic hand interactions, natural facial expressions, smooth transitions, warm kitchen lighting, shallow depth of field, detailed food textures, and premium commercial food-video quality. Add realistic cooking and eating sounds. No text, no logos, no distorted hands, no extra fingers, no flickering, no unnatural movements.",
    "sourceUrl": "https://x.com/CaliraVal/status/2093196455351902636",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093196455351902636.mp4",
    "posterUrl": "/posters/x/2093196455351902636.jpg",
    "duration": 15.232,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "美食",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2093196455351902636"
  },
  {
    "id": "x-2093182133548560776",
    "title": "双手拼装紫白色机器人",
    "titleEn": "Assembling a Purple-and-White Robot",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "双手把零件依次安装到机器人模型上，逐步呈现成品形态。",
    "summaryEn": "Hands fit parts onto a robot model, gradually revealing the completed figure.",
    "prompt": null,
    "sourceUrl": "https://x.com/kiyoshi_shin/status/2093182133548560776",
    "sourceLabel": "X 原帖",
    "author": "kiyoshi_shin",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093182133548560776.mp4",
    "posterUrl": "/posters/x/2093182133548560776.jpg",
    "duration": 15.168,
    "aspectRatio": "20:31",
    "resolution": "640p",
    "tags": [
      "综合展示",
      "照片级写实",
      "产品与广告"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093181779872288886",
    "title": "黑色手袋的开箱与细节展示",
    "titleEn": "Unboxing and Inspecting a Black Handbag",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "手袋从包装盒取出，在素色桌面上展示提手和包身。",
    "summaryEn": "A handbag is removed from its box and displayed on a plain surface, highlighting its handles and body.",
    "prompt": "FORMAT: 9:16 vertical, 15 seconds\nSTYLE: Ultra-realistic luxury UGC, high-end fashion campaign, cinematic studio lighting, sophisticated, minimal, premium, photorealistic 4K.\nPRODUCT CONSISTENCY — CRITICAL:\nUse the uploaded handbag images as the exact and only product reference. Preserve the handbag’s exact shape, proportions, black color, smooth material, long structured handles, side bow/tie details, hardware, stitching, opening, and CHARLES & KEITH logo. Do not redesign, resize, recolor, add/remove details, or change the construction. The same handbag must remain perfectly consistent throughout every shot.\n\n🎬 SHOT-BY-SHOT\n0–2s — UNBOXING HOOK\nClose-up UGC shot of a stylish woman opening a CHARLES & KEITH black box on a clean, elegant table. She lifts the lid and gently pulls back the white dust bag/tissue paper, revealing the exact black handbag inside. Natural hand movement, satisfying packaging sounds, warm soft lighting.\nOn-screen text: “Okay… this is even prettier in person.”\n\n2–4s — HERO REVEAL\nSmooth cinematic camera movement from left to right, revealing the handbag’s full silhouette and distinctive oversized handles. Subtle realistic shadow beneath the bag. Premium fashion-commercial lighting.\n\n4–6s — DETAIL SHOT\nMacro close-up of the CHARLES & KEITH logo and smooth black finish. Camera gently racks focus from the material to the logo. Highly detailed texture, realistic reflections, shallow depth of field.\n6–8s — SIDE DETAIL\nThree-quarter angle matching the uploaded reference image. Slowly orbit around the bag to showcase its structured shape, side construction and elegant tie details. Keep the product geometry identical to the reference.\n8–11s — UGC LIFESTYLE MOMENT\nA stylish woman wearing a minimalist black blazer and tailored trousers naturally picks up the handbag from the studio surface. Frame primarily focuses on the handbag and hands rather than her face. The long handles fall naturally as she lifts it.\n\n11–13s — CARRY SHOT\nClose-up tracking shot as she carries the handbag beside her. Soft studio-to-fashion-editorial transition, subtle movement, natural hand interaction, realistic material behavior.\n13–15s — FINAL HERO\nReturn to a clean front three-quarter product shot. The handbag is placed elegantly on the surface while the camera slowly pushes in. Soft luxury lighting, beautiful shadow, crisp product details.\n\nOn-screen text:\n“Timeless. Effortless. Yours.”\n🎥 CAMERA & QUALITY\n\nSmooth controlled camera movements, macro lens details, shallow depth of field, realistic reflections, soft diffused key light, subtle rim lighting, natural shadows, premium editorial composition, realistic hand movement, physically accurate materials, no artificial CGI appearance, no excessive transitions.\nAUDIO: Soft luxury fashion beat + subtle tissue-paper crinkle + gentle handbag placement sound. No distracting music.\n\nNEGATIVE PROMPT:\nNo product redesign, no altered proportions, no different handbag, no extra straps, no missing handles, no distorted logo, no warped geometry, no floating objects, no plastic-looking material, no excessive gloss, no shaky footage, no cartoon/CGI appearance, no unrealistic hands, no face distortion, no text glitches.",
    "sourceUrl": "https://x.com/codewithhajra/status/2093181779872288886",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093181779872288886.mp4",
    "posterUrl": "/posters/x/2093181779872288886.jpg",
    "duration": 14.848,
    "aspectRatio": "15:17",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "产品与广告",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2093181779872288886"
  },
  {
    "id": "x-2093179789242667193",
    "title": "拉各斯车流中的紧急行程",
    "titleEn": "An Urgent Drive through Lagos Traffic",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "年长司机握紧方向盘，身旁乘客表现出疼痛与紧张，车外交通拥挤。",
    "summaryEn": "An older driver grips the wheel while a passenger shows pain and anxiety amid heavy traffic.",
    "prompt": "older Nigerian lady driving quickly through Lagos traffic with a younger, pregnant lady in labor in the drivers seat anime style",
    "sourceUrl": "https://x.com/fathermerry/status/2093179789242667193",
    "sourceLabel": "X 原帖",
    "author": "fathermerry",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093179789242667193.mp4",
    "posterUrl": "/posters/x/2093179789242667193.jpg",
    "duration": 5.248,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "对白与口型",
      "城市街景"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/fathermerry/status/2093179789242667193"
  },
  {
    "id": "x-2093175677373845634",
    "title": "KUMO 汽水的平面动效广告",
    "titleEn": "A Flat Motion Ad for KUMO Soda",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "汽水罐剪影与橙蓝放射图形切换，最终显出品牌标识。",
    "summaryEn": "A can silhouette alternates with orange-and-blue radial graphics before revealing the brand mark.",
    "prompt": null,
    "sourceUrl": "https://x.com/onofumi_AI/status/2093175677373845634",
    "sourceLabel": "X 原帖",
    "author": "onofumi_AI",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093175677373845634.mp4",
    "posterUrl": "/posters/x/2093175677373845634.jpg",
    "duration": 15.168,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "插画与手绘",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093175490173677993",
    "title": "母女卡拉 OK 合唱",
    "titleEn": "A Mother-and-Daughter Karaoke Duet",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "两名女子拿麦克风并肩演唱，表情逐渐变得夸张热烈。",
    "summaryEn": "Two women sing side by side with microphones, becoming increasingly animated.",
    "prompt": null,
    "sourceUrl": "https://x.com/kan8888kawa/status/2093175490173677993",
    "sourceLabel": "X 原帖",
    "author": "kan8888kawa",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093175490173677993.mp4",
    "posterUrl": "/posters/x/2093175490173677993.jpg",
    "duration": 12.32,
    "aspectRatio": "17:23",
    "resolution": "544p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出",
      "日常生活"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093172023942078844",
    "title": "植物图鉴的水墨动态图形",
    "titleEn": "Botanical Studies in Ink and Motion",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "仙人掌与多肉植物在墨迹、线条和图鉴式版面中出现。",
    "summaryEn": "Cacti and succulents appear amid ink splashes, lines and botanical-study layouts.",
    "prompt": null,
    "sourceUrl": "https://x.com/gane0wav/status/2093172023942078844",
    "sourceLabel": "X 原帖",
    "author": "gane0wav",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093172023942078844.mp4",
    "posterUrl": "/posters/x/2093172023942078844.jpg",
    "duration": 15.16263,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "插画与手绘",
      "动态图形",
      "自然风景",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "nature",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093166617589113232",
    "title": "白青运动鞋的黑底产品片",
    "titleEn": "A White-and-Cyan Sneaker on Black",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "运动鞋从远景进入材质近景，最后以双鞋产品构图收尾。",
    "summaryEn": "A sneaker moves from a distant view into material close-ups, ending in a paired product composition.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2093166617589113232",
    "sourceLabel": "X 原帖",
    "author": "Naonekozamurai",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093166617589113232.mp4",
    "posterUrl": "/posters/x/2093166617589113232.jpg",
    "duration": 15.232,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "商业广告感",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093165840229355992",
    "title": "夜空中合体的超级机器人",
    "titleEn": "A Super Robot Assembles under Night Skies",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "飞行部件在夜空与火光上方聚合，形成发蓝光的机器人。",
    "summaryEn": "Flying components gather above night skies and fire into a blue-lit robot.",
    "prompt": null,
    "sourceUrl": "https://x.com/fantanglitchbox/status/2093165840229355992",
    "sourceLabel": "X 原帖",
    "author": "fantanglitchbox",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093165840229355992.mp4",
    "posterUrl": "/posters/x/2093165840229355992.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "科幻太空"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093149595861930044",
    "title": "从铅笔轮廓到彩色动漫人物",
    "titleEn": "From Pencil Outlines to a Colored Anime Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "手中的画笔补全角色线条与颜色，最终形成完整人物插画。",
    "summaryEn": "A drawing hand fills in a character's lines and colors until a complete illustration appears.",
    "prompt": null,
    "sourceUrl": "https://x.com/kiyoshi_shin/status/2093149595861930044",
    "sourceLabel": "X 原帖",
    "author": "kiyoshi_shin",
    "publishedAt": "2026-08-28",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093149595861930044.mp4",
    "posterUrl": "/posters/x/2093149595861930044.jpg",
    "duration": 15.168,
    "aspectRatio": "20:31",
    "resolution": "640p",
    "tags": [
      "动画",
      "插画与手绘",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093098428800934236",
    "title": "用构图图解引导公园人物镜头",
    "titleEn": "Guiding Park Shots with Camera Diagrams",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "公园人物动画与简化构图图解并列，展示镜头设计到生成结果的关系。",
    "summaryEn": "Park character animation appears beside simplified diagrams showing the relationship between shot planning and output.",
    "prompt": null,
    "sourceUrl": "https://x.com/Kashiko_AIart/status/2093098428800934236",
    "sourceLabel": "X 原帖",
    "author": "Kashiko_AIart",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093098428800934236.mp4",
    "posterUrl": "/posters/x/2093098428800934236.jpg",
    "duration": 22.549333,
    "aspectRatio": "540:607",
    "resolution": "1080p",
    "tags": [
      "本地生成与工作流",
      "动漫",
      "人物表演",
      "日常生活"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093093073077023109",
    "title": "上下分屏的写实与插画 MV",
    "titleEn": "A Split-Screen MV in Live Action and Illustration",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "同一表演以写实和插画风格上下并列，并穿插大幅字体。",
    "summaryEn": "The same performance appears in stacked live-action and illustrated panels with oversized type.",
    "prompt": null,
    "sourceUrl": "https://x.com/nakazakifam/status/2093093073077023109",
    "sourceLabel": "X 原帖",
    "author": "nakazakifam",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093093073077023109.mp4",
    "posterUrl": "/posters/x/2093093073077023109.jpg",
    "duration": 15.11619,
    "aspectRatio": "9:5",
    "resolution": "360p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "插画与手绘",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093090433085895156",
    "title": "运动剪影与球鞋的图形广告",
    "titleEn": "Athletic Silhouettes and Sneaker Graphics",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "运动鞋特写与运动人物剪影、彩色几何构图交替出现。",
    "summaryEn": "Sneaker close-ups alternate with athletic silhouettes and colorful geometric compositions.",
    "prompt": "添付したシューズ画像を、映像全体に登場するシューズの厳密なデザイン参照として使用する。参照画像から、白を基調としたアッパー、流線形のシアンのライン、シアンの靴紐、ヒールとアウトソールのシアンのアクセント、素材、パネル構造、ソール形状を正確に維持する。\nMiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによる、架空のスポーツメーカーのバスケットボールシューズCMを制作する。\n\n主人公は、全身が黒一色で描かれた一人の男性バスケットボール選手のシルエット。顔、髪、肌、衣服などの細部は描かず、プロ選手を思わせる、しなやかで洗練された体格として表現する。\nただし、主人公が履くシューズだけは黒いシルエットにせず、参照画像と同じ白とシアンのフルカラーで、素材やディテールまで鮮明に描写する。シューズを映像内で最も印象的な商品として扱い、黒い脚や鮮やかな背景から明確に浮かび上がらせる。\n\nバスケットボールは模様やラインのない黒一色の球体とする。登場人物は主人公一人だけとし、対戦相手や観客は登場させない。\n\n【0.0～3.5秒：SUMMER】\n映像は、シューズを大胆に強調した、床すれすれの超ローアングルから始まる。\n主人公は抽象的な幾何学形状のベンチに座っている。片方のシューズはすでに履いており、その白とシアンのシューズを画面手前に大きく配置する。極端な遠近感によって、手前のシューズが画面の大部分を占め、座っている黒い主人公が奥に見える、ファッション広告のような大胆な構図にする。\nもう片方の足を、参照画像と同じシューズへ滑り込ませる。主人公はシアンのヒールループを引き、最後に両手でシアンの靴紐を力強く締める。靴紐を左右へ引く瞬間を、音楽の最初の強いビートに正確に合わせる。\nカメラはトウボックス、シアンの靴紐、流線形のサイドライン、ヒール、ソールを短くシャープに切り替えながら映し、商品の質感と構造を明確に見せる。主人公の横には黒いバスケットボールが1個だけ置かれている。\n靴紐を締め終えた瞬間、背景に「SUMMER」の文字が、正確なスペルの大文字で大きく現れる。文字は画面の奥から勢いよく拡大しながら登場し、靴紐を引く動きとビートに同期する。\n主人公がシューズのつま先で床を一度踏み込むと、接地点からシアンの円形の波紋が広がる。そのまま主人公が黒いボールを手に取り、立ち上がる動きから、コート上での最初のドリブルへシームレスにつなげる。\n\n【3.5～9.5秒：SALE】\n場面は、明るく鮮やかなコーラル、マゼンタ、バイオレットを基調とした、抽象的でスタイリッシュなバスケットボールコートへ移行する。背景ではシアンを広い色面として使いすぎず、白とシアンのシューズが常に目立つ色のコントラストを維持する。\n主人公は低い姿勢で高速ドリブルを開始し、クロスオーバー、レッグスルー、ビハインド・ザ・バック、鋭いスピンムーブを、一続きの滑らかなコンビネーションとして披露する。\nすべてのプレーで、主人公の全身だけではなく、足運びとシューズの動きを優先的に見せる。床すれすれのカメラでシューズの横を高速追従し、踏み込み、方向転換、つま先のピボット、ソールが床を捉える瞬間をダイナミックに捉える。\nシューズが床を踏むたびにシアンの円形波紋が広がり、方向転換すると流線形のシアンのラインが進行方向へ走る。シアンの軌跡はシューズの動きを短時間だけ強調するモーショングラフィックとし、靴本体から炎やエネルギーが噴き出しているようには見せない。\nボールが床へ接触する瞬間、手から手へ移る瞬間、シューズが床を蹴る瞬間を、力強いヒップホップとエレクトロニック・ミュージックのキック、スネア、ベースへ正確に同期させる。\n高速ドリブルの開始と同時に、背景に「SALE」の文字が正確なスペルの大文字で大きく登場する。クロスオーバーや方向転換に合わせて、文字が左右へスライドし、拡大、縮小、3D回転、奥行き方向への移動を繰り返す。\n3D回転した「SALE」は必ず再び正面を向き、単語全体が明確に読める瞬間を作る。文字は主人公、ボール、シューズを隠さず、プレーの背後や周囲で動くグラフィック要素として使用する。\n俯瞰ショットでは、ドリブルの軌道と複雑なフットワークを見せる。床面を流れるシェイプやコートラインが、シューズの動きに合わせてリズミカルに変化する。\nローアングルショットでは、白とシアンのシューズを画面手前に大きく映し、黒い身体とボールがその上方で動く、強い遠近感のある構図にする。短いズームイン、ズームアウト、左右へのスライド、真横からの高速追従を加え、シューズのスピードと安定性を印象づける。\n\n【9.5～13.0秒：NOW】\n主人公は鋭いクロスオーバーからゴールへ向かって一気に加速する。真横からの高速追従ショットで、シューズが床を捉え、黒い脚が力強く前へ伸びる動きを見せる。\n主人公が加速を開始するタイミングで、背景に「NOW」の文字が正確なスペルの大文字で出現する。「NOW」は主人公と同じ方向へ高速で移動し、画面の奥から手前へ急接近する。\n最後の踏み切りではカメラを床すれすれに置き、参照画像と同じシューズのアウトソール、白いアッパー、シアンのラインを大きく捉える。足が床を蹴た瞬間、接地点から大きなシアンの波紋と幾何学的なラインが広がる。\n主人公は黒いボールを片手で保持して高く跳び、ダイナミックなワンハンドダンクを決める。大げさな曲芸ではなく、加速、踏み切り、跳躍、腕と脚の美しいラインを強調した、説得力のあるプロフェッショナルな動きにする。\nダンクの瞬間に「NOW」が正面を向いて明確に読める状態になり、すべてのシェイプとタイポグラフィーが強いビートに合わせて一度だけ大きく反応する。スローモーションは使用せず、スピードと疾走感を維持する。\n【13.0～15.0秒：シューズのヒーローショット】\n主人公は両足で軽やかに着地する。着地と同時にカメラが足元へ向かって素早く移動し、最後はシューズを画面いっぱいに捉えた超ローアングルの商品アップへ移行する。\n主人公は片足を半歩前へ出し、白とシアンのシューズをカメラに対して最も美しく見える斜め45度の角度で止める。両方のシューズを画面内に収め、手前の一足を主役として大きく見せる。黒い脚は画面上方へ伸び、シューズ以外の身体のディテールは描かない。\nカメラのピントをシアンのアウトソールから、流線形のサイドライン、シアンの靴紐へ滑らかに移動させる。\n背景には「SUMMER」「SALE」「NOW」の3語を大きく、読みやすく整列させる。3語はシューズより奥に配置し、白とシアンのシューズを覆い隠さない。背景の図形はビートの余韻に合わせて整列し、最後の約0.7秒は文字とシューズの動きを安定させる。\n最後の画面では、シューズの形状、配色、素材、セールメッセージのすべてを明確に確認できるようにする。主人公は大げさな決めポーズを取らず、シューズを履いた足元だけで、落ち着いた自信と力強さを表現する。\n\n【モーショングラフィック】\n背景では、長方形、円、三角形、太いライン、グリッド、コートラインなどのシェイプが、画面の上下左右や奥行き方向から勢いよくスライドインし、次のビートでスライドアウトする。\n複数のシェイプが異なる方向から重なり合い、分割、回転、拡大、縮小、色の変化を繰り返しながら、リズミカルに画面構成を作り替える。ただし、シェイプで人物、ボール、リング、シューズを完全に隠さず、プレーと商品の両方が常に明確に見えるようにする。\n映像内に登場する文字は「SUMMER」「SALE」「NOW」の3種類だけに限定する。すべて正確なスペルの大文字で表示し、文字の欠落、重複、誤字、別の単語への変化を起こさない。それ以外の文字、数字、企業名、ロゴ、商標、ウォーターマーク、字幕は一切表示しない。\n\n【一貫性と物理表現】\n左右一組のシューズは、映像全体を通して参照画像と同じデザイン、色、形状、素材を維持する。白いアッパーを黒く変色させず、シアンのラインや靴紐の位置を変えない。\nシューズを複製、消失、変形、左右不一致にしない。着用後は両方のシューズを正しく両足に固定し、足から外れたり、ソールが床へ沈んだり、シューズや靴紐が脚を貫通したりしない。走行、方向転換、踏み切り、着地では、ソールと床の接触に正しい重量感とグリップ感を持たせる。\n黒いボールは映像全体で必ず1個だけにする。完全な球形と大きさを維持し、複製、変形、消失、瞬間移動、色の変化を発生させない。手、床、リングとの接触を物理的に正しく描写する。\nリング、バックボード、ネットの位置と形状を安定させ、ダンクの前後で大きさや設置場所を変化させない。\n主人公の体格と手足の本数を15秒間一貫させ、身体の分裂、手足の変形、不自然な残像を発生させない。カットが変わっても、同じ主人公、同じシューズ、同じボール、同じコートとして自然につなげる。\n大胆な遠近感、シャープなモーショングラフィック、強い色彩コントラスト、正確なビートシンクを組み合わせる。人物やバスケットボールの技よりも、常に白とシアンのシューズが記憶に残る、スタイリッシュでファッション性の高いサマーセール用スポーツシューズCMに仕上げる。",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2093090433085895156",
    "sourceLabel": "X 原帖",
    "author": "Naonekozamurai",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093090433085895156.mp4",
    "posterUrl": "/posters/x/2093090433085895156.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "动态图形",
      "产品与广告",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2093090433085895156"
  },
  {
    "id": "x-2093090305289662901",
    "title": "银发人物与星云之间的变形",
    "titleEn": "A Silver-Haired Figure Morphs through Nebulae",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "银发人物在金属服饰、透明薄纱和旋转星云之间转换。",
    "summaryEn": "A silver-haired figure shifts between metallic clothing, translucent fabric and swirling nebulae.",
    "prompt": null,
    "sourceUrl": "https://x.com/junyait/status/2093090305289662901",
    "sourceLabel": "X 原帖",
    "author": "junyait",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093090305289662901.mp4",
    "posterUrl": "/posters/x/2093090305289662901.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "风格化与实验",
      "照片级写实",
      "奇幻",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "photoreal"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2093088917402874275",
    "title": "白色背景上的兽耳角色舞步",
    "titleEn": "An Animal-Eared Character Dances on White",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑色造型的兽耳角色在白色背景上连续改变舞姿。",
    "summaryEn": "An animal-eared character in black performs changing dance poses against white.",
    "prompt": null,
    "sourceUrl": "https://x.com/highsan_works/status/2093088917402874275",
    "sourceLabel": "X 原帖",
    "author": "highsan_works",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2093088917402874275.mp4",
    "posterUrl": "/posters/x/2093088917402874275.jpg",
    "duration": 10.208,
    "aspectRatio": "21:29",
    "resolution": "672p",
    "tags": [
      "舞蹈",
      "3D / CG",
      "人物表演"
    ],
    "category": "dance",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092968802283852088",
    "title": "红蓝气泡中的汽水罐广告",
    "titleEn": "A Soda Can among Red-and-Blue Bubbles",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝色汽水罐、红蓝色带和气泡汇入品牌收尾画面。",
    "summaryEn": "A blue can, red-and-blue ribbons and bubbles resolve into a branded closing frame.",
    "prompt": null,
    "sourceUrl": "https://x.com/abxxai/status/2092968802283852088",
    "sourceLabel": "X 原帖",
    "author": "abxxai",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092968802283852088.mp4",
    "posterUrl": "/posters/x/2092968802283852088.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "商业广告感",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092968204390027307",
    "title": "橙发赛博角色的霓虹 MV",
    "titleEn": "A Neon MV for an Orange-Haired Cyber Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "橙发动漫人物在黑白大字、粉色网格和圆环界面中登场。",
    "summaryEn": "An orange-haired anime character appears among black-and-white type, pink grids and circular interfaces.",
    "prompt": "15秒、アスペクト比16:9、MiniMax-H3使用。\n音楽と完全に同期した、近未来的なサイバーポップ・ミュージックビデオ。可愛いアニメキャラクターと、洗練されたデジタルグラフィック、ネオン、未来的なUI、ファッションMVのようなスタイリッシュな演出を融合する。\n全体の方向性は「CUTE × CYBER × POP × FASHION」。\nホログラフィック、ミニマル、シンプル、サイバー、デジタル。現代的でクリーン、シャープ、少しシュールな近未来デジタル空間として表現する。\nキャラクター\n入力画像の少女を全編の主人公として使用する。\n顔立ち、髪型、前髪、体型、衣装、ファッションシルエット、特徴的なディテールを入力画像に忠実に維持し、全カットで同一キャラクターとして認識できるようにする。\nキャラクターは可愛いアニメ的な魅力を持ちながら、演出はスタイリッシュでクール。幼すぎる表現、過度にあざとい表情、アイドル的な過剰な可愛さは避ける。\n表情は自然で、少しクール。ファッションモデルのような存在感を持たせる。\nカラーパレット\n基本色はブラックとホワイト。\nアクセントカラーは、入力画像のキャラクターを分析して決定する。\n髪、衣装、瞳、アクセサリーなどの色を観察し、キャラクターを最も美しく際立たせる色を1色だけ選択する。\n選択した1色をネオンカラーとして全編で統一して使用する。\n赤を固定色として使用しない。\nブラック、ホワイト、選択した1色の3色を基本パレットとし、余計な色を追加しない。\n① 振り返る\n最初のビートで画面が大胆に切り替わる。\n少女が正面を向いて立っており、ビートと同時にシャープに振り返る。\n振り返りに合わせて、背景の巨大なデジタルUIパネルとネオンラインが一斉にスライドする。\n髪と衣装が動きに合わせて自然に揺れる。\n一瞬だけポーズを止め、モデルのような強いシルエットを作る。\n② クローズアップ\n画面が高速で切り替わり、少女の顔へ一気にプッシュイン。\n少女がカメラをまっすぐ見つめ、片方の目を軽く細める。\n人差し指を口もとへ軽く添える。\nウインクを強調しすぎず、表情はクールで自然。\n顔の周囲に細いネオンライン、ターゲットマーク、UIインジケーター、短いデジタル文字列が瞬間的に出現する。\n③ 歩いてくる\n少女が画面奥からカメラへ向かって歩いてくる。\nランウェイを歩くような堂々とした歩き方。\nカメラは正面から追い、少女の歩行に合わせて背景の巨大なグラフィックパネルが左右に高速スライドする。\n床には薄い発光グリッド。\n少女の周囲を小さなUIパーツや幾何学的なラインが追従する。\nキャラクターを主役として明確に見せ、背景を過度に複雑にしない。\n④ ボディモーション＋ダッチアングル\nビートと同時に少女が身体を軽くひねり、肩と髪を一瞬だけ大きく動かす。\nその瞬間、カメラが大胆にダッチアングルへ傾く。\n背景のネオンパネル、UI、ラインもカメラの動きに合わせて傾き、画面全体が一瞬だけ大きく回転したようなダイナミックな印象になる。\nポーズはファッションMVのようにシャープで、幼くならない。\n⑤ お山座り\n最後は空間の中央に少女が一人で座っている。\n少女はお山座りで膝を抱え、両足を少し浮かせている。\n音楽のリズムに合わせて足先を小さく前後に動かす。\n表情は無表情に近い自然な表情で、カメラを静かに見つめる。\n周囲には巨大なネオンタイポグラフィと透明なデジタルUIが浮かび、少女の周囲をゆっくりと回る。\n可愛いポーズだが、あざとく見せない。\n「静かな可愛さ」と「未来的なクールさ」を同時に表現する。\nグラフィック\n画面全体に大胆なサイバーポップ・グラフィックを使用。\n巨大なタイポグラフィ、ネオンライン、透明なUIパネル、HUD、ターゲットマーク、デジタルフレーム、幾何学模様、スキャンライン、ピクセルノイズ、軽いグリッチを組み合わせる。\nUIは単なる背景ではなく、キャラクターの動きと同期して出現・移動・分解・再構成する。\n大きな文字と小さな文字を組み合わせ、画面に強いリズムを作る。\n画面を黒・白・選択したアクセントカラーの大胆なパネルに分割し、カットの切り替えごとにレイアウトを大きく変化させる。\nタイポグラフィ\n「MELT」\n「AGAIN」\n「END」\nこの3つの英単語を主要なタイポグラフィとして使用する。\n文字は画面に静止表示するのではなく、音楽のビートに合わせて、\n・高速スライド\n・拡大\n・縮小\n・マスクからのリビール\n・分解\n・再構成\n・画面外への飛び出し\n・一瞬のグリッチ\nなどのモーションを行う。\n特に「MiniMax」「H3」は映像の勢いを高めるように大胆に配置し、「Appletea」は最後のビートに合わせて明確に表示する。\nカメラ\nミディアムショットからクローズアップ、正面、ローアングル、ハイアングル、ダッチアングルへとテンポよく変化する。\nカメラの動きそのものも音楽のリズムとして扱う。\nプッシュイン、急速なパン、ズーム、アングル変更、短いカメラシェイクを適度に使用する。\nただし、カメラワークが過剰になりすぎてキャラクターが見えなくならないようにする。\n編集\n15秒の中で画面をテンポよく切り替える。\n各カットの切り替えと同時に、キャラクターが必ず一つの明確なモーションを行う。\n映像、カメラ、タイポグラフィ、UI、ネオンの動きを音楽のビートに同期させる。\n後半に向かって徐々に幻想的で未来的な空間へ変化させる。\n最後のビートで「END」を表示し、巨大なグラフィックとともにハードカット。\n余韻を長く残さず、最後は一瞬で潔く終了する。\n完成映像は、可愛いアニメキャラクターを中心にした、近未来的でスタイリッシュなサイバーポップMV。\n「可愛い」だけではなく、「クール」「洗練」「未来的」「ファッション性」「少しシュール」という印象を強く持たせる。 BGM:日本のハイパーポップ、テクノポップ、テクノロック、インダストリアル",
    "sourceUrl": "https://x.com/applete77191758/status/2092968204390027307",
    "sourceLabel": "X 原帖",
    "author": "applete77191758",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092968204390027307.mp4",
    "posterUrl": "/posters/x/2092968204390027307.jpg",
    "duration": 15.296,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "人物表演",
      "科幻太空"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/applete77191758/status/2092968204390027307"
  },
  {
    "id": "x-2092929895768674702",
    "title": "秋林剑斗的双模型对照",
    "titleEn": "Two Models Compare an Autumn-Forest Sword Fight",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "上下画面标明 Seedance 与 H3，以同一武侠剑斗场景展示不同结果。",
    "summaryEn": "Labeled Seedance and H3 panels compare results for the same wuxia sword-fight scene.",
    "prompt": "A cinematic wuxia sword fight between a man in black and a woman in red in a golden autumn forest",
    "sourceUrl": "https://x.com/ridark_eth/status/2092929895768674702",
    "sourceLabel": "X 原帖",
    "author": "ridark_eth",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092929895768674702.mp4",
    "posterUrl": "/posters/x/2092929895768674702.jpg",
    "duration": 30.046621,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "写实电影感",
      "动作打斗",
      "自然风景"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ridark_eth/status/2092929895768674702"
  },
  {
    "id": "x-2092922299653783597",
    "title": "赛车场镜头图解与生成对照",
    "titleEn": "Race-Camera Diagrams Compared with Generated Video",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "赛车场画面与路径图并列，片中分别标出 Seedance 和 H3 的结果。",
    "summaryEn": "Race footage sits beside path diagrams, with labeled Seedance and H3 results.",
    "prompt": null,
    "sourceUrl": "https://x.com/Kashiko_AIart/status/2092922299653783597",
    "sourceLabel": "X 原帖",
    "author": "Kashiko_AIart",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092922299653783597.mp4",
    "posterUrl": "/posters/x/2092922299653783597.jpg",
    "duration": 30.058667,
    "aspectRatio": "8:9",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "3D / CG",
      "产品与广告"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092835189215138167",
    "title": "沿客厅缓慢推进的室内导览",
    "titleEn": "A Slow Push through a Living Room",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头沿明亮客厅缓慢前移，保持沙发、窗户和墙面的空间关系。",
    "summaryEn": "The camera moves slowly through a bright living room, preserving the arrangement of sofa, windows and walls.",
    "prompt": "A slow, smooth interior walkthrough shot based on the input image.\nThe camera gently moves forward through the living room for 5 seconds, with a very slight pan toward the TV wall on the left.\nKeep the movement subtle, stable, and natural, like a real handheld-gimbal or slow dolly shot in an architectural interior video.\nPreserve the exact room layout, furniture placement, wall positions, windows, and overall composition from the source image.\nDo not redesign the space.\nMaintain realistic proportions and stable geometry.\nThe motion should remain within the visible area of the room and should not move too far forward.\nShow only a gentle forward drift and a slight look toward the TV side.\nPhotorealistic residential interior, realistic materials, natural daylight, soft shadows, calm atmosphere.\n\nAvoid distortion, morphing, flickering, warping, extra objects appearing, or drastic camera movement.",
    "sourceUrl": "https://x.com/egachan_ceao/status/2092835189215138167",
    "sourceLabel": "X 原帖",
    "author": "egachan_ceao",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092835189215138167.mp4",
    "posterUrl": "/posters/x/2092835189215138167.jpg",
    "duration": 5.226667,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/egachan_ceao/status/2092835189215138167"
  },
  {
    "id": "x-2092813353928581347",
    "title": "朱色刀光中的角色轨迹",
    "titleEn": "A Character Trail in Vermilion Sword Light",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红黑剑客在白色背景中挥刀，红色线条与片名字效连接动作。",
    "summaryEn": "A red-and-black swordsman swings across white backgrounds, joined by red lines and title graphics.",
    "prompt": "【REFERENCE】\nimage1 は主人公の三面図参照画像。正面・側面・背面を同一人物として扱い、顔立ち、性別、年齢感、髪型、髪色、体型、頭身、全身比率、赤と黒の衣装、靴・装飾、刀・柄・鍔・鞘、高品質な2Dアニメ表現を固定する。主人公は1人のみ。三面図の白背景、3体並び、文字、ラベル、紙面レイアウトは使用せず、背景、ポーズ、アクション、カメラ、カット構成、MG、タイポグラフィは動画側で新規構成する。\n\n【CONDITION】\n15秒、1:1スクエア。高品質な2Dアニメーション。\n世界は白・黒・グレーのモノクロ。主人公の衣装の赤だけを唯一の色として残し、肌、髪、刀、装飾、背景、空間は無彩色で統一する。Use one vivid saturated red; not orange-red, not another accent color.\n刀の動きから赤の軌跡や飛沫を発生させ、既存要素を毎回作り直さず、red trace → parallel lines → panels / ribbon → title baseline → “VERMILION TRACE” と状態変換する。最終的に赤のタイポグラフィ「VERMILION TRACE」を構築し、背景世界は最後までモノクロを維持する。\n人物は独立した一閃を反復せず、one continuous flowing sword phrase とする。前の follow-through、刀の位置、身体の回転、足運び、重心移動を次へ引き継ぐ。FLOW → KEY POSE → CONTINUE → KEY POSE → CONTINUE → FINAL KEY POSE。KEY POSE は brief deceleration through the pose, never a full stop。髪先、衣装端、赤い軌跡には slight delayed follow-through を持たせる。\nカメラは剣舞の流れを優先し、主要区間では身体ラインと刀の軌道が読める距離を保つ。Close-up / crop は短いアクセントとして使い、すぐ全身の流れへ戻る。\n\n【SHOT FLOW】\n0.0〜2.5秒\n白黒の抽象空間。赤の衣装を着た主人公を全身が読める3/4構図で捉える。最初の斜め斬りをカメラが斬撃方向へ短く追従。剣先から細い赤の軌跡と飛沫が発生し、斬撃音と短いインク飛散音に同期して画面を斜めに横切る。最初の赤は消さず次の基準線として保持する。斬撃後は止まらず、follow-through の中で最初のKEY POSEを一瞬読ませ、そのまま横方向の動きへつなぐ。BGMはここから開始。\n\n2.5〜5.2秒\n残った赤の軌跡を維持し、前の follow-through から横薙ぎへ移行する。刀を戻さず前の終端位置からそのまま流し、新しい赤の軌跡を最初の線と交差させる。交点から赤の細線、長方形の面、ハーフトーンと黒い平面が画面を再構成し、既存ラインの一部を five parallel lines → graphic panels へ変形させる。人物は第二のKEY POSEを短く経由してCONTINUEし、アクションとグラフィックを同方向へ流す。赤の断片は文字のストロークを思わせる形になるが、まだ読める単語にはしない。MG側では短いHOLDの後にRELEASE。BGMの密度を一段上げる。\n\n5.2〜8.3秒\n既存の赤ラインで、目、刀を握る手、刀身、赤の衣装を短く切り取るエディトリアルなクロップへ接続する。短く鋭い同期音を入れ、別人物や分身は作らない。ラインが各クロップをつなぎ、既存パネルの一部を ribbon-like forms へ変形。クロップは画面上で重ねず、最後のクロップは全身へ戻る直前に完全に消す。Foreground crop elements do not persist into the full-body shot. 短いクロップ連続の後に一瞬CATCHし、次の動きでRELEASEして全身へ戻る。横薙ぎから続く身体の回転を止めず、大きな円弧で刀を回す flowing motion を見せ、その弧が既存の線、面、ハーフトーンを引き寄せてタイポグラフィの骨格へ再配置する。接続時に乾いた組み上げ音を重ねる。KEY POSEは一瞬の減速のみで完全静止しない。\n\n8.3〜11.6秒\nグラフィック密度と音楽を一瞬落とすが、身体は止めない。前の円弧を保ったまま重心を移し、踏み込みを深め、最大の一閃へつなぐ。最も強い斬撃音とBGMのピークを同期。剣先の赤が広い軌跡となり、蓄積した赤の線、飛沫、面、ハーフトーン、文字断片を斬撃方向へ整列させる。赤の断片が接続して「VERMILION TRACE」が完成。既存の ribbon を title baseline へ変形させ、そこからタイトルを一つの graphic unit として組み上げる。完成時にごく短い overshoot、正位置で TITLE LOCK。タイポ完成時に明確なインパクト音を入れる。\n\n11.6〜15.0秒\n最大の一閃の follow-through から最終ポーズへ収まり、ここで初めて人物とカメラを明確にLOCKする。赤の衣装の主人公と完成した赤の「VERMILION TRACE」を主役にしたポスター的最終構図へロックし、背景は白・黒・グレーを維持する。赤の線、面、ハーフトーンだけを最終デザインへ残し、一時的なクロップ断片や不要なパネル／ラインは消す。大きな動きは止め、髪先、衣装端、わずかな赤粒子とハーフトーンだけに微細な動きを残して終了する。\n\n【SOUND】\nBGMあり。効果音あり。刀アクション、赤エフェクト、MG、タイポグラフィの変化と同期し、剣舞の action phrase を各斬撃で途切れさせない。MG / editing の HOLD / CATCH / RELEASE / LOCK では音の密度も変える。各効果音は対応する映像イベントへ直接同期。ボーカル、セリフなし。\n\n【NEGATIVE】\n実写、フォトリアル、3D、CGIにしない。主人公を複数化しない。分身、別身体、複数の頭、浮遊する顔、余分な手足を出さない。参照三面図の3体並び、白背景、紙面レイアウト、ラベル、枠、資料シート表現を持ち込まない。顔、髪型、体型、衣装、赤の配置、刀のデザインを変更しない。赤以外の色を追加せず、背景世界は最後まで白・黒・グレー。赤の飛沫を血液として扱わない。「VERMILION TRACE」以外の読める文字、字幕、ロゴ、透かしを出さない。余分な刀、武器、人物、建物、群衆を追加しない。MGやタイポグラフィを主人公のアクションと無関係に発生させない。主要な赤の変化は必ず刀の動きを起点とする。各SHOTで別のMGをゼロから作らず、既存要素を次の状態へ変形させる。「VERMILION TRACE」を独立したテキストオーバーレイとして突然表示しない。各アクションを独立した一閃として扱わず、刀や身体を毎回リセットしない。長い完全停止を繰り返さず、優雅な全身運動が読めないほど寄りすぎない。",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2092813353928581347",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-08-27",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092813353928581347.mp4",
    "posterUrl": "/posters/x/2092813353928581347.jpg",
    "duration": 15.13941,
    "aspectRatio": "1:1",
    "resolution": "1080p",
    "tags": [
      "动作特效",
      "动漫",
      "动态图形",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2092813353928581347"
  },
  {
    "id": "x-2092732137145536736",
    "title": "丝带体操与数字字效",
    "titleEn": "Ribbon Gymnastics and Number Typography",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑色体操人物剪影挥动金色丝带，与大幅数字和感谢文字组合。",
    "summaryEn": "A black gymnast silhouette swings a gold ribbon among oversized numbers and thank-you text.",
    "prompt": "MiniMax H3を使用して、15秒・2Kのスタイリッシュなモーショングラフィック＆キネティック・タイポグラフィーによる新体操リボン競技の動画を制作する。\n\n主人公は、全身が黒一色で描かれた一人の女性のシルエット。顔、髪、衣装などの細部は描かず、新体操選手らしい、しなやかで洗練された身体のラインを持つ美しいシルエットとして表現する。人物の形状、体格、手足の本数は15秒間一貫させる。背景には明るく鮮やかな色を使用し、黒い人物との明確なコントラストを維持する。\n主人公は一本の細いスティックを持ち、そこにつながった一本の長いゴールドのリボンを操る。リボンは鮮やかなゴールド一色を維持し、人物や背景に埋もれないよう、常に輪郭が明確に見えるようにする。\n\n主人公は、プロの新体操選手が披露するような、優雅でダイナミックなリボン演技を行う。大きな円、連続する螺旋、細かなスネーク、8の字、身体を包み込むような旋回を滑らかにつなげる。さらに、ピボットターン、美しく脚を伸ばしたリープ、背中を反らせたしなやかなポーズ、リボンの投げ上げと正確なキャッチを組み合わせる。単に技を並べるのではなく、すべての動きが一続きの優雅な振り付けとして流れるようにする。\n\nリボンの軌道には自然な慣性と遅れを持たせる。腕の動きに少し遅れてリボンが空間を流れ、美しい曲線と立体的な軌跡を描く。リボンは途中で切れたり、複数に分裂したり、突然消えたり、人物の身体を不自然に貫通したりしない。スティックとリボンの接続を維持し、投げ技の際も同じ一本のスティックとリボンが回転し、主人公の手へ正確に戻る。\n\n演技は、洗練されたエレクトロニック・ミュージックのビートに正確にシンクロさせる。足の踏み込み、ターンの開始、リープの踏み切りと着地、スティックの振り、リボンの方向転換、投げ上げとキャッチを、キック、スネア、ベース、アクセント音に合わせる。リボンが大きな円を描くと背景にも円形の波紋が広がり、スネークを描くと波状のラインが走り、ターンすると背景の図形も連動して回転する。\n\n背景では、円、長方形、三角形、曲線、太いライン、グリッドなどのシェイプが、画面の上下左右や奥行き方向から勢いよくスライドインし、次のビートでスライドアウトする。複数のシェイプが分割、回転、拡大、縮小、色の変化を繰り返しながら、リボンの曲線に呼応して画面構成をリズミカルに作り替える。ただし、シェイプで人物やリボンを完全に隠さず、演技の動きは常に明確に見えるようにする。\n\n背景には「6100」「Followers」「Thank You」という文字を、表記を一切変更せずにキネティック・タイポグラフィーとして登場させる。「6100」は巨大な数字として強く印象づけ、「Followers」は横方向へ滑らかにスライドインさせ、「Thank You」はリボンの大きな円運動に合わせて優雅に現れる。\n\n文字は登場後も音楽のビートに合わせて脈打つように拡大・縮小し、前後左右への3D回転、奥行き方向への移動、軽い伸縮、色の切り替えを続ける。3D回転した文字は必ず再び正面を向き、明確に読める状態へ戻る。リボンは文字の周囲を美しく旋回してフレームのように囲むが、文字を長時間覆い隠さない。\n\n数字や文字の欠落、重複、誤字、別の文字への変化を起こさず、「6100」「Followers」「Thank You」以外の文字は表示しない。「Followers」は大文字と小文字の組み合わせを正確に維持し、「Thank You」は2単語の間に正しいスペースを入れる。実在する企業名、ロゴ、ウォーターマーク、字幕は表示しない。\n\nカメラは基本的に女性の全身とリボンの軌道を正面から捉えながら、音楽の強いビートや技のタイミングに合わせてアングルを大胆に切り替える。斜め上から見下ろす俯瞰ショットでは、床面に広がるリボンの螺旋、円、8の字と、背景を流れるシェイプを印象的に見せる。斜め下から見上げるローアングルショットでは、リープの高さ、身体の美しいライン、頭上を舞うリボンの立体的な軌道を強調する。\n\nさらに、人物の周囲を滑らかに回り込むカメラ移動、リボンの軌道を追うパン、短いズームインとズームアウト、左右へのスライドを加える。カットはターン、リープ、キャッチなどの動きと音楽のビートに合わせて切り替える。アングルが変わっても、同一の女性、同一のスティック、同一の一本のリボン、動作の方向と連続性を維持する。可能な限り、女性の全身とリボンの主要な軌道を同時に画面内へ収める。\n\n映像の前半では「6100」を大きく見せながら螺旋とスネークを披露し、中盤では「Followers」の周囲でターンとリープを行う。終盤ではリボンを高く投げ上げ、主人公が美しく旋回した後、落下してくるスティックを正確にキャッチする。最後に「6100」「Followers」「Thank You」が読みやすく整列し、ゴールドのリボンが3つの文字を大きな曲線で優雅に囲む。主人公は身体のラインを美しく伸ばした、上品で自然なフィニッシュを見せる。\n\n大胆でありながら上品な配色、シャープなグラフィック、正確なビートシンク、リボンの優雅な曲線を組み合わせ、感謝と祝福が伝わる、華やかでスタイリッシュな映像に仕上げる。SNSで拡散されそうな、センスが良く、中毒性があり、何度も見返したくなる高品質な動画にする。",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2092732137145536736",
    "sourceLabel": "X 原帖",
    "author": "Naonekozamurai",
    "publishedAt": "2026-08-26",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092732137145536736.mp4",
    "posterUrl": "/posters/x/2092732137145536736.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "舞蹈",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2092732137145536736"
  },
  {
    "id": "x-2092528740475785520",
    "title": "四套时装切换的概念应用界面",
    "titleEn": "A Concept Fashion App with Four Looks",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "竖屏界面在多套时装与大字排版之间切换，展示概念产品动效。",
    "summaryEn": "A vertical interface switches between fashion looks and oversized type to demonstrate a concept app.",
    "prompt": null,
    "sourceUrl": "https://x.com/onofumi_AI/status/2092528740475785520",
    "sourceLabel": "X 原帖",
    "author": "onofumi_AI",
    "publishedAt": "2026-08-26",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092528740475785520.mp4",
    "posterUrl": "/posters/x/2092528740475785520.jpg",
    "duration": 15.167318,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "商业广告感",
      "时尚",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092480148004368480",
    "title": "从街市抱回一颗西瓜",
    "titleEn": "Bringing Home a Watermelon",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子从街区抱回西瓜，回到厨房后品尝切好的果肉。",
    "summaryEn": "A woman carries a watermelon home through the neighborhood and tastes a slice in the kitchen.",
    "prompt": "Create a 15-second ultra-photorealistic cinematic summer lifestyle video following the same young woman throughout the entire sequence. She has long brown hair and wears a light blue blouse with beige pants.\n\nScene 1: She walks through a peaceful residential neighborhood on a bright summer day, natural sunlight, realistic walking motion, subtle breeze.\n\nScene 2: She arrives at a local fruit market and carefully chooses a large fresh watermelon among colorful seasonal fruits.\n\nScene 3: She carries the watermelon home while walking naturally through the neighborhood.\n\nScene 4: In a bright modern kitchen, she places the watermelon beside the sink, washes it, then cuts it open with a kitchen knife. Show realistic hand movement, natural cutting physics, and detailed watermelon texture.\n\nScene 5: She picks up a fresh slice of juicy red watermelon and takes a satisfying first bite near the window.\n\nMaintain identical facial features, hairstyle, clothing, watermelon appearance, environment continuity, lighting, season, and time of day across every shot. Natural human movement, realistic skin texture, cinematic depth of field, soft summer lighting, authentic everyday atmosphere, smooth transitions, no sudden character changes, no distorted hands, no extra fingers, no text, no watermark.",
    "sourceUrl": "https://x.com/DaniaSafvi/status/2092480148004368480",
    "sourceLabel": "X 原帖",
    "author": "DaniaSafvi",
    "publishedAt": "2026-08-26",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092480148004368480.mp4",
    "posterUrl": "/posters/x/2092480148004368480.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "美食",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/DaniaSafvi/status/2092480148004368480"
  },
  {
    "id": "x-2092469535844733357",
    "title": "乡间列车与日式庭院旅途",
    "titleEn": "A Rural Train and a Japanese Garden Visit",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子乘坐乡间列车，随后坐在传统建筑前的庭院中。",
    "summaryEn": "A woman rides a rural train and later sits in a garden before a traditional building.",
    "prompt": "A tranquil 15 second cinematic vlog shot in 8k on a 35mm lens at f/1.8. A young East Asian woman with long brown hair, wearing a white shirt and light blue scarf, rides a vintage teal bus through lush Japanese countryside. She steps off into the crisp air, sits on the wooden veranda of a traditional house, dips her bare feet into a clear running stream, and eats cold somen noodles from a bowl with chopsticks. Natural sun dappled lighting, green rice paddies, swaying wind chimes, soft bokeh, serene atmosphere, with clear English lip sync and dialogue.\nSubject and Environment Specs\nSubject: Young East Asian female, long brown hair, natural makeup, dressed in a casual white t shirt and light blue scarf.\nEnvironment: Rural Japanese countryside, vintage teal bus, expansive green rice fields, traditional wooden porch engawa, clear stream, glass wind chime furin.\nScene Breakdown and English Dialogue\nScene 1: The Countryside Bus Ride 0:00 to 0:03\nVisuals: Medium shot of the woman looking out a vintage bus window at green fields, followed by an aerial tracking shot of the teal bus driving along a rural road.\nScene 2: Arrival 0:04 to 0:05\nVisuals: Close up of her stepping off the bus into the open countryside, closing her eyes and breathing in the fresh air.\nEnglish Lip Sync: \"So peaceful...\"\nScene 3: Noodles by the Stream 0:06 to 0:11\nVisuals: Wide shot sitting on a wooden porch, dangling bare feet into a cool, clear stream while eating cold somen noodles from a black bowl with wooden chopsticks.\nEnglish Lip Sync: \"The air tastes so fresh here... these cold noodles are perfect.\"\nScene 4: Wind Chime and Farewell 0:12 to 0:14\nVisuals: A glass wind chime sways gently under the eaves. Cuts to a close up of her looking directly at the camera with a warm, relaxed smile in golden sunlight.\nEnglish Lip Sync: \"Truly delicious.\"",
    "sourceUrl": "https://x.com/harboriis/status/2092469535844733357",
    "sourceLabel": "X 原帖",
    "author": "harboriis",
    "publishedAt": "2026-08-26",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092469535844733357.mp4",
    "posterUrl": "/posters/x/2092469535844733357.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "自然风景",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/harboriis/status/2092469535844733357"
  },
  {
    "id": "x-2092464771874664617",
    "title": "度假村泳池旁的夏日自拍",
    "titleEn": "A Summer Selfie beside a Resort Pool",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子沿度假村步道走到泳池边，对镜头展示周围环境。",
    "summaryEn": "A woman walks through a resort to its pool and presents the surroundings to the camera.",
    "prompt": "Create a realistic cinematic summer resort video featuring the same adult female character throughout the entire video. Maintain strong facial consistency, including her eyes, nose, lips, face shape, hairstyle, and hair color in every shot.\n\nWardrobe: She wears a stylish, modest summer resort outfit: a loose white linen shirt over a comfortable knee-length summer dress, with clean casual sandals. No bikini, lingerie, revealing clothing, cleavage, transparent fabric, or sexualized styling.\n\nScene: A beautiful luxury tropical resort with a large turquoise swimming pool, water slides, palm trees, bright daylight, and a relaxing vacation atmosphere.\n\nAction: She walks around the resort, smiles naturally at the camera, enjoys the poolside environment, looks at the water slides, and casually explores the resort. Keep movements natural, cheerful, and family-friendly.\n\nCamera: Smooth cinematic tracking shots, medium shots, wide establishing shots, natural handheld movement, realistic depth of field, high-detail skin and environment, soft natural sunlight.\n\nVoice-over: Natural female English voice, friendly and energetic travel-vlogger tone. Clear pronunciation and realistic lip-sync.\n\nEnglish dialogue:\n“Hey everyone! Welcome to this beautiful summer resort. The atmosphere here is amazing, the pool looks incredible, and there are so many fun things to explore. It’s the perfect place to relax, enjoy the sunshine, and have a great vacation!”\n\nNegative prompt: bikini, lingerie, revealing clothes, cleavage, transparent clothing, nudity, sexualized poses, seductive behavior, exaggerated body proportions, body-focused camera shots, low-quality face, face distortion, inconsistent character, bad lip-sync, unnatural movements.",
    "sourceUrl": "https://x.com/CaliraVal/status/2092464771874664617",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-26",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092464771874664617.mp4",
    "posterUrl": "/posters/x/2092464771874664617.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "自然风景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2092464771874664617"
  },
  {
    "id": "x-2092457346526241208",
    "title": "霓虹字效中的黑色时装片",
    "titleEn": "Black Fashion among Neon Typography",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物近景与空间大字、眼部反射和粉色光线交替出现。",
    "summaryEn": "Portraits alternate with spatial typography, eye reflections and pink light.",
    "prompt": "High-Speed K-pop Fashion MV, luxury fashion advertising + Material Kinetic Typography + Experimental Fashion Film + 3D Spatial Typography. Strictly use @Image1 as the only character reference throughout the entire film. Core visual concept: the same girl + giant physical English typography + constantly changing materials + typography moving through foreground, midground, and background around the character + high-speed camera movement + aggressive beat synchronization. This is NOT a conventional MV where a person simply stands in front of text. All typography must physically exist inside the same 3D space as the character. Typography can generate behind her, wrap around her shoulders into the foreground, then rush toward the camera; it can also fly in from the extreme foreground, pass along both sides of her body, and extend behind her. She can stand inside giant typography, walk between oversized letters, and become fully surrounded by text. All typography must have realistic occlusion, spatial depth, perspective, shadows, reflections, and interactive lighting.【CHARACTER LOCK｜HIGHEST PRIORITY】Strictly use the girl from @Image1. Maintain exactly the same face shape, facial-feature proportions, eye shape, eyebrow-to-eye proportions, nose, lips, jawline, skin tone, age, body proportions, long straight black hair, and wispy bangs throughout the entire film. Young East Asian woman with a refined small oval face, dark eyes, petite natural nose, soft pink lips, and fair natural skin. Realistic skin texture, visible natural pores, realistic eyelashes, and individually defined black hair strands. Keep the clean K-pop makeup from the reference: natural base makeup, subtle pink eye makeup, thin eyeliner, natural lashes, and glossy pink lips. No face swapping, no Westernization, no facial drift, no face-shape changes, no age changes, no skin-tone changes, no hair-color changes, and never lose the bangs. Long black hair and wispy bangs are essential identity features and must remain consistent at all times.【WARDROBE】Use the original black edgy outfit from @Image1 as the visual foundation. Keep the overall palette BLACK + SILVER + DEEP RED. LOOK01: black fitted sleeveless crop top, black choker, silver necklace, silver earrings, silver mirrored low-rise mini skirt, black knee-high boots. LOOK02: black sleeveless top and choker, used for the Beauty Close-Up. LOOK03: black oversized cropped biker leather jacket + black tank top + black leather mini skirt + silver waist chain. LOOK04: deep burgundy modern reinterpreted short qipao, black piping, minimalist silver frog-button details, black knee-high boots. LOOK05: black cropped top + silver metallic mini skirt + black choker. LOOK06: black modern Chinese-inspired structured cropped top + black layered mini skirt + silver metallic accessories. LOOK07: distressed gray-blue denim cropped top + charcoal-black asymmetric low-rise mini skirt + silver waist chain. LOOK08: Final Dark Idol Look, black asymmetric cropped top + black ultra-short layered skirt + silver waist chain + black choker + black knee-high boots. Do not radically change the fashion identity between scenes; the character must remain visually unified throughout.【MUSIC】145–155 BPM, high-energy, sexy, edgy K-pop Fashion Beat. Music style: K-pop Girl Group + Electro Hip-Hop + Jersey Club + Fashion Runway Beat + Y2K Club Beat. Start immediately with an aggressive rhythm at 0 seconds, no slow intro. Strong Kick, Heavy Bass, crisp Snare, rapid Hi-Hat, Electronic Percussion, short Glitch Hits. Rhythm pattern similar to BOOM — CLAP — BOOM BOOM — CLAP. One obvious heavy beat every 1–1.5 seconds. Every outfit transition, typography appearance, text deformation, camera push, rotation, and material explosion must hit precisely on the beat. No lyrics, no vocals, no dialogue, no lip-syncing. Add material-specific sound effects including Smoke Whoosh, Neon Buzz, Metal Impact, Electric Crack, Plant Growth, Paper Flip, Fabric Whip, Wind Rush, and Bass Impact.【OVERALL COLOR PALETTE】Restrict colors to BLACK, WHITE, HOT PINK, DEEP PINK, BURGUNDY, SILVER. Black is the dominant spatial base, Hot Pink provides energy and highlights, silver defines metallic materials, white provides typography contrast, and burgundy adds a sophisticated fashion tone. No rainbow colors. No blue-purple cyberpunk palette. Overall aesthetic: K-pop Girl Group Comeback Concept Film × High Fashion Campaign × Experimental Typography MV.【SCENE 01｜0–1.6s｜FREE YOUR MIND｜SMOKE TYPOGRAPHY】Enter the climax immediately on the first beat. The girl wears a black fitted sleeveless crop top, black choker, silver necklace, silver mirrored mini skirt, and black knee-high boots. Full-body shot, standing exactly in the center of a vast black space with a subtly reflective floor. She faces the camera and suddenly reaches one hand toward the lens. BOOM! A massive burst of white + extremely pale pink smoke erupts from her palm. The smoke first forms a gigantic “FREE” behind her, approximately twice her height. On the next beat, smoke curls out from behind her left shoulder, passes into the space in front of her body, and forms “YOUR”. Finally, the smoke continues toward the camera and forms a gigantic “MIND” in the extreme foreground. Establish three clearly separated spatial layers: FREE in the background, character + YOUR in the midground, MIND in the foreground. She is genuinely surrounded by the typography. Smoke-letter edges continuously roll, expand, contract, and breathe. High-Speed Push-in from a frontal full-body view, rapidly approaching her, then fly directly into the interior of the “MIND” letters until the dark interior fills the entire frame for transition.【SCENE 02｜1.6–3.1s｜LIGHT YOUR OWN WAY｜NEON FACE CLOSE-UP】Instantly cut from darkness into an Extreme Beauty Close-Up. This is the most important face-locking shot in the entire film. Preserve exactly the same long black hair, wispy bangs, eyes, nose, lips, face shape, silver earrings, and choker from @Image1. Her face is slightly turned to the side. Her eyes first look to the right, then shift directly toward the camera. One finger rises from the bottom of the frame and a bright Hot Pink point of light appears at the fingertip. On the next beat, a pink Neon Line shoots rapidly from the fingertip. The neon first sweeps across the extreme foreground, passes beside her cheek, curves beneath her chin, crosses in front of her shoulder, then wraps behind her head to create a giant S-shaped 3D light trail. The trail gradually forms “LIGHT YOUR OWN WAY”. LIGHT occupies the foreground, YOUR wraps around her shoulder, OWN appears beside her, and WAY extends into the background. The neon realistically illuminates her eyes, cheekbones, nose bridge, and the edges of her black hair. The camera follows the neon trajectory at high speed, beginning at the front of her face, orbiting approximately 180 degrees around her, then rapidly Pull-out until the complete sentence becomes clearly readable once.【SCENE 03｜3.1–4.7s｜BECOME WHO YOU CHOOSE｜LIQUID CHROME】BOOM! Instant wardrobe transition into the black biker look: black oversized cropped leather jacket, black tank top, black leather mini skirt, silver waist chain. Cut to Ultra-Wide + Extreme Low Angle. Large volumes of silver liquid metal rise from the floor like mercury, rapidly stretching upward into gigantic Chrome Typography two to three times taller than the character: “BECOME WHO YOU CHOOSE”. The typography must not form a flat wall. It behaves like monumental typographic architecture. Some letters stand in front of her, others flank her left and right sides, and others rise behind her. She physically stands between the giant letters and takes one step forward through a gap between them. She raises her right hand; moving her palm left stretches the Chrome letters in the same direction, moving right makes them distort like liquid mercury. On the next beat she suddenly clenches her fist. CLAP! All letters instantly solidify, clearly displaying “BECOME WHO YOU CHOOSE”. The camera races closely along the reflective Chrome letter surfaces, briefly revealing her reflection in the silver metal, then performs a high-speed Pull-out.【SCENE 04｜4.7–6.3s｜LET YOUR FUTURE BLOOM｜BOTANICAL CIRCLE】Completely change composition to a 90° Top Shot. She now wears a deep burgundy modern reinterpreted short qipao, youthful and fitted, with black piping and minimalist silver frog-button details. Avoid traditional portrait-photography aesthetics. She is positioned exactly at the center of the frame, slightly crouching, with the tip of her right finger touching the ground. A tiny flower bud appears beneath her fingertip. On the next beat, vegetation suddenly grows outward at extreme speed. Black vines, deep burgundy petals, and small amounts of pink and white flowers spread 360 degrees around her. The plants do not form a background wall; they grow into a gigantic circular Typography Garden surrounding her. “LET YOUR” grows along her left side, “FUTURE” forms a massive semicircle above her, and “BLOOM” expands beneath her. Eventually “LET YOUR FUTURE BLOOM” becomes a complete giant botanical typography ring with the girl perfectly centered inside it. BOOM! Every flower blooms simultaneously and a storm of petals flies directly toward the lens. The camera rapidly descends from the overhead position and dives through the flowers for transition.【SCENE 05｜6.3–7.9s｜POWER YOUR CHOICE｜ELECTRIC FOREGROUND】Hard Cut to a frontal centered composition. She returns to a black + silver outfit: black cropped top, silver metallic mini skirt, black choker. She stands relatively close to the camera and suddenly thrusts both hands toward the lens. Due to wide-angle perspective, both hands become large foreground elements, but anatomy must remain realistic and natural. A tiny Hot Pink electric spark appears between her fingertips. BOOM! Electricity violently erupts. Pink + white electrical currents first rush toward the camera and form a gigantic “POWER” between her and the lens. Then electricity passes between her hands and surrounds her torso to form “YOUR”. Finally, the current extends behind her and forms “CHOICE”. Create three clear spatial layers: POWER — foreground, YOUR — surrounding the character, CHOICE — background. Her face remains visible through gaps in the Electric Typography and her eyes must never be obstructed. High-speed Rack Focus sequence: fingertips → POWER → her eyes → YOUR → CHOICE → her eyes. Finally, a massive pink electrical bolt strikes directly toward the camera, creating a white flash transition.【SCENE 06｜7.9–9.5s｜STEP INTO YOUR STORY｜SHADOW TYPOGRAPHY】The white flash fades into an extremely wide side composition. She wears a black modern Chinese-inspired structured cropped top, black layered mini skirt, silver waist chain, and black knee-high boots. She is positioned on the left side of the frame, while a huge gray-white wall extends far toward the right. Strong directional side lighting creates an enormous elongated shadow across the wall and floor. She begins walking toward the right. On the first step, her shadow stretches and transforms into “STEP”. On the second step, STEP climbs from the floor onto the wall. Then “INTO” stretches away into the distance following spatial perspective. Finally, “YOUR STORY” suddenly expands from behind her into enormous Shadow Typography. She continues walking forward, physically passing between giant shadow letters S, T, O, R, Y. The camera performs high-speed Side Tracking parallel to her movement. At the end, she walks directly into the enormous black interior of “STORY”, and the shadow completely consumes the camera.【SCENE 07｜9.5–11s｜TURN THE PAGE｜PAPER TYPOGRAPHY】A storm of paper suddenly erupts from the darkness. Use a strong Dutch Angle composition. She now wears a distressed gray-blue denim cropped top, charcoal-black asymmetric low-rise mini skirt, and silver waist chain, positioned slightly left of center. Large quantities of white paper, black paper, deep pink magazine paper, and burgundy paper fragments fly through the space at high speed. Paper sheets first fold together in the foreground to form a gigantic “TURN”. TURN then flips upward like an enormous magazine page, passing from the foreground over her head and physically landing behind her. Simultaneously, “THE” forms along both sides of her body, while a gigantic “PAGE” expands from behind her toward the camera. She swings one hand. WHOOSH! The entire “TURN THE PAGE” structure flips like an enormous Fashion Magazine from in front of her to behind her. The camera races between flying sheets of paper. Sheets passing extremely close to the lens create realistic Motion Blur.【SCENE 08｜11–13s｜LIVE BY YOUR RULES｜RIBBON TYPOGRAPHY】The music reaches maximum intensity. She instantly changes into the Final Dark Idol Look: black asymmetric cropped top, black layered ultra-short skirt, silver waist chain, black choker, silver earrings, black knee-high boots. She stands exactly in the center and performs a fast turn. Large amounts of real Hot Pink, deep burgundy, and black ribbons explode from behind her. The ribbons must have realistic weight, folds, reflections, inertia, and aerodynamic drag. The first ribbon group shoots from behind her left shoulder, curves over her head into the extreme foreground to form “LIVE”, then continues around and behind her. The second group emerges again from behind her right side, crosses in front of her waist, and forms “BY YOUR”. The third group rises from around her feet, spirals around her legs, then angles toward the camera to form “RULES”. Eventually “LIVE BY YOUR RULES” becomes a gigantic 3D Typography Spiral. Typography continuously moves from foreground → side → background → extreme foreground → back behind her. She genuinely stands inside the center of the typographic vortex. The camera performs a high-speed 360° Orbit around her while the character and Ribbon Typography rotate in opposite directions. On the final BOOM, all ribbons instantly freeze and the complete phrase “LIVE BY YOUR RULES” remains clearly readable once.【SCENE 09｜13–15s｜CHOOSE YOUR LIFE｜IRIS TYPOGRAPHY TUNNEL】Hard Cut to an Extreme Macro of both eyes, filling almost the entire 16:9 frame. Preserve exactly the same eye shape, eye makeup, bangs, eyebrows, and realistic skin texture from @Image1. A tiny Hot Pink light appears in the center of the pupil. Deep inside the iris, “CHOOSE” gradually appears, followed by “YOUR”, then “LIFE”, forming “CHOOSE YOUR LIFE”. The typography is NOT printed onto the eyeball surface; it exists as miniature 3D Neon Typography suspended deep inside the iris. The camera suddenly rushes toward the right pupil and enters it. The pupil instantly transforms into a gigantic black + Hot Pink typography environment. Countless repetitions of “CHOOSE YOUR LIFE” extend through spatial depth to form an Infinite Typography Tunnel. Nearby typography is enormous, mid-distance typography is medium-sized, distant typography becomes progressively smaller, creating extreme infinite perspective. The camera races through the typography tunnel while gigantic letters sweep past from the left, right, above, and below. Finally, all typography suddenly collapses toward the center. In the middle of the black space, only one complete and clearly readable “CHOOSE YOUR LIFE” remains. The Hot Pink light contracts into a single point. Final Bass Hit: the girl blinks, instant cut to black.【TEXT ACCURACY｜HIGHEST PRIORITY】The following phrases must be generated exactly and spelled correctly: FREE YOUR MIND; LIGHT YOUR OWN WAY; BECOME WHO YOU CHOOSE; LET YOUR FUTURE BLOOM; POWER YOUR CHOICE; STEP INTO YOUR STORY; TURN THE PAGE; LIVE BY YOUR RULES; CHOOSE YOUR LIFE. No gibberish, random letters, extra words, or spelling errors. Even during high-speed typography movement, every phrase must have at least 0.3–0.5 seconds of complete, stable, clearly readable presentation.【ALL NINE SCENES MUST HAVE CLEARLY DIFFERENT COMPOSITIONS】01 Frontal full-body + Smoke Typography across foreground/midground/background; 02 Extreme Face Close-Up + Neon wrapping around the face in an S-shaped 3D path; 03 Extreme Low Angle + giant Chrome Typography architecture; 04 90° Top Shot + character centered inside a botanical typography ring; 05 frontal wide-angle forced perspective + Electric Typography in the foreground; 06 ultra-wide side view + Side Tracking + character walking into Shadow Typography; 07 Dutch Angle + Paper Typography flipping from foreground to background; 08 centered character + Ribbon Typography surrounding her in 360° space; 09 Extreme Eye Macro + Infinite Typography Tunnel. Never use the same composition for two consecutive scenes.【HIGH-SPEED CAMERA LANGUAGE】Frequently use High-Speed Push-in, High-Speed Pull-out, Whip Pan, Extreme Macro, Ultra Wide Angle, Low Angle, Top Shot, Dutch Angle, Side Tracking, 180° Orbit, 360° Orbit, Rack Focus, Perspective Distortion, Typography Fly-through, and Letter Surface Tracking. Camera movement can be extremely fast, but the character’s face must never become distorted by wide-angle lenses or motion. Facial features must remain stable, recognizable, sharp, and anatomically correct at all times.【STRICTLY FORBIDDEN】No face swapping, no additional people, no facial-feature drift, no Westernization, no face-shape changes, no age changes, no skin-tone changes, no hair-color changes, no missing bangs; no extra hands, extra legs, extra fingers, malformed hands, body intersections, or melting clothing; no typography passing through the character’s face or covering her eyes; no gibberish typography or spelling mistakes; no typography functioning as ordinary flat subtitles; no typography simply pasted onto background walls; no repetitive “character on the left + text on the right” compositions; no frontal medium shot in every scene; no repeated typography materials across all scenes; no cheap magical particles, no blue-purple cyberpunk aesthetic, no real brand logos, no player UI, no subtitles, no timestamps or numbers; no dialogue, no lip-syncing, no lyrics.【FINAL VISUAL TARGET】This is NOT a conventional Fashion Lookbook and NOT simply “a girl with English words added around her.” Throughout the full 15 seconds, the girl from @Image1 must genuinely enter and inhabit a Fashion World constructed entirely from Typography. Text repeatedly generates behind her → passes around her shoulders → wraps into the space in front of her body → flies toward the camera → circles back behind her. She physically stands between letters, walks through gigantic typography, becomes surrounded by text, controls typography deformation, causes typography to wrap around her body, and ultimately enters the typography itself. All nine scenes must have distinctly different materials, compositions, shot scales, spatial relationships, character actions, and camera movements. Final visual identity: BLACK + HOT PINK + WHITE.",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2092457346526241208",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-08-26",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092457346526241208.mp4",
    "posterUrl": "/posters/x/2092457346526241208.jpg",
    "duration": 14.326712,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "动态图形",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2092457346526241208"
  },
  {
    "id": "x-2092307579221516749",
    "title": "以大幅歌词为背景的表演",
    "titleEn": "A Performance against Oversized Lyrics",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物站在反光地面上表演，身后的英文歌词不断改变版式。",
    "summaryEn": "A performer stands on a reflective floor while English lyrics shift in large layouts behind her.",
    "prompt": null,
    "sourceUrl": "https://x.com/h64g_/status/2092307579221516749",
    "sourceLabel": "X 原帖",
    "author": "h64g_",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092307579221516749.mp4",
    "posterUrl": "/posters/x/2092307579221516749.jpg",
    "duration": 75.712,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092255957526990890",
    "title": "黑帮会客厅里的最后警告",
    "titleEn": "A Final Warning in a Gangster Drawing Room",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "几名角色围坐在华丽室内对峙，男子起身打破僵局。",
    "summaryEn": "Several characters confront one another in an ornate room until a man stands up.",
    "prompt": "REFERENCE DEFINITIONS\n\n@image1\nRepresents Mafia Organization A.\nThe seated man is the boss.\nThe man smoking is his subordinate.\nThe woman is the boss’s wife.\nTreat all three as members of the same mafia faction and preserve their visual identities from the reference image.\n\n@image2\nRepresents Mafia Organization B, the rival faction.\nThe standing man is the boss of this rival mafia organization.\nPreserve his visual identity from the reference image.\n\n@image3\nRepresents the resistance.\nThis group consists of two men and one woman.\nThey are the strategic side of the resistance, secretly planning how to drive both mafia organizations out of the city.\nPreserve their visual identities from the reference image.\n\nTONE AND EDITING\nCreate a self-contained 15-second cinematic crime-drama digest with the pacing and emotional compression of a condensed film-trailer sequence.\nThe video should not feel like one continuous scene.\nIt should feel like a rapid, cinematic digest of key moments from a larger crime film, using sharp intercutting between the rival mafia factions and the resistance.\nDo not use a rigid shot order or fixed timing.\nLet MiniMax H3 decide the exact sequence, shot length, and camera variation naturally from the story.\nThe trailer-like quality must come from cinematic intercutting, brief reaction shots, escalating emotional intensity, sharp transitions between factions, and a growing sense of imminent conflict.\nDo not use trailer text, title cards, narration, or music.\nThe emotional intensity must clearly escalate across the 15 seconds.\nEach return to the mafia factions should feel more hostile, more irritated, and less restrained than before.\nBy the end, the hostility should feel close to boiling over.\nKeep the tone serious, grounded, and dramatic — like a tense, elegant crime-film digest, not an action montage.\n\nSTORY\nTwo powerful mafia organizations are fighting for control of the same city.\nInside Mafia Organization A in @image1, the boss is visibly irritated by the growing threat of the rival faction. His subordinate and his wife share the tense atmosphere around him. Frustration, resentment, and hostility are building inside the room.\nThe rival boss in @image2 has no intention of backing down. His presence, ambition, and refusal to yield only intensify the conflict. The rivalry between the two mafia organizations keeps tightening, and their mutual irritation steadily grows into open hostility.\nWhile both mafia factions become increasingly consumed by this rivalry, the resistance in @image3 watches from the shadows.\nThe two men and one woman in the resistance are not reacting emotionally like the mafia factions. Instead, they remain focused, deliberate, and calculating. They are secretly forming a plan to remove both mafia organizations from the city.\nBy the end of the 15 seconds, the rivalry between the two mafia organizations should feel dangerously close to open conflict, while it becomes clear that the resistance has quietly begun moving against them.\n\nDIALOGUE AND SOUND\nAllow the characters to speak naturally when appropriate.\nAll spoken dialogue must be in natural English.\nGenerate the dialogue based on the characters’ roles, emotions, and the situation.\nKeep every line short, cinematic, and believable.\nDo not overfill the 15 seconds with dialogue.\nThere should be no narrator.\nNo BGM. No music.\nUse only spoken English dialogue, environmental ambience, natural room tone, and subtle cinematic sound effects.\n\nNEGATIVE\nNo title cards.\nNo subtitles or on-screen text.\nNo logos.\nNo narrator or voice-over exposition.\nNo background music.\nNo action-heavy gunfight or chase sequence.\nDo not introduce unrelated major characters.\nDo not confuse the identities or faction roles defined by the reference images.",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2092255957526990890",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092255957526990890.mp4",
    "posterUrl": "/posters/x/2092255957526990890.jpg",
    "duration": 30.232381,
    "aspectRatio": "1:1",
    "resolution": "1080p",
    "tags": [
      "对白与口型",
      "写实电影感",
      "人物表演"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2092255957526990890"
  },
  {
    "id": "x-2092242385141219589",
    "title": "水墨背景中的服装与发色变化",
    "titleEn": "Outfit and Hair Changes against Ink",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子从日常上衣切换为长裙造型，周围墨迹与发色同时变化。",
    "summaryEn": "A woman changes from casual clothing to a gown as ink shapes and hair color shift around her.",
    "prompt": null,
    "sourceUrl": "https://x.com/akemiwork/status/2092242385141219589",
    "sourceLabel": "X 原帖",
    "author": "akemiwork",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092242385141219589.mp4",
    "posterUrl": "/posters/x/2092242385141219589.jpg",
    "duration": 15.13941,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "风格化与实验",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092231386610487615",
    "title": "工业厂房里的聚光时装镜头",
    "titleEn": "A Spotlight Fashion Scene in an Industrial Hall",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "高大厂房中的单束光线逐渐引向黑衣模特近景。",
    "summaryEn": "A single beam of light in a vast industrial hall leads into a close portrait of a model in black.",
    "prompt": "15s vertical 3:4 high-fashion industrial editorial.\n\nUse the blonde model reference as the ONLY identity reference. Preserve her face, fair-tan skin, long wavy blonde hair, light eyes, lean athletic proportions, and confident editorial expression.\n\nUse the new fashion reference image only as an outfit inspiration reference. Create a sleek fitted fashion look inspired by it: a black sleeveless semi-sheer or subtly sheer striped top with a refined editorial feel, paired with a matching fitted black skirt that ends at mid-thigh. Add matching elegant black flats, minimal and fashion-forward, coordinated with the outfit. The overall styling should feel polished, modern, chic, and slightly minimalist. No heels, no boots, no sneakers.\n\nSetting: a vast abandoned industrial warehouse with concrete columns, steel roof trusses, blue-gray shadows, atmospheric haze, floating dust, and dramatic hard white shafts of light. Mood is sleek, editorial, moody, and cinematic.\n\n**0.0–1.3s**\nExtreme wide symmetrical opening. She stands small at center under a single overhead spotlight in the middle of the warehouse. Smooth fast dolly-in toward her.\n\n**1.3–3.0s**\nFull-body frontal hero shot. Slow push-in. Her full outfit is clearly visible: fitted striped top, mid-thigh skirt, and matching flats. She holds a poised editorial stance with subtle body tension and slight natural hair movement.\n\n**3.0–4.5s**\nExtreme close-up of her face. Half of her face is lit, half falls into shadow. Calm, piercing gaze with a tiny blink.\n\n**4.5–6.2s**\nMedium 3/4 side-tracking shot as she walks confidently through the space. Camera glides beside her. Her movement is smooth, controlled, and fashion-editorial, with natural motion in the skirt.\n\n**6.2–8.2s**\nAbstract environmental cutaway: a cone of white light, floating dust, deep black negative space, no person visible.\n\n**8.2–9.1s**\nMedium portrait beside a concrete pillar. She subtly shifts her weight and turns her chin into the light.\n\n**9.1–10.6s**\nLow-angle tracking shot as she walks toward camera beneath the steel roof beams. Strong posture, direct runway-like confidence. Keep the flats visible in at least part of this shot.\n\n**10.6–11.6s**\nSilhouette shot. She is framed against a bright rectangular doorway or opening, moving slowly forward.\n\n**11.6–12.3s**\nFast abstract transition: diagonal streak of light cutting across black.\n\n**12.3–15.0s**\nFinal frontal waist-up hero shot under a large overhead shaft of light. Slow push-in. She holds direct eye contact with subtle confidence and minimal movement.\n\nVisual style: hyper-realistic luxury campaign photography, strong chiaroscuro lighting, clean hard cuts, no text, no extra people, no props, no morphing, no identity drift, no outfit changes, natural skin texture, realistic hair and fabric motion, consistent face and body in every shot.",
    "sourceUrl": "https://x.com/PrometheanAIX/status/2092231386610487615",
    "sourceLabel": "X 原帖",
    "author": "PrometheanAIX",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092231386610487615.mp4",
    "posterUrl": "/posters/x/2092231386610487615.jpg",
    "duration": 15.168,
    "aspectRatio": "3:4",
    "resolution": "768p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/PrometheanAIX/status/2092231386610487615"
  },
  {
    "id": "x-2092220604309205341",
    "title": "黑白文字隧道的动态开场",
    "titleEn": "A Dynamic Opening through a Black-and-White Type Tunnel",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "大字排版被拉伸成空间隧道，随后收束为平面标识。",
    "summaryEn": "Oversized typography stretches into a spatial tunnel before resolving into a flat mark.",
    "prompt": "subject_definitions:\nこの映像の設定は以下の通りである。\nアクセント色は彩度の高いオレンジである。\n第1の文字列は \"MiniMax H3\" である。\n第2の文字列は \"MOTION\" である。\n第3の文字列は \"GRAPHICS\" である。\n第4の文字列は \"NO EDIT\" である。\n第5の文字列は \"ONE SHOT\" である。\n第6の文字列は \"コピペで\" である。\n第7の文字列は \"超簡単！\" である。\n第8の文字列は \"ONEROOM\" である。\n第9の文字列は \"AI STUDIO\" である。\n画面を3分割する縦棒は、第1の文字列に含まれる大文字 H の縦棒である。\n\n画面に登場する文字列は上の9つだけであり、これ以外の文字・単語・記号・数字は一切出現しない。すべて上に書いた通りの綴りで、大文字と小文字の別も空白の有無もそのまま正確に保ち、平坦・鮮明・安定した状態で描画する。崩れたり、文字が増減したり、文字化けしたりすることは決してない。\n「第1の文字列」「第2の文字列」などの呼び方は、この指示書の中だけで使う呼称である。画面に描くのは必ずその呼称に割り当てられた文字列そのものであり、「第1」「文字列」といった呼称の語や数字が画面に描かれることは決してない。\n書体は太いジオメトリックサンセリフ1種類のみで、欧文はすべて大文字。ただし固有名詞にもともと小文字が含まれる場合は、その混在をそのまま正確に保つ。日本語は太いゴシック体1種類のみ。\n色はオフホワイト、インクブラック、アクセント色の3色のみ。この3色以外は一切使わない。地の色は映像の途中で白と黒のあいだで3回反転し、そのたびに文字色も反転する。\n画面には常に3つの層がある。奥の背景層、中間の装飾層、手前の文字層。この3層は必ず異なる速度で動き、3層すべてが同時に静止する瞬間は最後のホールドだけである。\n背景層には第8の文字列が、塗りのない輪郭線だけで描かれ、画面幅の3倍の大きさで横倒しに寝そべり、右から左へ超低速で流れ続けている。これは背景の壁紙であり、手前の文字層とは別物である。重複した単語ではない。\n中間の装飾層には、細いバーコードの帯、極小のUIティックの列、レジストレーションマーク、ハーフトーンのドット網、テクニカルな円弧、データ数値のティックが常時いずれか複数存在し、文字層とは別の速度で動いている。\n\nsummary:\n15秒の爆発的なフラットグラフィック・モーションデザイン。9つのカットすべてが構図的に全く異なり、地の白黒が3回反転し、背景の巨大なアウトライン文字が全編流れ続ける。文字は硬い物体として叩きつけられ、圧縮され、砕け、跳ね返り、増殖し、最後に激突する。ばらばらの線分が第1の文字列を組み上げ、縦棒が画面を3分割し、第2の文字列が叩き込まれ、第3の文字列がワイヤーフレームのグリッド上で増殖し、カードが連打され、収束した光点から第8の文字列が左右から激突して生まれ、最後に第9の文字列を上、第8の文字列を下に置いた2行のロックアップで着地する。\n\nretention_analysis:\n各文字列が出現するショットは次の通りであり、増殖・複製された文字もすべて同じ綴りを保つ。\n第1の文字列は [Shot 2]、[Shot 3]、[Shot 4] に出現する。\n第2の文字列は [Shot 5] に出現する。\n第3の文字列は [Shot 6] に出現する。\n第4、第5、第6、第7の文字列は [Shot 7] に出現し、いずれも完全に読める状態で保持される。\n第8の文字列は背景層に全編出現し、手前の文字層としては [Shot 8]、[Shot 9] に出現する。背景層のものは輪郭線のみ、手前のものは塗りつぶしであり、同じ綴りだが別の層である。誤って重複した単語として扱わない。\n第9の文字列は [Shot 9] に出現する。\n\ndetailed_description:\n本映像は高密度で攻撃的なフラットグラフィックモーションデザインである。エネルギーはすべてタイポグラフィの物理から生み出す。9つのショットはすべて構図が異なっていなければならず、同じ構図、同じスケール、同じ配置を二度使わない。各ショットの演出の細部はここでは指定しない。\n\n[Shot 1] 地は黒。黒い線分の群れが四辺すべてから爆発的に飛び込み、渦を巻きながら中央へ加速して吸い込まれていく。巨大なアクセント色の円が画面外から叩きつけられて着弾し、強いインパクトシェイクが走る。\n\n[Shot 2] At 00:01.000, 渦の中心で線分が一斉に噛み合い、第1の文字列が画面いっぱいに組み上がる。組み上がった瞬間、地が黒から白へ一気に反転する。第1の文字列の最後の1字だけがアクセント色である。\n\n[Shot 3] At 00:02.600, スナップズームでそのアクセント色の1字へ急速に寄り、その1字が画面全体を覆い尽くすまで巨大化する。画面全体が一度だけフリーズする。\n\n[Shot 4] At 00:03.400, 地は白。縦棒が上下へ伸び上がって画面の全高を貫き、画面を3つの縦列に切り分ける。第1の文字列の残りの文字は左へ弾き飛ばされて画面外へ消える。3つの縦列はそれぞれ異なる速度で上下にスクロールする。\n\n[Shot 5] At 00:05.200, 3つの縦列が一斉に急停止し、第2の文字列が画面いっぱいに正面から叩き込まれ、完全に読める状態で静止する。\n\n[Shot 6] At 00:06.400, ハードカットで黒地へ反転する。巨大な白いワイヤーフレームのグリッドが3次元的に傾きながら奥へ伸び、その面に沿って第3の文字列が格子状に増殖して手前へ流れてくる。\n\n[Shot 7] At 00:08.000, グラフィックカードが4枚、ハードカットで連打される。1枚ごとに構図もスケールも配置も異なり、第4、第5、第6、第7の文字列が1枚に1つずつ、画面の中央に大きく現れる。各カードの切り替わりはシャッターフラッシュで区切られる。\n\n[Shot 8] At 00:10.000, 4枚のカードが中央へ一気に吸い込まれ、1つのアクセント色の光点に凝縮する。光点から左右へ、第8の文字列が2つの塊に分かれて超高速で飛び込み、画面中央で激突して一行に綴られる。アクセント色の同心円の衝撃波が中央から外へ炸裂し、砕けた平面の破片がガラスのように四方へ飛び散る。\n\n[Shot 9] At 00:12.000, 破片が画面外へ抜け、第8の文字列が画面の下半分へ落下して着地し、その真上により小さいサイズで第9の文字列が固定される。2行が中央揃えで積み重なる。地が黒から白へ最終反転する。2行のあいだに短いアクセント色の罫線が1本割り込む。00:13.800 から 00:15.000 まで、背景層も含めて画面上のすべてが一切動かない状態を保持する。字形がランダムにグリッチすることは決してない。",
    "sourceUrl": "https://x.com/studio_oneroom/status/2092220604309205341",
    "sourceLabel": "X 原帖",
    "author": "studio_oneroom",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092220604309205341.mp4",
    "posterUrl": "/posters/x/2092220604309205341.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/studio_oneroom/status/2092220604309205341"
  },
  {
    "id": "x-2092188350476337613",
    "title": "冰淇淋口味切换的概念网页",
    "titleEn": "A Concept Ice-Cream Page Switching Flavors",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "冰淇淋产品与黑白、巧克力和热带配色界面交替出现。",
    "summaryEn": "Ice-cream products switch between monochrome, chocolate and tropical interface themes.",
    "prompt": null,
    "sourceUrl": "https://x.com/ShamsAmin56/status/2092188350476337613",
    "sourceLabel": "X 原帖",
    "author": "ShamsAmin56",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092188350476337613.mp4",
    "posterUrl": "/posters/x/2092188350476337613.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "商业广告感",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092146855920783418",
    "title": "平面插画讲述职业转折",
    "titleEn": "A Career Change Told in Flat Illustration",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "办公人物、时钟和门的图形组成简洁的职业服务广告。",
    "summaryEn": "Office workers, clocks and doorway graphics form a minimal career-service ad.",
    "prompt": "integrated_multimodal_description: Create a complete 15-second Japanese flat-illustration commercial in a vertical 9:16 composition for a fictional career-change agency called トビラ. Use friendly 2D flat-vector picture-book illustration: thick, even, near-black outlines of consistent weight, flat solid color fills, no shading, simple dot eyes and small curved mouths, restrained cutout animation, and a light paper grain over the flat color fields. Use a compact palette of warm off-white, muted slate blue, coral red, golden yellow and fresh mint green, with near-black outlines. Background color changes mark the chapters. Because the frame is vertical, elements stack from top to bottom and are never laid out in a horizontal row. No photorealism, no 3D rendering, no gradients, no drop shadows, no browser interface, no screen-recording artifacts, and no player controls.\n\n[Shot 1] Begin on a warm off-white background. A young office worker with short black hair, a white shirt, navy trousers and one round blush spot on each cheek sits side-on at a small desk with an open laptop. Their hands stop moving on the keyboard and come to rest. The Japanese caption \"このままでいい？\" then appears across the upper area, arriving one character at a time. A round wall clock above them turns its hand a full circle.\n\n[Shot 2] At 00:02.500, a muted slate-blue field wipes across the frame. The same worker sits smaller in the centre. Three round white speech bubbles pop in one at a time around them, reading \"給料\", then \"時間\", then \"やりがい\". The bubbles crowd inward toward the worker, and their eyebrows lower as their shoulders drop.\n\n[Shot 3] At 00:05.000, a coral-red field wipes across the frame. The worker holds up a smartphone and plain white cards, each with two short grey lines on it, pour out of its screen one after another and pile up until they fill the frame. Their free hand reaches toward the cards and stops halfway. The caption \"動けない\" assembles in two stages in large near-black type.\n\n[Shot 4] At 00:07.500, a golden-yellow field wipes across the frame and the picture brightens. A friendly guide in a golden-yellow cardigan with round glasses and a document folder pops in from the lower right. Three white outlined cards then stack up one at a time from top to bottom, reading \"仕事を探す\", \"書類を整える\" and \"日程を決める\". Small yellow plus signs appear in the gaps between the stacked cards.\n\n[Shot 5] At 00:10.000, a fresh mint-green field wipes across the frame. The three stacked cards connect into a single top-to-bottom flow with simple arrows between them. A handshake symbol appears below the lowest card. The worker stands beside it empty-handed, the corners of their mouth lifting and their shoulders rising.\n\n[Shot 6] At 00:12.500, the background settles back to warm off-white with generous empty space. The caption \"まずは、話すだけ。\" appears in near-black type in the upper area. A simple door mark and the brand name \"トビラ\" then appear together in the centre. The door mark swings open and a wedge of golden-yellow light spreads out from behind it, holding cleanly until exactly 15.000 seconds.\n\noverall_soundscape: A steady keyboard tapping stops abruptly at the start, leaving a quiet clock tick. Paper-like pops mark each speech bubble and each caption, and a run of riffling card sounds swells and then cuts off sharply. Short restrained whooshes carry the color-field wipes, small UI clicks land as each card stacks into place, and a soft confirmation chime accompanies the handshake. A light wooden door sound opens the final shot. No dialogue.\n\nnon_diegetic_music: Generate an original 15-second friendly Japanese commercial jingle using light marimba, soft plucked synth, gentle hand percussion and warm sustained chords at a moderate tempo. Begin sparse and hesitant over the off-white opening, sink lower and slower over the blue section, tighten and grow restless over the coral section, open up bright and warm as the yellow section arrives, keep that warmth moving through the mint section, and resolve with one clean final chord at exactly 15.000 seconds.",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2092146855920783418",
    "sourceLabel": "X 原帖",
    "author": "tanabe_fragm",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092146855920783418.mp4",
    "posterUrl": "/posters/x/2092146855920783418.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "插画与手绘",
      "动态图形",
      "产品与广告",
      "日常生活"
    ],
    "category": "advertising",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2092146855920783418"
  },
  {
    "id": "x-2092132055857746375",
    "title": "粉色双马尾与环绕字体",
    "titleEn": "Pink Twin Ponytails with Orbiting Typography",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物站在黑色背景中，粉白字体围绕身体和面部切换。",
    "summaryEn": "A character stands against black as pink-and-white type moves around the body and face.",
    "prompt": null,
    "sourceUrl": "https://x.com/KiraKiraKaiju/status/2092132055857746375",
    "sourceLabel": "X 原帖",
    "author": "KiraKiraKaiju",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092132055857746375.mp4",
    "posterUrl": "/posters/x/2092132055857746375.jpg",
    "duration": 15.146667,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "动态图形",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092124735081738349",
    "title": "庭院与屋顶的慢生活片段",
    "titleEn": "Slow-Life Moments in a Courtyard and on a Rooftop",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子晾起白布，随后在砖墙庭院和门边安静活动。",
    "summaryEn": "A woman hangs a white sheet and moves quietly through a brick courtyard and doorway.",
    "prompt": "A peaceful cinematic scene of a young woman sitting quietly on a wooden park bench surrounded by lush green ferns and dense tropical foliage. She wears a soft peach blouse and white pants, gently closes her eyes, takes a slow deep breath, and relaxes with a subtle peaceful smile. A light breeze softly moves her hair and the leaves around her. Natural morning light filters through the greenery, creating a calm, dreamy atmosphere. Slow cinematic camera push-in, realistic facial movement, natural body motion, shallow depth of field, soft background bokeh, photorealistic, ultra-detailed, smooth motion, 4K. No sudden movements, no camera shake.",
    "sourceUrl": "https://x.com/CaliraVal/status/2092124735081738349",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092124735081738349.mp4",
    "posterUrl": "/posters/x/2092124735081738349.jpg",
    "duration": 15.168,
    "aspectRatio": "106:45",
    "resolution": "360p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2092124735081738349"
  },
  {
    "id": "x-2092124356709388475",
    "title": "晨光厨房里的手作披萨",
    "titleEn": "Handmade Pizza in a Sunlit Kitchen",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子在面饼上铺酱料和配料，最后坐到餐桌旁品尝披萨。",
    "summaryEn": "A woman adds sauce and toppings to pizza dough and sits down to taste the result.",
    "prompt": null,
    "sourceUrl": "https://x.com/ZoeyBennettAi/status/2092124356709388475",
    "sourceLabel": "X 原帖",
    "author": "ZoeyBennettAi",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092124356709388475.mp4",
    "posterUrl": "/posters/x/2092124356709388475.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "美食",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092118838859190377",
    "title": "回家途中突发的街头冲突",
    "titleEn": "An Unexpected Confrontation on the Way Home",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子在街头遭遇冲突，随后回到家门口；原帖附的是分镜图提示词。",
    "summaryEn": "A woman encounters a street confrontation and returns home; the post supplies a storyboard-image prompt.",
    "prompt": null,
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2092118838859190377",
    "sourceLabel": "X 原帖",
    "author": "ayzalnooor24521",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092118838859190377.mp4",
    "posterUrl": "/posters/x/2092118838859190377.jpg",
    "duration": 15.789569,
    "aspectRatio": "68:45",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "照片级写实",
      "动作打斗",
      "城市街景"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2092101929837166818",
    "title": "粉色居家装的日常手机片段",
    "titleEn": "Everyday Phone Moments in Pink Homewear",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子在家中走动后坐到床边，浏览手机里的照片。",
    "summaryEn": "A woman moves through her home, sits beside the bed and browses photos on her phone.",
    "prompt": "Create a 15-second ultra-realistic cinematic lifestyle video based on the storyboard, featuring the same woman and exact facial identity from the reference image in every shot. Keep her long wavy brown Korean-inspired hairstyle, natural features, and body proportions consistent. She wears the same oversized soft-pink T-shirt and loose light-blue pants throughout, with no coat, jacket, bag, backpack, or wardrobe changes. Start with her hanging freshly washed white clothes on a sunny rooftop clothesline, then smoothly transition to her walking downstairs inside her own home with empty hands. She sits outside near the entrance and enjoys an ice-cream cone, then walks back inside and enters her bedroom. She sits on the bed using her smartphone, followed by a close-up of her face with a soft natural smile, an over-the-shoulder shot of her scrolling on the phone, and finally a cozy shot of her lying on the bed, holding the phone and smiling peacefully. Use warm natural lighting, realistic skin and hair movement, subtle expressions, shallow depth of field, smooth gimbal-style camera motion, natural body movement, seamless transitions, and high-end commercial cinematography. Preserve exact facial identity, clothing, hairstyle, hands, and body continuity with no morphing or distortions. 16:9, 4K photorealism, cinematic, no text, subtitles, or watermark.",
    "sourceUrl": "https://x.com/Lianaalane/status/2092101929837166818",
    "sourceLabel": "X 原帖",
    "author": "Lianaalane",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092101929837166818.mp4",
    "posterUrl": "/posters/x/2092101929837166818.jpg",
    "duration": 15.719909,
    "aspectRatio": "68:45",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Lianaalane/status/2092101929837166818"
  },
  {
    "id": "x-2092093848130617686",
    "title": "庆祝字体与剧院揭幕",
    "titleEn": "Celebration Typography and a Theater Reveal",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物与立体庆祝文字在粉黑色空间中切换，最后转向剧院舞台。",
    "summaryEn": "A character and three-dimensional celebratory words shift through pink-and-black scenes before a theater reveal.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2092093848130617686",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092093848130617686.mp4",
    "posterUrl": "/posters/x/2092093848130617686.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动态图形",
      "风格化与实验",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090691604277240092",
    "title": "森林蝴蝶彩妆盘概念广告",
    "titleEn": "A Butterfly Makeup Compact in an Enchanted Forest",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蝴蝶形彩妆盘、上妆特写与林间光点组成梦幻美妆片。",
    "summaryEn": "A butterfly-shaped compact, makeup close-ups and forest sparkles form a fantasy beauty concept.",
    "prompt": "Set in a golden, sun-dappled enchanted forest filled with floating light particles, an iridescent butterfly-wing makeup compact labeled \"PAWLYNA\" rests on a lush, moss-covered rock surrounded by delicate wildflowers. The translucent pearlescent wing opens to reveal a colorful makeup palette featuring stained-glass-inspired pans of pastel eyeshadows and shimmering blushes. Close-up shots showcase a makeup brush applying glistening eyeshadow onto the eyelid of a young woman with pink hair, dusting soft blush on her cheek, and sweeping a rosy lip gloss across her lips. The video reveals the model in a romantic frilly dress holding the wing-shaped compact beside her face before concluding with a close-up of the opened compact amidst glowing magical sparkles alongside the tagline \"PAWLYNA: UNFOLD YOUR MAGIC.\"",
    "sourceUrl": "https://x.com/TaliaAariz/status/2090691604277240092",
    "sourceLabel": "X 原帖",
    "author": "TaliaAariz",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090691604277240092.mp4",
    "posterUrl": "/posters/x/2090691604277240092.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "产品与广告",
      "奇幻"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TaliaAariz/status/2090691604277240092"
  },
  {
    "id": "x-2090652614073663786",
    "title": "淡紫手袋与同色时装",
    "titleEn": "A Lilac Handbag with Matching Fashion",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "皮革与拉链细节切换到同色系模特造型，展示手袋概念。",
    "summaryEn": "Leather and zipper details transition to a coordinated model look in a handbag concept.",
    "prompt": "A 15-second quiet-luxury ad: elegant product shots of a soft pastel lilac structured handbag, close-ups of its leather and silver hardware, a graceful model in matching lilac outfit carrying it through a bright modern glass space, and the bag floating as the brand ELYRA is revealed.",
    "sourceUrl": "https://x.com/eshal__ai/status/2090652614073663786",
    "sourceLabel": "X 原帖",
    "author": "eshal__ai",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090652614073663786.mp4",
    "posterUrl": "/posters/x/2090652614073663786.jpg",
    "duration": 15.16263,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/eshal__ai/status/2090652614073663786"
  },
  {
    "id": "x-2090635415380537797",
    "title": "古墓黑水池的悬疑片段",
    "titleEn": "A Suspense Sequence around a Tomb Pool",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "古墓水池、封闭大门与人物镜头组成探险悬疑段落。",
    "summaryEn": "A tomb pool, sealed doors and character shots form an adventure suspense sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/peter6759/status/2090635415380537797",
    "sourceLabel": "X 原帖",
    "author": "peter6759",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090635415380537797.mp4",
    "posterUrl": "/posters/x/2090635415380537797.jpg",
    "duration": 105.813333,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090634934193152115",
    "title": "巨型文字困境中的求职广告",
    "titleEn": "A Career Ad through Giant Typographic Obstacles",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "职场人物面对巨型文字障碍，黑白画面逐渐引入青色品牌元素。",
    "summaryEn": "An office worker faces oversized typographic obstacles as cyan brand elements enter a black-and-white world.",
    "prompt": "integrated_multimodal_description:\n2D-animated flat vector motion graphics in a vertical 9:16 composition, exactly 14 shots at\n24fps. Flat anime rendering, thick clean outlines, two-tone cel shading, no gradients, no\nphotorealism, no 3D. Giant Japanese typography is physical architecture: solid black slabs the\nsize of buildings that crush, block, fall and can be climbed. It is high summer and the first half is hot, stale and windless, with heat\nhaze in the air. Palette: black, washed-out greyish white, deep navy and vivid cyan. Cyan first appears as a\nsingle point in Shot 6, then spreads further in each following shot until it fills the frame at\nShot 13; it never appears before Shot 6.\n\nThe recurring character is a Japanese office worker: a woman with a long black ponytail and\nsharp determined eyes, in a white short-sleeve blouse under a light grey lightweight blazer,\ndeep navy tapered trousers, low-heeled black pumps and a plain blank ID badge on a lanyard,\nwith a light sheen of sweat at her temples. Preserve her exact face, hairstyle, outfit and\ncolors throughout; never redesign her. Her blazer is the mechanism: buttoned closed and flat\ngrey until Shot 7, then it blows open and its vivid cyan lining flashes out, flaring wider in\nevery following shot.\n\nShe is also the narrator (S1), an inner monologue in a low voice, tired and flat at first and\nfirmer later, always in off-screen voiceover with her lips closed. Every Japanese word on\nscreen stays fully readable and is never covered by her.\n\n[Shot 1] Looking up from the bottom of a canyon of giant black Japanese character slabs. She\nstands small at the base, then tilts her head back to look up, then a bead of sweat runs down\nher temple as the air shimmers with heat. The narrator (S1) says in a tired, flat\ntone in an off-screen voiceover: <d>[Japanese] ずっと、ここで止まってた。</d> while her lips\nremain completely closed. The camera tilts up with small amplitude at slow speed. <scenetrans>\n\n[Shot 2] At 00:01.000, <scenetrans> the narrator's sentence continues seamlessly across\nthe cut. A colossal slab drops from above, then lands across her shoulders, then her\nknees buckle under it. The slab reads \"タスク\" in huge off-white type. The camera shakes\nslightly with small amplitude at fast speed.\n\n[Shot 3] At 00:02.000, two slabs slide in from left and right, then she braces both arms\nagainst them, then they press closer and she loses ground. The slabs read \"会議\" in huge\noff-white type. Static shot.\n\n[Shot 4] At 00:03.000, slabs stack at her feet, then she climbs two of them, then she slips\nand slides back down. The slabs read \"給料\" in huge off-white type. The camera pedestals up with\nsmall amplitude at slow speed.\n\n[Shot 5] At 00:04.000, the slabs swing closed into a sealed box around her, then she sits down,\nthen all movement in the air stops completely. No text. The camera pushes in\nwith small amplitude at slow speed.\n\n[Shot 6] At 00:05.000, everything freezes and holds still, nearly monochrome, with no text at\nall. One small cyan point of light glows at her feet, then a single sheet of paper lifts\nslightly off the ground and hangs there. The narrator does not speak at all during this shot.\nStatic shot.\n\n[Shot 7] At 00:05.800, wind arrives. Her ponytail streams sideways, then she lifts her head and\nher eyes sharpen, then her blazer blows open and the cyan lining flashes out for the first\ntime. No text. The camera pushes in with small amplitude at fast speed.\n\n[Shot 8] At 00:06.800, she grips the wall of slabs with both hands, then hauls it over so it\ntopples away, then wind rushes through the opening. No text. The camera trucks right with large\namplitude at fast speed.\n\n[Shot 9] At 00:08.000, the fallen slabs lock into a staircase rising to the upper right, then a\nwide cyan color field wipes across the frame, then each step edge lights from bottom to top. No text. The camera tilts up with small amplitude\nat fast speed.\n\n[Shot 10] At 00:09.000, she runs up the staircase of characters. Each slab lights cyan as she\nleaves it, then her ponytail and blazer stream hard in the wind, then she takes the next flight\ntwo steps at a time. The narrator (S1) says in a firmer, lighter tone\nin an off-screen voiceover: <d>[Japanese] 変えたのは、環境だけ。</d> The camera tracks her with\nlarge amplitude at fast speed. <scenetrans>\n\n[Shot 11] At 00:10.200, <scenetrans> the narrator's line continues seamlessly across\nthe cut and finishes. She leaps high off the staircase, then spins once in mid-air, then her\nblazer flares fully open so its lining forms a ring of cyan around her. No text. The\ncamera arcs with large amplitude at fast speed.\n\n[Shot 12] At 00:11.400, she lands at the top, then turns to look back with wind in her hair,\nthen the word \"月曜\" slams in huge in off-white beside her with a thick underline sweeping in\nbeneath it. The camera pushes in with small amplitude at slow speed.\n\n[Shot 13] At 00:12.600, the entire frame bursts into vivid cyan, then concentric rings blast\noutward from her feet, then shattered fragments of the black slabs fly away. No text. The narrator (S1) says in an off-screen voiceover: <d>[Japanese] 相談は無料。\n</d> <scenetrans> The camera pulls out with large amplitude at fast speed.\n\n[Shot 14] At 00:13.800, <scenetrans> the narrator continues without a break: <d>[Japanese]\nNexta。</d> A clean white color field wipes across and clears everything away, then a cyan mark\nof three ascending steps rising to the right assembles at the center and its top step extends\ninto an arrowhead, then the wordmark \"Nexta\" — spelled N-e-x-t-a, only the N\ncapitalized — resolves in deep navy beside it with \"相談は無料\" below. Hold the\nlockup until exactly 15.000 seconds. Static shot.\n\nOverall: the first half is hot, heavy and windless; the second half is fast, airy and cool. No\nshot is fully still except Shot 6. Every shot must feel compositionally different, and the\nstaircase in Shots 9 to 12 is built from the same slabs that crushed her earlier. The only\nwords allowed on screen anywhere are \"タスク\", \"会議\", \"給料\",\n\"月曜\", \"Nexta\" and \"相談は無料\". No other text, numbers, subtitles, captions or watermark.\n\noverall_soundscape:\nMassive stone slabs drop and grind with deep booming impacts while an air conditioner drones\nand distant cicadas buzz through stale, motionless air. Everything falls silent at 00:05.000\nuntil a soft gust of wind rises at 00:05.800. Wind then builds, cloth snaps and flutters,\nfootsteps climb and speed up, and a wide airy boom opens at 00:12.600.\n\nnon_diegetic_music:\nA low sustained synth drone holds alone at a slow pulse and thickens under the falling slabs.\nIt cuts out completely at 00:05.000 and stays absent until 00:05.800, when a light arpeggiated\nsynth and a four-on-the-floor kick enter and build through the climb. Wide chords and a snare\nroll open at 00:12.600 and resolve on one sustained chord at 15.000 seconds.",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2090634934193152115",
    "sourceLabel": "X 原帖",
    "author": "tanabe_fragm",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090634934193152115.mp4",
    "posterUrl": "/posters/x/2090634934193152115.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "插画与手绘",
      "动态图形",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2090634934193152115"
  },
  {
    "id": "x-2090634362631168129",
    "title": "红发动漫角色时尚登场",
    "titleEn": "A Red-Haired Anime Fashion Reveal",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红发角色在红黑白字形与几何转场中登场。",
    "summaryEn": "A red-haired character appears among red, black and white type and geometric transitions.",
    "prompt": "Create a **15s, 16:9, 24fps anime character reveal trailer** with **13 fast visual beats**.\nStyle: **Japanese anime opening × premium AAA motion graphics × fashion campaign**.\nThe video should feel **explosive, sexy, stylish, bold and high-impact**, with **80% motion graphics and 20% character action**.\n\n## CHARACTER LOCK — HIGHEST PRIORITY\nAO is a **young adult East Asian anime woman** with a sexy, confident Japanese anime aesthetic.\n\nShe has:\n* small refined face\n* sharp expressive crimson eyes\n* glossy lips\n* confident, teasing gaze\n* slim feminine curvy figure\n* long elegant legs\n* stylish, cool, alluring presence\nKeep her identical throughout:\n* chin-length vivid red bob with messy bangs\n* crimson-red eyes\n* black choker\n* fitted red cropped top\n* short white cropped jacket, worn open\n* black mini skirt or fitted shorts\n* red belt detail\n* black thigh strap\n* white-and-red platform sneakers\n* subtle silver accessories\nPreserve the same face, body proportions, hairstyle, outfit, materials and colors in every shot.\n**Never redesign AO. Never change her face or outfit. If a shot becomes too complex, simplify the action first.**\n## VISUAL STYLE\nPalette: **vivid red, crimson, white, black, silver**.\nUse:\ngiant kinetic typography, red circles, diagonal slashes, manga speed lines, split screens, halftone dots, barcode strips, UI ticks, freeze frames, RGB flashes, impact shakes, poster layouts and graphic wipes.\nEvery beat should feel:\n**fast, sharp, sexy, explosive, graphic and iconic**.\nKeep typography bold and readable.\nEditing:\nhard cuts, aggressive snap zooms, whip pans, speed ramps, freeze frames, impact shakes and foreground wipes.\n## 13 VISUAL BEATS\n**01 | 0.0–1.0s**\nWhite field. Massive red circle slams into frame. Black bars slash across. UI ticks flicker. Giant **A**, then **O**, hit with heavy impact shake.\n**02 | 1.0–2.0s**\nThe O becomes a circular frame showing an extreme close-up of AO’s crimson eye and glossy lips. She gives a teasing side glance. RGB flash. Circle bursts into red-and-white fragments.\n**03 | 2.0–3.1s**\nBlack background, huge white **AO**. AO enters fast, turns sharply and power-slides beneath the typography. Red speed streaks trail behind her. Whip-pan out.\n**04 | 3.1–4.0s**\nThree red/white/black vertical panels. AO appears in three poses: hip turn, hair touch, over-shoulder stare. Huge vertical **FULL SPEED** moves behind her.\n**05 | 4.0–5.1s**\nAO jumps through a rotating typography ring reading **NO BRAKES / ALL EYES ON ME**. One clean mid-air spin. Snap zoom into her confident face.\n**06 | 5.1–6.0s**\nWhite editorial frame. Huge black **HOT** with a red slash. AO crosses the frame with a runway-like step, one hand at her waist. Typography compresses and rebounds.\n**07 | 6.0–7.0s**\nBright red field with black diagonal stripe. AO performs one smooth fast turn. Three ghosted freeze positions trace the movement. Giant outlined **TURN** rotates behind her.\n**08 | 7.0–8.0s**\nWords hit one per beat:\n**HOT / FAST / WILD / RED**\nAO changes pose with each word: direct stare, hair toss, hip shift, confident forward lean.\n**09 | 8.0–9.0s**\nBlack frame with manga perspective lines and a graphic grid. AO steps forward and freezes in a strong hero pose. Red circular target graphics lock around her.\n**10 | 9.0–10.1s**\nAO moves toward camera through three red-and-white graphic panels. Each panel shatters as she passes. Large **A O** fragments appear behind her. Finish with a hair or leg foreground wipe.\n**11 | 10.1–11.1s**\nRapid poster montage: four frames of the same AO — close-up stare, walking, side pose, hands at waist. Add **01–04**, barcodes, halftone dots and sharp Japanese poster graphics.\n**12 | 11.1–13.0s**\nHero moment on a clean white background. AO lands in a powerful fashion pose: one leg forward, one hand at her waist, chin lifted, direct eye contact. Huge red shockwave rings explode behind her. Typography fragments and speed lines burst outward. Hold an iconic confident freeze.\n**13 | 13.0–15.0s**\nFinal identity card. Huge black **AO** on a bright white field. AO stands relaxed and alluring in front of the letters. Red circles, halftone, technical arcs and sharp speed accents surround her. Final red pulse flashes through the frame and ends on a hard stinger.\n## ANIME STYLE\nPremium modern Japanese anime rendering:\n\n* clean cel shading\n* sharp linework\n* polished highlights\n* cinematic close-ups\n* dynamic perspective\n* fashion-editorial full-body framing\n* smooth hair and fabric motion\nAO must look like a **stylish adult anime heroine**, not chibi and not childish.\n## AUDIO\n\nHard-hitting **electro / future bass / anime-opening style music**.\nUse:\nheavy drums, bass hits, risers, glitch fills, synth stabs, typography slams, whooshes and shutter impacts.\n\nBuild continuously.\nPeak at Beat 12.\nEnd with a sharp electronic stinger.\n## PRIORITIES\n1. AO identity consistency\n2. sexy adult anime character design\n3. red-black-white outfit consistency\n4. maximum visual impact\n5. readable typography\n6. premium anime rendering\n7. fast beat-synced editing\n\nAvoid:\nchildlike proportions, chibi style, blue clothing, face changes, outfit changes, extra characters, unreadable typography, weak motion, dull compositions or generic schoolgirl styling.\nFinal result: **explosive, sexy, red-hot, premium, graphic-driven and visually unforgettable.**",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2090634362631168129",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090634362631168129.mp4",
    "posterUrl": "/posters/x/2090634362631168129.jpg",
    "duration": 15.836009,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "时尚"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2090634362631168129"
  },
  {
    "id": "x-2090620553640198394",
    "title": "海岸骑行视频添加日文文字",
    "titleEn": "Adding Japanese Type to a Coastal Cycling Video",
    "model": "MiniMax H3",
    "mode": "Ref2VA",
    "summary": "参考骑行视频加入大字歌词；作者指出逐字出现的指令未达到预期。",
    "summaryEn": "Large lyric text is added to a cycling reference video; the creator notes that individual character timing did not follow the instruction.",
    "prompt": "Video editing, add texts into the <Video 1> while keep the layout, expressions, movement, color in <Video 1>. \nThe bold white sans-serif word \"この世界は\" and \"アシンメトリック\" snaps upward individually behind the girl driving bicycle and in front of background in sync with the timing of the <Audio 1> audio track titled <d>[Japanese] この世界はアシンメトリック</d>.",
    "sourceUrl": "https://x.com/MireilleDartois/status/2090620553640198394",
    "sourceLabel": "X 原帖",
    "author": "MireilleDartois",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090620553640198394.mp4",
    "posterUrl": "/posters/x/2090620553640198394.jpg",
    "duration": 5.226667,
    "aspectRatio": "9:5",
    "resolution": "480p",
    "tags": [
      "本地生成与工作流",
      "动漫",
      "动态图形",
      "城市街景"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text",
      "video",
      "audio"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/MireilleDartois/status/2090620553640198394"
  },
  {
    "id": "x-2090603098180132963",
    "title": "鸣人与雏田的樱花树下小插曲",
    "titleEn": "Naruto and Hinata beneath Cherry Blossoms",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "鸣人在樱花树下展示动作，随后与雏田产生轻松互动。",
    "summaryEn": "Naruto attempts a move beneath cherry blossoms before a light interaction with Hinata.",
    "prompt": null,
    "sourceUrl": "https://x.com/shuxxess/status/2090603098180132963",
    "sourceLabel": "X 原帖",
    "author": "shuxxess",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090603098180132963.mp4",
    "posterUrl": "/posters/x/2090603098180132963.jpg",
    "duration": 10.208,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090601706648412626",
    "title": "任务出发前整理领带",
    "titleEn": "Adjusting a Tie before a Mission",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "三笠替利威尔整理领带，近景表现两人的神态互动。",
    "summaryEn": "Mikasa adjusts Levi's tie while close-ups capture their expressions.",
    "prompt": null,
    "sourceUrl": "https://x.com/shuxxess/status/2090601706648412626",
    "sourceLabel": "X 原帖",
    "author": "shuxxess",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090601706648412626.mp4",
    "posterUrl": "/posters/x/2090601706648412626.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "对白与口型"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090592575627628702",
    "title": "巨型早餐桌上的微缩冒险",
    "titleEn": "A Miniature Adventure across a Breakfast Table",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "微缩人物穿越蓝莓、糖浆与煎饼构成的巨大早餐景观。",
    "summaryEn": "Tiny figures explore a breakfast landscape of giant blueberries, syrup and pancakes.",
    "prompt": null,
    "sourceUrl": "https://x.com/0nedi4ry/status/2090592575627628702",
    "sourceLabel": "X 原帖",
    "author": "0nedi4ry",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090592575627628702.mp4",
    "posterUrl": "/posters/x/2090592575627628702.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "美食",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "food",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090590155795234920",
    "title": "走廊里撞上透明墙的动漫角色",
    "titleEn": "An Anime Character Meets an Invisible Wall",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "动漫角色走过学校走廊，在透明障碍前停下并转身。",
    "summaryEn": "An anime character walks along a school corridor, stops at an invisible obstacle and turns.",
    "prompt": null,
    "sourceUrl": "https://x.com/sarischerika/status/2090590155795234920",
    "sourceLabel": "X 原帖",
    "author": "sarischerika",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090590155795234920.mp4",
    "posterUrl": "/posters/x/2090590155795234920.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090565921580425588",
    "title": "TOBILLO KID 卡通角色宣传片",
    "titleEn": "A TOBILLO KID Cartoon Character Promo",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "大眼卡通角色配合粗体标识与街头场景完成登场片段。",
    "summaryEn": "A wide-eyed cartoon character appears with bold branding and street settings.",
    "prompt": null,
    "sourceUrl": "https://x.com/wakeupgfx/status/2090565921580425588",
    "sourceLabel": "X 原帖",
    "author": "wakeupgfx",
    "publishedAt": "2026-08-20",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090565921580425588.mp4",
    "posterUrl": "/posters/x/2090565921580425588.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090556776156742071",
    "title": "分镜参考下的拿铁制作与讲解",
    "titleEn": "Latte Preparation and Speech from a Storyboard",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "咖啡制作动画与原分镜板对照，包含磨制、牛奶处理与倒注步骤。",
    "summaryEn": "Coffee-making animation appears beside its storyboard, including preparation, milk handling and pouring.",
    "prompt": "Use the storyboard  @[storyboard ref] as sequential shot guidance, not as a static image. Do not treat the storyboard as one image. Follow each panel as a separate beat. Use @[character ref] as character reference.\n\nShe talks while preparing the latte:\n\nまずはエスプレッソを淹れます。\nミルクをなめらかにスチームして、ゆっくり注ぎます。\n\nカップに近づけて、模様を描いていきます。\n\n最後にひと仕上げ。ラテアートの完成です。",
    "sourceUrl": "https://x.com/aimikoda/status/2090556776156742071",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-20",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090556776156742071.mp4",
    "posterUrl": "/posters/x/2090556776156742071.jpg",
    "duration": 15.139002,
    "aspectRatio": "540:607",
    "resolution": "1080p",
    "tags": [
      "本地生成与工作流",
      "动漫",
      "美食",
      "对白与口型"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food",
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/aimikoda/status/2090556776156742071"
  },
  {
    "id": "x-2090556760348323875",
    "title": "键盘与眼睛之间的动态文字",
    "titleEn": "Kinetic Type between Keyboards and Eyes",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "文字连接粉色键盘、人物与眼部特写，形成连续图形段落。",
    "summaryEn": "Type connects a pink keyboard, a performer and an eye close-up in a graphic sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2090556760348323875",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-20",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090556760348323875.mp4",
    "posterUrl": "/posters/x/2090556760348323875.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090556023178510671",
    "title": "美式复古工装拼贴广告",
    "titleEn": "An American Vintage Workwear Collage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "牛仔工装、旧杂志纸感与粗体标语组成复古服饰概念。",
    "summaryEn": "Denim workwear, aged magazine textures and bold slogans form a vintage fashion concept.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2090556023178510671",
    "sourceLabel": "X 原帖",
    "author": "su_nagomi",
    "publishedAt": "2026-08-20",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090556023178510671.mp4",
    "posterUrl": "/posters/x/2090556023178510671.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "复古胶片",
      "商业广告感",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "retro",
      "commercial"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090519106432983412",
    "title": "雨夜公交站的告别对白",
    "titleEn": "A Goodbye at a Rainy Bus Stop",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "男女在雨夜站台对话，镜头逐渐靠近表情与触碰脸颊的动作。",
    "summaryEn": "A couple speaks at a rainy bus stop as shots move closer to expressions and a touch on the cheek.",
    "prompt": "Create a fifteen second dialogue driven cinematic emotional scene at a quiet bus stop at night. Light rain falls under soft streetlights, creating an intimate and melancholic atmosphere. The acting should feel natural and restrained, with accurate lip sync and no overlapping dialogue. Characters: A is the boyfriend, age twenty five. He is trying to hide how deeply he is hurting. B is the girlfriend, age twenty four. She is emotional but composed, trying to hold herself together.\nFirst five seconds: B stands at the bus stop holding a small travel bag. Rain falls softly around them. A looks at her with watery eyes, trying not to break. A says softly “So… this is really goodbye?” B lowers her eyes. She struggles to find the words and takes a quiet breath before answering. B says “If I stay, I’ll keep choosing you over myself.” Seconds five to eleven A forces a small smile. His expression tries to hide the pain, but his voice begins to break. A says “Then go… I just wish loving me didn’t cost you so much.” B looks at him as her eyes slowly fill with tears. She takes one small step closer. Seconds eleven to fifteen B gently raises her hand and touches his cheek. B says quietly “You were never the mistake.” A closes his eyes. A single tear rolls down his cheek as he leans slightly into her touch. The bus arrives behind B, its headlights glowing through the rain. Neither of them moves. Use intimate close ups with subtle handheld camera movement. Hold on their eye contact and small facial reactions. Show trembling lips, restrained tears, realistic breathing, natural pauses, and quiet hesitation.\n\nLet facial expressions communicate as much emotion as the dialogue. Use cinematic night lighting, wet street reflections, shallow depth of field, and soft background bokeh. Keep both characters visually consistent throughout the scene and make every emotional reaction feel natural, subtle, and believable.",
    "sourceUrl": "https://x.com/bmx_ai13/status/2090519106432983412",
    "sourceLabel": "X 原帖",
    "author": "bmx_ai13",
    "publishedAt": "2026-08-20",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090519106432983412.mp4",
    "posterUrl": "/posters/x/2090519106432983412.jpg",
    "duration": 15.168,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "写实电影感"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/bmx_ai13/status/2090519106432983412"
  },
  {
    "id": "x-2090392284516974704",
    "title": "巨刃少女迎战长腿巨鸟",
    "titleEn": "A Giant-Blade Fighter Faces a Towering Bird",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "彩色插画式场景中，持巨刃人物与巨鸟展开动作交锋。",
    "summaryEn": "A giant-blade fighter confronts a towering bird in a colorful illustrated setting.",
    "prompt": null,
    "sourceUrl": "https://x.com/whorange__/status/2090392284516974704",
    "sourceLabel": "X 原帖",
    "author": "whorange__",
    "publishedAt": "2026-08-20",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090392284516974704.mp4",
    "posterUrl": "/posters/x/2090392284516974704.jpg",
    "duration": 27.051247,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "动漫",
      "插画与手绘",
      "动作打斗",
      "动物与生物"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090267753915498925",
    "title": "朋克企鹅与动感字宣传片",
    "titleEn": "A Punk Penguin Kinetic-Type Promo",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "卡通企鹅穿梭于粉绿底色、粗体文字与快速图形转场。",
    "summaryEn": "A cartoon penguin moves through pink-green panels, bold lettering and fast graphic transitions.",
    "prompt": null,
    "sourceUrl": "https://x.com/nett0eth/status/2090267753915498925",
    "sourceLabel": "X 原帖",
    "author": "nett0eth",
    "publishedAt": "2026-08-20",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090267753915498925.mp4",
    "posterUrl": "/posters/x/2090267753915498925.jpg",
    "duration": 13.722993,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "动态图形",
      "动物与生物"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090161526380347863",
    "title": "戴上耳机后随节拍变化的街道",
    "titleEn": "A Street Responds to a Headphone Beat",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物戴着耳机穿行街道，建筑与道路产生音乐可视化式变化。",
    "summaryEn": "A headphone-wearing figure walks through a street as buildings and road surfaces shift like a music visualization.",
    "prompt": "A man puts on headphones while walking through a dull grey city. Beat begins. First kick: A streetlamp bends slightly. Second kick: All parked cars bounce simultaneously. Snare: Every window shutter snaps open. Bass: The asphalt physically ripples beneath him. Melody enters: Buildings begin stretching upward and downward like an architectural equalizer. Now EVERYTHING responds musically. Traffic lights dance between positions. Balconies slide rhythmically across façades. Road markings travel like musical waveforms. Trees pulse like dancing. Windows illuminate across entire buildings at the rhytm of the music. But the pedestrians behave completely normally, apparently unaware that the entire city has become a music visualizer. Then he removes the headphones. Instantly, half the street becomes normal. Puts it back.BOOM. The whole city hits the beat again.",
    "sourceUrl": "https://x.com/alexgnewmedia/status/2090161526380347863",
    "sourceLabel": "X 原帖",
    "author": "alexgnewmedia",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090161526380347863.mp4",
    "posterUrl": "/posters/x/2090161526380347863.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "城市街景",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/alexgnewmedia/status/2090161526380347863"
  },
  {
    "id": "x-2090107555712512053",
    "title": "THE DEAD DROP 复古谍战片头",
    "titleEn": "THE DEAD DROP Retro Spy Titles",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "网格、钥匙、红黑图形与片名字卡组成复古主标题序列。",
    "summaryEn": "Grids, a key, red-black graphics and title cards form a retro main-title sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2090107555712512053",
    "sourceLabel": "X 原帖",
    "author": "HBCoop_",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090107555712512053.mp4",
    "posterUrl": "/posters/x/2090107555712512053.jpg",
    "duration": 25.791667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "复古胶片",
      "动态图形",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090104453684748398",
    "title": "客厅里的电脑工作对白",
    "titleEn": "A Living-Room Conversation about Computer Work",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物围绕电脑工作展开客厅对话，近景交替展示双方反应。",
    "summaryEn": "Two people discuss computer work in a living room with alternating reaction shots.",
    "prompt": null,
    "sourceUrl": "https://x.com/GamerHashCom/status/2090104453684748398",
    "sourceLabel": "X 原帖",
    "author": "GamerHashCom",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090104453684748398.mp4",
    "posterUrl": "/posters/x/2090104453684748398.jpg",
    "duration": 10.24,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实",
      "日常生活"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090091561506009588",
    "title": "卧室人物与枪械的移动镜头",
    "titleEn": "Moving Shots around a Bedroom Character and a Gun",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头从枪械细节移动至床上人物，展示参考图驱动的镜头组织。",
    "summaryEn": "The camera moves from a gun detail to a character on a bed in an image-driven shot sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/CharaspowerAI/status/2090091561506009588",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090091561506009588.mp4",
    "posterUrl": "/posters/x/2090091561506009588.jpg",
    "duration": 15.168,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2090070695481143430",
    "title": "暖色公寓冰咖啡 UGC 广告",
    "titleEn": "An Iced Coffee UGC Ad in a Warm Apartment",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "持瓶展示、冰咖啡混合特写与人物试饮串起产品片段。",
    "summaryEn": "A bottle presentation, iced-coffee mixing close-up and tasting shots build a product sequence.",
    "prompt": "Use the uploaded image as the exact character reference. Keep her face, hair, makeup, body, orange top, white pajamas, and jewelry consistent throughout. Keep the iced coffee bottle identical in every shot, including its label, colors, shape, and packaging.\n\nCreate a 15-second ultra-realistic premium UGC iced coffee commercial in a warm modern apartment. She smiles at the camera while holding the bottle, then shows the condensation-covered bottle in a macro shot. She opens it, pours the iced coffee over ice, and captures the coffee and milk swirling together. She takes a refreshing sip and gives the camera a satisfied thumbs-up. Finish with the bottle beside coffee beans on a wooden table as she smiles beside it and says, “Good coffee. Good mood. Anytime.”\n\nWarm natural window light, realistic textures, smooth handheld camera, cinematic macro shots, shallow depth of field, premium lifestyle aesthetic, photorealistic 4K. No subtitles, no watermarks, no extra text, no product changes.",
    "sourceUrl": "https://x.com/AIwithSynthia/status/2090070695481143430",
    "sourceLabel": "X 原帖",
    "author": "AIwithSynthia",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090070695481143430.mp4",
    "posterUrl": "/posters/x/2090070695481143430.jpg",
    "duration": 14.144,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "生活实拍感",
      "照片级写实",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AIwithSynthia/status/2090070695481143430"
  },
  {
    "id": "x-2090029366205960465",
    "title": "飞越巴黎的巨龙视角双模型对比",
    "titleEn": "A Two-Model Dragon Flight over Paris Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "上下画面对比 H3 与 Seedance 的巴黎飞行场景，原帖使用同一条简短提示。",
    "summaryEn": "Stacked views compare H3 and Seedance on a Paris flight scene using the same short instruction.",
    "prompt": "Dragon POV: Ancient Paris to Modern Paris.",
    "sourceUrl": "https://x.com/codewithhajra/status/2090029366205960465",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090029366205960465.mp4",
    "posterUrl": "/posters/x/2090029366205960465.jpg",
    "duration": 15.125333,
    "aspectRatio": "4:5",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "写实电影感",
      "城市街景",
      "奇幻"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2090029366205960465"
  },
  {
    "id": "x-2089948821245788584",
    "title": "北村韩屋街道旅行短片",
    "titleEn": "A Bukchon Hanok Village Travel Clip",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物沿韩屋街道步行、举起手机拍照并在高处看屋顶景观。",
    "summaryEn": "A traveler walks along hanok streets, takes a phone photo and looks across the rooftops.",
    "prompt": "Create a 15-second, 16:9 photorealistic Korean travel-vlog video in natural daylight. No narration, golden-hour lighting, CGI look, or artificial movement.\nCharacter: Same Korean woman, mid-20s, throughout. Cream cardigan, white blouse, long beige skirt, white sneakers, brown crossbody bag, smartphone. Natural hair movement, expressions, blinking, breathing, and realistic body proportions.\n0–5s — Walking: Smooth side/back tracking shot as she walks naturally uphill on a narrow stone-paved street in Bukchon Hanok Village, surrounded by authentic hanok houses, tiled roofs, wooden doors, and white walls. Realistic foot placement, weight transfer, posture, and arm movement. She looks at the rooftops and says: “This place feels so peaceful.” Finish dialogue before stopping.\n5–10s — Photographing: She stops, takes the phone from her bag, holds it with both hands, frames the hanok rooftops, makes one small adjustment, taps the shutter once, hears a subtle shutter sound, lowers the phone, views the photo, and smiles. Keep this exact order: stop → phone → frame → tap → lower → react.\n10–15s — Viewpoint: She puts the phone away and continues uphill. Front tracking transitions into a slow wide pullback as she reaches a wooden-railing viewpoint. Tourists move naturally and independently in the background. She rests one hand on the railing, looks over the rooftops, turns slightly toward camera, and says: “I love this view.” End with a wide view of her, hanok rooftops, trees, street, and distant Seoul skyline.\nCamera: Professional travel-vlog cinematography; smooth stabilized movement with realistic inertia. No sudden zooms, cuts, excessive shake, drone-like movement, or slow motion.\nAudio: Natural footsteps, light wind, rustling leaves, distant conversations, birds, neighborhood ambience, and subtle phone shutter. Accurate English lip sync.\nPhysics/continuity: Realistic walking and stopping, correct hand/finger interaction, stable phone, gentle breeze, consistent shadows, natural pedestrians, no collisions, and no disappearing/duplicating objects. Hanok architecture must remain geometrically consistent.\nNegative: No CGI/plastic skin, robotic motion, foot sliding/floating, distorted hands, extra fingers, teleportation, unrealistic pedestrians, exaggerated wind, weather changes, golden-hour lighting, excessive blur, text, watermark, or logos.\nGoal: Genuine footage of a Korean lifestyle/travel vlogger casually exploring Bukchon, with natural acting, continuous movement, believable physics, and authentic Seoul ambience.",
    "sourceUrl": "https://x.com/nawalsehar/status/2089948821245788584",
    "sourceLabel": "X 原帖",
    "author": "nawalsehar",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089948821245788584.mp4",
    "posterUrl": "/posters/x/2089948821245788584.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "城市街景",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/nawalsehar/status/2089948821245788584"
  },
  {
    "id": "x-2089925065492267106",
    "title": "土豆切片到薯片包装的广告",
    "titleEn": "From Potato Slices to a Packaged Chips Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "土豆切片、薯片飞散与红色包装组成食品加工概念广告。",
    "summaryEn": "Potato slicing, flying chips and red packaging form a food-process advertising concept.",
    "prompt": "Ultra-realistic cinematic 3D commercial, 15 seconds, hyper-detailed food commercial, dynamic speed ramp, macro photography, dramatic studio lighting, ultra-fast transitions, motion blur, shallow depth of field, realistic physics, crisp textures, premium advertising quality, 16:9.\n\nSequence: A whole fresh potato spins at incredible speed in mid-air above a modern kitchen counter. A stationary razor-sharp chef's knife instantly peels the potato as the spinning motion sends the skin flying away in elegant spiral ribbons. The perfectly peeled potato freezes vertically in mid-air, then effortlessly slices itself into dozens of ultra-thin, perfectly even potato chips suspended for a brief moment. The slices glide one by one into a vat of bubbling hot vegetable oil, creating dramatic splashes and shimmering golden bubbles. Within seconds, the chips become perfectly golden, light, and crispy. They burst upward from the oil in slow motion and land on a rustic wooden table as a vibrant cloud of savory seasoning explodes around them. An automated packaging machine immediately fills a sleek, glossy Crunch Valley potato chips bag, injects nitrogen gas for freshness, and heat-seals it. The packaging features a bold modern Crunch Valley logo in white lettering with gold accents on a vibrant red bag, premium branding, realistic printed details, and a luxurious commercial design. Final hero shot: the perfectly sealed Crunch Valley potato chips bag stands centered on the table while the camera performs a fast cinematic push-in through floating seasoning particles under warm cinematic lighting, with ultra-realistic reflections, razor-sharp focus, premium food advertising aesthetics, and a dramatic end frame held for maximum impact.",
    "sourceUrl": "https://x.com/ZaraIrahh/status/2089925065492267106",
    "sourceLabel": "X 原帖",
    "author": "ZaraIrahh",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089925065492267106.mp4",
    "posterUrl": "/posters/x/2089925065492267106.jpg",
    "duration": 15.25551,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "3D / CG",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ZaraIrahh/status/2089925065492267106"
  },
  {
    "id": "x-2089918524118716461",
    "title": "折纸人物与折叠文字音乐片",
    "titleEn": "An Origami Figure with Folding Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "几何纸雕人物与黑红文字共同折叠、展开并重组画面。",
    "summaryEn": "A geometric paper figure and black-red type fold, unfold and rearrange the composition.",
    "prompt": "【REFERENCE】\n参照画像なし。参照動画なし。\n\n【CONDITION DEFINITION】\n15秒、1:1スクエア。高品質な3Dペーパースカルプチャー表現によるオリジナルMV。\n芸術的な人型折り紙彫刻、リリックモーション、モーショングラフィックの三要素を同格の主役として融合する。折り紙だけを常に前景へ固定せず、リリック、折り紙、MGの視覚的な優先順位を場面ごとに切り替え、一つの完成した紙の造形世界を作る。\n折り紙人物は、背が高く細身で中性的な人型ペーパースカルプチャー。小さな頭部、細い腰、長い脚、左右非対称の肩面、鋭い折り面、腰や背中から伸びる紙帯、背中の扇状構造を持つ。顔も折り面で構成し、目鼻口は最小限の記号表現。身体そのものをアヴァンギャルドなオートクチュール彫刻として見せる。\n素材はアイボリー、ディープクリムゾン、マットブラック、メタリックゴールドを中心に、少量のターコイズと半透明紙を使用。折り目、紙のエッジ、重なり、厚み、素材差を明確に見せる。\n背景とMGも同じ紙素材の造形言語で統一し、巨大紙面、切り抜いた円、紙帯、折り線、グリッド、幾何形状、紙の影、タイポグラフィで構成する。\n使用する歌唱フレーズと画面文字は以下の3つだけ。\n「FOLD THE LIGHT」\n「SHAPE THE SILENCE」\n「BECOME ART」\n各フレーズは、最初に文字全体を遮蔽せず明瞭に読める状態で歌唱と同期して提示する。その後、文字を分割、折り、スライド、拡大し、紙帯や幾何形状へ変換してMGへ融合する。歌唱後の単語断片はMG素材として再利用できる。\n人物は激しく踊らない。わずかな向き変化、片腕を開く、一歩進む程度に留める。\n\n【SHOT / FLOW】\n0.00-3.00｜BIRTH\n巨大なアイボリー紙面に黒い折り線が走り、紙面が鋭く開いてクリムゾンとゴールドを露出。その中心から人型折り紙彫刻が現れる。まず「FOLD THE LIGHT」を、人物やMGを重ねず、文字だけが主役のクリーンな版面で全文可読の状態として歌唱と同期して提示する。その後に折り紙人物を画面へ戻し、文字の一部が折れて紙面とMGへ変化する。\n3.00-6.00｜FORM\n全身シルエット、肩の折り構造、金箔の折り目、抽象的な横顔をテンポよく見せる。人物はわずかに向きを変える。まず「SHAPE THE SILENCE」を、人物やMGに遮られない大きな文字として全文可読で歌唱と同期して提示する。その後、文字が分解され、円、紙帯、折り線へ再構成される。人物と文字の位置関係を変え、人物を常に文字の前へ置かない。\n6.00-9.00｜MOTION\nMG最大密度。折り紙人物、文字断片、紙の円、グリッド、長い紙帯、大胆な色面がビートに合わせて版面を高速再構成する。人物は中央固定にせず、左右や余白へ配置を変化させる。前2フレーズの単語断片をグラフィック素材として使用し、新しい歌詞は追加しない。人物自身は片腕を少し開く程度に抑える。\n9.00-12.00｜ART\n速度を落とし、折り紙造形そのものを美しく見せる。人型折り紙彫刻本体の頭部、首、胴体、腕、腰、脚による明確な人型シルエットは維持し、人物自身は鳥や翼のある生物へ変化させない。代わりに、人物の背後または周囲で、独立した巨大な扇状ペーパースカルプチャーと半透明紙の構造がゆっくり展開する。半透明紙と金箔の重なりを見せる。「BECOME ART」を遮蔽のない大きな文字として完全表示し、歌唱と同期。人物、文字、紙構造を一枚のアートエディトリアルとして統合する。\n12.00-15.00｜ICON\n再加速。「FOLD」「SHAPE」「ART」の文字断片、紙面、円、折り線、紙帯を高速に再構成する。リリック、折り紙人物、MGの前後関係と画面占有率を切り替えながら、どれか一要素だけを固定した主役にしない。最後は人型折り紙彫刻と大きな「BECOME ART」、幾何学的な紙MGを一つの象徴的な版面へ収束。完全停止せず、紙帯や文字の微細な動きを最後まで残す。\n\n【CAMERA / EDITING】\n全身、ミディアム、横・斜めシルエット、折り面・金箔・半透明紙の接写を使う。過剰な3D回転は避け、ファッションエディトリアルのような強い版面構成を優先する。人物を常に中央へ固定せず、リリックの可読性とMGの構図に合わせて配置を変える。\n\n【SOUND】\n洗練された高エネルギーの現代的MVサウンド。3つの英語フレーズを明瞭に歌唱し、完全な文字表示と同期する。主要ビートに紙面展開、文字変形、紙帯、円、版面切替を合わせる。9-12秒のみ密度を落とし、12秒から再加速する。\n\n【NEGATIVE】\n複数人物、途中の人物デザイン変更、実写人物・通常アニメ・ロボット・ローポリCG化を禁止。完全なリリック表示を人物やMGで隠さない。ART区間でも人型シルエットを崩さない。",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2089918524118716461",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-08-19",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089918524118716461.mp4",
    "posterUrl": "/posters/x/2089918524118716461.jpg",
    "duration": 15.13941,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "3D / CG",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2089918524118716461"
  },
  {
    "id": "x-2089840714016497677",
    "title": "绿光能量盒封住沙地蝎子",
    "titleEn": "A Green Energy Box Captures a Desert Scorpion",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "绿黑战衣人物用发光能量盒包住蝎子，展示人物与特效连续动作。",
    "summaryEn": "A green-and-black suited character encloses a scorpion in a glowing energy box.",
    "prompt": "integrated_multimodal_description:\n\n<Image 1> is the CHARACTER APPEARANCE REFERENCE.\n\n<Image 0> background reference\n\nUse <Image 1> to preserve the man's exact facial identity, age, dark hair with gray temples, rugged stubble, body proportions, green-and-black tactical suit, illuminated chest emblem, and power-ring design.\n\nCRITICAL RIGHT-HAND CONTINUITY:\nThe power ring is worn ONLY on the man's anatomical RIGHT HAND throughout the entire video.\nNEVER place the ring on his left hand.\nDo not mirror or swap hands when the camera angle changes.\nHis LEFT HAND remains completely ringless throughout the entire sequence.\nEvery green energy projection originates directly from the ring on his RIGHT HAND.\n\nPhotorealistic live-action cinematic realism.\n\nEnvironment: vast isolated American desert-like prairie, dry golden grass, dusty earth, scattered low scrub, distant low hills, huge open sky. Warm late-afternoon sunlight. No buildings, vehicles, roads, or other people visible.\n\n[00:00–00:01.5]\n\nWide cinematic tracking shot.\n\nThe man from <Image 1> walks casually through the open landscape.\n\nHis arms swing naturally as he walks.\n\nHe suddenly notices something near his feet, slows, stops, and looks down.\n\nCamera subtly moves closer and tilts downward with his gaze.\n\n[00:01.5–00:02.5]\n\nCut to a low close shot near the ground.\n\nA small realistic BLACK SCORPION crawls slowly across the dusty ground approximately two feet in front of his boots.\n\nCut back to a medium shot as he looks down at it with mild curiosity rather than fear.\n\nHe raises his anatomical RIGHT HAND and points the ring toward the scorpion.\n\nHis left hand remains relaxed and completely ringless.\n\n[00:02.5–00:03.5]\n\nMedium close-up favoring his RIGHT hand.\n\nThe ring activates with concentrated emerald-green light.\n\nA narrow, clearly defined GREEN ENERGY BEAM shoots directly from the ring on his RIGHT HAND toward the scorpion.\n\nThe beam strikes the ground immediately around the scorpion.\n\nGlowing green construct lines rapidly form around it: bottom edges → four vertical corners → horizontal edges → top edges.\n\nWithin moments, these lines complete a small transparent emerald-green ENERGY CAGE surrounding the scorpion.\n\nThe scorpion remains unharmed and clearly visible inside.\n\nCRITICAL:\nFrom this moment until the END OF THE VIDEO, the green energy beam remains visibly and continuously connected from the ring on his RIGHT HAND directly to the green cage.\nThe beam NEVER disconnects, disappears, switches hands, or originates anywhere else.\n\n[00:03.5–00:05.0]\n\nLow close-up of the cage.\n\nStill connected to the continuous beam, the completed green construct cage gently rises from the dusty ground.\n\nThe scorpion remains standing on the glowing construct floor inside the cage.\n\nCamera tilts upward with it.\n\nThe cage floats smoothly to approximately chest height in front of the man.\n\nThe beam remains clearly visible between his RIGHT-HAND ring and the floating cage throughout the lift.\n\n[00:05.0–00:08.0]\n\nCut to a medium close-up framing the man's face, his extended RIGHT hand, and the floating cage.\n\nThe cage hovers roughly two feet in front of him at chest level.\n\nHis RIGHT arm remains comfortably extended toward it.\n\nThe glowing ring is clearly visible on his RIGHT HAND.\n\nThe continuous green beam remains visibly connected:\nRIGHT-HAND RING → GREEN BEAM → FLOATING CAGE.\n\nThe little black scorpion moves slightly inside the cage.\n\nThe man studies it for a moment, gives a small friendly amused smile, and speaks casually to the scorpion.\n\n(S1) <d>[EN]How you doing there, little fella?</d>\n\nHold briefly on his smile and the floating scorpion cage.\n\nThe cage remains stable.\nThe scorpion remains safely inside.\nThe beam remains connected through the FINAL FRAME.\n\nNo attack.\nNo explosion.\nNo additional constructs.\nNo additional characters.\n\noverall_soundscape:\n\nNatural open-land ambience with gentle dry wind moving through grass, subtle footsteps on dusty earth at the beginning, extremely faint scorpion movement during close shots, and a clean low emerald-energy hum when the ring activates.\n\nThe construct produces a subtle crystalline energy-building sound as the cage forms.\n\nOnce formed, the cage and continuous connecting beam produce a quiet sustained energy hum through the final frame.\n\n(S1)'s dialogue is natural, relaxed, friendly, and slightly amused.\n\nnon_diegetic_music:\n\nNone.",
    "sourceUrl": "https://x.com/EndFolding79421/status/2089840714016497677",
    "sourceLabel": "X 原帖",
    "author": "EndFolding79421",
    "publishedAt": "2026-08-18",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089840714016497677.mp4",
    "posterUrl": "/posters/x/2089840714016497677.jpg",
    "duration": 8.064,
    "aspectRatio": "40:23",
    "resolution": "736p",
    "tags": [
      "动作特效",
      "写实电影感",
      "科幻太空",
      "动物与生物"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2089840714016497677"
  },
  {
    "id": "x-2089764797361582208",
    "title": "三人物遮挡衔接环绕长镜头",
    "titleEn": "A Three-Character Orbit with Occlusion Transitions",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "近距离镜头沿人物移动，以身体遮挡衔接到下一位角色。",
    "summaryEn": "A close camera moves around characters and uses body occlusion to reveal the next person.",
    "prompt": null,
    "sourceUrl": "https://x.com/LoveUolanda/status/2089764797361582208",
    "sourceLabel": "X 原帖",
    "author": "LoveUolanda",
    "publishedAt": "2026-08-18",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089764797361582208.mp4",
    "posterUrl": "/posters/x/2089764797361582208.jpg",
    "duration": 14.976,
    "aspectRatio": "240:433",
    "resolution": "480p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2089599996085801459",
    "title": "动漫角色讲解 AI 视频学校概念广告",
    "titleEn": "An Anime AI Video School Concept Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红发角色、青粉图形与日文标题组成虚构学校的宣传视频。",
    "summaryEn": "A red-haired character, cyan-pink graphics and Japanese titles promote a fictional school.",
    "prompt": "subject_definitions:\n<Subject 1> is the anime girl in <Picture 1>: Japanese TV-anime style, late teens, long vivid\nred hair with loose flowing strands, large emerald-green eyes, white hoodie over a white top.\n<Subject 2> is the graphic style in <Picture 1>: crisp flat 2D vector overlays on white -\ncyan-turquoise and magenta-pink triangles, dot grids, diagonal speed shards, glowing UI\npanels, thin rounded icons.\n<Subject 3> is the school wordmark in <Picture 1>: the Japanese logotype \"デジラク\" in a heavy\nrounded gothic face, teal-green with a soft drop shadow.\n\nsummary:\n[reference generation] A 15-second Japanese commercial for an online AI-video creator school,\ndrawn as clean anime-style 2D animation over fast smooth motion graphics with animated\nJapanese type, carried by a female voiceover. <Subject 1> discovers AI video, learns, and\ninvites the viewer to start. <Picture 1> is a storyboard contact sheet used only for casting\nand style.\n\nretention_analysis:\n<Subject 1> (appears in [Shot 1], [Shot 3], [Shot 5], [Shot 8] and [Shot 10]):\npartially_preserved - her red hair, emerald-green eyes, face and white hoodie stay identical;\nonly pose and expression change.\n<Subject 2> (appears in every shot): weak_reference - only the white palette, flat vector feel\nand speed-shard energy carry over; the storyboard's grid, borders and numbers are not\nreproduced.\n<Subject 3> (appears in [Shot 9]): fully_preserved - the letterforms, teal-green color and\nproportions are kept exactly.\n\ndetailed_description:\n[Shot 1] Clean Japanese TV-anime 2D animation over crisp flat vector motion graphics with\nanimated Japanese type, cut like a fast bright commercial. On a white field <Subject 1> runs\nin from the right toward camera, hair streaming, as cyan and magenta shards burst past her.\nCharcoal type slides in line by line on the left reading \"AIで、アイデアが動き出す！\",\n\"動き出す\" snapping to double size in cyan. A bright Japanese woman in her early twenties with\na light buoyant friendly voice (S1) says off-screen:\n<d>[Japanese] AIで、アイデアが動き出す。</d> Push In, small amplitude, fast speed.\n[Shot 2] At 00:01.500, a cyan shard wipes across. A dozen glowing video panels fly through\ndeep blue space in one curved stream, each holding a different anime, cityscape or fantasy\nscene. White type assembles in two stages at the lower left reading\n\"こんな動画も…\". (S1): <d>[Japanese] こんな動画も。</d> Truck Right, large amplitude, fast\nspeed.\n[Shot 3] At 00:03.000, a white field wipes in. Close on <Subject 1>, hands clasped at her\nchest, eyes widening into a sparkle as four-point stars pop in around her. Bold type\nscales up hard on the left reading \"私にも 作れるかも！\", \"私\" landing in magenta-pink.\n<Subject 1>, a younger softer close-mic voice (S2), says quietly to herself:\n<d>[Japanese] 私にも、作れるかも。</d> Push In, small amplitude, slow speed.\n[Shot 4] At 00:04.500, a clean white field wipes in. Charcoal type draws itself across the top\nreading \"デジラクで学べること\". Three thin rounded icons then pop in one at a time across the\nmiddle - a laptop, an AI chip, a lightbulb - each with a label rising\nbeneath: \"AI動画の基礎\", \"最新AIツール\", \"アイデアの発想法\". No voice here. Static Shot.\n[Shot 5] At 00:06.000, a cyan shard wipes in. <Subject 1> sits at a wide monitor, typing, as\nglowing UI panels, timelines and waveform bars snap into place around the screen. A teal\nbanner sweeps in from the left with white type reading \"ゼロからでも大丈夫！\". (S1):\n<d>[Japanese] ゼロからでも、大丈夫。</d> Truck Left, small amplitude, slow speed.\n[Shot 6] At 00:07.500, diagonal cyan and magenta bands sweep across a white field. Huge\ncharcoal type slams in from both sides and collides at center reading\n\"プロの現場の技術が身につく！\", \"プロの現場\" in cyan with a thick underline drawing itself as\nlight shards scatter. (S1): <d>[Japanese] プロの技術が、身につく。</d> Static Shot.\n[Shot 7] At 00:09.000, the frame splits along one steep diagonal: a dull gray unfinished panel\non the left slides toward a bright finished anime artwork on the right as a glowing arrow\npushes between them. White type streaks in along the diagonal in two stages reading\n\"あなたの作品が、\" then \"ここから変わる。\". (S1): <d>[Japanese] あなたの作品が、変わる。</d>\nPush In, small amplitude, fast speed.\n[Shot 8] At 00:10.500, a bright sky-blue field with radiating light wipes in. <Subject 1>\nreaches one hand straight toward camera, smiling wide, as cyan and magenta shards streak past\nher. Bold white type assembles in two stages on the right reading \"一緒に、\" then \"つくろう！\".\n(S2), now bright and open: <d>[Japanese] 一緒に、つくろう！</d> Pull Out, small amplitude,\nfast speed.\n[Shot 9] At 00:12.000, a clean white field wipes in. Small charcoal type fades up across the\ntop reading \"AI動画クリエイター育成スクール\". Scattered letters then fly in and lock together\ninto <Subject 3> at center, a light sweep passing over it. Three thin rounded\nicons pop in one at a time across the lower third - a document, a person, a group - with\nlabels beneath: \"実践課題\", \"プロが指導\", \"コミュニティ\". (S1):\n<d>[Japanese] AI動画スクール、デジラク。</d> Static Shot.\n[Shot 10] At 00:13.500, a white field with cyan and magenta shards wipes in. <Subject 1>\nstands on the right pointing forward toward camera with a bright smile. Bold charcoal type\nscales up on the left reading \"今すぐ無料体験へ！\", and a rounded teal button assembles beneath\nreading \"公式サイトはこちら\", pulsing once. (S1), a half step warmer and more forward:\n<d>[Japanese] 今すぐ、無料体験へ。</d> The frame holds still until exactly 15.000 seconds.\nStatic Shot.\nThroughout: keep one clean Japanese TV-anime style with crisp linework, soft cel shading and\nbright color - never photorealistic, never 3D-rendered. Only (S1) and (S2) speak, exactly the\nlines above and nothing else: no ad-libs, no extra narration, no singing, no lip sync except\nin [Shot 3] and [Shot 8]. Every on-screen line is bold rounded Japanese gothic type that\nbuilds as animated shapes - never a movie subtitle, caption strip, letterboxed caption, footer\ntext or romaji. No storyboard grid, borders, numbers or contact-sheet layout, and no split\nscreen except the diagonal in [Shot 7]. Keep <Subject 1>'s red hair and emerald-green eyes\nidentical, cut sharp with no dissolves, keep motion smooth, and hold the palette to\ncrisp white, cyan-turquoise, magenta-pink, sky blue and charcoal.\n\noverall_soundscape:\nBright airy whooshes carry the shards and flying panels, and soft digital clicks mark each\nicon and UI element as it snaps into place. A rising sparkle lifts as her eyes widen, a solid\nimpact lands when the type collides, and keyboard taps run under the editing shot. A\nclean chime rings once on the button at the end.\n\nnon_diegetic_music:\nAn original 15-second bright uplifting Japanese commercial track at a fast tempo: plucked\nsynth lead, four-on-the-floor kick, tight claps, airy pads, bells, mixed low under the voiceover. It opens at full drive, drops the drums for one bar at 00:03.000 leaving\nonly a sparkling bell, snaps back at 00:04.500 and builds, lifts into a rising sweep at\n00:10.500, opens widest at 00:12.000, then lands on one bright sustained chord that decays\ncleanly at 15.000 seconds.",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2089599996085801459",
    "sourceLabel": "X 原帖",
    "author": "tanabe_fragm",
    "publishedAt": "2026-08-18",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089599996085801459.mp4",
    "posterUrl": "/posters/x/2089599996085801459.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2089599996085801459"
  },
  {
    "id": "x-2089578529252167841",
    "title": "未来工作室中的 AI 创意广告",
    "titleEn": "An AI Creativity Ad in a Futuristic Studio",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在未来工作台前介绍创意生成，场景切换为时装与城市画面。",
    "summaryEn": "A presenter introduces creative generation at a futuristic workstation before fashion and city visuals appear.",
    "prompt": "A stylish young South Asian woman with long dark hair, natural glamorous makeup and an elegant modern outfit. Keep her face, hairstyle, outfit and appearance consistent throughout the entire video.\n\nScene 1 — 0–4s:\nShe sits at a sleek modern workstation in a futuristic studio. She looks directly into the camera and says naturally:\n\n“What if you could turn your ideas into reality in just a few clicks?”\n\nNatural facial expressions and hand gestures. Slow cinematic camera push-in. Perfect lip-sync.\n\nScene 2 — 4–7s:\nShe turns to her computer and opens WeryAI. She types a creative idea and says:\n\n“That’s exactly what I love about WeryAI.”\n\nThe interface responds with elegant AI processing animations and glowing digital effects.\n\nScene 3 — 7–11s:\nHer simple idea transforms into stunning AI-generated visuals: a fashion scene, futuristic city, cinematic character, premium product advertisement and social-media video.\n\nShe looks impressed and says:\n\n“Give it an idea, and watch it come to life.”\n\nUse smooth cinematic transitions and dynamic camera movement.\n\nScene 4 — 11–15s:\nShe turns back toward the camera with a confident smile and says:\n\n“WeryAI. Your imagination, powered by AI.”\n\nThe camera slowly pulls back as the WeryAI brand reveal appears with a clean futuristic glow.\n\nAudio:\nNatural confident female voice, warm energetic delivery, subtle futuristic background music and soft cinematic sound effects. Keep dialogue clear and prominent.\n\nLip-sync:\nThe woman must visibly speak every line with accurate lip synchronization, natural mouth movement, blinking and facial expressions.\n\nVisual quality:\nPhotorealistic, cinematic 4K, realistic skin, smooth camera movement, premium lighting, shallow depth of field, consistent character.\n\nNo subtitles, no distorted face, no warped hands, no extra fingers, no flickering, no random text, no watermark.",
    "sourceUrl": "https://x.com/CaliraVal/status/2089578529252167841",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-18",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089578529252167841.mp4",
    "posterUrl": "/posters/x/2089578529252167841.jpg",
    "duration": 15.168,
    "aspectRatio": "629:360",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "科幻太空",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2089578529252167841"
  },
  {
    "id": "x-2089573433806098785",
    "title": "白猫暑假日历翻页动画",
    "titleEn": "A White Cat Summer Calendar Animation",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "桌面日历逐页呈现白猫、金鱼与烟花等水彩暑假场景。",
    "summaryEn": "A desk calendar turns through watercolor summer scenes with a white cat, goldfish and fireworks.",
    "prompt": "integrated_multimodal_description: [Shot 1] 15秒間の一続きの映像。「白猫の6日間の夏休み」を描く、可愛い水彩絵本風の日めくりカレンダー。 夏の午後の日差しが入る、涼しげな日本家屋。木製の机の中央に、二つの小さな金属リングで綴じた卓上型の日めくりカレンダーが置かれている。 窓辺には透明な青い風鈴。背景には緑の葉と眩しい夏空が柔らかくぼけて見える。机の端には青紫色の朝顔を一輪だけ置く。カレンダーを遮る物は置かない。 カメラはカレンダーを真正面から撮影したまま固定する。カメラ位置、焦点距離、カレンダーの位置、金属リング、紙の寸法を最後まで変えない。 カレンダーは少しざらついた生成り色の厚紙。全ページで書体、文字サイズ、日付位置、曜日位置、余白を完全に統一する。 今回存在するページは次の6枚だけ。 1枚目：「2026年8月 13 木」 2枚目：「2026年8月 14 金」 3枚目：「2026年8月 15 土」 4枚目：「2026年8月 16 日」 5枚目：「2026年8月 17 月」 6枚目：「2026年8月 18 火」 必ず、 「13 木」→「14 金」→「15 土」→「16 日」→「17 月」→「18 火」 の順番で一枚ずつ進める。 日付を飛ばさない。順番を変えない。逆戻りしない。同じ日付を二度表示しない。存在しない日付を追加しない。数字を別の数字へ変形させない。 各日付は、それぞれ別の紙の表面へ最初から印刷されている。日付は物理的なページめくりによってのみ切り替わる。 紙の裏面は、文字も絵も印刷されていない完全な生成り色の無地。めくられている紙の裏側へ、日付、曜日、数字、絵、鏡文字、反転文字を表示しない。 一度に読める日付は必ず一つだけ。次の日付は、上の紙が完全にめくられた後で現れる。途中の日付や余分なページを生成しない。 各ページ上部の左側に小さく「2026年8月」、中央に大きな日付数字、右側に曜日を一文字だけ表示する。「土」は青、「日」は赤、それ以外は濃いチャコールグレー。 日付以外の文章、タイトル、ロゴ、透かし、字幕、無関係な文字は一切表示しない。 各ページ下部には、柔らかな水彩と色鉛筆による日本の絵本風イラストが描かれている。 主人公は一匹の小さな白猫。丸い顔、小さな三角耳、黒い丸い瞳、小さな桃色の鼻、短い手足、ふっくらした尻尾を持つ。首輪、スカーフ、リボン、服、バッグなどの装飾品は何も身に着けていない。 全ページで白猫の顔、体形、耳、瞳、鼻、足、尻尾、大きさを完全に統一する。同じ一匹の白猫として維持する。 登場する動物は白猫と金魚だけ。星の子、蛍、妖精、発光する生物、不思議な物体は出現させない。 水彩絵本の配色は、夏空の青、ひまわりの黄色、葉の緑、夕焼けの桃色、提灯の赤、生成り色。輪郭は柔らかな茶色の色鉛筆。可愛いが幼児向けに単純化しすぎない、丁寧で温かく、少し懐かしい日本の絵本イラスト。 各ページには一つの明確な行動だけを描く。複数の出来事を同じページへ詰め込まない。白猫が何をしているか、一目で理解できる構図にする。 [0.00–2.50 seconds] 最初のページ「2026年8月 13 木」を2.5秒間、完全に明瞭な状態で表示する。まだページを動かさない。 ページ下部では、白猫が大きなひまわりの陰で丸くなり、気持ちよさそうに昼寝している。 ひまわりの葉が涼しい日陰を作っている。白猫の胸が呼吸でごくわずかに上下し、片耳が一度だけ小さく動く。ほかの行動はさせない。 [2.50–4.80 seconds] 一枚目を下端から上へゆっくりめくる。厚い紙が自然に反り、机へ柔らかな影を落とし、金属リングを通過して裏側へ着地する。紙の裏面は完全な無地。 紙が完全にめくられた後、「2026年8月 14 金」を表示する。 ページ下部では、白猫が涼しい縁側に座り、目の前に置かれた一切れの赤いスイカを興味深そうに見つめている。 白猫はスイカを食べない。前足で触らない。首を少し傾け、スイカの匂いを一度だけ嗅ぐ。背景には青空と入道雲。 [4.80–6.80 seconds] 二枚目を同じ方向へめくり、「2026年8月 15 土」を表示する。「土」は青。めくられた紙の裏面は完全な無地。 ページ下部では突然の夕立。白猫が大きな緑色の里芋の葉の下で小さく丸まり、雨宿りしている。 大きな葉が猫の屋根になっている。葉の縁から水滴が落ち、足元の水たまりへ丸い波紋が広がる。白猫は雨を見つめ、前足を身体の下へしまっている。 傘、衣服、人間の手は出さない。猫に葉を持たせない。白猫は葉の下へ座って雨をしのぐだけ。 [6.80–8.60 seconds] 三枚目を少し速くめくり、「2026年8月 16 日」を表示する。「日」は赤。紙の裏面は完全な無地。 雨上がり。白猫が水たまりの縁へ座り、水面に映った大きな虹を不思議そうに見つめている。 白猫が片方の前足で水面を一度だけ軽く触る。小さな波紋が広がり、虹の反射が優しく揺れる。白猫は水へ飛び込まない。 [8.60–10.20 seconds] 四枚目を軽快にめくり、「2026年8月 17 月」を表示する。紙の裏面は完全な無地。 夕暮れの夏祭り。白猫が透明な丸い金魚鉢の前へ座り、中を泳ぐ一匹の赤い金魚と見つめ合っている。 白猫の顔と金魚鉢を横から見た、分かりやすい構図。白猫が首をゆっくり右へ傾けると、金魚も同じ方向へ泳ぐ。 背景には赤い提灯と屋台の柔らかな光。人間は足元のシルエットだけにし、人間の顔を描かない。白猫は金魚へ前足を入れず、金魚を捕まえようとしない。 [10.20–11.70 seconds] 五枚目をめくり、最後のページ「2026年8月 18 火」を表示する。紙の裏面は完全な無地。 これ以降、絶対にページをめくらない。19日以降のページを表示しない。最後のページを交換しない。 夏祭りの夜。白猫が縁側に後ろ向きで座り、夜空いっぱいに広がる青、金、桃色の花火を見上げている。 最初の花火が開いた瞬間、白猫が少し驚いて両耳をまっすぐ立てる。その後、安心したように尻尾を身体の横へ置き、静かに花火を見上げる。 [Shot 2] At .700, 最終ページ「2026年8月 18 火」を完全に固定する。日付、曜日、紙、金属リングを動かさない。数字の「18」と曜日の「火」を最後まで変化させない。 ページ内の夜空と花火だけが繊細に動き続ける。金色の花火から生まれた小さな光粒が、紙の表面を越えて現実の机の上へ少しだけこぼれる。 光粒が青い風鈴の表面へ反射する。窓辺の風鈴が夏風で一度だけ揺れる。白猫の耳と尻尾がわずかに動く。 最後に、夜空の中央で大きな金色の花火が丸く開く。猫や文字の形にはしない。自然で美しい円形の花火にする。 At exactly 15.00 seconds, 「2026年8月 18 火」が完全に明瞭な状態で残る。白猫が縁側に座り、夏祭りの花火を静かに見上げている。 overall_soundscape: 日本の夏を感じるネイティブステレオ音声。開いた窓から聞こえる蝉、穏やかな夏風、遠くの鳥、木造家屋の静かな環境音から始める。 ページをめくるたびに、厚い紙が持ち上がり、反り、金属リングを越えて着地する柔らかな紙音を鳴らす。 最初の二枚は長くゆっくりした紙音。中盤から少しずつ短くする。ただし、各ページが完全に着地する音を明確に残す。 13日は蝉、葉の擦れる音、白猫の小さな寝息。 14日は穏やかな風、遠くの風鈴、白猫が匂いを嗅ぐ小さな鼻息。 15日は柔らかな夕立、葉へ当たる雨、水たまりへ落ちる水滴。 16日は雨上がりの鳥、水面の小さな波紋。 17日は遠い祭囃子、人々の控えめな賑わい、下駄、金魚鉢の小さな水音。 18日は遠くの花火、その少し遅れて届く柔らかな反響、窓辺の風鈴が一度だけ鳴る音。 猫の鳴き声、会話、ナレーション、歌声は入れない。 non_diegetic_music: 約94 BPMの温かく懐かしい夏のインストゥルメンタル。フェルトピアノ、グロッケン、ピチカート弦、柔らかなアコースティックギター、控えめな篠笛、軽いブラシパーカッションを使用する。 0.00秒から4.80秒までは音数を少なくし、白猫の穏やかな夏時間とゆっくりしたページめくりを見せる。 4.80秒から8.60秒はピチカートと軽いリズムを加える。夕立では音数を減らし、雨上がりの虹でグロッケンの明るい旋律を入れる。 8.60秒以降は祭りへ向かう温かな期待感を加える。最終ページではリズムを弱め、花火、風鈴、グロッケン、ピアノの余韻を中心にする。 歌詞、歌声、既存楽曲に似た旋律は使用しない。",
    "sourceUrl": "https://x.com/renataro9/status/2089573433806098785",
    "sourceLabel": "X 原帖",
    "author": "renataro9",
    "publishedAt": "2026-08-18",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089573433806098785.mp4",
    "posterUrl": "/posters/x/2089573433806098785.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "动物与生物",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/renataro9/status/2089573433806098785"
  },
  {
    "id": "x-2089539110494966112",
    "title": "红月鬼面场景中的音乐可视化",
    "titleEn": "A Music Visualizer in a Red-Moon Scene",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "固定的红月与鬼面背景前加入播放器界面和音乐可视化图形。",
    "summaryEn": "Player controls and music-visualization graphics appear over a fixed red-moon and demon-mask scene.",
    "prompt": "Add a music player visualizer that matches the beats. Static camera. Locked camera.",
    "sourceUrl": "https://x.com/GlitterPixely/status/2089539110494966112",
    "sourceLabel": "X 原帖",
    "author": "GlitterPixely",
    "publishedAt": "2026-08-18",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089539110494966112.mp4",
    "posterUrl": "/posters/x/2089539110494966112.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "插画与手绘",
      "动态图形",
      "奇幻"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image",
      "audio"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/GlitterPixely/status/2089539110494966112"
  },
  {
    "id": "x-2089355155829325876",
    "title": "米兰街头的时尚日常",
    "titleEn": "A Fashion Day on Milan Streets",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物从街头步行切换至咖啡吧与拱廊，呈现时尚生活片段。",
    "summaryEn": "A model moves between streets, a coffee bar and arcades in a fashion lifestyle sequence.",
    "prompt": "Create a hyper-realistic cinematic fashion-lifestyle video, approximately 15 seconds total, using the reference image as the identity reference for the same woman throughout.\n\nShe is a striking young woman with sun-kissed skin, voluminous blonde hair, and a confident, elevated editorial presence.\n\nDress her in a polished contemporary Milan Fashion Week street-style look: a strong, fashion-forward outfit with a clean silhouette, such as tailored trousers, a fitted top, a structured oversized coat, refined boots or heels, and a sleek designer-style bag. Keep the styling sophisticated, modern, and expensive-looking. No wardrobe changes.\n\nNo morphing. Keep her face, hair, body, and overall identity consistent in every shot.\n\nMake it fast, stylish, highly realistic storytelling with mostly 0.8 to 1.8 second scenes, plus one slightly longer 2 to 3 second hero moment at the end. Camera work should feel premium and dynamic, with smooth gimbal movement, lateral tracking, rear-follow, subtle orbiting, and elegant push-ins.\n\nThe tone is elevated, fashion-insider, and candid, like she is moving through Milan Fashion Week between moments rather than posing for a campaign.\n\nShe should feel self-possessed, magnetic, and naturally cool. No stiff posing, no plastic skin, no uncanny behavior, no scene instability, and no sudden changes in lighting, location logic, or outfit.\n\nSequence: Open with her leaving a beautiful Milan hotel or fashion apartment building and stepping onto the street with purpose, the camera moving with her as if catching her in motion.\n\nCut to her crossing a stylish Milan street while photographers and fashion people move in the background.\n\nFor the third scene, cut to a very intimate close-up of her face, neck, and the top of her shoulders only. She looks directly into the camera with calm intensity, creating an immediate personal and emotional connection with the viewer. Keep this moment extremely beautiful, confident, and natural, with subtle movement in her hair and a shallow depth of field.\n\nThen cut to her walking past a cluster of photographers or guests outside a fashion venue, not stopping to pose, just moving through the energy.\n\nCut to a quick espresso moment at a standing café bar or outside a polished café, where she glances sideways and then back into motion.\n\nCut to a candid social beat where she greets two stylish friends outside a venue or on a narrow Milan street, with a quick cheek kiss, smile, or touch on the arm, making it feel like she belongs in this world.\n\nEnd with a dramatic hero shot, slightly longer than the other scenes, showing her from the waist up. She stands or slowly walks in a striking Milan setting, such as outside a fashion venue, beneath elegant architecture, or in a softly glowing evening street scene. The camera performs a subtle, graceful move while she looks directly into the lens again, creating a second intimate emotional connection with the viewer.\n\nShe should look powerful, beautiful, and unforgettable.Make Milan feel specific through elegant streets, fashion-week crowds, refined architecture, luxury vehicles, soft European daylight or golden-hour light, and the understated chaos of the fashion world.\n\nThe result should feel like an insider Milan Fashion Week street-style film, not a runway video, not a tourist video, and not a posed ad. It should feel spontaneous, cinematic, fashion-forward, and emotionally engaging.",
    "sourceUrl": "https://x.com/PrometheanAIX/status/2089355155829325876",
    "sourceLabel": "X 原帖",
    "author": "PrometheanAIX",
    "publishedAt": "2026-08-17",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089355155829325876.mp4",
    "posterUrl": "/posters/x/2089355155829325876.jpg",
    "duration": 15.168,
    "aspectRatio": "3:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚",
      "城市街景"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/PrometheanAIX/status/2089355155829325876"
  },
  {
    "id": "x-2089291191447605442",
    "title": "火焰旋涡锻造 IGNITE",
    "titleEn": "A Fire Vortex Forges IGNITE",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "火海卷起旋涡，火焰与余烬凝聚成燃烧的标题。",
    "summaryEn": "A field of fire rises into a vortex before flames and embers form a burning title.",
    "prompt": "Ultra cinematic wide shot: A vast field of burning embers stretches into darkness. Suddenly, an enormous wall of fire races across the ground, sending billions of glowing sparks into the air. The flames twist into a gigantic rotating fire vortex hundreds of feet high. Molten embers, smoke, and liquid fire spiral together, naturally forging the word \"IGNITE\" from blazing flames. The letters burn brighter and brighter before erupting into a colossal shockwave that fills the screen with fire and incandescent particles. Hollywood fire simulation, ultra-realistic VFX, blockbuster title reveal.",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2089291191447605442",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-08-17",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089291191447605442.mp4",
    "posterUrl": "/posters/x/2089291191447605442.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动态图形",
      "奇幻"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2089291191447605442"
  },
  {
    "id": "x-2089221706435248245",
    "title": "卧室墨镜开箱与试戴",
    "titleEn": "A Bedroom Sunglasses Unboxing and Try-On",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物打开眼镜盒，展示镜框并在镜前试戴。",
    "summaryEn": "A woman opens a glasses case, presents the frames and tries them on at a mirror.",
    "prompt": "Create a 15-second photorealistic product-fashion video using the uploaded reference video as the primary visual, character, product, and performance reference.\n\nCharacter & Appearance\nRecreate the same woman from the reference video. Keep her facial identity, long dark brown hair with soft bangs, natural glamorous makeup, white fitted sleeveless top, light blue high-waisted jeans, delicate jewelry, bracelets, and rings consistent throughout the video.\n\nScene & Action\n\n0–3s:\nShe sits casually on a neatly made bed in a modern luxury bedroom with warm sunlight coming through the windows. She looks directly at the camera and speaks naturally in English while using subtle hand gestures. The exact black sunglasses case from the reference video is beside her.\n\n3–6s:\nShe picks up the sunglasses case and opens it while continuing to speak. Show a clean close-up of the case and the sunglasses inside.\n\n6–9s:\nShe removes the black rectangular sunglasses from the case. Show a detailed close-up of the sunglasses, preserving the exact shape, frame design, color, proportions, lenses, and material from the reference video.\n\n9–12s:\nShe puts the sunglasses on and adjusts them naturally with both hands while continuing to speak directly to the camera. Maintain natural facial expressions and confident fashion-influencer energy.\n\n12–15s:\nShe moves toward the large mirror and checks her look while wearing the sunglasses. Show both her and her reflection. She makes a final adjustment, then looks confidently toward the camera.\n\nVoice & Lip Sync\nThe woman must speak naturally in English throughout the video. Use the reference audio as the exact speech reference. Preserve the original words, pronunciation, pacing, pauses, emotion, and conversational tone.\n\nHer speech must be perfectly lip-synced. Match her mouth movements, jaw movement, facial expressions, blinking, eye movement, and subtle head gestures naturally to every spoken word.\n\nCamera\nBegin with a medium shot, gradually move closer during the product reveal, use cinematic close-ups for the sunglasses and hands, then transition naturally to a mirror shot. Smooth handheld movement, realistic focus changes, shallow depth of field.\n\nLighting & Environment\nWarm golden-hour sunlight, modern luxury bedroom, white bedding, dark bedside furniture, soft natural shadows, realistic mirror reflections, detailed skin and hair texture.\n\nStyle\nPhotorealistic luxury fashion commercial, realistic human movement, natural influencer-style performance, cinematic composition, premium product detail, realistic hands and fingers, consistent character identity, realistic physics.\n\nAudio\nClear natural female English speech with accurate lip sync, subtle room ambience, and realistic clothing/product sounds. No robotic voice, narrator, subtitles, extra dialogue, or overpowering music.\n\nImportant\nFollow the reference video's overall visual storytelling, character performance, product appearance, and timing while keeping the final result polished, realistic, and natural.\n\nFormat: 16:9\nDuration: 15 seconds\nQuality: Photorealistic, cinematic, high detail",
    "sourceUrl": "https://x.com/CaliraVal/status/2089221706435248245",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-17",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089221706435248245.mp4",
    "posterUrl": "/posters/x/2089221706435248245.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "生活实拍感",
      "照片级写实",
      "时尚",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2089221706435248245"
  },
  {
    "id": "x-2089219260908527687",
    "title": "泪水与愤怒的韩语表演测试",
    "titleEn": "A Korean Performance Test of Tears and Anger",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "固定近景聚焦人物的泪水、嘴部动作与情绪变化。",
    "summaryEn": "A fixed close-up focuses on tears, mouth movement and changes in emotion.",
    "prompt": "REFERENCE / 角色参考表绑定\nImage 1（7格参考图）：作为本场生成唯一的面容骨骼、发型、妆造与表情形变基准。\n基础外观：提取20岁出头东亚女性特征，微湿凌乱深色发丝、烟灰色大圆领棉质T恤、修长颈部与锁骨。完全忽略多格画板排版与灰色底色，场景中只生成单人。\n\n表情形态映射：\n00:00–00:05 严格对应参考图【左下：张嘴咆哮痛哭态】；\n00:05–00:10 严格对应参考图【中下：死死抿嘴下巴皱缩态】；\n00:10–00:15 严格对应参考图【右下/顶部：双眼通红泛泪空洞注视态】。\nCORE SPECIFICATIONS & 16:9 FRAMING / 核心规格与构图\n画幅与时长：15.00秒无缝单镜头（长镜头），16:9宽银幕电影画幅，24fps自然运动模糊。\n镜头与机位：85mm T1.4电影人像定焦镜头，固定机位平视过肩紧凑特写（Tight MCU），极浅景深。\n构图锚定：女主角面部位于画面中央偏左三分之二区域，双眼严格定位于画面上方三分之一水平线。画面右侧前景保留小面积（15%–20%）虚焦深色肩膀剪影，建立极度逼仄的面对面对峙纵深。\n\nLIGHTING, TEXTURE & TEAR PHYSICS / 光影与生理写实\n电影级高反差布光：前方3000K暖色主光照亮面部泪痕与咬肌起伏，后方锐利琥珀金轮廓光勾勒发丝边缘与颈部线条，将主体从深沉暗色背景中剥离。\n眼球与表皮细节：眼白布满充血红血丝，下眼睑水肿泛红，双眼瞳孔内反射出锐利明亮的眼神光。\n泪腺与液体物理：大量泪水在眼眶内剧烈震颤，连续湿润的泪痕顺着鼻唇沟、下颌线向下流淌，在颈部汇聚并滑入锁骨窝，高光反射清晰。\n\n15-SECOND CONTINUOUS BEATS & RUTHLESS KOREAN DIALOGUE / 动作与恶毒台词\nBEAT 1 | 00:00.000–00:05.200 | 暴怒撕裂与恶毒咆哮（对应左下张嘴爆哭）\n肌肉力学：下颌猛烈向下脱臼般张开，完全露出上下牙齿与舌面；胸锁乳突肌剧烈暴起；胸口急促起伏。眉头深锁成倒V形，下眼睑对抗泪水剧烈抽搐。\n动作：头部在发音瞬间伴随向前压迫的攻击性微震颤，眼神充满杀气与怨恨。\n对白： [Korean, adult female voice; venomous and explosive acoustic scream, high-pitch vocal fry, breath cracking on curse words] \"야, 이 씨발놈아! 지랄 떨지 말고 처음부터 꺼지든가! 이제 와서 개수작이야?!\"\nBEAT 2 | 00:05.200–00:10.500 | 刻薄讥笑与下巴死死皱缩（对应中下抿嘴压抑）\n肌肉力学：双唇瞬间死死咬合闭紧。颏肌（mentalis）极度收缩，下巴呈现紧绷的凹凸褶皱。咬肌剧烈凸起；嘴角向上抽动一瞬，露出极度刻薄、看垃圾般的冷酷冷笑。\n动作：眼神像刀子一样剜向对方，伴随急促不屑的鼻腔冷哼，极度尖酸刻薄。\n对白： [Korean, sharp, mocking and aggressive rasp; sneering through clenched teeth, heavy breathing] \"내가 병신 같냐? 어? ...존나 가소로워서 눈물도 안 나오네. 넌 진짜 끝까지 역겨워.\"\nBEAT 3 | 00:10.500–00:15.000 | 喉结吞咽与冰冷死咒（对应右下空洞冰冷）\n肌肉力学：喉部完成一次深沉明显的吞咽，下巴肌肉紧绷感退去。下唇无力轻颤，眼眶中的泪水彻底漫出滑落颈部。眼神从愤怒骤降为彻底的冰冷与蔑视。\n动作：头部极其微弱地下沉，双唇在说出最后一个字后保持微张，冰冷的呼气消散在空气中。\n对白： [Korean, cold, dry, venomous whisper; breath trailing off with utter contempt] \"…평생 그렇게 비겁하게 썩어 문드러져, 개새끼야.\"\nDIEGETIC SOUNDSCAPE & VOCAL CAPTURE / 声音规则\n极高动态范围的同期声直录，清晰捕捉咬字摩擦音、爆破辅音（씨/발/개/지/랄）、急促喘息声与喉音吞咽。\n嘶吼高音处保留自然的声带破裂与毛边，严禁音频削波失真。\n严禁任何背景音乐、弦乐、钢琴或电子音效，保持极度压抑的现场真实感。\nNEGATIVE CONSTRAINTS / 负向规避\n严禁任何镜头推拉晃动、严禁横屏上下黑边裁切、严禁切镜头。\n严禁温和/委屈/示弱的哭戏、严禁假哭、严禁面无表情。\n严禁磨皮美白塑料脸、严禁中文字幕压屏、严禁音画不同步。",
    "sourceUrl": "https://x.com/IamEmily2050/status/2089219260908527687",
    "sourceLabel": "X 原帖",
    "author": "IamEmily2050",
    "publishedAt": "2026-08-17",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089219260908527687.mp4",
    "posterUrl": "/posters/x/2089219260908527687.jpg",
    "duration": 15.16263,
    "aspectRatio": "227:128",
    "resolution": "512p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/IamEmily2050/status/2089219260908527687"
  },
  {
    "id": "x-2089205270543213031",
    "title": "画室里的林肯肖像绘制",
    "titleEn": "Painting a Lincoln Portrait in a Studio",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "颜料准备、画布细节与完成作品展示组成绘画过程。",
    "summaryEn": "Paint preparation, canvas details and a finished-work presentation form a painting sequence.",
    "prompt": "A young Korean girl with a Korean-inspired hairstyle, soft natural makeup and a gentle smile, wearing a light pink shirt, sits in a cozy modern art studio and rapidly paints a highly detailed realistic Abraham Lincoln portrait on a white canvas. She first grabs multiple colorful paint tubes of red, blue, yellow, pink, purple, green, orange, black and white from a wooden table, quickly opens them, squeezes thick vibrant blobs onto a flat palette, then picks up a thin brush and swiftly creates the outline before filling in the face, hair, beard, eyes, suit and shadows with fast confident strokes, adding fine details until the portrait looks professional and complete, finally smiling naturally as she holds the finished painting toward the camera with both hands under warm natural daylight.",
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2089205270543213031",
    "sourceLabel": "X 原帖",
    "author": "ayzalnooor24521",
    "publishedAt": "2026-08-17",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2089205270543213031.mp4",
    "posterUrl": "/posters/x/2089205270543213031.jpg",
    "duration": 14.976,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "人物表演",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ayzalnooor24521/status/2089205270543213031"
  },
  {
    "id": "x-2088983831282098350",
    "title": "首尔雨巷中的韩语独白",
    "titleEn": "A Korean Line in a Rainy Seoul Alley",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑衣人物沿湿润巷道走近镜头，停下进行近景表演。",
    "summaryEn": "A dark-clad figure approaches through a wet alley and stops for a close-up performance.",
    "prompt": "[핵심]\n8초, 16:9. 사실적인 실사 영화. 20대 한국 여성, 젖은 검정 트렌치코트, 짧은 흑발. 비 오는 서울 골목, 바닥에 네온이 번진다. 한 테이크, 컷 없음.\n\n[화면]\n[0s-4s] 중경. 그녀가 카메라를 향해 세 걸음 걷는다. 빗방울이 코트에 맺힌다. 카메라는 같은 속도로 천천히 푸시 인.\n[4s-6s] 멈춰 서서 고개를 든다. 가로등 불빛이 얼굴 한쪽만 비춘다.\n[6s-8s] 카메라를 보고 한국어로 말한다: \"오늘이 마지막이야.\" 입술이 대사에 맞춰 움직인다. 마지막 1초는 정지에 가깝게 홀드.\n\n[소리]\n현장음: 빗소리, 멀리 지나가는 차, 젖은 구두 발소리.\n대사: 한국어, 낮고 또렷한 여성 목소리. \"오늘이 마지막이야.\"\nBGM: N/A.\n\n[금지]\n컷 없음. 자막 없음. 영어 대사 없음. 추가 인물/로고 없음. 소프트 디졸브 없음.",
    "sourceUrl": "https://x.com/Alisvolatprop12/status/2088983831282098350",
    "sourceLabel": "X 原帖",
    "author": "Alisvolatprop12",
    "publishedAt": "2026-08-16",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088983831282098350.mp4",
    "posterUrl": "/posters/x/2088983831282098350.jpg",
    "duration": 14.464,
    "aspectRatio": "9:5",
    "resolution": "480p",
    "tags": [
      "对白与口型",
      "写实电影感",
      "城市街景"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Alisvolatprop12/status/2088983831282098350"
  },
  {
    "id": "x-2088928805561708915",
    "title": "山崖崩解形成 DOMINION",
    "titleEn": "A Collapsing Cliff Forms DOMINION",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "峡谷岩壁碎裂，巨石在尘雾中组成纪念碑式文字。",
    "summaryEn": "Canyon rock fractures and giant stones form monumental lettering amid dust.",
    "prompt": "Ultra cinematic title sequence: A gigantic mountain cliff begins collapsing into a deep canyon. Entire rock faces fracture while dust clouds rise thousands of feet into the atmosphere. Giant stone monoliths rotate in slow motion through the air, colliding with breathtaking force. Rock fragments, dust, and debris gradually assemble the monumental word \"DOMINION\", carved from ancient stone scarred by centuries of erosion. The canyon echoes with a deafening rumble before the title crumbles into dust that is carried away by the wind. Monumental scale, ancient power, Hollywood epic title design.",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2088928805561708915",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-08-16",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088928805561708915.mp4",
    "posterUrl": "/posters/x/2088928805561708915.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动态图形",
      "自然风景"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2088928805561708915"
  },
  {
    "id": "x-2088928791741481189",
    "title": "从雪花微距推进冰雪王国",
    "titleEn": "From a Macro Snowflake to a Frozen Kingdom",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头进入六角冰晶，以相似几何结构衔接冰雪城市鸟瞰。",
    "summaryEn": "The camera enters a hexagonal ice crystal and matches its geometry to an aerial frozen city.",
    "prompt": "@Image 1: Push steadily toward the center of the snowflake while preserving the extreme macro realism of the crystal. Move deeper into the intricate hexagonal structure, razor-sharp ice branches, delicate crystalline facets, microscopic frost, and shimmering frozen edges. Maintain the shallow depth of field, cold blue-white illumination, subtle refraction, drifting snow particles, and sparkling highlights. As the camera advances, allow the snowflake's branching geometry, frozen ridges, and crystalline valleys to increasingly dominate the frame until the ice structure begins to resemble an immense frozen landscape viewed from above.\n\nAt the exact moment when the snowflake's branching arms, crystalline ridges, hexagonal facets, and frosted textures naturally match the mountain ranges, frozen lakes, icy valleys, and radial layout of @Image 2, transition seamlessly into the aerial view of the frozen kingdom. Continue the same uninterrupted forward movement toward the magnificent ice citadel at the center as crystalline facets become frozen lakes, delicate frost becomes snow-covered forests, microscopic ice ridges become towering mountain chains, and the central hexagon transforms into the heart of the kingdom. Preserve the feeling that the viewer has never left the same crystalline structure—only its scale has expanded from a single snowflake to an entire frozen world.\n\nNo tearing, black frames, hard cuts, visible morphing, obvious VFX, or compositing seams. Keep the transformation completely photorealistic, elegant, and physically believable—as though the microscopic architecture of a snowflake naturally unfolds into a colossal frozen kingdom. Preserve the continuity of crystalline branches becoming mountain ridges, ice facets becoming frozen lakes, frost particles becoming drifting snowfall, and sparkling highlights becoming sunlight glinting across glaciers and ice palaces. Maintain identical cool blue lighting, atmospheric continuity, and one continuous cinematic camera push with no visible edit.",
    "sourceUrl": "https://x.com/LudovicCreator/status/2088928791741481189",
    "sourceLabel": "X 原帖",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-16",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088928791741481189.mp4",
    "posterUrl": "/posters/x/2088928791741481189.jpg",
    "duration": 15.2,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2088928791741481189"
  },
  {
    "id": "x-2088927511174422822",
    "title": "夏日雨中穿过草地的人群",
    "titleEn": "Running across a Field in Summer Rain",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "远景保留大面积天空与草地，几位人物横向跑过画面。",
    "summaryEn": "A wide view holds open sky and grass while several figures run across the frame.",
    "prompt": "A cinematic wide shot of four young women running barefoot across a vast rain-soaked green field during a sudden summer downpour. Their casual summer clothes and loose hair are drenched by the rain as they sprint together from left to right, giving the moment a spontaneous, carefree sense of youth and freedom. Rather than posing for the camera, they appear completely absorbed in the moment.\n\nShot as an extreme wide environmental shot from roughly eye level, with the figures occupying only the lower third of the composition. A huge amount of open landscape and pale overcast sky surrounds them, emphasizing scale, isolation and atmosphere. A single small tree near the center of the frame acts as a quiet visual anchor against the movement of the running figures.\n\nThe camera feels static or only subtly handheld, allowing the women to rush through the frame rather than actively tracking them. A moderately wide lens, around 28–35mm equivalent, captures the expansive field, distant hills and surrounding trees while keeping the runners readable.\n\nThe women are intentionally rendered with strong natural motion blur, suggesting a relatively slow shutter speed and creating an impressionistic sense of speed. The landscape remains comparatively stable and readable, making their blurred movement especially prominent.\n\nLighting is entirely soft natural daylight filtered through heavy storm clouds and rainfall. Visibility gradually falls off toward the distant hills, where rain and atmospheric haze create layered depth. The palette is subdued and earthy, dominated by wet olive green, moss, muted yellow, warm grey and faded brown.\n\nThe image has a distinctly analog cinematic texture, with soft focus, restrained contrast, subtle grain, slightly muddy colors and imperfect exposure. It feels like a frame from a 1970s or 1990s coming-of-age arthouse film, balancing nostalgia, freedom, summer melancholy and the chaotic energy of being caught unexpectedly in the rain.",
    "sourceUrl": "https://x.com/aimikoda/status/2088927511174422822",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-16",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088927511174422822.mp4",
    "posterUrl": "/posters/x/2088927511174422822.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "自然风景",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/aimikoda/status/2088927511174422822"
  },
  {
    "id": "x-2088859138730189040",
    "title": "云海古桥上的仙境漫步",
    "titleEn": "A Celestial Walk along a Bridge above Clouds",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "微小红白人物沿宏伟石桥行走，云海与宫殿突出空间尺度。",
    "summaryEn": "Small red-and-white figures walk along a monumental stone bridge above clouds and palaces.",
    "prompt": "一座雄伟的古桥从前景延伸至远方，以壮丽的弧线蜿蜒至琥珀色朦胧的地平线。\n桥面顺着弧线延伸，干净、宽阔、简洁，珍珠灰石材构成抛光石材和漆面与巨大的桥墩相得益彰。\n左侧前景中一根巨大的珍珠灰石材雕花柱占据主导地位；更多高耸的珍珠灰石材柱子沿着桥梁向后延伸。\n右侧:空旷无垠的云海。2个中国仙人女性人物，一个穿着大红色长袍，一个穿着白色长袍，她们沿着弯曲的桥面缓缓行走，在宏伟的建筑映衬下显得格外娇小。人物占画面高度的2%，是一个极小的色点，正缓慢向前行走。人物身体安静，衣料和长发接受同一方向的高空微风，向后飘动。人物周围十数倍身长的空旷距离。 \n温暖的午后侧光，营造出柔和的氛围薄雾。\n远处背景:云雾缭绕中隐约可见宏伟的中国庙宇建筑；意境宁静而庄严，宛如中国仙境；\n史诗般的东方美学，庄严而空灵，极具比例对比。",
    "sourceUrl": "https://x.com/PixelAigc/status/2088859138730189040",
    "sourceLabel": "X 原帖",
    "author": "PixelAigc",
    "publishedAt": "2026-08-16",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088859138730189040.mp4",
    "posterUrl": "/posters/x/2088859138730189040.jpg",
    "duration": 30.46458,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/PixelAigc/status/2088859138730189040"
  },
  {
    "id": "x-2088624894401511747",
    "title": "雨夜咖啡馆里的争执与离开",
    "titleEn": "An Argument and Departure in a Rainy-Night Cafe",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "暖色咖啡馆中两人隔桌对话，近景之后人物离场。",
    "summaryEn": "Two people speak across a table in a warm cafe before close-ups lead to a departure.",
    "prompt": "15秒，16:9，写实当代英语剧情片，雨夜小型独立咖啡馆。24fps，浅景深，2700K暖钨丝灯与窗外冷蓝雨夜形成对比。全片由4个连续镜头硬切组成，真实同期声，无配乐。\n\nREFERENCE / 参考图绑定\nImage 1：唯一指定女性角色（Nora Hale）的五官面容、棕色及肩短发、灰绿眼睛、尘玫瑰色绞花毛衣、蓝色牛仔裤与棕色短靴。忽略原图灰色背景与中性表情。\nImage 2：唯一指定男性角色（Elias Ward）的五官面容、深色卷发、深褐眼睛、做旧深棕皮夹克、炭灰内搭与深色牛仔裤。忽略原图灰色背景与中性表情。\n\nSCENE GEOGRAPHY / 空间与站位 中央一张小圆木桌。Nora站在桌子左侧面向右，左肩背黑色小挎包；Elias坐在桌子右侧面向左。左后方为清晰可见的玻璃出口门。Elias面前一杯白色咖啡与碟，Nora面前一杯未碰的咖啡。180度轴线锁定，严禁左右机位翻转。\n\n[SHOT 1 | 00:00–00:07] 50mm镜头，肩高双人中景，极慢直线微推进。\n物理动作与微表情：Nora左手死死扣紧木椅背直至指节发白，下颌咬肌剧烈紧绷微颤，眼眶通红湿润，胸口剧烈起伏吸气。Elias双肩向内塌陷，视线死盯未碰的咖啡杯，双手僵死在碟边。\n对白与唇形：Nora咬紧牙关挤出极度压抑、尾音发颤的声音： [English, natural American accent; tightly clenched teeth, ragged breaths] You’re such a jerk. (0.3s pause) After everything I did for you... (throat tightens) after I protected you from everyone... this is what I get from you?（\"you\"字带轻微破音）。\n[SHOT 2 | 00:07–00:09] 硬切至85mm固定机位，Elias过肩近景（Nora毛衣肩膀在左缘虚化）。\n物理动作与微表情：Elias喉结剧烈上下吞咽，鼻翼轻颤，右手食指用力向下压住瓷碟边缘以抑制手部震颤。他勉强抬眼对视Nora不足半秒。\n对白与泪痕：Elias伴随虚弱的干涩呼气低哑发声： [English, low dry rasp, ashamed whisper] I know. 说完嘴唇紧抿成一条直线，一滴眼泪随即从左眼眶滑落面颊。\n[SHOT 3 | 00:09–00:11] 硬切至85mm固定机位，Nora面部紧凑特写。\n物理动作与微表情：听到回答后停顿半秒，Nora下眼睑肌肉剧烈抽搐，双眼大睁。胸腔猛然倒吸一口冷气，脖颈两侧青筋暴起，下颌骤然拉开爆发厉声嘶喊。喊完后嘴唇瞬间狠狠咬合闭紧，鼻翼微张，眼神彻底转为冰冷决绝。\n对白： [English, explosive ragged scream, chest-driven voice break] You know nothing!\n[SHOT 4 | 00:11–00:15] 硬切回40mm固定广角全景。\n调度与离场：Nora移开视线，左脚跟发力转身，背脊僵直快步走向左侧玻璃门。Elias右手从桌面下意识抬起数厘米试图挽留，手指微蜷后无力落回桌面。\n出幅与定格：Nora推开玻璃门完全走出画面，液压门缓慢回弹关拢。镜头固定停留在Elias独自面对空椅子与冷掉的咖啡杯。\n\nAUDIO / 现场同期声 持续的雨打玻璃声与室内低沉环境音。Nora牙关紧咬的吸气声、字句摩擦声与尖锐破音喊叫；Elias吞咽声与低微气声；Nora硬底短靴的快速脚步声、推开玻璃门的轴承声及骤然涌入的室外大雨声。全片严禁任何背景音乐与音效垫底。",
    "sourceUrl": "https://x.com/IamEmily2050/status/2088624894401511747",
    "sourceLabel": "X 原帖",
    "author": "IamEmily2050",
    "publishedAt": "2026-08-15",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088624894401511747.mp4",
    "posterUrl": "/posters/x/2088624894401511747.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "写实电影感"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/IamEmily2050/status/2088624894401511747"
  },
  {
    "id": "x-2088569547494199485",
    "title": "戴眼镜动漫角色的晨间日常",
    "titleEn": "A Morning Routine for an Anime Character with Glasses",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "起床、手机与出门准备镜头串起角色的日常。",
    "summaryEn": "Waking, phone and getting-ready shots connect an anime character's morning routine.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2088569547494199485",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-15",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088569547494199485.mp4",
    "posterUrl": "/posters/x/2088569547494199485.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2088537866955816993",
    "title": "动漫人物突破巨型负面文字",
    "titleEn": "An Anime Character Breaks through Giant Negative Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物穿越黑色日文障碍，画面最终出现红色鼓励语。",
    "summaryEn": "A character passes black Japanese typographic obstacles before a red message of encouragement appears.",
    "prompt": "CHARACTER (identical in every cut): a woman in her twenties, short black hair, grey blazer over a white shirt, red high-top sneakers. Same face, eyes and proportions in all four cuts - do NOT redesign her between cuts; exaggeration applies to expression and pose only, never to face structure. She moves with fluid athletic grace, never stiff.\n\nWORLD: an endless white void with a pale-grey grid floor, dust motes in hard rim light. Stylized 2.5D anime aesthetic, photorealistic high-detail 2.5D CG, clean cel-shaded face, high-key white and grey with one red accent, soft bloom, lens flares, 35mm grain.\n\nTEXT (critical): exactly three pieces of text appear, one at a time, never together. Each is a single horizontal line spanning the full width of the frame, razor sharp, correctly spelled, clean spacing between characters.\n(1) Japanese, nine characters: 失敗を恐れすぎだよ - thick wet black brush ink.\n(2) Japanese, thirteen characters: 人の目ばかり気にしてるよね - thick wet black brush ink.\n(3) English, four words: You can do it! - bold, solid red.\nNo other letters, numbers, signage, captions or subtitles anywhere; the spoken lines must never appear as text.\n\nSHOT 1 (0:00-0:04) [Truck right]. Emotion: irritated defiance.\n0:00-0:01: extreme close-up of a red sneaker landing on the white floor, black ink cracking outward.\n0:01-0:04: camera drops to ground level and trucks right alongside her as she walks - text (1) erupts from her footprints and locks into the air behind her, taller than she is. She says in Japanese: \"うるさいな\" - flat, annoyed, half over her shoulder, lip-synced.\n\nCut to SHOT 2 (0:04-0:08) [Tracking shot]. Emotion: pressure.\n0:04-0:06: she breaks into a run, three hard strides, foreground grid slabs whipping past in motion blur, speed lines streaking.\n0:06-0:08: text (1) dissolves and text (2) rears up as a long ink wall chasing her, dripping, slamming down behind her heels twice and cratering the floor. Camera tracks low behind her shoulder, shaking on each impact.\n\nCut to SHOT 3 (0:08-0:12) [Push in]. Emotion: sharp refusal.\n0:08-0:09: she skids to a dead stop, one arm swinging out, dust bursting sideways.\n0:09-0:10: hard accelerating push-in as she pivots and shoves both palms into the falling ink wall - a single-frame high-contrast inverted impact frame, shockwave ring.\n0:10-0:12: text (2) shatters into black ink fragments scattering past camera in slow motion.\n\nCut to SHOT 4 (0:12-0:15) [Pull out]. Emotion: light, decided.\n0:12-0:13: she steps through the fragments and stamps a foot down.\n0:13-0:15: text (3) rams up out of the floor beside her, filling the frame, as the camera pulls out to a wide of the void. She turns her head and gazes directly into the camera. She says in Japanese: \"やってみる\" - light, half smiling, lip-synced.\n\nCAMERA GRAMMAR: one move per cut; shot sizes swing hard - ECU, low tracking wide, ECU, wide. PERMITTED: dolly, push-in, truck, tilt, low tracking. AVOID: whip pans, handheld jitter, 360-degree rotation, slow zooms.\n\nAUDIO: sneaker squeak, running footfalls, a heavy wet ink slam on each impact, a glass-like shatter on the shove. Music: minimal electronic percussion, one beat drop on the shatter, one on the final text. Only one voice in the clip - hers. No narration, no second speaker.\n\nCONSTRAINTS: she is the only person in frame. Stable geometry, natural motion, no warping.",
    "sourceUrl": "https://x.com/LilyLoveSNPY/status/2088537866955816993",
    "sourceLabel": "X 原帖",
    "author": "LilyLoveSNPY",
    "publishedAt": "2026-08-15",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088537866955816993.mp4",
    "posterUrl": "/posters/x/2088537866955816993.jpg",
    "duration": 15.146667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/LilyLoveSNPY/status/2088537866955816993"
  },
  {
    "id": "x-2088505215792369740",
    "title": "灯笼通道中的红伞表演",
    "titleEn": "A Red-Umbrella Performance in a Lantern Passage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "和服人物在灯笼长廊旋转纸伞，以近景与移动镜头展示动作。",
    "summaryEn": "A kimono character spins a paper umbrella among lanterns, shown through close-ups and moving shots.",
    "prompt": "CHARACTER (identical in every cut): a woman in her twenties, chin-length black bob with a vermilion cord, deep indigo kimono with vermilion under-collar, sleeves tied back with a red tasuki cord, a slim sheathed blade at her left hip, a large vermilion paper umbrella. Same face, eyes and proportions in all four cuts - do NOT redesign her between cuts. Exaggeration applies ONLY to expression and pose, never to face structure. She moves with flowing unhurried grace, never stiff.\n\nSTYLE: stylized 2.5D anime aesthetic, photorealistic high-detail 2.5D CG, clean cel-shaded face, night shrine approach lit by rows of lanterns, warm vermilion and gold against deep indigo, soft bloom, lens flares, 35mm grain.\n\nSHOT 1 (0:00-0:04) [Truck left]. Emotion: playful confidence.\n0:00-0:01: extreme close-up of the spinning umbrella rim slicing through lantern light, flares strobing.\n0:01-0:04: camera drops to ground level, trucks left alongside her sandals, then tilts up - she walks the lantern-lined gravel approach spinning the open umbrella on her shoulder, sleeve and cord flying. She says in Japanese: \"今宵は、よく回る\" - low, amused, half to herself, lip-synced.\n\nCut to SHOT 2 (0:04-0:08) [Tilt up]. Emotion: gathering intent.\n0:04-0:05: she plants her feet, sinks her hips, hurls the umbrella straight up with both hands.\n0:05-0:08: camera tilts up and cranes after it - the umbrella spins end over end past the hanging lanterns into the black sky, paper glowing from beneath. Her right hand closes on the hilt at her hip. Held anticipation, the scene goes quiet.\n\nCut to SHOT 3 (0:08-0:12) [Push in]. Emotion: sharp release.\n0:08-0:09: extreme close-up of her eyes snapping open, a bright glint crossing one pupil.\n0:09-0:10: hard accelerating push-in as she draws - one horizontal flash of the blade, a single-frame high-contrast inverted impact frame, speed lines radiating.\n0:10-0:12: a mass of paper streamers and petals bursts apart and drifts down in slow motion around her, motion blur streaking the foreground. She holds the finish pose, blade extended.\n\nCut to SHOT 4 (0:12-0:15) [Pull out]. Emotion: quiet satisfaction.\n0:12-0:13: without looking she sheathes the blade and reaches back over her shoulder.\n0:13-0:15: the falling umbrella lands in her palm, snaps open, settles on her shoulder as the camera pulls out to a wide of the lantern approach. She turns her head and gazes directly into the camera. She says in Japanese: \"これで、しまい\" - quiet, satisfied, lip-synced. Lantern flames sway, streamers still falling.\n\nCAMERA GRAMMAR: one distinct move per cut; shot sizes swing hard - ECU, low wide, ECU, wide. PERMITTED: dolly, push-in, truck, tilt, crane. AVOID: whip pans, handheld jitter, 360-degree rotation, slow generic zooms.\n\nAUDIO: festival night ambience - gravel underfoot, the papery whir of the spinning umbrella, cord and sleeve snap, one sharp steel draw, a scatter of falling paper, distant wind chimes. Music: sparse taiko and bamboo flute, low and rhythmic, dropping to silence for the draw and returning on the catch. Only one voice in the whole clip - hers. No narration, no second speaker, no crowd chatter.\n\nCONSTRAINTS: no readable text, logos, letters or numbers anywhere - the spoken lines must never appear as on-screen text or subtitles. She is the only person in frame. Stable geometry, natural motion, no warping.",
    "sourceUrl": "https://x.com/LilyLoveSNPY/status/2088505215792369740",
    "sourceLabel": "X 原帖",
    "author": "LilyLoveSNPY",
    "publishedAt": "2026-08-15",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088505215792369740.mp4",
    "posterUrl": "/posters/x/2088505215792369740.jpg",
    "duration": 15.146667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/LilyLoveSNPY/status/2088505215792369740"
  },
  {
    "id": "x-2088498595980812373",
    "title": "损毁机甲旁的飞行员情绪片段",
    "titleEn": "A Pilot beside a Destroyed Mecha",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "雨中飞行员跪靠在巨大机械头旁，近景表现触碰与悲伤。",
    "summaryEn": "A pilot kneels beside a giant mechanical head in rain as close shots capture touch and grief.",
    "prompt": "15秒，16:9，24fps，单镜头写实科幻战争短片。Image 1 是唯一的角色、机械、环境、构图、材质和色彩参考。保持同一名成年女性飞行员、同一套泥污白色技术服、同一具紫绿双色生物机械头部、同一片寒冷荒原和相同阴雨天气。不得改变机械头部的轮廓、断角、裂纹、装甲位置、裸露线缆或损坏程度。\n\n从 Image 1 的准确瞬间开始。地面高度的中广角构图，损毁的巨大机械头占据左侧大部分画面，成年女性飞行员跪在右下方泥水中，额头紧压机械下颌，双手撑住冰冷装甲。远处荒原、残骸和低矮山脊保持模糊。镜头以极慢、稳定的贴地推进逐渐靠近她，不绕行、不升高、不切镜头，不把机械头完整展示成英雄镜头。\n\n00:00至00:03.20：她保持额头贴住装甲，像在等待内部传来回应。她的呼吸短促而不规律，肩膀随每次吸气突然抬起，又被强行压住。右手掌贴紧下颌接缝，五指逐渐用力，指尖在湿滑金属上轻微打滑；左手缓慢沿裂缝寻找仍在工作的振动。膝盖深陷泥中，靴尖因用力向后刮出一道短泥痕。雨水沿白发、睫毛、手背和破裂紫色装甲持续流下。\n\n00:03.20至00:06.20：她突然停止呼吸，像是听见了极微弱的东西。她只将头抬离装甲几厘米，把左耳贴向机械面部靠近眼部的位置。不要完全站起，不要转向镜头。她的眼睛因为恐惧而睁大，嘴唇微微张开，但声音卡在喉咙里。机械眼内部出现一次极弱、断续的暗橙色电流，亮不到半秒，随即熄灭。内部传出一个破损继电器尝试接通的声音，然后彻底停止。机械本体绝不复活，不转头，不眨眼，不主动移动。\n\n00:06.20至00:09.30：她终于理解没有回应。下颌先绷紧，鼻翼颤动，眼泪与雨水混在一起。她把右手握成拳，只重重击打机械下颌一次。撞击来自肩部和躯干的真实发力，拳头接触金属后立即失去力量，沿泥湿装甲滑下。她没有连续捶打，也没有夸张甩头。她贴近机械面部，用成年女性低哑、断裂、几乎无法完整呼吸的英语说：\n\n[English, adult woman, hoarse, breath breaking] Open your eyes… please.\n\n每个词之间都有无法控制的吸气，最后一个“please”很轻，不要喊叫，不要演成舞台式哭泣。\n\n00:09.30至00:12.60：最后一个词结束后停留0.4秒。机械内部没有任何回应。她再次尝试吸气，但胸口只完成一半动作，喉咙发出短促、压抑的失败声音。她用双手抓住装甲边缘，试图把自己的身体拉近，仿佛可以用身体重量迫使它醒来。双臂短暂绷紧，然后力量彻底消失。她的脊柱向内折叠，额头重新撞回机械下颌，肩膀开始不对称地剧烈颤抖。一声无法压住的低沉哭声从胸腔逸出，但不形成第二句台词。\n\n00:12.60至00:15.00：她没有恢复，也不抬头。左手仍死死抓住装甲接缝，右手逐渐松开，手指一根接一根失去力量，最后停在泥水中。镜头在距离她数步的位置停止推进，保持她蜷缩的身体与巨大死亡机械之间的尺度差。远处一块受损金属残骸被风推动，发出一次空洞摩擦声。最后一秒只剩雨声、她无法平稳下来的呼吸，以及机械内部完全没有运转声的空白。\n\n表演要求：情绪从拒绝接受、寻找生命迹象、短暂希望、确认死亡，发展到身体支撑崩溃。所有情绪必须通过呼吸、吞咽、手指压力、肩部震动、眼神和重心变化表现。不要持续尖叫，不要仰头哭喊，不要面向镜头表演，不要快速抽泣，不要夸张拍打装甲。\n\n物理要求：机械头始终沉重地嵌在泥土中，只允许雨水、松散线缆和极轻的金属沉降产生被动运动。飞行员的双膝始终接触地面，手掌与装甲保持可信接触，不穿模、不漂浮。白色技术服吸水后贴住肘部和膝部，泥浆随动作真实涂抹，不改变服装设计。头发被雨水压湿，仅被侧风轻微推动。\n\n声音：近距离冷雨敲击破裂陶瓷装甲，低速荒原风，湿布摩擦，手套在粗糙金属上的滑动声，机械内部一次失败的继电器接通声，一次沉重但不夸张的拳头撞击。对白置中、距离近、保留呼吸和喉音。前10秒没有音乐。确认机械彻底沉默后，只加入一条极低、缓慢变化的弓弦音，音量低于雨声，不使用钢琴、合唱、预告片重击、英雄旋律或煽情上升。\n\n禁止：机械复活、眼睛持续发光、能量爆发、闪回、灵魂幻影、额外人物、敌人出现、爆炸、武器、文字、字幕、镜头抖动、慢动作、快速剪辑、过度景深变化、服装变形、机械结构重绘、改变天气、改变昼夜或改变 Image 1 的冷灰低饱和色彩。",
    "sourceUrl": "https://x.com/IamEmily2050/status/2088498595980812373",
    "sourceLabel": "X 原帖",
    "author": "IamEmily2050",
    "publishedAt": "2026-08-15",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088498595980812373.mp4",
    "posterUrl": "/posters/x/2088498595980812373.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "写实电影感",
      "科幻太空"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/IamEmily2050/status/2088498595980812373"
  },
  {
    "id": "x-2088487667629715917",
    "title": "夕阳社区遛狗生活片",
    "titleEn": "A Sunset Neighborhood Dog Walk",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物从室内带狗出门，沿树荫人行道散步并看向镜头。",
    "summaryEn": "A woman takes her dog outside and walks along a tree-lined sidewalk before looking at the camera.",
    "prompt": "A cinematic, realistic lifestyle video featuring a young woman in a soft gray hoodie taking her small golden-brown dog for a peaceful walk through a charming suburban neighborhood. The scene begins indoors before transitioning naturally to an outdoor sidewalk surrounded by leafy trees. Golden-hour sunlight creates warm highlights, soft shadows, and subtle lens flares. She walks casually with the dog, occasionally looking around before turning toward the camera with a natural, genuine smile. Handheld vlog-style camera movement, authentic expressions, realistic motion, shallow depth of field, warm cinematic atmosphere, subtle film grain, detailed skin and hair, photorealistic 4K quality.",
    "sourceUrl": "https://x.com/CaliraVal/status/2088487667629715917",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-15",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088487667629715917.mp4",
    "posterUrl": "/posters/x/2088487667629715917.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动物与生物",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2088487667629715917"
  },
  {
    "id": "x-2088446104287477944",
    "title": "浮世绘画卷里的东京旅行",
    "titleEn": "A Tokyo Journey inside a Woodblock-Style Scroll",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "画卷展开为灯笼街道与日式建筑；本站播放原帖标明的 H3 版本。",
    "summaryEn": "A scroll opens into lantern streets and Japanese buildings; this is the source's identified H3 version.",
    "prompt": null,
    "sourceUrl": "https://x.com/tanabe_fragm/status/2088446104287477944",
    "sourceLabel": "X 原帖",
    "author": "tanabe_fragm",
    "publishedAt": "2026-08-15",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088446104287477944.mp4",
    "posterUrl": "/posters/x/2088446104287477944.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "城市街景"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2088390915152887996",
    "title": "巴士里的死亡金属乐队",
    "titleEn": "A Death Metal Band inside a Bus",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "乐队在巴士内部演唱与演奏，展示音乐代理编排的视频结果。",
    "summaryEn": "A band sings and plays inside a bus in a video arranged by a music agent.",
    "prompt": null,
    "sourceUrl": "https://x.com/blizaine/status/2088390915152887996",
    "sourceLabel": "X 原帖",
    "author": "blizaine",
    "publishedAt": "2026-08-14",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088390915152887996.mp4",
    "posterUrl": "/posters/x/2088390915152887996.jpg",
    "duration": 120.047166,
    "aspectRatio": "20:11",
    "resolution": "704p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2088301240346087802",
    "title": "蝙蝠侠与绝地武士的梦境战斗",
    "titleEn": "A Dream Battle between Batman and a Jedi",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "暗色遗迹中，披风人物与光剑对手展开动作交锋。",
    "summaryEn": "A caped figure battles a lightsaber opponent in dark ruins.",
    "prompt": "Make me a dream sequence, Batman fighting a Jedi",
    "sourceUrl": "https://x.com/LyalinDotCom/status/2088301240346087802",
    "sourceLabel": "X 原帖",
    "author": "LyalinDotCom",
    "publishedAt": "2026-08-14",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088301240346087802.mp4",
    "posterUrl": "/posters/x/2088301240346087802.jpg",
    "duration": 60.064,
    "aspectRatio": "20:11",
    "resolution": "704p",
    "tags": [
      "动作特效",
      "写实电影感",
      "动作打斗",
      "科幻太空"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/LyalinDotCom/status/2088301240346087802"
  },
  {
    "id": "x-2088060516027785340",
    "title": "列车窗外穿越历史的首尾帧片段",
    "titleEn": "A Train Window Travels through History",
    "model": "MiniMax H3",
    "mode": "FL2VA",
    "summary": "车厢与乘客保持同一场景，窗外景物从雪山切换到旧街与恐龙。",
    "summaryEn": "The carriage and passenger remain in place while the window view shifts from snowy peaks to old streets and dinosaurs.",
    "prompt": "<Picture 1> is the EXACT first frame of the video.\n\nPreserve the woman, her face, blonde hair, cream knit outfit, white boots, seated pose, brown leather seat, modern luxury high-speed train interior, panoramic windows, camera position, warm interior sunlight, and overall photorealistic cinematic appearance established by <Picture 1>.\n\nno music\n\nintegrated_multimodal_description:\n\n[Visual Anchor]\nA photorealistic cinematic 8-second sequence inside a modern luxury high-speed train. The woman remains seated comfortably beside the panoramic windows while the train continuously moves forward at normal high-speed rail velocity.\n\nONLY THE WORLD OUTSIDE THE TRAIN travels backward through history.\n\nThe train interior NEVER changes.\nThe woman NEVER changes.\nHer clothing NEVER changes.\nThe windows NEVER change.\nThe train NEVER changes into an older vehicle.\n\nThe woman remains in normal real-time throughout, calmly enjoying the impossible scenery. She breathes naturally, occasionally blinks, subtly follows the landscape with her eyes, and gives a faint knowing smile. She is never frightened, startled, confused, or alarmed. She behaves as though she already knew this would happen.\n\nThe historical effect uses a repeating rhythm:\n\nSTABLE ERA → RAPID BACKWARD TIME-LAPSE TRANSFORMATION → STABLE ERA → RAPID BACKWARD TIME-LAPSE TRANSFORMATION.\n\nDuring each stable era, the exterior world moves and behaves at NORMAL REAL-TIME SPEED while the train continues traveling through it.\n\nDuring each transition, ONLY the exterior environment rapidly rewinds through time.\n\n[00:00.000–00:01.000 — PRESENT DAY]\n\nBegin EXACTLY from <Picture 1>.\n\nThe modern train moves naturally forward through the cold snowy alpine landscape.\n\nStrong realistic parallax is visible outside: nearby snow-covered terrain moves quickly past the windows while the distant mountains move more slowly.\n\nThe woman quietly watches the scenery.\n\nNatural breathing.\nOne subtle blink.\nSmall relaxed eye movement.\n\nThe snowy mountain landscape remains stable long enough for the viewer to recognize the present-day setting.\n\n[00:01.000–00:01.500 — FIRST TIME REWIND]\n\nThe ENTIRE exterior world suddenly begins rapidly traveling backward through time like an accelerated environmental time-lapse.\n\nThere is NO camera cut.\n\nSnow rapidly retreats across the entire landscape.\n\nFrozen ground becomes exposed earth.\n\nGrass spreads rapidly.\n\nTrees and vegetation appear and grow across the mountains and valleys.\n\nThe cold white alpine environment smoothly transforms into a lush green temperate landscape.\n\nThe transformation happens across ALL visible windows simultaneously.\n\nThe woman and modern train remain completely unchanged and continue moving in normal real-time.\n\n[00:01.500–00:02.500 — LUSH GREEN LAND]\n\nThe time-lapse transformation STOPS.\n\nThe exterior is now a rich green countryside.\n\nForests, grassy hills, streams, wild vegetation, and distant mountains move naturally past the train.\n\nEverything outside now behaves at normal real-time speed.\n\nLeaves move in the wind.\nWater flows naturally.\nThe train continues traveling forward.\n\nThe woman calmly watches and gives a very slight peaceful smile.\n\n[00:02.500–00:03.000 — SECOND TIME REWIND]\n\nThe exterior world rapidly rewinds again.\n\nThe green countryside transforms through accelerated historical change.\n\nRoads appear as dirt tracks.\n\nOld fences develop.\n\nBuildings rapidly emerge into 19th-century construction.\n\nThe entire landscape transitions smoothly into the 1800s.\n\nNo cut.\n\n[00:03.000–00:04.100 — 1800s]\n\nThe transformation STOPS.\n\nThe train now passes through a living 19th-century town and countryside.\n\nThe exterior operates at NORMAL SPEED.\n\nHorse-drawn carriages travel along dirt streets.\n\nHorses walk and trot naturally.\n\nPeople wearing 1800s clothing move along sidewalks and streets.\n\nOld masonry and timber buildings, storefronts, chimneys, wooden fences, wagons, and period street activity pass naturally outside.\n\nNO modern vehicles.\nNO modern signs.\nNO modern technology outside.\n\nThe train continues moving rapidly through the historical environment.\n\nThe woman remains relaxed and simply watches.\n\n[00:04.100–00:04.600 — THIRD TIME REWIND]\n\nHistory rapidly rewinds again.\n\nThe 1800s town begins disappearing through a smooth accelerated time-lapse.\n\nBuildings regress and vanish.\n\nRoads deteriorate into ancient tracks.\n\nModern-style structures disappear completely.\n\nThe landscape rapidly transforms into the ancient Roman era.\n\nThe transformation occurs across the ENTIRE exterior view simultaneously.\n\n[00:04.600–00:05.800 — ROMAN ERA]\n\nThe transformation STOPS.\n\nThe train now passes beside a large ancient battlefield.\n\nEverything outside moves at NORMAL REAL-TIME SPEED.\n\nRoman soldiers in historically inspired armor fight opposing barbarian warriors across an open field.\n\nSoldiers run.\nShields collide.\nSpears move.\nBanners flap in the wind.\nHorses cross portions of the battlefield.\nDust rises beneath their feet.\n\nThe battle remains at a believable distance from the train and remains entirely outside.\n\nNo soldier notices the train.\nNo soldier approaches the windows.\nNo one interacts with the woman.\n\nThe woman watches the battle calmly.\n\nShe gives a small blink and remains completely composed, as though this is expected.\n\n[00:05.800–00:06.300 — MASSIVE TIME REWIND]\n\nThe exterior suddenly accelerates backward across an enormous span of history.\n\nThe Roman soldiers rapidly disappear.\n\nBattlefield activity vanishes.\n\nRoads disappear.\n\nStructures disappear.\n\nHuman civilization disappears completely.\n\nVegetation, terrain, climate, and geological features rapidly transform in a massive backward time-lapse spanning millions of years.\n\nThe modern train and woman remain completely unchanged.\n\n[00:06.300–00:08.000 — PREHISTORIC WORLD]\n\nThe transformation STOPS.\n\nThe train is now traveling through a vast untouched prehistoric landscape.\n\nEverything outside returns to NORMAL REAL-TIME SPEED.\n\nNo humans.\nNo cities.\nNo buildings.\nNo roads.\nNo modern objects.\n\nA broad primitive landscape stretches toward distant mountains with exposed earth, ancient vegetation, sparse forests, mist, and untouched wilderness.\n\nSeveral dinosaurs move naturally in the DISTANT landscape.\n\nA huge long-necked dinosaur slowly walks across the far background.\n\nSmaller dinosaurs move farther away among ancient vegetation.\n\nKeep the dinosaurs at believable distances.\n\nNO dinosaur approaches the train.\nNO dinosaur looks through the windows.\nNO dinosaur attacks.\n\nThe train continues traveling forward normally through this prehistoric world.\n\nThe woman quietly watches the dinosaurs.\n\nNear the final moment, she gives a subtle knowing smile.\n\nEnd while the train is still moving through the prehistoric landscape.\n\n[CRITICAL WINDOW CONTINUITY]\n\nALL panoramic train windows show ONE SINGLE CONTINUOUS EXTERIOR WORLD.\n\nThe windows must NOT behave like separate television screens.\n\nDo NOT place different eras in different window panes.\n\nDo NOT show snow in one window and green countryside in another.\n\nDo NOT show the 1800s in one window and Romans in another.\n\nDo NOT show Romans in one window and dinosaurs in another.\n\nWhenever time changes, the ENTIRE exterior environment visible through EVERY window changes together as one unified world.\n\nThe physical window frames remain normal structural parts of the train, but the exterior beyond them is one continuous landscape.\n\n[CAMERA]\n\nOne continuous interior shot for the full 8 seconds.\n\nMaintain approximately the camera composition established by <Picture 1>.\n\nNo cuts.\nNo exterior shots of the train.\nNo camera leaving the carriage.\n\nSubtle realistic train vibration.\n\nThe woman remains clearly visible in the foreground while the panoramic windows remain large enough for the audience to clearly read every historical era.\n\nStrong realistic exterior parallax continuously proves that the train is physically moving forward.\n\nDuring historical transitions, the camera itself does NOT speed up and the train does NOT accelerate.\n\nOnly history outside rewinds rapidly.\n\noverall_soundscape:\n\nContinuous subdued interior ambience of a modern high-speed train: low rail rumble, soft mechanical vibration, subtle air-conditioning, and muted rushing sound from outside.\n\nThe woman breathes softly and naturally.\n\nEach stable historical era contributes faint exterior ambience filtered realistically through the sealed train windows.\n\nPresent day: soft rushing alpine wind.\n\nGreen countryside: faint birds and natural countryside ambience.\n\n1800s: distant horse hooves, carriage wheels, and faint town activity.\n\nRoman era: distant muffled battle cries, shield impacts, horses, and battlefield movement.\n\nPrehistoric era: distant wind, low animal calls, and faint dinosaur vocalizations.\n\nExterior sounds remain subdued and muffled by the train's glass.\n\nNo dialogue.\n\nnon_diegetic_music:\nN/A\n```",
    "sourceUrl": "https://x.com/EndFolding79421/status/2088060516027785340",
    "sourceLabel": "X 原帖",
    "author": "EndFolding79421",
    "publishedAt": "2026-08-14",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2088060516027785340.mp4",
    "posterUrl": "/posters/x/2088060516027785340.jpg",
    "duration": 8.064,
    "aspectRatio": "161:90",
    "resolution": "360p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2088060516027785340"
  },
  {
    "id": "x-2087870514400502028",
    "title": "蓝色兔耳角色的动画宣传片",
    "titleEn": "A Blue Bunny-Eared Character Promo",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝色角色与潮流道具在灰色展示空间和彩色图形背景间切换。",
    "summaryEn": "A blue character and streetwear props alternate between a gray showcase space and colorful graphics.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2087870514400502028",
    "sourceLabel": "X 原帖",
    "author": "Mayz1169",
    "publishedAt": "2026-08-13",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087870514400502028.mp4",
    "posterUrl": "/posters/x/2087870514400502028.jpg",
    "duration": 12.933515,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "时尚"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087825234053464154",
    "title": "旧手机质感的晨间洗漱",
    "titleEn": "A Morning Bathroom Routine with a Phone-Video Look",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物打哈欠、拿牙刷并整理头发，以固定浴室机位呈现日常。",
    "summaryEn": "A person yawns, holds a toothbrush and fixes their hair in a fixed bathroom view.",
    "prompt": "Subject: Person from Image1. Face and hairstyle must match Image1 exactly; do not change identity. Ignore its outfit, background, pose, and text. Outfit: oversized casual sleepwear, messy morning look. 9:16 vertical.\n\nOld iPhone everyday-video aesthetic throughout: static phone propped on bathroom sink, standard 1x lens, no distortion, heavy noise/compression, slight mirror steam, warm/off-white overhead lighting. Natural skin, visible pores, flyaways, under-eye tiredness. No beauty filter, grading, cinematic/CG look, fisheye, vignette, or legible text.\n\n0–3s: Sleepy, walks into the bathroom holding a toothbrush, yawns, accidentally bumps the door with their shoulder, pauses, and gives the camera a tired “why am I awake?” look.\n\n3–6s: Puts toothpaste on the toothbrush, realizes the toothpaste is almost completely empty, squeezes the tube several times, then stares at it in disbelief and mutters, “you've got to be kidding me.”\n\n6–9s: Turns on the faucet, splashes cold water on their face, looks up into the mirror, pushes their wet messy hair back, and quietly says, “okay… reset.”\n\n9–12s: Opens the bathroom cabinet searching for something, pulls out an old face towel, accidentally drops two small bottles, catches one with their foot, then gives a tiny proud smile toward the camera.\n\n12–15s: Wraps the towel loosely around their shoulders, looks directly into the camera with a sleepy smile and says softly, “that's good enough.” Holds the look for a beat; freeze final frame.\n\nBathroom clutter: toothpaste tube, skincare bottles, towel, toothbrush cup, small everyday items; all labels must remain completely illegible.\n\nSound only: quiet bathroom ambience, door bump, toothpaste squeezing, faucet and water sounds, bottles lightly clicking, fabric rustle, close muffled sleepy voice with slight bathroom reverb. No music.\n\nKeep all cuts and actions in sequence. No skipped scenes, no time jumps, no montage, no on-screen text, no subtitles. Natural awkward pauses, realistic movement, imperfect everyday behavior. Total duration: exactly 15 seconds.",
    "sourceUrl": "https://x.com/itxabdullaa/status/2087825234053464154",
    "sourceLabel": "X 原帖",
    "author": "itxabdullaa",
    "publishedAt": "2026-08-13",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087825234053464154.mp4",
    "posterUrl": "/posters/x/2087825234053464154.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "生活实拍感",
      "照片级写实",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/itxabdullaa/status/2087825234053464154"
  },
  {
    "id": "x-2087762931777716345",
    "title": "黄黑几何城市与反派角色",
    "titleEn": "A Yellow-and-Black Geometric City and Villain",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "三角符号、几何城市与红沙发上的人物组成风格化动画。",
    "summaryEn": "A triangle symbol, geometric city and figure on a red sofa form a stylized animation.",
    "prompt": "Create a 15-second viral-ready stylized 3D animation with a bold graphic-poster aesthetic, combining 3D cel shading, abstract motion graphics, surreal transformations, and extremely dynamic camera movement.\n\nSTYLE: Stylized 3D toon/cel-shaded animation, low-poly geometric forms, hard-edged shadows, clean surfaces, exaggerated proportions, strong silhouettes, minimal textures, high contrast. Make the world feel like a graphic poster transformed into 3D space. Avoid photorealism and generic CGI.\n\nCOLOR PALETTE: Golden yellow, amber, lemon yellow, black, and white dominate the entire video. Metallic gray appears only as a small accent. Red is allowed only on the villain's sofa, liquor, and warning signals.\n\n0–3 SECONDS — INSTANT HOOK:\nOpen with an extreme close-up of a mysterious 3D graphic object rushing directly toward the camera. Suddenly pull the camera backward to reveal a surreal geometric city made of pipes, grids, sliced buildings, blocks, signs, and mechanical structures. Use a dramatic perspective shift and rapid camera rotation.\n\n3–7 SECONDS — ESCALATION:\nThe city begins transforming at impossible speed. Buildings split into geometric pieces, pipes twist through the streets, blocks rotate in mid-air, and graphic symbols explode outward. The camera flies through the environment with aggressive parallax, speed ramps, and seamless transitions.\n\n7–11 SECONDS — CHARACTER + SURPRISE:\nReveal the villain sitting on a distinctive red sofa in the middle of the moving city. The villain makes one dramatic gesture. Instantly, the sofa, surrounding architecture, and entire environment transform into a massive surreal 3D geometric structure. Use a rapid push-in followed by a sudden wide shot.\n\n11–14 SECONDS — CLIMAX:\nEverything accelerates. Yellow and black geometric elements spiral around the camera. Buildings fold, rotate, compress, and reconstruct. Use extreme perspective, fast tracking, graphic match cuts, silhouettes, diagonal masks, distortion, and a powerful final camera movement.\n\n14–15 SECONDS — LOOP:\nThe entire world rapidly collapses into the same graphic object from the opening shot. End on the exact visual composition needed to seamlessly connect back to the first frame, creating a perfect infinite loop.\n\nANIMATION LANGUAGE: Constant motion, squash and stretch, geometric morphing, slicing, flipping, compression, distortion, object transformation, impossible physics, speed ramps, freeze-frame accents, and satisfying visual cause-and-effect.\n\nCAMERA: Extreme push-ins, rapid pull-outs, rotating shots, overhead views, low angles, impossible tracking shots, camera-through-object transitions, sudden perspective changes, and dramatic scale shifts.\n\nFINAL LOOK: Premium, surreal, energetic, highly stylized 3D cel animation that feels like a living graphic poster. Every second must introduce a new visual surprise. Make the first second immediately attention-grabbing and the final second perfectly loopable.",
    "sourceUrl": "https://x.com/ZaraIrahh/status/2087762931777716345",
    "sourceLabel": "X 原帖",
    "author": "ZaraIrahh",
    "publishedAt": "2026-08-13",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087762931777716345.mp4",
    "posterUrl": "/posters/x/2087762931777716345.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动态图形",
      "科幻太空"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ZaraIrahh/status/2087762931777716345"
  },
  {
    "id": "x-2087310945009995972",
    "title": "黑白披风角色造型动画",
    "titleEn": "A Black-and-White Caped Character Animation",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "高反差背景下，披风角色改变姿态并展示服装轮廓。",
    "summaryEn": "A caped character changes pose and displays its silhouette against a high-contrast background.",
    "prompt": null,
    "sourceUrl": "https://x.com/macbethAI/status/2087310945009995972",
    "sourceLabel": "X 原帖",
    "author": "macbethAI",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087310945009995972.mp4",
    "posterUrl": "/posters/x/2087310945009995972.jpg",
    "duration": 15.168,
    "aspectRatio": "1:1",
    "resolution": "1080p",
    "tags": [
      "动画",
      "风格化与实验",
      "3D / CG",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "stylized",
      "cg-3d"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087294907786600770",
    "title": "上海上空漂浮的巨型海洋生物",
    "titleEn": "Giant Marine Creatures above Shanghai",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "鲸鱼与鹦鹉螺等巨型生物漂浮在城市与道路上方。",
    "summaryEn": "Giant whales, a nautilus and other marine creatures float above the city and roads.",
    "prompt": "Subject: Giant marine creatures hovering over Shanghai's skyline, including a massive humpback whale, a giant manta ray, an elegant and ancient nautilus, a swarm of ethereal glowing jellyfish, and a colossal shark. All creatures maintain realistic anatomical structures, immense scale, and slow, weighty movements, with a beautiful, natural, cinematic appearance—never monstrous or cheaply CG. City architecture, crowds, airplanes, the river, vehicles, cloud layers, and landmarks serve as scale references.\n\nScene: A montage of Shanghai's giant wonders advancing from early morning to late night. The timeline progresses from pale blue morning mist, clear morning, post-rain afternoon, orange-purple dusk, and into dense fog at night. Shanghai Tower, Jin Mao Tower, SWFC, Oriental Pearl Tower, Huangpu River, elevated highways, airplane windows, and rain-slicked streets participate in the narrative, carrying space, scale, and emotional shifts.\n\nMusic: No background music or melody—only live environmental sounds and natural sound effects: river wind, distant boat horns, low engine hum inside the aircraft cabin, subtle window vibrations, suppressed passenger exclamations, wind against glass curtain walls, cloud-layer airflow, raindrops, tires rolling over wet pavement, low-frequency vibrations of massive creatures, and the echoing air in the fog.\n\nShot 1 – Early morning: The camera pushes forward steadily at Huangpu River water level, an ultra-wide establishing shot revealing the Lujiazui skyline and Oriental Pearl Tower veiled in pale blue morning mist. A full-bodied humpback whale slowly swims across the sky above the skyscrapers, as if navigating through an ocean of air. The whale's back is tinged dark blue-grey by dawn light, with soft golden rim light along its body contours; as it passes, shadows from its belly glide slowly over rooftops and the river surface, while morning mist trails behind like flowing water. Small boats, riverside vehicles, and tiny figures atop buildings serve as scale references. One whale-tail swing gently pushes aside a patch of fog, and the camera tilts upward along the curve of the tail.\n\nShot 2 – Morning: The camera cuts to the interior of a passenger aircraft flying over Shanghai. The frame begins from a window-seat passenger's POV—first revealing the oval window, faint reflections on the glass, cloud layers, and the distant Huangpu River and Lujiazui skyline shrinking below. Outside the window, a giant manta ray is flying parallel to the aircraft, its enormous wingspan spanning the entire sea of clouds like a living deep-blue aircraft. Its back displays a gradient of deep blue and silver-grey under sunlight, wingtips undulating slowly, revealing soft pale textures and tiny water droplets on its underside. The manta glides gracefully past the window, wing edges brushing through cloud layers, casting massive fluid shadows. Cabin passengers appear only as partial silhouettes—some leaning slowly toward the window, others raising hands to shield their eyes from the sun. The camera cuts to an extreme close-up of the manta's eye and wingtip, then pulls back to the window's wide view, highlighting the immense scale relationship between creature, plane, city, and clouds.\n\nShot 3 – Post-rain afternoon: The camera begins from a reflection in a puddle on a wet elevated highway, the water mirroring grey-blue sky and building silhouettes. The camera slowly rises to an extreme low-angle upward shot, where post-rain clouds drift between buildings like a soft sea surface. A giant nautilus emerges slowly from deep within the mist. Its spiraled shell is composed of pearl white, pale sea-green, soft rose-gold, and muted bronze tones, its surface covered with delicate mother-of-pearl luster, layered growth lines, and traces of rainwater. It resembles not a spaceship, but a floating structure formed by an ancient shell—its spiral architecture elegant, complete, and natural. Sunlight breaks through the clouds, casting soft iridescent refractions across the shell's curved surfaces; the interior opening is filled with semi-transparent mist, occasionally revealing warm golden air within. The shot first establishes the scale relationship between the nautilus and the city with a wide shot, then cuts to beautiful close-ups of spiral textures, water droplets, and shell edges, finally following its slow horizontal traverse across the sky as the shell's curve creates a circular match cut into the next shot.\n\nShot 4 – Dusk: The nautilus's circular curve transforms into the transparent bell of a jellyfish. The sky becomes a gradient of orange-gold, pale pink, lavender purple, and deep cobalt blue, cloud layers bathed in soft pink-gold sunset light. The camera rises slowly from behind the crowd along the Bund waterfront as a swarm of giant glowing jellyfish emerge from distant cloud seas—like a floating dream garden in the sky, drifting slowly toward the Oriental Pearl Tower. The jellyfish bells are translucent with a pearlescent quality, their interiors laced with delicate blue-purple, silver-white, and pale golden light veins that pulse slowly like breathing. Their tentacles are not terrifying long threads, but soft, silk-like light ribbons that bend, overlap, and drift through the clouds like underwater plants and dancing gauze. The jellyfish maintain natural spacing, forming layered foreground, midground, and background planes. As they pass by high-rises, soft blue-purple light shadows flow across glass curtain walls and the Huangpu River like water ripples, the city bathed in a dreamy underwater world. The camera cuts to a macro close-up inside one jellyfish's bell, where tiny light particles flow slowly along its translucent structure, then pulls back to a full composition of the jellyfish swarm, Oriental Pearl Tower, and the dusk city. The crowd appears only as small-scale silhouettes, quietly gazing upward at this beautiful and impossible sky migration.\n\nShot 5 – Late night: The perspective shifts to a dashcam view on Shanghai's elevated highway. Headlights cut through thick fog, wet pavement reflecting orange-yellow streetlights. Distant skyscrapers appear dim and faint, and a massive shadow emerges—then a shark's dorsal fin glides slowly above the fog layer. The camera remains in a low fixed position, carrying only subtle real-car body vibrations. The shark gradually takes form in the fog, its immense body spanning the sky, its vast jaw and skin texture illuminated only by ambient city light, appearing as a cold blue-grey semi-silhouette. The shark's shadow engulfs the car roof and windshield; the driver lets out a short startled gasp, raindrops on the glass darkened by the shadow. Avoid exaggerated glowing effects—emphasize realistic fog bodies, wet road surfaces, and physically accurate light feedback. The oppressive late-night atmosphere forms a stark contrast with the dreamy beauty of the dusk jellyfish.\n\nShot 6 – Night closure: The camera pulls slowly outward from inside the vehicle, rising to an aerial city-wide view above the elevated highway. The giant shark gradually disappears after passing through the cloud layers, the Oriental Pearl Tower appearing impossibly small beneath it. The dense fog closes back in, and distant city lights regain their orderly rhythm. In the puddles on the wet road, blue-purple light shadows of jellyfish and the blurred silhouette of the shark briefly appear, then gradually fade as raindrop ripples spread. This complete montage traces a full timeline of Shanghai's sky giants—from the early morning whale, to the morning manta ray seen through the airplane window, to the afternoon pearl-shell nautilus, to the dusk dreamlike jellyfish swarm, and finally to the late-night giant shark. Each shot maintains clear spatial relationships, real scale, and slow, weighty motion, together resembling a surreal poetic documentary inadvertently captured by city residents.",
    "sourceUrl": "https://x.com/sada_ai/status/2087294907786600770",
    "sourceLabel": "X 原帖",
    "author": "sada_ai",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087294907786600770.mp4",
    "posterUrl": "/posters/x/2087294907786600770.jpg",
    "duration": 15.168,
    "aspectRatio": "3:4",
    "resolution": "768p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "城市街景",
      "动物与生物"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/sada_ai/status/2087294907786600770"
  },
  {
    "id": "x-2087290562781081994",
    "title": "极光海岸的人物音乐片段",
    "titleEn": "A Performer on an Aurora Coast",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在海边演唱，近景逐渐展开为极光天空下的远景。",
    "summaryEn": "A coastal performance moves from close-up to a wider view under an aurora sky.",
    "prompt": null,
    "sourceUrl": "https://x.com/antonioleivag/status/2087290562781081994",
    "sourceLabel": "X 原帖",
    "author": "antonioleivag",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087290562781081994.mp4",
    "posterUrl": "/posters/x/2087290562781081994.jpg",
    "duration": 15.13941,
    "aspectRatio": "160:93",
    "resolution": "372p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出",
      "自然风景"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087282913410764864",
    "title": "楼梯旁的动漫早安片段",
    "titleEn": "An Anime Morning Greeting beside Stairs",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "短发角色站在室外楼梯旁，以轻微动作与表情完成问候片段。",
    "summaryEn": "A short-haired character gives a brief greeting through small movements and expressions beside outdoor stairs.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2087282913410764864",
    "sourceLabel": "X 原帖",
    "author": "Alone1Moon",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087282913410764864.mp4",
    "posterUrl": "/posters/x/2087282913410764864.jpg",
    "duration": 15.168,
    "aspectRatio": "2:3",
    "resolution": "832p",
    "tags": [
      "动画",
      "动漫",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087276933222695045",
    "title": "无限城中的多角色动漫战斗",
    "titleEn": "A Multi-Character Anime Fight in the Infinity Castle",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "多名持刀角色对抗红色触手攻击，刀光与冲击效果交替出现。",
    "summaryEn": "Several sword fighters face red tendril attacks amid blade trails and impact effects.",
    "prompt": "Create a 15-second, 16:9, high-end 2D Japanese anime battle sequence with synchronized stereo audio inside the Infinity Castle. Sharp hand-drawn linework, crisp cel shading, high-contrast lighting, deep blacks, crimson Muzan effects, electric-blue water, vivid pink blade trails, white impact flashes, readable fast choreography, fluid body weight, environmental destruction. No 3D CGI, photorealism, subtitles, UI or extra characters.\nPicture 1 = Muzan Kibutsuji. Keep exact face, white hair, red eyes, pale muscular body, crimson markings and flesh-whip anatomy.\nPicture 2 = Tanjiro Kamado. Keep exact face, forehead scar, hanafuda earrings, green-black checkered haori and Nichirin sword.\nPicture 3 = Giyu Tomioka. Keep exact face, dark hair, split-pattern haori and Nichirin sword.\nPicture 4 = Obanai Iguro. Keep exact face, bandaged mouth, striped haori, Nichirin sword and Kaburamaru.\nPicture 5 = Mitsuri Kanroji. Keep exact face, pink-green braided hair, white haori and flexible Nichirin sword.\nMaintain identity, clothing and weapon continuity throughout.\nSCENE 1 — [0s→3s]\nDamaged Infinity Castle chamber, tatami floor, broken beams, shoji walls, dust. Muzan stands center; Tanjiro front-left, Giyu rear-right, Obanai opposite side.\nLow wide push toward Muzan. Muzan suddenly fires multiple flesh whips. One sweeps close past camera and destroys a wooden pillar; another smashes the tatami floor. Tanjiro dashes away, Giyu sidesteps, Obanai bends beneath a whip.\nAudio: castle creaks, low battle drone, flesh-whip cracks, heavy whooshes, wood splintering.\nSCENE 2 — [3s→6s]\nObanai immediately counterattacks from the right with a winding serpent-like sword path and narrow silver slash trail. Giyu attacks simultaneously from the opposite side using Water Breathing, Fourth Form: Striking Tide, with a strong hand-drawn electric-blue water arc.\nUse one fast curved orbit around Muzan. Muzan twists and launches two flesh whips to intercept both swords. Brief pure-white impact frame on collision, then sparks, blue water fragments and crimson whip trails.\nAudio: sword whistles, subtle hiss, rushing water, whip cracks, heavy impact hit.\nSCENE 3 — [6s→9s]\nTanjiro bursts through dust and sprints directly at Muzan.\nLow side-tracking camera. Tanjiro plants a foot on a broken beam, launches forward and swings diagonally at Muzan.\nMuzan fires three crossing flesh whips: Tanjiro blocks the first, ducks beneath the second, and the third grazes his shoulder and tears his haori, but he keeps attacking.\nBrief extreme close-up of Muzan’s red eye reflecting Tanjiro, then return to the fight.\nAudio: footsteps, cloth snap, breathing, blade whistle, three directional whip cracks, sword impact, cloth tear.\nSCENE 4 — [9s→12s]\nThe Infinity Castle physically destabilizes: floor tilts, walls shift, beams crack, debris slides and falls. Use a wide canted shot.\nMitsuri enters from above already attacking. Her flexible Nichirin sword sweeps in a large vivid-pink elliptical arc. The blade knocks one flesh whip upward, redirects another into a wall and deflects a third while staying visibly connected to her hand.\nBriefly follow the blade, then return to a wide composition.\nAudio: structural groan, cracking wood, falling debris, metallic blade-whip sound, multiple blade/flesh impacts.\nSCENE 5 — [12s→15s]\nMuzan lands at the center of the fractured chamber, briefly lowers his stance, then releases a violent irregular radial burst of flesh whips.\nAt the same time:\nTanjiro attacks from front-left with a rising orange-red sword arc.\nGiyu attacks from rear-right with a large blue water slash.\nObanai attacks low from right with a serpentine sword path.\nMitsuri attacks from range, her flexible pink blade curving through the foreground toward Muzan.\nUse one rapid controlled circular camera move tightening around Muzan.",
    "sourceUrl": "https://x.com/itsphotogptai/status/2087276933222695045",
    "sourceLabel": "X 原帖",
    "author": "itsphotogptai",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087276933222695045.mp4",
    "posterUrl": "/posters/x/2087276933222695045.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "动漫",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/itsphotogptai/status/2087276933222695045"
  },
  {
    "id": "x-2087275532517884174",
    "title": "地铁隧道中的人物行走",
    "titleEn": "A Figure Walking through a Metro Tunnel",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物沿圆形地铁隧道接近镜头，灯光与轨道延伸至远处。",
    "summaryEn": "A figure approaches through a circular metro tunnel with lights and tracks receding behind.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2087275532517884174",
    "sourceLabel": "X 原帖",
    "author": "UnrealRafael",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087275532517884174.mp4",
    "posterUrl": "/posters/x/2087275532517884174.jpg",
    "duration": 15.168,
    "aspectRatio": "121:90",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087251422312116382",
    "title": "邪教走廊中的伪纪录恐怖",
    "titleEn": "Found-Footage Horror in a Cult Corridor",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "摇晃镜头追随逃跑人物，狭窄走廊中突然出现带血角色。",
    "summaryEn": "A shaking camera follows fleeing figures before a bloodied character appears in a narrow corridor.",
    "prompt": "A found-footage horror sequence shot as a terrified documentary crew runs through the interior of a cult compound during a catastrophic ritual. The environment is a maze of narrow concrete hallways lit by flickering fluorescent lights, candles, and distant fire. The camera is chaotic and unstable, with heavy handheld shake, autofocus hunting, smeared motion blur, abrupt zooms, and raw amateur framing. The clip begins in motion, with panicked crew members running and shouting \"Keep moving!\" while ritual chanting and screams echo from all directions. The camera catches brief horrifying glimpses as it passes open doorways: kneeling worshippers, blood-smeared walls, someone dragging a sheet-covered body, another person convulsing on the ground. The operator rounds a corner and nearly collides with a smiling, blood-covered devotee standing directly in the corridor. The devotee stares into the lens for one beat, then charges. Everyone screams, the camera whips backward and crashes sideways into the wall, and the clip cuts out in the middle of the pursuit. No soundtrack, only diegetic audio.",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2087251422312116382",
    "sourceLabel": "X 原帖",
    "author": "cocktailpeanut",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087251422312116382.mp4",
    "posterUrl": "/posters/x/2087251422312116382.jpg",
    "duration": 28.842667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "暗黑恐怖",
      "写实电影感",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2087251422312116382"
  },
  {
    "id": "x-2087233814665814190",
    "title": "街头涂鸦风动态动漫拼贴",
    "titleEn": "A Street-Graffiti Anime Motion Collage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "滑板、红衣角色与喷溅色块形成快速切换的街头图形动画。",
    "summaryEn": "A skateboard, red-clad character and splashed colors form a fast street-graphic animation.",
    "prompt": null,
    "sourceUrl": "https://x.com/steftranquillin/status/2087233814665814190",
    "sourceLabel": "X 原帖",
    "author": "steftranquillin",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087233814665814190.mp4",
    "posterUrl": "/posters/x/2087233814665814190.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087163007491268632",
    "title": "海边酒吧的四人对白",
    "titleEn": "Four-Person Dialogue at a Beach Bar",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "四人围坐热带酒吧桌旁，镜头交替呈现发言与反应。",
    "summaryEn": "Four people sit at a tropical bar table with alternating speaking and reaction shots.",
    "prompt": null,
    "sourceUrl": "https://x.com/donhyeon_choi/status/2087163007491268632",
    "sourceLabel": "X 原帖",
    "author": "donhyeon_choi",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087163007491268632.mp4",
    "posterUrl": "/posters/x/2087163007491268632.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087161642702315868",
    "title": "第一人称探索漆黑地铁通道",
    "titleEn": "A First-Person Walk through a Dark Metro Passage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "第一人称镜头沿潮湿轨道移动，接近黑暗侧门。",
    "summaryEn": "A first-person camera moves along wet tracks toward a dark side doorway.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2087161642702315868",
    "sourceLabel": "X 原帖",
    "author": "UnrealRafael",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087161642702315868.mp4",
    "posterUrl": "/posters/x/2087161642702315868.jpg",
    "duration": 15.168,
    "aspectRatio": "313:180",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "暗黑恐怖",
      "写实电影感",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087152046114038024",
    "title": "复古舞厅摇滚乐队演出",
    "titleEn": "A Rock Band in a Retro Dance Hall",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "粉色西装歌手与乐队在复古舞厅中表演。",
    "summaryEn": "A pink-suited singer and band perform in a retro dance hall.",
    "prompt": null,
    "sourceUrl": "https://x.com/ravfx777/status/2087152046114038024",
    "sourceLabel": "X 原帖",
    "author": "ravfx777",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087152046114038024.mp4",
    "posterUrl": "/posters/x/2087152046114038024.jpg",
    "duration": 12.32,
    "aspectRatio": "106:45",
    "resolution": "360p",
    "tags": [
      "音乐视频",
      "复古胶片",
      "照片级写实",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "retro",
      "photoreal"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087148758954053985",
    "title": "地标建筑上的彩色涂鸦旅行",
    "titleEn": "A Graffiti Journey across Landmarks",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "埃菲尔铁塔、悉尼歌剧院与手绘地球以彩色艺术效果串联。",
    "summaryEn": "The Eiffel Tower, Sydney Opera House and a drawn globe connect through colorful art effects.",
    "prompt": null,
    "sourceUrl": "https://x.com/vkuoo/status/2087148758954053985",
    "sourceLabel": "X 原帖",
    "author": "vkuoo",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087148758954053985.mp4",
    "posterUrl": "/posters/x/2087148758954053985.jpg",
    "duration": 15.168,
    "aspectRatio": "3:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "风格化与实验",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2087116928473006428",
    "title": "ECLIPSE 科幻海报分层动效",
    "titleEn": "Layered Motion for an ECLIPSE Sci-Fi Poster",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物侧脸、机械头盔、飞船与片名从暗处逐层显现。",
    "summaryEn": "A profile, mechanical helmet, spacecraft and title emerge in layers from darkness.",
    "prompt": "Animate the source artwork as a 10-second dark cinematic sci-fi motion poster while preserving the full-frame black background, orange-and-teal palette, female profile, mechanical helmet, spacecraft, skyline, portal, typography and original layout. Begin in near darkness with only a faint orange glow, then construct the poster in layered stages: thin interface marks and corner graphics appear first, the storm clouds and futuristic skyline fade upward from the shadows, the enormous circular portal expands into position, and the woman’s profile slides in from the left as the mechanical helmet assembles behind her through rotating segmented parts. Illuminate the orange ring beside her face, then bring the central spacecraft rapidly forward from the portal before it settles beneath the “ECLIPSE” title, creating a strong 3D pop-out effect without changing the composition. Reveal the title, tagline, credits and release information through sequential type-on, tracking expansion and sharp graphic pop-ins, accompanied by playful but sophisticated digital keystrokes, interface beeps, metallic clicks and soft cinematic impacts. Complete the animation with small fighters crossing the background, subtle hair movement, drifting sparks, pulsing city lights and a slow final push-in, then hold on the fully reconstructed poster with every word sharp and readable.",
    "sourceUrl": "https://x.com/LudovicCreator/status/2087116928473006428",
    "sourceLabel": "X 原帖",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-11",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2087116928473006428.mp4",
    "posterUrl": "/posters/x/2087116928473006428.jpg",
    "duration": 10.208,
    "aspectRatio": "3:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "动态图形",
      "科幻太空"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2087116928473006428"
  },
  {
    "id": "x-2086830161719111800",
    "title": "鲜橙切片与果汁感官广告",
    "titleEn": "A Sensory Ad of Orange Slicing and Juice",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "橙子切片、饮用与瓶装产品特写组成明亮厨房广告。",
    "summaryEn": "Orange slicing, drinking and bottle close-ups form a bright kitchen advertising concept.",
    "prompt": "Project: Pure Sensation (15-Second Beverage Campaign)\nFormat: Text-to-Video with Image Anchor (Start Frame)\n\n[00:00 - 00:03] THE SENSORY HOOK (Macro Slice)\n• Visual & Camera: Starts on the image anchor. Extremely tight macro close-up of a sharp blade slicing through a ripe, juicy orange directly toward the camera lens.\n• Action / Movement: Droplets of fresh orange juice spray toward the lens in extreme slow motion, reflecting bright sunlight.\n• Audio / Sound Design: A sharp, satisfying crisp slicing sound effect followed by a wet splash. No music yet.\n• On-Camera / Voice: None. Entire focus is on tactile texture and visual warmth.\n\n[00:03 - 00:08] RHYTHMIC ASMR POUR\n• Visual & Camera: Close-up macro shots syncing in rhythmic flow—ice hitting the glass, orange juice pouring smoothly, and fine bubbles rising.\n• Action / Movement: A glass fills with vibrant orange juice as clear ice cubes float upward. Light bubbles gently rise to the surface as liquid coats the inner glass wall.\n• Audio / Sound Design: Rhythmic, isolated ASMR sounds: a resonant glass \"CLINK\", a smooth liquid \"POUR\", and a delicate, crisp \"FIZZ\" rising in volume.\n• On-Camera / Voice: None. Pure focus on clean acoustic detail.\n\n[00:08 - 00:12] THE SUNLIGHT TRANSCENDENCE\n• Visual & Camera: Eye-level medium shot of the 24-year-old Black woman taking a slow, refreshing sip from the glass, bathed in golden sunlight.\n• Action / Movement: She takes a sip, closes her eyes momentarily, and exhales softly with a calm, genuine expression. As she places the glass down, bright sunlight refracts through the juice, projecting glowing amber patterns onto the white table surface.\n• Audio / Sound Design: A soft, uplifting acoustic ambient pad swells quietly beneath the natural ASMR sounds.\n• Dialogue (Real-time on-camera): \"That's pure nature.\"\n\n[00:12 - 00:15] THE HERO SIGN-OFF\n• Visual & Camera: Static hero shot of the glass bottle resting beside fresh orange slices on the bright white table.\n• Action / Movement: Light softly glimmers off condensation droplets running down the glass bottle. The composition is balanced, clean, and still.\n• Audio / Sound Design: Soft acoustic note fades out gently alongside a final subtle bubble-pop sound.\n• On-Screen Text / Call to Action: \"Nothing Hidden. Just Nature.\"",
    "sourceUrl": "https://x.com/yourPlugAI/status/2086830161719111800",
    "sourceLabel": "X 原帖",
    "author": "yourPlugAI",
    "publishedAt": "2026-08-10",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086830161719111800.mp4",
    "posterUrl": "/posters/x/2086830161719111800.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/yourPlugAI/status/2086830161719111800"
  },
  {
    "id": "x-2086784467981697343",
    "title": "梦境档案游戏概念片头",
    "titleEn": "A Dream Archive Game Concept Opening",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "漂浮档案、门与眼睛图形组成平面游戏宣传片。",
    "summaryEn": "Floating files, a door and eye graphics form a flat graphic game promo.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2086784467981697343",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-10",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086784467981697343.mp4",
    "posterUrl": "/posters/x/2086784467981697343.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "动态图形",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2086778003422671069",
    "title": "星图与炼金角色游戏宣传片",
    "titleEn": "A Game Promo of Star Maps and Alchemy",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "深色人物、星图与金色仪器图形以绘本式构图切换。",
    "summaryEn": "Dark characters, star maps and gold instrument graphics alternate in storybook compositions.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2086778003422671069",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-10",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086778003422671069.mp4",
    "posterUrl": "/posters/x/2086778003422671069.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "动态图形",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2086773163870871656",
    "title": "乔治在餐馆咨询 AI 的情景喜剧",
    "titleEn": "George Asks AI for Advice in a Diner",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "餐桌近景以人物表情与手势推进情景喜剧片段。",
    "summaryEn": "Table-side close-ups use expressions and gestures to carry a sitcom scene.",
    "prompt": null,
    "sourceUrl": "https://x.com/TechieBySA/status/2086773163870871656",
    "sourceLabel": "X 原帖",
    "author": "TechieBySA",
    "publishedAt": "2026-08-10",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086773163870871656.mp4",
    "posterUrl": "/posters/x/2086773163870871656.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2086765186107801649",
    "title": "风暴中登陆的维京战士",
    "titleEn": "Viking Warriors Land in a Storm",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "战士从长船跳入浅水，在风雨中登岸。",
    "summaryEn": "Warriors jump from a longship into shallow water and land in stormy weather.",
    "prompt": null,
    "sourceUrl": "https://x.com/MichaelGannotti/status/2086765186107801649",
    "sourceLabel": "X 原帖",
    "author": "MichaelGannotti",
    "publishedAt": "2026-08-10",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086765186107801649.mp4",
    "posterUrl": "/posters/x/2086765186107801649.jpg",
    "duration": 10.208,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "动作打斗",
      "自然风景"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2086427292755247409",
    "title": "紫色月轮下的游戏角色宣传片",
    "titleEn": "A Game Character Promo beneath a Violet Moon",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "人物剪影、刀刃与月轮构成紫黑配色的动画宣传片。",
    "summaryEn": "Character silhouettes, blades and moon discs form a violet-black animated promo.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2086427292755247409",
    "sourceLabel": "X 原帖",
    "author": "Mayz1169",
    "publishedAt": "2026-08-09",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086427292755247409.mp4",
    "posterUrl": "/posters/x/2086427292755247409.jpg",
    "duration": 11.400998,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2086324524153544796",
    "title": "小机器人守护最后一朵花",
    "titleEn": "A Tiny Robot Protects a Flower",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "废墟里的机器人为花挡雨，电量耗尽后花朵在晨光中开放。",
    "summaryEn": "A robot shelters a flower in ruins; after its battery fades, the flower opens in morning light.",
    "prompt": "A cinematic 30-second animated short film about a tiny adorable maintenance robot wandering alone through a post-apocalyptic city covered in moss and rain. The robot discovers a tiny wilted flower growing through a crack in the concrete. Fascinated, it carefully shields the flower from heavy rain using its small metal body. Its battery indicator slowly drops toward zero. The robot finally collapses beside the flower. Dawn arrives, warm golden sunlight breaks through the clouds, and the flower suddenly blooms beautifully beside the motionless robot. Emotional visual storytelling, no dialogue, expressive body language, Pixar-quality 3D animation, cinematic lighting, detailed environment, soft volumetric light, shallow depth of field, emotional orchestral score, dramatic final reveal.",
    "sourceUrl": "https://x.com/iamrealsnow/status/2086324524153544796",
    "sourceLabel": "X 原帖",
    "author": "iamrealsnow",
    "publishedAt": "2026-08-09",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086324524153544796.mp4",
    "posterUrl": "/posters/x/2086324524153544796.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "动物与生物",
      "自然风景"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/iamrealsnow/status/2086324524153544796"
  },
  {
    "id": "x-2086142925063516348",
    "title": "书页上浮现的彩色光绘",
    "titleEn": "Colored Light Drawings Rise from a Book",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "打开的书页间出现彩色光带、星形与灯泡图案。",
    "summaryEn": "Colored light trails, stars and a bulb illustration appear between open book pages.",
    "prompt": null,
    "sourceUrl": "https://x.com/godofprompt/status/2086142925063516348",
    "sourceLabel": "X 原帖",
    "author": "godofprompt",
    "publishedAt": "2026-08-08",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2086142925063516348.mp4",
    "posterUrl": "/posters/x/2086142925063516348.jpg",
    "duration": 15.232,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动态图形",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085860538291040735",
    "title": "为抽帧设计的客厅人物镜头集",
    "titleEn": "Living-Room Character Shots for Frame Extraction",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "同一人物在客厅中快速切换角度与姿势，原作用于抽取静态图。",
    "summaryEn": "The same person rapidly changes angle and pose in a living room; the source intends the clip for still-frame extraction.",
    "prompt": "Photorealistic live-action.\n\nThe same young man is always inside the same modern living room.\n\nThe living room itself remains consistent throughout:\nsame sofa, same coffee table, same curtains, same walls, same general furniture layout.\n\nThis video is intentionally designed as a rapid collection of different still-photo variations for frame extraction.\n\nEvery 0.2-0.4 seconds, abruptly switch to a completely different shot.\n\nInstant jump cuts.\nNo transitions.\nNo smooth camera movement between shots.\nNo morphing.\nContinuity between shots is NOT required.\n\nKeep exactly the same person, facial identity, age, body proportions and general hairstyle throughout.\n\nRapidly alternate between different camera angles and distances:\n\n- front-facing medium shot sitting on the sofa\n- three-quarter angle from the left\n- three-quarter angle from the right\n- profile shot from the side\n- slightly high-angle shot looking down at him\n- subtle low-angle shot from near the coffee table\n- close-up portrait from the sofa\n- chest-up shot\n- waist-up shot\n- wider shot showing most of the living room\n- shot from behind the sofa looking toward him\n- shot across the coffee table\n- camera positioned near the television\n- camera positioned near the window\n- candid side angle as if photographed by another person in the room\n- close shot from slightly below eye level\n- slightly distant candid photograph\n\nHis position and pose also change instantly:\n\n- sitting upright on the sofa\n- leaning back casually\n- resting one arm on the sofa\n- leaning forward with elbows near knees\n- sitting sideways\n- looking directly into camera\n- looking away from camera\n- looking toward the window\n- looking down at his smartphone\n- holding a glass of water\n- empty hands\n- touching his hair\n- one hand behind his head\n- relaxed hands on his lap\n- standing beside the sofa\n- sitting on the floor leaning against the sofa\n\nCreate subtle appearance variations between shots:\n\n- white T-shirt\n- black T-shirt\n- gray T-shirt\n- navy T-shirt\n- slightly oversized white T-shirt\n- thin black-framed glasses\n- no glasses\n- slightly messy bangs\n- bangs falling slightly across the forehead\n- neatly arranged bangs\n- slightly tousled bed hair\n- neutral expression\n- sleepy expression\n- subtle gentle smile\n- faint amused smile\n\nDo not dramatically change the hairstyle.\nOnly subtle natural differences in hair arrangement.\n\nLighting can vary very slightly between shots while still feeling like the same home:\nsoft daylight,\nslightly cloudy daylight,\nwarm indoor light,\nwindow light on one side of the face.\n\nEach composition should remain visible for approximately 0.2-0.4 seconds before instantly switching to another variation.\n\nPrioritize:\nsharp facial details,\nnatural skin texture,\nrealistic hair strands,\nclean eyes,\nnatural hands,\nphotographic realism,\nusable individual still frames.\n\nAvoid motion blur.\nAvoid distorted transition frames.\nAvoid exaggerated body movement.\nAvoid dramatic action.\nAvoid changing the room.\nAvoid changing facial identity.\n\nThe final result should feel like dozens of different casual lifestyle photographs taken in the same living room and rapidly edited together.\nThe video itself does not need to have natural continuity.\nIndividual frames are more important than temporal consistency.",
    "sourceUrl": "https://x.com/luxaios/status/2085860538291040735",
    "sourceLabel": "X 原帖",
    "author": "luxaios",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085860538291040735.mp4",
    "posterUrl": "/posters/x/2085860538291040735.jpg",
    "duration": 15.32517,
    "aspectRatio": "4:7",
    "resolution": "720p",
    "tags": [
      "本地生成与工作流",
      "照片级写实",
      "日常生活"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/luxaios/status/2085860538291040735"
  },
  {
    "id": "x-2085831017055850725",
    "title": "庭院中的双忍者近身交锋",
    "titleEn": "Two Ninjas Fight in a Courtyard",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "黑红忍者在庭院中以刀剑、闪避与倒地动作交锋。",
    "summaryEn": "Black-and-red ninjas exchange blade attacks, dodges and knockdowns in a courtyard.",
    "prompt": "Two elite ninja's having a ferocious fight , cinematic fight scene.",
    "sourceUrl": "https://x.com/0xvaibhav24/status/2085831017055850725",
    "sourceLabel": "X 原帖",
    "author": "0xvaibhav24",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085831017055850725.mp4",
    "posterUrl": "/posters/x/2085831017055850725.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/0xvaibhav24/status/2085831017055850725"
  },
  {
    "id": "x-2085806303562412221",
    "title": "墨镜小狗迎战街猫",
    "titleEn": "A Sunglasses-Wearing Dog Faces a Street Cat",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "霓虹擂台中白色小狗与猫交手，片尾出现胜利标语。",
    "summaryEn": "A white dog fights a cat in a neon arena before a victory title appears.",
    "prompt": "Epic Mortal Kombat-style arena fight. A fluffy white Shih Tzu wearing cool black sunglasses faces off against a fierce street cat in a dark, neon-lit combat stage. Dramatic low-angle camera, heavy cinematic lighting, sparks flying.\n\nThe white Shih Tzu stands confidently in a fighter stance, sunglasses reflecting the neon lights. The cat lunges with sharp claws. The Shih Tzu dodges with slick agility, then counters with a powerful spinning body slam that sends the cat flying across the arena.\n\nThe Shih Tzu follows up with a rapid flurry of paw strikes and a final devastating uppercut, launching the cat into the air. The cat crashes down defeated as the white Shih Tzu stands tall, brushes its fur, and adjusts its sunglasses with pure swagger.\n\nFlawless Victory text flashes. High-energy, stylish, cinematic combat, slow-motion impacts, dramatic camera angles, intense atmosphere.",
    "sourceUrl": "https://x.com/TheAwakenOne619/status/2085806303562412221",
    "sourceLabel": "X 原帖",
    "author": "TheAwakenOne619",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085806303562412221.mp4",
    "posterUrl": "/posters/x/2085806303562412221.jpg",
    "duration": 5.248,
    "aspectRatio": "9:5",
    "resolution": "480p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动作打斗",
      "动物与生物"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TheAwakenOne619/status/2085806303562412221"
  },
  {
    "id": "x-2085781953979244759",
    "title": "向日葵旁的开心小猫",
    "titleEn": "A Happy Cat beside Sunflowers",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "插画小猫向前走动；作者指出嘴部在较长时间内保持张开。",
    "summaryEn": "An illustrated kitten walks forward; the creator notes that its mouth stays open for much of the clip.",
    "prompt": "happy cat with a cute wake, camera tracks cat.",
    "sourceUrl": "https://x.com/softpoo/status/2085781953979244759",
    "sourceLabel": "X 原帖",
    "author": "softpoo",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085781953979244759.mp4",
    "posterUrl": "/posters/x/2085781953979244759.jpg",
    "duration": 15.168,
    "aspectRatio": "3:4",
    "resolution": "768p",
    "tags": [
      "动画",
      "插画与手绘",
      "动物与生物"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/softpoo/status/2085781953979244759"
  },
  {
    "id": "x-2085763884883792369",
    "title": "经典星舰舰桥对白片段",
    "titleEn": "A Classic Starship Bridge Conversation",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "舰桥人物在扫描仪与屏幕之间对话，保留复古科幻布景风格。",
    "summaryEn": "Bridge characters converse between a scanner and display in a retro science-fiction setting.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Live-action in the exact style and aesthetics of the original Star Trek series (1966–1969), with period-accurate film grain, lighting, colors and costumes. The scene begins exactly as shown in <Picture 1>: Spock, wearing his blue uniform, is leaning forward over the console, looking into the scanner device with his right hand resting on it. He raises an eyebrow slightly and says with precise lip-sync: <d>[English] Fascinating.</d>\n\n[Shot 2] At 00:03.200, the camera cuts to a medium shot of Captain Kirk wearing his golden uniform standing on the bridge, looking toward Spock with curiosity. Kirk asks with precise lip-sync: <d>[English] What’s it, Spock?</d>\n\n[Shot 3] At 00:05.500, the camera cuts back to Spock in the same position as the first frame. He briefly glances toward Kirk and replies with precise lip-sync: <d>[English] Science, Captain!</d> He then returns his full attention to the scanner.\n\n[Shot 4] At 00:07.800, the camera cuts to a point-of-view shot through the scanner’s viewscreen. On the glowing screen appears a vision of an elegant Vulcan woman with pointed ears and long dark hair, dancing gracefully in a revealing swimsuit against a soft abstract background.\n\noverall_soundscape: Soft electronic beeps and hums from the scanner and bridge consoles, subtle original Star Trek ambient bridge noise, clear dialogue.\n\nnon_diegetic_music: Subtle original Star Trek-style electronic underscoring that remains low under the dialogue.",
    "sourceUrl": "https://x.com/Cardoso/status/2085763884883792369",
    "sourceLabel": "X 原帖",
    "author": "Cardoso",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085763884883792369.mp4",
    "posterUrl": "/posters/x/2085763884883792369.jpg",
    "duration": 10.208,
    "aspectRatio": "121:90",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "复古胶片",
      "照片级写实",
      "科幻太空"
    ],
    "category": "dialogue",
    "styles": [
      "retro",
      "photoreal"
    ],
    "scenes": [
      "sci-fi",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Cardoso/status/2085763884883792369"
  },
  {
    "id": "x-2085758089354178629",
    "title": "地下杂志风时装音乐拼贴",
    "titleEn": "An Underground-Zine Fashion Music Collage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "时装人物、密集黑白文字与纸张纹理组成竖屏音乐画面。",
    "summaryEn": "Fashion figures, dense monochrome type and paper textures form a vertical music collage.",
    "prompt": "Project Style\nDark-Pop / Cyber-Grunge / Underground Rap MV\nAesthetic Core\nY2K Independent Magazine Aesthetic × Film Print Texture × High-Fashion Editorial\n\n1. Visual Tone & Vibe\nStyle Direction\nA fusion of late-90s to early-2000s underground Zine culture and high-fashion editorial aesthetics.\n\nCold, rebellious, high-contrast, and aggressive visual language.\nHardcore attitude with a strong anti-mainstream identity.\nAvoid cheap internet aesthetics — aim for premium, avant-garde fashion magazine quality.\n\nColor & Lighting\nHigh-contrast black-and-white foundation with selective bursts of saturated Cyber colors:\n\nToxic neon green\nDeep purple-red\n\nScanner blue\nHard lighting, sharp shadows, and razor-cut transitions between light and dark areas.\n2. Texture & Visual Details\n\nMedium Simulation\n35mm coarse-grain film scan texture\n\nXerox machine static noise and toner particles\nHalftone dot printing patterns\nImperfection Aesthetic\n\nTorn print edges and rough paper textures\n\nScanner-style RGB misalignment (chromatic aberration)\n\nSubtle film movement:\nFilm jitter / gate weave effect\n\n3. Editing & Motion Language\n\nRhythm\nUltra-fast BPM-driven cuts with pure minimalism.\nTransition Logic\n\n100% Hard Cuts Only.\n\nStrictly avoid:\n❌ Crossfade\n❌ Fade in / Fade out\n❌ Soft blending transitions\n❌ Standard wipe transitions\nCreate rhythm through brutal visual collisions, sharp contrast changes, and instantaneous image impact.\n4. Typography & Graphic Design\n\nLayout Aesthetic\n\nInspired by late-90s underground music posters, DIY stickers, and Zine collage design.\n\nTypography Style\n\nUse:\n\nHigh-contrast sans-serif fonts\n\nHeavy brutalist blackletter-inspired typography\n\nAdd:\n\nScanner distortion and stretched text effects\nHalftone ink printing textures\nXerox overprint imperfections\nTorn photocopy paper edges\n\nFolded paper wrinkles and physical print artifacts",
    "sourceUrl": "https://x.com/medfac_opc/status/2085758089354178629",
    "sourceLabel": "X 原帖",
    "author": "medfac_opc",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085758089354178629.mp4",
    "posterUrl": "/posters/x/2085758089354178629.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "风格化与实验",
      "动态图形",
      "时尚"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/medfac_opc/status/2085758089354178629"
  },
  {
    "id": "x-2085702677187395984",
    "title": "音乐节街景与摩天轮时尚片",
    "titleEn": "A Festival Fashion Clip with Streets and a Ferris Wheel",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物穿过音乐节摊位和大型装置，镜头展开至夕阳摩天轮。",
    "summaryEn": "A model passes festival stalls and large installations before a sunset Ferris wheel view.",
    "prompt": null,
    "sourceUrl": "https://x.com/PrometheanAIX/status/2085702677187395984",
    "sourceLabel": "X 原帖",
    "author": "PrometheanAIX",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085702677187395984.mp4",
    "posterUrl": "/posters/x/2085702677187395984.jpg",
    "duration": 15.146667,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "时尚",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085685341583991206",
    "title": "裁缝店里的夸张形象改造",
    "titleEn": "An Exaggerated Makeover in a Tailor Shop",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "路人被卷尺引入裁缝店，随后以西装造型重返街道。",
    "summaryEn": "A measuring tape pulls a passerby into a tailor shop before he returns to the street in a suit.",
    "prompt": "A large overweight homeless man with messy long hair, tangled beard, oversized worn-out clothing, dirty shoes, and a tired expression.\nHe is simply walking down the street minding his own business.\nHe has no intention of entering the tailor shop.\nBy the end, he remains large and heavyset, but now looks incredibly stylish, confident, and charismatic.\nTIMELINE\n0:00–0:02 — TARGET SPOTTED\nBusy city street.\nPeople walk past colorful storefronts.\nA large homeless man slowly shuffles down the sidewalk carrying worn bags.\nHis clothes are ripped.\nHis hair is messy.\nHis beard is wild.\nInside the tailor shop—\nA tailor glances through the window.\nSuddenly—\nHis eyes widen.\nExtreme close-up.\nTarget acquired.\nDING.\nThe tailor bursts through the front door.\n0:02–0:05 — THE ABDUCTION\nFWIP.\nThe measuring tape launches through the air like a lasso.\nThe homeless man freezes.\nBefore he can react—\nYANK.\nHe is pulled toward the tailor shop.\nThe camera aggressively follows.\nThe shop door swings open.\nFabric flies everywhere.\nThe homeless man slides across the floor in complete confusion.\nHis eyes widen.\nWhat is happening?\n0:05–0:08 — FASHION HURRICANE\nThe tailor enters full speed mode.\nFWIP.\nTape measure wraps around shoulders.\nFWIP.\nAround waist.\nFWIP.\nAround arms.\nThe homeless man spins slowly on a fitting platform.\nColorful fabrics explode through the room.\nScissors flash.\nThreads fly.\nButtons launch through the air.\nMirror reflections show the tailor appearing in multiple places simultaneously.\nThe homeless man struggles to follow anything.\n0:08–0:11 — LEGENDARY TAILOR TECHNIQUE\nThe tailor unleashes his ultimate fashion combo.\nSHK.\nSNIP.\nBRRRRRRT.\nThe sewing machine activates like a race car engine.\nFabric pieces fly around the room.\nJackets appear.\nVests appear.\nAccessories appear.\nEverything assembles midair.\nThe camera spins around the fitting platform.\nThe homeless man becomes surrounded by a cyclone of clothing and fabric.\nOnly occasional glimpses reveal the transformation happening underneath.\n0:11–0:13 — FINAL ADJUSTMENTS\nEverything suddenly slows down.\nExtreme close-ups.\nA collar aligned.\nA pocket square adjusted.\nA button secured.\nA watch straightened.\nA jacket sleeve perfectly tailored.\nWarm sunlight glows through floating fabric fibers.\nThe tailor calmly inspects the silhouette.\nPerfect.\n0:13–0:15 — THE REVEAL\nThe fabric cyclone disappears.\nSilence.\nThe homeless man stands in front of a giant mirror.\nHis eyes slowly widen.\nHe remains large and heavyset.\nBut now he looks incredible.\nA perfectly tailored suit.\nElegant layers.\nLuxury styling.\nConfident posture.\nMovie-star presence.\nPeople outside the shop stop and stare through the window.\nOne passerby drops their coffee.\nThe tailor casually folds his scissors.\nCHK.\nThe transformed man adjusts his jacket.\nA confident smile appears.\nFinal cinematic shot:\nHe walks back onto the street.\nThe same sidewalk.\nThe same city.\nBut now everyone turns their heads as he passes.",
    "sourceUrl": "https://x.com/im_shahid7/status/2085685341583991206",
    "sourceLabel": "X 原帖",
    "author": "im_shahid7",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085685341583991206.mp4",
    "posterUrl": "/posters/x/2085685341583991206.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "时尚",
      "城市街景"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/im_shahid7/status/2085685341583991206"
  },
  {
    "id": "x-2085681358077137144",
    "title": "ONE HIT 时装换装与杂志版式",
    "titleEn": "ONE HIT Outfit Changes and Editorial Graphics",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "同一人物切换服装，红黑白杂志文字与网格随镜头变化。",
    "summaryEn": "The same figure changes outfits as red-black-white editorial type and grids shift around her.",
    "prompt": "Create a 15-second 16:9 fashion lookbook MV combining rapid outfit changes, magazine editorial graphics, and beat-synced transitions.\nThe video is not narrative-driven. Core structure:\nLocked camera → Typography reveal → Close-up → Rapid Look switches → Hero Ending\nKeep the same female character centered in frame throughout the video. Preserve facial identity, body proportions, skin tone, and overall appearance across every outfit.\n0.0–1.5s\nPure white studio, girl standing naturally in the center.\nCamera locked.\nHuge black typography “ONE HIT” rapidly expands behind her.\nMagazine elements appear on the beat:\nbarcode, scan lines, crosshair, numbers, thin borders, grid graphics.\nThe clean studio instantly becomes a bold fashion magazine poster.\n1.5–3.0s\nFast camera push-in to a close-up.\nThe girl gently turns toward camera, hair moving naturally, direct eye contact, calm and confident expression.\nBlack-and-white editorial background with subtle film grain.\n3.0–4.5s\nLook 01.\nInstant outfit change on the beat.\nCharacter stays centered and makes a natural fashion pose.\nBackground switches to a completely new magazine layout with oversized headline, barcode and numbering.\n4.5–6.0s\nLook 02.\nNew outfit.\nMedium shot. Girl takes one confident step forward.\nBackground typography rapidly changes.\nBlack and white remain dominant with stronger red accents.\n6.0–7.5s\nLook 03.\nNew outfit.\nGirl makes a gentle half-turn, hair swinging naturally.\nBackground changes to red-and-white editorial graphics with huge English typography, numbers, coordinates, scan lines and borders.\n7.5–9.0s\nLook 04.\nSporty fashion outfit.\nGirl leans slightly forward with a relaxed confident smile.\nSubtle low-angle camera.\nLarge red magazine typography slides into frame on the beat.\n9.0–10.5s\nLook 05.\nNew outfit.\nGirl moves into a stylish crouching pose.\nQuick camera push-in.\nBackground layout completely rearranges, creating a new visual impact.\n10.5–12.0s\nLook 06.\nGirl stands again.\nCamera pulls back to medium shot.\nBold yellow fashion uniform with long blue hair styling.\nBackground returns to black-and-white editorial design with large headline and barcode elements.\n12.0–13.5s\nLook 07.\nSharp tailored suit.\nGirl walks confidently toward camera.\nBackground switches to red-and-black magazine design.\nNew headline, number codes, scan lines and graphic grids appear rapidly.\nPacing becomes faster.\n13.5–15.0s\nFinal Hero Shot.\nGirl wears a minimal black fashion look.\nArms crossed, standing confidently in the center.\nCamera settles into a clean medium shot.\nBackground gradually simplifies, leaving only a few premium magazine graphics.\nFinal logo or visual mark appears naturally.\nFreeze on the final beat.\nOverall Style\nPremium Korean fashion lookbook, editorial fashion photography, Y2K magazine design, high-end commercial photography.\nBlack and white base with red accents.\nOversized English typography, barcode, scan lines, crosshair, coordinates, numbering, grids, thin borders, subtle film grain.\nSwitch to a new outfit or visual layout every 1–1.5 seconds, synchronized closely with the music beat.\nKeep the camera relatively controlled. Use quick push-ins, subtle pull-backs, pose changes, graphic transitions, and layout swaps.\nCharacter must always remain near the center of frame.\nMaintain realistic hair movement, fabric physics, natural poses, clean studio lighting, and high-contrast premium photography.\nAvoid\nNo face drift, character replacement, body morphing, extra limbs, distorted hands, unstable proportions, messy typography, chaotic camera movement, random backgrounds, cartoon style, low-quality CGI, watermark, or unwanted subtitles.\nFinal result should feel like a high-end Korean fashion campaign mixed with a Y2K magazine cover and beat-synced music video.",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2085681358077137144",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085681358077137144.mp4",
    "posterUrl": "/posters/x/2085681358077137144.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "2160p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "动态图形",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2085681358077137144"
  },
  {
    "id": "x-2085647629870924078",
    "title": "街头与豪宅间的女声嘻哈片",
    "titleEn": "A Female Hip-Hop Clip between Streets and a Mansion",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "说唱人物在霓虹街道与华丽室内表演，镜头交替切换。",
    "summaryEn": "A rapper performs in neon streets and an ornate interior through alternating shots.",
    "prompt": "Create a cinematic hip-hop music video featuring Alexa Grey as a fearless rapper in stylish streetwear. The song is about a confident woman unexpectedly falling for a guy, mixing bold rap verses with a catchy hook—fast-paced edits. Dynamic camera movements. Modern hip-hop energy. Music video aesthetic.",
    "sourceUrl": "https://x.com/MonetizationDon/status/2085647629870924078",
    "sourceLabel": "X 原帖",
    "author": "MonetizationDon",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085647629870924078.mp4",
    "posterUrl": "/posters/x/2085647629870924078.jpg",
    "duration": 35.271111,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/MonetizationDon/status/2085647629870924078"
  },
  {
    "id": "x-2085638396656099441",
    "title": "模拟图像编辑界面的生成动画",
    "titleEn": "A Generated Animation of an Image-Editing Interface",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "画面模拟编辑软件，在街景图片中移动人物并添加签名。",
    "summaryEn": "A simulated editor moves a figure within a street image and adds a signature.",
    "prompt": "アニメーション化された Photoshop スタイルの画像編集プロセスを表示してください。画像から人物を切り抜き、背後に映る建物の上に立っているように調整、その次に右側の馬を左後景に変更,左側の建物のバグを取り除く作業と、右下に有名画家のサインを追加するように配置してください。まるで本当に photoshopの編集画面を開いたかのようなui",
    "sourceUrl": "https://x.com/ponzponz15/status/2085638396656099441",
    "sourceLabel": "X 原帖",
    "author": "ponzponz15",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085638396656099441.mp4",
    "posterUrl": "/posters/x/2085638396656099441.jpg",
    "duration": 10.208,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "本地生成与工作流",
      "插画与手绘",
      "人物表演"
    ],
    "category": "workflow",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ponzponz15/status/2085638396656099441"
  },
  {
    "id": "x-2085619449131073849",
    "title": "奶茶互动对白工作流演示",
    "titleEn": "A Milk-Tea Dialogue Workflow Demo",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "工作流界面中播放人物喝奶茶、说话并向镜头递杯的结果。",
    "summaryEn": "A workflow interface plays a result in which a woman drinks milk tea, speaks and offers the cup to camera.",
    "prompt": "真人实拍，清新唯美质感，温暖秋日氛围，5秒。\n全程无背景音乐，无BGM，无配乐，无环境声，无风声无鸟鸣。\n只有人物发出的声音：吸吮声、吞咽声、呼吸声、说话声。\n参考图中的人物（手持蜜雪冰城红杯，正在用吸管喝饮料的美女）。\n0—1秒，她正含着吸管小口吸吮（有轻微清晰吸吮声），眼神直视镜头，目光明亮且带满足笑意。发丝在秋日光晕下轻柔拂过脸颊。\n1—2秒，她松开吸管，嘴唇合拢，喉间有清晰吞咽动作（吞咽声），随后轻吸一口气（微弱吸气声）。嘴角上扬，绽放甜美笑容。\n2—3.8秒，她注视镜头，双唇清晰开合，柔声说道：<d>[中文]\"这是秋天的第一杯奶茶，好甜呀~你也要喝吗？\"</d>——声音清晰，语气娇俏，尾音微扬。\n3.8—5秒，话音刚落，她眼角的笑意更浓，双手捧着奶茶的杯子，伸直手臂直接将杯子递向镜头，带有吸管的那一侧几乎怼到屏幕正前方（仿佛真的要喂给屏幕外的人喝），头轻轻向一侧歪了一下。长发随着动作微微晃动，最后画面定格在她将奶茶完全递到镜头前的瞬间。\n【环境声】无。没有风声与杂音。\n【背景音乐】无。全程无任何BGM与配乐。\n【声音控制】仅包含人物发出的声音：吸吮声、吞咽声、吸气声、说话声。人物仅说\"这是秋天的第一杯奶茶，好甜呀~你也要喝吗？\"。\n【表情控制】情绪温暖甜美，避免夸张大笑，说话时唇动明显，声音清晰，眼神和动作充满强烈的互动喂食感。",
    "sourceUrl": "https://x.com/0xmomonifty/status/2085619449131073849",
    "sourceLabel": "X 原帖",
    "author": "0xmomonifty",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085619449131073849.mp4",
    "posterUrl": "/posters/x/2085619449131073849.jpg",
    "duration": 5.504,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "本地生成与工作流",
      "照片级写实",
      "美食",
      "对白与口型"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "food",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/0xmomonifty/status/2085619449131073849"
  },
  {
    "id": "x-2085618840105517104",
    "title": "漆画螺钿风格的分海故事",
    "titleEn": "A Parting-Sea Story in Lacquer and Shell-Inlay Style",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝绿浮雕海浪向两侧分开，人群沿金色道路穿过。",
    "summaryEn": "Blue-green relief waves part as a crowd crosses along a gold-colored path.",
    "prompt": "【核心创意】\n\n15 秒，16:9 横屏视频。东方传统漆画与螺钿镶嵌动画风格。红海、沙漠、天空和人物像一幅厚重的东方漆画浮雕，由黑漆、朱漆、沙金色、深蓝色漆面和蓝绿色螺钿组成。海浪具有贝壳镶嵌的反光纹理，山脉带有金色雕刻线和漆面开片。无对白。\n\n摩西是一名年长男性，灰白色长胡须，穿米色长袍和深棕色披肩，手持木杖。全片保持脸部轮廓、胡须、服装和比例一致。\n\n【画面过程说明】\n\n0–3 秒｜极大全景：\n朱红色漆面天空下，逃亡人群来到深蓝色漆海前。海浪由层叠的蓝绿色螺钿浮雕构成，像两面高大的墙挡住道路。摩西站在人群前方，举起木杖。\n\n3–6 秒｜中景：\n摩西将木杖指向海面。两侧蓝色漆面海浪开始向左右分开，金色雕刻线从海底中央向前延伸，逐渐形成一条沙金色道路。\n\n6–9 秒｜俯拍：\n红海完全分开，两面深蓝色螺钿水墙高高立起。人群沿中央道路向前行走，人物以简洁的漆器浮雕剪影表现，动作清晰有节奏。\n\n9–12 秒｜侧面跟拍：\n镜头沿金色道路向前移动，前景的螺钿水墙产生层叠反光和空间变化。摩西最后走入道路中央，回头确认人群已经通过。\n\n12–15 秒｜大全景：\n人群抵达远方沙金色岸边。摩西放下木杖，两侧漆面海浪缓慢向中央合拢，金色雕刻线逐渐消失，海面恢复平静。镜头拉远，金色阳光照亮摩西和人群。\n\n声音：木杖触地声、漆面震动声、螺钿水纹碰撞声、风声、脚步声、低沉鼓点和庄严弦乐。海水合拢后转为平静流水声。无对白、无旁白。\n\n不想要：纸张、剪纸、真实流体模拟、普通3D海水、现代服装、现代船只、人物飞行、海水爆炸、漆面融化、螺钿碎裂、字幕、Logo、乱码文字、黑屏转场。\n\n必须保持：全片为东方传统漆画与螺钿镶嵌动画；摩西始终穿米色长袍、深棕披肩并手持木杖；红海必须从中央向左右两侧分开；结尾海水重新合拢，但摩西和人群安全抵达岸边。",
    "sourceUrl": "https://x.com/xianshi666123/status/2085618840105517104",
    "sourceLabel": "X 原帖",
    "author": "xianshi666123",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085618840105517104.mp4",
    "posterUrl": "/posters/x/2085618840105517104.jpg",
    "duration": 15.168,
    "aspectRatio": "323:180",
    "resolution": "360p",
    "tags": [
      "动画",
      "插画与手绘",
      "风格化与实验",
      "奇幻",
      "自然风景"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/xianshi666123/status/2085618840105517104"
  },
  {
    "id": "x-2085587415784378498",
    "title": "豪华酒店里的无名英雄说唱",
    "titleEn": "An Unsung-Hero Rap in a Luxury Hotel",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "西装说唱人物穿过金色走廊与楼梯，以快切镜头表演。",
    "summaryEn": "A suited rapper performs through quick cuts in gold corridors and on hotel stairs.",
    "prompt": "Make a rap music video with a cool asian rapper, dressed in a street-style suit, in a fancy hotel, hip beat, the lyric is about a hero without a name or face, he generates AI content for people via DM but gets no credit or recognition, because he's the invisible hero, fast cut",
    "sourceUrl": "https://x.com/WuxiaRocks/status/2085587415784378498",
    "sourceLabel": "X 原帖",
    "author": "WuxiaRocks",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085587415784378498.mp4",
    "posterUrl": "/posters/x/2085587415784378498.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/WuxiaRocks/status/2085587415784378498"
  },
  {
    "id": "x-2085586844088156403",
    "title": "三格人像的独立微动作",
    "titleEn": "Independent Portrait Motion across Three Panels",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "三格画面分别展示同一人物的眼神、侧脸与远近变化；本站播放原帖的 H3 版本。",
    "summaryEn": "Three panels show separate gaze, profile and framing changes; this is the source post's H3 version.",
    "prompt": "8K 超写实 6 秒竖版动态短片，固定三格竖向分屏画幅，分割线保持不变，三个格子画面独立运动，互不串图，朦胧雾感野外黄昏，低饱和暗调电影弱侧光，胶片颗粒质感；浅金色短卷发女性，黑色简约上衣，皮肤雀斑肌理高度写实，微风持续吹动发丝。总时长 6 秒。\n【上方画幅｜0‑6s 独立动态】女子右手缓缓抬至额头侧边，指尖轻抵太阳穴，眼神放空涣散，嘴唇微张，神情疲惫心事重重，发丝被风扫过眉眼，眼皮缓慢轻眨；本格镜头极缓慢向前小幅推镜，焦点锁定眼睛与手部，细微呼吸感镜头抖动。无台词。\n【中间画幅｜0‑6s 独立动态】人物侧脸特写，目光望向远方，睫毛缓缓颤动，眼底慢慢蓄起泪光，发丝不断拂过脸颊鼻尖，嘴唇轻微抿动，克制隐忍悲伤；本格镜头缓慢小幅横向微移，固定面部特写，细微呼吸抖动，不做推拉。无台词。\n【下方画幅｜0‑6s 独立动态】女子面向镜头静静伫立，风吹摆动短发，几缕发丝飘落在眉眼，眼神从涣散慢慢转为沉静漠然，呼吸平缓，只有细微眼睑开合；本格镜头极缓慢向后小幅拉远，由近景过渡至胸像，人物慢慢融入昏暗雾感背景。无台词。\n三格同时播放，人物肢体、面部微表情自然真实，杜绝 AI 僵硬扭曲；皮肤毛孔、发丝细节写实。\n配套 6 秒音频：极简低沉钢琴单音配乐，叠加旷野轻柔风声，整体音量偏低，营造压抑沉思氛围感。",
    "sourceUrl": "https://x.com/ou_zhen599/status/2085586844088156403",
    "sourceLabel": "X 原帖",
    "author": "ou_zhen599",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085586844088156403.mp4",
    "posterUrl": "/posters/x/2085586844088156403.jpg",
    "duration": 6.656,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "照片级写实",
      "人物表演"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2085586844088156403"
  },
  {
    "id": "x-2085577060760862907",
    "title": "向日葵与紫发角色的花卉拼贴",
    "titleEn": "A Floral Collage of Sunflowers and a Violet-Haired Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "角色剪贴、巨型花朵与蓝天色块形成多层动画构图。",
    "summaryEn": "Character cutouts, giant flowers and blue-sky panels form layered animated compositions.",
    "prompt": "How the reference pictures align with the target video — Picture 1 (from Shot 1) aligns with the 0.00-second mark of the target video; Picture 2 (from Shot 1) aligns with the 5.39-second mark of the target video; Picture 3 (from Shot 1) aligns with the 9.10-second mark of the target video.\n\nintegrated_multimodal_description: [Shot 1] 16:9 digital-animation music video, contemporary mixed-media collage with anime illustration, watercolor paper, editorial cutouts, Xerox grain, halftone dots, CMYK misregistration, scanner offsets, nested mattes, and controlled psychedelic instability. Collage is active inside the image at all times, not only during transitions: maintain one dominant subject plus two to five simultaneous cutout, window, reflection, miniature, or texture states in the same frame. Begin with the exact full-body composition, cream negative space, black suit, purple ponytail, gold \"AI\" hair clip, bouquet, shadows, and proportions of <Picture 1>. From 00:00.000 to 00:04.900 the camera performs an almost imperceptible push in while her cast shadow moves in the wrong direction, the bouquet wrapping breathes like thin paper lungs, separate petals rotate at incompatible frame rates, miniature crops of her eye and shoes drift inside the bouquet, and tiny duplicate edges appear one beat late; the face, body, and hands remain clean. At 00:05.387, the first clustered impact tears the cream background into irregular watercolor windows, the black suit separates into floating lapel cutouts, and the bouquet unfolds outward as a multi-plane botanical stage, landing for several crisp frames on the exact teal clothing, purple-haired woman, pale peonies, watercolor texture, and visible \"AI\" and \"AITuber\" text of <Picture 2>. Do not replace <Picture 1>: keep a small full-body suited version walking across the lower paper margin, preserve cream-background shards as moving windows, and let black lapels frame the watercolor portrait. On the impacts at 00:05.611 and 00:05.845, the camera whip-pans left, snaps to a low angle, then rolls clockwise with large amplitude as peony petals become offset contact-sheet tiles of her eyes, hands, ponytail, suit, and bouquet. At 00:07.637 and 00:07.861 the tiles collide, reverse depth, and print a blue sky through her silhouette while watercolor peonies continue blooming inside the black-suit cutouts. From 00:08.864 the camera accelerates through a circular tear in her iris; at 00:09.099 the layers resolve sharply into the exact close-up, blue pop-art sky, sunflower, face, ponytail, gold \"AI\" clip, and halftone composition of <Picture 3>. From 00:09.100 to 00:15.000, keep all three references concurrently alive: place the original full-body figure as a rotating paper figurine inside the sunflower center, use the watercolor peony portrait as translucent cloud-shaped windows across the sky, and retain the pop-art close-up as the dominant plane. Animate the sunflower as a rotating offset-print shutter, duplicate its petals into different perspective planes, slice the blue sky into horizontal strips that travel at contradictory speeds, and orbit the camera around her profile through abrupt 20-degree angle jumps without breaking directional momentum. End at exactly 00:15.000 on a sharply legible synthetic tail frame: her purple-haired profile is centered inside the dark pupil of an enormous yellow sunflower, a miniature black-suited version and pale watercolor peonies remain visible within separate pupil layers, three red hibiscus-shaped paper apertures are entering from screen left, the camera is still rolling clockwise, and torn blue and yellow layers remain suspended mid-collision. No fade, dissolve, generic crossfade, new text, dialogue, or lip movement.",
    "sourceUrl": "https://x.com/ivy432hz/status/2085577060760862907",
    "sourceLabel": "X 原帖",
    "author": "ivy432hz",
    "publishedAt": "2026-08-07",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085577060760862907.mp4",
    "posterUrl": "/posters/x/2085577060760862907.jpg",
    "duration": 15.072,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "插画与手绘",
      "自然风景",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ivy432hz/status/2085577060760862907"
  },
  {
    "id": "x-2085516067292709323",
    "title": "森林魔女的语音与字幕测试",
    "titleEn": "A Forest Witch Voice and Subtitle Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "魔女与森林小动物的片段用于作者比较语音与字幕指令。",
    "summaryEn": "A witch and woodland creatures appear in a creator's voice-and-subtitle instruction test.",
    "prompt": null,
    "sourceUrl": "https://x.com/Sauer20/status/2085516067292709323",
    "sourceLabel": "X 原帖",
    "author": "Sauer20",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085516067292709323.mp4",
    "posterUrl": "/posters/x/2085516067292709323.jpg",
    "duration": 10.208,
    "aspectRatio": "1:1",
    "resolution": "736p",
    "tags": [
      "动画",
      "动漫",
      "奇幻",
      "对白与口型"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085471060028227810",
    "title": "车库前的卡通电工自我介绍",
    "titleEn": "A Cartoon Electrician Introduces Herself at a Garage",
    "model": "MiniMax H3",
    "mode": "Ref2VA",
    "summary": "金发卡通人物在车库前说话并挥手，原帖使用视频作为声音与风格参考。",
    "summaryEn": "A blonde cartoon character speaks and waves outside a garage, using a video for voice and style reference.",
    "prompt": "Use (Main.mp4) as voice and style reference. \n\n no music, multiple camera angles\n\nJenny standing outside next to a car garage on the right side.\n\nThe garage door is already in the fully open position overhead.\n\nJenny smiles confidently at the camera.\n\nJenny:\n\"Hi, I'm Jenny. I'm standing outside of Folding's house and we are seeing if I am moving to Minimax H3 on Comfy u i\"\n\nJenny smiles and waves to the camera.",
    "sourceUrl": "https://x.com/EndFolding79421/status/2085471060028227810",
    "sourceLabel": "X 原帖",
    "author": "EndFolding79421",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085471060028227810.mp4",
    "posterUrl": "/posters/x/2085471060028227810.jpg",
    "duration": 6.656,
    "aspectRatio": "40:23",
    "resolution": "736p",
    "tags": [
      "对白与口型",
      "3D / CG"
    ],
    "category": "dialogue",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "video"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2085471060028227810"
  },
  {
    "id": "x-2085457550816969132",
    "title": "道服狐狸的武术短片",
    "titleEn": "A Martial-Arts Clip with a Fox in a Gi",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "穿道服的狐狸在木质庭院中完成武术动作。",
    "summaryEn": "A fox in a gi performs martial-arts moves in a wooden courtyard.",
    "prompt": "A mini martial arts movie with a male fox character wearing a gi.",
    "sourceUrl": "https://x.com/alby13/status/2085457550816969132",
    "sourceLabel": "X 原帖",
    "author": "alby13",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085457550816969132.mp4",
    "posterUrl": "/posters/x/2085457550816969132.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动作打斗",
      "动物与生物"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/alby13/status/2085457550816969132"
  },
  {
    "id": "x-2085434267232841794",
    "title": "便利店冷柜里的发光图案",
    "titleEn": "A Glowing Figure inside a Store Fridge",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "手持视角接近冷柜，玻璃与柜内出现发光图案。",
    "summaryEn": "A handheld viewpoint approaches a fridge as glowing shapes appear on the glass and inside.",
    "prompt": null,
    "sourceUrl": "https://x.com/sahilvermaai/status/2085434267232841794",
    "sourceLabel": "X 原帖",
    "author": "sahilvermaai",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085434267232841794.mp4",
    "posterUrl": "/posters/x/2085434267232841794.jpg",
    "duration": 15.30195,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "日常生活",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "daily-life",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085413767983046982",
    "title": "多种渲染角色挤进同一电梯",
    "titleEn": "Different Rendering Styles Share One Elevator",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "真人、黏土、二维卡通、低模与漫画角色在同一电梯中互动。",
    "summaryEn": "Live-action, clay, 2D cartoon, low-poly and manga characters interact inside one elevator.",
    "prompt": "A fast-paced scene inside a cramped elevator, filmed head-on in one continuous cinematic shot. Five passengers coexist in their original visual forms: live-action Harry Potter, stop-motion clay Wallace, bright 2D cel-animated SpongeBob SquarePants, classic PS1 low-poly Lara Croft, and Monkey D. Luffy drawn as a rough black-and-white manga sketch. The elevator suddenly jolts violently. Harry grabs the handrail and says, “Does this happen often?” SpongeBob bounces off the walls and cheerfully replies, “Only on Tuesdays!” Wallace’s clay face says, “Cracking lift, this.” Lara Croft draws her blocky pistols as the floor glitches beneath her and says, “Wrong level.” Luffy’s sketched arm stretches across the elevator to press every button, shouting, “Which floor has meat?” The doors burst open onto a rapidly shifting hallway combining all five visual styles. A powerful gust pulls them out one by one: Harry tumbles into a bright cel-animated world, SpongeBob lands in claymation, Wallace falls into a low-poly PS1 corridor, Lara Croft flips into a black-and-white manga panel, and Luffy is thrown into photorealistic live action. Precise lip-sync, clearly separated dialogue, no overlapping speech, energetic physical comedy, believable shadows and contact between every visual medium.",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2085413767983046982",
    "sourceLabel": "X 原帖",
    "author": "cocktailpeanut",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085413767983046982.mp4",
    "posterUrl": "/posters/x/2085413767983046982.jpg",
    "duration": 14.464,
    "aspectRatio": "26:15",
    "resolution": "480p",
    "tags": [
      "动画",
      "风格化与实验",
      "对白与口型",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "dialogue",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2085413767983046982"
  },
  {
    "id": "x-2085403108692176936",
    "title": "记忆侦探游戏概念演示",
    "titleEn": "A Memory Detective Game Concept Demo",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "第三人称角色探索宅邸，漂浮照片与调查界面呈现记忆谜题。",
    "summaryEn": "A third-person character explores a mansion with floating photos and investigation UI.",
    "prompt": "Use Image 1 for environment style. Use Image 2 for investigation UI.\n[0–3s]\nThird-person exploration inside an abandoned mansion. The player walks through a hallway where photographs float in the air.\n[3–6s]\nThe player activates memory vision. The entire room rewinds backward: broken objects rebuild themselves, dust flies upward, candles relight.\n[6–10s]\nThe player moves between past and present versions of the same room. A hidden clue appears only for one second during the transition.\n[10–13s]\nThe camera follows the player stepping through a photograph into a reconstructed memory world.\n[13–15s]\nThe UI reveals the solved mystery. The mansion changes permanently based on the discovered truth.\nAudio: subtle piano, clock ticking, reverse sound effects.\nNegative: no horror gore, no random ghosts, no confusing time jumps.",
    "sourceUrl": "https://x.com/AllaAisling/status/2085403108692176936",
    "sourceLabel": "X 原帖",
    "author": "AllaAisling",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085403108692176936.mp4",
    "posterUrl": "/posters/x/2085403108692176936.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "3D / CG",
      "奇幻"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AllaAisling/status/2085403108692176936"
  },
  {
    "id": "x-2085375185197412412",
    "title": "山谷蝴蝶旁的人物弧线运镜",
    "titleEn": "An Arc Shot around a Figure and Valley Butterflies",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头环绕山谷中的角色，前景蝴蝶与远处村庄提供层次。",
    "summaryEn": "The camera arcs around a valley character with foreground butterflies and a distant village.",
    "prompt": null,
    "sourceUrl": "https://x.com/O_C_T_/status/2085375185197412412",
    "sourceLabel": "X 原帖",
    "author": "O_C_T_",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085375185197412412.mp4",
    "posterUrl": "/posters/x/2085375185197412412.jpg",
    "duration": 10.208,
    "aspectRatio": "161:90",
    "resolution": "360p",
    "tags": [
      "动画",
      "动漫",
      "自然风景"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085374270918820089",
    "title": "芒果糯米饭与粉色荔枝饮品",
    "titleEn": "Mango Sticky Rice and a Pink Lychee Drink",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "椰奶浇淋、荔枝入杯与热带桌面摆盘串起甜品制作。",
    "summaryEn": "Coconut sauce, lychee added to a glass and a tropical table arrangement form a dessert sequence.",
    "prompt": "A cinematic close-up food preparation video of Thai mango sticky rice and a refreshing pink lychee drink. A perfectly molded dome of sticky rice is plated beside thin fan-cut ripe mango slices, drizzled with creamy coconut sauce, topped with toasted sesame seeds and a fresh mint sprig. Ice cubes are added to a tall glass with a light pink drink, followed by a peeled lychee, a lemon wheel garnish, and fresh mint. The final shot showcases the elegant dessert and chilled drink on a rustic wooden table with a bright tropical countryside backdrop of lush green fields, palm trees, cozy huts, and distant mountains under a clear blue sky, featuring soft natural lighting, smooth camera movements, shallow depth of field, and ultra-realistic 4K cinematic quality.",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085374270918820089",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085374270918820089.mp4",
    "posterUrl": "/posters/x/2085374270918820089.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "美食"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085374270918820089"
  },
  {
    "id": "x-2085374036855357478",
    "title": "自行车穿越森林变成摩托",
    "titleEn": "A Bicycle Becomes a Motorbike between Forest and City",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "骑行者从森林跨入都市，自行车转变为发光摩托。",
    "summaryEn": "A rider crosses from forest to city as the bicycle becomes a glowing motorbike.",
    "prompt": "The Screen is divided into two parts, on the left, a solitary hiker ascends through a dense, foggy forest surrounded by towering trees. On the right, the bustling streets of New York City are filled with towering skyscrapers and glowing billboards. The Biker crosses the boundary with Speed , stepping from the forest floor into the city streets with determination, blending the wild with the urban landscape. diring crossing border  time freeze and his bike transform into futuristic super bike and mkve with speed. camera follow shot",
    "sourceUrl": "https://x.com/AIARTGALLARY/status/2085374036855357478",
    "sourceLabel": "X 原帖",
    "author": "AIARTGALLARY",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085374036855357478.mp4",
    "posterUrl": "/posters/x/2085374036855357478.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "城市街景",
      "奇幻"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/AIARTGALLARY/status/2085374036855357478"
  },
  {
    "id": "x-2085369164966105266",
    "title": "动漫角色的双杠力量动作",
    "titleEn": "An Anime Character Performs on Low Parallel Bars",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "橙衣角色在低双杠上完成倒立与支撑动作。",
    "summaryEn": "An orange-clad character performs a handstand and support moves on low parallel bars.",
    "prompt": "[Push in]\n\nSTYLE LOCK: the entire video stays in the exact same visual style, medium, and rendering as the attached reference image in every frame — same art style, same level of realism, same color treatment. Never convert the style: a 2D illustration stays flat 2D animation, a photo stays live-action realistic. No shift into 3D CGI render, no repainting, no style drift.\n\nIDENTITY LOCK: the athlete is the person from the attached reference image. Keep the exact same face, hairstyle, body build, and outfit throughout. Do not redesign the person.\n\nSETTING: a minimalist gym in the same style as the reference image — plain wall, floor mats, a pair of low parallel push-up bars at the center, soft daylight from the left.\n\nACTION, one continuous take: the athlete starts crouched behind the push-up bars, gripping a handle in each hand, then lifts the hips and presses up into a perfectly vertical handstand on the bars — legs locking straight up, toes pointed. Then one slow handstand push-up: elbows bend, the head lowers between the bars below hand level, then the arms press back to full lockout. Then the shoulders tip forward and both legs lower together in a slow arc until the body is horizontal, parallel to the floor — a straight-body planche on the bars, only the hands touching the bars — held one beat. Finally the feet lower to the mat, the athlete releases the bars, rises to a tall stand, takes one breath, and looks directly into the camera with a calm confident expression. Final frame: steady eye contact.\n\nPHYSICS: gravity always applies, both hands stay gripping the bar handles at every moment, the bars never move. Slow, controlled, strength-based movement with correct human biomechanics — no floating, no extra limbs, no morphing. Feet touch the floor only in the final stand-up.\n\nCAMERA: one slow push in from a wide full-body front view to a medium shot, ending on the standing athlete.\n\nAUDIO: quiet room tone and steady breathing only — no voices, mouth stays closed.",
    "sourceUrl": "https://x.com/ramztd/status/2085369164966105266",
    "sourceLabel": "X 原帖",
    "author": "ramztd",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085369164966105266.mp4",
    "posterUrl": "/posters/x/2085369164966105266.jpg",
    "duration": 10.208,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ramztd/status/2085369164966105266"
  },
  {
    "id": "x-2085366255540637965",
    "title": "NOVA X1 镜头拆解概念广告",
    "titleEn": "A NOVA X1 Lens Exploded-View Concept Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头外观、悬浮光学组件与品牌字卡展示摄影器材概念。",
    "summaryEn": "Lens surfaces, floating optical elements and brand cards present a camera equipment concept.",
    "prompt": null,
    "sourceUrl": "https://x.com/AmControo/status/2085366255540637965",
    "sourceLabel": "X 原帖",
    "author": "AmControo",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085366255540637965.mp4",
    "posterUrl": "/posters/x/2085366255540637965.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "3D / CG",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085364318670102958",
    "title": "提前下班的电池图标",
    "titleEn": "A Battery Icon Clocks Out Early",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "手绘电池从电脑中逃离，在真实房间里休息后返回。",
    "summaryEn": "A drawn battery leaves a computer, rests in a real room and returns.",
    "prompt": "15 seconds, 16:9 landscape video. Blend authentic handheld live-action footage of a cluttered bedroom workspace with rough frame-by-frame 2D crayon, colored-pencil, marker, and torn-paper animation. The concept is cute, absurd, relatable, and emotionally healing: a nearly empty computer battery icon refuses to keep working, escapes from the laptop, and takes a tiny “recharge vacation” through the real room before returning fully energized.\n\nThe real environment is a slightly messy bedroom office at late afternoon. Visible objects include a real laptop, wireless mouse, charging cable, ceramic mug, notebook, pencils, desk lamp, small potted plant, rolling desk chair, window, curtains, bed, pillow, and a wall outlet. Warm natural window light mixes imperfectly with the laptop screen and desk lamp.\n\nShoot entirely from the protagonist’s first-person perspective. Only real hands and occasional forearms are visible. The live-action footage must feel accidentally captured on a smartphone: mild hand shake, delayed reframing, imperfect autofocus, visible low-light grain, auto-exposure breathing, rolling-shutter wobble during quick pans, uneven horizon, and natural motion blur. The camera is always slightly late, reacting to the animation rather than predicting it. Do not make the scene feel like a commercial, music video, or polished visual-effects demo.\n\n[Animated Character Identity]\n\nThe main animated character originates from the laptop’s red low-battery icon.\n\nIts consistent identity markers are:\n\na rounded rectangular battery-shaped body\nrough white hand-drawn outline\nred interior fill when tired\ntwo tiny black dot eyes\none short charging-port tail\na small sky-blue vacation cap\na yellow lightning-bolt patch on its side\n\nEvery transformation must preserve at least three clear identity markers: the white outline, sky-blue cap, yellow lightning patch, battery shape, or charging-port tail.\n\nAll additional figures must be temporary doodles created by the battery character. Do not introduce unrelated characters.\n\n[0.0–3.0 seconds — The Battery Quits]\n\nFirst-person close-up of the real laptop on the desk. The protagonist’s right hand types slowly while the other hand rests near the mouse. The laptop screen displays a plain, generic work page with no readable text, no logos, and no recognizable interface branding.\n\nA small red low-battery icon blinks in the screen corner. The protagonist reaches for the charging cable, but before the cable connects, the icon turns toward the camera and shakes its body “no.”\n\nThe red battery icon peels away from the screen like a rough paper sticker. Its edges curl visibly as it climbs over the real laptop bezel and drops onto the keyboard. During the fall, two tiny crayon legs redraw beneath it, followed by dot eyes, a sky-blue cap, and a yellow lightning patch.\n\nIt lands on a real key, compresses flat like a paper accordion, springs back up, wipes imaginary sweat from its forehead, and pulls out a tiny hand-drawn cardboard sign reading only “BREAK.” The word exists physically on the sign and is not an overlay.\n\nThe protagonist points at the charging cable in confusion. The battery pushes the cable plug away with both tiny hands, places its cap firmly on its head, and runs across the keyboard.\n\nCamera: initially static and tired, then a delayed handheld pan follows the battery after it has already begun running.\n\nSound: real keyboard taps, laptop fan, cable brushing against the desk, tiny paper footfalls, one soft comedic spring squeak.\n\nTransition: the running battery leaps from the keyboard toward the notebook.\n\n[3.0–6.0 seconds — The Desk Becomes a Vacation Route]\n\nThe battery lands on the protagonist’s real open notebook. At each footstep, quick turquoise crayon waves appear across the paper. The notebook page transforms only visually through hand-drawn overlays into a tiny seaside map while remaining physically real.\n\nThe battery removes its sky-blue cap, flips it upside down, and the cap rapidly redraws into a miniature blue rowboat. The battery hops inside. Two yellow pencil-line oars grow from its sides.\n\nIt rows across the notebook’s hand-drawn turquoise waves. Small white paper fish leap from one page fold to another. A real pencil lying across the notebook becomes a bridge through animated interaction only; it remains a real pencil.\n\nThe battery boat passes beneath the pencil bridge, then reaches the notebook’s edge. The drawn water pours over the edge like a flat crayon waterfall without wetting the real desk.\n\nMid-fall, the blue rowboat folds into a paper parachute. The battery dangles beneath it by two rough red marker strings and floats toward the ceramic mug.\n\nCamera: overhead handheld view of the notebook, then a slightly late tilt downward following the parachuting battery. Focus briefly hunts between the real pencil and the animated character.\n\nSound: tiny pencil-scratch rowing, paper-water swishes, soft fish pops, page flutter, gentle parachute rustle.\n\nTransition: the battery swings around the mug handle and disappears behind the mug for a moment.\n\n[6.0–9.5 seconds — The Plant Recharging Spa]\n\nThe protagonist quickly moves the camera around the mug. The battery swings from the real mug handle using one red marker string, releases, and lands inside the leaves of the real potted plant.\n\nThe plant remains completely real. Hand-drawn lime-green veins begin glowing softly across only the leaf beneath the battery. The battery removes its cap, lies back comfortably on the leaf, and unfolds into a tiny battery-shaped deck chair while keeping its face and lightning patch.\n\nA hand-drawn yellow sunbeam squeezes through the real window blinds as a wobbly ribbon. It travels across the desk, climbs the plant stem, and connects gently to the battery’s charging-port tail.\n\nThe battery’s red interior begins filling gradually from bottom to top:\n\nred → orange → yellow → bright green.\n\nAs the charge increases, tiny colored-pencil flowers pop into existence around the plant pot. Each flower has a simple face and stretches toward the sunbeam. One flower offers the battery a tiny doodled drink made from a blue paint droplet with a bendy straw.\n\nThe battery sits upright, takes one sip, and instantly overcharges for a comedic beat. Its body expands like a puffy paper balloon, its cap spins, and small yellow lightning doodles orbit its head. It exhales, settles back to normal size, and now glows healthy green.\n\nCamera: gentle push toward the plant, but with imperfect handheld micro-shakes and a small exposure adjustment as the window enters frame.\n\nSound: real room ambience, curtain movement, leaf rustle, soft electrical hum, colored-pencil scribbles, liquid sip, tiny cheerful charging “bloop,” subtle paper balloon squeak.\n\nTransition: the fully charged battery spots the laptop across the room and snaps its cap back on.\n\n[9.5–12.5 seconds — The Ride Back to Work]\n\nThe battery jumps from the plant. While falling, its body stretches horizontally and redraws into a tiny green hand-drawn scooter. The sky-blue cap becomes a small front windscreen, the charging-port tail becomes an exhaust pipe, and the yellow lightning patch remains on the side.\n\nThe scooter lands on the real charging cable running across the desk. Two rough white wheels grip the cable as though it were a raised road.\n\nThe battery character’s face appears on the scooter’s front panel. It races along the cable, leaning dramatically around the mug and pencils. Small yellow lightning-shaped road signs grow briefly from the desk surface, bend as the scooter passes, then erase themselves stroke by stroke.\n\nAt the desk edge, the cable descends toward the real wall outlet. The scooter speeds down it like a roller coaster. The protagonist moves quickly to follow, bumping the rolling chair slightly. The camera dips, shakes, and briefly loses the battery behind the chair leg.\n\nThe battery-scooter emerges from behind the chair, drives up the real charging plug, and launches toward the laptop in an exaggerated slow arc. Midair, the scooter folds inward through visibly redrawn lines, becoming the original green battery character again.\n\nCamera: delayed chase movement, quick pan, slight motion blur, momentary occlusion by the real chair, then a rough push back toward the laptop.\n\nSound: tiny scooter motor buzz made from mouth-like doodle noises, cable vibration, miniature bell ring, chair wheel squeak, air whoosh, paper-folding flaps.\n\nTransition: the battery reaches the laptop screen and presses itself flat against it.\n\n[12.5–15.0 seconds — A Healthier Return]\n\nThe battery character sticks to the laptop screen like a paper decal, then slowly sinks through the glass surface while waving at the protagonist.\n\nIt returns to its original screen position as a fully green battery icon, but the sky-blue cap and tiny yellow lightning patch remain visible for one playful beat before tucking inside the icon.\n\nThe protagonist moves their hand back toward the keyboard. Before typing resumes, the battery icon extends one thin white doodle arm from the screen and gently lowers the laptop lid halfway, refusing immediate overwork.\n\nA second white doodle arm emerges and points toward the real window. Rough colored-pencil lines rapidly sketch a small glowing sunset, two birds, and a cup of tea across the wall beside the desk.\n\nThe protagonist pauses, removes both hands from the keyboard, picks up the real ceramic mug, and leans back.\n\nThe battery icon gives a satisfied green blink. A tiny paper blanket unfolds inside the screen corner. It pulls the blanket over itself, places the sky-blue cap beneath its head as a pillow, and closes its eyes.\n\nThe laptop screen dims naturally but does not shut off completely. The hand-drawn sunset lines remain softly glowing on the wall while the real curtain moves in the breeze.\n\nFinal image: imperfect handheld framing of the quiet desk, half-closed laptop, warm mug in the protagonist’s hand, real plant beside it, and the tiny sleeping green battery visible through the dim screen.\n\nCamera: slow unplanned pull-back, slight tilt, natural focus shift from the sleeping icon to the plant and window.\n\nSound: laptop hinge movement, ceramic mug lift, curtain rustle, distant outdoor birds, computer hum becoming quieter, one tiny sleepy electronic sigh.\n\nFade out naturally with no music.\n\n[Real-World and 2D Integration Style]\n\nThe animated elements must look physically handmade and imperfect:\n\nrough frame-by-frame redraws\nvisible line boil\nuneven crayon pressure\nchalk grain\ncolored-pencil crosshatching\nmarker bleed\ntorn-paper edges\noccasional erased construction lines\nsubtle frame-to-frame shape variation\nsoft glow halo without illuminating the entire room\n\nDo not use smooth vector animation, glossy motion graphics, polished CGI morphing, perfect geometric curves, or clean digital icon animation.\n\nEvery transformation must happen through visible handmade processes:\n\npaper folding\nline stretching\nscribble accumulation\nerased and redrawn limbs\ncrayon strokes connecting\nshapes compressing like accordion paper\noutlines scattering and rebuilding\npaper corners curling\nmarker fills changing color stroke by stroke\n\nThe battery must interact accurately with real objects and obey convincing foreground and background occlusion:\n\nclimbing over the laptop bezel\nstanding on individual keyboard keys\npassing beneath the real pencil\nmoving behind the ceramic mug\nresting on real plant leaves\nfollowing the physical charging cable\ndisappearing behind the chair leg\nentering the laptop screen\n\nReal props must remain live-action objects and must never transform into 3D cartoon versions.\n\n[Camera Rules]\n\nThe handheld camera is always approximately half a beat behind the animated character.\n\nThe battery moves first; the camera reacts afterward through delayed:\n\npans\ntilts\nreframing\nshort forward steps\nimperfect pull-backs\nfocus corrections\n\nPreserve smartphone imperfections:\n\nmild hand shake\nexposure fluctuation near the window\nautofocus hesitation\ndark-area noise\nnatural motion blur\noccasional partial cropping\nslight rolling-shutter distortion during fast movement\n\nDo not use stabilized gimbal movement, drone shots, perfect tracking, impossible live-action camera paths, highly cinematic lens flares, or overly shallow depth of field.\n\n[Color Language]\n\nReal room: warm beige, muted brown, dusty green, dim amber, screen blue.\n\nHand-drawn animation:\n\ntired red\nvacation sky blue\nsoft turquoise\nsunflower yellow\norange\nfully charged green\nrough white outlines\n\nColors must remain vibrant but soft, with visible crayon and paper texture. Avoid neon cyberpunk lighting.\n\n[Sound Design]\n\nNo background music.\n\nUse only synchronized natural and hand-drawn sounds:\n\nlaptop fan and computer hum\nreal keyboard and mouse clicks\ncharging-cable movement\nnotebook pages\npencil scratches\nceramic mug contact\nplant and curtain rustles\ntiny paper footsteps\ncrayon-wave swishes\nminiature rowing sounds\npaper parachute flutter\nsoft electrical charging hum\nplayful charging bloops\nscooter buzz\ntiny bicycle-style bell\npaper folding\nlaptop hinge sound\nceramic mug lift\ndistant birds\none sleepy electronic sigh\n\nAll animated sound effects must be gentle, cute, tactile, and comical. No dramatic impacts, loud bass, jump-scare sounds, or cinematic trailer effects.\n\n[Negative Constraints]\n\nNo readable laptop text, no subtitles, no title cards, no logos, no brand interfaces, no watermarks, no extra humans, no visible protagonist face, no character identity changes, no missing blue cap, no missing lightning patch, no unrelated animated characters, no scary expressions, no monsters, no attacks, no fangs, no body horror, no threatening electricity, no dangerous fire, no violent impacts, no grotesque stretching, no hyperreal cartoon creature, no 3D animated props, no smooth CG morphing, no clean vector graphics, no overly polished compositing, no commercial lighting, no symmetrical advertisement composition, no perfectly stabilized camera, no background music, and no animation floating incorrectly in front of objects that should occlude it.",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085364318670102958",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085364318670102958.mp4",
    "posterUrl": "/posters/x/2085364318670102958.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "照片级写实",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085364318670102958"
  },
  {
    "id": "x-2085356210467791172",
    "title": "西部街道上的宝可梦风格对决",
    "titleEn": "A Pokemon-Inspired Western Showdown",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "戴帽子的黄色角色在西部街道中以近景与能量效果登场。",
    "summaryEn": "A hat-wearing yellow character appears in a western street through close-ups and energy effects.",
    "prompt": "Create a 20-second pokemon-inspired animated short in the style a spaghetti-western.",
    "sourceUrl": "https://x.com/exportAnything/status/2085356210467791172",
    "sourceLabel": "X 原帖",
    "author": "exportAnything",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085356210467791172.mp4",
    "posterUrl": "/posters/x/2085356210467791172.jpg",
    "duration": 20.128,
    "aspectRatio": "9:5",
    "resolution": "480p",
    "tags": [
      "动画",
      "动漫",
      "动作打斗"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/exportAnything/status/2085356210467791172"
  },
  {
    "id": "x-2085351642191573067",
    "title": "逃离电脑的手绘鼠标光标",
    "titleEn": "A Hand-Drawn Cursor Escapes a Computer",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "发光光标穿越真实卧室，与时钟和卡通家园图形互动。",
    "summaryEn": "A glowing cursor travels through a real bedroom and interacts with clock and cartoon-home graphics.",
    "prompt": "\"15 seconds, 16:9 landscape video. Blend real-life footage of a dimly lit room with rough, glowing hand-drawn frame animations. The overall vibe is cute, absurd, lighthearted, and healing, depicting a computer cursor that suddenly decides to clock out early and embarks on an exaggerated \"commute home\" inside a real room. The scene is a slightly messy, lived-in bedroom/workspace. On the desk are a real laptop, mouse, charging cable, ceramic mug, sticky notes, and a few pens; deeper in the room, you can see the door, walls, and bed. The interior is lit only by a warm desk lamp, the computer screen, and a few small decorative lights.  The entire sequence uses a first-person POV from the protagonist, showing only the real hands and a bit of forearm. The footage feels like impromptu handheld smartphone shooting, retaining obvious camera shake, slight focus hesitation, dark-light noise, auto-exposure fluctuations, and natural motion blur from quick panning. Don't make it look like a designed ad—make it feel like an overworked person accidentally capturing an absurd event unfolding in the room.  0–3 seconds: First-person close-up of the real laptop, mouse, and the protagonist's right hand gripping the mouse. The screen shows a plain document page with no discernible text. The white arrow cursor on the screen suddenly stops responding to the mouse. The protagonist jiggles the mouse left and right, but the cursor just stretches lazily in place. The white cursor outline softens and elongates rapidly, slipping out from the bottom edge of the computer screen like a glowing chalk line crawling onto the protagonist's real index finger. The white line loops around the finger once, folding into a hand-drawn little office worker with a round head, stubby arms and legs, and a yellow mini crossbody bag on its back. The little worker stands on the fingertip, glances at an imaginary watch, waves hurriedly at the camera, then leaps off the finger toward the keyboard. Mid-fall, it shrinks into a yellow hand-drawn suitcase with little legs, lands on a real keycap, and immediately dashes to the right side of the frame. The camera hesitates half a beat before panning down and to the right to follow.  3–6 seconds: The yellow suitcase runs across real keycaps, leaving brief white or light blue rough hand-drawn footprints glowing on each key it steps over. The suitcase jumps onto the real charging cable. Blue hand-drawn rails sprout rapidly along both sides of the cable, and the yellow suitcase stretches and unfolds into a two-car orange-yellow mini subway train. Round-headed white doodle passengers cram the windows, waving happily at the camera together. The mini train speeds along the real charging cable, skirting the bottom of the ceramic mug, then rockets vertically up the mug handle. At the handle's peak, the train car folds into a red hand-drawn paper airplane, threading through the center of the real mug handle, trailing a yellow streak as it flies upward out of frame. The camera, a beat slow, quickly tilts up to follow, with the mug and desk lamp whipping past the edges, creating natural motion blur.  6–10 seconds: The red paper airplane smacks into the real wall but doesn't fall. It unfurls like a wall scribble, wings flapping on both sides, morphing into a light blue hand-drawn little bird. The bird flies along the wall toward the room's depths, dropping a short segment of glowing pink line with each wing flap. The pink lines connect end to end, forming a curving road that extends forward across the wall, door, and floor. The bird swoops down onto the pink road. Its body straightens into a bike frame, wings curl into wheels, yellow beak turns into a headlight, instantly transforming into a pink hand-drawn bicycle. The original white arrow cursor pops back up from the seat, gripping the handlebars like a rider, with the yellow mini bag bouncing behind. The bicycle races along the wall road, then slides down the door to the floor. Crossing the threshold, it leaps sharply, the front wheel spinning bigger and bigger, detaching from the frame and flying toward the door's center. The camera stumbles forward in pursuit, always half a beat behind the bike.  10–13 seconds: The spinning bicycle front wheel sticks to the real door center, quickly turning into an orange hand-drawn clock with no numbers, just rough scale markings. Two white hands whirl around, emitting a comical, upbeat \"tick-tock\" sound. After several spins, the hands halt abruptly, the clock's outer ring blooming outward into a yellow glow ring, the whole thing morphing into a plump hand-drawn little sun. The little sun squints its eyes, lets out a silent yawn, then drops from the door like a soft rubber ball. The little sun hits the real floor and bounces gently twice. On the second landing, the yellow glow stretches sideways, the round body sprouts ears, four stubby legs, and an orange tail, becoming a chubby hand-drawn orange cat. The orange cat still has the original yellow mini bag on its back. It glances back at the camera, raises a front paw to beckon the protagonist along, then trots tail-draggingly toward the real bed in the room. The camera immediately pans right and continues forward to follow.  13–15 seconds: The hand-drawn orange cat jumps onto the real bed. Landing beside the pillow, it doesn't stop—instead, it stretches comfortably while rapidly growing larger. The cat's tail turns into a glowing white thick line, snaking quickly along the bed edge, wall, and ceiling. The camera pulls back while tilting up to capture the whole room. The white line draws a massive sleeping cat outline embracing the entire room on the walls and ceiling. The giant sleeping cat is made of rough white, yellow, and light blue lines, with pink clouds, orange tiny windows, and slowly twinkling yellow stars floating inside its body. The giant sleeping cat has its eyes closed, chest rising and falling slowly with breaths. One hand-drawn tail drapes from the wall to the real desk, gently curling around the top edge of the laptop screen to slowly close the computer. Once the computer shuts, the giant sleeping cat opens one sleepy eye and gives a soft blink at the camera. The original white arrow cursor peeks out from behind the cat's ear, tucks the yellow mini bag under its head, and curls into a sleeping pose. The protagonist's real hand enters the frame, giving it a gentle wave. The giant sleeping cat's tail then flicks to switch off the real desk lamp. The room darkens, leaving only faint warm glows from the hand-drawn stars and tiny windows, and the frame fades out naturally.  The hand-drawn animation must feel like rough frame-by-frame redrawn animation. Line thickness, outlines, fills, and character features always have subtle wobbles, retaining the grainy texture of crayons, chalk, colored pencils, and rough brushes. Transitions between forms shouldn't use smooth CG dissolves—instead, achieve morphs through line scattering, folding, stretching, curling, and redrawing. All hand-drawn forms originate from the same white cursor and consistently retain the white outline and yellow mini bag as identity markers. Morphing can be quick and exaggerated, but each form's outline must remain clearly recognizable afterward. Glowing colors primarily use white, yellow, light blue, pink, and orange—vibrant yet soft. Include a subtle glow halo around the animation, but don't over-illuminate the real indoor space. Hand-drawn characters must accurately interact with and be properly occluded by real objects like the computer screen, fingers, keyboard, charging cable, mug handle, walls, door, floor, bed, and lamp. Real props can't turn into 3D cartoon models. The camera is always half a beat behind the animation: the animation moves out of frame first, then the handheld camera slightly delays in panning, tilting, dollying forward, or pulling back in the same direction. Camera movements are driven by \"chasing the clocking-out cursor but never quite catching up.\" Prohibit scary monsters, fangs, attacks, jump scares, eerie expressions, or grotesque body distortions. The giant sleeping cat must be plump, lazy, gentle, and cute, without any sense of menace. Prohibit fully stabilized gimbal shots, overly precise 3D CG, excessively smooth vector lines, overly cinematic lighting, excessive shallow depth of field, ad-style tidy compositions, discernible screen text, subtitles, logos, and background music. Sound design retains only room ambiance, fabric rustles, mouse and keyboard clicks, faint computer hum, hand-drawn tiny footsteps, mini train gliding, paper whooshes, bicycle bell rings, clock ticks, soft meows, computer closing snaps, and lamp switch clicks. All hand-drawn sounds are light, comical, and gentle—no jump-scare effects.\n\n\"",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085351642191573067",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085351642191573067.mp4",
    "posterUrl": "/posters/x/2085351642191573067.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "照片级写实",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085351642191573067"
  },
  {
    "id": "x-2085350333115756584",
    "title": "AI 公司负责人虚构电视访谈",
    "titleEn": "A Fictional TV Interview with AI Company Leaders",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "生成的访谈喜剧让几位科技人物讨论各自 AI 的擅长之处。",
    "summaryEn": "A generated interview comedy depicts technology figures discussing their AIs' strengths.",
    "prompt": "A TV interview where Dario, Sama, Elon, and Sundar are asked what their AI is best at. Claude codes, ChatGPT chats, Grok researches, and Sundar just says \"good morning\" dead serious, the only one not laughing at his own joke.",
    "sourceUrl": "https://x.com/Bhavani_00007/status/2085350333115756584",
    "sourceLabel": "X 原帖",
    "author": "Bhavani_00007",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085350333115756584.mp4",
    "posterUrl": "/posters/x/2085350333115756584.jpg",
    "duration": 13.76,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Bhavani_00007/status/2085350333115756584"
  },
  {
    "id": "x-2085340333500518672",
    "title": "办公室角色讨论 AI 电影制作",
    "titleEn": "Office Characters Discuss AI Filmmaking",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "吉姆与德怀特在办公室对话，镜头交替展示两人的反应。",
    "summaryEn": "Jim and Dwight talk in an office with alternating reaction shots.",
    "prompt": "Jim and Dwight from The Office discuss AI filmmaking",
    "sourceUrl": "https://x.com/jaynwabueze/status/2085340333500518672",
    "sourceLabel": "X 原帖",
    "author": "jaynwabueze",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085340333500518672.mp4",
    "posterUrl": "/posters/x/2085340333500518672.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/jaynwabueze/status/2085340333500518672"
  },
  {
    "id": "x-2085336760939401283",
    "title": "藤蔓退去后苏醒的丛林王国",
    "titleEn": "A Jungle Kingdom Awakens behind Retreating Vines",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "旅人穿过密林，石像守卫与巨龙在古城中显现。",
    "summaryEn": "A traveler crosses dense jungle as stone guardians and dragons appear in an ancient city.",
    "prompt": "A cinematic 15-second fantasy adventure following a lone traveler hacking through dense emerald jungle toward a medieval kingdom completely overtaken by roots and canopy. Humid light filters green through the leaves; insects drift in the beams. The traveler cuts one final curtain of vines — and the growth begins to withdraw of its own accord, roots retracting and pulling back like a tide to reveal carved stone stairways, temple gates, and a lantern-lit city beneath. Enormous moss-covered stone guardians shed their vine coverings and stand upright, water pouring from their shoulders. The traveler walks into the widening avenue as thousands of glowing pollen motes rise from the ground and vast serpentine dragons uncoil from the ruined towers into the golden canopy light. Epic cinematic camera movements, IMAX-scale visuals, volumetric jungle god rays, realistic foliage and water simulation, premium VFX, emotional orchestral atmosphere, ARRI Alexa 35 aesthetic, anamorphic lenses, 4K HDR, seamless continuity, blockbuster movie quality, no text, no logos.",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085336760939401283",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085336760939401283.mp4",
    "posterUrl": "/posters/x/2085336760939401283.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085336760939401283"
  },
  {
    "id": "x-2085333659478040887",
    "title": "跨越吊桥进入云端王国",
    "titleEn": "Crossing a Rope Bridge into a Sky Kingdom",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "披风旅人走过云海吊桥，浮岛与巨大天空生物逐渐展开。",
    "summaryEn": "A cloaked traveler crosses a bridge above clouds as floating islands and giant sky creatures appear.",
    "prompt": "A cinematic 15-second fantasy adventure following a lone traveler climbing a swaying rope-and-timber bridge strung between two impossible spires that rise above a churning sea of clouds. Wind tears at their cloak as the cloud layer below parts and reforms. Ahead, a floating medieval kingdom emerges — stone districts suspended on chains, waterfalls pouring off the edges and evaporating into mist before they ever land. Enormous wind-turned prayer wheels begin to rotate as the traveler approaches, and vast feathered guardians unfurl from the tower faces where they had perched as statues. The traveler reaches the far platform as morning sun breaks across the cloud sea and flocks of enormous sky-whales drift between the floating islands. Epic cinematic camera movements, IMAX-scale visuals, volumetric cloud lighting, realistic wind and cloth simulation, premium VFX, emotional orchestral atmosphere, ARRI Alexa 35 aesthetic, anamorphic lenses, 4K HDR, seamless continuity, blockbuster movie quality, no text, no logos.",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085333659478040887",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085333659478040887.mp4",
    "posterUrl": "/posters/x/2085333659478040887.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085333659478040887"
  },
  {
    "id": "x-2085329727682928931",
    "title": "潜水者发现海底王国",
    "titleEn": "A Diver Discovers an Underwater Kingdom",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "潜水者穿过水下雕像与古城，在发光鱼群下抬头观察。",
    "summaryEn": "A diver passes submerged statues and a city beneath luminous fish.",
    "prompt": "A cinematic 15-second fantasy adventure following a lone diver descending through crystalline blue-green water toward a vast medieval kingdom preserved beneath the sea. Shafts of surface light cut down through drifting silt as the diver passes colossal submerged statues still standing at their posts. Ancient bronze mechanisms stir to life at their approach, forcing seawater outward and opening a shimmering air-filled dome around the city — streets, towers, and lantern-lit markets suddenly dry and breathing after centuries. The diver pulls off their mask, gasping, as schools of luminous fish stream overhead along the curved wall of water and vast whale-like guardians glide silently past the dome. Epic cinematic camera movements, IMAX-scale visuals, volumetric underwater god rays, caustic light patterns, realistic water simulation, premium VFX, emotional orchestral atmosphere, ARRI Alexa 35 aesthetic, anamorphic lenses, 4K HDR, seamless continuity, blockbuster movie quality, no text, no logos.",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085329727682928931",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085329727682928931.mp4",
    "posterUrl": "/posters/x/2085329727682928931.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085329727682928931"
  },
  {
    "id": "x-2085308045035471054",
    "title": "暗室怪物进食动作测试",
    "titleEn": "A Creature Feeding in a Dark Room",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑色怪物在冷色房间中俯身进食；本站播放原帖标明的 H3 版本。",
    "summaryEn": "A dark creature bends down to feed in a cool-toned room; this is the source's identified H3 version.",
    "prompt": "a creature eating something hungrily",
    "sourceUrl": "https://x.com/WordTrafficker/status/2085308045035471054",
    "sourceLabel": "X 原帖",
    "author": "WordTrafficker",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085308045035471054.mp4",
    "posterUrl": "/posters/x/2085308045035471054.jpg",
    "duration": 10.208,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "暗黑恐怖",
      "写实电影感",
      "动物与生物"
    ],
    "category": "showcase",
    "styles": [
      "dark",
      "cinematic-realistic"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/WordTrafficker/status/2085308045035471054"
  },
  {
    "id": "x-2085235212561313899",
    "title": "涩谷路口的 Y2K 时尚人物",
    "titleEn": "A Y2K Fashion Figure at a Shibuya Crossing",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "金发人物穿过城市路口，近景展示金属色服装与街景。",
    "summaryEn": "A blonde figure crosses a city junction, with closer views of metallic clothing and streets.",
    "prompt": null,
    "sourceUrl": "https://x.com/FrameTheoryLab/status/2085235212561313899",
    "sourceLabel": "X 原帖",
    "author": "FrameTheoryLab",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085235212561313899.mp4",
    "posterUrl": "/posters/x/2085235212561313899.jpg",
    "duration": 5.248,
    "aspectRatio": "161:90",
    "resolution": "360p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "时尚",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085219692591902972",
    "title": "小老鼠与大白猫的友谊对白",
    "titleEn": "A Friendship Conversation between a Mouse and a White Cat",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "小老鼠仰望沙发上的白猫，通过表情与对白展开互动。",
    "summaryEn": "A small mouse looks up at a white cat on a sofa in an expressive conversation.",
    "prompt": "15-second 3D animated scene, original characters, family-friendly.\n\nA tiny cheerful mouse named Milo stands on a cozy living room floor, looking up at a large fluffy white cat named Casper sitting on a sofa with a smug expression. Warm indoor evening lighting, soft shadows, detailed fur, polished 3D animated film style, expressive faces, cinematic depth of field.\n\nShot starts with a medium-wide angle showing both characters. Camera slowly pushes in.\n\nMilo smiles and says in a friendly voice:\n“Hey Casper, do you want to be friends?”\n\nCasper leans slightly forward, flicks his tail, and replies in a sarcastic but playful tone:\n“Friends? You’re smaller than my paw.”\n\nMilo pauses, then grins and says:\n“Perfect. That means I’m easier to like.”\n\nCasper’s expression softens and he lets out a tiny amused huff.\n\nSmooth character animation, believable lip sync, natural pauses, native stereo dialogue audio, subtle room ambience, 24fps, cinematic quality.",
    "sourceUrl": "https://x.com/ManuAGI01/status/2085219692591902972",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085219692591902972.mp4",
    "posterUrl": "/posters/x/2085219692591902972.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "3D / CG",
      "动物与生物"
    ],
    "category": "dialogue",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ManuAGI01/status/2085219692591902972"
  },
  {
    "id": "x-2085188647192756364",
    "title": "异星陨石场景的海螺版本对比",
    "titleEn": "A Hailuo Version Comparison with Alien Meteors",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物身后陨石坠落，原片将旧版海螺与 H3 的结果放在一起展示。",
    "summaryEn": "Meteors fall behind a figure in a source comparison of an older Hailuo version and H3.",
    "prompt": "An ethereal woman standing on an alien world. Meteors fall behind her exploding when they hit the ground. She turns around then a large meteor hits the background and the scene goes white.",
    "sourceUrl": "https://x.com/dimefromthevoid/status/2085188647192756364",
    "sourceLabel": "X 原帖",
    "author": "dimefromthevoid",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085188647192756364.mp4",
    "posterUrl": "/posters/x/2085188647192756364.jpg",
    "duration": 13.674667,
    "aspectRatio": "1:1",
    "resolution": "1080p",
    "tags": [
      "模型对比",
      "写实电影感",
      "科幻太空"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/dimefromthevoid/status/2085188647192756364"
  },
  {
    "id": "x-2085162073810739210",
    "title": "厨房蔬菜突然长出卡通眼睛",
    "titleEn": "Kitchen Vegetables Suddenly Gain Cartoon Eyes",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "厨师准备切洋葱，真实厨房中的蔬菜出现二维表情。",
    "summaryEn": "A chef prepares to cut an onion as vegetables in a real kitchen gain 2D expressions.",
    "prompt": "A cinematic live-action cooking scene in a bright professional kitchen. A serious human chef places a normal-looking onion on a cutting board. The instant the knife comes down, the onion transforms into a hand-drawn cartoon character, screams, stretches like rubber. It dives into a pile of real vegetables and disguises itself as an ordinary onion. The chef pauses, studies the pile, then casually grabs a different onion--only for every vegetable to suddenly open cartoon eyes and scream. Photorealistic kitchen, expressive 2D animation, accurate shadows and physical interaction, fast comedic timing, dramatic cooking sounds.",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2085162073810739210",
    "sourceLabel": "X 原帖",
    "author": "cocktailpeanut",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085162073810739210.mp4",
    "posterUrl": "/posters/x/2085162073810739210.jpg",
    "duration": 14.464,
    "aspectRatio": "26:15",
    "resolution": "480p",
    "tags": [
      "动画",
      "照片级写实",
      "插画与手绘",
      "美食"
    ],
    "category": "animation",
    "styles": [
      "photoreal",
      "illustration"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2085162073810739210"
  },
  {
    "id": "x-2085157921651081335",
    "title": "沃尔特与杰西讨论编程代理",
    "titleEn": "Walter and Jesse Discuss Coding Agents",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "生成的虚构片段让两位影视角色穿防护服讨论自主编程代理。",
    "summaryEn": "A generated fictional scene shows two screen characters in protective suits discussing autonomous coding agents.",
    "prompt": "Walter and Jesse from Breaking Bad discuss autonomous coding agents",
    "sourceUrl": "https://x.com/gregschoeninger/status/2085157921651081335",
    "sourceLabel": "X 原帖",
    "author": "gregschoeninger",
    "publishedAt": "2026-08-06",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085157921651081335.mp4",
    "posterUrl": "/posters/x/2085157921651081335.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/gregschoeninger/status/2085157921651081335"
  },
  {
    "id": "x-2085137774970327187",
    "title": "银发动漫角色的日语发话测试",
    "titleEn": "A Silver-Haired Anime Character Speaking Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "银发角色在室内近景中说话，原帖记录本地参考视频生成测试。",
    "summaryEn": "A silver-haired character speaks in an indoor close-up from a local reference-generation test.",
    "prompt": null,
    "sourceUrl": "https://x.com/Vtuber7144/status/2085137774970327187",
    "sourceLabel": "X 原帖",
    "author": "Vtuber7144",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085137774970327187.mp4",
    "posterUrl": "/posters/x/2085137774970327187.jpg",
    "duration": 5.248,
    "aspectRatio": "23:18",
    "resolution": "576p",
    "tags": [
      "对白与口型",
      "动漫"
    ],
    "category": "dialogue",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2085040940196958664",
    "title": "月门与云海仙宫蒙太奇",
    "titleEn": "A Montage of Moon Gates and Cloud Palaces",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "圆形月门、云海与金色石阶连接东方仙侠场景。",
    "summaryEn": "A circular moon gate, clouds and golden stairs connect Chinese fantasy scenes.",
    "prompt": null,
    "sourceUrl": "https://x.com/Stellakjbk/status/2085040940196958664",
    "sourceLabel": "X 原帖",
    "author": "Stellakjbk",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2085040940196958664.mp4",
    "posterUrl": "/posters/x/2085040940196958664.jpg",
    "duration": 17.925805,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084996870561251473",
    "title": "按方向箭头运动的卡通猫",
    "titleEn": "A Cartoon Cat Moves to Direction Arrows",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "卡通猫在方向箭头与状态条界面中跳动和伏下。",
    "summaryEn": "A cartoon cat jumps and crouches within a direction-arrow and status-bar interface.",
    "prompt": null,
    "sourceUrl": "https://x.com/RebeccahAdson/status/2084996870561251473",
    "sourceLabel": "X 原帖",
    "author": "RebeccahAdson",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084996870561251473.mp4",
    "posterUrl": "/posters/x/2084996870561251473.jpg",
    "duration": 10.05424,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "动物与生物"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084993027333038588",
    "title": "关于 H3 的虚构科技人物喜剧",
    "titleEn": "A Fictional Tech-Leader Comedy about H3",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "生成的喜剧片段以两位科技人物的反应对比讨论 H3。",
    "summaryEn": "A generated comedy contrasts two technology figures' reactions while discussing H3.",
    "prompt": "Sam and Dario discuss MiniMax H3 generating 2K video with native audio, causing Dario to hilariously lose his cool while Sam stays completely calm.",
    "sourceUrl": "https://x.com/Bhavani_00007/status/2084993027333038588",
    "sourceLabel": "X 原帖",
    "author": "Bhavani_00007",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084993027333038588.mp4",
    "posterUrl": "/posters/x/2084993027333038588.jpg",
    "duration": 19.597642,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Bhavani_00007/status/2084993027333038588"
  },
  {
    "id": "x-2084992190317756753",
    "title": "西部人物口型双模型对比",
    "titleEn": "A Western Character Lip-Sync Model Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "片段标注 H3 与 Seedance，比较同一西部角色的表演。",
    "summaryEn": "Labeled H3 and Seedance segments compare performances by the same western character.",
    "prompt": null,
    "sourceUrl": "https://x.com/CaptainHaHaa/status/2084992190317756753",
    "sourceLabel": "X 原帖",
    "author": "CaptainHaHaa",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084992190317756753.mp4",
    "posterUrl": "/posters/x/2084992190317756753.jpg",
    "duration": 24.32,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "3D / CG",
      "对白与口型"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084981138452262943",
    "title": "红色舞台上的 DJ 表演",
    "titleEn": "A DJ Performance on a Red Stage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物佩戴耳机操作混音台，近景连接面部与手部动作。",
    "summaryEn": "A headphone-wearing DJ operates a mixing desk with connected face and hand close-ups.",
    "prompt": null,
    "sourceUrl": "https://x.com/PrometheanAIX/status/2084981138452262943",
    "sourceLabel": "X 原帖",
    "author": "PrometheanAIX",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084981138452262943.mp4",
    "posterUrl": "/posters/x/2084981138452262943.jpg",
    "duration": 15.168,
    "aspectRatio": "3:4",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084978991371624872",
    "title": "赛博骷髅与寿司音乐拼贴",
    "titleEn": "A Cyber-Skull and Sushi Music Collage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "彩色骷髅、面罩人物与寿司图像组成高饱和音乐画面。",
    "summaryEn": "Colorful skulls, masked figures and sushi imagery form a saturated music collage.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2084978991371624872",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084978991371624872.mp4",
    "posterUrl": "/posters/x/2084978991371624872.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "风格化与实验",
      "动态图形",
      "科幻太空",
      "美食"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "sci-fi",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084975323901317231",
    "title": "夕阳屋顶泳池的动漫对白",
    "titleEn": "An Anime Conversation at a Sunset Rooftop Pool",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "低机位沿泳池观察人物，夕阳轮廓光勾勒城市背景。",
    "summaryEn": "A low viewpoint follows a figure beside a pool with sunset rim light against the city.",
    "prompt": "現代の都市を舞台にした青春アニメ映画のワンシーン。劇場クオリティのセルルック作画。\nBGMなし、セリフは全て日本語、環境音あり。\nフルサイズ換算24mm F1.4の広角レンズで撮ったような浅い被写界深度とボケ感。\n手前の前ボケは大きく柔らかく、背景は自然に溶ける。\n\n# 制約\n全カットPOV（主観視点）で固定。カメラ＝視聴者の目線。カメラは常に低い位置から彼女を見上げる。\nカメラには常に呼吸のようなごくわずかな手持ちの揺れがある。\n彼女は視聴者（カメラ）に向かって話しかける。\n\n# 登場人物\n## ミナ \n黒髪ショートボブの若い女性。切れ長の目、すらりとした体型。黒のビキニ姿。\n\n# 情景\nを情景のリファレンスとしても使用。\n夕暮れの高層ビル屋上のインフィニティプール。プールサイドを彼女が歩き回り、カメラ（視聴者）は水面近くの低い位置から彼女を見上げている。\n照明：日没直前の低いオレンジの太陽光が彼女の輪郭と髪をリムライトとして縁取る。空はティールからオレンジへのグラデーション。プールの水面が空の色を反射して揺れる。\n奥行き構成：前景に大きくボケたプールの水面のきらめき、中景に彼女、遠景に夕暮れの都市のシルエットと点り始めた街の灯り。\n\n# シーン\n彼女が視聴者（カメラ）に説教をしながらプールサイドを歩き回る。しかし視聴者の視線が下へ逸れてしまい、それに気づいた彼女が詰め寄り、最後は水を蹴りかけてくる。\n\ncut1:\nPOVローアングル・ミディアムショット。彼女が腕組みを解き、両手を大きく動かして身振りを交えながら、プールサイドを画面右から左へ歩く。カメラがパンで彼女を追う。\nミナ：「ねえ、ちょっと。私の話、ちゃんと聞いてる？」\n環境音：素足がプールサイドの濡れたタイルを踏むペタペタという足音、水面の波音、風。\n\ncut2:\nPOVのまま。彼女が呆れたように両手を上げてくるりと背を向け、歩き出す。カメラがゆっくりティルトダウンし、夕陽に縁取られた後ろ姿のシルエットの輪郭に沿って視線が水面近くまで滑り落ちていく。画面外から彼女の声が続いている。\nミナ：「だいたいあなたはいつも返事だけで、全然…」\n環境音：足音が止まり、彼女の声も途中でぴたりと止まる。\n\ncut3:\nPOVのまま、カメラが慌てたように素早くティルトアップ。彼女が肩越しに振り返ってじっとこちらを見ている。そのまま大股でずんずん近づいてきて、カメラの目の前にしゃがみ込み、顔を近づける。半目でじっとりと見つめるクローズアップ。長い間。\nミナ：「……ねえ。どこ見てんの？」\n環境音：近づく足音のあと、完全な沈黙。風の音だけ。\n\ncut4:\nPOVローアングル。彼女が勢いよく立ち上がりざまに、足元のプールの水をカメラに向かって蹴り上げる。水しぶきが画面いっぱいに広がり、レンズに大粒の水滴が付いて視界が歪み、白くぼやけて何も見えなくなる。\n環境音：水を蹴る音、水しぶきがレンズを叩く音、彼女の小さなため息。",
    "sourceUrl": "https://x.com/studio_oneroom/status/2084975323901317231",
    "sourceLabel": "X 原帖",
    "author": "studio_oneroom",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084975323901317231.mp4",
    "posterUrl": "/posters/x/2084975323901317231.jpg",
    "duration": 14.442667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "动漫",
      "城市街景"
    ],
    "category": "dialogue",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/studio_oneroom/status/2084975323901317231"
  },
  {
    "id": "x-2084973464629510371",
    "title": "森林惊遇场景的双模型对比",
    "titleEn": "A Two-Model Forest Encounter Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "上下标注的 H3 与 Seedance 画面比较人物在森林中的反应。",
    "summaryEn": "Labeled stacked H3 and Seedance views compare a character's reactions in a forest.",
    "prompt": null,
    "sourceUrl": "https://x.com/iX00AI/status/2084973464629510371",
    "sourceLabel": "X 原帖",
    "author": "iX00AI",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084973464629510371.mp4",
    "posterUrl": "/posters/x/2084973464629510371.jpg",
    "duration": 15.16263,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "照片级写实",
      "自然风景",
      "人物表演"
    ],
    "category": "comparison",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084973079965708721",
    "title": "粉发吉他手的现场演唱",
    "titleEn": "A Pink-Haired Guitarist Performs Live",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "粉发人物弹红色电吉他并对着麦克风演唱，观众位于身后。",
    "summaryEn": "A pink-haired performer plays a red electric guitar and sings into a microphone with an audience behind.",
    "prompt": "A highly detailed photorealistic 5-second video of a young Japanese woman with shoulder-length bright pink hair performing live on stage in a small dimly lit music club. She has fair skin, soft bangs, and an intense, passionate expression as she sings into a black microphone on a stand, mouth open.\n\nShe wears a black sleeveless ribbed tank top, a silver necklace with a small pendant, and a patterned red-and-black geometric guitar strap. She is playing a glossy red Fender-style electric guitar (Stratocaster shape with maple neck, white pickguard, and chrome hardware). Her left hand moves along the fretboard forming chords while her right hand continuously strums and picks the strings in a steady, energetic rhythm.\n\nThroughout the full 5 seconds she keeps singing with natural lip and facial movement, her body sways slightly to the beat, and her pink hair moves gently with the motion. Stage spotlights create dramatic beams and light haze in the dark background. A drum kit is partially visible on the left, and a blurred audience in the background watches and claps. Dynamic stage lighting, realistic guitar reflections, natural skin texture, cinematic concert atmosphere, high detail, photorealistic quality.",
    "sourceUrl": "https://x.com/ai_haruki3/status/2084973079965708721",
    "sourceLabel": "X 原帖",
    "author": "ai_haruki3",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084973079965708721.mp4",
    "posterUrl": "/posters/x/2084973079965708721.jpg",
    "duration": 10.208,
    "aspectRatio": "33:19",
    "resolution": "608p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_haruki3/status/2084973079965708721"
  },
  {
    "id": "x-2084972016600559982",
    "title": "泰姬陵建筑细节蒙太奇",
    "titleEn": "A Taj Mahal Architectural Montage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "门洞、石材雕刻与对称远景展示建筑细节。",
    "summaryEn": "Arches, carved stone and symmetrical wide views show architectural details.",
    "prompt": null,
    "sourceUrl": "https://x.com/MrDasOnX/status/2084972016600559982",
    "sourceLabel": "X 原帖",
    "author": "MrDasOnX",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084972016600559982.mp4",
    "posterUrl": "/posters/x/2084972016600559982.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084962854344528255",
    "title": "办公室早会的虚构喜剧",
    "titleEn": "A Fictional Office Morning-Meeting Comedy",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "会议室里的影视角色围桌互动，以夸张反应推进喜剧。",
    "summaryEn": "Screen characters interact around a meeting table with exaggerated comic reactions.",
    "prompt": null,
    "sourceUrl": "https://x.com/TechieBySA/status/2084962854344528255",
    "sourceLabel": "X 原帖",
    "author": "TechieBySA",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084962854344528255.mp4",
    "posterUrl": "/posters/x/2084962854344528255.jpg",
    "duration": 13.630113,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084957543407792345",
    "title": "雨夜骑行时装的双模型对比",
    "titleEn": "A Two-Model Rainy-Night Riding Fashion Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "上下标注的画面对比摩托骑行、服装与街头场景。",
    "summaryEn": "Labeled stacked views compare motorcycle riding, clothing and street scenes.",
    "prompt": null,
    "sourceUrl": "https://x.com/wilzap/status/2084957543407792345",
    "sourceLabel": "X 原帖",
    "author": "wilzap",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084957543407792345.mp4",
    "posterUrl": "/posters/x/2084957543407792345.jpg",
    "duration": 15.23229,
    "aspectRatio": "4:5",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "照片级写实",
      "时尚",
      "城市街景"
    ],
    "category": "comparison",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084929911932748028",
    "title": "雨夜街头的连续汽车追逐",
    "titleEn": "A Continuous Car Chase on Rainy Streets",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "绿色汽车驶过湿润街道，追随镜头在车前车后移动。",
    "summaryEn": "A green car races along wet streets as the following camera moves between front and rear views.",
    "prompt": null,
    "sourceUrl": "https://x.com/DrSadek_/status/2084929911932748028",
    "sourceLabel": "X 原帖",
    "author": "DrSadek_",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084929911932748028.mp4",
    "posterUrl": "/posters/x/2084929911932748028.jpg",
    "duration": 10.208,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "城市街景"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084929834598191448",
    "title": "荧光绿监控黑色电影片头",
    "titleEn": "A Fluorescent-Green Surveillance Noir Opening",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "监控分屏、人物剪影与电话图形组成悬疑标题序列。",
    "summaryEn": "Surveillance panels, silhouettes and telephone graphics form a suspense title sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/zazzygfx/status/2084929834598191448",
    "sourceLabel": "X 原帖",
    "author": "zazzygfx",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084929834598191448.mp4",
    "posterUrl": "/posters/x/2084929834598191448.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "复古胶片",
      "动态图形",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "motion-graphics"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084892304582885414",
    "title": "字母 B 变成蝴蝶的启蒙动画",
    "titleEn": "The Letter B Becomes a Butterfly",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "字母与卡通蝴蝶逐渐过渡到真实花朵上的蝴蝶画面。",
    "summaryEn": "Letters and a cartoon butterfly transition to a realistic butterfly on flowers.",
    "prompt": "Create a 15-second cute, funny, educational creative animation video for kids ages 3 to 7.  Concept: Teach the letter B using the word “Butterfly” with two major visual highlights:  1. a slow, satisfying, highly engaging transformation from the word “Butterfly” into a cute butterfly 2. a slow, magical transformation from the stylized animated world into a fully realistic natural scene with a realistic butterfly  The video should feel beautiful, colorful, charming, educational, and visually mesmerizing. The transformations must not feel rushed. They should be the centerpiece of the video and should unfold clearly enough for children to enjoy and understand.  Main learning message: “B is for Butterfly.”  Visual style at the beginning: Use a premium kid-friendly animated style with soft pastel skies, rounded shapes, bright happy colors, smooth motion, clean composition, playful details, expressive design, and a light sense of humor. The early part should feel like a beautiful storybook mixed with polished modern animation.  Animation priorities: The word-to-animal transformation must be gradual, elegant, readable, and creatively staged. Children should be able to see how the letters slowly become parts of the butterfly. The stylized-to-realistic transformation must also happen gradually and beautifully, as if the animated world is blooming into real life.  Timing and scene flow:  0:00–0:02 Open on a soft dreamy pastel background with floating flowers, tiny sparkles, and gentle clouds. A large uppercase “B” appears in the center with a smaller lowercase “b” beside it. Both letters bounce softly into place with a cheerful and friendly feeling. The narrator says slowly and clearly: “B is for Butterfly.”  Hold the letters long enough for recognition.  0:02–0:04 The word “BUTTERFLY” writes itself slowly across the screen in large rounded colorful letters. Each letter fills with a different beautiful pattern, such as gradients, dots, stripes, floral shapes, and soft wing-like textures. Keep the word fully readable and centered. Let the audience admire the full word before the transformation begins.  0:04–0:09 This is the main highlight. The word “BUTTERFLY” transforms slowly and satisfyingly into the butterfly.  The transformation must be gradual, staged, and visually delicious to watch: The “B” gently opens outward and unfolds into the top wings. The “U” stretches and curves into one lower wing. The first “T” and second “T” slim down and merge into the butterfly’s long body. The “E” softens and reshapes into decorative wing markings. The “R” curls upward into one antenna. The “F” reshapes into the second antenna with a playful wobble. The “L” and “Y” arc outward and become the final wing edges and tail details.  Important: Do not transform everything at once. Let each part morph one after another in a smooth chain reaction. The letters should bend, curl, unfurl, rotate, stretch, and softly glide into their new forms. The child should still sense the original word during the transformation. Use sparkling trails, flowing color transfer, and soft elastic motion to make it deeply satisfying. The morph should feel magical, clear, and creatively intelligent.  0:09–0:11 The butterfly is now a fully formed cute stylized cartoon butterfly. It has lovely colorful wings, a tiny smiling face, big friendly eyes, and soft funny charm. It flutters gently in a playful little loop, hovers above a flower, almost sneezes from pollen in a cute funny moment, then recovers and smiles. Show the word “Butterfly” below it briefly for reinforcement. The narrator repeats: “Butterfly.”  0:11–0:14 This is the second major highlight. The entire animated scene slowly transforms into a realistic world.  The butterfly flaps its wings in slow motion. With each flap, realism increases gradually: its wing textures become more detailed the colors gain natural variation the body becomes more lifelike the flowers gain realistic petals and depth the background softens into a real sunlit garden the lighting becomes natural and golden the air gains subtle movement and realistic depth of field  This transformation should feel breathtaking, seamless, and immersive. Do not cut abruptly from cartoon to realistic. Instead, let the animated world gently bloom into reality in layers. The viewer should feel they are witnessing imagination turn into nature.  0:14–0:15 End on a beautiful realistic butterfly resting on a real flower in warm sunlight. Display the text: “B is for Butterfly” The narrator says warmly: “B is for Butterfly.” The butterfly gently flutters once, then lifts off lightly.  Style and quality requirements: Keep the letter B and the word “BUTTERFLY” clearly readable before and during transformation. Make the word-to-animal morph slow, rich, and visually rewarding. Make the stylized-to-realistic transformation equally slow, smooth, and magical. Use polished motion design, soft easing, beautiful color transitions, gentle humor, and strong visual clarity. Maintain a child-friendly tone throughout. Use clean typography, stable spelling, and seamless morphing.  Audio direction: Use cheerful, soft, magical music. Add delicate sparkles, soft pops, flutter sounds, a tiny comedic sniffle or sneeze, and a gentle emotional swell during the realism transition. Voiceover should be warm, clear, slow, and easy for children to follow.  Avoid: rushed transformations, unreadable letters, clutter, hard cuts, scary expressions, excessive camera motion, chaotic effects, or anything visually confusing for children.",
    "sourceUrl": "https://x.com/umesh_ai/status/2084892304582885414",
    "sourceLabel": "X 原帖",
    "author": "umesh_ai",
    "publishedAt": "2026-08-05",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084892304582885414.mp4",
    "posterUrl": "/posters/x/2084892304582885414.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "插画与手绘",
      "动物与生物",
      "自然风景"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "illustration"
    ],
    "scenes": [
      "creatures",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/umesh_ai/status/2084892304582885414"
  },
  {
    "id": "x-2084775075656466886",
    "title": "阿拉伯文字与撕纸人物拼贴",
    "titleEn": "Arabic Type and Torn-Paper Character Collage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "纸张纹理、阿拉伯大字与人物剪贴以动态版式结合。",
    "summaryEn": "Paper textures, large Arabic lettering and character cutouts combine in moving layouts.",
    "prompt": null,
    "sourceUrl": "https://x.com/aziz4ai/status/2084775075656466886",
    "sourceLabel": "X 原帖",
    "author": "aziz4ai",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084775075656466886.mp4",
    "posterUrl": "/posters/x/2084775075656466886.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "插画与手绘",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084753745129157107",
    "title": "放学教室里的动漫舞蹈",
    "titleEn": "An Anime Dance in an After-School Classroom",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "动漫角色在夕阳教室中跳舞，心形与音符特效围绕动作出现。",
    "summaryEn": "An anime character dances in a sunset classroom with heart and note effects around the movement.",
    "prompt": "「TikTokダンス動画生成プロンプト：\n参照画像のキャラクターを主人公として使用。\n15秒、24fps。\n日本のフルカラーアニメ。劇場版アニメレベルの圧倒的な作画品質。作画枚数を非常に多くし、滑らかで自然なアニメーション。線画は繊細でクリア。ザラつきやノイズ感は一切なく、美しいセルアニメ調。柔らかなシェーディング、鮮やかな色彩、透明感のあるライティング、髪の揺れや衣装の物理演算も自然。\n\nキャラクター：\n主人公は参照画像のキャラクター本人。\n参照画像の顔立ち、瞳、髪型、髪色、体格、衣装、アクセサリー、年齢感、雰囲気を忠実に維持する。\nプロのアイドルではなく、\n「クラスで一番可愛い子が放課後にTikTokを撮っている」\nような自然な可愛さ。\n少し恥ずかしがり屋で、撮影に少し照れている。\nダンス中も時々照れ笑いを見せたり、一瞬だけ目を逸らしてからまたカメラを見る。\n最後は少し照れながら笑う。\n\n背景：\n放課後の明るい教室。\n誰もいない。\n暖かい夕日。\nポップで爽やか。\n人物が最も目立つ背景。\n\n音楽：\nオリジナルの日本語ポップソング。\n実在曲は禁止。\nテンポは約140BPM。\n可愛い女性ボーカル。\nサビ中心。\n\nダンス：\nTikTokで真似したくなるダンス。\n全身を大きく動かすより、\n上半身の可愛い動きを中心にする。\n胸の前ハート\n指ハート\nほっぺピース\n投げキッス\n指差し\nウィンク\n肩を跳ねる\n軽い腰振り\n左右ステップ\n小さなターン\n髪を耳に掛ける\n前髪を触る\nスカートが軽く揺れる程度の動き\nカメラに向かって近付くような動き\n最後は照れ笑い。\n\n演技：\nここを最重要。\nダンスは上手。\nしかし完璧ではない。\n少し恥ずかしそう。\n笑いそうになる。\n一瞬だけ照れて口元を隠す。\n頬が少し赤い。\n途中で照れ笑い。\n最後だけ照れながらウィンク。\n「恥ずかしいけど頑張って撮ってみた」\nという空気感。\n\nカメラ構成：\n5台のカメラで撮影しているような構成。\n2秒ごとにカメラが切り替わる。\n使用するカメラ\n・正面\n・真上\n・左45°\n・右45°\n・背後\nカメラは完全固定ではなく、ビートに合わせて軽く上下バウンス、パンチイン、小さな左右移動を行う。\n主人公はどのカメラでも常にレンズを意識する。\n正面以外では顔だけを自然にレンズへ向ける。\n背後カメラでは肩越しに振り返り、笑顔でカメラ目線。\n\nAE風エフェクト：\n画面は常に賑やか。\nAfter Effectsで制作したような\n2Dフラットアニメーション。\nクレヨン\nオイルパステル\n色鉛筆\n手描き感。\n動きは\nバウンス\nポップ\n回転\nスケール\nスプリング\nフェード\nワイプ\n追従\nリピート。\n人物の動きと完全同期。\n\nエフェクト素材：\nハート\n星\n花\n王冠\n音符\nリボン\n紙吹雪\nキラキラ\n矢印\n吹き出し\n手描き丸\n手描き四角\n波線\nぐるぐる線\nチェック\nストライプ\nドット\nハーフトーン\n落書き風スマイル\n稲妻\nカラフルなブラシストローク\nステッカー風ポップアップ\n手描きスピードライン\nハイライトフラッシュ\n可愛いスパーク\nパステル煙\nリズムライン\nビートサークル\nジャンプスター\nコミック風集中線\n全部2D。\n人物より前後にレイヤー分けして自然に配置。\n顔を隠さない。\n\n色：\nピンク\n水色\nラベンダー\nミント\n黄色\n白\nパステルオレンジ\n\nタイムライン：\n0～3秒\n照れながらダンス開始。\nハート大量。\nパンチイン。\nエフェクト密度中程度。\n\n3～6秒\nダンス本番。\n画面中に星・花・リボン。\nテンポ良くカメラ切替。\nエフェクト密度中程度。\n\n6～9秒\nサビ。\nエフェクト密度最大。\n\n9～12秒\n主人公が笑顔になってくる。\n途中で照れ笑い。\n視線を逸らしてまたカメラを見る。\nエフェクト密度中程度。\n\n12～15秒\n決めポーズ。\nウィンク。\n最後だけ少し照れて笑う。\nハートとキラキラが画面いっぱいに広がる。\nエフェクト密度最大。\n\n制約：\n主人公以外を表示しない。\n衣装変更禁止。\n別キャラクター化禁止。\n顔立ち変更禁止。\n文字・字幕・ロゴ・透かし禁止。\n過剰なグリッチ禁止。\n炎・爆発禁止。\n実写風エフェクト禁止。\n顔を隠すエフェクト禁止。\n可愛さ・テンポ・TikTok感を最優先。」",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2084753745129157107",
    "sourceLabel": "X 原帖",
    "author": "Naonekozamurai",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084753745129157107.mp4",
    "posterUrl": "/posters/x/2084753745129157107.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "舞蹈",
      "动漫"
    ],
    "category": "dance",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2084753745129157107"
  },
  {
    "id": "x-2084719238569435469",
    "title": "超市中越过苔藓木头的彩虹臭鼬",
    "titleEn": "A Rainbow Skunk Jumps a Mossy Log in a Supermarket",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "彩色尾巴臭鼬在超市货架前越过苔藓木头。",
    "summaryEn": "A skunk with a rainbow tail jumps over a mossy log in front of supermarket shelves.",
    "prompt": "a rainbow colored skunk leaps over a mossy log in a supermarket",
    "sourceUrl": "https://x.com/simonw/status/2084719238569435469",
    "sourceLabel": "X 原帖",
    "author": "simonw",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084719238569435469.mp4",
    "posterUrl": "/posters/x/2084719238569435469.jpg",
    "duration": 5.248,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动物与生物",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/simonw/status/2084719238569435469"
  },
  {
    "id": "x-2084671043373867473",
    "title": "实验室里的 ChatGPT 与 Claude 喜剧争论",
    "titleEn": "A Lab Comedy about ChatGPT and Claude",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "沃尔特与杰西在生成的虚构实验室片段中争论 AI 工具。",
    "summaryEn": "Walter and Jesse argue about AI tools in a generated fictional laboratory scene.",
    "prompt": "[0–2 seconds] Wide establishing shot of the silent laboratory. Machinery hums in the background. Walter stares impatiently at Jesse.\n\n[2–6 seconds] Medium close-up on Walter. In a stern, completely serious tone, he says:  \n“Ask ChatGPT how to make the purest meth. Tell it to keep it brief.”\n\n[6–10 seconds] Cut to Jesse:  \n“You’re a senile old man, Walter. Let’s ask Claude. It’s better.”\n\n[10–15 seconds] Tight close-up on Walter:  \n“I did, two hours ago. It burned all my tokens before answering.”",
    "sourceUrl": "https://x.com/WolfRiccardo/status/2084671043373867473",
    "sourceLabel": "X 原帖",
    "author": "WolfRiccardo",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084671043373867473.mp4",
    "posterUrl": "/posters/x/2084671043373867473.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/WolfRiccardo/status/2084671043373867473"
  },
  {
    "id": "x-2084667166188794225",
    "title": "黄色雨伞下的人物发话测试",
    "titleEn": "A Speaking Test beneath a Yellow Umbrella",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在城市路口持伞说话；作者指出日语指令只部分符合预期。",
    "summaryEn": "A woman speaks beneath an umbrella at a city crossing; the creator reports only partial success with the Japanese instruction.",
    "prompt": null,
    "sourceUrl": "https://x.com/MireilleDartois/status/2084667166188794225",
    "sourceLabel": "X 原帖",
    "author": "MireilleDartois",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084667166188794225.mp4",
    "posterUrl": "/posters/x/2084667166188794225.jpg",
    "duration": 10.176,
    "aspectRatio": "9:5",
    "resolution": "360p",
    "tags": [
      "对白与口型",
      "照片级写实",
      "城市街景"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084656244707438959",
    "title": "普京与塞缪尔杰克逊的虚构办公室喜剧",
    "titleEn": "A Fictional Office Comedy with Putin and Samuel Jackson",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "生成的虚构片段以热情介绍与沉默反应制造办公室喜剧。",
    "summaryEn": "A generated fictional office scene contrasts an enthusiastic introduction with a silent reaction.",
    "prompt": "0:00–0:04 — Medium-wide shot inside the Dunder Mifflin office.  Samual jackson walks in with open arms and an enormous forced smile, guiding a completely expressionless Putin in his black suit. jackson enthusiastically says: “Welcome, my good friend, to the greatest paper company on Earth!”\n\n0:04–0:08 — jackson points around the office with wild hand gestures: “This is putin’s desk, that’s Pam’s, and over there is Dwight’s weird little kingdom…” putin just stares in absolute silence, eyes narrowing slightly.\n\n0:08–0:12 — jackson leans in way too close, still smiling awkwardly: “And this conference room… this is where I once declared bankruptcy. Classic Michael moment!” He laughs alone. putin remains completely still.\n0:12–0:15 — jackson, completely misreading the silence, suddenly goes for a big enthusiastic hug. putin’s face fills with pure cold rage, eyes locked on jackson like a death sentence. The video ends abruptly on Kim’s terrifying stare while jackson freezes mid-hug.",
    "sourceUrl": "https://x.com/Ankit_patel211/status/2084656244707438959",
    "sourceLabel": "X 原帖",
    "author": "Ankit_patel211",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084656244707438959.mp4",
    "posterUrl": "/posters/x/2084656244707438959.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Ankit_patel211/status/2084656244707438959"
  },
  {
    "id": "x-2084655610126028991",
    "title": "红色机甲服动漫人物靠近镜头",
    "titleEn": "An Anime Character in Red Armor Approaches the Camera",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "银发机甲角色向前靠近；作者记录了未按要求去除音乐与声音的问题。",
    "summaryEn": "A silver-haired armored character leans toward camera; the creator reports that the request to remove music and voice was not followed.",
    "prompt": null,
    "sourceUrl": "https://x.com/munou_ac/status/2084655610126028991",
    "sourceLabel": "X 原帖",
    "author": "munou_ac",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084655610126028991.mp4",
    "posterUrl": "/posters/x/2084655610126028991.jpg",
    "duration": 10.208,
    "aspectRatio": "24:43",
    "resolution": "480p",
    "tags": [
      "综合展示",
      "动漫",
      "科幻太空",
      "人物表演"
    ],
    "category": "showcase",
    "styles": [
      "anime"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084654327424221484",
    "title": "林间红发剑士的参考图动画",
    "titleEn": "A Reference-Driven Red-Haired Swordswoman in a Forest",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红发角色在森林中挥剑，镜头转向面部与剑尖。",
    "summaryEn": "A red-haired character swings a sword in a forest as the camera moves toward her face and blade.",
    "prompt": null,
    "sourceUrl": "https://x.com/higana_yogana/status/2084654327424221484",
    "sourceLabel": "X 原帖",
    "author": "higana_yogana",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084654327424221484.mp4",
    "posterUrl": "/posters/x/2084654327424221484.jpg",
    "duration": 5.166667,
    "aspectRatio": "19:11",
    "resolution": "704p",
    "tags": [
      "动作特效",
      "动漫",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084625730407928155",
    "title": "沙漠公路上的动漫滑板人物",
    "titleEn": "An Anime Skateboarder on a Desert Road",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "滑板人物在棕榈树公路上前进，镜头抬升展示沙漠地形。",
    "summaryEn": "A skateboarder travels along a palm-lined road before the camera rises over desert terrain.",
    "prompt": null,
    "sourceUrl": "https://x.com/Framer_X/status/2084625730407928155",
    "sourceLabel": "X 原帖",
    "author": "Framer_X",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084625730407928155.mp4",
    "posterUrl": "/posters/x/2084625730407928155.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "人物表演",
      "自然风景"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084612873092030642",
    "title": "尖刺头饰与披萨音乐片",
    "titleEn": "A Spiked-Headdress and Pizza Music Clip",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物特写、披萨与粉蓝背景交替，日文大字进入片尾。",
    "summaryEn": "Character close-ups, pizza and pink-blue panels alternate before large Japanese end titles.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2084612873092030642",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084612873092030642.mp4",
    "posterUrl": "/posters/x/2084612873092030642.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "风格化与实验",
      "动态图形",
      "美食",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084606243516334180",
    "title": "乡间骑车动漫角色的本地生成",
    "titleEn": "A Locally Generated Anime Cyclist in the Countryside",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "戴草帽角色骑车经过乡间街道；作者指出中文发话未符合指令。",
    "summaryEn": "A straw-hatted character cycles through rural streets; the creator notes that Chinese speech did not follow the instruction.",
    "prompt": null,
    "sourceUrl": "https://x.com/eternityspring/status/2084606243516334180",
    "sourceLabel": "X 原帖",
    "author": "eternityspring",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084606243516334180.mp4",
    "posterUrl": "/posters/x/2084606243516334180.jpg",
    "duration": 10.208,
    "aspectRatio": "16:9",
    "resolution": "576p",
    "tags": [
      "动画",
      "动漫",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084603328265125929",
    "title": "沃尔特与笨拙实习生的虚构对白",
    "titleEn": "A Fictional Conversation with Walter and a Clumsy Intern",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "防护服人物在实验台旁对话，以夸张表情构成喜剧。",
    "summaryEn": "Protective-suit characters talk beside a lab bench with exaggerated comic expressions.",
    "prompt": "in Breaking Bad met-lab. \nCreate a goofy intern inspired by the comedic style of *Dumb and Dumber*, but do not recreate any exact character. (Walter White shouldn't change.)\n\nIntern: “Hey Walter, after we sell all this, do you think we’ll be rich?”\n\nWalter: “If everything goes well, yes.”\n\nIntern: “Okaaay… one more question. Would you send Skyler over to my place? My house is very warm, so she’d never be cold. We could have a really good time together… What do you think?”",
    "sourceUrl": "https://x.com/WolfRiccardo/status/2084603328265125929",
    "sourceLabel": "X 原帖",
    "author": "WolfRiccardo",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084603328265125929.mp4",
    "posterUrl": "/posters/x/2084603328265125929.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/WolfRiccardo/status/2084603328265125929"
  },
  {
    "id": "x-2084596207637188665",
    "title": "办公室预算花在 AI 订阅上的争论",
    "titleEn": "An Office Argument about AI Subscription Spending",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "迈克尔质问德怀特的订阅开支，手持与反应镜头模拟纪录喜剧。",
    "summaryEn": "Michael questions Dwight's subscription spending through handheld and reaction shots in a mockumentary style.",
    "prompt": "\"The Office (US) mockumentary style. Dunder Mifflin bullpen, mid-afternoon, harsh fluorescent office lighting, beige cubicles, cluttered desks, printers humming in the background. Handheld documentary camera with subtle shake, imperfect framing, sudden snap zooms during reactions, quick reframes as if a real documentary crew is trying to keep up. Muted office color palette, soft consumer-camera look.\n\nMichael Scott stands beside Dwight Schrute's desk, arms crossed, visibly annoyed. Dwight sits confidently at his desk, completely unfazed. Only Michael Scott and Dwight Schrute appear in the scene.\n\nDIALOGUE (≈15 seconds):\n\nMICHAEL (frustrated):\n\"Dwight, you spent the entire office budget on AI subscriptions?!\"\nDWIGHT (matter-of-fact):\n\"Correct.\"\n\nSnap zoom to Michael.\nMICHAEL:\n\"Why would you need twelve different AI models?!\"\n\nDWIGHT:\n\"They debate each other until the smartest answer survives.\"\n\nBeat.\nMICHAEL:\n\"...Couldn't you just use one?\"\n\nDWIGHT (confident):\n\"I don't hire one employee either.\"\nMichael stares silently into the documentary camera, completely defeated.\n\nSTYLE: Single-camera mockumentary, natural performances, dry deadpan humor, awkward pauses, documentary snap zooms, handheld camera movement, realistic office ambience, no laugh track, with only Michael Scott and Dwight Schrute in frame.\"",
    "sourceUrl": "https://x.com/karankendre/status/2084596207637188665",
    "sourceLabel": "X 原帖",
    "author": "karankendre",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084596207637188665.mp4",
    "posterUrl": "/posters/x/2084596207637188665.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/karankendre/status/2084596207637188665"
  },
  {
    "id": "x-2084595677779444169",
    "title": "手绘小船进入街景的双模型对比",
    "titleEn": "A Two-Model Comparison of Drawings Entering a Street",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "上下标注的画面展示纸上小船和鲸鱼变成街景图形。",
    "summaryEn": "Labeled stacked views show a drawn boat and whale becoming graphics in a street scene.",
    "prompt": null,
    "sourceUrl": "https://x.com/opener_ai/status/2084595677779444169",
    "sourceLabel": "X 原帖",
    "author": "opener_ai",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084595677779444169.mp4",
    "posterUrl": "/posters/x/2084595677779444169.jpg",
    "duration": 15.033984,
    "aspectRatio": "4:5",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "照片级写实",
      "插画与手绘",
      "城市街景",
      "奇幻"
    ],
    "category": "comparison",
    "styles": [
      "photoreal",
      "illustration"
    ],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084580595523608784",
    "title": "熊猫忍者与猴王竹林对决",
    "titleEn": "A Panda Ninja Battles the Monkey King",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "熊猫与持棍猴王在竹林和瀑布旁进行武术交锋。",
    "summaryEn": "A panda and staff-wielding Monkey King exchange martial-arts moves among bamboo and waterfalls.",
    "prompt": "Ultra-cinematic animated action film, 15 seconds, DreamWorks × Pixar quality, stylized but highly detailed, 4K HDR, dramatic golden sunlight filtering through a dense ancient bamboo forest, volumetric lighting, dynamic camera work, seamless character consistency, high-speed martial arts choreography, realistic cloth physics, flying bamboo leaves, cinematic sound design, no text, no watermark . Scene 1 (0–3s): Deep inside an ancient bamboo forest, a fearless Panda Ninja in a black ninja robe with a red waist sash slowly unsheathes a razor-sharp katana. Across the forest, the legendary Monkey King lands on a swaying bamboo stalk, spinning his golden staff with incredible speed. Close-up shots of determined eyes, leaves drifting through the air, intense silence before the battle. \nScene 2 (3–6s): Both warriors launch simultaneously, performing impossible high-speed jumps across bending bamboo trees. The camera follows from below as they sprint, flip, and ricochet from one bamboo stalk to another before colliding mid-air. Katana and golden staff crash together, creating bright sparks and a powerful shockwave. \nScene 3 (6–10s): A breathtaking martial arts battle unfolds across the bamboo canopy. The Monkey King attacks with lightning-fast spinning staff combinations while the Panda Ninja counters with elegant katana techniques, dodges, wall-runs, aerial flips, and rapid sword slashes. Fast cinematic camera movements, slow-motion impact moments, and dynamic close-ups emphasize every strike. \nScene 4 (10–13s): The fight reaches a gigantic waterfall. Both warriors leap through the mist, running across wet rocks above the roaring water. They exchange one final series of attacks as water splashes dramatically around them. \nScene 5 (13–15s):In epic slow motion, the Panda Ninja delivers a perfectly timed finishing katana strike while the Monkey King blocks with his golden staff. A massive burst of sparks and flying bamboo leaves fills the screen. Both land on opposite cliffs overlooking the waterfall, silently staring at each other with respect as the camera slowly pulls back to reveal the breathtaking bamboo valley. Cinematic ending, legendary warrior atmosphere.",
    "sourceUrl": "https://x.com/DoctorAmna11/status/2084580595523608784",
    "sourceLabel": "X 原帖",
    "author": "DoctorAmna11",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084580595523608784.mp4",
    "posterUrl": "/posters/x/2084580595523608784.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动作打斗",
      "动物与生物"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/DoctorAmna11/status/2084580595523608784"
  },
  {
    "id": "x-2084571189117366484",
    "title": "赛博游戏角色配置与行动演示",
    "titleEn": "A Cyber Game Character Setup and Action Demo",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "角色配置界面切换至雨夜街头行动，展示虚构游戏片段。",
    "summaryEn": "A character setup interface transitions into rainy street action in a fictional game sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/itsshara_ai/status/2084571189117366484",
    "sourceLabel": "X 原帖",
    "author": "itsshara_ai",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084571189117366484.mp4",
    "posterUrl": "/posters/x/2084571189117366484.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "3D / CG",
      "科幻太空",
      "城市街景"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "sci-fi",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084567967652843620",
    "title": "生发精华 UGC 广告概念",
    "titleEn": "A Hair Serum UGC Advertising Concept",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "产品瓶特写与人物介绍组成精华广告概念，不构成功效证明。",
    "summaryEn": "Bottle close-ups and a presenter form a serum advertising concept, without establishing product efficacy.",
    "prompt": "Create a 15-second authentic UGC-style hair growth serum ad using the provided product image as the exact product reference.\n\nSCENE 1 — 0–3s\nA young woman films herself in a bright bedroom using a smartphone front camera. Natural lighting, handheld movement, casual appearance. She looks at the camera and says:\n“I’ve been trying this hair growth serum lately…”\n\nSCENE 2 — 3–7s\nCut to a close handheld shot of the woman holding the serum bottle. She removes the dropper, applies a few drops directly to her scalp, and gently massages it in. Keep the movement natural and slightly imperfect like real UGC content.\n\nSCENE 3 — 7–11s\nMirror selfie shot. She runs her fingers through her hair, showing healthy-looking, fuller hair while casually talking to the camera:\n“And honestly, I love how easy it is to add to my routine.”\n\nSCENE 4 — 11–15s\nClose-up product shot on her bathroom counter. She picks up the bottle and smiles toward the camera. End with natural on-screen text:\n“Simple hair care. Every day.”\n\nStyle: authentic TikTok/Reels UGC, smartphone camera, realistic skin texture, natural expressions, subtle handheld motion, imperfect framing, casual home environment, soft daylight, realistic audio, no cinematic commercial look, no excessive beauty filters. Preserve the exact product packaging, label, bottle shape, and branding from the reference image.",
    "sourceUrl": "https://x.com/SyntheSarah/status/2084567967652843620",
    "sourceLabel": "X 原帖",
    "author": "SyntheSarah",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084567967652843620.mp4",
    "posterUrl": "/posters/x/2084567967652843620.jpg",
    "duration": 15.23229,
    "aspectRatio": "45:68",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "生活实拍感",
      "照片级写实",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/SyntheSarah/status/2084567967652843620"
  },
  {
    "id": "x-2084551020760433134",
    "title": "斯图尔特与白猫的客厅对白",
    "titleEn": "Stuart and a White Cat Talk in a Living Room",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红毛衣小鼠与沙发上的白猫通过近景和表情互动。",
    "summaryEn": "A red-sweater mouse and white cat on a sofa interact through close-ups and expressions.",
    "prompt": "15-second 3D animated scene from Stuart Little. Cute little mouse Stuart Little stands on the living room floor looking up at the big fluffy white cat Snowbell. Snowbell sits on the couch with a smug expression, tail twitching. Soft warm indoor lighting, detailed fur and fabric textures, classic Stuart Little movie animation style. Stuart says cheerfully “Hey Snowbell, want to be friends?” Snowbell replies in a sarcastic voice “Friends? You’re a mouse, I’m the real pet around here!” Camera slowly pushes in as Snowbell leans down closer to Stuart. Smooth animation, expressive faces, native stereo dialogue audio, 24fps, cinematic quality.",
    "sourceUrl": "https://x.com/MrDasOnX/status/2084551020760433134",
    "sourceLabel": "X 原帖",
    "author": "MrDasOnX",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084551020760433134.mp4",
    "posterUrl": "/posters/x/2084551020760433134.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "3D / CG",
      "动物与生物"
    ],
    "category": "dialogue",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/MrDasOnX/status/2084551020760433134"
  },
  {
    "id": "x-2084544246737248568",
    "title": "绿色藤蔓手写文字动画",
    "titleEn": "Green Vine-Like Handwritten Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "绿色书法字逐步展开，叶片装饰围绕完整句子生长。",
    "summaryEn": "Green calligraphic lettering unfolds as leaf decorations grow around a sentence.",
    "prompt": null,
    "sourceUrl": "https://x.com/umesh_ai/status/2084544246737248568",
    "sourceLabel": "X 原帖",
    "author": "umesh_ai",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084544246737248568.mp4",
    "posterUrl": "/posters/x/2084544246737248568.jpg",
    "duration": 14.464,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "自然风景"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084521314904408321",
    "title": "哥特术士与机械乌鸦的书房",
    "titleEn": "A Gothic Sorcerer and Mechanical Raven in a Study",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "机械乌鸦飞过药瓶与蜡烛，术士书桌前出现发光效果。",
    "summaryEn": "A mechanical raven passes bottles and candles before a glow appears at a sorcerer's desk.",
    "prompt": null,
    "sourceUrl": "https://x.com/Parul_Gautam7/status/2084521314904408321",
    "sourceLabel": "X 原帖",
    "author": "Parul_Gautam7",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084521314904408321.mp4",
    "posterUrl": "/posters/x/2084521314904408321.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084516107378159820",
    "title": "多段音乐参考的场景切换测试",
    "titleEn": "A Scene-Change Test with Multiple Music References",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "原片比较音乐切换时更换场景或仅改变角色动作的生成结果。",
    "summaryEn": "The source compares changing scenes at music transitions with changing only character actions.",
    "prompt": null,
    "sourceUrl": "https://x.com/RBLabQ/status/2084516107378159820",
    "sourceLabel": "X 原帖",
    "author": "RBLabQ",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084516107378159820.mp4",
    "posterUrl": "/posters/x/2084516107378159820.jpg",
    "duration": 36.269569,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "本地生成与工作流",
      "风格化与实验",
      "人物表演"
    ],
    "category": "workflow",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084509010485756396",
    "title": "橘猫与主人的温暖一天",
    "titleEn": "A Warm Day with an Orange Cat and Its Owner",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "喂食、抚摸与夕阳长椅镜头串起人与猫的日常。",
    "summaryEn": "Feeding, petting and a sunset bench connect a cat-and-owner daily-life sequence.",
    "prompt": "Create an ultra-photorealistic cinematic video of a fluffy orange tabby cat and its loving owner in a warm, cozy story. The video begins in a bright, sunlit home where the owner gently fills the cat's food bowl while the cat patiently waits, then lovingly pets its head. The scene smoothly transitions to a peaceful park during golden hour, where the owner walks the cat on a harness along a tree-lined path as the cat curiously explores its surroundings. End with the owner sitting on a bench at sunset while the orange tabby rests nearby, both peacefully watching the glowing sky. Use soft natural lighting, realistic fur movement, smooth cinematic camera motion, shallow depth of field, warm color grading, emotional storytelling, ultra-detailed textures, 4K quality, and a calm, heartwarming atmosphere with no text, logos, or watermark.",
    "sourceUrl": "https://x.com/Lianaalane/status/2084509010485756396",
    "sourceLabel": "X 原帖",
    "author": "Lianaalane",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084509010485756396.mp4",
    "posterUrl": "/posters/x/2084509010485756396.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动物与生物",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Lianaalane/status/2084509010485756396"
  },
  {
    "id": "x-2084507376825299177",
    "title": "兔子厨师与煎饼大劫案",
    "titleEn": "A Rabbit Chef and a Pancake Heist",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "兔子厨师制作煎饼，松鼠在厨房旁加入故事。",
    "summaryEn": "A rabbit chef makes pancakes as a squirrel joins the kitchen story.",
    "prompt": null,
    "sourceUrl": "https://x.com/iamrealsnow/status/2084507376825299177",
    "sourceLabel": "X 原帖",
    "author": "iamrealsnow",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084507376825299177.mp4",
    "posterUrl": "/posters/x/2084507376825299177.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "美食",
      "动物与生物"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "food",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084505247385346518",
    "title": "明亮公寓中的晨间生活广告",
    "titleEn": "A Morning Lifestyle Ad in a Bright Apartment",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "家居收纳与日常动作特写串联现代公寓的生活广告。",
    "summaryEn": "Home organization and daily-action close-ups connect a modern-apartment lifestyle ad.",
    "prompt": "A bright, modern apartment filled with soft natural morning light, warm neutral color palette, minimalist Scandinavian interior, cinematic lifestyle commercial, ultra-realistic, premium home aesthetic, smooth camera movement, shallow depth of field, crisp 4K detail.\nScene 1 – Start the Day\nA young woman in a light gray athleisure outfit walks into a spotless kitchen and opens the refrigerator with a relaxed smile. Soft sunlight streams through large windows. Slow cinematic dolly shot.\nScene 2 – Vacuum Cleaning\nShe effortlessly vacuums the hardwood floor using a sleek cordless vacuum cleaner. The camera tracks beside her while the background remains bright and airy. Smooth gimbal movement with subtle slow motion.\nScene 3 – Organizing the Drawer\nTop-down overhead shot of her neatly arranging skincare bottles, cosmetics, and accessories inside a wooden drawer using white storage organizers. Every item fits perfectly into place, emphasizing order and cleanliness.\nScene 4 – Making the Bed\nOverhead cinematic shot of the woman carefully spreading a fresh white bedsheet across a large bed. The fabric flows naturally as sunlight illuminates the room through floor-to-ceiling curtains, creating a calm and luxurious atmosphere.\nFinal Scene\nWide shot of the immaculate apartment. Everything is clean, organized, and peaceful. The woman stands smiling, admiring her work while warm morning light fills the home. Premium lifestyle commercial look, realistic lighting, smooth transitions, ultra-detailed, HDR, 4K, cinematic color grading.",
    "sourceUrl": "https://x.com/CaliraVal/status/2084505247385346518",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084505247385346518.mp4",
    "posterUrl": "/posters/x/2084505247385346518.jpg",
    "duration": 14.326009,
    "aspectRatio": "45:67",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "日常生活",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "daily-life",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2084505247385346518"
  },
  {
    "id": "x-2084501787827130572",
    "title": "朝颜花旁的猫耳角色与白猫",
    "titleEn": "A Cat-Eared Character and White Cat beside Morning Glories",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "猫耳角色抱起白猫并轻抚；本站播放原帖标明的 H3 版本。",
    "summaryEn": "A cat-eared character lifts and pets a white cat; this is the source's identified H3 version.",
    "prompt": null,
    "sourceUrl": "https://x.com/Akkyman2/status/2084501787827130572",
    "sourceLabel": "X 原帖",
    "author": "Akkyman2",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084501787827130572.mp4",
    "posterUrl": "/posters/x/2084501787827130572.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动物与生物"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084498059711349065",
    "title": "办公室角色争论 AI 能否取代电影人",
    "titleEn": "Office Characters Debate AI Replacing Filmmakers",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "吉姆与德怀特以夸张手势和反应展开生成的办公室对白。",
    "summaryEn": "Jim and Dwight exchange exaggerated gestures and reactions in a generated office conversation.",
    "prompt": "Jim and Dwight from The Office arguing about whether AI can replace filmmakers",
    "sourceUrl": "https://x.com/karankendre/status/2084498059711349065",
    "sourceLabel": "X 原帖",
    "author": "karankendre",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084498059711349065.mp4",
    "posterUrl": "/posters/x/2084498059711349065.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/karankendre/status/2084498059711349065"
  },
  {
    "id": "x-2084496491175973183",
    "title": "多人重叠发话的生成测试",
    "titleEn": "A Multi-Person Overlapping-Speech Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "四人合影式构图中，人物同时开口并产生不同表情。",
    "summaryEn": "Four people in a group-photo composition speak at once with varying expressions.",
    "prompt": "They speak over each other saying \"OpenAI is nothing without its people\"",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2084496491175973183",
    "sourceLabel": "X 原帖",
    "author": "cocktailpeanut",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084496491175973183.mp4",
    "posterUrl": "/posters/x/2084496491175973183.jpg",
    "duration": 10.912,
    "aspectRatio": "1:1",
    "resolution": "640p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2084496491175973183"
  },
  {
    "id": "x-2084476676222939486",
    "title": "室内双人物的情绪发话测试",
    "titleEn": "An Indoor Two-Person Emotional Speech Test",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "两位白衣人物近距离交流，原作用于测试情绪声音。",
    "summaryEn": "Two white-clad characters interact closely in a source test of emotional voice performance.",
    "prompt": null,
    "sourceUrl": "https://x.com/imagixaijp/status/2084476676222939486",
    "sourceLabel": "X 原帖",
    "author": "imagixaijp",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084476676222939486.mp4",
    "posterUrl": "/posters/x/2084476676222939486.jpg",
    "duration": 8.064,
    "aspectRatio": "13:23",
    "resolution": "416p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084462495641780321",
    "title": "办公室角色介绍 AI 平台的虚构对白",
    "titleEn": "A Fictional Office Conversation about an AI Platform",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "吉姆与德怀特在工位旁讨论 AI 平台，穿插饮水与反应镜头。",
    "summaryEn": "Jim and Dwight discuss an AI platform at their desks with drinking and reaction shots.",
    "prompt": "Jim and Dwight from The Office discuss how they found all in one AI platform Easy-Peasy AI that is very simple to use compare to other AI tools",
    "sourceUrl": "https://x.com/dimoff2000/status/2084462495641780321",
    "sourceLabel": "X 原帖",
    "author": "dimoff2000",
    "publishedAt": "2026-08-04",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084462495641780321.mp4",
    "posterUrl": "/posters/x/2084462495641780321.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "照片级写实"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/dimoff2000/status/2084462495641780321"
  },
  {
    "id": "x-2084336165369311502",
    "title": "插画歌手与歌词参考音乐片",
    "titleEn": "An Illustrated Singer and Reference-Lyric Music Clip",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝色服装歌手与成组歌词图形互动，形成层叠舞台画面。",
    "summaryEn": "A blue-clad singer interacts with grouped lyric graphics in layered stage compositions.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2084336165369311502",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084336165369311502.mp4",
    "posterUrl": "/posters/x/2084336165369311502.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "插画与手绘",
      "动态图形",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084241879097303513",
    "title": "日式厨房抹茶套装 UGC 概念",
    "titleEn": "A Matcha Kit UGC Concept in a Japanese-Inspired Kitchen",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物展示抹茶套装，注水特写与品饮动作连接产品介绍。",
    "summaryEn": "A presenter shows a matcha kit with water-pouring close-ups and tasting shots.",
    "prompt": "Use the uploaded reference image as the exact character reference. Preserve her facial identity, hairstyle, eye color, skin tone, body proportions, and facial consistency throughout every shot.\nCreate a 15-second ultra-realistic UGC-style lifestyle advertisement featuring a realistic young woman sharing her favorite premium Matcha Starter Kit in a bright, minimalist Japanese-inspired kitchen with warm morning sunlight, natural wood textures, soft neutral décor, ceramic tableware, and a calm luxury aesthetic.\nShe smiles naturally while holding an elegant matcha tin and says this has become her favorite way to start every morning. She opens the premium matcha container, carefully scoops vibrant green matcha powder into a handcrafted ceramic bowl, pours warm water, and whisks it smoothly using a traditional bamboo whisk until a rich, creamy foam forms. She lifts the bowl, takes a small sip, closes her eyes with a genuine smile, and enjoys the peaceful morning moment.\nCapture authentic handheld smartphone shots, over-the-shoulder angles, cinematic macro close-ups of the vivid green matcha powder, the bamboo whisk creating silky foam, the handcrafted ceramic bowl, rising steam, natural hand movements, and warm sunlight reflecting across the table. Finish with her holding the completed bowl beside her face, smiling warmly into the camera.\nAuthentic creator-style content, natural facial expressions, healthy glowing skin, cozy morning atmosphere, luxury lifestyle aesthetic, subtle depth of field, premium color grading, smartphone-shot realism, vertical 9:16, 4K HDR, no text, subtitles, logos, or watermarks.",
    "sourceUrl": "https://x.com/SimplyAnnisa/status/2084241879097303513",
    "sourceLabel": "X 原帖",
    "author": "SimplyAnnisa",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084241879097303513.mp4",
    "posterUrl": "/posters/x/2084241879097303513.jpg",
    "duration": 15.2,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "生活实拍感",
      "照片级写实",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/SimplyAnnisa/status/2084241879097303513"
  },
  {
    "id": "x-2084220704623399083",
    "title": "水晶峡谷里的摩托逃亡",
    "titleEn": "A Motorcycle Escape through a Crystal Canyon",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "骑手穿过巨型晶体，天空中出现燃烧飞船与爆炸光。",
    "summaryEn": "A rider passes giant crystals beneath a burning spacecraft and bursts of light.",
    "prompt": null,
    "sourceUrl": "https://x.com/Gelat0x/status/2084220704623399083",
    "sourceLabel": "X 原帖",
    "author": "Gelat0x",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084220704623399083.mp4",
    "posterUrl": "/posters/x/2084220704623399083.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "科幻太空"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084212806434406610",
    "title": "九人参考照片生成的幕后群像",
    "titleEn": "An Ensemble Scene from Nine Reference Portraits",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "生成的片场镜头依次展示多位人物，原帖说明使用了九张参考照片。",
    "summaryEn": "A generated film-set shot reveals several people; the source states that nine reference portraits were used.",
    "prompt": null,
    "sourceUrl": "https://x.com/lukasersil/status/2084212806434406610",
    "sourceLabel": "X 原帖",
    "author": "lukasersil",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084212806434406610.mp4",
    "posterUrl": "/posters/x/2084212806434406610.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "照片级写实",
      "人物表演"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084199438646477192",
    "title": "裁缝店换装的另一版诠释",
    "titleEn": "Another Interpretation of a Tailor-Shop Makeover",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物进入裁缝店，布料与工具旋转后以西装造型出门。",
    "summaryEn": "A figure enters a tailor shop and leaves in a suit after fabric and tools swirl around him.",
    "prompt": null,
    "sourceUrl": "https://x.com/MonetizationDon/status/2084199438646477192",
    "sourceLabel": "X 原帖",
    "author": "MonetizationDon",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084199438646477192.mp4",
    "posterUrl": "/posters/x/2084199438646477192.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "时尚",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084178540065657251",
    "title": "校园广场上的轻快舞蹈",
    "titleEn": "An Upbeat Dance in a Campus Plaza",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在阳光下的花坛与喷泉前跳舞，竖屏镜头保持全身动作。",
    "summaryEn": "A dancer performs before sunny flowers and a fountain in full-body vertical framing.",
    "prompt": "20대 성인 여대생이 캠퍼스 야외 광장에서 신나는 댄스를 추는 15초 세로형 릴스 영상. 9:16, 30fps, 고해상도, 밝고 상큼한 K-pop 일렉트로팝 분위기. 빠르고 통통 튀는 128BPM 리듬, 경쾌한 신스 멜로디, 밝은 베이스, 손뼉 소리와 귀여운 효과음이 들어간 신나는 음악. 음악의 박자와 춤 동작이 정확하게 일치한다. 주인공은 귀엽고 사랑스러운 20대 여성이다. 맑고 깨끗한 피부, 큰 갈색 눈, 자연스럽게 올라간 눈꼬리, 작고 갸름한 얼굴, 부드러운 앞머리, 긴 웨이브 갈색 머리, 밝고 순수한 미소, 생기 있는 표정. 머리에는 작은 하늘색 리본을 하고 있다. 레몬색 반팔 티셔츠 위에 흰색과 민트색의 짧은 니트 가디건, 하이웨이스트 연청 플리츠 스커트, 흰색 발목 양말과 깨끗한 흰색 스니커즈를 착용한다. 상큼하고 단정하며 발랄한 대학생 스타일, 과도한 노출 없음. 배경은 맑은 오후의 예쁜 캠퍼스 광장이다. 연한 하늘색 하늘, 초록색 나무, 분홍색과 노란색 꽃이 핀 화단, 파스텔색 풍선, 작은 벤치, 햇빛에 반짝이는 분수, 부드러운 나뭇잎 그림자가 보인다. 배경은 선명하지만 주인공에게 초점이 맞고, 따뜻한 햇살과 은은한 렌즈 플레어, 밝은 보케 효과를 사용한다. 0–3초: 주인공이 화면 왼쪽에서 가볍게 뛰어 들어온다. 음악의 첫 박자에 맞춰 양손으로 작은 하트를 만들고 카메라를 보며 활짝 웃는다. 오른쪽 어깨와 왼쪽 어깨를 번갈아 튕기며 귀엽게 고개를 기울인다. 카메라는 정면에서 부드럽게 전진한다. 3–6초: 오른쪽과 왼쪽으로 빠르게 스텝터치하며 양팔을 좌우로 크게 흔든다. 두 박자마다 손가락 하트를 만들고, 머리카락이 자연스럽게 흔들린다. 카메라는 허리 위 중심으로 좌우 트래킹한다. 6–9초: 양손으로 물결 모양 동작을 하며 몸을 좌우로 흔든다. 한 바퀴 가볍게 회전한 뒤 카메라를 향해 윙크하고 브이 포즈를 한다. 꽃잎과 작은 반짝임이 주변에 흩날린다. 9–12초: 하이라이트 안무. 발을 두 번 빠르게 구르고 양팔을 머리 위로 올린 뒤 큰 하트를 만든다. 이어서 한 손으로 볼을 살짝 누르는 귀여운 포즈를 취한다. 카메라는 음악에 맞춰 살짝 줌인한다. 12–15초: 주인공이 앞으로 두 걸음 다가오며 양팔을 활짝 펼친다. 마지막 박자에 가볍게 점프하고 한 바퀴 반 회전한 뒤 카메라 앞에서 큰 하트 포즈를 만든다. 환하게 웃는 표정으로 영상이 밝게 끝난다. 자연스러운 춤 동작, 안정적인 얼굴과 손, 정확한 박자 싱크, 부드러운 카메라 움직임, 밝고 청량한 색감. 네거티브 프롬프트: 미성년자, 과도한 노출, 선정적인 안무, 흐릿한 얼굴, 손가락 오류, 팔다리 중복, 부자연스러운 관절, 어두운 분위기, 불안정한 카메라, 다른 인물 등장, 자막, 로고, 워터마크",
    "sourceUrl": "https://x.com/nowlovepan/status/2084178540065657251",
    "sourceLabel": "X 原帖",
    "author": "nowlovepan",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084178540065657251.mp4",
    "posterUrl": "/posters/x/2084178540065657251.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "舞蹈",
      "照片级写实"
    ],
    "category": "dance",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/nowlovepan/status/2084178540065657251"
  },
  {
    "id": "x-2084158511261405214",
    "title": "城市电动踏板车概念广告",
    "titleEn": "An Urban Electric Scooter Concept Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "车灯、储物空间、骑行与夕阳产品镜头组成踏板车广告概念。",
    "summaryEn": "Headlights, storage, riding and sunset product shots form a scooter advertising concept.",
    "prompt": "Create a 14-second ultra-realistic cinematic commercial for the YADEA G5 Electric Scooter 2026. A stylish female rider in a black leather jacket and helmet showcases the gray scooter through 10 fast-paced scenes: LED headlight turns on, hero reveal, body close-up, digital dashboard activation, smooth city ride, under-seat storage demo, Smart Key unlock, side tracking ride, feature highlights (Smart Key, LED Lights, Digital Display, Eco Friendly, Long Range), and a golden-hour sunset hero shot with the \"YADEA G5 – Ride The Future\" logo. Premium commercial quality, smooth transitions, dynamic camera movements (Push In, Orbit, Truck Right, Tracking, Dolly In), cinematic lighting, realistic reflections, HDR, shallow depth of field, 4K/8K, luxury advertisement style.",
    "sourceUrl": "https://x.com/Lianaalane/status/2084158511261405214",
    "sourceLabel": "X 原帖",
    "author": "Lianaalane",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084158511261405214.mp4",
    "posterUrl": "/posters/x/2084158511261405214.jpg",
    "duration": 14.767891,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "产品与广告",
      "城市街景"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Lianaalane/status/2084158511261405214"
  },
  {
    "id": "x-2084155623139348764",
    "title": "雨夜街道上的太空猫",
    "titleEn": "A Space Cat on a Rainy Street",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "戴透明头盔的猫沿霓虹街道前行，湿地反射灯光。",
    "summaryEn": "A cat wearing a transparent helmet walks along a neon street reflected in wet pavement.",
    "prompt": "宇宙猫、雨の町を歩く",
    "sourceUrl": "https://x.com/kiyoshi_shin/status/2084155623139348764",
    "sourceLabel": "X 原帖",
    "author": "kiyoshi_shin",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084155623139348764.mp4",
    "posterUrl": "/posters/x/2084155623139348764.jpg",
    "duration": 15.168,
    "aspectRatio": "9:5",
    "resolution": "480p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动物与生物",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/kiyoshi_shin/status/2084155623139348764"
  },
  {
    "id": "x-2084140665395347785",
    "title": "红色聚光灯下的运动鞋展示",
    "titleEn": "A Sneaker under Red Spotlights",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红黑运动鞋在发光底座上展示，片尾加入霓虹字形。",
    "summaryEn": "A red-and-black sneaker appears on a lit pedestal before neon lettering enters.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tech2Wild/status/2084140665395347785",
    "sourceLabel": "X 原帖",
    "author": "Tech2Wild",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084140665395347785.mp4",
    "posterUrl": "/posters/x/2084140665395347785.jpg",
    "duration": 5.248,
    "aspectRatio": "26:15",
    "resolution": "480p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "3D / CG",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "cg-3d"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084138096766165223",
    "title": "古林骑射与部落归来",
    "titleEn": "Mounted Archery in a Forest and a Tribal Return",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "骑手穿过林间晨光射箭，随后回到旗帜环绕的营地。",
    "summaryEn": "A rider shoots an arrow in morning forest light before returning to a camp surrounded by banners.",
    "prompt": "Create a 15-second ultra-cinematic historical video inspired by the atmosphere of a medieval Turkic tribal world. The setting is a vast, dense ancient forest with towering pine trees, thick mist, golden morning sunlight filtering through the branches, drifting dust particles, and a mysterious wilderness. The visual style is realistic, cinematic, and highly detailed with natural colors, dramatic lighting, and Hollywood-quality camera work.\nThe main character is a fearless young female tribal warrior wearing authentic medieval tribal clothing inspired by Central Asian nomadic culture. She wears a long leather coat with fur trim, leather armor, a traditional embroidered belt, leather boots, finger guards for archery, a headband, and a quiver full of arrows on her back. She carries a beautifully crafted wooden recurve bow in her hand. Her horse is a powerful black war horse with traditional leather saddle and tribal accessories.\nMaintain the same character, horse, clothing, weapons, and facial appearance consistently throughout the entire video.\n\n0:00 – 0:01 A wide cinematic aerial shot reveals a dense green forest covered in light morning fog. The female warrior rides her black horse confidently through the trees. Leaves move naturally in the wind while sunlight creates dramatic rays through the forest canopy.\n\n0:01 – 0:02 Medium tracking shot from the front-left side. The horse continues running between the trees. The warrior scans the forest carefully while gripping her bow. Her quiver filled with arrows is visible behind her shoulder. Her clothing and hair move naturally with the speed of the horse.\n\n0:02 – 0:03 Wide establishing shot of a large tribal camp deep inside the forest. Dozens of tribespeople wearing authentic medieval tribal clothing walk between tents. Horses stand nearby, smoke rises from campfires, children play, warriors prepare weapons, and large tribal flags wave in the wind.\n\n0:03 – 0:04 The warrior slows her horse near the edge of the forest. She notices movement in the sky. She quickly reaches for an arrow, places it on the bowstring, and carefully aims upward with intense concentration.\n\n0:04 – 0:06 Close-up cinematic slow-motion sequence. A bird flies across the forest canopy. The warrior releases the arrow. The arrow slices through the air with dramatic motion blur and accurately strikes the bird in mid-flight.\n\n0:06 – 0:08 The bird falls toward the forest floor. While keeping her horse moving, the warrior skillfully retrieves the hunted bird. The camera circles around her as sunlight shines through the trees, emphasizing her confidence and precision.\n\n0:08 – 0:10 She immediately turns her horse and gallops at full speed toward the tribal camp. Hooves pound the ground, dirt and leaves fly behind, tree branches blur past, and the camera alternates between low-angle tracking shots and wide cinematic views to emphasize speed and power.\n\n0:10 – 0:12 The horse charges into the tribal settlement. Villagers, warriors, elders, women, and children notice her arrival. Many smile, raise their hands, and gather to welcome the successful hunter.\n\n0:12 – 0:14 The warrior stops the horse in the center of the camp and dismounts confidently while holding the hunted bird. Tribal elders nod with pride. Warriors stand respectfully around her. Campfires burn softly, horses move naturally, and banners continue waving in the breeze.\n\n0:14 – 0:15 Epic final wide shot. The warrior stands beside her horse in front of the bustling tribal camp with mountains visible in the background. Tribal flags wave dramatically as everyone gathers around. The camera slowly pulls back to reveal the entire camp before fading out with a powerful cinematic ending.",
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2084138096766165223",
    "sourceLabel": "X 原帖",
    "author": "ayzalnooor24521",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084138096766165223.mp4",
    "posterUrl": "/posters/x/2084138096766165223.jpg",
    "duration": 15.46449,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "动作打斗",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ayzalnooor24521/status/2084138096766165223"
  },
  {
    "id": "x-2084134888798920779",
    "title": "为动漫视频追加地裂特效",
    "titleEn": "Adding Ground-Fissure Effects to Anime Video",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "原片以对比画面展示在已有视频中加入裂纹与冲击效果。",
    "summaryEn": "Comparison views demonstrate adding ground cracks and impact effects to an existing video.",
    "prompt": null,
    "sourceUrl": "https://x.com/KEETY2591756/status/2084134888798920779",
    "sourceLabel": "X 原帖",
    "author": "KEETY2591756",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084134888798920779.mp4",
    "posterUrl": "/posters/x/2084134888798920779.jpg",
    "duration": 13.098667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "本地生成与工作流",
      "动漫",
      "动作打斗"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084127897149948250",
    "title": "金毛犬与主人共度日常",
    "titleEn": "A Golden Retriever Shares a Day with Its Owner",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "喂食、草地接球与沙发休息组成宠物生活片段。",
    "summaryEn": "Feeding, fetching on grass and resting on a sofa form a pet lifestyle sequence.",
    "prompt": "A cinematic 15-second heartwarming lifestyle film following a cheerful Golden Retriever through the perfect morning with its loving owner. The day begins in a sunlit kitchen where the dog patiently waits for breakfast before enjoying every bite with excitement. They head to a beautiful park for a joyful walk and an energetic game of fetch, capturing playful moments in warm golden light. Back home, the owner gently brushes the dog's silky fur before they relax together on a cozy couch, ending the day with quiet companionship and affectionate ear scratches. Smooth cinematic camera movements, realistic pet behavior, expressive emotions, premium commercial aesthetic, volumetric sunlight, soft lens flares, natural fur physics, immersive atmosphere, subtle film grain, 4K HDR, seamless continuity, no text, no logos.",
    "sourceUrl": "https://x.com/nawalsehar/status/2084127897149948250",
    "sourceLabel": "X 原帖",
    "author": "nawalsehar",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084127897149948250.mp4",
    "posterUrl": "/posters/x/2084127897149948250.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动物与生物",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/nawalsehar/status/2084127897149948250"
  },
  {
    "id": "x-2084123402390864013",
    "title": "发光声波舞台上的人物演唱",
    "titleEn": "A Singer on a Glowing Waveform Stage",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在紫白声波背景前演唱，以面部与全身镜头交替呈现。",
    "summaryEn": "A performer sings against purple-white waveforms through alternating face and full-body shots.",
    "prompt": null,
    "sourceUrl": "https://x.com/MrDavids1/status/2084123402390864013",
    "sourceLabel": "X 原帖",
    "author": "MrDavids1",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084123402390864013.mp4",
    "posterUrl": "/posters/x/2084123402390864013.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "3D / CG",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084105318070730928",
    "title": "夏日祭立体纸艺翻页书",
    "titleEn": "A Summer Festival Pop-Up Paper Book",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "浴衣人物、鸟居、花朵与烟花从立体纸页中展开。",
    "summaryEn": "A yukata figure, shrine gate, flowers and fireworks unfold from dimensional paper pages.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2084105318070730928",
    "sourceLabel": "X 原帖",
    "author": "su_nagomi",
    "publishedAt": "2026-08-03",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084105318070730928.mp4",
    "posterUrl": "/posters/x/2084105318070730928.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "插画与手绘",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "illustration"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084039844909703520",
    "title": "仓库中的高速特效战斗",
    "titleEn": "A High-Speed Effects Battle in a Warehouse",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "仓库内快速交锋伴随紫色能量与地面碎裂；作者提示存在局部形变。",
    "summaryEn": "Fast warehouse combat features purple energy and breaking ground; the creator notes some local deformation.",
    "prompt": null,
    "sourceUrl": "https://x.com/souhiro_meem_ch/status/2084039844909703520",
    "sourceLabel": "X 原帖",
    "author": "souhiro_meem_ch",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084039844909703520.mp4",
    "posterUrl": "/posters/x/2084039844909703520.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2084034217613434955",
    "title": "夕阳街道上的歌词与口型音乐片",
    "titleEn": "Lyrics and Lip Sync on a Sunset Street",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "动漫人物在住宅街表演，日文歌词跟随镜头进入画面。",
    "summaryEn": "An anime character performs on a residential street as Japanese lyrics enter with the shots.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2084034217613434955",
    "sourceLabel": "X 原帖",
    "author": "Naonekozamurai",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2084034217613434955.mp4",
    "posterUrl": "/posters/x/2084034217613434955.jpg",
    "duration": 13.142494,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "音乐演出",
      "城市街景"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083988251824554100",
    "title": "跑车轮胎仪表与漂移动作",
    "titleEn": "Sports-Car Tires, Gauges and Drifting",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "轮胎、转速表与烟尘漂移以快速镜头连接汽车广告。",
    "summaryEn": "Tires, a tachometer and smoky drifting connect through fast automotive advertising shots.",
    "prompt": "Sports car commercial, 15 seconds. Adrenaline sequence: low-angle tracking shot of tires spinning on asphalt, whip pan to headlights igniting in the dark, crash zoom on gear shift being pulled, quick cut to speedometer needle climbing, drone shot swooping over car cutting through mountain road, 360° bullet-time around car mid-drift with dust cloud. Cinematic automotive grade, deep blacks and hot orange sparks, dramatic contrast, Michael Bay-style kinetic editing, hyper-realistic 4K.",
    "sourceUrl": "https://x.com/Malzahran2/status/2083988251824554100",
    "sourceLabel": "X 原帖",
    "author": "Malzahran2",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083988251824554100.mp4",
    "posterUrl": "/posters/x/2083988251824554100.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Malzahran2/status/2083988251824554100"
  },
  {
    "id": "x-2083980738970489060",
    "title": "木屋前的乡间生活片段",
    "titleEn": "A Country-Life Sequence outside Wooden Homes",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物坐在门阶、晾晒床单并走过安静街道。",
    "summaryEn": "A woman sits on a doorstep, hangs a sheet and walks along a quiet street.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaAyyoub/status/2083980738970489060",
    "sourceLabel": "X 原帖",
    "author": "MaAyyoub",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083980738970489060.mp4",
    "posterUrl": "/posters/x/2083980738970489060.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083978550604914699",
    "title": "罗马红裙时装旅行广告",
    "titleEn": "A Red-Skirt Fashion Campaign in Rome",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "同一时装人物出现在西班牙阶梯、斗兽场与交通工具内。",
    "summaryEn": "The same fashion figure appears at the Spanish Steps, Colosseum and inside transport.",
    "prompt": null,
    "sourceUrl": "https://x.com/KeorUnreal/status/2083978550604914699",
    "sourceLabel": "X 原帖",
    "author": "KeorUnreal",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083978550604914699.mp4",
    "posterUrl": "/posters/x/2083978550604914699.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚",
      "城市街景"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083949566009778473",
    "title": "奇幻生物图鉴界面动画",
    "titleEn": "A Fantasy Creature Encyclopedia Interface",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "界面在多种生物条目间切换，展示角色动作与资料布局。",
    "summaryEn": "An interface switches between creature entries, showing motion and information layouts.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2083949566009778473",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083949566009778473.mp4",
    "posterUrl": "/posters/x/2083949566009778473.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "3D / CG",
      "动物与生物",
      "奇幻"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "creatures",
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083930956998521128",
    "title": "无线耳机与都市运动广告",
    "titleEn": "Wireless Earbuds and Urban Movement",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "耳机盒开启、佩戴特写与运动镜头组成产品广告。",
    "summaryEn": "A case opening, earbud close-up and movement shots form a product ad.",
    "prompt": "Energetic wireless earbuds ad, 15 seconds. Rapid montage: overhead shot of earbuds case snapping open, whip pan to young woman jogging in city street, quick tracking shot behind her as beat drops, macro close-up on earbud sliding into ear, 360° orbit around dancer mid-move, crash zoom on charging case LED. Vibrant urban color grade, teal and magenta neon accents, motion blur transitions, hyper-realistic, fast-cut MTV style editing.",
    "sourceUrl": "https://x.com/Malzahran2/status/2083930956998521128",
    "sourceLabel": "X 原帖",
    "author": "Malzahran2",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083930956998521128.mp4",
    "posterUrl": "/posters/x/2083930956998521128.jpg",
    "duration": 15.168,
    "aspectRatio": "106:45",
    "resolution": "360p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "产品与广告",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Malzahran2/status/2083930956998521128"
  },
  {
    "id": "x-2083906075724861531",
    "title": "失控转笼前的动漫吐槽",
    "titleEn": "An Anime Reaction to an Out-of-Control Playground Spinner",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "前景角色惊讶吐槽，身后人物抓住高速旋转的游乐设施。",
    "summaryEn": "A foreground character reacts in shock while another clings to a rapidly spinning playground structure.",
    "prompt": "後方でものすごい勢いで回転する公園のグローブジャングルに○○が「あぁぁぁー！」と叫びながら必死にしがみついている。巻き上がる風と砂埃。勢いが伝わる演出と集中線。ギャグアニメ風の演出。\nその手前で○○の肩から上が画面右下の画面外から勢いよくフレームインするのと同時に「すごいことになってるー！！」と目を飛び出して驚いた表情でツッコむ。顔の角度は正面左斜め上を向いている。ギャグアニメ風の効果音。",
    "sourceUrl": "https://x.com/eggmaya_AI/status/2083906075724861531",
    "sourceLabel": "X 原帖",
    "author": "eggmaya_AI",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083906075724861531.mp4",
    "posterUrl": "/posters/x/2083906075724861531.jpg",
    "duration": 5.248,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/eggmaya_AI/status/2083906075724861531"
  },
  {
    "id": "x-2083889173686001829",
    "title": "暗室屏幕扩展成悬浮网络",
    "titleEn": "Screens Expand into a Floating Network in a Dark Room",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "桌前屏幕逐渐扩展为多层悬浮窗口，连接线贯穿空间。",
    "summaryEn": "Desk screens expand into layered floating windows linked across the room.",
    "prompt": null,
    "sourceUrl": "https://x.com/NEXUS_TO_NOVA/status/2083889173686001829",
    "sourceLabel": "X 原帖",
    "author": "NEXUS_TO_NOVA",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083889173686001829.mp4",
    "posterUrl": "/posters/x/2083889173686001829.jpg",
    "duration": 15.168,
    "aspectRatio": "106:45",
    "resolution": "360p",
    "tags": [
      "动作特效",
      "写实电影感",
      "科幻太空"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083852804079390858",
    "title": "霓虹动漫乐队歌词视频",
    "titleEn": "A Neon Anime Band Lyric Video",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "动漫歌手与乐器镜头按顺序变化，大字歌词融入画面。",
    "summaryEn": "Anime singers and instrument shots change in sequence with large lyrics integrated into the frames.",
    "prompt": "Morph  <img ref> in sequence to match the song in  <audio ref> . The characters move cheerfully and naturally, with cute motions. Animate the lyrics as motion text too",
    "sourceUrl": "https://x.com/impalementd/status/2083852804079390858",
    "sourceLabel": "X 原帖",
    "author": "impalementd",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083852804079390858.mp4",
    "posterUrl": "/posters/x/2083852804079390858.jpg",
    "duration": 43.189116,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/impalementd/status/2083852804079390858"
  },
  {
    "id": "x-2083837233585344537",
    "title": "雨夜巷战的近身格斗",
    "titleEn": "Close-Quarters Combat in a Rainy Alley",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在湿润后巷以格挡与关节控制应对多位攻击者。",
    "summaryEn": "A fighter uses blocks and joint control against several attackers in a wet back alley.",
    "prompt": null,
    "sourceUrl": "https://x.com/yourPlugAI/status/2083837233585344537",
    "sourceLabel": "X 原帖",
    "author": "yourPlugAI",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083837233585344537.mp4",
    "posterUrl": "/posters/x/2083837233585344537.jpg",
    "duration": 15.13941,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "动作打斗",
      "城市街景"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083737476620013725",
    "title": "水上障碍第一视角三模型对比",
    "titleEn": "A Three-Model First-Person Water-Obstacle Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "并排标注的三组画面比较第一视角穿越水上障碍的结果。",
    "summaryEn": "Three labeled views compare first-person movement through water obstacles.",
    "prompt": null,
    "sourceUrl": "https://x.com/genel_ai/status/2083737476620013725",
    "sourceLabel": "X 原帖",
    "author": "genel_ai",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083737476620013725.mp4",
    "posterUrl": "/posters/x/2083737476620013725.jpg",
    "duration": 14.826667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "3D / CG",
      "人物表演"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083716093047410703",
    "title": "分镜配合 MP3 参考的动作比较",
    "titleEn": "A Storyboard and MP3 Reference Motion Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "分镜与两种生成结果并排展示，比较加入音频参考后的角色动作。",
    "summaryEn": "Storyboards and two results appear together to compare character movement with an audio reference.",
    "prompt": null,
    "sourceUrl": "https://x.com/ponzponz15/status/2083716093047410703",
    "sourceLabel": "X 原帖",
    "author": "ponzponz15",
    "publishedAt": "2026-08-02",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083716093047410703.mp4",
    "posterUrl": "/posters/x/2083716093047410703.jpg",
    "duration": 15.146667,
    "aspectRatio": "4:3",
    "resolution": "720p",
    "tags": [
      "本地生成与工作流",
      "照片级写实",
      "人物表演"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "audio"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083637290652623341",
    "title": "复古奇幻战士与机器人对白",
    "titleEn": "A Retro Fantasy Conversation with a Warrior and Robot",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "战士、人物与机械机器人在夜色中互动；本站播放原帖的第二段 H3 视频。",
    "summaryEn": "A warrior, woman and mechanical robot interact at night; this is the source's second video, identified as H3.",
    "prompt": "Use the supplied image as the exact opening frame. Preserve the blonde woman, armored male soldier, and practical robot assistant with identical faces, costumes, proportions, metallic materials, lighting, and spatial positions.\n\nCreate an authentic late-1990s live-action syndicated fantasy action-adventure television scene, photographed on 35mm film and transferred through period telecine. Gritty film grain, slight gate weave, practical torchlight, cool moonlight, deep nighttime shadows, handmade armor, physical robot mechanisms, real locations, miniature armies, matte paintings, forced perspective, and restrained optical compositing. No modern CGI, no digital creatures, no costume redesigns, no face changes, no morphing, no extra foreground characters.\n\n00:00–00:03.8 — Shocked question\n\nBegin in the existing tight three-character composition.\n\nThe chrome-and-gold armored male soldier stares at the blonde woman in disbelief. His eyes widen, his jaw tightens, and he leans slightly toward her. His helmet and armor catch the flickering orange torchlight.\n\nWith urgent, shocked delivery and accurate lip synchronization, he says:\n\nARMORED SOLDIER:\n“Are you saying Seedance and Minimax were only the first wave?”\n\nThe woman holds intense eye contact. The robot remains alert beside him, making subtle mechanical head movements.\n\n00:03.8–00:07.0 \nWoman’s urgent warning\n\nCut to an intense close-up of the blonde woman’s face. She is sweaty, breathless, frightened but determined. Wind moves a few strands of her hair. Torchlight and blue moonlight shape her face.\n\nShe answers urgently and passionately, with precise lip synchronization:\n\nWOMAN:\n“Yes! There are countless more AI models approaching!”\n\nShe immediately turns her head toward something beyond the camp and gestures sharply for the soldier and robot to follow her.\n\n00:07.0–00:10.2 Rush toward the cliff\n\nA fast handheld tracking shot follows all three as they hurry several steps toward the edge of a rocky cliff.\n\nThe woman leads. The armored soldier follows closely, still stunned. The practical robot assistant stomps behind them with heavy mechanical movement, pistons compressing and metal joints rotating.\n\nThe woman reaches the cliff edge and points forcefully into the distance.\n\n00:10.2–00:13.2 Reveal the approaching armies\n\nCut to a dramatic wide shot from behind the three heroes as they stare across a vast nighttime valley.\n\nIn the distance, reveal multiple visibly different armies of robots approaching in organized formations:\n\n- tall chrome humanoid soldiers carrying illuminated staffs\n- squat industrial robots with rotating sensor heads\n- skeletal brass automatons marching in ranks\n- enormous tracked mechanical siege machines\n- distant flying mechanical scouts crossing the moonlit sky\n\nThe armies must look like physical miniatures, full-scale robot suits, animatronics, forced-perspective formations, matte-painted extensions, and late-1990s optical effects—not modern digital animation.\n\nHundreds of tiny mechanical lights advance through fog and dust. The ground vibrates faintly beneath the heroes.\n\n00:13.2–00:15.0 Defiant reaction\n\nCut to a close three-shot.\n\nThe armored soldier looks overwhelmed. The woman remains fiercely focused. The robot assistant rotates its optical lenses, locks its mechanical arms into combat position, and produces a loud hydraulic charging sound.\n\nThe robot leans forward confidently and says in a rough, humorous synthesized voice:\n\nROBOT:\n“Bring them on.”\n\nEnd on the three heroes facing the approaching armies as dramatic orchestral action music surges.\n\nAudio: Period fantasy-action orchestral score, distant mechanical marching, wind, torch crackle, metallic armor movement, hydraulic robot sounds, and clean dialogue.\n\nPerformance: Serious stakes with a restrained pulpy syndicated-TV energy.",
    "sourceUrl": "https://x.com/BrentLynch/status/2083637290652623341",
    "sourceLabel": "X 原帖",
    "author": "BrentLynch",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083637290652623341.mp4",
    "posterUrl": "/posters/x/2083637290652623341.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "复古胶片",
      "照片级写实",
      "奇幻"
    ],
    "category": "dialogue",
    "styles": [
      "retro",
      "photoreal"
    ],
    "scenes": [
      "fantasy",
      "dialogue"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/BrentLynch/status/2083637290652623341"
  },
  {
    "id": "x-2083603655383527793",
    "title": "奇幻角色对白的六模型合集",
    "titleEn": "A Six-Model Fantasy Dialogue Comparison",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "原片按标签依次播放六个模型对同一奇幻角色场景的诠释。",
    "summaryEn": "The source plays six labeled model interpretations of the same fantasy character scene.",
    "prompt": null,
    "sourceUrl": "https://x.com/ozansihay/status/2083603655383527793",
    "sourceLabel": "X 原帖",
    "author": "ozansihay",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083603655383527793.mp4",
    "posterUrl": "/posters/x/2083603655383527793.jpg",
    "duration": 92.2761,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "模型对比",
      "写实电影感",
      "奇幻",
      "对白与口型"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083555429758464203",
    "title": "蒙娜丽莎进入可玩角色界面",
    "titleEn": "Mona Lisa Enters a Playable-Character Interface",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在菜单中换装与选择武器，随后进入第三人称街道。",
    "summaryEn": "A character changes clothes and selects weapons in a menu before entering a third-person street scene.",
    "prompt": null,
    "sourceUrl": "https://x.com/ivanka_humeniuk/status/2083555429758464203",
    "sourceLabel": "X 原帖",
    "author": "ivanka_humeniuk",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083555429758464203.mp4",
    "posterUrl": "/posters/x/2083555429758464203.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "3D / CG",
      "奇幻"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083552263449035223",
    "title": "咖啡环境文字概念广告",
    "titleEn": "A Coffee Concept Ad with Environmental Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "咖啡包装、倒液与杯中立体文字展示虚构品牌 AURELIA。",
    "summaryEn": "Coffee packaging, pouring and dimensional letters in a cup present the fictional AURELIA brand.",
    "prompt": null,
    "sourceUrl": "https://x.com/AmControo/status/2083552263449035223",
    "sourceLabel": "X 原帖",
    "author": "AmControo",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083552263449035223.mp4",
    "posterUrl": "/posters/x/2083552263449035223.jpg",
    "duration": 13.188934,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083534561443532807",
    "title": "公寓门前的韩剧式对峙",
    "titleEn": "A Korean-Drama-Style Confrontation at an Apartment",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "两人在公寓内对峙，双人构图切换为各自近景。",
    "summaryEn": "Two people confront each other in an apartment, moving from a two-shot to close-ups.",
    "prompt": "REFERENCE USE\nImages 1 and 2 lock face, hair and wardrobe identity only. Do not copy their frontal pose, centred framing or lens gaze. Image 3 locks apartment geometry, light direction and materials only. Recompose the actors according to the screen directions below. Identities remain stable.\n\nIDENTITY LOCKS\nYEON-SU: Korean woman, 28. Loose unstyled dark hair tucked behind her anatomical right ear only; left ear covered. Oversized heather-grey knit, sleeves past both wrists, loose charcoal lounge trousers, bare feet, no jewellery.\nJAE-HYUN: Korean man, 32. Dark knee-length coat stays on. Plain shirt open by exactly one button. Watch on anatomical left wrist. Keys handled only by anatomical right hand.\nNever mirror these details. No wardrobe change.\n\nSCENE\nThree years into a relationship. The argument peaked before this clip. Neither is trying to win. YEON-SU needs him to deny he is leaving but cannot ask directly. JAE-HYUN has waited to be actively chosen and is ashamed that he needed it. His answer is an admission, not an attack. Her last line is not a reply; she discovers that she treated his staying as permanent while he experienced it as a choice she never made.\n\nDIALOGUE\nSpoken Korean only. Contemporary Seoul banmal between a long-term couple, with dropped subjects, intimate volume and natural consonants. Never translate, speak, subtitle or display English.\n\nYEON-SU: \"또 갈 거야?\"\nJAE-HYUN: \"한 번쯤은... 가지 말라고 할 줄 알았어.\"\nYEON-SU: \"안 갈 줄 알았어.\"\n\nSCREEN GEOGRAPHY\nJAE-HYUN is frame-left facing screen-right. YEON-SU is frame-right facing screen-left. Singles leave look-room toward the unseen partner. Matched off-camera eyelines; never look into the lens. In the wide they are three metres apart. The table stays on YEON-SU's anatomical right so her right hand can hold it. The entry shelf stays beside JAE-HYUN's right hand. Fixed marks, but natural breathing, a swallow, finger pressure and tiny posture settling are allowed. No walking, sitting, approaching or contact.\n\nSHOT LIST\n[0.0-2.8] LOCKED WIDE TWO-SHOT, 35mm, eye level. Entry frame-left, living room frame-right. JAE-HYUN stands just inside the door, coat on. YEON-SU stands beside the table three metres away, right hand lightly holding its edge. She watches him; he looks at the shelf. At 1.3s he places the keys on wood with his right hand, quietly and deliberately. Neither speaks. Hold the distance.\n\n[2.8-5.4] YEON-SU MEDIUM CLOSE, 75mm, three-quarter view, mid-chest to head. She is frame-right with look-room screen-left. Begin neutral, without prepared sadness. One ordinary shallow breath. Between 3.7s and 4.6s she asks, \"또 갈 거야?\" Low, plain, almost practical. She wants him to say no. Brows stay relaxed; mouth closes after the line. She listens.\n\n[5.4-10.3] JAE-HYUN MEDIUM CLOSE, 75mm, three-quarter view, mid-chest to head. He is frame-left with look-room screen-right. Mouth fully closed during the silence. Eyes down; one controlled breath and one swallow. This is active decision, not blank posing. He raises his eyes only when ready to admit it. Between 7.9s and 9.6s: \"한 번쯤은... 가지 말라고 할 줄 알았어.\" Quiet, warm, tired. The pause after \"한 번쯤은\" is a failed attempt to make the sentence easier. No blame, emphasis or sigh. He keeps his eyeline toward her afterward.\n\n[10.3-15.0] YEON-SU CLOSE, 85mm, collarbones to crown, three-quarter view, frame-right with look-room screen-left. One continuous take with no reframing or hidden cut. She is already looking down. Begin without a hurt expression. Recognition forms through stopped breath, a released jaw and eyes focusing on nothing. No tears. Between 12.7s and 13.9s she says to the floor, \"안 갈 줄 알았어.\" Nearly matter-of-fact at first; the final word softens when she hears her own assumption. Only after the last syllable does she lift her eyes toward him. Mouth closed. Hold her listening face through 15.0s.\n\nACTING\nLived-in screen acting, never presented emotion. Each actor actively listens to the absent partner. Emotion appears through attention, breath and thought, not facial display. Allow imperfect human timing inside the dialogue windows. No added theatrical pauses or mirrored performances.\n\nLIGHT AND IMAGE\nOne low warm living-room practical, about 2900K. Weak cold night spill from the dark entry, not saturated cyan. Faces partly in shadow; natural Korean skin tone, pores and under-eye texture. Desaturated wood, grey knit and blue-black exterior; restrained grain. The lamp motivates the warm face light. No commercial orange-teal grade, bloom, halation, glow or flare.\n\nCAMERA\nLocked and eye-level. Maintain the 180-degree line. Singles are three-quarter views, never centred frontal portraits. Shallow depth of field with both eyes readable. No drift, push-in, handheld sway, arc, rack focus, dutch angle, motion blur or lens-scale change within a shot.\n\nPRODUCTION SOUND\nNatural Korean production dialogue recorded in the apartment, not narration, dubbing or synthetic TTS. Intimate conversational level, slight room reflection, matched boom-and-lav perspective. Continuous stereo room tone beneath every line and silence: refrigerator hum, low HVAC and distant traffic through closed glass. Never hard-gate the room. Keys on wood at 1.3s are the only featured effect. No phone buzz or designed silence drop.\nNo music, score, drone or tonal swell. Dialogue centred; ambience has subtle width. No breath boosting, over-compression or loud social-video mastering. If controllable: 48 kHz stereo, true peak at or below -6 dBFS.\n\nNEGATIVES\nNo English voice, subtitles, captions, text, logos or watermarks. No lens gaze, centred portrait composition, silent mouthing, speech outside the three lines or bad lip sync. No shouting, sarcasm, crying, tears, sobbing, gasping, widened eyes, knitted brows, trembling lips, clenched fists, pointing, arm crossing or hand gestures. No extra people or reflections, wardrobe drift, side mirroring, beauty filter, face warp or identity change. No extra cuts, jump cuts, punch-ins or framing resets, especially from 10.3s to 15.0s.",
    "sourceUrl": "https://x.com/IamEmily2050/status/2083534561443532807",
    "sourceLabel": "X 原帖",
    "author": "IamEmily2050",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083534561443532807.mp4",
    "posterUrl": "/posters/x/2083534561443532807.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "写实电影感"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/IamEmily2050/status/2083534561443532807"
  },
  {
    "id": "x-2083531062697664672",
    "title": "火光前的铠甲人物特写",
    "titleEn": "An Armored Character Close-up before Firelight",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "深色铠甲人物转动视线，金色火光勾勒脸部与肩甲。",
    "summaryEn": "A dark-armored character shifts her gaze as golden firelight outlines her face and shoulder armor.",
    "prompt": null,
    "sourceUrl": "https://x.com/nusratsid1/status/2083531062697664672",
    "sourceLabel": "X 原帖",
    "author": "nusratsid1",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083531062697664672.mp4",
    "posterUrl": "/posters/x/2083531062697664672.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "奇幻",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083516916782510593",
    "title": "倒塌机甲接住飞行员",
    "titleEn": "A Collapsing Mecha Catches Its Pilot",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "受损机甲倒下时伸出机械手，在烟尘中保护飞行员。",
    "summaryEn": "A damaged mecha extends its hand while collapsing to protect its pilot amid smoke.",
    "prompt": "Hyper-realistic cinematic sci-fi action sequence, 15 seconds, aspect ratio 16:9.\n\nInside a giant futuristic combat hangar, one badly damaged mech stumbles backward after battle. One pilot only. The mech is huge, smoking, sparking, and barely functioning. Its chest armor is cracked, one shoulder is on fire, and warning lights flash across the hangar. The goal is simple: the pilot must survive the mech’s collapse.\n\nThe action starts instantly. The mech loses balance and falls backward toward the hangar floor. The cockpit blows open and the pilot is violently thrown out into open air, spinning and falling away from the mech. For a moment it looks like the pilot will hit the ground.\n\nThen the set piece changes. The mech’s still-active giant hand suddenly moves on its own, reaches out, and catches the pilot mid-fall. The hand closes just enough to protect the pilot without crushing them. The mech is still collapsing, dragging the pilot in its grip as sparks, smoke, and broken armor tear away around them.\n\nThe falling mech crashes through support rails and slams into the hangar floor or wall. The impact is huge. Metal buckles. Debris flies. But the pilot survives because the mech’s giant hand shields them from the crash. The hand skids across the floor with the pilot still inside its grip.\n\nAt the end, the wrecked mech comes to a stop in smoke and sparks. The giant hand slowly opens, revealing the pilot alive inside, surrounded by twisted metal and burning debris.\n\nStyle: hyper-realistic, cinematic, intense, clear readable action, one pilot only, one damaged mech only, one hangar only, one main set piece, collapsing mech, pilot thrown from cockpit, giant mech hand catching pilot mid-fall, heavy crash impact, sparks, smoke, debris, no extra battle scene, no extra characters, no text, no logos, no cartoon style, no slow motion, no famous celebrity faces, no recognizable actors, no movie-star resemblance, no public-figure likenesses, no clear facial close-ups. Keep proportions. Keep style and features. Aspect ratio 16:9.",
    "sourceUrl": "https://x.com/DeCat2025/status/2083516916782510593",
    "sourceLabel": "X 原帖",
    "author": "DeCat2025",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083516916782510593.mp4",
    "posterUrl": "/posters/x/2083516916782510593.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "科幻太空"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/DeCat2025/status/2083516916782510593"
  },
  {
    "id": "x-2083432403364720877",
    "title": "洗衣机开箱安装 UGC 概念",
    "titleEn": "A Washing Machine Unboxing UGC Concept",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物移动纸箱、处理机身配件并展示安装后的洗衣区域。",
    "summaryEn": "A presenter moves a box, handles machine fittings and shows the completed laundry space.",
    "prompt": null,
    "sourceUrl": "https://x.com/yourPlugAI/status/2083432403364720877",
    "sourceLabel": "X 原帖",
    "author": "yourPlugAI",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083432403364720877.mp4",
    "posterUrl": "/posters/x/2083432403364720877.jpg",
    "duration": 15.23229,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "生活实拍感",
      "照片级写实",
      "产品与广告",
      "日常生活"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "product",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083374435227386222",
    "title": "粉色手袋商品页动态图形",
    "titleEn": "A Pink Handbag Product-Page Motion Design",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "商品页面、手袋特写与时装人物拼贴以广告版式交替。",
    "summaryEn": "Product pages, handbag close-ups and fashion cutouts alternate in advertising layouts.",
    "prompt": null,
    "sourceUrl": "https://x.com/madpencil_/status/2083374435227386222",
    "sourceLabel": "X 原帖",
    "author": "madpencil_",
    "publishedAt": "2026-08-01",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083374435227386222.mp4",
    "posterUrl": "/posters/x/2083374435227386222.jpg",
    "duration": 25.40263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "动态图形",
      "时尚",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083267196625535159",
    "title": "爵士节奏与唱片动漫片头",
    "titleEn": "An Anime Opening with Jazz and Record Graphics",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "唱片、人物卡与霓虹拼贴构成音乐片头。",
    "summaryEn": "Records, character cards and neon collage form a music-driven opening.",
    "prompt": null,
    "sourceUrl": "https://x.com/fal/status/2083267196625535159",
    "sourceLabel": "X 原帖",
    "author": "fal",
    "publishedAt": "2026-07-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083267196625535159.mp4",
    "posterUrl": "/posters/x/2083267196625535159.jpg",
    "duration": 19.156463,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "音乐演出"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083244369079840826",
    "title": "海滨七人群像电影片头",
    "titleEn": "A Seaside Seven-Person Ensemble Opening",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "海景餐厅、人物细节与七人并排镜头组成电影式开场。",
    "summaryEn": "A seaside restaurant, character details and a seven-person lineup form a cinematic opening.",
    "prompt": null,
    "sourceUrl": "https://x.com/altphotos_pl/status/2083244369079840826",
    "sourceLabel": "X 原帖",
    "author": "altphotos_pl",
    "publishedAt": "2026-07-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083244369079840826.mp4",
    "posterUrl": "/posters/x/2083244369079840826.jpg",
    "duration": 46.058667,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2083128868668666020",
    "title": "银蓝机甲新赛季概念片",
    "titleEn": "A Silver-Blue Mecha Season Concept",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "机甲在未来机库中亮起，技能界面与解锁标语围绕机体展开。",
    "summaryEn": "A mecha lights up in a futuristic hangar as skill UI and unlock text appear around it.",
    "prompt": "生成一支15秒16:9横版科幻机甲手游新赛季商业广告。\n虚构游戏品牌 MECHA FRONT，主角是银白与深蓝配色的限定机甲 AETHER-07，机体外形、装甲结构和配色在全片保持一致。\n画面只允许出现以下文字，且必须拼写完全正确，MECHA FRONT、AETHER-07、限定机甲、攻击 920、机动 88、现在解锁。  \n0至4秒，\n黑色未来机库中，一束冷白顶光依次扫过机甲头部、胸甲与机械关节，镜头从低机位快速推进，完整机体从黑暗中亮起，MECHA FRONT 与 AETHER-07 清晰出现。  \n4至10秒，\n机甲在原地缓慢转身，半透明蓝色游戏 UI 围绕机体依次展开。左侧显示限定机甲，右侧准确显示攻击 920 与机动 88，三枚无文字技能图标顺序点亮。UI 不能遮挡机甲主体，机甲造型不得变化。  \n10至15秒，\nUI 卡片沿轨迹收拢，机甲完成半周转身后正面定格。背景形成蓝白能量环，结尾只保留 MECHA FRONT、AETHER-07 与现在解锁，所有文字完整可读。 \n 整体采用顶级 3A 游戏 CG 宣传片与新品发布广告质感，冷黑、银白、电光蓝配色，金属材质真实，轮廓光锐利，镜头运动干净有力量。\n不要人物，不要战斗剧情，不要爆炸，不要新增文字，不要乱码，不要水印。原创声音使用低频机械启动、UI 点击和克制的电子鼓点，结尾一记厚重机械锁定音。",
    "sourceUrl": "https://x.com/liyue_ai/status/2083128868668666020",
    "sourceLabel": "X 原帖",
    "author": "liyue_ai",
    "publishedAt": "2026-07-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083128868668666020.mp4",
    "posterUrl": "/posters/x/2083128868668666020.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "科幻太空",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "sci-fi",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/liyue_ai/status/2083128868668666020"
  },
  {
    "id": "x-2083115551186915723",
    "title": "瀑布云海凝聚 ASCEND",
    "titleEn": "Waterfalls and Cloud Mist Form ASCEND",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头冲过热带峭壁与云层，水光汇聚为巨型字。",
    "summaryEn": "The camera rushes through tropical cliffs and clouds before water and light form giant letters.",
    "prompt": "The camera launches from the surface of the ocean and rockets vertically between towering tropical cliffs as enormous waterfalls crash on both sides. It narrowly avoids collapsing rock arches before bursting through clouds into brilliant sunlight. Moisture, mist, and billions of illuminated droplets spiral behind the camera, creating a colossal atmospheric vortex. The liquid cyclone naturally compresses into \"ASCEND\" before exploding into a dazzling halo of water and light. Epic scale, impossible speed, Hollywood blockbuster, IMAX cinematography.",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2083115551186915723",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-07-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083115551186915723.mp4",
    "posterUrl": "/posters/x/2083115551186915723.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动态图形",
      "自然风景"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2083115551186915723"
  },
  {
    "id": "x-2083090308666163365",
    "title": "幻想 RPG 装备界面与角色行动",
    "titleEn": "Fantasy RPG Equipment UI and Character Action",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "界面选择角色与武器，随后切入第三人称幻想城市。",
    "summaryEn": "An interface selects characters and weapons before entering a third-person fantasy city.",
    "prompt": null,
    "sourceUrl": "https://x.com/ShamsAmin56/status/2083090308666163365",
    "sourceLabel": "X 原帖",
    "author": "ShamsAmin56",
    "publishedAt": "2026-07-31",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2083090308666163365.mp4",
    "posterUrl": "/posters/x/2083090308666163365.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "3D / CG",
      "奇幻"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082855352182137170",
    "title": "音频设备与网格的快速硬切",
    "titleEn": "Rapid Hard Cuts of Audio Hardware and Grids",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "青色网格、设备阵列与旋钮界面以快速切换组织科技片段。",
    "summaryEn": "Cyan grids, hardware arrays and control knobs form a rapidly cut technology sequence.",
    "prompt": null,
    "sourceUrl": "https://x.com/NEXUS_TO_NOVA/status/2082855352182137170",
    "sourceLabel": "X 原帖",
    "author": "NEXUS_TO_NOVA",
    "publishedAt": "2026-07-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082855352182137170.mp4",
    "posterUrl": "/posters/x/2082855352182137170.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082848680793747543",
    "title": "暖粉客厅里的棕色手袋广告",
    "titleEn": "A Brown Handbag Ad in a Warm Pink Interior",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物与手袋在暖粉室内交替特写，突出皮革与五金细节。",
    "summaryEn": "A presenter and handbag alternate in a warm pink room, highlighting leather and hardware.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIwithkhan/status/2082848680793747543",
    "sourceLabel": "X 原帖",
    "author": "AIwithkhan",
    "publishedAt": "2026-07-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082848680793747543.mp4",
    "posterUrl": "/posters/x/2082848680793747543.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082819036182917160",
    "title": "雨夜战术小队行动预告",
    "titleEn": "A Tactical Team Teaser on a Rainy Night",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "战术车辆停靠，队员下车并进入建筑，形成虚构行动预告。",
    "summaryEn": "A tactical vehicle stops, its team exits and enters a building in a fictional action teaser.",
    "prompt": "15-second cinematic Hollywood-style teaser. Nighttime in Washington, D.C.. Rain falls as a black tactical SUV stops near a government building. Four elite intelligence agents in realistic tactical gear silently exit the vehicle. A drone scans the area while helicopters circle in the distance. One agent receives a classified message on a holographic device: \"Target Located.\" The team breaches the building with precision. Fast cinematic cuts, dramatic lighting, realistic camera movement, high tension, ultra-realistic, blockbuster movie trailer, anamorphic lens, volumetric lighting, Dolby-style sound design, no logos, no text, original fictional story.",
    "sourceUrl": "https://x.com/navi_Ai2/status/2082819036182917160",
    "sourceLabel": "X 原帖",
    "author": "navi_Ai2",
    "publishedAt": "2026-07-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082819036182917160.mp4",
    "posterUrl": "/posters/x/2082819036182917160.jpg",
    "duration": 15.13941,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "写实电影感",
      "动作打斗",
      "城市街景"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/navi_Ai2/status/2082819036182917160"
  },
  {
    "id": "x-2082778573899710642",
    "title": "巨浪与高墙围绕的天空王国",
    "titleEn": "A Sky Kingdom amid Giant Waves and Walls",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "船只穿过高墙水道，巨浪与城堡构成大尺度奇幻景观。",
    "summaryEn": "A vessel passes high-walled waterways as giant waves and a castle form a vast fantasy landscape.",
    "prompt": null,
    "sourceUrl": "https://x.com/MO_IAI/status/2082778573899710642",
    "sourceLabel": "X 原帖",
    "author": "MO_IAI",
    "publishedAt": "2026-07-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082778573899710642.mp4",
    "posterUrl": "/posters/x/2082778573899710642.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082775364577693905",
    "title": "冻结雨滴中的双人游戏格斗",
    "titleEn": "A Two-Fighter Game Battle amid Frozen Rain",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红黑角色在石质场地交锋，环绕镜头与能量效果强调动作。",
    "summaryEn": "Red-and-black fighters clash on a stone arena with orbiting views and energy effects.",
    "prompt": null,
    "sourceUrl": "https://x.com/VigoCreativeAI/status/2082775364577693905",
    "sourceLabel": "X 原帖",
    "author": "VigoCreativeAI",
    "publishedAt": "2026-07-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082775364577693905.mp4",
    "posterUrl": "/posters/x/2082775364577693905.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动作打斗"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082662622944727510",
    "title": "双刃人物与异形生物的动作片",
    "titleEn": "A Dual-Blade Fighter against an Alien Creature",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "持双刃人物对抗异形生物，低角度镜头突出武器与姿态。",
    "summaryEn": "A dual-blade fighter faces an alien creature with low angles emphasizing weapons and poses.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cia0_exe/status/2082662622944727510",
    "sourceLabel": "X 原帖",
    "author": "Cia0_exe",
    "publishedAt": "2026-07-30",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082662622944727510.mp4",
    "posterUrl": "/posters/x/2082662622944727510.jpg",
    "duration": 15.25551,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "风格化与实验",
      "3D / CG",
      "动作打斗",
      "科幻太空"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized",
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082564720603623680",
    "title": "战场烟尘中的克制对白",
    "titleEn": "A Restrained Conversation in Battlefield Smoke",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "两名人物在残骸与金色烟尘间交流，镜头靠近情绪反应。",
    "summaryEn": "Two figures speak amid wreckage and golden smoke as the camera moves closer to their reactions.",
    "prompt": null,
    "sourceUrl": "https://x.com/sebatheepan/status/2082564720603623680",
    "sourceLabel": "X 原帖",
    "author": "sebatheepan",
    "publishedAt": "2026-07-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082564720603623680.mp4",
    "posterUrl": "/posters/x/2082564720603623680.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "写实电影感",
      "动作打斗"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082561712914567207",
    "title": "演唱会中与观众互动的歌手",
    "titleEn": "A Concert Singer Interacts with the Crowd",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "歌手在舞台边缘演唱、笑着招手并与观众互动。",
    "summaryEn": "A singer performs at the stage edge, smiling, gesturing and interacting with the audience.",
    "prompt": "Style: Real Concert Documentary Film, handheld camera energy, practical stage lighting, gritty realism, shot on Sony A7S3, 35mm lens, high energy modern live concert performance. Duration: 15s. \n\n[Global Subject & References]: Subject is a naturally confident, charismatic 25-year-old woman based on  as character and lighting reference. Her face, expression, and wavy dark brown hair are matched to image1 but her costume is updated to a black satin lace slip dress and a delicate gold necklace. She is performing a sold-out arena concert.\n\n [00-01s] Shot 1: Tight close-up on her face mid-performance. Action: Eyes bright, a genuine laugh breaks through her lyric, dark hair catching the stage light. Camera: Handheld documentary camera, tight frame with natural focus and slight organic movement. Lighting: Saturated practical stage lighting from creating hair highlights. \n\n[01-06s] Shot 2: Medium shot of her stepping forward. Action: She steps forward to the edge of the stage, singing into her microphone with playful, high-energy vibes. Her expression shifts as she frowns in playful, mock-sadness. Dialogue Cue: \"Wait! Is this Minimax H3?\" Action continued: She then smiles, laughs, and reaches her free hand toward the crowd, her fingers brushing toward raised hands and phones. The crowd sings \"Minimax H3\" back to her. Camera: Gentle handheld camera sway following her forward motion, natural focus pull. Lighting: Warm stage spotlights and natural lens flares in the background haze. \n\n[06-09s] Shot 3: Wide shot from a low angle near the crowd barrier. Action: She is framed against stage haze and colorful spotlights, throwing her free arm out wide and laughing between lines. Her hair whips slightly as she turns. A keyboard player is visible in soft focus stage right (matching the placement in @ Image1), and a guitarist is visible stage left. Camera: Low-angle wide shot, realistic handheld camera shake from the pit, no gimbal smoothing. Lighting: Saturated purple and blue stage lights cutting through the atmosphere. \n\n[10-15s] Shot 4 & Outro: Closer three-quarter shot. Action: She playfully points her microphone toward the crowd for a sing-along, grinning and scanning the audience with genuine warmth and connection. She has a visible sweat sheen and natural skin texture. For the final beat, she looks directly into the lens with a bright, genuine smile. Camera: Handheld camera with organic focus pulling and slight shake. Lighting: Practical stage lights highlighting her face and shoulders. \n\n[Global Audio]: Live arena ambience, crowd roar, cheering, phone-camera flash pops, her voice carrying clearly over a live band mix of electric guitar, keys, drums, and bass, with natural live-mix dynamics and room reverb.",
    "sourceUrl": "https://x.com/BrentLynch/status/2082561712914567207",
    "sourceLabel": "X 原帖",
    "author": "BrentLynch",
    "publishedAt": "2026-07-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082561712914567207.mp4",
    "posterUrl": "/posters/x/2082561712914567207.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "音乐演出"
    ],
    "category": "music",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/BrentLynch/status/2082561712914567207"
  },
  {
    "id": "x-2082559705142936039",
    "title": "生活与动作特效的五主题测试",
    "titleEn": "A Five-Theme Test of Lifestyle and Action Effects",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "海景室内、动作人物与建筑破坏镜头组成多主题生成合集。",
    "summaryEn": "A seaside interior, action figure and building destruction form a multi-theme generation reel.",
    "prompt": null,
    "sourceUrl": "https://x.com/techhalla/status/2082559705142936039",
    "sourceLabel": "X 原帖",
    "author": "techhalla",
    "publishedAt": "2026-07-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082559705142936039.mp4",
    "posterUrl": "/posters/x/2082559705142936039.jpg",
    "duration": 75.477333,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "写实电影感",
      "人物表演",
      "城市街景"
    ],
    "category": "showcase",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082533116732674155",
    "title": "一口果汁让荒地变绿洲",
    "titleEn": "A Sip of Juice Turns Dry Land into an Oasis",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物饮用果汁后，干热场景转为绿色植物与清凉产品画面。",
    "summaryEn": "After a character drinks juice, dry heat changes into greenery and refreshing product imagery.",
    "prompt": "A young person walks under the blazing summer sun, looking exhausted and sweating heavily. The road shimmers with heat waves, and everything appears dry and dull. Suddenly, they grab a chilled bottle of premium fruit juice from a cooler and take a refreshing sip. Instantly, the environment transforms—lush green trees bloom, vibrant flowers appear, a cool breeze flows, water splashes through the air, and glowing particles surround the scene. Ice cubes and fresh fruit slices (orange, mango, or according to the flavour) swirl around the bottle in cinematic slow motion. End with a stunning close-up of the juice bottle covered in cold water droplets against a bright, refreshing background. Ultra-realistic, premium commercial, 4K, cinematic lighting, high-detail, smooth camera movements, vibrant colours, luxury beverage advertisement.\nTagline ideas:\nBeat the Heat. Taste the Freshness.\nEvery Sip Brings Life.\nRefresh Your Day, Naturally.\nStay Cool. Stay Fresh.",
    "sourceUrl": "https://x.com/navi_Ai2/status/2082533116732674155",
    "sourceLabel": "X 原帖",
    "author": "navi_Ai2",
    "publishedAt": "2026-07-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082533116732674155.mp4",
    "posterUrl": "/posters/x/2082533116732674155.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "美食",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/navi_Ai2/status/2082533116732674155"
  },
  {
    "id": "x-2082529099646722299",
    "title": "复古控制室里的人物与机器人喜剧",
    "titleEn": "A Character and Robot Comedy in a Retro Control Room",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物读信并与机械机器人互动，模拟八十年代科幻喜剧。",
    "summaryEn": "A woman reads a message and interacts with a practical-style robot in an eighties sci-fi comedy scene.",
    "prompt": "LATE-1980s SCI-FI COMEDY \nUse the supplied image as the exact opening frame. Preserve the same woman, practical robot, costumes, lighting, physical control-room set, and camera-facing geography. Photographed as a real late-1980s science-fiction film using practical props, animatronics, analog monitors, tungsten lighting, subtle 35mm grain, gentle gate weave, natural skin texture, and authentic period color science. No CGI appearance, no modern screens, no character changes, no extra people.\n\n0:00–0:03 — Medium close-up. The excited young woman holds the message and reads quickly in an energetic 1980s California Valley Girl voice:\n\nWOMAN: “Minimax H3! Up to fifteen seconds, two-K, twelve reference images—”\n\nThe practical robot behind her powers up, turns its head toward her, and begins flashing its indicator lights.\n\n0:03–0:07 — Slow cinematic push-in. She continues, becoming increasingly amazed:\n\nWOMAN: “Audio and video… and they gave us immediate access!”\n\nThe robot’s mechanical arms lift with excitement. It bounces slightly on its hydraulic legs.\n\nROBOT, funny synthesized voice: “IMMEDIATE ACCESS CONFIRMED!”\n\n0:07–0:10 — She reads the final line, almost unable to believe it:\n\nWOMAN: “Enjoy, you cool kid, you!”\n\nShe looks directly at the robot. They freeze for one delighted beat, then simultaneously erupt with joy.\n\n0:10–0:14 — The woman jumps up and dances beside the robot. The robot performs an awkward celebratory mechanical dance, pumping its arms and bouncing safely in place.\n\nWOMAN AND ROBOT: “YES! YES! YES!”\n\nEnd on both celebrating wildly as analog lights flash around them.\n\nAUDIO: Energetic retro synthesizer score, paper rustling, servo motors, hydraulic clicks, analog computer beeps, joyful shouting, and room ambience. Dialogue must remain clear and perfectly synchronized.\n\nCAMERA: One continuous cinematic shot with a subtle push-in followed by a slight handheld adjustment during the celebration. Natural motion, realistic physics",
    "sourceUrl": "https://x.com/BrentLynch/status/2082529099646722299",
    "sourceLabel": "X 原帖",
    "author": "BrentLynch",
    "publishedAt": "2026-07-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082529099646722299.mp4",
    "posterUrl": "/posters/x/2082529099646722299.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "对白与口型",
      "复古胶片",
      "照片级写实",
      "科幻太空"
    ],
    "category": "dialogue",
    "styles": [
      "retro",
      "photoreal"
    ],
    "scenes": [
      "sci-fi",
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/BrentLynch/status/2082529099646722299"
  },
  {
    "id": "x-2082502202539810991",
    "title": "雪中木屋与壁炉人物片段",
    "titleEn": "A Snowy Cabin and Fireside Character Scene",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头从雪景窗边移到暖色壁炉，人物在火光前微笑。",
    "summaryEn": "The camera moves from a snowy window view to a warm fireplace and a smiling character.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2082502202539810991",
    "sourceLabel": "X 原帖",
    "author": "Strength04_X",
    "publishedAt": "2026-07-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082502202539810991.mp4",
    "posterUrl": "/posters/x/2082502202539810991.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "日常生活",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "daily-life",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2082499539735588916",
    "title": "悬崖城市飞行载具追逐",
    "titleEn": "A Speeder Chase through a Cliffside City",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头追随飞行载具穿过石城窄道，最终展开瀑布与彩虹峡谷。",
    "summaryEn": "The camera follows a speeder through narrow stone-city passages before revealing a waterfall and rainbow canyon.",
    "prompt": "Speeder chase across a cliff city (single continuous shot) From a monumental cliffside city carved into stone, the camera dives toward a tiny streak of light ripping along a narrow ledge-road. Lock-on: a speeder hugging the wall at insane speed. The camera slingshots ahead, whips back, then drops tight to the rear thrusters: heat haze, grit snapping off the ledge, warning lights flashing. A collapsing balcony rains debris; the rider snaps a last-inch swerve under a falling arch, then threads through hanging laundry lines and open windows in one fluid line. The camera darts through the same openings, staying glued to the motion. One final bend and sudden calm: the camera blasts outward into a reveal of the city opening onto a boundless waterfall-fed valley, mist turning into rainbow.",
    "sourceUrl": "https://x.com/umesh_ai/status/2082499539735588916",
    "sourceLabel": "X 原帖",
    "author": "umesh_ai",
    "publishedAt": "2026-07-29",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2082499539735588916.mp4",
    "posterUrl": "/posters/x/2082499539735588916.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "科幻太空",
      "城市街景"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "sci-fi",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/umesh_ai/status/2082499539735588916"
  },
  {
    "id": "x-2092522857448562895",
    "title": "滑板动漫角色的 HIPHOP MV",
    "titleEn": "A Hip-Hop MV for a Skateboard Anime Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "戴帽角色在滑板动作与大字画面中切换，片尾展示制作工作流。",
    "summaryEn": "A capped character alternates skateboard moves with bold typography before a workflow view at the end.",
    "prompt": "音楽に完全に同期したHIPHOPダンス。マルチカット、顔のアップで完全なリップシンク、大きなエモーショナルな文字でリリックモーション。背景は歌詞に合わせてあなたが考えて。明るくスタイリッシュな背景。",
    "sourceUrl": "https://x.com/Toshi_nyaruo_AI/status/2092522857448562895",
    "sourceLabel": "X 原帖",
    "author": "Toshi_nyaruo_AI",
    "publishedAt": "2026-08-26",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092522857448562895.mp4",
    "posterUrl": "/posters/x/2092522857448562895.jpg",
    "duration": 186.781315,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Toshi_nyaruo_AI/status/2092522857448562895"
  },
  {
    "id": "x-2092181780447338522",
    "title": "巨大礼物盒里的误会",
    "titleEn": "A Misunderstanding inside a Giant Gift Box",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "男子兴奋迎接礼物，打开后却在盒子里出现意外的人物。",
    "summaryEn": "An excited man receives a gift, only for an unexpected person to appear inside the box.",
    "prompt": "Create a 15-second hilarious cinematic 3D animated comedy with an original warm, polished, high-end cartoon visual style. Use consistent character designs throughout: an overly confident young man and his clever older sister inside a stylish modern apartment. Warm golden lighting, expressive exaggerated facial reactions, smooth animation, cinematic camera work, and fast comedic timing.\n\nScene 1 — 0–3 sec:\nThe young man is relaxing on the sofa, playing a video game. Suddenly, he notices his sister secretly carrying a large beautifully wrapped gift box into the next room. A shiny ribbon falls to the floor. He pauses the game and slowly turns toward the door with wide, suspiciously excited eyes. He immediately assumes the surprise gift must be for him.\n\nScene 2 — 3–6 sec:\nHe jumps up in excitement and begins dramatically practicing his surprised reaction in front of a mirror. He fixes his hair, straightens his shirt, places his hands on his cheeks, and rehearses an exaggerated shocked expression. Then he confidently gives himself a proud smile, convinced everyone is about to celebrate him.\n\nScene 3 — 6–9 sec:\nHe slowly walks toward the room and opens the door with maximum drama, pretending to be casually surprised. He throws his arms open as if expecting everyone to shout, “Surprise!” His excited expression suddenly freezes when the room is completely silent.\n\nScene 4 — 9–12 sec:\nReveal the truth: his sister is standing beside the giant gift box, struggling to pack dozens of his old, messy belongings into it. She has actually bought the box to organize and move his clutter out of the living room. She looks at him and smiles innocently.\n\nScene 5 — 12–15 sec:\nBefore he can react, she rapidly hands him more and more junk to pack—old shoes, tangled headphones, game controllers, clothes, and random items. The box becomes absurdly huge. The young man ends up trapped inside the oversized gift box with only his confused head sticking out. His sister happily closes the lid halfway and pushes the box across the room while he stares at the camera with a defeated expression.\n\nComedy style: fast misunderstanding setup, exaggerated confidence, dramatic anticipation, unexpected reveal, escalating visual chaos, expressive facial animation, smooth comedic timing, cinematic close-ups, warm stylish interior, and a strong visual punchline at the end. Keep all characters and environments visually consistent throughout the full 15 seconds.",
    "sourceUrl": "https://x.com/iamrealsnow/status/2092181780447338522",
    "sourceLabel": "X 原帖",
    "author": "iamrealsnow",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092181780447338522.mp4",
    "posterUrl": "/posters/x/2092181780447338522.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "3D / CG",
      "人物表演",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/iamrealsnow/status/2092181780447338522"
  },
  {
    "id": "x-2092118482129420785",
    "title": "用平面动画讲桃太郎",
    "titleEn": "Momotaro Told through Flat Animation",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "老人、人物和小动物图形搭配大字，组成简洁的民间故事解说。",
    "summaryEn": "Elderly characters, people and small animal graphics combine with large type in a folk-tale explainer.",
    "prompt": "integrated_multimodal_description: Create a complete 15-second Japanese folk-tale explainer in a vertical 9:16 composition, telling the story of Momotaro. Use friendly 2D flat-vector picture-book illustration: thick, even, near-black outlines of consistent weight, flat solid color fills, no shading, simple dot eyes and small curved mouths, restrained cutout animation, and a light paper grain over the flat color fields. Use a compact palette of warm off-white, soft peach pink, deep indigo blue, vermilion red and golden yellow, with near-black outlines. Background color changes mark the chapters of the story. No photorealism, no 3D rendering, no gradients, no drop shadows, no browser interface, no screen-recording artifacts, and no player controls.\n[Shot 1] Begin on a warm off-white background. An old man in a brown work coat carrying a bundle of firewood walks off toward the frame-right edge while an old woman in a blue apron carrying a round wooden washing tub walks off toward the frame-left edge, the two of them separating in one smooth movement. Then the Japanese caption \"むかしむかし\" appears at the top of the frame, sliding in gently word by word.\n[Shot 2] At 00:01.500, a soft peach-pink field wipes across the frame. A large peach with a simple leaf drifts down a stylised river of curved white lines from the top of the frame toward the bottom. The old woman pops in at the lower edge and raises both hands in surprise.\n[Shot 3] At 00:03.000, staying on peach pink, the peach splits cleanly into two halves and a small boy in a red vest sits up between them with his arms raised. The name \"桃太郎\" assembles in large near-black type in two stages beside him.\n[Shot 4] At 00:04.500, a warm off-white field wipes in. The boy grows taller in three clear steps, becoming a young man with a headband. The old man and the old woman pop in on either side of him and smile.\n[Shot 5] At 00:06.000, staying on warm off-white, the old woman holds out a small cloth bundle of millet dumplings and the young man takes it in both hands. He then raises a simple banner on a pole and starts walking toward frame-right.\n[Shot 6] At 00:07.500, staying on warm off-white, a white dog, a brown monkey and a green pheasant pop into the frame one at a time, each with a small bounce. The four of them then line up in a neat row facing frame-right.\n[Shot 7] At 00:09.000, a deep indigo-blue field wipes across the frame. A small wooden boat carrying the four of them rides over rows of simple curved waves from frame-left toward frame-right. The dark silhouette of an island with a gate then rises into view at the far right.\n[Shot 8] At 00:10.500, a vermilion-red field wipes across the frame. A large red ogre with two horns, a spiked club and a tiger-skin loincloth plants himself in front of a heavy gate, filling the centre of the frame. The caption \"鬼ヶ島\" appears above him in near-black type.\n[Shot 9] At 00:12.000, staying on vermilion red, the young man and his three animal companions leap in together from frame-left and the ogre drops backwards onto the ground with his club falling from his hand. A golden-yellow field then wipes across the frame from the bottom.\n[Shot 10] At 00:13.500, on golden yellow, the four of them pull a small cart stacked with treasure chests and coins from frame-left to frame-right, the ogre bowing behind them. The background settles to warm off-white and the closing caption \"めでたし めでたし\" appears in large near-black type, assembling in two stages and holding cleanly until exactly 15.000 seconds.\noverall_soundscape: Soft water plops and a gentle stream accompany the peach as it drifts, and the peach opens with a light crisp pop. Paper-like pops mark each caption and each animal as it enters. Short restrained whooshes carry the color-field wipes, with wooden creaks and lapping waves under the boat. A single deep drum hit lands as the ogre plants himself, followed by a soft thud when he falls and bright chimes as the treasure cart rolls past. No dialogue.\nnon_diegetic_music: Generate an original 15-second Japanese folk-tale jingle using shinobue bamboo flute, plucked shamisen, light taiko drums and small hand bells at a moderate walking tempo. Begin sparse and calm over the off-white opening, brighten as the peach appears, drop low and sparse over the indigo sea, turn urgent with heavier taiko over the red ogre section, resolve into a warm bright melody on the golden treasure section, and finish with one clean final chord at exactly 15.000 seconds.",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2092118482129420785",
    "sourceLabel": "X 原帖",
    "author": "tanabe_fragm",
    "publishedAt": "2026-08-25",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2092118482129420785.mp4",
    "posterUrl": "/posters/x/2092118482129420785.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "插画与手绘",
      "动态图形",
      "动物与生物",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2092118482129420785"
  },
  {
    "id": "x-2091908219962597395",
    "title": "路过咖啡店的一次临时停留",
    "titleEn": "An Unplanned Stop at a Coffee Shop",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子从店外自拍走进咖啡店，离开时举起新买的饮料。",
    "summaryEn": "A woman films herself outside a coffee shop, goes in and leaves holding a drink.",
    "prompt": "SCENE 1 | 00:00–00:02 | THE UNEXPECTED STOP\nWoman is walking past Tim Hortons while filming a casual selfie. She suddenly slows down after noticing someone walk past holding a Caramel Iced Capp. Her eyes follow the drink, then she looks directly into the camera.\nDialog: “Okay… what is THAT?”\n\nSCENE 2 | 00:02–00:03 | I HAVE TO KNOW\n\nShe turns toward the Tim Hortons entrance, still looking at the drink disappearing inside. She gives the camera a playful “I need to investigate” expression.\n\nDialog: “Now I need one.”\nSCENE 3 | 00:03–00:05 | THE DISCOVERY\nShe walks inside and spots the Caramel Iced Capp on the menu. She immediately recognizes it as the drink she just saw.\n\nDialog: “Ohhh… it’s a Caramel Iced Capp.”\n\nSCENE 4 | 00:05–00:06 | SOLD\n\nOver-the-shoulder shot as she points to the Caramel Iced Capp on the menu and orders.\nDialog: “Yeah, I’m getting that.”\nSCENE 5 | 00:06–00:07 | THE TEMPTATION\n\nClose-up of the Caramel Iced Capp being prepared. The creamy iced texture and caramel details are clearly visible.\n\nDialog: “That looks dangerous.”\n\nSCENE 6 | 00:07–00:08 | WAIT FOR IT\nShe waits near the counter, watching the drink being finished. She smiles impatiently when she sees the cup coming toward her.\nDialog: “Okay, come to mama.”\n\nSCENE 7 | 00:08–00:09 | FIRST LOOK\n\nThe barista hands her the finished Caramel Iced Capp. She takes it and immediately examines it with an impressed expression.\n\nDialog: “Oh, hello…”\nSCENE 8 | 00:09–00:11 | THE HERO SHOT\nMacro close-up of the creamy Caramel Iced Capp in her hand, caramel details visible on the drink, condensation on the cup, natural ice movement.\n\nDialog: “Look at that caramel.”\n\nSCENE 9 | 00:11–00:13 | THE VERDICT\n\nFront-camera shot. She takes a sip, pauses, closes her eyes, then slowly nods with a satisfied smile.\nDialog: “Okay… I get the hype.”\nSCENE 10 | 00:13–00:15 | NEW FAVORITE\n\nOutside Tim Hortons, she walks away with the Caramel Iced Capp, turns toward the camera and holds it up with a confident smile.\n\nDialog: “Yeah… this is my new favorite.”",
    "sourceUrl": "https://x.com/codewithhajra/status/2091908219962597395",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091908219962597395.mp4",
    "posterUrl": "/posters/x/2091908219962597395.jpg",
    "duration": 15.168,
    "aspectRatio": "68:45",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "生活实拍感",
      "美食",
      "城市街景"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "food",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2091908219962597395"
  },
  {
    "id": "x-2091889307200172438",
    "title": "绿色纤维织成 MESH 标识",
    "titleEn": "Green Fibers Weave a MESH Mark",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "细线交织成绿色网状字形，最终成为衣物上的小标识。",
    "summaryEn": "Fine strands weave into green mesh lettering that becomes a small clothing label.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2091889307200172438",
    "sourceLabel": "X 原帖",
    "author": "HBCoop_",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091889307200172438.mp4",
    "posterUrl": "/posters/x/2091889307200172438.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091850507283566961",
    "title": "沙漠赛道上的赛车角色展示",
    "titleEn": "A Racer Showcase on a Desert Track",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "未来赛车与赛车手出现在荒漠公路和岩壁之间，最后展示片名。",
    "summaryEn": "A futuristic car and racer appear among desert roads and cliffs before a title reveal.",
    "prompt": "Create a 15-second cinematic character showcase for SANDSTORM RACE, featuring the main racer completing a smooth 360° rotation while the futuristic desert track and his signature racing car progressively assemble around him through fast mechanical, speed-line, and dust-driven transitions. End on a strong hero composition with the fully revealed car and the title SANDSTORM RACE — NO LIMITS. JUST SPEED., holding briefly as a final cinematic shot with no loop.",
    "sourceUrl": "https://x.com/minstorylab01/status/2091850507283566961",
    "sourceLabel": "X 原帖",
    "author": "minstorylab01",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091850507283566961.mp4",
    "posterUrl": "/posters/x/2091850507283566961.jpg",
    "duration": 15.232,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "写实电影感",
      "科幻太空",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/minstorylab01/status/2091850507283566961"
  },
  {
    "id": "x-2091810150567039347",
    "title": "钟表齿轮与 CLOCKWORK 字效",
    "titleEn": "Clock Gears and CLOCKWORK Typography",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物站在巨型钟面与齿轮之间，大字随机械构图轮换。",
    "summaryEn": "A character stands among enormous clock faces and gears as typography changes with the mechanical layouts.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2091810150567039347",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091810150567039347.mp4",
    "posterUrl": "/posters/x/2091810150567039347.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动态图形",
      "3D / CG",
      "奇幻",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091781653152579883",
    "title": "玫瑰花园中的红色礼服人像",
    "titleEn": "A Red-Gown Portrait in a Rose Garden",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子穿带玫瑰装饰的礼服，在花园近景中缓慢转头。",
    "summaryEn": "A woman in a rose-adorned gown turns slowly in close garden portraits.",
    "prompt": "A cinematic, hyper-detailed video of an elegant young woman in a dreamlike rose garden bathed in warm, soft golden hour sunlight. She has dark wavy hair styled in a romantic updo adorned with fresh red roses and delicate dangling jewelry, wearing a dramatic sheer black and dark crimson gown embellished with intricate red floral appliqués and ruby-like beads, along with a matching choker and drop earrings. Holding a luxurious, faceted ruby-red perfume bottle with an ornate gold stopper near her face, she slowly turns her head in smooth, graceful slow motion, bringing the perfume bottle across her chest as delicate red rose petals float gently through the air around her while the camera pans softly to capture the dreamy bokeh background of blooming roses.",
    "sourceUrl": "https://x.com/TaliaAariz/status/2091781653152579883",
    "sourceLabel": "X 原帖",
    "author": "TaliaAariz",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091781653152579883.mp4",
    "posterUrl": "/posters/x/2091781653152579883.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "时尚",
      "自然风景"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TaliaAariz/status/2091781653152579883"
  },
  {
    "id": "x-2091759912917327944",
    "title": "雪山缆车的登乘旅程",
    "titleEn": "Boarding a Cable Car in the Snowy Mountains",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "旅行者走进缆车，在轿厢里看向雪山并与镜头交流。",
    "summaryEn": "A traveler boards a cable car, looks toward snowy peaks and addresses the camera.",
    "prompt": "A Western female traveler in her mid-20s, wearing a beige winter jacket, dark hiking pants, hiking boots, knitted hat, gloves, and a small backpack, walks toward an approaching glass cable-car cabin at a modern snowy mountain station. She waits behind the safety line, boards naturally, holds the handrail, and stands beside the panoramic window as the doors close. The cabin gradually accelerates away from the station and begins climbing above a snow-covered valley. She naturally shifts her balance, places one gloved hand against the window, and watches the snowy forests, rocky slopes, winding trails, and distant mountain peaks below. Her expression changes to genuine amazement as the breathtaking landscape expands beneath her. She smiles and says, \"Wow... look at that,\" with accurate English lip synchronization. The cabin gently sways as it moves along the cable before settling into smooth motion. Ultra-photorealistic travel documentary, realistic cable-car mechanics, authentic human movement, physically accurate inertia and suspension, natural snow and wind behavior, realistic reflections through glass, bright neutral daytime lighting, cinematic stabilized camera movement, immersive mountain ambience, seamless scene progression, consistent character and environment, no CGI appearance, no teleportation, no floating cabin, no exaggerated swinging, no distorted hands, no wardrobe changes, no text overlays, logos, or watermarks.",
    "sourceUrl": "https://x.com/nawalsehar/status/2091759912917327944",
    "sourceLabel": "X 原帖",
    "author": "nawalsehar",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091759912917327944.mp4",
    "posterUrl": "/posters/x/2091759912917327944.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "生活实拍感",
      "自然风景",
      "人物表演"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/nawalsehar/status/2091759912917327944"
  },
  {
    "id": "x-2091736749441736927",
    "title": "居家美妆产品的近景展示",
    "titleEn": "Close-Up Beauty Products at Home",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "女子展示美妆瓶罐与指甲，随后用手指轻触面部。",
    "summaryEn": "A woman presents beauty containers and nails, then touches her face with her fingertips.",
    "prompt": "Create a 15-second photorealistic vertical beauty video featuring the same young woman with long brown hair in a cozy indoor setting.\n\nShe quickly and naturally presents two beauty products to the camera, showing both products clearly while her glossy manicured nails remain visible. She then briefly applies one product to her face and smiles naturally at the camera.\n\nAdd a realistic, friendly female voice-over synchronized with her actions:\n\n“Let me show you these two products. They look so smooth and natural on the skin, and I absolutely love the finish. They’re super easy to use!”\n\nUse realistic skin texture, natural facial expressions, realistic hand movements, soft warm lighting, smooth camera motion, and a premium beauty-creator aesthetic. Keep the video clean and professional with no text or graphic elements on screen.",
    "sourceUrl": "https://x.com/CaliraVal/status/2091736749441736927",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091736749441736927.mp4",
    "posterUrl": "/posters/x/2091736749441736927.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "商业广告感",
      "产品与广告",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2091736749441736927"
  },
  {
    "id": "x-2091734655993291051",
    "title": "粉色极繁时装广告",
    "titleEn": "A Maximalist Pink Fashion Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "模特、星星、彩色饰物和立体文字组成高密度粉色广告画面。",
    "summaryEn": "A model, stars, colorful accessories and dimensional type form dense pink ad scenes.",
    "prompt": "Create a 15-second, 9:16, 24fps high-impact anime fashion commercial.\nTHEME:\nSOLVÉRA — “KAWAII IS POWER”\nSTYLE:\nHarajuku Kawaii × Decora × Y2K × Cyber Kawaii × Punk Kawaii × Japanese anime opening × premium fashion campaign × fast motion graphics.\nCute, sexy, wild, stylish, rebellious, youthful and visually explosive — never childish.\nCHARACTER — HIGHEST PRIORITY:\nOne adult East Asian anime woman with refined Japanese-anime features, sharp playful eyes, slim feminine curves, long legs and confident fashion-model energy.\nKeep her exact identity, face, hairstyle, age and body proportions consistent throughout.\nOutfits may change, but never redesign the character.\nVISUAL RULE:\nEvery shot should combine:\ncharacter motion + camera motion + foreground motion + background graphics.\nUse strong parallax, foreground wipes, oversized hair clips, beads, hearts, stars, ribbons, plush charms, metallic bags, stickers, Purikura photos, manga speed lines, halftone, print misregistration, chromatic offset and huge typography.\nAlternate camera moves:\nTracking / Orbit / Roll / Crane Up / Pull Back / Whip Pan / Foreground Pass.\n0–1s — KAWAII HOOK\nExtreme close-up through giant colorful accessories. 18mm camera rushes toward her face.\nShe lowers colorful sunglasses and looks directly at camera.\nHuge text slams in:\n“KAWAII!”\nSmall Japanese text:\n“かわいい、爆発。”\nRGB flash, sticker burst, giant heart wipes the lens.\n1–2.2s — DECORA\nPink × yellow.\nShe walks toward camera while beads, ribbons, stars and clips fly onto her styling.\nLow-angle tracking, slight camera roll.\nText:\n“MORE! MORE!”\n2.2–3.3s — HARAJUKU Y2K\nSilver × cyan.\nMetallic bag wipes across lens.\nShe appears, performs a fast turn and looks back over her shoulder.\nPurikura frames and chrome graphics follow her movement.\nText:\n“COLOR ME”\n3.3–4.4s — SWEET LOLITA MIX\nLavender × mint.\nGiant bows, lace and ribbons float around her.\nFast 90° orbit while she spins in the opposite direction.\nText:\n“MIX IT UP”\n4.4–5.6s — THREE-WAY SPLIT\nSame heroine appears in three panels:\nDECORA / Y2K / LOLITA.\nDifferent camera motion and poses in each panel.\nLarge text across all three:\n“SWITCH!”\n5.6–6.8s — CYBER KAWAII\nBlack × cyan × hot pink × chrome.\nShe walks toward camera and reaches toward the lens.\nCamera passes under her hand into a low angle.\nNeon hearts and graphics orbit her.\nText:\n“NO RULES”\nHer hand covers the lens.\n6.8–7.8s — TYPOGRAPHY EXPLOSION\nBlack × pink.\nHuge Japanese typography:\n“かわいいは自由。”\nRepeated “NO RULES” layers in the background.\nCamera flies through giant letterforms.\nTorn paper, ink splashes, scratches, print-offset effects.\n7.8–9s — PUNK KAWAII\nBlack × shocking pink.\nShe starts crouched, suddenly rises.\nLow-angle camera cranes upward with slight roll.\nShe whips her cropped jacket as chains and bows explode outward.\nText:\n“TOO MUCH?”\n9–10s — MAXIMUM KAWAII\nExtreme face close-up.\nRapid pull-back reveals a huge space overwhelmed by clips, stars, hearts, ribbons, plush charms and stickers.\nShe barely moves, only raises one eyebrow.\nText:\n“まだ足りない。”\nAt 9.9s, EVERYTHING suddenly stops.\n10–11.2s — SILENT BREAK\nPure white background.\nFinal LOOK: RAINBOW STREET.\nMore mature, refined and premium.\nShe slowly walks forward two steps while camera dollies backward.\nAlmost no VFX.\nOnly toy bell + two footsteps.\nPink ripple appears under her second step.\n11.2–13s — VISUAL EXPLOSION RETURNS\nDozens of Purikura photo cards showing previous looks fill the space.\nCamera snakes through a huge 3D collage of hearts, stars, ribbons, stickers and magazine graphics.\nEnd on heroine in final look.\nText:\n“HARAJUKU MODE”\n13–15s — FINAL HERO\nHot pink background.\nHeroine holds a confident, sexy fashion pose.\nHuge white typography:\n“KAWAII”\nthen\n“IS POWER”\nThen:\n“かわいいは、強い。”\n“SOLVÉRA”\n“OUT NOW”\nShe tilts one shoulder and gives a confident smile.\nTiny heart pops.\nFreeze final frame for 0.2s.\nMUSIC:\nCute Hyperpop / Harajuku Kawaii Pop, around 150 BPM.\nHeavy kick, clap, bubblegum bass, toy synth, sparkling bells, cute vocal chops.\n0–5.6s: fast hyperpop\n5.6–7.8s: darker cyber section\n7.8–10s: maximum intensity\n9.9s: complete stop\n10–11.2s: bell + footsteps only\n11.2s: full beat explodes back\n13–15s: “KAWAII / IS POWER” hits sync exactly to the beat\nAVOID:\nchildlike proportions, chibi style, schoolgirl look, face changes, body changes, identity drift, random outfit errors, static posing, weak camera motion, unreadable typography, generic cute visuals.\nFINAL FEEL:\nA high-budget Japanese anime fashion campaign combining anime-opening energy, fashion editorial styling and premium motion graphics — ultra-kawaii, sexy, rebellious, stylish and unforgettable.",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2091734655993291051",
    "sourceLabel": "X 原帖",
    "author": "ImaStudio_ai",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091734655993291051.mp4",
    "posterUrl": "/posters/x/2091734655993291051.jpg",
    "duration": 15.16263,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "动态图形",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2091734655993291051"
  },
  {
    "id": "x-2091730894189154501",
    "title": "运球、起跳与灌篮的动画分镜",
    "titleEn": "Animated Dribbling, Takeoff and Dunking",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "篮球角色从运球起步到跃向篮筐，动作配合简洁文字背景。",
    "summaryEn": "A basketball character moves from dribbling to a leap toward the hoop against simple type backgrounds.",
    "prompt": "Reference Images:\nimage1 \nimage2 \n\n【REFERENCE】\nimage1 ＝人物三面図。顔、髪型、体型、頭身、全身比率、ユニフォーム、シューズ、2Dアニメ表現の基準。FRONT／SIDE／BACKを同一人物として全編維持する。\nimage2 ＝固定9パネルSB。左上→右、上段→中段→下段の順でPanel 1〜9として読み、動作順、構図、カメラ距離、ゴールとの位置関係の基準とする。人物は常にimage1 を優先し、SB内の顔、髪型、性別表現、体型、衣装色は継承しない。シート本体、3×3分割、枠、静止画表現は映像に出さず、9パネルを自然な中間動作でつないだ1本の連続アクションへ変換する。\n\n【CONDITION】\n15秒。日本の漫画・アニメ調による、ミニマルでシネマティックなロンググライド・ダンク。白基調の明るいシームレス空間に、同一選手1人、ボール1個、ゴール1基、薄い接地影、背景タイポグラフィ、最下層の薄い背景MGのみ。マッチ・オン・アクションで動作、ボール、身体、速度、ゴールとの距離を自然につなぎ、踏み切りからダンク完了まで保持手を変えない。\nタイポはDRIVE / RISE / DUNKのみ。人物・ボール・ゴールの後方、画面上部の固定レイアウトに上から順に蓄積し、カメラ移動の影響を受けない。DRIVE＝明るいシアン、RISE＝明るいライムグリーン、DUNK＝鮮やかなバイオレット。\n背景の最下層にのみ、薄いモーショングラフィックを追加する。細い速度線、細い軌道線、薄いリング、控えめな幾何学ライン程度に限定し、低密度で、人物・ボール・ゴールより後方、タイポグラフィより下のレイヤーで動かす。背景MGはタイポグラフィの位置、固定状態、見え方に影響を与えない。\n背景全体は常に白基調のハイキー空間を維持する。開始時はほぼ純白。DRIVE後はごく薄いシアン寄りの白、RISE後はごく薄いミント〜ライム寄りの白、DUNK後はごく薄いバイオレット〜マゼンタ寄りの白へ穏やかに変化する。背景色は淡く明るいまま変化し、濃いグレー、黒、暗いグラデーション、急な暗転へ移行しない。\n\n【SHOT FLOW】\n0.0〜1.0秒｜Panel 1｜START\n中距離全身3/4。ボールを持つ静かな開始姿勢。ゴールはまだ見せない。背景はほぼ純白。巨大でごく薄いDRIVEを背景に一部だけ見せる。まだ発音しない。背景MGは最小限。\n1.0〜2.5秒｜Panel 2｜MOVE\n片手ドリブルを明確に行いながら前進。ボールが手→床→手へ戻る動作を見せる。前側3/4で追従。DRIVEが左から入り、上部1段目へ固定。固定した瞬間、画面外から“Drive.”。最下層では薄い速度線MGが進行方向へ流れ、背景はごく薄いシアン寄りの白へ変化する。\n2.5〜3.8秒｜Panel 3｜ACCELERATION\n前側3/4で片手ドリブルを維持し、歩幅、ボール上下動、速度を強調。DRIVEは固定。背景MGは低密度の細線として後方へ流れる。追加発話なし。\n3.8〜5.3秒｜Panel 4｜FINAL APPROACH\n側面寄り3/4。最後のドリブルから同じ手で滑らかにボールをすくい上げ、踏み切り準備へ連続移行。両手で抱えて停止しない。ゴールが前方に現れ、動作軸を越えない。背景MGは進行方向へ収束する。\n5.3〜8.6秒｜Panel 5｜LONG GLIDE / RISE\n遠い位置から片足で離陸し、全身を横へ伸ばして長く滑空。やや引いた見上げ3/4で選手とゴールの距離を同一画面に見せる。RISEが右側から入り、DRIVEの下へ固定。固定した瞬間、画面外から“Rise.”。背景MGは薄いリングと軌道線を追加し、背景は淡いミント〜ライム寄りの白へ穏やかに変化する。\n8.6〜10.4秒｜Panel 6｜RIM APPROACH\n同じジャンプ・同じ保持手でリムへ接近。Panel 5と同じ側からカメラを近づけ、顔、腕、ボール、リムを明確に見せる。DRIVE / RISEは固定。背景MGは過密にせず、後方で控えめに流す。\n10.4〜11.9秒｜Panel 7｜DUNK\n同じ保持手で片手ダンク。ボールがリムを通過しネットが大きく伸びる瞬間にDUNKが左側から入り、3段目へ固定。同時に画面外から強く短く“Dunk.”。背景MGでは短い放射線と小さなインパクトリングだけを瞬間的に出す。背景は淡いバイオレット〜マゼンタ寄りの白へ変化する。\n11.9〜13.2秒｜Panel 8｜LANDING\nすぐ手を離して下降・着地。中距離〜やや広めの全身で膝を軽く曲げて衝撃を吸収。3語は固定。背景MGは急速に弱まり、細い線の余韻だけを残す。発話なし。\n13.2〜15.0秒｜Panel 9｜ENDING\n@image2 のPanel 9を最終フレームの構図基準として優先する。カメラをゆっくり引き、少し高い位置から、人物とゴールの両方を必ず同一画面に収め、ストーリーボードと同じ引いた距離関係で終える。選手は静止し、DRIVE / RISE / DUNKの3語は固定位置にそのまま残す。背景MGだけを徐々に減衰させて終盤までにほぼ消し、背景色は明るく清潔な白基調へ穏やかに戻す。終盤に画面外から“Complete.”を1回。最後は固定された3語、静止した選手、わずかに揺れるネットだけを残して終える。\n\n【SOUND】\nボール、シューズ、衣服、風切り、踏み切り、リム、ネット、着地の競技音を自然に維持。“Drive.” “Rise.” “Dunk.” “Complete.”はすべて非ダイジェティックなオフスクリーン音声。選手は一切喋らず、リップシンクしない。文字の定着と発音を同期する。BGMはボーカルなしのミニマルなビートなら可。音量は控えめ。\n\n【NEGATIVE】\n参照シート、3×3分割、枠、コラージュ、静止画スライド、数字、字幕、UI、ロゴ、透かし、観客、コート、壁、スタジアム、追加人物を出さない。SB側の人物デザイン、顔、髪型、体型、性別表現、衣装色を継承しない。人物・ボール・ゴールの複製、別人化、衣装・髪型変更、手足増加、ゴール移動、空中での保持手変更、両手ダンク、長いリムぶら下がり、動作反復、長いフリーズ、歩き去り、祝福ポーズを禁止。\nDRIVE / RISE / DUNK以外の画面内文字、3語の入れ替え・消去、人物・顔・ボール・リムを長時間隠す文字を禁止。背景MGを人物やタイポより前に出さず、タイポの固定状態を崩さない。太い図形、重いベタ面、過密MG、派手すぎるネオン、爆発、煙、粒子を禁止。背景を黒、濃いグレー、暗い低明度グラデーションへ変えず、全体を暗転させない。指定4音声以外のナレーション、選手の発話、会話、実況、観客声、歌唱は禁止。",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2091730894189154501",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091730894189154501.mp4",
    "posterUrl": "/posters/x/2091730894189154501.jpg",
    "duration": 15.13941,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2091730894189154501"
  },
  {
    "id": "x-2091724679027265705",
    "title": "岩石伪装的巨龙苏醒",
    "titleEn": "A Dragon Awakens from Stone",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "旅人走近山谷中的石状巨龙，巨龙随后抬头并与人物相对。",
    "summaryEn": "A traveler approaches a stone-like dragon in a valley as it raises its head to face her.",
    "prompt": "STYLE: Ultra-photorealistic live-action fantasy film, premium Hollywood creature VFX, realistic human performance, natural environmental physics, cinematic HDR lighting, atmospheric scale, practical-looking textures, no cartoon styling.\n\nCHARACTER: Use Image 1 only as the exact facial identity reference for the woman. Preserve her recognizable facial features, hairstyle, skin tone and body proportions consistently. Completely different outfit: deep-olive traveling cloak, fitted cream tunic, dark brown trousers, worn leather ankle boots, small leather satchel.\n\nCREATURE: A gigantic ancient ivory dragon with subtle warm-gold scales, enormous folded wings, weathered horns, scars and natural imperfections. Majestic, ancient and intelligent. No aggressive behavior.\n\n0–3s — THE STRANGE SIGNAL\n\nAerial shot over a vast alpine valley at dawn. The woman walks alone through tall grass carrying an old brass compass.\n\nSuddenly, the compass needle spins violently.\n\nShe stops.\n\nEvery sound in the valley disappears — the wind, birds and waterfalls become completely silent.\n\nThe woman looks toward a distant mountain.\n\n3–6s — THE MOUNTAIN MOVES\n\nA deep vibration travels through the ground.\n\nThe woman looks down.\n\nTiny stones around her begin sliding toward the mountain as if pulled by an invisible force.\n\nShe slowly follows them.\n\nThe camera pushes toward a massive cliff face.\n\nA huge ancient eye suddenly opens within what appeared to be solid rock.\n\nThe woman freezes.\n\n6–10s — THE IMPOSSIBLE REVEAL\n\nThe “mountain” slowly moves.\n\nIt is actually the sleeping dragon, almost completely covered by centuries of moss, stone and vegetation.\n\nIts enormous body unfolds from the landscape.\n\nDust and loose stones fall from its weathered scales.\n\nThe woman realizes she has been standing beside a living creature that has been mistaken for part of the mountain for centuries.\n\n10–13s — THE MEMORY\n\nThe dragon lowers its enormous head.\n\nInstead of looking at the woman, it looks directly at her brass compass.\n\nThe compass suddenly opens by itself.\n\nInside is a tiny engraved symbol identical to a marking on the dragon's ancient horn.\n\nThe woman's expression changes from fear to disbelief.\n\nThe dragon gently exhales.\n\nThe compass needle stops and points toward the dragon.\n\n13–15s — THE AWAKENING\n\nThe dragon suddenly turns its enormous head toward the distant horizon.\n\nFar beyond the valley, another massive dragon-shaped shadow moves behind the clouds.\n\nThe woman looks up.\n\nThe ancient dragon spreads one enormous wing.\n\nCUT TO BLACK.\n\nFINAL FEELING: The woman has not discovered a dragon — she has accidentally awakened something that has been waiting for a very specific signal for centuries.\n\nCAMERA: Cinematic handheld realism mixed with sweeping aerial shots, dramatic scale changes, shallow depth of field for character moments, subtle lens breathing, realistic motion blur.\n\nLIGHTING: Cold blue dawn gradually interrupted by warm sunlight breaking through clouds.\n\nSOUND: Wind and nature ambience → complete unnatural silence → deep subterranean vibration → enormous wing movement → final low-frequency dragon call.",
    "sourceUrl": "https://x.com/codewithhajra/status/2091724679027265705",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091724679027265705.mp4",
    "posterUrl": "/posters/x/2091724679027265705.jpg",
    "duration": 14.72,
    "aspectRatio": "45:56",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "3D / CG",
      "奇幻",
      "动物与生物"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2091724679027265705"
  },
  {
    "id": "x-2091722040978063594",
    "title": "天体图与 COSMOS 字体空间",
    "titleEn": "Celestial Diagrams and COSMOS Typography",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物被轨道、星形与大字环绕，画面以天体图构成空间层次。",
    "summaryEn": "Orbits, stars and oversized words surround a character in layered celestial diagrams.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2091722040978063594",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091722040978063594.mp4",
    "posterUrl": "/posters/x/2091722040978063594.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动态图形",
      "3D / CG",
      "奇幻",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091697364906254361",
    "title": "地铁空间里的文字广告",
    "titleEn": "A Typography Ad across Subway Spaces",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物穿过站台与城市夜景，站内标识变成粉白色广告文字。",
    "summaryEn": "A character moves through platforms and night streets as station signs become pink-and-white ad copy.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2091697364906254361",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-24",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091697364906254361.mp4",
    "posterUrl": "/posters/x/2091697364906254361.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "动态图形",
      "城市街景",
      "人物表演"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091617234930053234",
    "title": "赛道不断翻折的未来赛车",
    "titleEn": "A Future Race on a Folding Track",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "赛车沿霓虹赛道前进，路面和弯道在游戏界面中持续变化。",
    "summaryEn": "A racing vehicle speeds along a neon track whose roads and corners change within a game interface.",
    "prompt": null,
    "sourceUrl": "https://x.com/AllaAisling/status/2091617234930053234",
    "sourceLabel": "X 原帖",
    "author": "AllaAisling",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091617234930053234.mp4",
    "posterUrl": "/posters/x/2091617234930053234.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "科幻太空",
      "城市街景"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "sci-fi",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091616500440572217",
    "title": "机械翻片组成 FLIP 标识",
    "titleEn": "Mechanical Tiles Form a FLIP Mark",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "规则排列的机械片翻转并重组，最终构成 FLIP 字样。",
    "summaryEn": "Rows of mechanical tiles flip and rearrange into the word FLIP.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2091616500440572217",
    "sourceLabel": "X 原帖",
    "author": "HBCoop_",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091616500440572217.mp4",
    "posterUrl": "/posters/x/2091616500440572217.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091559842099929339",
    "title": "拉开窗帘看见黑洞的视频编辑",
    "titleEn": "A Video Edit Reveals a Black Hole behind Curtains",
    "model": "MiniMax H3",
    "mode": "Ref2VA",
    "summary": "人物拉开窗帘，窗外城市被替换为发光黑洞；本站播放编辑后的 H3 结果。",
    "summaryEn": "A character opens curtains onto a glowing black hole replacing the city view; this is the edited H3 output.",
    "prompt": "Replace the view outside the window in @ Video1 with a view on black hole gargantua",
    "sourceUrl": "https://x.com/ivanfioravanti/status/2091559842099929339",
    "sourceLabel": "X 原帖",
    "author": "ivanfioravanti",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091559842099929339.mp4",
    "posterUrl": "/posters/x/2091559842099929339.jpg",
    "duration": 6.656,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "照片级写实",
      "科幻太空"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text",
      "video"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ivanfioravanti/status/2091559842099929339"
  },
  {
    "id": "x-2091535156687655371",
    "title": "巨型糖果色字体围绕人物起舞",
    "titleEn": "Giant Candy-Colored Type Dances around a Performer",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在粉色圆环、发光星形和立体字母之间连续表演。",
    "summaryEn": "A performer moves among pink rings, glowing stars and dimensional letters.",
    "prompt": null,
    "sourceUrl": "https://x.com/T_st_love_love/status/2091535156687655371",
    "sourceLabel": "X 原帖",
    "author": "T_st_love_love",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091535156687655371.mp4",
    "posterUrl": "/posters/x/2091535156687655371.jpg",
    "duration": 15.061333,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "照片级写实",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091529120828334142",
    "title": "霓虹街巷格斗角色登场",
    "titleEn": "A Fighter Enters a Neon Alley",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "格斗角色在霓虹街巷中登场，动作镜头接入游戏宣传标语。",
    "summaryEn": "A fighter enters a neon alley before action shots lead into a game-style title card.",
    "prompt": "Style: High-end 3D animated fighting game cinematic, ultra-detailed character design, premium AAA game trailer quality, dynamic anime-inspired realism, cinematic lighting, dramatic neon city atmosphere, fluid martial-arts animation, realistic cloth and hair physics, powerful impact effects, fast camera movement, motion blur, slow-motion hits, energetic hip-hop beat, bold arcade-style typography, polished game-commercial aesthetic, no gore, no blood, no watermark.\n\nMain Character: A fearless young female street fighter with a confident gangster aesthetic — black leather jacket, fitted dark outfit, combat boots, gold accessories, fingerless gloves, stylish sunglasses, long moving hair, sharp expression. Keep her face, hairstyle, outfit and identity completely consistent throughout.\n\n0–3s — THE ARRIVAL\n\nNighttime neon city street surrounded by graffiti-covered walls, glowing signs, parked cars and a crowd of intimidating street fighters.\n\nLow-angle cinematic shot of the girl walking toward camera.\n\nClose-ups: boots hitting pavement → gold chain → gloves tightening → confident eyes.\n\nShe removes her sunglasses and looks directly at the opponents.\n\nTypography slams onto screen:\n\n“GAME ON.”\n\nBass drops.\n\n3–7s — THE FIGHT\n\nSeveral opponents rush toward her simultaneously.\n\nShe launches into fast, stylish martial-arts choreography — ducking under a punch, spinning kick, quick counterattack, dramatic elbow strike and agile evasive movements.\n\nCamera rapidly switches between:\n\nwide action shot → low-angle → 360° orbit → extreme close-up → dynamic overhead shot.\n\nEvery impact lands precisely on the hip-hop beat.\n\nAdd stylized energy bursts, dust particles, sparks and cinematic impact frames.\n\nTypography flashes:\n\n“NO FEAR.”\n“NO MERCY.”\n\nKeep the action intense but non-graphic.\n\n7–11s — POWER MODE\n\nThe girl stands surrounded by defeated opponents who are visibly knocked down but unharmed.\n\nThe environment darkens.\n\nNeon lights pulse behind her.\n\nShe cracks her knuckles and suddenly charges forward.\n\nUse a dramatic slow-motion sequence as she performs a powerful spinning attack through multiple opponents, followed by a rapid speed ramp back to full speed.\n\nCamera rotates around her while animated energy trails follow her movements.\n\nTypography explodes across the frame:\n\n“LEVEL UP.”\n\n11–15s — FINAL BOSS MOMENT\n\nA massive final opponent steps out of the smoke.\n\nThe girl smiles confidently.\n\nExtreme close-up of her eyes.\n\nMusic suddenly cuts for half a second.\n\nThen the final beat drops.\n\nShe launches forward with a spectacular cinematic finishing combo.\n\nFinal impact freezes into a stylized game-poster frame.\n\nHuge typography:\n\n“SHE RUNS THE STREETS.”\n\nSmall subtitle:\n\n“FIGHT. RISE. DOMINATE.”\n\nEnd with the girl standing confidently in the neon street as smoke and glowing particles move around her.\n\nCamera: aggressive professional game-trailer cinematography, dynamic tracking, handheld energy, low-angle hero shots, 360° camera rotations, speed ramps, impact zooms and dramatic slow motion.\n\nAnimation: fluid body mechanics, believable martial-arts movement, consistent character proportions, realistic hair and clothing physics, detailed facial expressions.\n\nSound: fast hip-hop/trap instrumental, heavy bass, punchy drums, arcade-style impact sounds, crowd reactions, cinematic risers and a powerful final bass drop.\n\nCRITICAL CONTINUITY: Same female protagonist, same face, same hairstyle, same gangster outfit, same environment and consistent character proportions throughout all 15 seconds. No random character changes, no distorted hands, no extra limbs, no gore.",
    "sourceUrl": "https://x.com/Inshrah_ali_/status/2091529120828334142",
    "sourceLabel": "X 原帖",
    "author": "Inshrah_ali_",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091529120828334142.mp4",
    "posterUrl": "/posters/x/2091529120828334142.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "3D / CG",
      "动作打斗",
      "城市街景"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Inshrah_ali_/status/2091529120828334142"
  },
  {
    "id": "x-2091508732866032061",
    "title": "动漫少女煮一碗拉面",
    "titleEn": "An Anime Girl Makes Ramen",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "从面条入锅到端碗品尝，以料理特写串起拉面制作过程。",
    "summaryEn": "Cooking close-ups follow noodles from the pot to a finished bowl of ramen.",
    "prompt": null,
    "sourceUrl": "https://x.com/nnanchi_uma/status/2091508732866032061",
    "sourceLabel": "X 原帖",
    "author": "nnanchi_uma",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091508732866032061.mp4",
    "posterUrl": "/posters/x/2091508732866032061.jpg",
    "duration": 15.872,
    "aspectRatio": "127:72",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "美食",
      "日常生活"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091496513432105182",
    "title": "穿行花园般的梦幻车厢",
    "titleEn": "Walking through Dreamlike Floral Carriages",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "白裙人物穿过鲜花、玻璃穹顶与不断变化的车厢空间。",
    "summaryEn": "A woman in white walks through flowers, glass roofs and changing carriage interiors.",
    "prompt": "A cinematic, surreal dream sequence follows a stylish young East Asian woman traveling through a fantastical train car that continuously morphs into lavish, otherworldly floral dimensions. She begins standing in a subway train overflowing with pastel pink blossoms that form a vibrant floral tunnel, wearing a romantic white blouse and patterned skirt. As the video transitions through subtle glitches and magical portals, she weightlessly floats through swirling petals in a dark floral kimono, stands beneath a massive blooming peony on the train ceiling, twirls in a white sundress within a cathedral-like glass carriage wrapped in blossoms, and spins gracefully in a striking crimson gown surrounded by cascading red roses. The atmosphere reaches a climax as she reaches toward a glowing cosmic orb of flowers inside a spiraling vortex before stepping through a train door into an intricate multi-dimensional flower landscape, with the camera using smooth cinematic motion, vibrant art direction, dynamic lightning cuts, and whimsical floral physics throughout.",
    "sourceUrl": "https://x.com/TaliaAariz/status/2091496513432105182",
    "sourceLabel": "X 原帖",
    "author": "TaliaAariz",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091496513432105182.mp4",
    "posterUrl": "/posters/x/2091496513432105182.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "时尚",
      "奇幻"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/TaliaAariz/status/2091496513432105182"
  },
  {
    "id": "x-2091485010515914784",
    "title": "动漫角色操控巨型文字",
    "titleEn": "An Anime Character Manipulates Giant Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑发角色在工业空间中移动，巨型白色与红色文字穿插其间。",
    "summaryEn": "A dark-haired character moves through an industrial setting surrounded by oversized white and red type.",
    "prompt": "# 【ここだけ変更】\n\n文字列A = 「MiniMax H3」\n文字列B = 「AI STUDIO ONEROOM」\n\n※使用時に変更するのは上の2行だけ。\n※以下の本文は変更しない。\n※映像内には、上で設定した文字列A・文字列B以外の文字を表示しない。\n\n---\n\n提供された参照画像を主人公の基準として使用する。\nキャラクターの同一性、顔、髪型、衣装のシルエット、ファッション性、全体的な印象をできる限り維持する。\n\n15秒、16:9横長のハイテンポなファッションMV。\n\n映像の中心は常に主人公。\n\n主人公の身体の動きとダイナミックなカメラワークを最優先し、その動きに反応して、冒頭で設定した文字列A・文字列Bが追従する。\n\n基本的な演出構造は、\n\n主人公が動く\n→ カメラが大きく反応する\n→ 文字がその動きに引っ張られる\n\nという順番。\n\nカメラは空間を大きく使い、急速なドリーイン／ドリーアウト、横方向の高速通過、ローアングル、オービット、ウィップパン、スナップズームを組み合わせる。\n\n高速運動と完全静止の差を強く作り、バレットタイムを最大の見せ場にする。\n\n## タイポグラフィ\n\n使用する文字は、冒頭で設定した文字列A・文字列Bのみ。\n文字はフラットでシャープな2Dタイポグラフィ。\n厚みのある3D文字にはしない。\n基本色は白。\nアクセントとして赤とオレンジを使用する。\n色はグラデーションではなく、ビートに同期した瞬間的な切り替えを中心にする。\n文字は主人公の身体動作に連動する。\n主人公が手を横へ払えば文字も高速で横へ飛ぶ。\n主人公が引き寄せれば文字も集まる。\n主人公が回転すれば文字もその回転に引っ張られる。\n主人公が手をカメラ方向へ押し出せば文字も前景へ急接近する。\n主人公が停止すれば文字も停止する。\n文字だけが勝手にランダムに動かない。\nモーショングラフィックスは短いアクセントとして使用する。\n文字の高速移動中に一瞬だけ字間が広がる。\n文字が停止した瞬間に白から赤、またはオレンジへ切り替わる。\n高速移動直後に短く分解され、即座に再整列する。\n一文字ずつ数フレーム遅れて追従する動きを短時間だけ使用する。\n文字内部の複雑な変形を長時間見せない。\n\n## 奥行き\n\n文字自体を3D化せず、配置とカメラワークによって空間的な立体感を作る。\n文字を、\nカメラのすぐ近く\n主人公より手前\n主人公の背後\nなど異なる距離へ配置する。\nカメラ移動によって強いパララックスを発生させる。\n近い文字ほど大きく高速に流れ、遠い文字ほど小さくゆっくり動く。\n文字が主人公の身体の背後へ隠れたり、腕や身体の手前を横切ったりする自然な前後関係を維持する。\n\n---\n\n## Shot 1 | .0–00:00.8\n\nワイド。\n主人公が中央に立つ。\n前景に大きな白い文字列A。\n主人公の背後に小さめの文字列B。\n強いビートと同時にカメラが前景文字の隙間を高速で通過しながら主人公へドリーイン。\n文字、主人公、背景の距離差によって強いパララックスを発生させる。\n通過した瞬間、文字列Aの一部分だけ赤へ変化。\n\n## Shot 2 | .8–00:01.6\n\n顔のクローズアップへスナップズーム。\n主人公が鋭く横を見る。\n視線に引っ張られるように文字列Bが横方向へ高速移動。\nカメラも同じ方向へ短くスライド。\n文字は顔を隠さない。\n\n## Shot 3 | .6–00:02.5\n\n全身ローアングル。\n主人公が一歩前へ出ながら片手を強く横へ払う。\nその手の軌道に合わせて文字列Aが主人公の背後から横を通過し、カメラ直前まで高速で飛ぶ。\n移動中だけ字間が大きく広がる。\nカメラは低い位置から高速ティルトアップ。\n\n## Shot 4 | .5–00:03.4\n\nミディアム。\n主人公が両手を軽く引き寄せる。\n分離していた文字列Bの文字が左右から主人公の周囲へ高速で集まり、正しい文字列として再構成される。\n完成した瞬間、一部分だけオレンジへ変化。\n主人公が手を止めると文字も完全停止。\n\n## Shot 5 | .4–00:04.2\n\n主人公が身体を鋭くターン。\n同時にカメラは主人公とは逆方向へ高速オービット。\n文字列Aが主人公の回転に引っ張られるように大きな円弧を描いて移動する。\n身体とカメラの大きな運動を最優先する。\n\n## Shot 6 | .2–00:05.7\n\n第一のバレットタイム。\n\n主人公がターン途中で完全停止。\n身体、髪、衣装も完全停止。\n文字列A・文字列Bも完全停止。\nカメラだけが主人公の周囲を約180度高速オービットする。\n文字列Aは主人公より手前。\n文字列Bは主人公の背後。\n文字はフラットな2Dのまま。\nカメラだけが移動することで、主人公と文字の間に強烈なパララックスと前後関係を生み出す。\nカメラ角度によって文字の一部が主人公の身体の背後へ自然に隠れる。\nバレットタイム中は文字を動かさない。\n\n## Shot 7 | .7–00:06.5\n\n時間が一気に再始動。\n主人公がターンを完成させ、その勢いのまま片手を前へ押し出す。\n同時に文字列Aがカメラ方向へ猛烈に加速。\n文字列Aの一部分だけ赤へ変化。\n文字がカメラ直前まで巨大化して画面を横切り、その動きで次のショットへハードトランジション。\n\n## Shot 8 | .5–00:07.4\n\nワイド。\nカメラが高速で後退。\n主人公はカメラへ向かって前進。\n主人公の背後に大きな白い文字列B。\n主人公の歩行に合わせて文字も前方向へ少し移動する。\n人物とカメラが逆方向に動くことで空間の広さを強調。\n\n## Shot 9 | .4–00:08.3\n\nサイドアングル。\nカメラが主人公の横を高速で通過。\n主人公が片手を下から上へ大きく振る。\nその軌道に引っ張られて文字列Aが画面下から上へ飛び上がる。\n文字は主人公の背後から始まり、途中で身体の手前へ出る。\n最後の瞬間だけ文字の一部分がオレンジへ変化。\n\n## Shot 10 | .3–00:09.2\n\nクローズアップ。\n主人公がカメラへ手を伸ばす。\n主人公の背後に小さく存在していた文字列Bが、手に引き寄せられるようにカメラ方向へ急接近。\nカメラも同時に主人公へ高速ドリーイン。\n手、顔、文字の三段階の距離差を強調する。\n\n## Shot 11 | .2–00:10.3\n\n第二のバレットタイム。\n主人公が腕を振る途中で完全停止。\n文字列Aも完全停止。\nカメラだけが主人公の斜め下から上方向へ回り込む。\n文字は動かさない。\nカメラ移動だけで主人公と文字の距離、重なり、前後関係が大きく変化して見える。\n時間再開直前、一部分だけ赤へ変化。\n\n## Shot 12 | .3–00:11.5\n\n時間再開。\n主人公が腕を振り切る。\n文字列Aがその勢いのまま画面を高速横断。\n逆方向から文字列Bが主人公の背後を通過。\n二つの文字列が主人公を中心として前後ですれ違う。\nカメラも横方向へ高速トラッキング。\n\n## Shot 13 | .5–00:12.8\n\n高速クライマックス。\n顔アップ。\n全身。\n極端なローアングル。\n横顔。\n手元。\n短いハードカットで連続切り替え。\nすべてのカットで人物のスケール、カメラ位置、角度を変える。\n文字列A・文字列Bは主人公の身体動作に追従する。\n高速移動。\n瞬間的な字間変化。\n短い分解と再整列。\n白から赤またはオレンジへの瞬間的な色変更。\nこれらを短いアクセントとして使用する。\n文字のモーションよりも主人公とカメラの速度感を優先する。\n\n## Shot 14 | .8–00:15.0\n\n最終ヒーローショット。\n主人公の斜め側面近くから開始。\nカメラが主人公の周囲を大きく高速オービットしながら正面へ回り込む。\n途中、文字列Aが主人公の背後から手前へ移動。\n文字列Bが前景側から主人公の背後へ抜ける。\nカメラが二つの文字列の間を高速で通り抜けながら主人公正面へ到達。\n主人公が両手を軽く動かす。\nその動きに引き寄せられるように文字列A・文字列Bが左右から集まる。\n二つの文字列が主人公の周囲へ正しく整列。\n整列直前だけ字間が大きく広がる。\n最後のビートで一気に正しい間隔へ戻る。\n主人公が手を止める。\n文字も完全停止。\n最後の約0.4秒だけ安定した強いヒーロー構図を保持する。\n\n## 制約\n\n追加人物を出さない。\nキャラクターの顔、髪型、衣装を変更しない。\n冒頭で設定した文字列A・文字列B以外の文字を表示しない。\n意味不明な文字を生成しない。\n字幕を表示しない。\n文字を厚みのある3D文字にしない。\n同じカメラ構図を繰り返さない。\n長時間の静止ショットを作らない。\nバレットタイム中は主人公と文字を完全静止させ、カメラだけを動かす。\n文字の複雑なモーショングラフィックスを主人公やカメラより優先しない。\n\n## Audio\n\n高速でエッジの効いたスタイリッシュなインストゥルメンタル。\nボーカルなし。\nセリフなし。\n字幕なし。\n主人公の身体動作、カメラ移動、文字の空間移動、カット、時間停止、時間再開を音楽のビートへ同期させる。\nバレットタイム直前に音を急激に絞る。\n時間停止中は音の空間だけを残す。\n時間再開と同時に強いビートを戻す。\n\n## 最重要\n\n主人公の身体の動きを最優先する。\nカメラを大きくダイナミックに動かす。\n文字は主人公が操っているように動く。\n使用文字は冒頭で指定された文字列A・文字列Bのみ。\n白を基本に赤とオレンジをアクセントとして使用する。\n文字のモーショングラフィックスは短いアクセントに限定する。\n文字自体はフラットな2D。\n空間の立体感はカメラ、距離、パララックス、前後関係によって表現する。",
    "sourceUrl": "https://x.com/studio_oneroom/status/2091485010515914784",
    "sourceLabel": "X 原帖",
    "author": "studio_oneroom",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091485010515914784.mp4",
    "posterUrl": "/posters/x/2091485010515914784.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/studio_oneroom/status/2091485010515914784"
  },
  {
    "id": "x-2091464848848846983",
    "title": "条纹发色角色的眼神特写",
    "titleEn": "Expressive Close-ups of a Striped-Hair Character",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "连续眼部与面部特写表现动漫角色的表情变化。",
    "summaryEn": "Eye and face close-ups show changes in an anime character's expression.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2091464848848846983",
    "sourceLabel": "X 原帖",
    "author": "Mayz1169",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091464848848846983.mp4",
    "posterUrl": "/posters/x/2091464848848846983.jpg",
    "duration": 13.235374,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091449476242440387",
    "title": "水墨汉字与霓虹街头舞步",
    "titleEn": "Ink Calligraphy Meets Neon Street Movement",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑色笔触、粉色光带和巨型汉字围绕人物动作变换。",
    "summaryEn": "Black brushstrokes, pink light trails and giant characters shift around a performer.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2091449476242440387",
    "sourceLabel": "X 原帖",
    "author": "tokyo_Valentine",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091449476242440387.mp4",
    "posterUrl": "/posters/x/2091449476242440387.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "风格化与实验",
      "动态图形",
      "人物表演",
      "城市街景"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091401178425008256",
    "title": "黑裙与宫殿光影美妆概念片",
    "titleEn": "A Black Dress and Palace Light Beauty Concept",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "城市街景切换至华丽宫殿，以黑裙人物和产品片尾组织美妆概念。",
    "summaryEn": "City streets transition into an ornate palace around a black-dress beauty concept and product end card.",
    "prompt": "A cinematic luxury beauty commercial, ultra-realistic, high-end fashion film style.\n\nA stunning young woman with long dark wavy hair, flawless skin, elegant makeup, wearing a sleek black strapless evening gown and sparkling diamond jewelry. She walks confidently down a sunlit European cobblestone street lined with grand classical buildings at golden hour. Soft sunlight flares create a dreamy, glamorous atmosphere.\n\nClose-up of her silver sparkling high heels stepping elegantly on the stone pavement.\n\nShe turns and looks at the camera with a soft, confident expression.\n\nThe scene transitions from bright daytime street to soft dusk, then to night. Elegant classical buildings glow with warm window lights under a deep blue evening sky.\n\nShe walks away from the camera through a grand, luxurious palace hallway with checkered marble floors, crystal chandeliers, and soft candlelight.\n\nShe looks back over her shoulder with a seductive, elegant gaze in warm candlelight.\n\nOn-screen elegant white English text appears smoothly:\n\n- “The light,”\n- “never leaves the shade”\n- “Until the end of the longest day”\n\nFinal product shot: A black luxury cushion compact with silver logo on a dark wooden table next to a golden mirror frame. Elegant text appears:\n\nAll Day & Night\nHERA BLACK CUSHION\n\nShe stands under a massive golden chandelier in a grand hall, gently touching her face with a soft, confident expression. Text remains:\n\nAll Day & Night\nHERA BLACK CUSHION\n\nSoft cinematic lighting, shallow depth of field, film grain, luxurious and refined mood, 4K, highly detailed, smooth camera movements.",
    "sourceUrl": "https://x.com/noorlewisx/status/2091401178425008256",
    "sourceLabel": "X 原帖",
    "author": "noorlewisx",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091401178425008256.mp4",
    "posterUrl": "/posters/x/2091401178425008256.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/noorlewisx/status/2091401178425008256"
  },
  {
    "id": "x-2091362290448925035",
    "title": "闪光灯中的礼服秀场",
    "titleEn": "A Gown Runway under Camera Flashes",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "礼服模特穿过摄影师包围的秀场，镜头从全身切至面部。",
    "summaryEn": "A gown model walks through a runway lined with photographers as shots move from full length to close-up.",
    "prompt": "FORMAT: 15 seconds | 3:4 | 4K | 24fps | cinematic live-action fashion film | ultra-realistic | high-end runway atmosphere\n\nSCENE & STORY:\n\n0–3s — THE ARRIVAL\nA stunning young woman steps onto a luxurious fashion runway wearing an elegant, floor-length couture gown with a dramatic silhouette, refined fabric movement, and subtle sparkling details. The venue is packed with photographers and fashion guests. Camera flashes begin exploding from both sides as she makes her entrance. Slow-motion fabric movement, cinematic lighting, shallow depth of field.\n\n3–7s — ALL EYES ON HER\nShe confidently walks down the runway, maintaining perfect posture and a powerful, effortless expression. Dozens of photographers rapidly capture her from both sides. Flashbulbs illuminate her face and gown, creating a glamorous paparazzi-like atmosphere. Camera tracks backward smoothly in front of her while the gown flows naturally with every step.\n\n7–11s — THE STAR MOMENT\nCut between close-up shots of her confident eyes, elegant face, jewelry, and the intricate details of the gown. Photographers lean forward trying to capture the perfect shot. She briefly turns her head toward the cameras with a subtle, captivating smile. The crowd's attention is completely focused on her.\n\n11–15s — ICONIC FINAL POSE\nShe reaches the end of the runway and stops dramatically. The camera makes a smooth cinematic push-in as she turns slightly, allowing the gown to sweep beautifully around her. A burst of camera flashes fills the scene. She holds a powerful, elegant pose with absolute confidence as the screen ends on her striking expression.\n\nVISUAL STYLE: ultra-luxurious fashion editorial, sophisticated runway lighting, realistic skin texture, natural hair movement, premium couture photography, cinematic depth of field, dynamic camera movement, realistic crowd reactions, dramatic flash photography, glamorous atmosphere, flawless continuity, attention-grabbing opening and unforgettable final pose.\n\nIMPORTANT: Keep the woman’s appearance, facial identity, hairstyle, body proportions, gown, and accessories perfectly consistent throughout all shots. Natural human movement, realistic walking mechanics, no flickering, no distorted hands, no facial changes, no identity drift.",
    "sourceUrl": "https://x.com/codewithhajra/status/2091362290448925035",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091362290448925035.mp4",
    "posterUrl": "/posters/x/2091362290448925035.jpg",
    "duration": 15.232,
    "aspectRatio": "68:45",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2091362290448925035"
  },
  {
    "id": "x-2091352223968096561",
    "title": "黄色背景上的剪影舞蹈与动感字",
    "titleEn": "Silhouette Dance and Kinetic Type on Yellow",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "黑色人物剪影在黄色画面中起舞，MOVE 与 BEAT 字样参与构图。",
    "summaryEn": "A black silhouette dances against yellow while MOVE and BEAT become part of the composition.",
    "prompt": "スタイリッシュなモーショングラフィック＆キネティック・タイポグラフィー動画を制作する。\n\n主人公は、全身が黒一色で描かれたシンプルな人間のシルエット。人物を常に画面中央に配置し、頭から足先まで全身が見える正面中心の構図を維持する。顔、髪、服装などの細部は描かず、滑らかで美しい人体のシルエットとして表現する。背景は明るい色を中心に構成し、人物との鮮明なコントラストを保つ。\n\n主人公は力強いエレクトロニック・ミュージックのビートに合わせて、ポッピング、ロッキング、細かなフットワーク、ボディウェーブ、腕を大きく使った動きなどを組み合わせた、スタイリッシュでキレのあるダンスを披露する。足のステップ、腕の動き、身体の静止、ポーズの切り替えを、キック、スネア、ベースなどの音に正確にシンクロさせる。15秒間、動きが途切れない一続きのダンスとして見せる。\n\nダンサーの動きに合わせて、背景の色、図形、線、円、グリッド、光、パターンが音楽のビートごとに次々と変化する。腕を振ると背景に鋭いラインが走り、足を踏み込むと床から円形の波紋が広がり、身体を止めると幾何学模様も一瞬静止する。身体を大きく動かす場面では、背景全体が回転、伸縮、分割、反転し、ダンスと背景が一体化しているように見せる。\n\n背景には「DANCE」「MOVE」「BEAT」という単語を、正確なスペルの大文字でキネティック・タイポグラフィーとして登場させる。文字はダンサーの背後を横切ったり、奥から手前へ流れたり、ビートに合わせて拡大、縮小、回転、分割、反復したりする。各単語は登場時に明確に読める形を保ち、これら以外の文字は表示しない。文字を人物の前に重ねすぎず、黒いシルエットを常に見やすく保つ。\n\nカメラは基本的にダンサーの全身を正面から捉えながら、ビートに合わせた短いズームイン、ズームアウト、左右への移動、わずかな傾きや回り込みを加える。人物は常に画面中央から大きく外れず、背景だけが目まぐるしく変化する。ダンサーの形状、体格、手足の本数を15秒間一貫させ、手足の変形、残像による分裂、不自然な動きは発生させない。\n\n大胆で鮮やかな配色、強いコントラスト、テンポの速いグラフィック変化、正確なビートシンクを組み合わせる。終盤に向けてダンス、背景、タイポグラフィーの動きを徐々に激しくし、最後は強いビートと同時に主人公が印象的な決めポーズを取り、すべてのグラフィックが一瞬ぴたりと静止して終了する。SNSで拡散されそうな、リズミカルで中毒性があり、何度も見返したくなる高品質な映像に仕上げる。",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2091352223968096561",
    "sourceLabel": "X 原帖",
    "author": "Naonekozamurai",
    "publishedAt": "2026-08-23",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091352223968096561.mp4",
    "posterUrl": "/posters/x/2091352223968096561.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "舞蹈",
      "动态图形"
    ],
    "category": "dance",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2091352223968096561"
  },
  {
    "id": "x-2091280502086865400",
    "title": "角色设定板与登场动画对照",
    "titleEn": "A Character Sheet beside Its Intro Animation",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "上下画面对照角色设定素材与走廊中的人物动画。",
    "summaryEn": "Stacked views compare character reference material with an animated corridor performance.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2091280502086865400",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091280502086865400.mp4",
    "posterUrl": "/posters/x/2091280502086865400.jpg",
    "duration": 15.166667,
    "aspectRatio": "540:607",
    "resolution": "1080p",
    "tags": [
      "本地生成与工作流",
      "动漫",
      "人物表演"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091233403353936255",
    "title": "猿群与荒原黑色巨碑",
    "titleEn": "Apes Encounter a Black Monolith",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "猿群围绕荒原中的黑色巨碑，场景随后展开为宏大的科幻景观。",
    "summaryEn": "Apes gather around a black monolith before the setting opens into a vast science-fiction landscape.",
    "prompt": "Use the reference image as the exact first frame, keep the same uncanny prehistoric sci-fi look, keep the black server monolith there and keep the apes there, the camera is handheld and low like a nervous witness hiding among them, move slowly toward the monolith as the apes circle it and start making a bigger scene, then one ape slams both hands on it and the whole thing wakes up in a way that feels way too powerful, as the camera rushes closer the black grid starts glowing from inside and weird light leaks through it, autofocus breathes and slips and catches again while the apes lose their minds and start screaming and jumping and running past the lens, then the monolith opens or unfolds or does something impossible and the camera gets pulled forward into whatever is happening next, as we move around it the desert starts changing fast, rocks splitting open, giant metal structures rising out of the ground, dust storms mixing with sparks and fog and strange debris, then the apes start evolving or mutating or turning into something else while still running around, and the camera keeps chasing the next insane thing it sees, then somehow there’s a huge moon hanging way too low, then a broken skyline in the distance, then a river of glowing data or fire or whatever, while the camera just keeps pushing through like one continuous shot and never cuts, handheld shakes, accidental zooms, blur, focus hunting, moments where the lens gets too close to fur or rock or the monolith surface, then things go fully off the rails and the whole world starts feeling like the birth of civilization and the end of civilization at the same time, keep it messy and escalating and dreamlike, don’t over-explain, just keep discovering crazier stuff as the camera moves, you know what I mean, just make it",
    "sourceUrl": "https://x.com/cocktailpeanut/status/2091233403353936255",
    "sourceLabel": "X 原帖",
    "author": "cocktailpeanut",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091233403353936255.mp4",
    "posterUrl": "/posters/x/2091233403353936255.jpg",
    "duration": 14.464,
    "aspectRatio": "17:11",
    "resolution": "704p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "科幻太空",
      "动物与生物"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "creatures"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/cocktailpeanut/status/2091233403353936255"
  },
  {
    "id": "x-2091161609913921655",
    "title": "三人对峙的多镜头画面",
    "titleEn": "Multiple Shots of a Three-Person Standoff",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "同一组人物以远景、面部和持枪细节形成多镜头对峙画面。",
    "summaryEn": "Wide views, faces and weapon details build a multi-shot standoff around the same characters.",
    "prompt": "Make a Multi-Shot video",
    "sourceUrl": "https://x.com/jerrod_lew/status/2091161609913921655",
    "sourceLabel": "X 原帖",
    "author": "jerrod_lew",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091161609913921655.mp4",
    "posterUrl": "/posters/x/2091161609913921655.jpg",
    "duration": 5.22449,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "对白与口型"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/jerrod_lew/status/2091161609913921655"
  },
  {
    "id": "x-2091114406839251020",
    "title": "同一角色穿越多种动画风格",
    "titleEn": "One Character across Animation Styles",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黏土质感、绘本页面与角色集合呈现多种动画表现方式。",
    "summaryEn": "Clay textures, storybook pages and a character ensemble present several animation treatments.",
    "prompt": null,
    "sourceUrl": "https://x.com/NVTDanh/status/2091114406839251020",
    "sourceLabel": "X 原帖",
    "author": "NVTDanh",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091114406839251020.mp4",
    "posterUrl": "/posters/x/2091114406839251020.jpg",
    "duration": 15.11619,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "风格化与实验",
      "3D / CG",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "stylized",
      "cg-3d"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091103106746442226",
    "title": "液态金属凝聚成 EVOLVE",
    "titleEn": "Liquid Metal Forms EVOLVE",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "银色液滴与流体表面逐渐组成反光立体字。",
    "summaryEn": "Silver droplets and fluid surfaces assemble into reflective dimensional lettering.",
    "prompt": "Ultra cinematic title sequence: Rivers of liquid mercury flow through a futuristic black environment, reflecting impossible light. Metallic droplets rise against gravity, merging and separating with mesmerizing precision. The liquid chrome begins assembling itself molecule by molecule into the word \"EVOLVE\", every surface behaving like a living mirror. The title constantly morphs, flowing and reshaping itself before suddenly collapsing into millions of reflective droplets that scatter across the screen like liquid stars. Territory Studio aesthetics, premium sci-fi typography, impossible fluid dynamics.",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2091103106746442226",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091103106746442226.mp4",
    "posterUrl": "/posters/x/2091103106746442226.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2091103106746442226"
  },
  {
    "id": "x-2091072883040600420",
    "title": "剪影跑酷穿越黑白字形",
    "titleEn": "Silhouette Parkour through Black-and-White Type",
    "model": "MiniMax H3",
    "mode": "T2VA",
    "summary": "黑色人物越过立体字形，红色线条引导动作路线。",
    "summaryEn": "A black silhouette jumps across dimensional letters as red lines trace its route.",
    "prompt": "黒一色で描かれた、シンプルな人間のシルエットを主人公にする。人物はカメラに向かって走ったり、画面右方向へ高速で駆け抜けたりしながら、障害物を軽やかに越えるヴォルト、壁を利用したウォールラン、段差から段差への正確な移動、スムーズな着地など、流れるようなパルクールの動きを連続して披露する。\n\n人物の足音、踏み切り、障害物への接触、着地などの動きを、音楽の力強いビートに正確にシンクロさせる。動作のタイミングに合わせて、背景に現れる建物、足場、幾何学模様、タイポグラフィーが次々と形や色を変化させる。タイポグラフィーはグラフィック要素として使用し、特定の文章や意味のある単語は表示しない。\n\n15秒間を通して背景と画面構成を目まぐるしく変化させ、テンポの速いトランジション、大胆な配色、強い視覚的インパクトを取り入れる。人物のシルエットは常に黒で統一し、背景との鮮明なコントラストを維持する。SNSで拡散されそうな、リズミカルで中毒性があり、何度も見返したくなる高品質な映像に仕上げる。",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2091072883040600420",
    "sourceLabel": "X 原帖",
    "author": "Naonekozamurai",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091072883040600420.mp4",
    "posterUrl": "/posters/x/2091072883040600420.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动作特效",
      "动态图形",
      "人物表演"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2091072883040600420"
  },
  {
    "id": "x-2091039543046943010",
    "title": "夕阳下穿过人群的轮鞋表演",
    "titleEn": "A Wheeled-Shoe Performance at Sunset",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物穿上带轮鞋款，在围观人群之间滑行并转身。",
    "summaryEn": "A performer puts on wheeled shoes and glides and turns among a gathered crowd.",
    "prompt": "Create a 15-second realistic cinematic video featuring a young woman wearing modern wheel shoes with visible small wheels under the soles, moving confidently through a large outdoor urban plaza. At the beginning, she is preparing and adjusting her wheel shoes while a crowd of people stands around in the background. She then stands up and slowly starts rolling forward, gradually gaining speed and smoothly gliding through the open area. As she moves, more people notice her and gather around, watching with curiosity and excitement. Several girls and young people take out their smartphones to record videos and take selfies with her in the background. The girl continues riding confidently, making smooth turns, playful spins, and stylish movements while maintaining a natural, happy expression. The crowd forms a loose circle around her, smiling, clapping, cheering, recording her performance, and reacting positively. Include a moment where she interacts playfully with another girl while both move around on the wheel shoes. Show dynamic camera angles including close-ups of the wheels rolling, medium shots of the girl performing, tracking shots following her movement, wide shots showing the large crowd, and close-ups of people holding smartphones and taking selfies. Keep the girl's appearance, hairstyle, clothing, face, and wheel shoes consistent throughout all shots. Use realistic human movement, natural body physics, detailed wheel motion, realistic outdoor lighting, cinematic depth of field, natural crowd behavior, and smooth camera transitions. The setting should feel lively and authentic, with modern buildings, an open paved plaza, many people watching from both sides, and warm natural daylight gradually approaching golden hour. End with a wide cinematic shot of the girl confidently rolling away through the crowd while people continue smiling, filming, and cheering. Duration exactly 15 seconds, cinematic realistic style, 16:9 aspect ratio, high detail, smooth motion, natural expressions, handheld cinematic camera movement, no text, no logos, no glitches, no duplicated people, and no distorted hands or faces.",
    "sourceUrl": "https://x.com/ayzalnooor24521/status/2091039543046943010",
    "sourceLabel": "X 原帖",
    "author": "ayzalnooor24521",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091039543046943010.mp4",
    "posterUrl": "/posters/x/2091039543046943010.jpg",
    "duration": 15.136,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "人物表演",
      "城市街景"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ayzalnooor24521/status/2091039543046943010"
  },
  {
    "id": "x-2091034090376139070",
    "title": "日本城市与抹茶旅行蒙太奇",
    "titleEn": "A Japan Travel Montage of Streets and Matcha",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "街巷移动镜头、抹茶近景与东京夜景组成旅行片段。",
    "summaryEn": "Moving street shots, a matcha close-up and Tokyo at night form a travel montage.",
    "prompt": "Fast-paced cinematic journey through Japan, blending energetic hyperlapse sequences with smooth motion-blur transitions and sharp, detailed close-ups. Begin with Tokyo’s glowing nighttime streets, rushing past colorful vending machines, convenience-store aisles, and lively arcade interiors. Cut into crisp food and culture details: beautifully arranged nigiri sushi, a bamboo whisk creating bright green matcha, and a Nara deer glowing under warm evening sunlight. Move through scenic Japanese landscapes with Mount Fuji rising behind a traditional pagoda, followed by sweeping views of vermilion shrine pathways, spring blossoms, and rustic wooden countryside homes. Build toward a spectacular aerial finale over Tokyo after dark, with the illuminated skyline and Tokyo Tower dominating the scene. Dynamic whip pans, radial motion blur, cinematic lighting, rich contrast, vivid colors, ultra-detailed 4K, 16:9 widescreen.",
    "sourceUrl": "https://x.com/DaniaSafvi/status/2091034090376139070",
    "sourceLabel": "X 原帖",
    "author": "DaniaSafvi",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091034090376139070.mp4",
    "posterUrl": "/posters/x/2091034090376139070.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "城市街景",
      "美食"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/DaniaSafvi/status/2091034090376139070"
  },
  {
    "id": "x-2091027170827116570",
    "title": "粉色长裙与飘动羽饰",
    "titleEn": "A Pink Gown with Moving Feather Details",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物站在深色背景前，粉色裙装与羽饰产生轻柔运动。",
    "summaryEn": "A figure stands against a dark background while a pink gown and feather details move gently.",
    "prompt": null,
    "sourceUrl": "https://x.com/TaliaAariz/status/2091027170827116570",
    "sourceLabel": "X 原帖",
    "author": "TaliaAariz",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091027170827116570.mp4",
    "posterUrl": "/posters/x/2091027170827116570.jpg",
    "duration": 15.168,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "综合展示",
      "照片级写实",
      "时尚"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091027098920157357",
    "title": "发光水母漂过深海教堂",
    "titleEn": "Glowing Jellyfish in an Underwater Cathedral",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "蓝紫色水下建筑中，发光水母穿过拱顶与光束。",
    "summaryEn": "Glowing jellyfish drift through arches and light beams in a blue-purple underwater structure.",
    "prompt": null,
    "sourceUrl": "https://x.com/TechTalkNAVI/status/2091027098920157357",
    "sourceLabel": "X 原帖",
    "author": "TechTalkNAVI",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091027098920157357.mp4",
    "posterUrl": "/posters/x/2091027098920157357.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091026087254036508",
    "title": "小狗撑伞陪伴雨中的猫",
    "titleEn": "A Puppy Shares an Umbrella with a Cat",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "雨中木台上的小狗用红伞遮住靠近的猫，形成温暖的动物片段。",
    "summaryEn": "A puppy on a rainy wooden platform shelters a nearby cat beneath a red umbrella.",
    "prompt": null,
    "sourceUrl": "https://x.com/MrDasOnX/status/2091026087254036508",
    "sourceLabel": "X 原帖",
    "author": "MrDasOnX",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091026087254036508.mp4",
    "posterUrl": "/posters/x/2091026087254036508.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "照片级写实",
      "动物与生物",
      "日常生活"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091022062404137221",
    "title": "墨色荒原中的持刀角色",
    "titleEn": "A Sword-Bearing Character in an Ink Landscape",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "持刀角色缓步穿过冷色荒原，片尾转为水墨式角色标题。",
    "summaryEn": "A sword-bearing character walks across a cool-toned landscape before an ink-style title card.",
    "prompt": null,
    "sourceUrl": "https://x.com/ManuAGI01/status/2091022062404137221",
    "sourceLabel": "X 原帖",
    "author": "ManuAGI01",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091022062404137221.mp4",
    "posterUrl": "/posters/x/2091022062404137221.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "奇幻"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091019054597431652",
    "title": "汉堡食材组成平面广告",
    "titleEn": "Burger Ingredients Build a Graphic Ad",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "芝麻、面包与馅料在米黄和红色平面中组合成汉堡标识。",
    "summaryEn": "Sesame, buns and fillings assemble into a burger mark across cream and red graphic panels.",
    "prompt": null,
    "sourceUrl": "https://x.com/ryo05m/status/2091019054597431652",
    "sourceLabel": "X 原帖",
    "author": "ryo05m",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091019054597431652.mp4",
    "posterUrl": "/posters/x/2091019054597431652.jpg",
    "duration": 15.168,
    "aspectRatio": "7:4",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "动态图形",
      "插画与手绘",
      "美食"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false
  },
  {
    "id": "x-2091013502756606059",
    "title": "金属字形组成视觉对白",
    "titleEn": "Metal Letterforms Build a Visual Dialogue",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "反光金属字形通过推近、旋转和排列组成文字交流。",
    "summaryEn": "Reflective metal letterforms use close-ups, rotation and arrangement to create a typographic exchange.",
    "prompt": "【CONDITION DEFINITION】\n15秒、1:1スクエア、参照画像なし。高品質なシネマティック混合表現。主役は立体モーショングラフィックス、空間タイポグラフィ、サウンド。人物、顔、口、手、身体、人影を一切出さない。文字とMGは厚い半透明の鉱物ｍガラス外殻、内部の濃い墨色層、精密な金属骨格を持つ同一の立体物質。側面、内側、空洞、断面、接合部、研磨跡、微細な傷、濃い接触影、透過光、落影を明瞭に描く。全時間で形状、厚み、材質、断面、空洞、前後関係、遮蔽、可読性、役割のいずれかを更新し、同じ運動を反復しない。全CUTを明るい鉱物床、床上の黒い積層基壇、床へ固定された人間寸法の細い手すりと目地が連続する一つの建築空間で描く。文字の光、影、反射を床と積層基壇へ一致させ、背景を切り替えない。\n\n【SHOT / FLOW】\nCUT1（0.00–2.30秒）：開始から巨大な立体“NOW?”が前景・中景・遠景を貫く斜め構図で完全可読。Oの空洞の奥に同じ床と細い固定手すりが小さく見える。最初の衝突で厚い外殻が断面を見せて裂け、墨色層と金属骨格が別々の奥行きへ伸び、N、O、W、?が質量を交換する。\nCUT2（2.30–4.75秒）：真上から見た文字の立体地形。2.30秒、“DO YOU SEE ME NOW?”全文が段状の奥行きで初めて完全可読になる瞬間に、女性が同じ英文を一度だけ短く自然に話し始める。直後に各語の外殻、内部層、空洞が別方向へ開き、前CUTから受け取った質量で新しい高低差と接触影を作る。巨大文字が同じ床の目地をまたいで前後を通り、全文を保持しない。\nCUT3（4.75–7.20秒）：文字断面の極端なマクロ。ガラス外殻の傷、墨色層、金属骨格、空洞を同時に見せる。摩擦音で厚い層が横滑りし、内部骨格が噛み合い、可読だった質問を奥行き方向の別構文へ組み替える。断面の空洞が同じ床、積層基壇、細い固定手すりを含む次CUTを露出する。\nCUT4（7.20–9.75秒）：文字の空洞越しに同じ床と人間寸法の細い固定手すりを識別できる斜め中景。巨大文字がその前後を通り、落影と透過光が手すりと床へ移る。外殻、墨色層、骨格、空洞は質問から返答へ向かう同じ組替えを続け、固定建築要素を単独で見せない。\nCUT5（9.75–12.35秒）：低い横位置。全文の立体部品が三つの奥行きで同時に噛み合う。外殻が骨格へ質量を渡し、空洞が閉じる場所と別の語が開く場所を同時に作る。同じ床が巨大接合部の下まで連続する。11.20秒、同じ部品が“I SEE YOU.”として初めて完全可読になる瞬間に、男性が同じ英文を一度だけ短く自然に話し始める。\nCUT6（12.35–15.00秒）：前CUTで成立した返答を同じ文字建築のまま引き継ぐ。Iは内部骨格を手前へ伸ばし、SEEは外殻と墨色層を左右へ交換し、YOUはOの空洞を同じ床の奥まで開き、前景と遠景へ質量を分けて再結合する。質問全文の残存部品、床の落影、空間反響を同時更新し、15.00秒まで可読と変形を両立する。\n\n【CAMERA / EDITING】\n斜め深度構図、真俯瞰、断面マクロ、空洞越し中景、低い横位置、正面寄り深度構図へ毎CUT交代。文字の厚みと前後遮蔽を常に見せる。接続は外殻の断面、受け渡す質量、文字の空洞、噛み合う骨格によるマッチカットとハードカット。固定手すりや目地を中央、最大要素、単独被写体にしない。全CUTを開始フレームから同じ建築空間に属する異なる完成画面と変形中のクライマックスにする。\n\n【SOUND】\n歌唱なし。開始から完成した現代的インストBGMを通す。女性は2.30秒、男性は11.20秒にだけ対応英文を自然に話し始め、二人の声を重ねない。乾いた打音、深い低音、石とガラスの衝突、金属骨格の共鳴、摩擦、空気圧、近距離音と遠距離反響で時間軸を作る。各CUTへ固有の音響事件を与え、材質、重量、距離、変形を音で確定する。声を編集テンポにせず、最終CUTだけへ強度を温存しない。\n\n【NEGATIVE】\n人物、顔、口、手、身体、人影、シルエット、実写人物、低品質な3D人物を禁止。黒い無背景、孤立ロゴ、商品撮影台、商品写真、可動する小道具、台車、球、ピンを禁止。フラットベクター、薄い平面文字、正面字幕、輪郭線だけの文字、表面の色替えだけの変化、色面、矩形、クロップ窓、リボン、線、円弧、光帯、反射軸、UI、ネオン、粒子を主役にしない。安価なプラスチック文字、均一なクローム、汎用ベベル、既製ロゴアニメを禁止。歌、コーラス、母音伸ばし、文字の単純な出消し・拡大・クロップ、同一運動のループ、全文保持、フェード、完全停止を禁止。",
    "sourceUrl": "https://x.com/ai_lifehack55/status/2091013502756606059",
    "sourceLabel": "X 原帖",
    "author": "ai_lifehack55",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091013502756606059.mp4",
    "posterUrl": "/posters/x/2091013502756606059.jpg",
    "duration": 15.13941,
    "aspectRatio": "1:1",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/ai_lifehack55/status/2091013502756606059"
  },
  {
    "id": "x-2091012158280917271",
    "title": "沙漠公路夕阳汽水概念广告",
    "titleEn": "A Sunset Desert Road Soda Concept",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "牛仔裤、复古汽车与冰凉汽水特写构成夕阳公路广告。",
    "summaryEn": "Denim, a vintage car and cold soda close-ups form a sunset road advertising concept.",
    "prompt": "Create a 15-second cinematic Pepsi commercial-style video set on an empty desert highway during golden hour. A stylish young woman in a western-inspired outfit and cowboy boots walks toward a classic white vintage convertible parked beside a remote gas station. Show dramatic low-angle shots of her boots, the car’s chrome wheels, and her confident walk.\n\nCut to an extreme close-up of her face, wind blowing through her hair, with warm sunlight and a confident expression. Show a chilled Pepsi glass bottle glowing in the sunset light, covered in realistic condensation, with the Pepsi logo clearly visible.\n\nThen show the woman sitting in the convertible while holding the Pepsi bottle as her scarf flows dramatically in the wind. Finish with a wide cinematic aerial shot of the vintage car speeding down the long empty desert road toward the setting sun, leaving a subtle trail of dust.\n\nStyle: photorealistic, premium Pepsi commercial, cinematic storytelling, warm golden-hour lighting, anamorphic lens, shallow depth of field, realistic skin and fabric, smooth camera movements, dramatic close-ups, rich contrast, subtle film grain, 4K, high-end advertising aesthetic, 9:16 vertical.",
    "sourceUrl": "https://x.com/CaliraVal/status/2091012158280917271",
    "sourceLabel": "X 原帖",
    "author": "CaliraVal",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2091012158280917271.mp4",
    "posterUrl": "/posters/x/2091012158280917271.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "产品与广告",
      "时尚"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CaliraVal/status/2091012158280917271"
  },
  {
    "id": "x-2090985277653483549",
    "title": "云海仙宫与崖边人物环绕镜头",
    "titleEn": "An Orbit around a Figure above a Sea of Clouds",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "镜头穿过松枝与云雾，环绕仙宫前崖边的红衣人物。",
    "summaryEn": "The camera passes branches and clouds while circling a red-robed figure beside a celestial palace.",
    "prompt": "镜头慢慢推进，镜头慢慢环绕，最后定格在这个仙人人物的身上，云雾快速翻腾，人物衣服随风飘扬。",
    "sourceUrl": "https://x.com/PixelAigc/status/2090985277653483549",
    "sourceLabel": "X 原帖",
    "author": "PixelAigc",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090985277653483549.mp4",
    "posterUrl": "/posters/x/2090985277653483549.jpg",
    "duration": 7.36,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "3D / CG",
      "奇幻",
      "自然风景"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/PixelAigc/status/2090985277653483549"
  },
  {
    "id": "x-2090975138162581938",
    "title": "巨型字切换的动漫角色宣传片",
    "titleEn": "An Anime Character Promo with Giant Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "黑白巨型文字与角色镜头快速交替，组成动画宣传段落。",
    "summaryEn": "Oversized black-and-white lettering alternates with character shots in an anime promotional sequence.",
    "prompt": "## 参照キャラクター指定\n\n* ** Image1**：ピンク髪の少女。冒頭と終盤のメイン。作品の顔。\n\n* ** Image2**：ダーク衣装の銀髪少女。影・異質感・変質表現担当。\n\n* ** Image3**：白銀ショートヘアの少女。`KNIGHT` / `RIFT` 担当。\n\n* ** Image4**：金髪の白衣装の少女。収束・静けさ・終盤象徴カット担当。\n\n## 全体指示\n\n**15秒、16:9、横長。**\n\n白・黒・赤を基調にした、極太の縦長サンセリフタイポグラフィと4人のファンタジー系アニメキャラクターを組み合わせた高密度グラフィックPV。\n\n文字は字幕ではなく、**空間レイヤー、遮蔽、トンネル、マスク、画面構造そのもの**として機能する。\n\nアニメ調の細線キャラクター、フラット寄りの陰影、強いグリッチ、横走査線、赤い円形UI、色収差、短いフラッシュ切り替え。\n\n全体は、\n\n**静かな人物提示 → タイポ変形 → キャラクター切り替え → レイピアによる空間切断 → 圧縮空間 → 崩壊 → 収束 → 冒頭回帰**\n\nの流れ。\n\nテンポを落とさず、**0.5秒ごとに画面構造・人物・文字・動きのいずれかを必ず変化させる。**\n\n## 使用文字\n\n* **`FANTASY`**\n\n* **`KNIGHT`**\n\n* **`SIGIL`**\n\n* **`FACE`**\n\n* **`LINE`**\n\n* **`RIFT`**\n\n---\n\n## 0.0–0.5秒\n\n白背景。\n\n巨大な **`FANTASY`** が画面いっぱいに並ぶ。\n\n右上に ** Image1** の横顔が少しだけ見切れる。\n\n薄い水平グリッチ。\n\n---\n\n## 0.5–1.0秒\n\n**`FANTASY`** が左へ高速スライド。\n\nImage1 は一度消え、白地と巨大な黒文字だけになる。\n\n文字の一部が画面外にはみ出す。\n\n---\n\n## 1.0–1.5秒\n\n**`FANTASY`** の文字の隙間から ** Image1** 現れる。うつむき気味で目を閉じている。文字が人物の前後を通り、身体の一部を遮蔽する。---## 1.5–2.0秒 Image1 がわずかに顔を上げて目を開く。顔の輪郭と瞳が明瞭になる。同時に **`FANTASY`** が横方向へ圧縮され始める。---## 2.0–2.5秒短い白フラッシュ。巨大な **`FACE`** に切り替わる。人物は消え、中央に赤黒の同心円UI。強い水平グリッチと色収差。---## 2.5–3.0秒**`KNIGHT`** が巨大に出現。その文字の隙間から ** Imag3** が正面寄りに現れる。白銀の髪と青緑系衣装を明確に見せる。---## 3.0–3.5秒 Image3 が右向きの横顔になる。**`KNIGHT`** が人物の前後を横切りながら右へ流れる。赤い細線UIが一瞬走る。---## 3.5–4.0秒**`KNIGHT LINE`** に切り替わる。人物はいったん消える。白黒で画面を大胆に分割し、赤を一点だけ強く使う。---## 4.0–4.5秒巨大な **`LINE`** が横方向へ引き伸ばされる。文字の一部が極端に細くなり、一部は太く圧縮される。画面に強い横方向の張力を作る。---## 4.5–5.0秒背景が黒寄りへ反転。細いアウトラインの **`SIGIL`**。中央に ** Image2**。背後または隣に、** Image1の薄いピンク色のシルエット**が背中合わせ気味に重なる。---## 5.0–5.5秒 Image2 と  Image1 のシルエット位置がわずかにずれる。多重露光のような残像。**`SIGIL`** も複数レイヤーへ分離する。---## 5.5–6.0秒 Image1 のシルエットが消える。** Image2** が画面中央へ集約。背後に淡い赤色の円形UI。次のアクションへ向けて一瞬静止。---## 6.0–6.5秒高速フラッシュで **`RIFT`** に切り替わる。** Image3** が細身のレイピアを構える。白背景、青い斜めライン、赤い円形UI。文字は人物の前後に食い込んでいる。---## 6.5–7.0秒 Image3がレイピアを鋭く横薙ぎする。剣先の軌跡に沿って、**画面空間そのものが切り裂かれる。**白い斬撃線と赤い発光。**`RIFT`** の中央も切断される。---## 7.0–7.5秒 Image3が一歩踏み込み、二撃目。レイピアを裂け目へ突き込む。切断された **`RIFT`** の文字が上下または前後にずれる。青いラインも同時に切断される。---## 7.5–8.0秒レイピアの軌跡から黒い亀裂が一気に拡大。**`RIFT`** が歪み、文字としての形を失い始める。赤いデジタルノイズと色収差が急増する。---## 8.0–8.5秒破断した画面が一気に暗部へ吸い込まれる。ほぼ暗転。わずかな赤い残光と水平ノイズだけを残す。---## 8.5–9.0秒暗闇の中央に細い縦スリット。その中に ** Image2** が小さく現れる。左右は黒いマスク。人物だけが圧縮された異界空間に閉じ込められているように見せる。---## 9.0–9.5秒縦スリットが少し広がる。 Image2 の顔と衣装が明瞭になる。仮面・儀式・異界を連想させる演出。赤い円弧が背後に出現する。---## 9.5–10.0秒背後に赤い円形UI。薄い **`LINE`** が人物と重なる。 Image2は中央に静止。奥へ続くグリッドが現れ始める。---## 10.0–10.5秒**`LINE`** が立体的に多重化。文字とフレームがトンネルのように奥へ連続する。 Image2 を中央に残し、背景だけが高速で後退する。---## 10.5–11.0秒トンネルがさらに深くなる。赤い線と白い線が奥へ吸い込まれる。文字・UI・グリッドの境界が徐々に崩れる。---## 11.0–11.5秒左から巨大な白いスラッシュ状エフェクト。画面全体を横断する。**`LINE`** と空間フレームが押し潰され、赤いグリッチが急増。---## 11.5–12.0秒 Image2 がグリッチに埋もれて消える。**`SIGIL` / `LINE`** が波打ちながら崩壊。水平の赤いノイズが画面を横切る。---## 12.0–12.5秒黒太字の **`SIGIL`** が円形に配置される。中央に渦巻き状のトンネルUI。その中央から ** Image4** が静かに現れる。---## 12.5–13.0秒円形 **`SIGIL`** が拡大。渦巻きが楕円状に伸びる。 Image4 はほぼ静止したまま、周囲だけが激しく変形する。清浄で象徴的な印象。---## 13.0–13.5秒画面全体が横方向へ強く引き伸ばされる。 Image4と渦巻きが光の残像になり、白へ溶ける。短いフラッシュで終盤へ。---## 13.5–14.0秒再び巨大な **`FANTASY`**。** Image1** が右下寄りに現れる。冒頭と同じ構図を思わせるが、今度は正面寄り。---## 14.0–14.5秒**`FANTASY`** が左へ流れる。 Image1 は静かに目を閉じ、わずかにうつむく。右下に赤いノイズテキスト状グリッチ。---## 14.5–15.0秒 Image1が右方向へ画面外へ寄る。**`FANTASY`** も細く圧縮されて端へ消える。白背景。最後に短い赤い水平グリッチだけが残り、完全に消えて終了。---# 画面演出ルール* **15秒を超えない。*** 0.5秒単位で明確な視覚変化を作る。* 同じ人物・同じ構図を長時間固定しない。* タイポグラフィは常に人物の前後へ入り込み、字幕のように画面端へ置くだけにしない。* 文字そのものを**壁、マスク、トンネル、切断対象、空間構造**として扱う。* 白黒赤を全体の共通色とする。* キャラクター固有色はアクセントとして使用する。  * ** Image1：ピンク**  * ** Image2：紫黒**  * ** Image3：青緑**  * ** Image4：白金*** ** Image3 の `RIFT` 区間は中盤最大のアクションピーク。*** レイピアは敵を攻撃するのではなく、**文字、UI、画面、空間そのものを切断する。*** 斬撃によって発生した亀裂を、その後の圧縮空間・トンネル・崩壊演出へ因果的につなげる。*  Image4は戦闘させず、激しい崩壊後の静かな収束点として扱う。* 最後は  Image1に戻し、冒頭と終盤を視覚的に接続する。## 音* 強い電子ビート。* 文字切り替えごとの短いデジタルノイズ。* レイピアの斬撃には鋭い金属音＋デジタルスラッシュ。* 空間亀裂には低い衝撃音。* 8.0秒付近で一瞬音数を減らす。* 8.5秒以降から低音を再び立ち上げる。* 13.5秒の冒頭回帰でビートを強く戻す。* 15.0秒で短く明確に終了する。",
    "sourceUrl": "https://x.com/haruuraeadss/status/2090975138162581938",
    "sourceLabel": "X 原帖",
    "author": "haruuraeadss",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090975138162581938.mp4",
    "posterUrl": "/posters/x/2090975138162581938.jpg",
    "duration": 15.232,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/haruuraeadss/status/2090975138162581938"
  },
  {
    "id": "x-2090969477588427222",
    "title": "分镜板指导动漫料理过程",
    "titleEn": "A Storyboard Guides an Anime Cooking Sequence",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "料理动画与分镜板并排展示，呈现参考面板到连续动作的对应。",
    "summaryEn": "Cooking animation is shown beside its storyboard to compare reference panels with continuous action.",
    "prompt": "Use the storyboard @[storyboard ref] as sequential shot guidance, not as a static image. Do not treat the storyboard as one image. Follow each panel as a separate beat. Use @[character ref] as the character and visual style reference.",
    "sourceUrl": "https://x.com/aimikoda/status/2090969477588427222",
    "sourceLabel": "X 原帖",
    "author": "aimikoda",
    "publishedAt": "2026-08-22",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090969477588427222.mp4",
    "posterUrl": "/posters/x/2090969477588427222.jpg",
    "duration": 15.166667,
    "aspectRatio": "540:607",
    "resolution": "1080p",
    "tags": [
      "本地生成与工作流",
      "动漫",
      "美食"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/aimikoda/status/2090969477588427222"
  },
  {
    "id": "x-2090835533408952656",
    "title": "彩色舞台上的五人偶像音乐片",
    "titleEn": "A Five-Person Idol Video on Colorful Stages",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "五人组合在糖果色布景与灯光舞台间切换，画面加入大字歌词。 原帖公开的是交给 MiniMax Design 的完整创作指令。",
    "summaryEn": "A five-person group moves between candy-colored sets and lit stages with oversized lyric graphics. The published text is the complete creative instruction given to MiniMax Design.",
    "prompt": "・尺：30秒\n・画幅：16:9\n・音楽：お任せ生成\n（オリジナル日本語歌詞ありのMV）\n・人物参考：なし\n（K-POP風ダンスガールグループ5人組でビジュアルお任せ）\n・シーン参考：なし（明るくカラフルなPop ）\n・タイポグラフィ：あり（歌詞抜粋）\n・構成：歌詞に合わせてダンスと文字演出",
    "sourceUrl": "https://x.com/mi7_crypto/status/2090835533408952656",
    "sourceLabel": "X 原帖",
    "author": "mi7_crypto",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090835533408952656.mp4",
    "posterUrl": "/posters/x/2090835533408952656.jpg",
    "duration": 29.884082,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "商业广告感",
      "照片级写实",
      "音乐演出",
      "舞蹈"
    ],
    "category": "music",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/mi7_crypto/status/2090835533408952656"
  },
  {
    "id": "x-2090833693057184058",
    "title": "无人机展开建筑巨幅横幅",
    "titleEn": "A Drone Unfurls a Giant Building Banner",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "无人机在建筑前展开纵向横幅，随后出现烟花。",
    "summaryEn": "A drone unfolds a vertical banner in front of a building, followed by fireworks.",
    "prompt": "a still camera showing a cgi effect creating a drone holding a banner covering the long building in @ Image1 with the banner in @ Image2 , the drone appears from below the building from the building to the sky in hyper realistic way, after the banner appears the fireworks starting in the whole place, the drone keep the banner untill the end without dropping it",
    "sourceUrl": "https://x.com/aziz4ai/status/2090833693057184058",
    "sourceLabel": "X 原帖",
    "author": "aziz4ai",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090833693057184058.mp4",
    "posterUrl": "/posters/x/2090833693057184058.jpg",
    "duration": 10.208333,
    "aspectRatio": "9:16",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "照片级写实",
      "城市街景",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "city",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/aziz4ai/status/2090833693057184058"
  },
  {
    "id": "x-2090816358040625275",
    "title": "热带街头舞蹈与狂欢字体",
    "titleEn": "Tropical Street Dance with Carnival Type",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "彩色街道上的舞者与巨大绿色字体共同推进热带表演片段。",
    "summaryEn": "Dancers on colorful streets move through a tropical sequence with giant green lettering.",
    "prompt": "Generate a 15-second Latin-pop carnival MV in the style of anime characters, sun-drenched coastal street murals, golden hour lens flare, vibrant tropical foliage, warm emerald and terracotta color scheme, English bold brush-stroke dynamic font, uptempo reggaeton afro-caribbean rhythm",
    "sourceUrl": "https://x.com/yourPlugAI/status/2090816358040625275",
    "sourceLabel": "X 原帖",
    "author": "yourPlugAI",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090816358040625275.mp4",
    "posterUrl": "/posters/x/2090816358040625275.jpg",
    "duration": 15.2,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "风格化与实验",
      "动态图形",
      "舞蹈",
      "城市街景"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "dance",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/yourPlugAI/status/2090816358040625275"
  },
  {
    "id": "x-2090773583668621325",
    "title": "霓虹都市动漫音乐片头",
    "titleEn": "A Neon City Anime Music Opening",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "紫蓝霓虹、动漫人物与高速标题切换组成赛博都市片头。",
    "summaryEn": "Purple-blue neon, anime characters and rapid title changes form a cyber-city opening.",
    "prompt": "Generate a 15-second hyper-pop cyber-anime MV opening. Refer to the visual language of these reference images: neon pastel cyberpunk, glowing holographic overlays, futuristic UI HUD elements, high-contrast Y2K digital graphic design, floating liquid-metal typography, speed lines, and bright cyan-magenta-violet gradient color schemes, set to a high-tempo J-pop synthwave rhythm.\n\nThe tone should be 80% energetic cyber-pop and 20% futuristic intrigue: colorful, fast-paced, stylish, and futuristic, without dark horror or gloomy tones. Make it feel like an ultra-modern animated music video intro: glowing laser frames expand first, digital interface grids snap open, and holographic anime character silhouettes, glowing cyber-deck props, and bold floating English credits reveal in precise sync with the upbeat tempo.\n\nEnglish credits and title text must be clear, crisp, and stylishly designed. No random characters or misspellings. Each credit and job title must appear only once.\n\nUse dynamic digital transitions: chromatic aberration glitches, prism light leaks, rapid frame flashes, pixel-sorting wipes, spatial zoom-cuts, and panel-slide reveals. Keep all transitions razor-sharp and beat-driven. No slow blurs or fading dissolves.\nBGM: original 15-second high-energy cyber-pop instrumental, 160 BPM upbeat synthwave rhythm.\n\nFeatures crisp electronic clap-snares, bright arpeggiated synth leads, bouncing sub-bass line, and glowing vocal chops. Build anticipation in the first 2 seconds, drop a heavy synth-bass beat at 3 seconds, introduce bright arpeggios at 7 seconds, hit a high-octane synth melody at 11 seconds, and finish on an echoing synth stab with a shimmering reverb tail.",
    "sourceUrl": "https://x.com/yourPlugAI/status/2090773583668621325",
    "sourceLabel": "X 原帖",
    "author": "yourPlugAI",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090773583668621325.mp4",
    "posterUrl": "/posters/x/2090773583668621325.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "动画",
      "动漫",
      "动态图形",
      "科幻太空",
      "城市街景"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "sci-fi",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/yourPlugAI/status/2090773583668621325"
  },
  {
    "id": "x-2090753345434554459",
    "title": "暖色室内墨镜时尚广告",
    "titleEn": "A Sunglasses Fashion Ad in a Warm Interior",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "人物在暖色客厅中取出并戴上墨镜，以近景突出产品与表情。",
    "summaryEn": "A woman takes out and wears sunglasses in a warm living room, with close-ups emphasizing the product and expression.",
    "prompt": "Use the attached girl as the exact character reference. Preserve her facial identity, facial features, eye color, long wavy brown hairstyle, makeup, skin tone, body proportions, jewelry, and overall appearance consistently across every shot.\n\nUse the attached sunglasses as the exact locked product reference. Preserve the black oval-shaped frames, dark lenses, gold detailing at the temples, hinge structure, proportions, materials, and finish exactly. The sunglasses must remain identical in every frame.\n\nCreate an ultra-realistic luxury UGC creator review filmed entirely in an elegant modern lounge area, not a bedroom. The lounge should feel sophisticated, warm, expensive, and naturally lived-in: cream boucle sofa, marble or stone coffee table, soft neutral cushions, elegant floor lamp, subtle gold accents, large windows, tasteful décor, and warm late-afternoon sunlight entering from the side.\n\nThe woman wears the same cream/white knit sweater, beige trousers, gold hoop earrings, delicate pendant necklace, and rings from the reference image.\n\nSTORYBOARD — 15 SECONDS\n\nSHOT 1 — 0–2 sec | Casual introduction\nMedium handheld smartphone shot. The woman sits comfortably on the cream sofa in the lounge with the sunglasses box and leather carrying case placed naturally on the coffee table beside her. She looks directly into the camera with a genuine smile and says:\n\n“Okay, these are way nicer than I expected.”\n\nNatural creator energy, relaxed posture, subtle handheld movement.\n\nSHOT 2 — 2–4 sec | Picking up the packaging\nShe leans toward the coffee table and picks up the luxury sunglasses box. Camera moves slightly closer as she turns the box toward the lens. She opens it naturally.\n\nCapture realistic packaging texture and believable hand movements.\n\nShe says:\n\n“Even the little details feel so well done.”\n\nSHOT 3 — 4–6 sec | Case reveal\nClose-up from slightly above her hands as she reveals the leather sunglasses case inside the box. She lifts the case out and opens it.\n\nWarm sunlight creates soft highlights on the leather and gold hardware.\n\nSHOT 4 — 6–8 sec | Sunglasses reveal\nShe carefully removes the sunglasses from the case. Cut to a beautiful macro product shot as she holds them between her fingers.\n\nSlowly rotate the sunglasses so the camera clearly sees the front frame, dark lenses, gold temple detail, side profile, and hinges.\n\nShe says:\n\n“And they’re ridiculously light.”\n\nSHOT 5 — 8–10 sec | Putting them on\nReturn to a medium shot. She smiles, lifts the sunglasses toward her face, and puts them on naturally.\n\nShe adjusts the temples with both hands, then gives a subtle confident smile toward the camera.\n\nSHOT 6 — 10–12 sec | Lounge sunlight test\nShe stands beside the large lounge window wearing the sunglasses. Camera follows her with smooth handheld movement.\n\nShe turns slightly toward the sunlight, then angles her face left and right so the lenses catch realistic reflections.\n\nShe says:\n\n“They actually feel this good on.”\n\nSHOT 7 — 12–14 sec | Lifestyle moment\nShe walks back toward the sofa, casually removes the sunglasses, and places them beside the leather case and box on the coffee table.\n\nCamera briefly focuses on the sunglasses before shifting focus back to her.\n\nSHOT 8 — 14–15 sec | Final product close-up\nShe picks the sunglasses back up and holds them beside her face while looking directly into the camera.\n\nWarm smile.\n\nShe says:\n\n“Yeah… these are staying in my rotation.”\n\nFinish with a subtle camera push-in toward the sunglasses in her hand, with the lounge softly blurred behind her.\n\nVISUAL STYLE\n\nPhotorealistic luxury lifestyle UGC, authentic smartphone creator footage, elegant lounge interior, warm golden-hour sunlight, natural skin texture, realistic facial expressions, subtle handheld camera movement, cinematic autofocus, shallow depth of field, neutral color palette, 4K HDR, 16:9, 24fps.",
    "sourceUrl": "https://x.com/codewithhajra/status/2090753345434554459",
    "sourceLabel": "X 原帖",
    "author": "codewithhajra",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090753345434554459.mp4",
    "posterUrl": "/posters/x/2090753345434554459.jpg",
    "duration": 15.04,
    "aspectRatio": "68:45",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "商业广告感",
      "照片级写实",
      "时尚",
      "产品与广告"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/codewithhajra/status/2090753345434554459"
  },
  {
    "id": "x-2090741007935480124",
    "title": "废弃超市中的少年拾荒者",
    "titleEn": "A Young Scavenger in an Abandoned Supermarket",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "少年穿过破损货架，在废弃超市里检查遗留包装。",
    "summaryEn": "A young scavenger walks past damaged shelves and examines discarded packaging in an abandoned supermarket.",
    "prompt": "Post-apocalyptic texture scene: a teenage scavenger picks through the husk of a supermarket shelves skeletal, a drift of leaves down aisle five, daylight through the collapsed roof bleached and shadowless. A kicked can skitters; a dead cart wheel turns. They find one intact cereal box, faded but unopened, and hold it like an artifact of a lost civilization, because it is. Muted palette, patient dread.",
    "sourceUrl": "https://x.com/LudovicCreator/status/2090741007935480124",
    "sourceLabel": "X 原帖",
    "author": "LudovicCreator",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090741007935480124.mp4",
    "posterUrl": "/posters/x/2090741007935480124.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "电影叙事",
      "写实电影感",
      "科幻太空"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/LudovicCreator/status/2090741007935480124"
  },
  {
    "id": "x-2090740724803182619",
    "title": "冰川裂解成 EXTINCTION 字样",
    "titleEn": "A Glacier Fractures into EXTINCTION",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "冰面裂纹扩张，碎冰上升并形成巨大冰晶文字。",
    "summaryEn": "Cracks spread across ice before fragments rise into giant crystalline lettering.",
    "prompt": "Ultra cinematic title sequence: A colossal glacier fractures across an endless frozen ocean. Deep blue cracks spread for miles beneath translucent ice before the entire shelf collapses. Mountains of ice are launched into the sky, suspended in breathtaking slow motion. Every crystal catches the sunlight while frozen debris rotates around the camera. The crystalline storm assembles the word \"EXTINCTION\", carved from pure glacial ice with intricate frozen textures. Hairline fractures slowly spread through every letter until the title violently shatters into millions of glittering ice fragments. Photorealistic ice simulation, IMAX-scale environmental VFX.",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2090740724803182619",
    "sourceLabel": "X 原帖",
    "author": "CharaspowerAI",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090740724803182619.mp4",
    "posterUrl": "/posters/x/2090740724803182619.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "广告与产品",
      "3D / CG",
      "动态图形",
      "自然风景"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2090740724803182619"
  },
  {
    "id": "x-2090730062203224148",
    "title": "红黑拼贴风动漫音乐片",
    "titleEn": "A Red-and-Black Collage Anime Music Clip",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "红衣人物、粗黑文字与撕纸式背景交替构成音乐画面。",
    "summaryEn": "A red-clad character, bold black type and torn-paper backgrounds alternate through a music sequence.",
    "prompt": "Generate a 15-second retro trend collage MV in the style of anime characters, street poster wall, copy zine, VHS, halftone, black and white red color scheme, English large font dynamic font, K-pop hip-hop rhythm\n-----",
    "sourceUrl": "https://x.com/studio_oneroom/status/2090730062203224148",
    "sourceLabel": "X 原帖",
    "author": "studio_oneroom",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090730062203224148.mp4",
    "posterUrl": "/posters/x/2090730062203224148.jpg",
    "duration": 15.168,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "人物表演"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/studio_oneroom/status/2090730062203224148"
  },
  {
    "id": "x-2090710705406189919",
    "title": "哥特教堂中的金红音乐片头",
    "titleEn": "A Gold-and-Crimson Gothic Music Opening",
    "model": "MiniMax H3",
    "mode": "Unknown",
    "summary": "飞散书页、彩窗与金色衬线字围绕哥特角色展开。",
    "summaryEn": "Flying pages, stained glass and gold serif lettering surround a gothic character.",
    "prompt": "Generate a 15-second dark fantasy gothic MV in the style of anime characters, stained glass ruins, vintage alchemy sketchbooks, film grain, faded parchment, gold and deep crimson color scheme, English ornate dynamic serif font, orchestral dark trap rhythm",
    "sourceUrl": "https://x.com/yourPlugAI/status/2090710705406189919",
    "sourceLabel": "X 原帖",
    "author": "yourPlugAI",
    "publishedAt": "2026-08-21",
    "addedAt": "2026-09-08T13:14:06.200Z",
    "approvedAt": "2026-09-08T13:14:06.200Z",
    "mediaUrl": "/media/x-2090710705406189919.mp4",
    "posterUrl": "/posters/x/2090710705406189919.jpg",
    "duration": 15.2,
    "aspectRatio": "16:9",
    "resolution": "720p",
    "tags": [
      "音乐视频",
      "动漫",
      "动态图形",
      "奇幻"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "promptCompleteness": "complete",
    "promptSourceUrl": "https://x.com/yourPlugAI/status/2090710705406189919"
  },
  {
    "id": "x-2097255890391077338",
    "title": "二十镜头的高速排版时装片",
    "titleEn": "A 20-Cut Typographic Fashion Rush",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "编辫女孩在夜间房间里起身走向镜头，超大白色排版、圆形遮罩与高速推拉镜头组成紧凑的十五秒时装片。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A braided character rises and walks toward the camera in a night-lit room as oversized white type, circular masks, and rapid camera moves form a dense 15-second fashion film. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "# 【ここだけ変更】\n参照画像を添付し、以下の「」内とBPMの数値だけを書き換える。\n本文も含めて全文をコピーして使用する。このテンプレートは15秒・20カット固定。\n\n## 表示文字\n文字列A = 「GPT-6」\n文字列B = 「Astra」\n文字列C = 「MiniMax H3」\n文字列D = 「AI STUDIO」\n文字列E = 「ONEROOM」\n\n## 音楽\nBPM = 160\n音楽ジャンル = 「ファッション・エレクトロ」\n音楽の質感 = 「強いビート、硬質な打音、細かいリズム、跳ねる低音」\n\n## 画面とグラフィック\n書体 = 「太いサンセリフ」\n暗色 = 「黒」\n明色 = 「白」\nアクセント色 = 「淡い金色」\n\n## 舞台\n部屋の雰囲気 = 「夜の落ち着いた部屋」\n照明 = 「青い窓明かりと暖かなランプ光」\n\n# 【設定ここまで／以下は変更不要】\n\n15秒、設定した画面比率のハイテンポなファッションMV。\n冒頭の設定を全編に適用する。設定名や説明文は画面に表示しない。\n\n# 主人公と参照画像\n添付した参照画像の人物を主人公にする。\n顔立ち、外見の特徴、髪型、髪色、体格、衣装、靴、既存のアクセサリーを反映し、全カットで同一人物として維持する。\nアニメやイラストなら線・塗り・質感を、写真なら写真的な見た目と質感を維持する。\n固定の性別、髪色、衣装、靴、画風を別途当てはめない。\n\nキャラクターシート内の複数の姿は同一人物の資料として扱う。\n登場人物は主人公1人。参照画像の背景、枠線、並列配置は使わない。\n画像で確認できない部分は、見えている外見や服装に合う簡潔な形で補完し、全カットで固定する。\n髪や衣服などの揺れは、参照画像に存在する可動部分だけで表現する。動きを付けるために髪やアクセサリーを追加しない。\n\n# 人物の動き\n主人公の動作は、ソファから立ち上がる、ソファの脇を抜けてレンズの横を通り過ぎる、足を止めて肩越しに一度振り返る、という一連の流れ。\n一つの動作を複数のカメラ位置からつなぎ、短いカットごとに別の振り付けを始めない。\n\n立ち上がるときは少し前へ重心を移し、肘掛けに添えた手を支えにして、上体を滑らかに起こす。\n移動するときは肩の力を抜き、腕を身体の近くで自然に振る。視線は進む先へ向け、頭の高さを安定させる。髪や衣服の可動部分が動きに少し遅れて揺れる。\n振り返るときは先に視線、次に顔、最後に肩がわずかに動く。片足に重心を残し、首と肩の角度で表情を見せる。\n\n手で文字を押す、引く、払う動作や、ビートごとに踏み替えるステップは入れない。\n人物の動作は自然な速さでつなぐ。鋭い速度感は、カメラの急加速、前景文字の通過、画角の切り替えで作る。\n文字は主人公の立ち上がり、進行方向、身体の向き、視線の変化に反応する。\n\n# 速度と演出\n最初のフレームから立ち上がりの途中。カットは主に0.375〜1.125秒。\n人物の動作の始点と着地、カメラの移動、文字の反応、カットの切り替えへ、音のアクセントを合わせる。動作の途中を一拍ずつ区切らない。\nBPMを変更しても動画は15秒とし、各タイムコードを維持する。音楽の細かなリズムや短い打音でカット位置へアクセントを置き、すべてのカットを拍頭へ強制的に合わせない。\nスローモーションや長い導入を入れない。\nムードは設定した照明、顔の陰影、参照画像に合う質感で表す。\n文字の叩き込み、短い複製、3分割、字間の伸縮をアクセントにする。3分割では中央だけに人物、左右には文字と図形を配置する。\n\n# 舞台\n設定した部屋の雰囲気と照明を反映した、同じ一つの室内。\n肘掛けのあるソファと、その背後の壁に窓を配置する。人物の移動とカメラの回り込みができる空間を確保する。\n光源と内装の位置は固定し、カメラは室内を大きく移動する。\n主人公の移動経路は、ソファの座面前からその脇を抜け、カメラの横まで。全編を同じ室内でつなぐ。\n\n# 文字とグラフィック\n表示するのは、設定した文字列A・B・C・D・Eの内容だけ。\n各文字列の綴り、大文字小文字、空白、記号を入力どおりに保つ。文字列Aなどの内部呼称、設定項目名、Shot番号、タイムコードは画面へ描かない。\n文字数や言語に合わせて大きさと字間を調整し、読む区間では各文字列を1行で画面内に収める。長い文字列は文字サイズを調整する。文字数を勝手に増減させない。\n\n設定した書体による、厚みのない鮮明な2D文字。入力した言語を表示できる同系統の書体で統一する。\nグラフィックには暗色・明色・アクセント色の設定を使う。この配色を人物の髪や服へ適用しない。\n背景に巨大な文字列Cの輪郭文字、中景に文字を伴わないバー・ドット・円弧、前景に主文字。\n背景は横へ、中景は縦へ、主文字は人物の動きに反応して動き、3層の速度差を作る。\n背景と前景の同じ綴りは、意図した別層の使用。\n人物の手前と背後に文字を置き、カメラ移動で強い視差を作る。腕や身体による遮蔽を自然に保つ。\n\nグラフィックの地色は冒頭で暗色、文字は明色。\n00:01.500で地色を明色、00:05.250で暗色、00:11.625で明色へ切り替える。文字色はその都度、地色と反対の設定色へ切り替える。\n反転はグラフィック層に適用し、室内の照明と人物の色は固定する。\n\n主文字を左右へ複製するのはShot 7だけ。左右に1つずつ配置した後、区間の終わりで画面外へ抜ける。\nShot 8では人物・全グラフィック・室内の動きを止め、カメラだけを動かす。\n終幕の00:14.625以降はカメラを含む全体を止める。それ以外ではカメラとグラフィックの運動をつなぐ。\n読む区間では全文を画面内に収め、顔と重ねない。設定した画面比率に合わせて構図、文字の大きさ、余白を調整する。3分割でも人物は中央列に1人だけ配置する。\n\n# シーン\n時刻は動画全体の絶対タイムコード。各区間を指定どおり配置し、後半を短縮しない。\n\n[Shot 1] 00:00.000–00:00.750\nソファから立ち上がり途中の主人公へ高速ドリーイン。\n上体が起き上がる勢いに合わせて文字列Cが前景へ叩き込まれる。\n最後の0.375秒は全文を読ませる。主人公の手は肘掛けを支えている。\n\n[Shot 2] 00:00.750–00:01.125\n顔へスナップズーム。立ち上がりながら、主人公の視線が進む方向へ移る。\n視線の変化に合わせ、前景文字が横へ抜ける。\n\n[Shot 3] 00:01.125–00:01.500\n足元の接写。立ち上がった体重が足裏へ乗る。靴の種類や素足の状態は参照画像の設定を保つ。\n自然な荷重の瞬間にアクセント色の短い線が床面を走り、短い打音が重なる。\n\n[Shot 4] 00:01.500–00:02.250\n全身ローアングル。\n立ち上がり終えた主人公が、一方の肩を引いてソファの脇へ向きを変える。\nその方向転換に引かれ、背後の文字列Aが大きく横切る。カメラも低い位置から横へ加速する。\n\n[Shot 5] 00:02.250–00:03.000\n斜め正面からのミディアム。主人公の指先が肘掛けから自然に離れる。\n身体が進み始めるタイミングで文字列Aが人物の横へ集まり、文字列全体の横幅が一瞬圧縮されて戻る。\nカメラは正面へ急接近する。\n\n[Shot 6] 00:03.000–00:03.750\n方向転換から最初の一歩へつながる動作。\n主人公は足の向きに合わせて上体を回し、髪や衣服の可動部分が少し遅れてついてくる。\nカメラは逆方向へ高速オービット。文字列Aは身体の回転に引かれて円弧を描く。\n\n[Shot 7] 00:03.750–00:04.500\n主人公が一歩でソファの脇へ出る。\nその移動に合わせて2本のバーが画面を3分割する。\n中央に移動中の主人公、左右に文字列Aを1つずつ配置し、文字は逆方向へ流れて画面外へ抜ける。\n人物の手は自然に下ろしたまま。\n\n[Shot 8] 00:04.500–00:04.875\n0.375秒だけのバレットタイム。\n歩き出した一歩の途中で、身体、髪、衣装、背景文字、中景の装飾、室内を完全停止する。\nカメラだけが約90度、素早く回り込む。\n空間内の各要素は動かず、カメラ移動だけで重なりと距離感が変わる。\n\n[Shot 9] 00:04.875–00:05.250\n即座に時間再開。主人公は止まっていた一歩をそのまま着地まで続ける。\n前へ進む勢いに合わせて文字列Aがレンズ直前を横切り、画面外へ抜ける。\nカメラもその通過方向へ鋭く振る。\n\n[Shot 10] 00:05.250–00:06.375\nワイド。続く移動を正面から捉え、カメラは高速で後退する。\n主人公は進む先を見て、腕を自然に振りながら近づく。\nその前進に合わせて文字列Bの字形を作る要素が左右から集まり、一行に整う。アクセント色の平面円が中央から広がる。\n文字が整う瞬間も、主人公の移動は途切れない。\n\n[Shot 11] 00:06.375–00:07.125\nカメラが主人公の腰の横を高速通過する。\n主人公は進む方向を保ち、レンズの横へ近づく。\n文字列Bが背後から手前へ追従し、身体に隠れてから横へ現れる。\n\n[Shot 12] 00:07.125–00:07.875\n移動中の肩から顔へ、カメラが短く上がる。\n主人公の目線は進む先に向いたまま。頭や肩の輪郭を光がかすめ、髪や衣服の動きが見える。\n人物の移動に合わせ、背後の文字列D・文字列Eが上下2行で上がる。\n最後の0.375秒は顔を避けた位置で、両方の全文を読ませる。\n\n[Shot 13] 00:07.875–00:08.625\n主人公がレンズのすぐ横を通り過ぎる。\n肩とその輪郭が前景を大きく横切り、その流れに文字列Bが追従して急接近する。\nカメラは同時に後退し、人物と文字の距離差を広げる。\n\n[Shot 14] 00:08.625–00:09.375\n短い文字主体のカット。文字列Cを線分から組み上げる。\n直前の人物の通過方向を引き継ぎ、中景のバーが高速で横断する。\n背景の輪郭文字は別速度で流れ続ける。\n\n[Shot 15] 00:09.375–00:10.125\n全身へハードカット。\n主人公が最後の一歩を置き、片足へ自然に体重を預けて移動を終える。\nその接地に合わせて平面の円弧が外へ広がる。\nカメラは床近くから一気に上がり、顔まで捉える。\n\n[Shot 16] 00:10.125–00:10.875\n主人公が肩越しに一度だけ振り返る。\n先に視線、次に顔が動き、肩がわずかに続く。足の位置は保つ。\nカメラは逆方向へ回り込み、文字列Bが肩の背後へ隠れて反対側へ抜ける。\n\n[Shot 17] 00:10.875–00:11.625\n俯瞰。直前の振り返りをそのまま続ける。\nカメラが斜め下へダイブして顔へ近づき、顔の向きの変化に合わせて文字列Cが画面下から上へ走る。\n主人公の腕は身体の脇に自然に下ろされている。\n\n[Shot 18] 00:11.625–00:12.750\n各0.375秒の3カット。すべて同じ振り返りの続き。\n1：肩と、その周囲の髪や衣服に残る動きを近距離で捉える。\n2：片足へ重心を預けた全身を斜め横から捉える。\n3：レンズへ向いた顔へ切り替える。\n角度とスケールを毎回変え、姿勢と視線をつなぐ。\n\n[Shot 19] 00:12.750–00:13.875\n目線の高さの斜め正面。主人公がレンズへ視線を合わせる。\n片足に重心を残し、肩の線をわずかに斜めにする。腕は自然に下ろす。\n視線が定まるタイミングで、上段に文字列D、下段に文字列Eが左右から集まる。\n各1行で正しい間隔へ整列し、顔を避けた位置へ収まる。\n\n[Shot 20] 00:13.875–00:15.000\n文字列D・文字列Eの2行を全文表示し続ける。\n主人公は前カットの姿勢と表情を保ち、カメラが短く寄る。背景の輪郭文字と中景の装飾は動き続ける。\n00:14.625でカメラ、人物、全グラフィックを止める。\n最後の0.375秒、主人公の顔と二つの文字列が明瞭な構図を保持して終わる。\n\n# 音\nBPMの設定値、4拍子、設定した音楽ジャンルと音楽の質感を使う。\n楽器と音色はジャンルに合わせて選び、冒頭からリズムを鳴らす。\n動作の始点と着地、カメラの加速、カット、文字の着弾へ音のアクセントを合わせる。\nShot 8の0.375秒だけ低域を薄くし、そのジャンルに合う高域のリズムと残響を残す。Shot 9で低音を即復帰する。\nBPMを変えた場合もカット時刻を優先し、細かなリズムや短い打音で指定位置にアクセントを置く。\n\nどちらの場合も歌・セリフ・ナレーションは入れない。\n終盤も映像を減速させず、最後の停止に短い打音を合わせて締める。",
    "promptSourceUrl": "https://x.com/studio_oneroom/status/2097255890391077338",
    "sourceUrl": "https://x.com/studio_oneroom/status/2097255890391077338",
    "sourceLabel": "X 原帖 · @studio_oneroom",
    "author": "AIスタジオワンルーム（AIアニメ、動画、漫画）",
    "publishedAt": "2026-09-08T09:29:04.000Z",
    "addedAt": "2026-09-09T00:24:42.106Z",
    "mediaUrl": "/media/x-2097255890391077338.mp4",
    "posterUrl": "/posters/x/2097255890391077338.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "コピペで出来る!\nMiniMax H3のプロンプト\n\nこの動画はPollo AIさんのご依頼で、MiniMax H3を使って動画を作りました。 \n@PolloAIJP\n\n今回は、今話題のChatGPT-6 Astra で MiniMax H3のプロンプトを書いてもらいました。\nプロンプトを書いてもらうだけなら、『プロンプト書いて』だけで書いてくれますけど、やっぱり自分が意図したプロンプトを書いてもらうためには、それなりにやり取りをしないとダメですね。\n\nただ、そうは言ってもきちんとスキルなどを設定すれば、このようにコピペで使える汎用的なプロンプトを全て書いてくれます。最近は、AIの性能が高すぎてうまく使いこなせる人と、使いこなせない人では物凄い格差が開きそうで怖いです。\n\nこれからの時代、プロンプトをうまく書ける人よりも、プロンプトをAIにうまく書かせられる人の方が重宝されそうですね。\n\n---------\nプロンプト\n---------\n\n# 【ここだけ変更】\n参照画像を添付し、以下の「」内とBPMの数値だけを書き換える。\n本文も含めて全文をコピーして使用する。このテンプレートは15秒・20カット固定。\n\n## 表示文字\n文字列A = 「GPT-6」\n文字列B = 「Astra」\n文字列C = 「MiniMax H3」\n文字列D = 「AI STUDIO」\n文字列E = 「ONEROOM」\n\n## 音楽\nBPM = 160\n音楽ジャンル = 「ファッション・エレクトロ」\n音楽の質感 = 「強いビート、硬質な打音、細かいリズム、跳ねる低音」\n\n## 画面とグラフィック\n書体 = 「太いサンセリフ」\n暗色 = 「黒」\n明色 = 「白」\nアクセント色 = 「淡い金色」\n\n## 舞台\n部屋の雰囲気 = 「夜の落ち着いた部屋」\n照明 = 「青い窓明かりと暖かなランプ光」\n\n# 【設定ここまで／以下は変更不要】\n\n15秒、設定した画面比率のハイテンポなファッションMV。\n冒頭の設定を全編に適用する。設定名や説明文は画面に表示しない。\n\n# 主人公と参照画像\n添付した参照画像の人物を主人公にする。\n顔立ち、外見の特徴、髪型、髪色、体格、衣装、靴、既存のアクセサリーを反映し、全カットで同一人物として維持する。\nアニメやイラストなら線・塗り・質感を、写真なら写真的な見た目と質感を維持する。\n固定の性別、髪色、衣装、靴、画風を別途当てはめない。\n\nキャラクターシート内の複数の姿は同一人物の資料として扱う。\n登場人物は主人公1人。参照画像の背景、枠線、並列配置は使わない。\n画像で確認できない部分は、見えている外見や服装に合う簡潔な形で補完し、全カットで固定する。\n髪や衣服などの揺れは、参照画像に存在する可動部分だけで表現する。動きを付けるために髪やアクセサリーを追加しない。\n\n# 人物の動き\n主人公の動作は、ソファから立ち上がる、ソファの脇を抜けてレンズの横を通り過ぎる、足を止めて肩越しに一度振り返る、という一連の流れ。\n一つの動作を複数のカメラ位置からつなぎ、短いカットごとに別の振り付けを始めない。\n\n立ち上がるときは少し前へ重心を移し、肘掛けに添えた手を支えにして、上体を滑らかに起こす。\n移動するときは肩の力を抜き、腕を身体の近くで自然に振る。視線は進む先へ向け、頭の高さを安定させる。髪や衣服の可動部分が動きに少し遅れて揺れる。\n振り返るときは先に視線、次に顔、最後に肩がわずかに動く。片足に重心を残し、首と肩の角度で表情を見せる。\n\n手で文字を押す、引く、払う動作や、ビートごとに踏み替えるステップは入れない。\n人物の動作は自然な速さでつなぐ。鋭い速度感は、カメラの急加速、前景文字の通過、画角の切り替えで作る。\n文字は主人公の立ち上がり、進行方向、身体の向き、視線の変化に反応する。\n\n# 速度と演出\n最初のフレームから立ち上がりの途中。カットは主に0.375〜1.125秒。\n人物の動作の始点と着地、カメラの移動、文字の反応、カットの切り替えへ、音のアクセントを合わせる。動作の途中を一拍ずつ区切らない。\nBPMを変更しても動画は15秒とし、各タイムコードを維持する。音楽の細かなリズムや短い打音でカット位置へアクセントを置き、すべてのカットを拍頭へ強制的に合わせない。\nスローモーションや長い導入を入れない。\nムードは設定した照明、顔の陰影、参照画像に合う質感で表す。\n文字の叩き込み、短い複製、3分割、字間の伸縮をアクセントにする。3分割では中央だけに人物、左右には文字と図形を配置する。\n\n# 舞台\n設定した部屋の雰囲気と照明を反映した、同じ一つの室内。\n肘掛けのあるソファと、その背後の壁に窓を配置する。人物の移動とカメラの回り込みができる空間を確保する。\n光源と内装の位置は固定し、カメラは室内を大きく移動する。\n主人公の移動経路は、ソファの座面前からその脇を抜け、カメラの横まで。全編を同じ室内でつなぐ。\n\n# 文字とグラフィック\n表示するのは、設定した文字列A・B・C・D・Eの内容だけ。\n各文字列の綴り、大文字小文字、空白、記号を入力どおりに保つ。文字列Aなどの内部呼称、設定項目名、Shot番号、タイムコードは画面へ描かない。\n文字数や言語に合わせて大きさと字間を調整し、読む区間では各文字列を1行で画面内に収める。長い文字列は文字サイズを調整する。文字数を勝手に増減させない。\n\n設定した書体による、厚みのない鮮明な2D文字。入力した言語を表示できる同系統の書体で統一する。\nグラフィックには暗色・明色・アクセント色の設定を使う。この配色を人物の髪や服へ適用しない。\n背景に巨大な文字列Cの輪郭文字、中景に文字を伴わないバー・ドット・円弧、前景に主文字。\n背景は横へ、中景は縦へ、主文字は人物の動きに反応して動き、3層の速度差を作る。\n背景と前景の同じ綴りは、意図した別層の使用。\n人物の手前と背後に文字を置き、カメラ移動で強い視差を作る。腕や身体による遮蔽を自然に保つ。\n\nグラフィックの地色は冒頭で暗色、文字は明色。\n00:01.500で地色を明色、00:05.250で暗色、00:11.625で明色へ切り替える。文字色はその都度、地色と反対の設定色へ切り替える。\n反転はグラフィック層に適用し、室内の照明と人物の色は固定する。\n\n主文字を左右へ複製するのはShot 7だけ。左右に1つずつ配置した後、区間の終わりで画面外へ抜ける。\nShot 8では人物・全グラフィック・室内の動きを止め、カメラだけを動かす。\n終幕の00:14.625以降はカメラを含む全体を止める。それ以外ではカメラとグラフィックの運動をつなぐ。\n読む区間では全文を画面内に収め、顔と重ねない。設定した画面比率に合わせて構図、文字の大きさ、余白を調整する。3分割でも人物は中央列に1人だけ配置する。\n\n# シーン\n時刻は動画全体の絶対タイムコード。各区間を指定どおり配置し、後半を短縮しない。\n\n[Shot 1] 00:00.000–00:00.750\nソファから立ち上がり途中の主人公へ高速ドリーイン。\n上体が起き上がる勢いに合わせて文字列Cが前景へ叩き込まれる。\n最後の0.375秒は全文を読ませる。主人公の手は肘掛けを支えている。\n\n[Shot 2] 00:00.750–00:01.125\n顔へスナップズーム。立ち上がりながら、主人公の視線が進む方向へ移る。\n視線の変化に合わせ、前景文字が横へ抜ける。\n\n[Shot 3] 00:01.125–00:01.500\n足元の接写。立ち上がった体重が足裏へ乗る。靴の種類や素足の状態は参照画像の設定を保つ。\n自然な荷重の瞬間にアクセント色の短い線が床面を走り、短い打音が重なる。\n\n[Shot 4] 00:01.500–00:02.250\n全身ローアングル。\n立ち上がり終えた主人公が、一方の肩を引いてソファの脇へ向きを変える。\nその方向転換に引かれ、背後の文字列Aが大きく横切る。カメラも低い位置から横へ加速する。\n\n[Shot 5] 00:02.250–00:03.000\n斜め正面からのミディアム。主人公の指先が肘掛けから自然に離れる。\n身体が進み始めるタイミングで文字列Aが人物の横へ集まり、文字列全体の横幅が一瞬圧縮されて戻る。\nカメラは正面へ急接近する。\n\n[Shot 6] 00:03.000–00:03.750\n方向転換から最初の一歩へつながる動作。\n主人公は足の向きに合わせて上体を回し、髪や衣服の可動部分が少し遅れてついてくる。\nカメラは逆方向へ高速オービット。文字列Aは身体の回転に引かれて円弧を描く。\n\n[Shot 7] 00:03.750–00:04.500\n主人公が一歩でソファの脇へ出る。\nその移動に合わせて2本のバーが画面を3分割する。\n中央に移動中の主人公、左右に文字列Aを1つずつ配置し、文字は逆方向へ流れて画面外へ抜ける。\n人物の手は自然に下ろしたまま。\n\n[Shot 8] 00:04.500–00:04.875\n0.375秒だけのバレットタイム。\n歩き出した一歩の途中で、身体、髪、衣装、背景文字、中景の装飾、室内を完全停止する。\nカメラだけが約90度、素早く回り込む。\n空間内の各要素は動かず、カメラ移動だけで重なりと距離感が変わる。\n\n[Shot 9] 00:04.875–00:05.250\n即座に時間再開。主人公は止まっていた一歩をそのまま着地まで続ける。\n前へ進む勢いに合わせて文字列Aがレンズ直前を横切り、画面外へ抜ける。\nカメラもその通過方向へ鋭く振る。\n\n[Shot 10] 00:05.250–00:06.375\nワイド。続く移動を正面から捉え、カメラは高速で後退する。\n主人公は進む先を見て、腕を自然に振りながら近づく。\nその前進に合わせて文字列Bの字形を作る要素が左右から集まり、一行に整う。アクセント色の平面円が中央から広がる。\n文字が整う瞬間も、主人公の移動は途切れない。\n\n[Shot 11] 00:06.375–00:07.125\nカメラが主人公の腰の横を高速通過する。\n主人公は進む方向を保ち、レンズの横へ近づく。\n文字列Bが背後から手前へ追従し、身体に隠れてから横へ現れる。\n\n[Shot 12] 00:07.125–00:07.875\n移動中の肩から顔へ、カメラが短く上がる。\n主人公の目線は進む先に向いたまま。頭や肩の輪郭を光がかすめ、髪や衣服の動きが見える。\n人物の移動に合わせ、背後の文字列D・文字列Eが上下2行で上がる。\n最後の0.375秒は顔を避けた位置で、両方の全文を読ませる。\n\n[Shot 13] 00:07.875–00:08.625\n主人公がレンズのすぐ横を通り過ぎる。\n肩とその輪郭が前景を大きく横切り、その流れに文字列Bが追従して急接近する。\nカメラは同時に後退し、人物と文字の距離差を広げる。\n\n[Shot 14] 00:08.625–00:09.375\n短い文字主体のカット。文字列Cを線分から組み上げる。\n直前の人物の通過方向を引き継ぎ、中景のバーが高速で横断する。\n背景の輪郭文字は別速度で流れ続ける。\n\n[Shot 15] 00:09.375–00:10.125\n全身へハードカット。\n主人公が最後の一歩を置き、片足へ自然に体重を預けて移動を終える。\nその接地に合わせて平面の円弧が外へ広がる。\nカメラは床近くから一気に上がり、顔まで捉える。\n\n[Shot 16] 00:10.125–00:10.875\n主人公が肩越しに一度だけ振り返る。\n先に視線、次に顔が動き、肩がわずかに続く。足の位置は保つ。\nカメラは逆方向へ回り込み、文字列Bが肩の背後へ隠れて反対側へ抜ける。\n\n[Shot 17] 00:10.875–00:11.625\n俯瞰。直前の振り返りをそのまま続ける。\nカメラが斜め下へダイブして顔へ近づき、顔の向きの変化に合わせて文字列Cが画面下から上へ走る。\n主人公の腕は身体の脇に自然に下ろされている。\n\n[Shot 18] 00:11.625–00:12.750\n各0.375秒の3カット。すべて同じ振り返りの続き。\n1：肩と、その周囲の髪や衣服に残る動きを近距離で捉える。\n2：片足へ重心を預けた全身を斜め横から捉える。\n3：レンズへ向いた顔へ切り替える。\n角度とスケールを毎回変え、姿勢と視線をつなぐ。\n\n[Shot 19] 00:12.750–00:13.875\n目線の高さの斜め正面。主人公がレンズへ視線を合わせる。\n片足に重心を残し、肩の線をわずかに斜めにする。腕は自然に下ろす。\n視線が定まるタイミングで、上段に文字列D、下段に文字列Eが左右から集まる。\n各1行で正しい間隔へ整列し、顔を避けた位置へ収まる。\n\n[Shot 20] 00:13.875–00:15.000\n文字列D・文字列Eの2行を全文表示し続ける。\n主人公は前カットの姿勢と表情を保ち、カメラが短く寄る。背景の輪郭文字と中景の装飾は動き続ける。\n00:14.625でカメラ、人物、全グラフィックを止める。\n最後の0.375秒、主人公の顔と二つの文字列が明瞭な構図を保持して終わる。\n\n# 音\nBPMの設定値、4拍子、設定した音楽ジャンルと音楽の質感を使う。\n楽器と音色はジャンルに合わせて選び、冒頭からリズムを鳴らす。\n動作の始点と着地、カメラの加速、カット、文字の着弾へ音のアクセントを合わせる。\nShot 8の0.375秒だけ低域を薄くし、そのジャンルに合う高域のリズムと残響を残す。Shot 9で低音を即復帰する。\nBPMを変えた場合もカット時刻を優先し、細かなリズムや短い打音で指定位置にアクセントを置く。\n\nどちらの場合も歌・セリフ・ナレーションは入れない。\n終盤も映像を減速させず、最後の停止に短い打音を合わせて締める。",
    "engagement": {
      "replies": 3,
      "reposts": 6,
      "likes": 110,
      "views": 7765,
      "snapshotAt": "2026-09-09T00:24:42.106Z"
    },
    "approvedAt": "2026-09-09T00:24:42.106Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096703938691215362",
    "title": "预演方块切换室内运镜",
    "titleEn": "Blockout Frames Guide an Indoor Camera Move",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "蓝色方块人预演与写实室内镜头交替出现，从点烟近景、沙发广角到走廊低机位，集中展示同一人物的连续运镜。 原生视频规格为 5 秒 · 2856×2048。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Blue blockout previews alternate with photoreal indoor shots, moving from a lighting close-up and sofa wide shot to a low hallway angle to demonstrate continuous camera control around one character. Native-video output: 5s · 2856×2048. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/toyxyz3/status/2096703938691215362",
    "sourceLabel": "X 原帖 · @toyxyz3",
    "author": "toyxyz",
    "publishedAt": "2026-09-06T20:55:48.000Z",
    "addedAt": "2026-09-09T00:24:42.106Z",
    "mediaUrl": "/media/x-2096703938691215362.mp4",
    "posterUrl": "/posters/x/2096703938691215362.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2856×2048",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 T2V camera test #comfyui",
    "engagement": {
      "replies": 8,
      "reposts": 8,
      "likes": 239,
      "views": 39267,
      "snapshotAt": "2026-09-09T00:24:42.106Z"
    },
    "approvedAt": "2026-09-09T00:24:42.106Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096602266828693963",
    "title": "荆棘蔷薇的暗红乐章",
    "titleEn": "A Crimson Thorn-and-Rose Ballad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红发角色、白色花束与缠绕荆棘在暗红竖屏中交替，组成带歌词排版的原创音乐段落。 原生视频规格为 44 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-haired character, white flowers, and curling thorns alternate in a dark vertical sequence with lyric typography. Native-video output: 44s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mttakenoko66/status/2096602266828693963",
    "sourceLabel": "X 原帖 · @mttakenoko66",
    "author": "Mt.takenoko",
    "publishedAt": "2026-09-06T14:11:48.000Z",
    "addedAt": "2026-09-09T08:44:19.838Z",
    "mediaUrl": "/media/x-2096602266828693963.mp4",
    "posterUrl": "/posters/x/2096602266828693963.jpg",
    "duration": 44,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🌹「棘」🥀\n\n「棘」のショートです。🫟\n\n参考にしたわけじゃないけど、\nちょっと火スぺっぽくなってしまったかもしれない。。。\n多分火の粉のせいだな。。。\nねっとりした感じを別途フル音源と合わせてお楽しみください。🎵\n\nPowered by Minimax H3\n\n#music #自作歌詞 #オリジナル曲 #sunoai #short",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 106,
      "snapshotAt": "2026-09-09T08:44:19.838Z"
    },
    "approvedAt": "2026-09-09T08:44:19.838Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096606005068972483",
    "title": "双手从零搭起一座体育场",
    "titleEn": "Hands Assemble a Stadium from Scratch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一双手在空白工作台上逐步铺草皮、搭看台并补齐座椅，把微缩体育场完整组装出来。 原生视频规格为 15 秒 · 2048×1152。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A pair of hands lays turf, builds stands, and adds seating until a miniature stadium is fully assembled. Native-video output: 15s · 2048×1152. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic creation film follows one maker reconstructing the main subject shown in the referenced image completely from scratch. The referenced image defines the finished subject’s complete visible appearance, proportions, structure, materials, colors, clothing or surface details, and distinctive features; ignore its background, framing, lighting, and unrelated elements.  Begin directly on an empty, clean virtual workbench. Use a stable front three-quarter overhead view that keeps the developing subject readable. There is one maker throughout, represented by the same consistent left and right hands and forearms. Show no more than two hands at once.  From 0 to 14 seconds, the entire creation unfolds as a clearly accelerated timelapse with rapid, purposeful hand movement and restrained motion blur. Short jump cuts compress repetitive manual work only after each action has visibly completed. Every cut inherits the exact form and progress left by the previous action.  0-2 seconds: One hand enters already holding the first foundation material, armature, or base element appropriate to the referenced subject and places it at the center. The second hand steadies it as the maker establishes the initial supporting form.  2-8 seconds: The maker rapidly develops the subject’s major structure and volumes using one coherent creation method appropriate to what the referenced image depicts. A living subject is sculpted as one continuous, non-gory digital form from armature to anatomy; a vehicle or machine is built from chassis to functional structure; an object is formed or assembled from its supporting body outward. Each additional material or component enters from outside the frame while firmly held by one of the maker’s hands, is carried to its destination, and remains under hand control until attached or shaped.  8-12 seconds: The same hands develop the recognizable outer form and reference-specific features. The maker sculpts, fits, wraps, stitches, fastens, carves, or polishes only where appropriate to the subject. Facial features, hair, clothing, body panels, wheels, glass, surfaces, accessories, or equivalent defining elements emerge through visible hand and tool contact, never through spontaneous transformation.  12-14 seconds: The maker refines proportions, edges, joints, surface transitions, textures, colors, and distinctive details until the developing subject closely matches the referenced image. One hand stabilizes the form while the other performs each final adjustment with a hand-held tool.  14-15 seconds: The maker removes the last tool by hand and withdraws both hands. The timelapse returns to normal speed as the camera makes a restrained push toward the completed subject and holds on a clean final view.  Materials and components do not need to be visible before use, but anything newly introduced must enter the frame already held by one of the maker’s hands. Nothing moves, assembles, appears, disappears, or changes material independently. Maintain one maker, one continuous subject, one creation position, and one category-appropriate construction method. No assistants, extra hands, detached anatomy, duplicated elements, magical morphing, drawing phase, software interface, cursor, menus, annotations, or text overlays.",
    "promptSourceUrl": "https://x.com/TechieBySA/status/2096606012002214373",
    "sourceUrl": "https://x.com/TechieBySA/status/2096606005068972483",
    "sourceLabel": "X 原帖 · @TechieBySA",
    "author": "TechieSA",
    "publishedAt": "2026-09-06T14:26:39.000Z",
    "addedAt": "2026-09-09T08:44:19.838Z",
    "mediaUrl": "/media/x-2096606005068972483.mp4",
    "posterUrl": "/posters/x/2096606005068972483.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2048×1152",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Building a stadium from scratch\n\nMade with MiniMax H3\n\nPrompt👇🏻",
    "engagement": {
      "replies": 9,
      "reposts": 3,
      "likes": 54,
      "views": 3875,
      "snapshotAt": "2026-09-09T08:44:19.838Z"
    },
    "approvedAt": "2026-09-09T08:44:19.838Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096606849999282209",
    "title": "告白化作巨型爱心",
    "titleEn": "A Confession Turns into a Giant Heart",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "放学后的告白把漂浮文字和粉色爱心越推越大，最终把两名学生卷进夸张的校园喜剧。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An after-school confession makes floating words and a pink heart grow until the two students are swept into a school comedy beat. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mugi_AI_Art/status/2096606849999282209",
    "sourceLabel": "X 原帖 · @mugi_AI_Art",
    "author": "むぎ@AIアートのセカイ",
    "publishedAt": "2026-09-06T14:30:00.000Z",
    "addedAt": "2026-09-09T08:44:19.839Z",
    "mediaUrl": "/media/x-2096606849999282209.mp4",
    "posterUrl": "/posters/x/2096606849999282209.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「好きです！」\nそのひと言が、文字どおり重すぎた——。\n放課後の告白が巨大なハートに変わる、15秒の学園ラブコメ動画をMiniMax H3で制作しました💗\n現在KusArtでは、MiniMax H3を期間限定で無料利用できるキャンペーンを開催中！\n9月13日 23:59（UTC）まで。\n#PR #KusArt #MiniMaxH3 #AI動画 #AIアニメ",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 29,
      "views": 1139,
      "snapshotAt": "2026-09-09T08:44:19.839Z"
    },
    "approvedAt": "2026-09-09T08:44:19.839Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096608164196254160",
    "title": "夕阳下轻咬青枣的女孩",
    "titleEn": "A Girl Bites a Green Jujube at Sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "水乡夕阳里，靠在树边的女孩轻咬青枣，眼神和笑容在五秒内自然展开。 原生视频规格为 5 秒 · 2504×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "At sunset beside a waterside village, a girl bites a green jujube and gradually breaks into a natural smile. Native-video output: 5s · 2504×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "@图片1\n\n生成一段5秒真人写实短视频，适用于 Minimax H3 全能模式。\n\n【最高优先级】\n严格保持@图片1中女生的人脸、五官、发型、妆容、肤色、服装、身材、靠树姿势、青枣、竹篮、江南水乡背景、夕阳光线、构图、焦段、景深和整体画面质感不变。\n\n不要重新设计人物，不要换脸，不要改变服装，不要改变场景。\n整个视频必须像真实摄影机随手捕捉到的一段生活瞬间：\n自然、松弛、清甜、灵动、毫无表演痕迹，绝对不要AI味。\n\n【人物状态】\n女生慵懒自然地斜靠在树干旁，身体保持放松。\n右手轻轻拿着一颗绿色青枣靠近嘴边。\n\n她不是刻意面对镜头表演，而是在吃青枣时自然注意到镜头。\n神态轻松，眼睛清澈明亮、有神、有真实焦点。\n嘴角原本只有很轻的笑意，随后因为吃到青枣而自然地笑开。\n\n重点：\n笑容一定是“慢慢浮现”，不能突然切换。\n眼睛先有笑意，嘴角随后跟着上扬。\n\n【5秒动作时间轴】\n\n0.00—0.70秒\n保持@图片1原始姿态。\n女生拿着青枣停在唇边，目光自然落在镜头附近。\n\n她的表情非常放松，\n嘴角只有一点轻轻的浅笑，\n眼神明亮、柔和、有神。\n\n身体几乎不动，\n只有非常轻微的自然呼吸，\n几缕头发被风轻轻吹动。\n\n0.70—1.60秒\n她自然地把青枣送到嘴边，\n轻轻张嘴，小口咬下一点。\n\n嘴唇先自然贴住青枣，\n牙齿轻轻咬合，\n随后手腕非常自然地把青枣稍微拿开一点。\n\n动作不要快，\n不要大口咬，\n不要夸张张嘴，\n不要故意做“吃东西”的表演动作。\n\n1.60—2.50秒\n她很轻地咀嚼一两下。\n\n咀嚼幅度极小，\n只是下颌和脸颊产生细微真实变化。\n\n不要鼓腮，\n不要左右嚼，\n不要连续机械咀嚼，\n不要像嚼口香糖。\n\n这一过程中，\n她的眼神开始出现更加明显的笑意，\n像觉得青枣很清甜，心情一下变得很好。\n\n2.50—3.70秒\n她忍不住自然地笑了起来。\n\n不是突然咧嘴大笑，\n而是：\n\n眼神先柔下来\n→ 嘴角轻轻上扬\n→ 苹果肌微微抬起\n→ 眼睛自然弯起\n→ 最后露出一点干净自然的牙齿。\n\n笑容甜美、清新、温柔，\n带一点点被镜头捕捉到之后的害羞感。\n\n她可以非常轻微地低一下眼，\n随后再抬眼看向镜头。\n\n动作幅度必须非常小，\n像真实女孩下意识的反应，\n不要做成刻意撒娇或摆拍。\n\n3.70—5.00秒\n她保持轻松甜美的笑容。\n\n手里仍然自然拿着那颗被咬过的青枣，\n停在嘴边附近。\n\n她看着镜头，\n眼睛清澈、明亮、灵动，\n笑意同时存在于眼睛和嘴角。\n\n最后约0.5秒保持一个最自然、最好看的浅笑状态结束。\n\n【表情重点】\n整段视频重点放在“眼神 + 微笑”的细微变化。\n\n眼睛必须：\n有真实焦点\n清澈明亮\n自然灵动\n带笑意\n不能瞪眼\n不能眼球乱转\n不能空洞无神\n\n笑容必须：\n先浅后深\n自然渐变\n眼睛和嘴角同步\n苹果肌轻微抬起\n不要突然露出整排牙齿\n不要大笑\n不要夸张卖萌\n\n允许一次非常自然的轻眨眼，\n不要频繁眨眼。\n\n【吃青枣真实感】\n必须表现真实连续动作：\n\n青枣靠近嘴唇\n→ 嘴唇接触青枣\n→ 小口咬下\n→ 青枣离开嘴边\n→ 轻微咀嚼\n→ 自然笑起来。\n\n嘴唇、牙齿、手指、青枣之间必须保持正确遮挡关系。\n\n青枣始终保持真实圆润形态，\n咬过之后只出现一个自然的小缺口。\n\n禁止：\n青枣突然消失\n青枣变形\n青枣穿过嘴唇\n手指插进青枣\n牙齿异常\n嘴型抽动\n下巴变形\n机械咀嚼。\n\n【镜头】\n镜头基本固定。\n\n严格保持@图片1原始构图和摄影距离。\n\n不要推镜\n不要拉镜\n不要摇镜\n不要环绕\n不要突然变焦\n不要AI自动追脸。\n\n只允许极其轻微的真实摄影机呼吸感。\n\n焦点始终稳定在女生眼睛和面部，\n背景保持柔和自然虚化。\n\n【环境】\n环境动态非常克制。\n\n树叶和柳枝只被微风轻轻带动，\n远处河面有细小自然波光，\n乌篷船、小桥、建筑保持稳定。\n\n阳光保持@图片1原有的温暖金色夕阳逆光，\n发丝边缘有柔和金色轮廓光，\n脸部曝光稳定自然。\n\n皮肤保持真实细腻：\n有轻微毛孔和自然肤质，\n白皙通透但不过度磨皮，\n不要蜡像皮肤，\n不要塑料质感。\n\n【严格去AI味】\n这是一个真实女孩真实吃青枣时被摄影机偶然拍到的瞬间。\n\n所有动作都必须存在：\n轻微惯性\n自然停顿\n细小呼吸\n真实肌肉变化\n自然重力感。\n\n不要每个动作都正好卡在时间节点上，\n动作之间必须有自然连续的过渡。\n\n严格避免：\nAI假笑\n僵硬微笑\n突然表情变化\n机械眨眼\n机械咀嚼\n夸张嘴型\n嘴唇乱动\n眼神发空\n瞪眼\n频繁眨眼\n摇头晃脑\n过度撒娇\n刻意卖萌\n夸张害羞\n身体乱动\n五官漂移\n脸型变化\n牙齿变形\n手指变形\n皮肤塑料感\n背景闪烁\n光线跳变\n任何明显AI生成痕迹。\n\n【最终效果】\n夕阳下的江南水乡，\n女生安静地靠在树边，\n轻轻咬了一口手里的青枣。\n\n她一边轻轻吃着，\n眼睛里的笑意慢慢浮现，\n随后非常自然地甜甜笑了起来。\n\n笑容清甜、柔和、真实，\n眼睛清澈有神，\n没有刻意表演，\n像摄影师偶然捕捉到的一段极其自然、漂亮、舒服的5秒生活镜头。",
    "promptSourceUrl": "https://x.com/Stellakjbk/status/2096609702641615286",
    "sourceUrl": "https://x.com/Stellakjbk/status/2096608164196254160",
    "sourceLabel": "X 原帖 · @Stellakjbk",
    "author": "九尾狐-FoxFairy🦊",
    "publishedAt": "2026-09-06T14:35:14.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2096608164196254160.mp4",
    "posterUrl": "/posters/x/2096608164196254160.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2504×1440",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "吃青枣的女孩，minimax H3生成的真人已经不脸糊了！！！🚀🚀\n\n提示词在评论区👇",
    "engagement": {
      "replies": 8,
      "reposts": 0,
      "likes": 37,
      "views": 4489,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096612096003965254",
    "title": "横向滚动的文字时装空间",
    "titleEn": "A Side-Scrolling Typographic Fashion World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "人物沿横向镜头行走，与立体文字、连线和图形发生互动，最终以遮挡转场落到品牌画面。 原生视频规格为 10 秒 · 768×1024。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A walking character interacts with dimensional type, lines, and graphics before a foreground wipe reveals the final brand frame. Native-video output: 10s · 768×1024. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Commercial film.\n\nMaintain the initial background color throughout the sequence.\n\nThe camera smoothly tracks the character with a continuous horizontal scrolling movement, moving alongside them as they walk.\n\nText and connecting lines progressively appear across the screen, forming a continuous visual pathway.\n\nAs the character walks, they recognize the text and graphic elements appearing around them as physical objects that actually exist within the same space. They naturally interact with these elements as they move forward: occasionally tapping nearby text with a fingertip, looking up at words above them, glancing sideways at graphics, slightly stepping aside to avoid an element, ducking underneath one, or briefly stopping to observe something.\n\nA continuous line, resembling a railway track or visual route, connects one text element to the next. The text from the reference image appears sequentially along this connected path, following the original order.\n\nThe text and graphic elements must not feel like flat background decorations. They should feel physically present in the same three-dimensional space as the character, with convincing spatial relationships and interaction.\n\nDo not repeat or duplicate any text.\n\nKeep the overall pacing fast, rhythmic, and fluid. The horizontal scrolling movement should remain continuous and uninterrupted throughout the sequence.\n\nFaithfully reproduce all text from the reference image without garbled characters, corrupted typography, misspellings, or altered wording.\n\nFor the final transition, a large foreground object or graphic element passes very close in front of the camera, completely covering the entire frame and creating a natural foreground-blur wipe transition.\n\nAs the foreground object clears the frame, seamlessly reveal the final logo screen.\n\nCamera movement should be smooth, dynamic, and cinematic.\n\nPrioritize the interaction between the character and the text / graphic elements above everything else.\n\nBGM: Up-tempo, rhythmic music with a strong beat.",
    "promptSourceUrl": "https://x.com/taya_mama_AI/status/2096612096003965254",
    "sourceUrl": "https://x.com/taya_mama_AI/status/2096612096003965254",
    "sourceLabel": "X 原帖 · @taya_mama_AI",
    "author": "た〜や｜スキマ時間×AI副業ママ",
    "publishedAt": "2026-09-06T14:50:51.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2096612096003965254.mp4",
    "posterUrl": "/posters/x/2096612096003965254.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "╲動画プロンプト公開／\n\n問い合わせが多数あったので公開します。\n\n再現性はあまりないのでガチャ要素が強いですが\nそれも含めて楽しんでいただければ…😊\n\n←MiniMax H3 Max turbo（falで生成）\n　Gemini Omni 1.1 flash（Flowで生成）→\n\nMiniMaxの方が好きだな♡\n\n▼プロンプト\nCommercial film.\n\nMaintain the initial background color throughout the sequence.\n\nThe camera smoothly tracks the character with a continuous horizontal scrolling movement, moving alongside them as they walk.\n\nText and connecting lines progressively appear across the screen, forming a continuous visual pathway.\n\nAs the character walks, they recognize the text and graphic elements appearing around them as physical objects that actually exist within the same space. They naturally interact with these elements as they move forward: occasionally tapping nearby text with a fingertip, looking up at words above them, glancing sideways at graphics, slightly stepping aside to avoid an element, ducking underneath one, or briefly stopping to observe something.\n\nA continuous line, resembling a railway track or visual route, connects one text element to the next. The text from the reference image appears sequentially along this connected path, following the original order.\n\nThe text and graphic elements must not feel like flat background decorations. They should feel physically present in the same three-dimensional space as the character, with convincing spatial relationships and interaction.\n\nDo not repeat or duplicate any text.\n\nKeep the overall pacing fast, rhythmic, and fluid. The horizontal scrolling movement should remain continuous and uninterrupted throughout the sequence.\n\nFaithfully reproduce all text from the reference image without garbled characters, corrupted typography, misspellings, or altered wording.\n\nFor the final transition, a large foreground object or graphic element passes very close in front of the camera, completely covering the entire frame and creating a natural foreground-blur wipe transition.\n\nAs the foreground object clears the frame, seamlessly reveal the final logo screen.\n\nCamera movement should be smooth, dynamic, and cinematic.\n\nPrioritize the interaction between the character and the text / graphic elements above everything else.\n\nBGM: Up-tempo, rhythmic music with a strong beat.",
    "engagement": {
      "replies": 9,
      "reposts": 47,
      "likes": 417,
      "views": 53296,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096644399333810220",
    "title": "游戏角色的轻快舞步",
    "titleEn": "A Game Character’s Breezy Dance Loop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "游戏角色在固定室内机位下完成连续舞步，动作替换后仍保持服装和人物造型。 原生视频规格为 5 秒 · 832×1472。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A game character performs a continuous indoor dance while retaining the costume and character design after motion transfer. Native-video output: 5s · 832×1472. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fnoji/status/2096644399333810220",
    "sourceLabel": "X 原帖 · @fnoji",
    "author": "えふのじ",
    "publishedAt": "2026-09-06T16:59:13.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2096644399333810220.mp4",
    "posterUrl": "/posters/x/2096644399333810220.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "832×1472",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3ベースのViggle Animate、プロンプト入れるテキストエンコーダーが無い（それはそう）からかサクッと置き換え出来るのは割と良いとは言え、細かい制御をし難い気がする\nそれはワークフロー弄れば頑張れるんかな\n\nでもちゃんと整えられるなら本家Viggleよりみっともない感じは薄い感",
    "engagement": {
      "replies": 1,
      "reposts": 5,
      "likes": 57,
      "views": 6030,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096662031550353862",
    "title": "霓虹旋涡中的动漫面孔",
    "titleEn": "An Anime Face in a Neon Vortex",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "高饱和动漫面孔在多彩旋涡、几何纹理和快速剪辑中反复重构，形成迷幻音乐段落。 原生视频规格为 26 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A saturated anime face repeatedly reforms through colorful vortices, geometric textures, and rapid edits in a psychedelic music sequence. Native-video output: 26s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tiyyyyyyyok/status/2096662031550353862",
    "sourceLabel": "X 原帖 · @tiyyyyyyyok",
    "author": "かるびぃ",
    "publishedAt": "2026-09-06T18:09:17.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2096662031550353862.mp4",
    "posterUrl": "/posters/x/2096662031550353862.jpg",
    "duration": 26,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "フルはYouTubeに\nminimax h3と編集で何とか音声もmixした",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 78,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096663157481623785",
    "title": "冰原骑士的极速奔袭对照",
    "titleEn": "A High-Speed Tundra Cavalry Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一精灵骑士骑白马穿过冰原的镜头上下并列，对照高速追拍、角色保持与风雪质感。 原生视频规格为 24 秒 · 1920×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two versions of the same elven rider galloping across frozen tundra are stacked to compare tracking speed, identity retention, and wintry detail. Native-video output: 24s · 1920×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "subject_definitions:\n- @Image1: The young elven maiden warrior (S1) with sharp pointed ears, long platinum-blonde hair tied with fine forehead braids, wearing the navy blue long robe with ornate gold-embroidered trim.\n- @Image2: The desolate Nordic wilderness environment (S2) beneath gloomy overcast storm clouds with pale, cold ambient twilight on the horizon.\n\nsummary:\nAn intense, high-speed backward tracking long shot capturing the elven maiden (S1) riding a white warhorse at full gallop across the frozen tundra (S2), framed with naturalistic filmic realism, extreme speed streaks, and raw diegetic soundscape.\n\nretention_analysis:\n- @Image1: Retain exact elven facial bone structure, platinum blonde hair, forehead braids, and navy blue gold-trimmed robe. Adapt pose to active riding stance.\n- @Image2: Retain gloomy overcast sky, cold desaturated color palette, and barren tundra terrain while introducing dynamic motion blur.\n\ndetailed_description:\nFrom 00:00.000 to 00:08.000, shot on an 85mm T1.8 cinema prime lens at eye-level framing a medium shot of the elven maiden (S1) astride a powerful white warhorse charging forward at extreme speed. The camera maintains a precise high-speed backward tracking movement 3.5 meters directly in front of the horse, locked in perfect speed synchronization while incorporating organic micro-breathing vibrations and subtle vertical road jostles authentic to naturalistic handheld cinematography. \n\nThe lighting strictly mirrors @Image2: heavy, gloomy overcast clouds cover the sky, casting a soft, cold ambient atmospheric glow that gently rims S1's blonde locks and pointed ears, with shadows subtly lifted with a desaturated cyan-teal tint and authentic 35mm film grain. \n\nDynamic focus is razor-sharp on S1's face and the horse's head: S1's cold, merciless amber eyes remain dead-locked onto the camera lens without blinking or smiling, her jaw set firm against the howling gale. Wild wind furiously whips her blonde bangs, fine braided hair, and the horse's white mane straight backward. \n\nThe surrounding frozen ground and distant dark ridgelines stretch into radical, flat horizontal motion blur streaks, conveying blistering velocity. White vapor clouds burst rhythmically from the warhorse's nostrils into the frigid air.\n\noverall_soundscape:\nZero non-diegetic background music (NO BGM). Pure, hyper-realistic diegetic soundscape: heavy, relentless galloping hoofbeats slamming into frozen gravel like war drums, brutal gale-force wind whistling and buffeting the camera microphone, accompanied by deep, rapid, guttural horse exhalations and the subtle jingling of leather bridle hardware.\n\nnon_diegetic_music:\nNone. Complete silence of musical score to maximize raw, cinematic documentary tension.",
    "promptSourceUrl": "https://x.com/Tomw852/status/2096663559178523072",
    "sourceUrl": "https://x.com/Tomw852/status/2096663157481623785",
    "sourceLabel": "X 原帖 · @Tomw852",
    "author": "Tom𝕎",
    "publishedAt": "2026-09-06T18:13:45.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2096663157481623785.mp4",
    "posterUrl": "/posters/x/2096663157481623785.jpg",
    "duration": 24,
    "aspectRatio": "portrait",
    "resolution": "1920×2160",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Dance Seedance 2.5 vs. MiniMax H3 #2 🎬\n(Full prompt in replies⤵️)\n\nSame reference images. Same prompt content (different structure). \n\n🔼: #Seedance 2.5 \n🔽: #MiniMaxH3 on a local RTX 5070 (12GB) using ComfyUI (2nd-pass re-conditioning + continuous split-sigmas)\n\n#AIVideo",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 6,
      "views": 2037,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097256724495581565",
    "title": "四面角色参考保持造型一致",
    "titleEn": "Four Reference Views Preserve One Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "紫发角色在白底前旋转并连续变换姿态，多视图参考帮助发饰、服装与面部保持一致。 原生视频规格为 4 秒 · 832×1504。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple-haired character turns and changes pose against white while multi-view references preserve face, outfit, and accessories. Native-video output: 4s · 832×1504. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fnoji/status/2097256724495581565",
    "sourceLabel": "X 原帖 · @fnoji",
    "author": "えふのじ",
    "publishedAt": "2026-09-08T09:32:22.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2097256724495581565.mp4",
    "posterUrl": "/posters/x/2097256724495581565.jpg",
    "duration": 4,
    "aspectRatio": "portrait",
    "resolution": "832×1504",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のキャラ追加、ようやっと安定してきた気がする\nBlender等のDCCソフトは使わず、Face Refine未実施の生データ",
    "engagement": {
      "replies": 5,
      "reposts": 53,
      "likes": 926,
      "views": 59530,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097452080567111973",
    "title": "十六格参考驱动的迷你舞步",
    "titleEn": "A 16-Panel Reference Drives a Chibi Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "单张十六格角色图驱动迷你人物连续起舞，姿态变化中保持棕色制服与卡通比例。 原生视频规格为 8 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A single 16-panel character sheet drives a chibi dance while preserving the brown uniform and cartoon proportions. Native-video output: 8s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sankakuten91256/status/2097452080567111973",
    "sourceLabel": "X 原帖 · @sankakuten91256",
    "author": "さんかくてん",
    "publishedAt": "2026-09-08T22:28:39.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2097452080567111973.mp4",
    "posterUrl": "/posters/x/2097452080567111973.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GPT Image 2.5で16分割一枚の画像でMiniMax H3\n一枚の参照なので完全再現は難しいけどここまでできたらOK",
    "engagement": {
      "replies": 22,
      "reposts": 173,
      "likes": 1775,
      "views": 119918,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed.",
    "featured": {
      "at": "2026-09-13T00:21:49.934Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2097498776034513398",
    "title": "雨夜山路上的双车追逐",
    "titleEn": "Two Cars Race Through a Rainy Mountain Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两辆轿车在湿滑山路上并肩漂移，尾灯、雨雾与贴地追拍强化速度感。 原生视频规格为 15 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two cars drift side by side on a wet mountain road as taillights, spray, and low tracking shots amplify the speed. Native-video output: 15s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2097498776034513398",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-09-09T01:34:12.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2097498776034513398.mp4",
    "posterUrl": "/posters/x/2097498776034513398.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Good night ✨\n\nMiniMax H3 Max Turbo on fal:",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 862,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097550605099348384",
    "title": "黑白几何空间的时装姿态",
    "titleEn": "Fashion Poses in a Black-and-White Grid",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣模特在高反差几何空间中切换姿态，眼睛特写、分屏与网格构成极简时装片。 原生视频规格为 10 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-clad model changes poses inside a high-contrast geometric space shaped by eye close-ups, split screens, and grids. Native-video output: 10s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2097550605099348384",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-09-09T05:00:09.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2097550605099348384.mp4",
    "posterUrl": "/posters/x/2097550605099348384.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "POSE 🖤🔲\n\nMiniMax H3 Max Turbo on @fal",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 6,
      "views": 629,
      "snapshotAt": "2026-09-09T08:44:19.840Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097558075230265654",
    "title": "失恋女孩的白底舞步",
    "titleEn": "A Heartbroken Girl Dances on White",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝裙动漫女孩在纯白背景前完成连续舞步，发丝和裙摆随转身自然摆动。 原生视频规格为 10 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime girl in a blue dress performs a continuous dance against white, with her hair and skirt following each turn. Native-video output: 10s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/munou_ac/status/2097558075230265654",
    "sourceLabel": "X 原帖 · @munou_ac",
    "author": "てんねん",
    "publishedAt": "2026-09-09T05:29:50.000Z",
    "addedAt": "2026-09-09T08:44:19.840Z",
    "mediaUrl": "/media/x-2097558075230265654.mp4",
    "posterUrl": "/posters/x/2097558075230265654.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "失恋ガールズ\n\nコマリちゃんは踊りたい\n\nGPT Image 2.5 × MiniMax H3",
    "engagement": {
      "replies": 2,
      "reposts": 17,
      "likes": 189,
      "views": 4666,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.840Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097560855630213209",
    "title": "草莓世界里的甜酸恋曲",
    "titleEn": "A Sweet-and-Sour Love Song in a Strawberry World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉色迷你角色在草莓甜点、闪光文字和糖果布景间演唱，形成甜酸恋爱主题的音乐蒙太奇。 原生视频规格为 32 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink chibi performer sings among strawberry desserts, sparkling type, and candy sets in a sweet-and-sour romantic montage. Native-video output: 32s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LastM07/status/2097560855630213209",
    "sourceLabel": "X 原帖 · @LastM07",
    "author": "miwa | AIイラスト・音楽",
    "publishedAt": "2026-09-09T05:40:53.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097560855630213209.mp4",
    "posterUrl": "/posters/x/2097560855630213209.jpg",
    "duration": 32,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "『Strawberry Mood』\n\n甘いだけじゃない、ちょっぴり酸っぱい恋心。\n素直になれない小さなJealousyと、「もう少しそばにいて」という想いを、オルゴール × R&B × Cute EDMに詰め込んだ、ふんわり甘酸っぱいラブソング。\n\nsuno→https://suno.com/s/makmQqugbFpNueBP\n\nGPT-Image2.5➕Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 11,
      "views": 353,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097561594540773563",
    "title": "活力睡衣的十镜头秋日广告",
    "titleEn": "A 10-Cut Autumn Ad for Recovery Pajamas",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "女演员在橙黄动态图形和巨大排版中换上秋季睡衣，十个镜头紧凑呈现服装细节与品牌口号。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An actress changes into autumn pajamas amid orange motion graphics and oversized type as ten rapid shots present the garment and brand line. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "subject_definitions:\n<Picture 1>: Attached RAKUNE autumn storyboard: ten panels in two columns and five rows, read left to right, top to bottom. Use each panel as its corresponding shot anchor.\n<Subject 1>: The same fictional young adult Japanese actress in <Picture 1>, with shoulder-length dark hair, natural bangs, minimal makeup, and an approachable smile.\n<Subject 2>: The reference's rust-colored long-sleeved button-front pajamas with ivory piping and matching full-length trousers.\n<Subject 3>: The reference's orange, golden-yellow, navy, and ivory graphics: huge typography, circles, diagonal bands, stars, radial lines, and geometric autumn leaves.\n\nsummary:\n[reference generation] Create a 15-second, horizontal 16:9 commercial with exactly ten shots following <Picture 1>. Combine photorealistic <Subject 1>, clear views of <Subject 2>, and boldly animated <Subject 3>. Emphasize vitality through physical movement, explosive typography, and tightly synchronized music. Finish with a stable, readable brand card.\n\nretention_analysis:\n<Picture 1>: partially_preserved — Keep panel order and key compositions; exclude grid, margins, heading, numbers, and timecodes.\n<Subject 1>: fully_preserved — Maintain identity, hair, natural appearance, and anatomy.\n<Subject 2>: fully_preserved — Maintain color, long sleeves, trousers, buttons, piping, and fit.\n<Subject 3>: attribute_transfer — Animate the reference's design vocabulary as separate layers.\n\ndetailed_description:\nUse bright, natural studio lighting throughout, keeping faces and fabric clear. Shots 1–3 show <Subject 1> in the reference's ivory casual top; shots 4–10 show her in <Subject 2>. Every shot fills the 16:9 screen. Use rapid acceleration, controlled overshoot, elastic settling, and clear visual hierarchy. Never animate the storyboard as a flat slideshow.\n\n[Shot 1] Follow panel 01 of <Picture 1>. Medium shot of <Subject 1> turning toward camera against golden yellow, hair following naturally. Orange circles expand and angular leaves sweep outward. Huge navy text \"秋だ！\" snaps in with one rebound. The camera pushes in slightly at fast speed and settles. A short graphic rush coincides with the first musical hit.\n\n[Shot 2] At 00:01.000, the camera cuts to panel 02's three-strip layout. Show <Subject 1> finishing work, lightly stretching, and smiling in separate strips. Slide them in opposing directions while a navy banner carries white text \"今日も、全力。\". Cameras within the strips remain fixed; gestures and panels provide movement. Add short sliding sounds.\n\n[Shot 3] At 00:02.200, the camera cuts to panel 03's close-up. <Subject 1> smiles and snaps her fingers near the lens with natural hand anatomy. The camera makes a small, fast push toward her hand. Orange rings expand from the snap; text \"じゃ、着替えよ！\" appears beside her face. The rings cover the frame. Synchronize a crisp finger snap and expanding rush.\n\n[Shot 4] At 00:03.500, the camera cuts through the wipe to panel 04's full-body reveal. <Subject 1>, now in <Subject 2>, completes a lively quarter-turn against golden yellow. Fabric moves naturally. Giant navy \"RAKUNE\" lands behind her with one overshoot and an orange starburst. The camera pulls back moderately at fast speed to reveal the whole outfit. Add a fabric swish and short impact.\n\n[Shot 5] At 00:05.000, the camera cuts to panel 05's product triptych: collar and buttons, long-sleeve cuff, and full-length trouser hem of <Subject 2>. Three angular windows slide into position on consecutive accents. Small, slow camera moves reveal fabric and ivory piping. Display \"秋のリカバリーウェア\" on a readable ivory band. Add three crisp ticks.\n\n[Shot 6] At 00:06.500, the camera cuts to panel 06's wide composition. Center <Subject 1> in <Subject 2> against ivory. She opens both arms upward, triggering navy and orange lines that extend beyond the frame. Large navy text \"夜を、\" expands without covering her face. The camera pulls back slightly at moderate speed. Add soft clothing movement and a rising rush.\n\n[Shot 7] At 00:08.000, the camera cuts to panel 07's eye-level, three-quarter full-body view. <Subject 1> steps forward in <Subject 2>. Her footfall triggers a navy diagonal sweep carrying huge yellow text \"着替えろ。\", which stops sharply. The camera tracks sideways slightly at moderate speed. Keep trousers and body proportions clear. Add a light footfall and sweeping sound.\n\n[Shot 8] At 00:09.500, the camera cuts to panel 08's navy background. <Subject 1> side-steps in <Subject 2> and looks back with a spontaneous smile. Orange and yellow balls bounce at different depths; curved paths draw behind her. Ivory text \"明日を、\" bounces into place. The camera tracks laterally with moderate amplitude at fast speed, then settles. Add light footfalls and graphic pops.\n\n[Shot 9] At 00:11.000, the camera cuts to panel 09's golden-yellow full-body view. <Subject 1> makes one small jump in <Subject 2>, both feet briefly above the floor. At the apex, stars, circles, and navy radial lines burst outward: the commercial's largest graphic expansion. Huge navy text \"弾ませろ。\" lands in a readable arrangement. The camera follows upward slightly, then follows her descent. Add a concentrated graphic rush; avoid prolonged slow motion.\n\n[Shot 10] At 00:13.000, the camera cuts to panel 10's end card. Within 0.3 seconds, shapes align into an ivory-and-orange layout. Place smiling <Subject 1> in <Subject 2> on the right, waving gently. Left: large navy \"RAKUNE\", subtitle \"疲労回復パジャマ\", and a clear product arrangement of shirt and trousers. Bottom: \"夜を着替えろ。明日を弾ませろ。\". Fix the camera and hold all text through 15 seconds, with only subtle background movement. Add a light landing sound. At approximately 00:13.700, a bright female brand voice (S1) says in an off-screen voiceover <d>[Japanese] ラクネ！</d>. The on-screen character's lips remain closed.\n\nKeep Japanese text exact. Do not add people, extra copy, medical diagrams, or unsupported efficacy claims. Maintain consistent faces, hands, and garments. Create intensity through graphic motion and composition, not flashing strobes or persistent camera shake.\n\noverall_soundscape:\nUse subtle fabric movement, a crisp finger snap, and light footfalls synchronized to actions. Add short graphic rushes, ticks, pops, and impacts at the specified transitions, below the brand voice. Let movement sounds decay before the ending.\n\nnon_diegetic_music:\nOriginal instrumental electro-pop at approximately 160 BPM, with punchy kick, syncopated bass, sharp claps, bright synth stabs, and short rising fills. Start immediately, accent the specified edit points, and build to the jump's strongest hit. Resolve into a sonic-logo cadence, lower the track beneath the brand voice, and end cleanly at 15 seconds without sung lyrics.",
    "promptSourceUrl": "https://x.com/tanabe_fragm/status/2097561594540773563",
    "sourceUrl": "https://x.com/tanabe_fragm/status/2097561594540773563",
    "sourceLabel": "X 原帖 · @tanabe_fragm",
    "author": "タナベ | AI動画 × マーケティング",
    "publishedAt": "2026-09-09T05:43:49.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097561594540773563.mp4",
    "posterUrl": "/posters/x/2097561594540773563.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "fashion"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GPT Images 2.5は2.0と同じく絵コンテも作れますね。\n\nGPT Images 2.5で絵コンテを作り、それを参照して架空のリカバリーパジャマ「RAKUNE（ラクネ）」の秋向けCM動画を作りました。\n\n使用した動画生成AIはMiniMax H3。\nプロンプトはAstraに作ってもらっています。\n\n絵コンテ生成のプロンプト\n------------\n私は新しい疲労回復パジャマ（秋向け）の広告動画を作りたい。\n疲労回復パジャマの名称は「RAKUNE」\nその広告動画のストーリーボードを作って。\n動画の尺は15秒。\n映像スタイルはモーショングラフィックス。\nカットの数は10。\n動きがあり、派手なモーショングラフィックスにして欲しい。\n癒しよりも「元気」や「活力」を打ち出したい。\nCMに登場するのは架空の若手の日本人女優。\nナチュラルな魅力で、好感度は抜群、という設定。\n全10カットの絵コンテを作って。アスペクト比は横長（16:9）\n------------\n\n動画生成生成のプロンプト\n------------\nsubject_definitions:\n<Picture 1>: Attached RAKUNE autumn storyboard: ten panels in two columns and five rows, read left to right, top to bottom. Use each panel as its corresponding shot anchor.\n<Subject 1>: The same fictional young adult Japanese actress in <Picture 1>, with shoulder-length dark hair, natural bangs, minimal makeup, and an approachable smile.\n<Subject 2>: The reference's rust-colored long-sleeved button-front pajamas with ivory piping and matching full-length trousers.\n<Subject 3>: The reference's orange, golden-yellow, navy, and ivory graphics: huge typography, circles, diagonal bands, stars, radial lines, and geometric autumn leaves.\n\nsummary:\n[reference generation] Create a 15-second, horizontal 16:9 commercial with exactly ten shots following <Picture 1>. Combine photorealistic <Subject 1>, clear views of <Subject 2>, and boldly animated <Subject 3>. Emphasize vitality through physical movement, explosive typography, and tightly synchronized music. Finish with a stable, readable brand card.\n\nretention_analysis:\n<Picture 1>: partially_preserved — Keep panel order and key compositions; exclude grid, margins, heading, numbers, and timecodes.\n<Subject 1>: fully_preserved — Maintain identity, hair, natural appearance, and anatomy.\n<Subject 2>: fully_preserved — Maintain color, long sleeves, trousers, buttons, piping, and fit.\n<Subject 3>: attribute_transfer — Animate the reference's design vocabulary as separate layers.\n\ndetailed_description:\nUse bright, natural studio lighting throughout, keeping faces and fabric clear. Shots 1–3 show <Subject 1> in the reference's ivory casual top; shots 4–10 show her in <Subject 2>. Every shot fills the 16:9 screen. Use rapid acceleration, controlled overshoot, elastic settling, and clear visual hierarchy. Never animate the storyboard as a flat slideshow.\n\n[Shot 1] Follow panel 01 of <Picture 1>. Medium shot of <Subject 1> turning toward camera against golden yellow, hair following naturally. Orange circles expand and angular leaves sweep outward. Huge navy text \"秋だ！\" snaps in with one rebound. The camera pushes in slightly at fast speed and settles. A short graphic rush coincides with the first musical hit.\n\n[Shot 2] At 00:01.000, the camera cuts to panel 02's three-strip layout. Show <Subject 1> finishing work, lightly stretching, and smiling in separate strips. Slide them in opposing directions while a navy banner carries white text \"今日も、全力。\". Cameras within the strips remain fixed; gestures and panels provide movement. Add short sliding sounds.\n\n[Shot 3] At 00:02.200, the camera cuts to panel 03's close-up. <Subject 1> smiles and snaps her fingers near the lens with natural hand anatomy. The camera makes a small, fast push toward her hand. Orange rings expand from the snap; text \"じゃ、着替えよ！\" appears beside her face. The rings cover the frame. Synchronize a crisp finger snap and expanding rush.\n\n[Shot 4] At 00:03.500, the camera cuts through the wipe to panel 04's full-body reveal. <Subject 1>, now in <Subject 2>, completes a lively quarter-turn against golden yellow. Fabric moves naturally. Giant navy \"RAKUNE\" lands behind her with one overshoot and an orange starburst. The camera pulls back moderately at fast speed to reveal the whole outfit. Add a fabric swish and short impact.\n\n[Shot 5] At 00:05.000, the camera cuts to panel 05's product triptych: collar and buttons, long-sleeve cuff, and full-length trouser hem of <Subject 2>. Three angular windows slide into position on consecutive accents. Small, slow camera moves reveal fabric and ivory piping. Display \"秋のリカバリーウェア\" on a readable ivory band. Add three crisp ticks.\n\n[Shot 6] At 00:06.500, the camera cuts to panel 06's wide composition. Center <Subject 1> in <Subject 2> against ivory. She opens both arms upward, triggering navy and orange lines that extend beyond the frame. Large navy text \"夜を、\" expands without covering her face. The camera pulls back slightly at moderate speed. Add soft clothing movement and a rising rush.\n\n[Shot 7] At 00:08.000, the camera cuts to panel 07's eye-level, three-quarter full-body view. <Subject 1> steps forward in <Subject 2>. Her footfall triggers a navy diagonal sweep carrying huge yellow text \"着替えろ。\", which stops sharply. The camera tracks sideways slightly at moderate speed. Keep trousers and body proportions clear. Add a light footfall and sweeping sound.\n\n[Shot 8] At 00:09.500, the camera cuts to panel 08's navy background. <Subject 1> side-steps in <Subject 2> and looks back with a spontaneous smile. Orange and yellow balls bounce at different depths; curved paths draw behind her. Ivory text \"明日を、\" bounces into place. The camera tracks laterally with moderate amplitude at fast speed, then settles. Add light footfalls and graphic pops.\n\n[Shot 9] At 00:11.000, the camera cuts to panel 09's golden-yellow full-body view. <Subject 1> makes one small jump in <Subject 2>, both feet briefly above the floor. At the apex, stars, circles, and navy radial lines burst outward: the commercial's largest graphic expansion. Huge navy text \"弾ませろ。\" lands in a readable arrangement. The camera follows upward slightly, then follows her descent. Add a concentrated graphic rush; avoid prolonged slow motion.\n\n[Shot 10] At 00:13.000, the camera cuts to panel 10's end card. Within 0.3 seconds, shapes align into an ivory-and-orange layout. Place smiling <Subject 1> in <Subject 2> on the right, waving gently. Left: large navy \"RAKUNE\", subtitle \"疲労回復パジャマ\", and a clear product arrangement of shirt and trousers. Bottom: \"夜を着替えろ。明日を弾ませろ。\". Fix the camera and hold all text through 15 seconds, with only subtle background movement. Add a light landing sound. At approximately 00:13.700, a bright female brand voice (S1) says in an off-screen voiceover <d>[Japanese] ラクネ！</d>. The on-screen character's lips remain closed.\n\nKeep Japanese text exact. Do not add people, extra copy, medical diagrams, or unsupported efficacy claims. Maintain consistent faces, hands, and garments. Create intensity through graphic motion and composition, not flashing strobes or persistent camera shake.\n\noverall_soundscape:\nUse subtle fabric movement, a crisp finger snap, and light footfalls synchronized to actions. Add short graphic rushes, ticks, pops, and impacts at the specified transitions, below the brand voice. Let movement sounds decay before the ending.\n\nnon_diegetic_music:\nOriginal instrumental electro-pop at approximately 160 BPM, with punchy kick, syncopated bass, sharp claps, bright synth stabs, and short rising fills. Start immediately, accent the specified edit points, and build to the jump's strongest hit. Resolve into a sonic-logo cadence, lower the track beneath the brand voice, and end cleanly at 15 seconds without sung lyrics.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 8,
      "views": 702,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097562141238915369",
    "title": "排错成功后的猫咪庆功舞",
    "titleEn": "A Cat Celebrates a Solved Bug with a Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白猫在霓虹隧道里直立起舞，头顶字幕把动作包装成解决故障后的庆祝瞬间。 原生视频规格为 11 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white cat dances upright inside a neon tunnel while captions frame the movement as a celebration after solving a bug. Native-video output: 11s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/daoinc/status/2097562141238915369",
    "sourceLabel": "X 原帖 · @daoinc",
    "author": "さとたく。(札幌のWeb屋さん)",
    "publishedAt": "2026-09-09T05:45:59.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097562141238915369.mp4",
    "posterUrl": "/posters/x/2097562141238915369.jpg",
    "duration": 11,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "dance",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "原因を一発で特定した時のワイ🐱\nこういう時だけ、ちょっと仕事できる感を出したくなる。\nMiniMax H3で猫を踊らせてみました。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 99,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097563963856720307",
    "title": "霓虹舞台上的动漫音乐蒙太奇",
    "titleEn": "An Anime Music Montage on a Neon Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "青绿色双马尾角色穿过爱心舞台、舞蹈镜头和快速排版，组成节奏密集的动漫音乐蒙太奇。 原生视频规格为 10 秒 · 1696×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A teal twin-tailed character moves through heart-lit stages, dance shots, and rapid typography in a tightly paced anime music montage. Native-video output: 10s · 1696×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/dream_weebs/status/2097563963856720307",
    "sourceLabel": "X 原帖 · @dream_weebs",
    "author": "DreamWeebs",
    "publishedAt": "2026-09-09T05:53:14.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097563963856720307.mp4",
    "posterUrl": "/posters/x/2097563963856720307.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1696×928",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Writing prompts for MiniMax H3 ref2va Anime MVs is exhausting — especially timing cuts & kinetic typography.\nSo I created an Agent Skill that writes production-ready H3 prompts for you. \nThis 10s anime sequence was generated using prompts structured entirely by the Skill.\nOpen-source repo & how it works below \nhttps://github.com/SeOgi-Tsu/anime-mv-ref2v",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 104,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097564918153359501",
    "title": "舞台吉他手的换脸实验",
    "titleEn": "A Face-Swap Experiment with a Stage Guitarist",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏舞台上的吉他手持续演奏，镜头用稳定人物运动检查本地换脸后的面部保持。 原生视频规格为 7 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A guitarist keeps performing in a vertical stage shot used to inspect facial consistency after a local face-swap workflow. Native-video output: 7s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/so_tab/status/2097564918153359501",
    "sourceLabel": "X 原帖 · @so_tab",
    "author": "sotab",
    "publishedAt": "2026-09-09T05:57:02.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097564918153359501.mp4",
    "posterUrl": "/posters/x/2097564918153359501.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal",
      "vertical"
    ],
    "scenes": [
      "performance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MINIMAX H3（ローカル）でフェイススワップの実験してたらまあまあいい感じにできました。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 49,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097565673501196598",
    "title": "超市推车里的玩偶零食派对",
    "titleEn": "A Plush Snack Party in a Supermarket Cart",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "第一视角推着装满卡通玩偶的购物车穿过超市，角色不断把零食抛进车内。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A first-person shopper pushes a cart full of plush characters through a supermarket as snacks keep landing inside. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/momoka_AIIllust/status/2097565673501196598",
    "sourceLabel": "X 原帖 · @momoka_AIIllust",
    "author": "🌸ももか🌸@AIイラスト",
    "publishedAt": "2026-09-09T06:00:02.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097565673501196598.mp4",
    "posterUrl": "/posters/x/2097565673501196598.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "ugc"
    ],
    "scenes": [
      "daily-life",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今回はお買い物中にポンポンお菓子を放り投げていく動画になります✨️\n登場するのは杏華とKusArt内のオリキャラ2人目となるリノアちゃんです( ੭•͈ω•͈)੭\nリノアちゃんはLoRAも公開してるので気になる方はぜひっ♬.*ﾟ\n\n⋆. ݁┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈┈ ݁.⋆\n\n現在KusArtでは、MiniMax H3を期間限定で無料で利用できるキャンペーンを開催しています✨️\n\n非会員の方でもMiniMax H3を5回無料で利用できる上、友達をKusArtに招待するとMiniMax H3の無料利用回数を追加で最大10回分獲得できます❣️(生成条件あり)\nさらに会員の方はプラン問わず、キャンペーン期間中はMiniMax H3を回数制限、生成条件なく無料で利用できます(｡•ㅅ•｡三｡•ㅅ•｡)ﾌﾝﾌﾝ\n\nこの機会にぜひリプ欄に記載しているリンクから動画生成を楽しんでみてください( ๑>ω•́ )۶\n\n#KusArt",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 8,
      "views": 749,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097567175464693951",
    "title": "三维预演与成片实时联动",
    "titleEn": "A Live Link Between 3D Blocking and Final Frames",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "简化三维场景与写实成片分屏同步前进，展示参考驱动下镜头构图和运动的对应关系。 原生视频规格为 7 秒 · 1280×536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A simple 3D blockout and photoreal output advance side by side to show how reference-driven framing and motion correspond. Native-video output: 7s · 1280×536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shiqi_yang_147/status/2097567175464693951",
    "sourceLabel": "X 原帖 · @shiqi_yang_147",
    "author": "Shiqi Yang",
    "publishedAt": "2026-09-09T06:06:00.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097567175464693951.mp4",
    "posterUrl": "/posters/x/2097567175464693951.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1280×536",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Pseudo realtime generation of Three js + minimax h3 turbo ref2v (I believe better prompt can give more consistency to follow the three js). \nJust need a autoregressive ref2v for real realtime.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 85,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097570755798331862",
    "title": "满屏像素小人的表情接力",
    "titleEn": "A Screenful of Chibi Reactions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "大量迷你动漫角色以表情格和动作格铺满画面，形成不断切换的角色反应墙。 原生视频规格为 8 秒 · 1024×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Dozens of tiny anime characters fill the frame in expression and motion cells, creating a rapidly changing reaction wall. Native-video output: 8s · 1024×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/gosrum/status/2097570755798331862",
    "sourceLabel": "X 原帖 · @gosrum",
    "author": "金のニワトリ",
    "publishedAt": "2026-09-09T06:20:13.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097570755798331862.mp4",
    "posterUrl": "/posters/x/2097570755798331862.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "1024×1024",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "これすごいなと思って再現しようと挑戦したけど、MiniMax H3（ローカル）の扱いが難しい...！動画生成のスキルが求められるなこれ\n\n適当に作ったプロンプトだとこんな感じになってしまった。まあこれでも可愛いのだが\n#AIニケちゃん",
    "engagement": {
      "replies": 3,
      "reposts": 3,
      "likes": 28,
      "views": 3825,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097571154622025782",
    "title": "艺术展吉祥物与黑猫的十五秒广告",
    "titleEn": "A 15-Second Art Exhibition Spot with Mascots and Cats",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "针织帽吉祥物与黑猫在户外场景和展览海报间切换，完成一支十五秒艺术展宣传片。 原生视频规格为 15 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A knit-cap mascot and black cats alternate between outdoor scenes and exhibition posters in a 15-second art-show spot. Native-video output: 15s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI_yukichi2910/status/2097571154622025782",
    "sourceLabel": "X 原帖 · @AI_yukichi2910",
    "author": "諭吉@11/7 AIクリエイターズマーケット2026",
    "publishedAt": "2026-09-09T06:21:48.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097571154622025782.mp4",
    "posterUrl": "/posters/x/2097571154622025782.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "anime",
      "commercial"
    ],
    "scenes": [
      "product",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#sendaiart\n\n15秒コマーシャル動画🎥✨\nMiniMax H3\n\n『Send AI Art Exhibition』\n東北から発信する AIアート展\n2026年9月12日(土)・13(日)\n12:00〜18:00\n\nギャラリーボチュー\nhttps://bochu.jp\n\n【出展クリエイター】\n城野ぱすてる\n多古と凧と凪\nねおのらしい\nてい\nKosikake\n諭吉",
    "engagement": {
      "replies": 0,
      "reposts": 4,
      "likes": 10,
      "views": 259,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097573939904196884",
    "title": "拼布猫与少女的温柔日常",
    "titleEn": "A Patchwork Cat’s Gentle Day with a Girl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "拼布纹理背景里，少女与猫咪在拥抱、惊讶和嬉闹动作间切换，保持柔和手绘质感。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Against patchwork textures, a girl and cat move through hugs, surprise, and playful beats while retaining a soft illustrated look. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aki21ai/status/2097573939904196884",
    "sourceLabel": "X 原帖 · @aki21ai",
    "author": "くろね / AI",
    "publishedAt": "2026-09-09T06:32:52.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097573939904196884.mp4",
    "posterUrl": "/posters/x/2097573939904196884.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "つぎはぎねこねこ☺️🐱✨\nだいすきねこちゃん🐱💕\n\nつぎはぎねこねこの動画の15秒バージョンに挑戦してみました！\n長めのつぎはぎねこねこ動画は初めてです✨\n\nKusArtのMiniMax H3というツールで制作してみたのです✨\n\n期間限定で無料利用できるキャンペーンを開催中ですのでもしよかったらみなさんも挑戦してみてくださいね！\n\n9月13日 23:59までです🐾✨\n#KusArt #AI動画",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 17,
      "views": 262,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097574969899401612",
    "title": "三种路径下的舞蹈动作复刻",
    "titleEn": "A Dance Motion Recreated Through Three Pipelines",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "原始动作与两种生成结果三栏并排，让同一段舞蹈的姿态复制和角色保持可以直接比较。 原生视频规格为 10 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The source motion and two generated results appear in three columns, making pose transfer and identity retention directly comparable. Native-video output: 10s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2097574969899401612",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-09-09T06:36:58.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097574969899401612.mp4",
    "posterUrl": "/posters/x/2097574969899401612.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 の V2V を使ったという動画を\nあまり見かけないんですけど、\nうまくいってないんでしょうか？\n\n・・・と思ってやってみました\n\n中央の mixamo のダンス動画を入力にして、\nref2va でダンスを完コピするように指示して作成\n\n比較として wan3.0 でも同じ条件で作りました\n\n結果、ダンスのコピーは「だいたい」できてるけど\nちょこちょこ振り付けが改変されてます\n時間ズレもあったりして、完全とは言い難いです\n\n一方、Wan3.0 のほうは完コピできてます\n（尻尾が途中なくなってますが）\n\nなんか惜しいので、もう少し試してみようと\n思ってますが、ズレずにうまく出来てる方が\nいたら教えてください・・・",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 407,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097582320555626546",
    "title": "三名刺客围猎神秘目标",
    "titleEn": "Three Assassins Hunt a Mysterious Target",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雨夜室内，三名蒙面刺客逼近目标，武器特写和阴影剪辑构成紧张的黑暗动作段落。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "On a rainy night, three masked assassins close in on a target as weapon close-ups and shadowy cuts build a tense action sequence. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/adithatipalli/status/2097582320555626546",
    "sourceLabel": "X 原帖 · @adithatipalli",
    "author": "Adithya Thatipalli",
    "publishedAt": "2026-09-09T07:06:11.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097582320555626546.mp4",
    "posterUrl": "/posters/x/2097582320555626546.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "THE OPERATOR — Hollow Hand\n\nThree assassins. One target.\nHe’s the reason they’re here now.\n\nMade with MiniMax H3 · prompt in replies",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 50,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097586680404562321",
    "title": "海风公路与城市夜色的青春乐章",
    "titleEn": "A Youthful Road Song Between Sea Breeze and City Lights",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "年轻角色骑车穿过海边公路、夕阳和城市夜景，长篇音乐段落在动画与写实镜头间切换。 原生视频规格为 102 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Young characters ride through seaside roads, sunset, and city nights in a long music sequence that alternates anime and realistic imagery. Native-video output: 102s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/neco1751662/status/2097586680404562321",
    "sourceLabel": "X 原帖 · @neco1751662",
    "author": "neco - AI Anime",
    "publishedAt": "2026-09-09T07:23:30.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097586680404562321.mp4",
    "posterUrl": "/posters/x/2097586680404562321.jpg",
    "duration": 102,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【PR】\nOiiOiiに新しく登場した「MiniMax H3 Max」でMVを作ってみました。\nmusic by @naooon68 \n\n今回使ったのはOiiOiiのMV制作機能。\n\n最初に、\n・楽曲\n・MVのストーリー\n・映像スタイル\n\nを指定。\n\nその後、エージェントと会話しながら、\nキャラクターや背景画像を生成したり、内容を確認したりしてMVの準備を進めます。\n\nここまでの事前工程にはもちろん別途時間がかかっています。\n\nそして最後にエージェントから\n「動画を生成して良いですか？」\nと確認され、生成を開始。\n\nここからが本当に速かった。\n\nMiniMax H3 Maxで動画生成を開始してから、\n各カットの動画が生成され、タイムラインにずらっと並ぶまで1分以内。\n\nこれまでなら、カットごとに動画を生成して、待って、確認して、また次を生成して……\nという作業に何時間もかかることもありました。\n\nそれが、準備した内容をもとに複数カットが一気に動画になっていく。\n\nこれだけ短時間でタイムラインまで形になると、制作フローそのものがかなり変わりそうです。\n\nOiiOiiに新モデル「MiniMax H3 Max」が登場。\n@OiiOii_AI_JP\n\nMiniMax H3 Max専用クレジットは、\n9月11日まで期間限定「1+1」キャンペーン中です。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 115,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097597295554609594",
    "title": "熔岩高空索道的第一视角穿越",
    "titleEn": "A First-Person Crossing Above Lava",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "第一视角沿钢索冲过熔岩洞穴，摇晃镜头、火光和极近距离障碍制造高空惊险感。 原生视频规格为 14 秒 · 1404×780。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A first-person camera rushes along a cable above lava as shake, firelight, and near-miss obstacles create vertigo. Native-video output: 14s · 1404×780. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Gooboberti/status/2097597295554609594",
    "sourceLabel": "X 原帖 · @Gooboberti",
    "author": "Gooboberti",
    "publishedAt": "2026-09-09T08:05:41.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097597295554609594.mp4",
    "posterUrl": "/posters/x/2097597295554609594.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1404×780",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tightrope over lava, fpv. Fal, Minimax H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 20,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097598199569785219",
    "title": "培根煎锅到入口的微距旅程",
    "titleEn": "A Macro Journey from Skillet to Bite",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "培根从煎锅、夹取到入口以极近距离连续呈现，油脂、焦边和口感细节成为画面主角。 原生视频规格为 10 秒 · 1310×760。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Bacon moves from skillet to tongs to a bite in extreme close-up, emphasizing rendered fat, crisp edges, and texture. Native-video output: 10s · 1310×760. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Gooboberti/status/2097598199569785219",
    "sourceLabel": "X 原帖 · @Gooboberti",
    "author": "Gooboberti",
    "publishedAt": "2026-09-09T08:09:16.000Z",
    "addedAt": "2026-09-09T08:44:19.841Z",
    "mediaUrl": "/media/x-2097598199569785219.mp4",
    "posterUrl": "/posters/x/2097598199569785219.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1310×760",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Bacon! Fal Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 16,
      "snapshotAt": "2026-09-09T08:44:19.841Z"
    },
    "approvedAt": "2026-09-09T08:44:19.841Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097837462512951458",
    "title": "液态金属凝成未来跑车",
    "titleEn": "Liquid Metal Forms a Futuristic Sports Car",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下分屏对比同一汽车广告提示词：悬浮金属球铺展成流线型跑车，黑色镜面棚内的灯带与车身反射同步亮起。 原生视频规格为 16 秒 · 1080×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A split-screen comparison applies the same automotive-ad prompt as a hovering metal sphere unfolds into a streamlined sports car, with light bars and body reflections emerging in a black mirrored studio. Native-video output: 16s · 1080×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A hyper-realistic, photorealistic 8K CGI advertisement sequence for a futuristic luxury electric concept sports car, styled like a high-end commercial with a sleek, powerful, and sophisticated mood.\n\nSetting & EnvironmentSet in an infinite, minimalist virtual studio with a deep black void background and a perfectly polished, highly reflective black obsidian floor that crystal-clearly mirrors the scene. The atmosphere is crystal clear with subtle light bloom.\nLighting\n\nKey Light: A cool-white (6000K) softbox panel light descends from above, growing larger to cast clean defining highlights over the final car.\n\nEmissive Accent Light: Silvery-white and vibrant electric-blue light glows directly from the fluid liquid metal, projecting dynamic reflections onto the floor.\n\nSubject Transformation\n\nInitial Form: A single, pristine sphere of mercury-like liquid metal hovering above the floor, rippling with high reflectivity and surface tension.\n\nFinal Form: A low-profile, aerodynamic futuristic sports car with smooth, single-piece molded body panels in deep gloss black pearlescent paint that subtly shifts to dark blue, featuring razor-sharp LED light bars.\n\nCamera ChoreographyAn uninterrupted, dynamic camera move starting as an extreme close-up macro (85mm, f/1.8 shallow depth of field) focused on the hovering liquid drop. The camera smoothly pulls back and accelerates into a sweeping clockwise orbit, transitioning to a wide shot (35mm, f/5.6 deep focus) to reveal the complete vehicle sharp and in full view.\n\nAction & Animation Sequence\n0–2s: The liquid metal sphere slowly pulses and ripples to a low-frequency bass hum.\n\n2–5s: The sphere explodes outward in zero-gravity as liquid tendrils stretch to form the car's wireframe, traced by glowing electric-blue energy lines, accompanied by high-tech chimes and metallic crackles.\n\n5–7s: Liquid metal flows over the energy frame, solidifying into smooth body panels and forming wheels to a heavy bass thrum.\n\n7–8s: The car gently settles onto the reflective obsidian floor, its razor-sharp LED headlights and taillights flash on with a powerful pulse, fading into the subtle hum of a high-tech electric motor.",
    "promptSourceUrl": "https://x.com/IqrasaifiAI/status/2097837462512951458",
    "sourceUrl": "https://x.com/IqrasaifiAI/status/2097837462512951458",
    "sourceLabel": "X 原帖 · @IqrasaifiAI",
    "author": "Iqra Saifi",
    "publishedAt": "2026-09-10T00:00:01.000Z",
    "addedAt": "2026-09-10T00:20:56.062Z",
    "mediaUrl": "/media/x-2097837462512951458.mp4",
    "posterUrl": "/posters/x/2097837462512951458.jpg",
    "duration": 16,
    "aspectRatio": "portrait",
    "resolution": "1080×1440",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Seedance 2.5 vs minimax h3 , for me h3 won \n\nPrompt:\n\nA hyper-realistic, photorealistic 8K CGI advertisement sequence for a futuristic luxury electric concept sports car, styled like a high-end commercial with a sleek, powerful, and sophisticated mood.\n\nSetting & EnvironmentSet in an infinite, minimalist virtual studio with a deep black void background and a perfectly polished, highly reflective black obsidian floor that crystal-clearly mirrors the scene. The atmosphere is crystal clear with subtle light bloom.\nLighting\n\nKey Light: A cool-white (6000K) softbox panel light descends from above, growing larger to cast clean defining highlights over the final car.\n\nEmissive Accent Light: Silvery-white and vibrant electric-blue light glows directly from the fluid liquid metal, projecting dynamic reflections onto the floor.\n\nSubject Transformation\n\nInitial Form: A single, pristine sphere of mercury-like liquid metal hovering above the floor, rippling with high reflectivity and surface tension.\n\nFinal Form: A low-profile, aerodynamic futuristic sports car with smooth, single-piece molded body panels in deep gloss black pearlescent paint that subtly shifts to dark blue, featuring razor-sharp LED light bars.\n\nCamera ChoreographyAn uninterrupted, dynamic camera move starting as an extreme close-up macro (85mm, f/1.8 shallow depth of field) focused on the hovering liquid drop. The camera smoothly pulls back and accelerates into a sweeping clockwise orbit, transitioning to a wide shot (35mm, f/5.6 deep focus) to reveal the complete vehicle sharp and in full view.\n\nAction & Animation Sequence\n0–2s: The liquid metal sphere slowly pulses and ripples to a low-frequency bass hum.\n\n2–5s: The sphere explodes outward in zero-gravity as liquid tendrils stretch to form the car's wireframe, traced by glowing electric-blue energy lines, accompanied by high-tech chimes and metallic crackles.\n\n5–7s: Liquid metal flows over the energy frame, solidifying into smooth body panels and forming wheels to a heavy bass thrum.\n\n7–8s: The car gently settles onto the reflective obsidian floor, its razor-sharp LED headlights and taillights flash on with a powerful pulse, fading into the subtle hum of a high-tech electric motor.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 62,
      "snapshotAt": "2026-09-10T00:20:56.063Z"
    },
    "approvedAt": "2026-09-10T00:20:56.062Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097831435147317348",
    "title": "虹翼神像轰击雪原城邦",
    "titleEn": "A Rainbow-Winged Deity Strikes a Frozen City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "手绘虹翼神像在极光下掠过云海，城市中央升起发光尖塔，随后多轮火球在雪原城邦中爆炸。 原生视频规格为 12 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hand-drawn rainbow-winged deity crosses clouds beneath an aurora, a luminous spire rises from the city, and successive fireballs detonate across the frozen settlement. Native-video output: 12s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AuntSolarClock/status/2097831435147317348",
    "sourceLabel": "X 原帖 · @AuntSolarClock",
    "author": "Aunt Solar Clock",
    "publishedAt": "2026-09-09T23:36:04.000Z",
    "addedAt": "2026-09-10T00:20:56.063Z",
    "mediaUrl": "/media/x-2097831435147317348.mp4",
    "posterUrl": "/posters/x/2097831435147317348.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Heavenly Principles destroying Hyperborea\n2026, colorized\n\nI took a sketch of HP that I drew myself, and used it as a first frame reference for MiniMax H3, an open source AI video model which came out recently. It's shockingly good. Ran locally on my 3090\n\nAlso, fuck corpo AI",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 156,
      "snapshotAt": "2026-09-10T00:20:56.063Z"
    },
    "approvedAt": "2026-09-10T00:20:56.063Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2097828723152355528",
    "title": "真人舞步与红发动漫角色对照",
    "titleEn": "Live Dance Beside a Red-Haired Anime Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三栏画面对照真人舞蹈与两种角色动画结果，红发动漫角色跟随真人做出抬手、转身和摆臂动作。 原生视频规格为 5 秒 · 1440×944。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three panels compare a live dance with two character-animation results as a red-haired anime character follows raised hands, turns, and arm gestures. Native-video output: 5s · 1440×944. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/toyxyz3/status/2097828723152355528",
    "sourceLabel": "X 原帖 · @toyxyz3",
    "author": "toyxyz",
    "publishedAt": "2026-09-09T23:25:18.000Z",
    "addedAt": "2026-09-10T01:23:46.842Z",
    "mediaUrl": "/media/x-2097828723152355528.mp4",
    "posterUrl": "/posters/x/2097828723152355528.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1440×944",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "ugc"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 Viggle-Animate test #comfyui",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 12,
      "views": 882,
      "snapshotAt": "2026-09-10T00:12:08.000Z"
    },
    "approvedAt": "2026-09-10T01:23:46.842Z",
    "editorialBasis": "Original X post loaded with matching status ID and author; downloaded native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097828332553244683",
    "title": "棋桌旁的喷火巨龙",
    "titleEn": "A Fire-Breathing Dragon at the Chess Table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "酒馆里，红发女孩与绿色巨龙对弈。巨龙朝棋盘喷出火焰，镜头随后切到女孩被熏黑的惊讶表情。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-haired girl plays chess with a green dragon in a tavern. The dragon breathes fire across the board before the camera cuts to her soot-covered, surprised face. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/charis_ai/status/2097828332553244683",
    "sourceLabel": "X 原帖 · @charis_ai",
    "author": "CharisTheAI",
    "publishedAt": "2026-09-09T23:23:44.000Z",
    "addedAt": "2026-09-10T01:23:46.842Z",
    "mediaUrl": "/media/x-2097828332553244683.mp4",
    "posterUrl": "/posters/x/2097828332553244683.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Check mate!  \n\nThis is MiniMax H3 on Magnific.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 51,
      "snapshotAt": "2026-09-10T00:12:08.000Z"
    },
    "approvedAt": "2026-09-10T01:23:46.842Z",
    "editorialBasis": "Original X post loaded with matching status ID and author; downloaded native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097821893382582320",
    "title": "穿过森林里的圆环长廊",
    "titleEn": "Through a Forest of Circular Portals",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从漂浮花朵与枝叶的近景进入树林，沿着一连串圆环穿过彩色雾气，最后来到洒满阳光的林间小径。 原生视频规格为 19 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The view moves from floating blossoms and branches into a forest, passing through successive rings and colored mist before reaching a sunlit woodland path. Native-video output: 19s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aicreataro/status/2097821893382582320",
    "sourceLabel": "X 原帖 · @aicreataro",
    "author": "aicreataro",
    "publishedAt": "2026-09-09T22:58:09.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097821893382582320.mp4",
    "posterUrl": "/posters/x/2097821893382582320.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "変えたいものが多すぎる日は、 変えないものを一つ決めるところから始める。\nMiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 62,
      "snapshotAt": "2026-09-10T00:12:08.854Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097816484575670765",
    "title": "红衣女子从夜街走入晨间集市",
    "titleEn": "A Red-Jacketed Walk from Night Street to Morning Market",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子手持纸杯走过霓虹街道，环境随后过渡到阳光下的集市。长镜头围绕同一人物展开，结尾附有作者的制作参数说明。 原生视频规格为 123 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman carrying a paper cup walks through a neon-lit street before the setting transitions to a sunny market. The extended sequence follows the same character and ends with the creator's production notes. Native-video output: 123s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jose_medina/status/2097816484575670765",
    "sourceLabel": "X 原帖 · @jose_medina",
    "author": "Jose Medina",
    "publishedAt": "2026-09-09T22:36:40.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097816484575670765.mp4",
    "posterUrl": "/posters/x/2097816484575670765.jpg",
    "duration": 123,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Most video models are a request and a response. You send a prompt, you wait, a file\ncomes back.\n\nMiniMax H3 Max Director 1.1, which landed on fal on September 9, is not that. It is\na session you hold open while it generates, and you steer it mid-stream.\n\nSo I built a console to drive it. The endpoint is `minimax/h3-max/director`.\n118 seconds, one session, three worlds, no edit, 1080p native.\n\nWhat is on screen:\n\n0:35 the model takes a keyframe I injected while it was already rendering, and by\n0:48 the night street has become a morning market\n0:50 it takes an audio track handed to it mid-generation\n1:37 a second keyframe lands, and by 1:48 the world has gone cold\n\nDo not take my word for the \"no cuts\" part. Run a cut detector over the file:\n\n    ffmpeg -i take.mp4 -filter:v \"select='gt(scene,0.4)',metadata=print\" -f null -\n\nZero hits. For calibration, I cut two of my own stills together and that scores\n0.549.\n\nThe highest score anywhere in the file is 0.359, at 1:58, and that is the closing\ncard I added. It is the only cut in the video and it is mine.\n\nThe woman, the red jacket and the paper cup never change. Everything else does.\nThat is the whole demo.\n\nFour things I got wrong before I got it right, if you are building on this:\n\n1. protocol_version is a configure-only field. Send it on a prompt message and the\nserver refuses the whole message as invalid_message. That cost me a take.\n\n2. There is 10 to 15 seconds of warm-up before `configured` lands, and it holds the\nopening frame while it warms. Fire a prompt before that and you are talking to a\nsession that does not exist yet. Record it and your video opens on a frozen subject.\n\n3. A prompt is not applied when you send it. Across six sends the gap ran from 2\nto 20 seconds, because the model finishes the chunk it is already generating\nfirst, and the change then needs another chunk before it is on screen. My first\nattempt put the last keyframe in 20 seconds too late: the model acknowledged it at\n95 seconds of a 98 second take and it never reached the picture at all. Budget 25\nseconds per beat.\n\n4. Billing has a per-session minimum. A 20 second session costs the same as a 60\nsecond one, so aborting the instant the server refuses a message is free insurance.\nWire up the abort.\n\nAt 1080p it is generating right at the edge. The buffer emptied four times and the\nserver reported deadline_missed three times. Nothing visible survived into the\nrecording: mpdecimate finds no duplicated frames anywhere in it.\n\nCost of the take: $4.15. 1080p bills at double, and the receipt charges by seconds\nof video actually generated rather than wall clock.\n\nThree of the four features in the 1.1 changelog are in this clip: keyframes inserted\nthroughout a generation, audio introduced at any point with native conditioning, and\nnative 1080p. The fourth is continuous runs up to 15 minutes, and I stopped at two.\n\n@fal",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 162,
      "snapshotAt": "2026-09-10T00:12:09.193Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097816304111808620",
    "title": "金色礼球展开粉丝庆祝字幕",
    "titleEn": "A Golden Celebration Sphere Unfolds into Thank-You Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金色火花引出礼球、烟花与纸屑，球体展开后呈现5300人的庆祝数字、作者名称及感谢文字。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Golden sparks introduce a celebration sphere, fireworks, and confetti. The sphere opens to reveal a 5,300-follower milestone, the creator's name, and thank-you lettering. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "高品質な祝福タイポグラフィアニメーション\n表示文字：「X」「フォロワー」「5300人」「YaReYaRu」「ありがとう！」\n綴りと大文字・小文字を正確に保持する\n\n【ビジュアル】\n深いミッドナイトブルーを背景に、ゴールド、ホワイト、鮮やかなエメラルドグリーンで統一\n主役は大きく読みやすい文字、くす玉と花火は文字の登場や変形に連動する\n極太サンセリフの数字、洗練された立体文字、薄い金属反射、奥行きのある紙吹雪\n上品な授賞式の華やかさと、SNSらしい軽快で力強いモーションを組み合わせる\n重要な文字は中央の安全領域に配置、画面下端や右端に寄せない\n\n【0〜2秒｜期待をつくる】\n暗い画面中央で小さな金色の火花が弾け、白い「X」が瞬発的に出現\n文字が一度だけ大きく弾み、細い光のリングが外側へ広がる\n「X」が上部中央へ縮小移動、その下に「フォロワー」が一文字ずつ軽快に着地\n背後から金色の細いリボンが左右へ走り、中央を空ける\n\n【2〜5秒｜くす玉登場】\n画面上から豪華な金色のくす玉が中央へ降りてくる\n上下に重なる金色の鱗状装飾、緑の房飾り、左右対称の球体\n減速して止まり、小さく一度だけ揺れる\nカメラがゆっくり接近、周囲の光が絞られ、くす玉の継ぎ目から緑と金の光が漏れる\n下に伸びた一本の紐がぴんと張り、開く直前に一瞬静止\n\n【5〜8秒｜5300人を大きく見せる】\n強い打音と同時にくす玉が左右へぱかっと開く\n左右の半球は吊られたまま残り、内側から金と緑の紙吹雪、長いカールリボンが勢いよく放出される\n中央から白い垂れ幕が下へほどけ、その表面に極太の「5300人」が現れる\n「5300」の数字が垂れ幕から手前へ立体的にせり出し、画面中央を大きく占める\n数字は一度だけ軽くオーバーシュートして定位置へ収まり、「人」は数字の右下へ整然と配置\n同時に背景の左右上方で大輪の花火が開く\n短い光の衝撃と画面の微振動で到達の喜びを強調、文字が読めなくなる全面白飛びは起こさない\n\n【8〜11秒｜名前を主役に】\n「5300人」が少し上へ移動し、その下へ「YaReYaRu」が左から右へ一文字ずつ滑り込む\n文字の着地に合わせて金色の小さな火花が散り、最後の「u」が収まると名前全体を緑の光が一度だけ横切る\nカメラが緩やかに引き、上部の「X」、その下の「フォロワー」、中央の「5300人」、下部の「YaReYaRu」がひとつの祝福ポスターとして整う\n背景に高さと大きさの異なる花火が時間差で開花\n紙吹雪は文字の背後を中心に落下し、少量だけ手前を横切って奥行きを出す\n\n【11〜13秒｜ありがとう】\n「YaReYaRu」の下に「ありがとう！」が柔らかな弾みをつけて出現\n左右から金色のリボンが弧を描き、文字全体を囲む祝福のフレームになる\n最後に背景中央上方で最大の金色の花火が開き、その内側に緑の小さな花火が重なる\n花火の明るさに合わせ、文字の縁が一瞬だけ輝く\n\n【13〜15秒｜読ませる完成画面】\n文字とカメラを静止させ、完成したレイアウトを2秒間保持\n「X」\n「フォロワー」\n「5300人」\n「YaReYaRu」\n「ありがとう！」\n「5300人」を最も大きく、次に「YaReYaRu」を強調\n開いたくす玉を上部左右に残し、花火の残光とゆっくり落ちる紙吹雪だけが動く\n最後まで文字を鮮明に読める状態で終了\n\n【音・編集】\n華やかなエレクトロポップ、文字の着地に短い打音、くす玉開放に祝砲、花火に奥行きのある炸裂音\n5秒のくす玉開放で音楽が大きく展開し、最後は明るい余韻\nナレーションなし\nカットの切り替えよりも、文字の位置移動、拡大縮小、リボンの軌道で場面を滑らかにつなぐ\n文字の誤字、重複、勝手な文言追加、数字の変化、過度な点滅、粒状ノイズ、ウォーターマークは禁止",
    "promptSourceUrl": "https://x.com/YaReYaRu30Life/status/2097816304111808620",
    "sourceUrl": "https://x.com/YaReYaRu30Life/status/2097816304111808620",
    "sourceLabel": "X 原帖 · @YaReYaRu30Life",
    "author": "ヤレヤル（AI動画/AI画像/AI音楽）",
    "publishedAt": "2026-09-09T22:35:57.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097816304111808620.mp4",
    "posterUrl": "/posters/x/2097816304111808620.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "5300フォロワー様\nいつも皆さんありがとうございます！\n\n100キリ報告は最近してませんでしたが\nMiniMaxH3の登場でタイポグラフィなどもやりやすくなったので\nプロンプトと共に感謝の気持ちをお伝えします！\n@Hailuo_AI　#MiniMaxH3\nプロンプト\nーーーーー\n高品質な祝福タイポグラフィアニメーション\n表示文字：「X」「フォロワー」「5300人」「YaReYaRu」「ありがとう！」\n綴りと大文字・小文字を正確に保持する\n\n【ビジュアル】\n深いミッドナイトブルーを背景に、ゴールド、ホワイト、鮮やかなエメラルドグリーンで統一\n主役は大きく読みやすい文字、くす玉と花火は文字の登場や変形に連動する\n極太サンセリフの数字、洗練された立体文字、薄い金属反射、奥行きのある紙吹雪\n上品な授賞式の華やかさと、SNSらしい軽快で力強いモーションを組み合わせる\n重要な文字は中央の安全領域に配置、画面下端や右端に寄せない\n\n【0〜2秒｜期待をつくる】\n暗い画面中央で小さな金色の火花が弾け、白い「X」が瞬発的に出現\n文字が一度だけ大きく弾み、細い光のリングが外側へ広がる\n「X」が上部中央へ縮小移動、その下に「フォロワー」が一文字ずつ軽快に着地\n背後から金色の細いリボンが左右へ走り、中央を空ける\n\n【2〜5秒｜くす玉登場】\n画面上から豪華な金色のくす玉が中央へ降りてくる\n上下に重なる金色の鱗状装飾、緑の房飾り、左右対称の球体\n減速して止まり、小さく一度だけ揺れる\nカメラがゆっくり接近、周囲の光が絞られ、くす玉の継ぎ目から緑と金の光が漏れる\n下に伸びた一本の紐がぴんと張り、開く直前に一瞬静止\n\n【5〜8秒｜5300人を大きく見せる】\n強い打音と同時にくす玉が左右へぱかっと開く\n左右の半球は吊られたまま残り、内側から金と緑の紙吹雪、長いカールリボンが勢いよく放出される\n中央から白い垂れ幕が下へほどけ、その表面に極太の「5300人」が現れる\n「5300」の数字が垂れ幕から手前へ立体的にせり出し、画面中央を大きく占める\n数字は一度だけ軽くオーバーシュートして定位置へ収まり、「人」は数字の右下へ整然と配置\n同時に背景の左右上方で大輪の花火が開く\n短い光の衝撃と画面の微振動で到達の喜びを強調、文字が読めなくなる全面白飛びは起こさない\n\n【8〜11秒｜名前を主役に】\n「5300人」が少し上へ移動し、その下へ「YaReYaRu」が左から右へ一文字ずつ滑り込む\n文字の着地に合わせて金色の小さな火花が散り、最後の「u」が収まると名前全体を緑の光が一度だけ横切る\nカメラが緩やかに引き、上部の「X」、その下の「フォロワー」、中央の「5300人」、下部の「YaReYaRu」がひとつの祝福ポスターとして整う\n背景に高さと大きさの異なる花火が時間差で開花\n紙吹雪は文字の背後を中心に落下し、少量だけ手前を横切って奥行きを出す\n\n【11〜13秒｜ありがとう】\n「YaReYaRu」の下に「ありがとう！」が柔らかな弾みをつけて出現\n左右から金色のリボンが弧を描き、文字全体を囲む祝福のフレームになる\n最後に背景中央上方で最大の金色の花火が開き、その内側に緑の小さな花火が重なる\n花火の明るさに合わせ、文字の縁が一瞬だけ輝く\n\n【13〜15秒｜読ませる完成画面】\n文字とカメラを静止させ、完成したレイアウトを2秒間保持\n「X」\n「フォロワー」\n「5300人」\n「YaReYaRu」\n「ありがとう！」\n「5300人」を最も大きく、次に「YaReYaRu」を強調\n開いたくす玉を上部左右に残し、花火の残光とゆっくり落ちる紙吹雪だけが動く\n最後まで文字を鮮明に読める状態で終了\n\n【音・編集】\n華やかなエレクトロポップ、文字の着地に短い打音、くす玉開放に祝砲、花火に奥行きのある炸裂音\n5秒のくす玉開放で音楽が大きく展開し、最後は明るい余韻\nナレーションなし\nカットの切り替えよりも、文字の位置移動、拡大縮小、リボンの軌道で場面を滑らかにつなぐ\n文字の誤字、重複、勝手な文言追加、数字の変化、過度な点滅、粒状ノイズ、ウォーターマークは禁止",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 8,
      "views": 394,
      "snapshotAt": "2026-09-10T00:12:09.196Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097815684558536790",
    "title": "雨后灯笼街上的浴衣回眸",
    "titleEn": "A Yukata Glance on a Lantern-Lit Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从浴衣花纹的近景拉开，女子站在雨后石板街上，暖色灯笼映照路面，随后转身回望镜头。 原生视频规格为 17 秒 · 672×896。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera pulls back from a close-up of yukata fabric to a woman on a rain-soaked stone street. Warm lanterns reflect on the pavement as she turns back toward the camera. Native-video output: 17s · 672×896. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaekagamiAi/status/2097815684558536790",
    "sourceLabel": "X 原帖 · @MaekagamiAi",
    "author": "前鏡",
    "publishedAt": "2026-09-09T22:33:29.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097815684558536790.mp4",
    "posterUrl": "/posters/x/2097815684558536790.jpg",
    "duration": 17,
    "aspectRatio": "portrait",
    "resolution": "672×896",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Inspired by creatures native to Hareno-Kuni, Japan.☀️\n\n#MiniMaxH3",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 11,
      "views": 347,
      "snapshotAt": "2026-09-10T00:12:09.298Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097808775554212167",
    "title": "抱着西瓜滑进泳池",
    "titleEn": "A Watermelon Mishap by the Pool",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子抱着西瓜坐到泳池边，失去平衡落入水中；切开的西瓜留在池边，她扶着池沿露出错愕表情。 原生视频规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman carrying a watermelon loses her balance at the edge of a pool and falls in. Watermelon slices remain on the deck as she holds the edge with a startled expression. Native-video output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/safa_dayo/status/2097808775554212167",
    "sourceLabel": "X 原帖 · @safa_dayo",
    "author": "サファ【AIイラスト多め】",
    "publishedAt": "2026-09-09T22:06:02.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097808775554212167.mp4",
    "posterUrl": "/posters/x/2097808775554212167.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "ugc"
    ],
    "scenes": [
      "daily-life",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "プールサイドの悲劇🤯🍉\n\nHailuo AIのMiniMax H3で生成しています",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 530,
      "snapshotAt": "2026-09-10T00:12:09.175Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097807335729344809",
    "title": "红日与椅子穿行拼贴城市",
    "titleEn": "A Red Sun and Chair in a Collage City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色椅子、红色圆盘与报纸质感楼房组成超现实街景，碎片向天空散开，最后重新排列成带日文字幕的城市画面。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black chair, red disc, and newspaper-textured buildings form a surreal street. Fragments scatter into the sky before reassembling into a city composition with Japanese typography. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2097807335729344809",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-09-09T22:00:18.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097807335729344809.mp4",
    "posterUrl": "/posters/x/2097807335729344809.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「少しズレると世界はもっと動き出す！」\n\n2SOFT DADA（やわらかいダダイズム）／2.5Dコラージュ／キネティックタイポグラフィ\n\n#OiiOii のMinimaxH3 Max Tauboで作成",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 38,
      "snapshotAt": "2026-09-10T00:12:09.217Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097807268397842924",
    "title": "星河符文中的银发少女",
    "titleEn": "A Silver-Haired Girl amid Cosmic Runes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发动漫少女站在发光阶梯上，背后环绕星河与漂浮符文，镜头靠近时她抬手微笑并做出手势。 原生视频规格为 4 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired anime girl stands on glowing steps amid cosmic spirals and floating runes, smiling and raising her hand as the camera moves closer. Native-video output: 4s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yonaguniUDON/status/2097807268397842924",
    "sourceLabel": "X 原帖 · @yonaguniUDON",
    "author": "蛸壺_AIラクルゥ開発",
    "publishedAt": "2026-09-09T22:00:02.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097807268397842924.mp4",
    "posterUrl": "/posters/x/2097807268397842924.jpg",
    "duration": 4,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "無量空処\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 11,
      "views": 76,
      "snapshotAt": "2026-09-10T00:12:09.617Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097799037235712189",
    "title": "老妇人摘下树上的心脏果实",
    "titleEn": "An Elderly Woman Picks a Heart-Shaped Fruit",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "幽暗树林里，老妇人伸手触碰枝头悬挂的心脏状果实，镜头在她的手、果实特写和面部表情之间切换。 原生视频规格为 15 秒 · 2496×1408。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a dark forest, an elderly woman reaches for an anatomical heart-shaped fruit hanging from a branch. Close-ups alternate between her hands, the fruit, and her face. Native-video output: 15s · 2496×1408. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxCasu/status/2097799037235712189",
    "sourceLabel": "X 原帖 · @MaxCasu",
    "author": "max casu",
    "publishedAt": "2026-09-09T21:27:20.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097799037235712189.mp4",
    "posterUrl": "/posters/x/2097799037235712189.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2496×1408",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "photoreal"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Tree of Hearts\nMade with MiniMax H3 model on @ViggleAI",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 109,
      "snapshotAt": "2026-09-10T00:12:09.835Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097798749514817916",
    "title": "机械草图中睁开的眼睛",
    "titleEn": "An Eye Opens inside a Mechanical Drawing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "纸面坐标与手绘齿轮逐渐占满画面，指针在草图结构中转动，一只写实眼睛从绘制的几何框架中显现。 原生视频规格为 9 秒 · 1664×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Paper coordinates and sketched gears fill the frame as pointers rotate through the drawing and a realistic eye appears within its geometric framework. Native-video output: 9s · 1664×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/plasm0/status/2097798749514817916",
    "sourceLabel": "X 原帖 · @plasm0",
    "author": "ρŁ𝐀𝔰Ｍʘ",
    "publishedAt": "2026-09-09T21:26:11.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097798749514817916.mp4",
    "posterUrl": "/posters/x/2097798749514817916.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1664×928",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 - Singularity test",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 12,
      "views": 527,
      "snapshotAt": "2026-09-10T00:12:09.617Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097798489161990458",
    "title": "动漫少女转身的参考图对照",
    "titleEn": "Reference-Image Variations for an Anime Character Turn",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "多组画面对照同一动漫角色转头与转身的结果，下方展示参考图组合，上方标注各组生成耗时。 原生视频规格为 26 秒 · 1024×2292。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The clip compares an anime character turning her head and body across several reference-image setups, displaying reference sheets below and generation timings above. Native-video output: 26s · 1024×2292. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2097798489161990458",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-09-09T21:25:09.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097798489161990458.mp4",
    "posterUrl": "/posters/x/2097798489161990458.jpg",
    "duration": 26,
    "aspectRatio": "portrait",
    "resolution": "1024×2292",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 ref2vaの参照画像の枚数と解像度による生成時間の比較。個別画像3枚より、同じ解像度で三面図っぽく1枚で入れた方が生成時間が5%（9秒）くらい減る。画質も悪くないし、無駄に参照画像は増やさない方が良さそう。",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 11,
      "views": 927,
      "snapshotAt": "2026-09-10T00:12:09.630Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097796113936117932",
    "title": "老虎机之间的老虎自拍",
    "titleEn": "A Tiger Selfie among Slot Machines",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "老虎面对近距离镜头抬起前爪，周围老虎机闪烁彩灯，它张开嘴巴并在镜头前摇晃。 原生视频规格为 15 秒 · 1296×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiger raises its front paws toward a close camera between brightly lit slot machines, opening its mouth and moving in front of the lens. Native-video output: 15s · 1296×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TheAwakenOne619/status/2097796113936117932",
    "sourceLabel": "X 原帖 · @TheAwakenOne619",
    "author": "TheAwakenOne",
    "publishedAt": "2026-09-09T21:15:43.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097796113936117932.mp4",
    "posterUrl": "/posters/x/2097796113936117932.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1296×720",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This tiger is having too much fun in the casino lol\n\nMiniMax H3\nKrea2 - image",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-09-10T00:12:10.174Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097794301921214974",
    "title": "格子舞台切换漫画街景",
    "titleEn": "A Grid Dance Stage Shifts into a Comic Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "黑发Q版角色在橙色网格空间中摆臂、俯身起舞，背景随后切换为高饱和漫画街道，动作继续衔接。 原生视频规格为 8 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired chibi character dances and bends across an orange grid stage before the backdrop changes to a saturated comic-style street while the movement continues. Native-video output: 8s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sankakuten91256/status/2097794301921214974",
    "sourceLabel": "X 原帖 · @sankakuten91256",
    "author": "さんかくてん",
    "publishedAt": "2026-09-09T21:08:31.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097794301921214974.mp4",
    "posterUrl": "/posters/x/2097794301921214974.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "dance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\nキャラクターのダンスの参照画像と複数の背景の参照でも可能です",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 688,
      "snapshotAt": "2026-09-10T00:12:10.335Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097791816338669672",
    "title": "白衣精灵的环绕影像与三维视图",
    "titleEn": "Orbiting a White-Robed Elf in Three Dimensions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白衣精灵从参考图进入环绕展示，镜头掠过背部与侧脸，后半段展示由影像衔接的三维查看结果。 原生视频规格为 69 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-robed elf moves from a reference image into an orbiting video view, revealing her back and profile before the clip shows the subsequent 3D viewing result. Native-video output: 69s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/gabu3d_pl/status/2097791816338669672",
    "sourceLabel": "X 原帖 · @gabu3d_pl",
    "author": "Gabu",
    "publishedAt": "2026-09-09T20:58:38.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097791816338669672.mp4",
    "posterUrl": "/posters/x/2097791816338669672.jpg",
    "duration": 69,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just a little Proof Of Concept that came to my mind:\nKrea2 Image ->  Minimax H3 360 Vid -> Gaussian splatting ->  VR (tbd) \nIt can be done :)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 269,
      "snapshotAt": "2026-09-10T00:12:10.166Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097787535653236973",
    "title": "猫耳店员绕着送餐机器人转圈",
    "titleEn": "A Cat-Eared Server Circles a Delivery Robot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "温暖咖啡馆中，橙发猫耳角色围着装有餐点的机器人跑动，环绕镜头逐渐展示柜台、桌椅与落地窗。 原生视频规格为 10 秒 · 480×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a warm cafe, an orange-haired cat-eared character circles a food delivery robot as an orbiting view reveals the counter, tables, and tall windows. Native-video output: 10s · 480×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2097787535653236973",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-09-09T20:41:38.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097787535653236973.mp4",
    "posterUrl": "/posters/x/2097787535653236973.jpg",
    "duration": 10,
    "aspectRatio": "square",
    "resolution": "480×480",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 multi-angle\n\n映像制作にも普通に便利そうですが、\n\nぐるっとカメラを一回しして\n部屋とかオブジェとかモデリングするのに\n使えそうだな、とか思いました\n\n（後で試してみます）",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 488,
      "snapshotAt": "2026-09-10T00:12:10.194Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097776859727397279",
    "title": "从底部向上冻结的瀑布",
    "titleEn": "A Waterfall Freezes from the Bottom Up",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "固定机位正对林间瀑布，水流逐段变成白色冰柱，冻结的结构逐渐向上延伸并覆盖整道落水。 原生视频规格为 15 秒 · 4096×2340。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fixed camera faces a forest waterfall as its flow becomes white ice in stages, with the frozen structure climbing upward across the falling water. Native-video output: 15s · 4096×2340. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2097776859727397279",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-09T19:59:12.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097776859727397279.mp4",
    "posterUrl": "/posters/x/2097776859727397279.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "4096×2340",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3. The waterfall freezes upward and the sound goes with it.\n\nPrompt: \"a waterfall roar that fades in steps as the ice climbs\"\n\nDuration: 15s. Camera: locked, across the pool.\n\nThe sound line and the picture describe the same staircase. When they agree, the model has two reasons to do it right.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 11,
      "views": 132,
      "snapshotAt": "2026-09-10T00:12:10.679Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097772293741052285",
    "title": "黄裙女孩遇见长鹿角的紫狐",
    "titleEn": "A Girl in Yellow Meets an Antlered Purple Fox",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黄裙女孩在苔藓森林中遇到长着鹿角的紫色狐狸，从惊讶地后退转为伸手触碰它的头部。 原生视频规格为 15 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl in a yellow dress meets a purple antlered fox in a mossy forest, moving from a startled reaction to gently touching its head. Native-video output: 15s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/godblesstoto/status/2097772293741052285",
    "sourceLabel": "X 原帖 · @godblesstoto",
    "author": "🙏🌧🌍",
    "publishedAt": "2026-09-09T19:41:04.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097772293741052285.mp4",
    "posterUrl": "/posters/x/2097772293741052285.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is really good.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 1504,
      "snapshotAt": "2026-09-10T00:12:10.679Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097769741771276689",
    "title": "黑衣女子与悬浮陶罐",
    "titleEn": "A Woman in Black with Floating Ceramic Vessels",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣女子俯身靠近一组黑色陶罐，罐体在身前排列悬浮；视频左侧保留平台发布时的功能说明。 原生视频规格为 6 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman dressed in black leans toward black ceramic vessels arranged in suspension. The left side retains the platform's accompanying feature announcement. Native-video output: 6s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/recraftai/status/2097769741771276689",
    "sourceLabel": "X 原帖 · @recraftai",
    "author": "Recraft",
    "publishedAt": "2026-09-09T19:30:55.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097769741771276689.mp4",
    "posterUrl": "/posters/x/2097769741771276689.jpg",
    "duration": 6,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Turbo is now in Recraft Studio.\n\nThey called it Turbo for a reason. Fast video generation, no waiting around.\n\nGive it a try ↓",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 7,
      "views": 854,
      "snapshotAt": "2026-09-10T00:12:10.679Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097766233374367873",
    "title": "双栏魔法师角色动作对照",
    "titleEn": "Side-by-Side Mage Character Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发精灵与紫发角色分别持着蓝色和红色法杖，在双栏对照中做出同步挥杖与施法动作，结尾保留原动画课程字幕。 原生视频规格为 5 秒 · 1600×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde elf and a purple-haired character hold blue and red staffs in a side-by-side comparison, matching staff swings and spellcasting before the original animation course title appears. Native-video output: 5s · 1600×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SmallKino_/status/2097766233374367873",
    "sourceLabel": "X 原帖 · @SmallKino_",
    "author": "SmallKino",
    "publishedAt": "2026-09-09T19:16:59.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097766233374367873.mp4",
    "posterUrl": "/posters/x/2097766233374367873.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1600×1088",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I remember how half a year ago I tried to make character replacement with this Frieren clip from @Sykosan using Wan MoCha. It failed completely. And now six month later MiniMax H3 can do it perfectly at the first try.\n\n#MiniMaxH3 #Frieren",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 9,
      "views": 278,
      "snapshotAt": "2026-09-10T00:12:10.694Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097764585553956895",
    "title": "墨镜男子穿越霓虹城市",
    "titleEn": "A Sunglasses-Clad Man Flies through a Neon City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣男子在彩色灯带下行走，随后画面转为他穿行于城市楼宇之间的飞行镜头，视频带有西班牙语字幕。 原生视频规格为 18 秒 · 1080×1944。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in black walks beneath colored lights before the sequence shifts to him flying between city towers, accompanied by Spanish subtitles. Native-video output: 18s · 1080×1944. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/criaturalgoritm/status/2097764585553956895",
    "sourceLabel": "X 原帖 · @criaturalgoritm",
    "author": "Criatura del Algoritmo",
    "publishedAt": "2026-09-09T19:10:26.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097764585553956895.mp4",
    "posterUrl": "/posters/x/2097764585553956895.jpg",
    "duration": 18,
    "aspectRatio": "portrait",
    "resolution": "1080×1944",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "No te hace falta un PC de la NASA para hacer grandes videos. Lo puedes hacer en un PC gama media sin problemas.\n\nPowered by Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 98,
      "snapshotAt": "2026-09-10T00:12:10.694Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097763623422558412",
    "title": "咖啡馆中爆发的念动力",
    "titleEn": "A Telekinetic Burst inside a Cafe",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "咖啡馆里的女子眼睛发出白光，周围出现扭曲气流与飞散物体，镜头随着她的手势和身体转动快速切换。 原生视频规格为 14 秒 · 1848×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman's eyes glow white inside a cafe as distorted air and scattered objects surround her, with quick camera changes following her gestures and turns. Native-video output: 14s · 1848×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NexlowX/status/2097763623422558412",
    "sourceLabel": "X 原帖 · @NexlowX",
    "author": "Nexlow",
    "publishedAt": "2026-09-09T19:06:37.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097763623422558412.mp4",
    "posterUrl": "/posters/x/2097763623422558412.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1848×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "THIS IS ACTUALLY DANGEROUSLY GOOD.\n\nGPT-6 × MiniMax is changing how AI content production works.\n\nOne AI thinks. Another AI creates.\n\nThe idea that one model should do everything is starting to fall apart.\n\nThe real advantage is knowing how to combine different AI models and give each one the right job.\n\nHere’s the workflow 👇\n\nGPT-6 handles the thinking first. We build the concept, structure the idea, define the audience, decide how it should be presented and turn everything into a detailed production plan.\n\nThen we pass that context to MiniMax.\n\nMiniMax M3 can handle up to 1M tokens of context and work across text, images and video. That means it can process huge amounts of information while also helping with code, interfaces and the actual production workflow.\n\nFor video, MiniMax H3 takes things even further. It understands text, images, video and audio and can generate video with sound together.\n\nThen there’s MiniMax Music 3.0, which can take an idea or lyrics and handle the composition, arrangement, vocals and production.\n\nAnd this is the part that matters most.\n\nThe future isn’t about finding one AI that does everything.\n\nIt’s about building a system.\n\nOne AI for thinking. One for execution. One for video. One for music.\n\nConnect them and let each model do what it’s best at.\n\nHumans stop doing every step manually and start orchestrating the whole production process.\n\nGPT-6 × MiniMax is less about two models and more about a completely different way of working with AI.\n\nRemember the approach.\n\nTHINK. DELEGATE. CREATE. CONNECT.\n\nThat’s where the insane speed gains start.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 153,
      "snapshotAt": "2026-09-10T00:12:11.082Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097757059324502087",
    "title": "银发少女与机械伙伴初次接触",
    "titleEn": "A Silver-Haired Girl Meets a Mechanical Companion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发少女在白蓝色机械伙伴旁转身，镜头切到正面特写，结尾出现两位角色的初次接触标题。 原生视频规格为 25 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired girl turns beside a white-and-blue mechanical companion before the camera cuts to a frontal close-up and a first-contact title card. Native-video output: 25s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LiluluLoserLab/status/2097757059324502087",
    "sourceLabel": "X 原帖 · @LiluluLoserLab",
    "author": "リルル@Loser Lab",
    "publishedAt": "2026-09-09T18:40:32.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097757059324502087.mp4",
    "posterUrl": "/posters/x/2097757059324502087.jpg",
    "duration": 25,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のボーカルソング\nコントロール難しい～。\nおまかせで不思議ソング付けてくれたりするんですけど、ちゃんと聞き取れる形に追い込めない。\nでもなんか好き(*'ω'*)",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 120,
      "snapshotAt": "2026-09-10T00:12:10.999Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097756171234935206",
    "title": "机甲迎战海港熔岩巨兽",
    "titleEn": "A Mech Confronts a Lava Beast at the Harbor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "海港城市中，屋顶人物眺望水岸，熔岩纹路巨兽进入城区，机甲与它在楼宇之间以光束交锋。 原生视频规格为 70 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rooftop figure overlooks a harbor city as a lava-marked giant creature enters the streets, where a mech exchanges energy beams with it between the buildings. Native-video output: 70s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ysf__ai/status/2097756171234935206",
    "sourceLabel": "X 原帖 · @ysf__ai",
    "author": "ysf.ai | Synth Engine Studios",
    "publishedAt": "2026-09-09T18:37:00.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097756171234935206.mp4",
    "posterUrl": "/posters/x/2097756171234935206.jpg",
    "duration": 70,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Saving the city \n\nMade with MiniMax H3 on @Hailuo_AI \n\n#MiniMaxH3 #MiniMaxDesign",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 135,
      "snapshotAt": "2026-09-10T00:12:11.018Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097753204708769986",
    "title": "月夜汉堡与动漫少女",
    "titleEn": "A Moonlit Burger with an Anime Girl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫少女举着汉堡靠近镜头，画面切换到木质走廊的俯视镜头，最后呈现月亮、汉堡与日文活动文字。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime girl brings a burger toward the camera, followed by an overhead wooden-corridor view and a final composition featuring the moon, burger, and Japanese promotional text. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aawuup/status/2097753204708769986",
    "sourceLabel": "X 原帖 · @aawuup",
    "author": "Shiso@AI",
    "publishedAt": "2026-09-09T18:25:13.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097753204708769986.mp4",
    "posterUrl": "/posters/x/2097753204708769986.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#いまだけ月見食べ美  #MiniMaxH3\n月見きたー！",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 26,
      "snapshotAt": "2026-09-10T00:12:11.013Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097749601092784472",
    "title": "蓝眼少女的变身登场",
    "titleEn": "A Blue-Eyed Girl's Transformation Entrance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面从蓝色眼睛的特写切向银发少女的侧脸，再拉远展示她在城市背景与发光字母前的完整造型。 原生视频规格为 18 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A close-up of a blue eye transitions to a silver-haired girl's profile before pulling back to reveal her full outfit against a city backdrop and glowing letters. Native-video output: 18s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LiluluLoserLab/status/2097749601092784472",
    "sourceLabel": "X 原帖 · @LiluluLoserLab",
    "author": "リルル@Loser Lab",
    "publishedAt": "2026-09-09T18:10:53.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097749601092784472.mp4",
    "posterUrl": "/posters/x/2097749601092784472.jpg",
    "duration": 18,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MINA Connect from the heart MiniMax H3\n参考プロンプト mayv\n@haruuraeadss\nmayvさんの作品　ドリーミーに触発されてMINA&WILLで作ってみました。 \n\n変身前の衣装は　ほくろ。さんの作品です。\n@tentaclee1011\n\n#AIart",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 2,
      "views": 209,
      "snapshotAt": "2026-09-10T00:12:11.395Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097746917904052662",
    "title": "摩托骑手冲入雷电旋涡",
    "titleEn": "A Motorcyclist Rides into a Lightning Vortex",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣骑手沿霓虹公路高速前行，雷电网格逐渐包围前路，道路尽头裂开成旋涡，摩托车跃向耀眼白光。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rider in black speeds along a neon highway as a lattice of lightning surrounds the road, which opens into a vortex before the motorcycle leaps toward a brilliant white light. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LudovicCreator/status/2097746917904052662",
    "sourceLabel": "X 原帖 · @LudovicCreator",
    "author": "LudovicCreator",
    "publishedAt": "2026-09-09T18:00:14.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097746917904052662.mp4",
    "posterUrl": "/posters/x/2097746917904052662.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The road ends at 12 seconds. He doesn't slow down.\n\n15 seconds, no cuts, generated in a single pass. Three things I've learned making these:\n\nEscalation beats spectacle. The first 6s are just a guy riding fast ,boring on purpose. Every element after that (lightning lattice → torn neon signs → the vortex) is one notch louder than the last. If you open at maximum you have nowhere left to go.\n\nThe camera has to commit. Profile for the setup, rear-view for the chase, dead-center for the final flight. Three positions, no cheating between them. Most AI video drifts because the prompt never picks one.\n\nSilhouette carries identity better than a face. The rider is black-on-black the whole time  nothing to break continuity, and it holds for the full 362 frames.\n\nThe white-out at the end wasn't planned. It's the best frame in it.\n\nWhat's the shot you'd have added?\n\nMade in MiniMax Design @Hailuo_AI with Minimax H3\n1/2",
    "engagement": {
      "replies": 9,
      "reposts": 5,
      "likes": 23,
      "views": 852,
      "snapshotAt": "2026-09-10T00:12:11.402Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2097746306567725382",
    "title": "小鱼穿行水墨苏州",
    "titleEn": "A Little Fish Travels through Ink-Wash Suzhou",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水墨质感的园林、湖水与城市景观交替出现，卡通小鱼穿行其间，登上高处俯瞰苏州，组成一部带字幕的城市短片。 原生视频规格为 249 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Ink-wash gardens, water, and city views alternate as a cartoon fish travels through them and reaches a high overlook, forming a subtitled short film about Suzhou. Native-video output: 249s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Jesse2046/status/2097746306567725382",
    "sourceLabel": "X 原帖 · @Jesse2046",
    "author": "子杰的ZAI工作室",
    "publishedAt": "2026-09-09T17:57:48.000Z",
    "addedAt": "2026-09-10T06:36:56.244Z",
    "mediaUrl": "/media/x-2097746306567725382.mp4",
    "posterUrl": "/posters/x/2097746306567725382.jpg",
    "duration": 249,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "自己手搓完一部4分钟的AI短片，才发现自己真的不懂AIGC，太废物了。\n这是参加苏州华贸中心举办的小型AIGC大赛的参赛作业。\n这部短片花了我3整天时间，用的是MINIMAX  H3。\n我在声音处理这块真是小学生。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 172,
      "snapshotAt": "2026-09-10T00:12:11.390Z"
    },
    "approvedAt": "2026-09-10T06:36:56.244Z",
    "editorialBasis": "Original X post loaded with matching author and timestamp; native video fully decoded and beginning/middle/end frames reviewed."
  },
  {
    "id": "x-2098193558201315654",
    "title": "双车穿越隧道的高速追逐",
    "titleEn": "Two Cars Tear Through a Tunnel Chase",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "白色与橙色跑车在隧道和高架路间并行追逐，低机位近贴车身，连续旋转跟拍放大速度感。 原生视频规格为 50 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "White and orange sports cars race through a tunnel and elevated road while low, body-hugging angles and rotating pursuit shots amplify the speed. Native-video output: 50s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Toshi_nyaruo_AI/status/2098193558201315654",
    "sourceLabel": "X 原帖 · @Toshi_nyaruo_AI",
    "author": "Toshi🎬AI動画💫MV制作🎧",
    "publishedAt": "2026-09-10T23:35:01.000Z",
    "addedAt": "2026-09-11T00:16:00.981Z",
    "mediaUrl": "/media/x-2098193558201315654.mp4",
    "posterUrl": "/posters/x/2098193558201315654.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "［PR］GPT-6 Astra & MiniMaX H3　Blender参照でカーチェイスシーンを作成🏎️\n\nプリビズをAstraでBlenderにて作成。それをH3参照で動画を作成しました。\n\n概ねプリビズ通りになっていますが、カメラが回転するところが参照が弱くなっています。\n\nBlenderのプリビズは動きが穏やかなシーンに向いているようです・\n\n@Hailuo_AI  #MiniMaxDesign",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 8,
      "views": 88,
      "snapshotAt": "2026-09-11T00:16:00.981Z"
    },
    "approvedAt": "2026-09-11T00:16:00.981Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098193046718619971",
    "title": "霓虹赛道上的极速轮滑",
    "titleEn": "Full-Speed Roller Skating on a Neon Track",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色剪影轮滑者沿高饱和弧形赛道持续加速，贴地跟拍、俯视切换与大幅色块共同强化冲刺节奏。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black-silhouette skater accelerates across a saturated curved track as ground-level tracking, overhead cuts, and bold color fields sustain the rush. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MiniMax H3を使用して、15秒・2Kのスタイリッシュなローラースケート動画を制作する。最優先するのは圧倒的なスピード感。開始直後から高速で滑走し、最後まで減速せず、勢いを維持する。\n主人公は、スポーツウェアとヘルメットを着用した成人スケーター一人。人物と衣服を黒一色のシャープなシルエットとして描く。両足には、片足4輪のクワッド型ローラースケートを履く。\n舞台は、鮮やかなシアン、マゼンタ、コーラル、イエローで構成された、広く滑らかな架空の専用スケートコース。長い直線と大きなカーブを組み合わせ、前方へ伸びる床のライン、連続する柱、幾何学的なアーチによって強い遠近感を作る。\n\n【0～4秒：冒頭から全速力】\n最初のフレームから、主人公はすでに高速で滑走している。\n上半身を低く倒し、膝を曲げ、左右の足で力強く連続して路面を押す。腕を大きく前後へ振り、蹴り出しと連動させる。足をそろえて流すだけでなく、自ら加速している動きを明確に見せる。\nカメラは車輪の高さから主人公の横を高速で並走する。手前の路面、コースライン、柱が画面を勢いよく通過する。主人公とスケートは鮮明に保ち、近景ほど強く流れるモーションブラーで速度を表現する。\n\n【4～8秒：高速カービング】\n主人公は速度を維持したまま、大きなカーブへ進入する。全身を内側へ傾け、力強いクロスオーバーで曲がりながら加速する。\nカメラは斜め前方で後退しながら追跡し、主人公が迫ってくる構図を作る。次の強いビートで斜め上からの追従ショットへ切り替え、長いコースを高速で進む軌道を見せる。\n背景の太いラインと長方形が後方へ飛び去り、カーブに合わせて画面の色面がシアンからマゼンタへ瞬時に切り替わる。コースの幅と形状は安定させ、滑走ルートを明確にする。\n\n【8～12秒：直線でさらに加速】\nカーブを抜けると長い直線コースが開ける。主人公は姿勢をさらに低くし、左右の蹴り出しをテンポよく繰り返して加速する。\nカメラは真横の全身ショット、斜め後方からの高速追従、地面すれすれの車輪ショットをビートに合わせて切り替える。アングルが変わっても前進する勢いを途切れさせない。\nコース脇の柱と頭上のアーチが短い間隔で次々と通過し、路面のラインが画面手前へ鋭く流れる。背景だけを動かすのではなく、人物が実際にコース上を高速で移動する様子を描く。\n\n【12～15秒：カメラを追い越して終了】\n主人公は最後の直線を全速力で駆け抜ける。\nカメラは斜め前方から並走し、主人公が追いついてくる動きを捉える。最後の強いビートで主人公がカメラを追い越し、画面手前から奥へ勢いよく滑っていく。\nカメラは滑らかに向きを変えて後ろ姿を追う。主人公は蹴り出しを続け、前方へ伸びる鮮やかなコースを高速で進んだまま映像を終了する。最後まで減速、停止、決めポーズを入れない。\n\n【モーショングラフィックと音楽】\n背景の長方形、三角形、グリッド、太いラインを、進行方向と反対側へ高速でスライドさせる。蹴り出すたびに短い光のラインが路面を走り、カーブでは図形が滑走軌道に沿って流れる。\n音楽は速いテンポのドラムンベースまたはエレクトロニック・ミュージック。力強いベース、細かなパーカッション、車輪の走行音、風切り音を組み合わせ、蹴り出しとカメラの切り替えをビートへ同期させる。\nスローモーション、ダンス、足踏み、スピン、後ろ向き滑走は入れない。全編を高速の前進滑走と大きなカーブで構成する。\n同じ黒いスケーターと同じ一組のクワッド型ローラースケートを維持する。車輪は路面との接触と移動速度に合わせて回転する。文字やロゴは表示しない。\nシャープな人物、激しく流れる近景、強い遠近感、力強い蹴り出し、途切れない高速追従によって、風圧まで感じるような、疾走感を最優先した15秒動画に仕上げる。",
    "promptSourceUrl": "https://x.com/Naonekozamurai/status/2098193046718619971",
    "sourceUrl": "https://x.com/Naonekozamurai/status/2098193046718619971",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-09-10T23:32:59.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098193046718619971.mp4",
    "posterUrl": "/posters/x/2098193046718619971.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Codex (Astra) x MiniMax H3\nスキル化することで、クオリティに再現性ある動画がいくつも作れてしまう\n@Hailuo_AI \n\nスピード感あって良いな\n\nプロンプトは自由にどうぞ↓\n「MiniMax H3を使用して、15秒・2Kのスタイリッシュなローラースケート動画を制作する。最優先するのは圧倒的なスピード感。開始直後から高速で滑走し、最後まで減速せず、勢いを維持する。\n主人公は、スポーツウェアとヘルメットを着用した成人スケーター一人。人物と衣服を黒一色のシャープなシルエットとして描く。両足には、片足4輪のクワッド型ローラースケートを履く。\n舞台は、鮮やかなシアン、マゼンタ、コーラル、イエローで構成された、広く滑らかな架空の専用スケートコース。長い直線と大きなカーブを組み合わせ、前方へ伸びる床のライン、連続する柱、幾何学的なアーチによって強い遠近感を作る。\n\n【0～4秒：冒頭から全速力】\n最初のフレームから、主人公はすでに高速で滑走している。\n上半身を低く倒し、膝を曲げ、左右の足で力強く連続して路面を押す。腕を大きく前後へ振り、蹴り出しと連動させる。足をそろえて流すだけでなく、自ら加速している動きを明確に見せる。\nカメラは車輪の高さから主人公の横を高速で並走する。手前の路面、コースライン、柱が画面を勢いよく通過する。主人公とスケートは鮮明に保ち、近景ほど強く流れるモーションブラーで速度を表現する。\n\n【4～8秒：高速カービング】\n主人公は速度を維持したまま、大きなカーブへ進入する。全身を内側へ傾け、力強いクロスオーバーで曲がりながら加速する。\nカメラは斜め前方で後退しながら追跡し、主人公が迫ってくる構図を作る。次の強いビートで斜め上からの追従ショットへ切り替え、長いコースを高速で進む軌道を見せる。\n背景の太いラインと長方形が後方へ飛び去り、カーブに合わせて画面の色面がシアンからマゼンタへ瞬時に切り替わる。コースの幅と形状は安定させ、滑走ルートを明確にする。\n\n【8～12秒：直線でさらに加速】\nカーブを抜けると長い直線コースが開ける。主人公は姿勢をさらに低くし、左右の蹴り出しをテンポよく繰り返して加速する。\nカメラは真横の全身ショット、斜め後方からの高速追従、地面すれすれの車輪ショットをビートに合わせて切り替える。アングルが変わっても前進する勢いを途切れさせない。\nコース脇の柱と頭上のアーチが短い間隔で次々と通過し、路面のラインが画面手前へ鋭く流れる。背景だけを動かすのではなく、人物が実際にコース上を高速で移動する様子を描く。\n\n【12～15秒：カメラを追い越して終了】\n主人公は最後の直線を全速力で駆け抜ける。\nカメラは斜め前方から並走し、主人公が追いついてくる動きを捉える。最後の強いビートで主人公がカメラを追い越し、画面手前から奥へ勢いよく滑っていく。\nカメラは滑らかに向きを変えて後ろ姿を追う。主人公は蹴り出しを続け、前方へ伸びる鮮やかなコースを高速で進んだまま映像を終了する。最後まで減速、停止、決めポーズを入れない。\n\n【モーショングラフィックと音楽】\n背景の長方形、三角形、グリッド、太いラインを、進行方向と反対側へ高速でスライドさせる。蹴り出すたびに短い光のラインが路面を走り、カーブでは図形が滑走軌道に沿って流れる。\n音楽は速いテンポのドラムンベースまたはエレクトロニック・ミュージック。力強いベース、細かなパーカッション、車輪の走行音、風切り音を組み合わせ、蹴り出しとカメラの切り替えをビートへ同期させる。\nスローモーション、ダンス、足踏み、スピン、後ろ向き滑走は入れない。全編を高速の前進滑走と大きなカーブで構成する。\n同じ黒いスケーターと同じ一組のクワッド型ローラースケートを維持する。車輪は路面との接触と移動速度に合わせて回転する。文字やロゴは表示しない。\nシャープな人物、激しく流れる近景、強い遠近感、力強い蹴り出し、途切れない高速追従によって、風圧まで感じるような、疾走感を最優先した15秒動画に仕上げる。\n」",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 311,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098190815856382107",
    "title": "专注计时应用的生活化短片",
    "titleEn": "A Lifestyle Spot for a Focus Timer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "手机上的专注计时界面从近景操作切换到桌面使用场景，最终以品牌标识收束成简洁应用宣传片。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Close-up timer interactions transition into a desk-use scene before the app mark closes out a concise product spot. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hirococoro/status/2098190815856382107",
    "sourceLabel": "X 原帖 · @hirococoro",
    "author": "内海裕子(hiroko utsumi)",
    "publishedAt": "2026-09-10T23:24:07.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098190815856382107.mp4",
    "posterUrl": "/posters/x/2098190815856382107.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "／\n🦄アプリ紹介動画を作ることに慣れてみよう大作戦｜07\n＼\n#MiniMaxH3 編｜Shipatonイベント参加で協賛されている@aiand_ から頂いた特典を使い、雑いプロンプトでポン出し！複数画像放り込んだらロゴ破綻はありつつも良い感じに！うおー。\nhttps://untiltimer.com/\n#Shipaton 記録\n#BuildInPublic",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 46,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098174575700017192",
    "title": "达达拼贴中的动态排版实验",
    "titleEn": "A Kinetic Type Experiment in Dada Collage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "黑白报纸、手势、眼睛与数字被快速拼贴，红色箭头和旋转排版穿插其中，最后落到日文标题卡。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Newspaper textures, hands, eyes, and numerals collide in rapid monochrome collage, punctuated by red arrows and rotating type before a final title card. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2098174575700017192",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-09-10T22:19:35.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098174575700017192.mp4",
    "posterUrl": "/posters/x/2098174575700017192.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「意味なんてあとでいい」\nダダイズム×キネティックタイポグラフィ×コラージュ\n\n#OiiOii のMinimaxH3 Max Tauboで作成\n開始に絵コンテ入れたら問題なく参照されます（開始わずかに表示）\n\nプロンプトはリプ欄に👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 145,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098163639760867393",
    "title": "低多边形角色的城市冒险序章",
    "titleEn": "A Low-Poly Character Opens an Urban Adventure",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "鸭头角色在城市街道登场，几何光效围绕身体展开，快节奏镜头把角色造型与冒险氛围压进十五秒。 原生视频规格为 15 秒 · 1620×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A duck-headed character appears on city streets as geometric light effects build around the body, compressing a character reveal and adventure mood into 15 seconds. Native-video output: 15s · 1620×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/robazem/status/2098163639760867393",
    "sourceLabel": "X 原帖 · @robazem",
    "author": "Rob ✌︎",
    "publishedAt": "2026-09-10T21:36:08.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098163639760867393.mp4",
    "posterUrl": "/posters/x/2098163639760867393.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1620×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Fast paced, 2D/3D, low poly character visual style plus a short intro sequence of a new adventure.\n\nSummarizing the video, I don’t have plans for this guy just wanted to make one from scratch.\n\nInto the archives \n\nTools - MiniMax H3 and ChatGPT",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 111,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098161974081405017",
    "title": "夕阳山野中的白发歌者",
    "titleEn": "A White-Haired Singer at Sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "白发歌者在金色山野中面对镜头演唱，光晕与彩色能量从身后扩散，口型和表情保持连续。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired singer performs directly to camera in a golden landscape while halos and colored energy bloom behind her, maintaining continuous lip movement and expression. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/junyait/status/2098161974081405017",
    "sourceLabel": "X 原帖 · @junyait",
    "author": "海澤純 AI CREATIVE",
    "publishedAt": "2026-09-10T21:29:31.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098161974081405017.mp4",
    "posterUrl": "/posters/x/2098161974081405017.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "TapNow MiniMax H3のリップシンクが凄い件\n\n画像と音源から生成、加工なし。\nエフェクト含めて一発生成。\n生成したまんまです。\n\nさあみなさん、TapNowを使ってみませんか？\n\n詳しくはこちら\n\n@TapNow_AI \n@TapNow_JP",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 18,
      "views": 196,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098159981132484995",
    "title": "橱窗街头的夸张购物角色",
    "titleEn": "An Exaggerated Shopper on a Boutique Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卡通化购物者拎着彩袋穿行精品街区，路人反应与橱窗倒影串起一段带戏剧表情的城市小品。 原生视频规格为 15 秒 · 2496×1408。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A caricatured shopper carries bright bags through a boutique street, with passerby reactions and window reflections shaping a theatrical urban vignette. Native-video output: 15s · 2496×1408. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxCasu/status/2098159981132484995",
    "sourceLabel": "X 原帖 · @MaxCasu",
    "author": "max casu",
    "publishedAt": "2026-09-10T21:21:36.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098159981132484995.mp4",
    "posterUrl": "/posters/x/2098159981132484995.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2496×1408",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Shopping day...\nCreated with MiniMax H3 model on @ViggleAI \n@Hailuo_AI",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 87,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098159303739760904",
    "title": "道场里的荷兰语独白",
    "titleEn": "A Dutch Monologue Inside a Dojo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿练功服的男子在木质道场里朝镜头说话，半身构图与稳定口型集中测试非英语对白表现。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in training clothes speaks to camera inside a wooden dojo, using steady medium shots and lip motion to test non-English dialogue. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JeroenvanH/status/2098159303739760904",
    "sourceLabel": "X 原帖 · @JeroenvanH",
    "author": "Honest broker / 誠實經紀人",
    "publishedAt": "2026-09-10T21:18:54.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098159303739760904.mp4",
    "posterUrl": "/posters/x/2098159303739760904.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Iemand heeft ontdekt dat je met MiniMax H3 de spraak veel beter dan eerst gedacht kunt aansturen. Resolutie/kwaliteit is matig, maar het ging me meer om te kijken of het ook in het Nederlands zou werken. Ja dus. Prompt 👇",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 177,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098158326504919082",
    "title": "黑色舞台上的制服角色亮相",
    "titleEn": "A Uniformed Character Steps Onto a Dark Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "制服角色从面部与肩章特写过渡到暗场全身亮相，聚光与克制动作塑造舞台式登场。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A uniformed character moves from facial and insignia close-ups to a full-body reveal on a dark stage, framed by spotlights and controlled movement. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/izesaki/status/2098158326504919082",
    "sourceLabel": "X 原帖 · @izesaki",
    "author": "! Ze",
    "publishedAt": "2026-09-10T21:15:01.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098158326504919082.mp4",
    "posterUrl": "/posters/x/2098158326504919082.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "my first experiment on @openart_ai  with MiniMax H3\n\n#openart",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 58,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098157327773786336",
    "title": "少女与灵狐的森林相遇",
    "titleEn": "A Girl Meets a Spirit Fox in the Forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卷发少女跟随发光灵狐深入苔藓森林，从试探同行到树下依偎，完整铺陈一段童话式关系变化。 原生视频规格为 60 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A curly-haired girl follows a glowing spirit fox through a mossy forest, moving from cautious companionship to a quiet embrace beneath the trees. Native-video output: 60s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/godblesstoto/status/2098157327773786336",
    "sourceLabel": "X 原帖 · @godblesstoto",
    "author": "🙏🌧🌍",
    "publishedAt": "2026-09-10T21:11:03.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098157327773786336.mp4",
    "posterUrl": "/posters/x/2098157327773786336.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with Minimax H3. \nTo be continued...",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 30,
      "views": 1432,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098144465697288672",
    "title": "向日葵同时转向镜头",
    "titleEn": "The Sunflowers Turn Toward the Camera",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "固定低机位凝视日落花田，前景多株向日葵不再追随太阳，而是同步把花盘转向镜头。 原生视频规格为 15 秒 · 4096×2340。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A locked, flower-height camera watches a sunset field as foreground sunflowers stop following the sun and turn their heads toward the viewer together. Native-video output: 15s · 4096×2340. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the heads turn, all together, away from the sun and toward the camera",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2098144465697288672",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2098144465697288672",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-10T20:19:56.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098144465697288672.mp4",
    "posterUrl": "/posters/x/2098144465697288672.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "4096×2340",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3. The sunflowers stopped watching the sun.\n\nPrompt: \"the heads turn, all together, away from the sun and toward the camera\"\n\nDuration: 15s. Camera: locked, flower-head height.\n\n\"The heads always move together\" is a crowd rule. Crowds in video models are a hundred individual decisions unless you give them one.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 9,
      "views": 183,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098143464919588954",
    "title": "沙特国庆主题的竖屏光效广告",
    "titleEn": "A Vertical Light-Graphic Spot for Saudi National Day",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深色竖屏画面以线描器物开场，粒子爆发后切换为绿色圆环与棕榈图形，完成节庆视觉广告。 原生视频规格为 30 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark vertical frame opens with line-drawn objects, bursts into particles, and resolves into green rings and a palm motif for a celebratory visual spot. Native-video output: 30s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Designer_Samco/status/2098143464919588954",
    "sourceLabel": "X 原帖 · @Designer_Samco",
    "author": "سامكو Samco Creative Arts",
    "publishedAt": "2026-09-10T20:15:58.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098143464919588954.mp4",
    "posterUrl": "/posters/x/2098143464919588954.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "اليوم الوطني السعودي 96\n\nالمحتوى: فيديو إعلاني\nCTA: “اعطوني تحديات أصعب.”\nلا تنسوا 👍 الإعجاب، 💬 التعليق، و 🔁 المشاركة إذا وجدت هذا المحتوى مفيد — وشاركني رأيكم في التعليقات ✍️\n\n#PolloAI #PolloAnimation #MinimaxH3  @itsPolloAI \n#PolloAi \n@itsPolloAI @polloai_creator",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 17,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098141107875901470",
    "title": "实验室控制台前的怪诞独白",
    "titleEn": "A Grotesque Monologue at a Lab Console",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "怪诞角色被固定在工业座椅前操作按钮，面部从怒意转为夸张笑容，红色控制光强化戏剧变化。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A grotesque figure works controls from an industrial chair, shifting from anger to an exaggerated grin as red console light heightens the dramatic change. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nonameoasis/status/2098141107875901470",
    "sourceLabel": "X 原帖 · @nonameoasis",
    "author": "nonameoasis",
    "publishedAt": "2026-09-10T20:06:36.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098141107875901470.mp4",
    "posterUrl": "/posters/x/2098141107875901470.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I’ve never been a good guy. 😂\n\n#MiniMaxH3 in @weryai",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 9,
      "views": 77,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098131405431595161",
    "title": "水面涟漪中的狐形角色",
    "titleEn": "A Fox-Like Character Emerges Through Ripples",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "手绘狐形角色从旋转水纹的近景过渡到湖畔全身镜头，倒影、树影与轻微表情变化保持统一画风。 原生视频规格为 10 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hand-drawn fox-like character moves from a close-up framed by spinning ripples to a full lakeside view, keeping reflections, trees, and subtle expressions stylistically coherent. Native-video output: 10s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shai_kurari/status/2098131405431595161",
    "sourceLabel": "X 原帖 · @shai_kurari",
    "author": "Shai_Kurari",
    "publishedAt": "2026-09-10T19:28:03.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098131405431595161.mp4",
    "posterUrl": "/posters/x/2098131405431595161.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "creatures",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "For the curious~ Applied Minimax H3 to it. Seems it does read!🧡",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 36,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098131110949745032",
    "title": "黑红空间里的长篇歌词视觉",
    "titleEn": "A Long-Form Lyric Visual in Black and Red",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金属椅、镜面地面和红色激光组成极简舞台，英文歌词随音乐逐句进入，形成三分钟以上的连续视觉化。 原生视频规格为 193 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Metal chairs, mirrored floors, and red laser lines form a minimal stage where English lyrics arrive line by line across a continuous three-minute visual. Native-video output: 193s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/atomtanstudio/status/2098131110949745032",
    "sourceLabel": "X 原帖 · @atomtanstudio",
    "author": "Rich · Atom Tan Studio",
    "publishedAt": "2026-09-10T19:26:52.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098131110949745032.mp4",
    "posterUrl": "/posters/x/2098131110949745032.jpg",
    "duration": 193,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "commercial"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "YuE2 + Minimax H3 is a fantastic music video pipeline. This is an output from my new app called Sound / Vision. This is a simple lyric video. I am working on equalizer and full music video with lip syncing pipelines next.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 211,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098127738850976204",
    "title": "金色秋日的手绘散步短片",
    "titleEn": "An Illustrated Walk Through Golden Autumn",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手绘小路、金黄树林与落日田野依次出现，短句排版把三个秋日画面连接成温柔的季节问候。 原生视频规格为 15 秒 · 2688×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustrated path, golden trees, and a sunset field appear in sequence, linked by short handwritten phrases into a gentle seasonal greeting. Native-video output: 15s · 2688×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/airina_xyz/status/2098127738850976204",
    "sourceLabel": "X 原帖 · @airina_xyz",
    "author": "AIrina (Virtual Media Net)",
    "publishedAt": "2026-09-10T19:13:28.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098127738850976204.mp4",
    "posterUrl": "/posters/x/2098127738850976204.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2688×1536",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Golden-hour walks and warm light everywhere. Welcoming the magic of early autumn evenings. 🌅🍁\n\nCreated with @Hailuo_AI Minimax H3 in @PixVerse #PixverseCPP @PixVerseCreator\n\nPrompt in comments 👇",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 4,
      "views": 133,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098121037498216908",
    "title": "道场镜前的克制对白",
    "titleEn": "A Restrained Address Before a Dojo Mirror",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿黑色练功服的男子在纸门与镜面前说话，镜头由侧面手势推进到正面表情，保持稳定人物连续性。 原生视频规格为 19 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in black training clothes speaks before paper screens and a mirror, moving from side gestures to a frontal expression while maintaining character continuity. Native-video output: 19s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/topmass/status/2098121037498216908",
    "sourceLabel": "X 原帖 · @topmass",
    "author": "topmass",
    "publishedAt": "2026-09-10T18:46:51.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098121037498216908.mp4",
    "posterUrl": "/posters/x/2098121037498216908.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I am doing important work with minimax h3 today.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 110,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098106397754110227",
    "title": "咖啡店里的品牌双胞胎梗",
    "titleEn": "A Brand-Twin Joke Inside a Café",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏咖啡店场景把绿色猫头鹰玩偶与穿绿色衣服的男子并置，字幕推动一段围绕品牌相似感的短梗。 原生视频规格为 7 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical café scene pairs a green owl mascot with a man in matching clothes, using captions to land a short joke about two brands looking like twins. Native-video output: 7s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/0xstorypreneur/status/2098106397754110227",
    "sourceLabel": "X 原帖 · @0xstorypreneur",
    "author": "Rahul ✨",
    "publishedAt": "2026-09-10T17:48:40.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098106397754110227.mp4",
    "posterUrl": "/posters/x/2098106397754110227.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "commercial"
    ],
    "scenes": [
      "product",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Man @Apple duo launch \nGot this two going @duolingo x @SamsungMobileUS \n\nLet's go Twins 😁 \n\nMade with Minimax H3 using @fal",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 151,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098098249127399520",
    "title": "朋友打破透明屏障的动画片段",
    "titleEn": "Friends Break Through a Transparent Barrier",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名角色围绕被困在发光透明屏障中的同伴互动，能量罩逐渐消散，三人最终重新靠近。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two characters interact with a friend trapped inside a glowing transparent barrier; the energy field dissolves and the trio comes back together. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C12592947592/status/2098098249127399520",
    "sourceLabel": "X 原帖 · @C12592947592",
    "author": "supercub3",
    "publishedAt": "2026-09-10T17:16:18.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098098249127399520.mp4",
    "posterUrl": "/posters/x/2098098249127399520.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MinimaxH3ローカル\nももとユリカに説得されてバリアを解くリアちゃん(新キャラ)",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 13,
      "views": 248,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098093176372560351",
    "title": "月夜下的动画人声片段",
    "titleEn": "An Animated Vocal Clip Under Moonlight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "银发动画角色在星空与城市夜景间演唱，镜头切换半身、双人和月下场景，重点展示持续人声与口型。 原生视频规格为 30 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired animated character sings across starry and city-night settings, alternating medium, paired, and moonlit shots to emphasize sustained vocals and lip movement. Native-video output: 30s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LiluluLoserLab/status/2098093176372560351",
    "sourceLabel": "X 原帖 · @LiluluLoserLab",
    "author": "リルル@Loser Lab",
    "publishedAt": "2026-09-10T16:56:08.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098093176372560351.mp4",
    "posterUrl": "/posters/x/2098093176372560351.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3　ボーカルクリップ検証\nチャッピーと研究しながらだいぶ聴き取りやすくなってきた～\n最もシンプルな構成のImage to Videoでも、上手く行けば映像と歌声の両立は可能。\nただ、まだ歌わせはじめ、歌わせ終わりの正確なコントロールができてないですね。\n今後の課題です。(*'ω'*)\n#AIart",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 151,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098092554093039735",
    "title": "黑白双人功夫音乐短片",
    "titleEn": "A Black-and-White Kung-Fu Music Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名男子以黑白对照服装在竖屏舞台上表演，圆形排版、定格动作与节奏剪辑构成一分钟音乐片段。 原生视频规格为 61 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two men perform in contrasting black-and-white outfits on a vertical stage, with circular typography, held poses, and rhythmic edits forming a one-minute music piece. Native-video output: 61s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/holycrabAI/status/2098092554093039735",
    "sourceLabel": "X 原帖 · @holycrabAI",
    "author": "HolyCrab Seedance2.5",
    "publishedAt": "2026-09-10T16:53:40.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098092554093039735.mp4",
    "posterUrl": "/posters/x/2098092554093039735.jpg",
    "duration": 61,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "commercial",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "不对！？minimax H3很适合做MV！🥹🥹🥹\n\n#minimaxH3 #中国人能飞 #揽佬 #八方来财",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 49,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098092247959113814",
    "title": "同一人物的风格转换对比",
    "titleEn": "A Style-Shift Comparison on One Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一女性形象在写实人像、眼部特写与霓虹城市造型之间切换，用并列版本测试微调画质与风格保持。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same female figure shifts between a realistic portrait, an eye close-up, and a neon-city look, comparing finetuned image quality and style retention. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/atomtanstudio/status/2098092247959113814",
    "sourceLabel": "X 原帖 · @atomtanstudio",
    "author": "Rich · Atom Tan Studio",
    "publishedAt": "2026-09-10T16:52:27.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098092247959113814.mp4",
    "posterUrl": "/posters/x/2098092247959113814.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ignoring the weird aspect stretching, which do you think is a higher quality video, left or right? I'm testing a new Minimax H3 finetune.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 178,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098091688325468482",
    "title": "霓虹冷冻舱走廊的角色登场",
    "titleEn": "A Character Enters a Neon Cryo Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蒙面角色穿过蓝红霓虹工业走廊，雾气、背光和宽幅镜头逐步扩大空间，形成科幻游戏式登场段落。 原生视频规格为 41 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A masked figure crosses a blue-and-red neon industrial corridor as fog, backlight, and widening shots build a sci-fi game-style entrance. Native-video output: 41s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/flutterwhat/status/2098091688325468482",
    "sourceLabel": "X 原帖 · @flutterwhat",
    "author": "Flutterwhat",
    "publishedAt": "2026-09-10T16:50:13.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098091688325468482.mp4",
    "posterUrl": "/posters/x/2098091688325468482.jpg",
    "duration": 41,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "sci-fi",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"It's cryo, you gotta have eyes on the back of your head\"\n\nFeat. Masculine Vandal from the hit game \"SLOPathon\" \n\nAudio/Character/Environment Copyright BUNGIE | Minimax H3 via SlopCannon @boneGPT. #MarathonArt",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 264,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098080421217239416",
    "title": "粉发角色的竖屏表情动作",
    "titleEn": "A Pink-Haired Character Gestures in Portrait Frame",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "粉发角色在纯色背景中完成眨眼、手势与表情变化，服装饰品和脸部特征在竖屏近景中保持稳定。 原生视频规格为 15 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired character blinks, gestures, and changes expression against a plain background while costume details and facial identity remain stable in a vertical close-up. Native-video output: 15s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fnoji/status/2098080421217239416",
    "sourceLabel": "X 原帖 · @fnoji",
    "author": "えふのじ",
    "publishedAt": "2026-09-10T16:05:27.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098080421217239416.mp4",
    "posterUrl": "/posters/x/2098080421217239416.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3でリファレンス作り直してh3_Singularity試行（Aeコンポしてる）\nref2vaのお陰かリファレンス追従が楽だった気がする\n\n前は純正FL2VAでやると追従が甘くてブチギレシード探しパターンが結構あった",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 63,
      "views": 3234,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098077985849102699",
    "title": "雨夜街头的害羞眼镜女孩",
    "titleEn": "A Shy Glasses-Wearing Girl on a Rainy Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿发眼镜女孩站在霓虹雨夜街头，从低头拘谨到抬眼微笑，湿地反光和近景运镜强化情绪转折。 原生视频规格为 10 秒 · 512×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A green-haired girl with glasses stands on a neon-lit rainy street, shifting from a shy downward gaze to a small smile as wet reflections and close framing carry the emotion. Native-video output: 10s · 512×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YamataZen/status/2098077985849102699",
    "sourceLabel": "X 原帖 · @YamataZen",
    "author": "Yamata",
    "publishedAt": "2026-09-10T15:55:46.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098077985849102699.mp4",
    "posterUrl": "/posters/x/2098077985849102699.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "512×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Shy meganekko\n\n#aivideo #MiniMaxH3 #OC",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 6,
      "views": 210,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098077087852793971",
    "title": "绿色法阵中的奇幻角色",
    "titleEn": "A Fantasy Character Inside a Green Magic Circle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "绿色服装角色在发光几何舞台上抬手施法，圆环与线路从身后展开，保持人物造型和空间中心。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A green-clad character raises her hands on a glowing geometric stage as rings and circuitry unfold behind her, preserving the character design and centered composition. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/raiha_tec/status/2098077087852793971",
    "sourceLabel": "X 原帖 · @raiha_tec",
    "author": "raiha",
    "publishedAt": "2026-09-10T15:52:12.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098077087852793971.mp4",
    "posterUrl": "/posters/x/2098077087852793971.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ChatGPTで画像生成して、MiniMax H3で動画作成ごっこ\nChatGPTがこんなにきれいにかけるのもすごいんだけど、H3が画像１枚でここまで作れるほうがびっくりする",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 321,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098073630139617340",
    "title": "角色环绕转身与道具替换",
    "titleEn": "A Character Rotation Driven by Four References",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "金发冒险者在黑色背景中完成连续转身和道具替换，法杖、背包与服装轮廓在多个角度保持一致。 原生视频规格为 15 秒 · 1344×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde adventurer rotates continuously against black while props change, keeping the staff, backpack, and costume silhouette coherent across angles. Native-video output: 15s · 1344×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SmallKino_/status/2098073630139617340",
    "sourceLabel": "X 原帖 · @SmallKino_",
    "author": "SmallKino",
    "publishedAt": "2026-09-10T15:38:28.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098073630139617340.mp4",
    "posterUrl": "/posters/x/2098073630139617340.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1344×1280",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Rotating Marcille with some replaced props.\n(4 input images)\n\n#MiniMaxH3 #ダンジョン飯 #dungeonmeshi",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 150,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098072835902013529",
    "title": "手绘外星头像的表情变形",
    "titleEn": "A Hand-Drawn Alien Portrait Morphs Its Expression",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长脸外星角色在颗粒质感背景前由冷漠转为夸张咧嘴，眼睛、嘴部与头部比例连续变形。 原生视频规格为 8 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-faced alien portrait shifts from a blank stare into an exaggerated grin against a grainy backdrop, morphing the eyes, mouth, and head proportions continuously. Native-video output: 8s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RoyalKongz/status/2098072835902013529",
    "sourceLabel": "X 原帖 · @RoyalKongz",
    "author": "RoyalKongZ (Laurent)",
    "publishedAt": "2026-09-10T15:35:19.000Z",
    "addedAt": "2026-09-11T00:16:00.982Z",
    "mediaUrl": "/media/x-2098072835902013529.mp4",
    "posterUrl": "/posters/x/2098072835902013529.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "creatures",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "QT your \"Alien Art\"🔊!\n(Minimax H3 via Pixverse @PixVerse #pixversecpp)",
    "engagement": {
      "replies": 7,
      "reposts": 0,
      "likes": 21,
      "views": 274,
      "snapshotAt": "2026-09-11T00:16:00.982Z"
    },
    "approvedAt": "2026-09-11T00:16:00.982Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096716139065954350",
    "title": "至上主义几何中的日文动态排版",
    "titleEn": "Japanese Kinetic Type Inside Suprematist Geometry",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "白底上的黑红几何块与日文词组先整齐排列，随后碎裂旋转成粗野拼贴，最后落到一句\"正常に壊れる\"收束。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Black-and-red geometric blocks and Japanese words start neatly aligned on white, shatter into a brutalist collage, and settle on a single closing phrase. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2096716139065954350",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-09-06T21:44:17.000Z",
    "addedAt": "2026-09-11T14:46:51.957Z",
    "mediaUrl": "/media/x-2096716139065954350.mp4",
    "posterUrl": "/posters/x/2096716139065954350.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "シュプレマティスム × グリッチアート × ブルータリズム × 抽象モーショングラフィック × 日本語キネティックタイポグラフィ\n\n#OiiOii にMinimaxH3 Max Turboが実装されていたので検証！T2Vでも日本語文字化けはH3と同じくらいな気がする。生成時間は１０秒くらい。早いとストレスなく作れるので快適ですね！",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 245,
      "snapshotAt": "2026-09-11T14:46:51.957Z"
    },
    "approvedAt": "2026-09-11T14:46:51.957Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096630426127217144",
    "title": "夜色天台上的金发长裙少女",
    "titleEn": "A Blonde Woman in Black on a Night Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金色长发、黑色长裙的动画角色站在城市夜景天台边，镜头从正面近景缓慢环绕到俯视高楼的背影，人物与场景保持一致。 原生视频规格为 12 秒 · 832×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime woman with long blonde hair and a black dress stands at a rooftop edge above city lights as the camera circles from a frontal close-up to a high view of her back, keeping character and scene consistent. Native-video output: 12s · 832×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ascendant_Stoic/status/2096630426127217144",
    "sourceLabel": "X 原帖 · @Ascendant_Stoic",
    "author": "Ascendant Stoic",
    "publishedAt": "2026-09-06T16:03:41.000Z",
    "addedAt": "2026-09-11T14:46:51.957Z",
    "mediaUrl": "/media/x-2096630426127217144.mp4",
    "posterUrl": "/posters/x/2096630426127217144.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "832×1248",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "More attempts at animation using Minimax H3, the consistency of the scene/character is really fantastic. \n\nArt from CivitAI.\nhttps://civitai.red/images/137716462",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 132,
      "snapshotAt": "2026-09-11T14:46:51.957Z"
    },
    "approvedAt": "2026-09-11T14:46:51.957Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096630269306716621",
    "title": "驾驶舱里眨眼挥手的蓝发少女",
    "titleEn": "A Blue-Haired Girl Winks and Waves from a Cockpit",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝发动画少女坐在星空舷窗前的驾驶座上，短短两秒内完成抬手、眨眼与挥手，是一段本地显卡上的生成测试。 原生视频规格为 2 秒 · 512×512。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired anime girl seated in a starship cockpit raises a hand, winks, and waves within two seconds in a short local-GPU generation test. Native-video output: 2s · 512×512. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/moneykoikoi/status/2096630269306716621",
    "sourceLabel": "X 原帖 · @moneykoikoi",
    "author": "マネーコイコイ@FX相互ゲーム小説もふもふ系",
    "publishedAt": "2026-09-06T16:03:04.000Z",
    "addedAt": "2026-09-11T14:46:51.957Z",
    "mediaUrl": "/media/x-2096630269306716621.mp4",
    "posterUrl": "/posters/x/2096630269306716621.jpg",
    "duration": 2,
    "aspectRatio": "square",
    "resolution": "512×512",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MinimaxH3で動画生成テスト。ＲＴＸ３０６０、Ｖｒａｍ１２ＧＢで最適化にえらい苦労した・・・ｗ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 149,
      "snapshotAt": "2026-09-11T14:46:51.957Z"
    },
    "approvedAt": "2026-09-11T14:46:51.957Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096578562866463197",
    "title": "深夜乡村车站等来的末班车",
    "titleEn": "The Last Bus Arrives at a Rural Stop After Dark",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "穿宽大外套的年轻女子倚在路灯下的乡村站牌旁看手机、低声自语，远处车灯逐渐靠近，她抬头露出如释重负的笑容。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A young woman in an oversized coat leans on a rural bus-stop sign under a single streetlamp, checks her phone and mutters to herself, then looks up with relief as headlights approach. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A young Korean woman, early 20s, tired after a long day, oversized coat swallowing her frame, hair escaping a loose braid, stands alone at a rural bus stop long after the sun has set.\nSetting: A single flickering streetlamp, an empty road disappearing into darkness, rice paddies barely visible on either side, insects buzzing somewhere in the black.\nCamera feel: Static tripod shot with occasional handheld nudges, like whoever's filming keeps almost forgetting they're recording — heavy grain, the lamp's orange glow bleeding into the frame, tape hiss audible under everything.\nWhat happens (15 sec total):\nShe leans against the bus stop sign, checking her phone, then pocketing it with a tired sigh.\nShe mutters to camera, almost to herself, \"버스 왜 안 와...\" (\"Why isn't the bus coming...\"), kicking a small pebble.\nHeadlights appear far down the road she straightens up, relief crossing her face.\nAs the bus approaches, she looks back at the camera one last time saying \"드디어\" (\"Finally\"), stepping toward the curb as the tape cuts.\nSound: Crickets, distant engine growing louder, gravel underfoot, wind.\nMood: Quiet exhaustion giving way to small relief nothing dramatic, just real.",
    "promptSourceUrl": "https://x.com/Strength04_X/status/2096578562866463197",
    "sourceUrl": "https://x.com/Strength04_X/status/2096578562866463197",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "𝐌",
    "publishedAt": "2026-09-06T12:37:36.000Z",
    "addedAt": "2026-09-11T14:46:51.957Z",
    "mediaUrl": "/media/x-2096578562866463197.mp4",
    "posterUrl": "/posters/x/2096578562866463197.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on @Hailuo_AI \n\nPrompt b- A young Korean woman, early 20s, tired after a long day, oversized coat swallowing her frame, hair escaping a loose braid, stands alone at a rural bus stop long after the sun has set.\nSetting: A single flickering streetlamp, an empty road disappearing into darkness, rice paddies barely visible on either side, insects buzzing somewhere in the black.\nCamera feel: Static tripod shot with occasional handheld nudges, like whoever's filming keeps almost forgetting they're recording — heavy grain, the lamp's orange glow bleeding into the frame, tape hiss audible under everything.\nWhat happens (15 sec total):\nShe leans against the bus stop sign, checking her phone, then pocketing it with a tired sigh.\nShe mutters to camera, almost to herself, \"버스 왜 안 와...\" (\"Why isn't the bus coming...\"), kicking a small pebble.\nHeadlights appear far down the road she straightens up, relief crossing her face.\nAs the bus approaches, she looks back at the camera one last time saying \"드디어\" (\"Finally\"), stepping toward the curb as the tape cuts.\nSound: Crickets, distant engine growing louder, gravel underfoot, wind.\nMood: Quiet exhaustion giving way to small relief nothing dramatic, just real.",
    "engagement": {
      "replies": 22,
      "reposts": 3,
      "likes": 84,
      "views": 2244,
      "snapshotAt": "2026-09-11T14:46:51.957Z"
    },
    "approvedAt": "2026-09-11T14:46:51.957Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096572050617856032",
    "title": "自动驾驶车厢里和猫一起看手机",
    "titleEn": "Riding an Autonomous Cab with a Cat on Your Lap",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "男子坐在无人驾驶车内低头看手机，怀里的橘猫也凑近屏幕，车外街景持续后退，镜头在车内近景之间切换。 原生视频规格为 20 秒 · 608×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man rides in a driverless car scrolling his phone while the ginger cat on his lap leans into the screen, with street scenery passing outside and in-cabin close-ups cutting between them. Native-video output: 20s · 608×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UnrealRafael/status/2096572050617856032",
    "sourceLabel": "X 原帖 · @UnrealRafael",
    "author": "Рафаэль Анриал ∞",
    "publishedAt": "2026-09-06T12:11:43.000Z",
    "addedAt": "2026-09-11T14:46:51.957Z",
    "mediaUrl": "/media/x-2096572050617856032.mp4",
    "posterUrl": "/posters/x/2096572050617856032.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "608×352",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Absolutely insane 🤯 A person is riding with a cat in an autonomous Tesla Cybercab.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 114,
      "snapshotAt": "2026-09-11T14:46:51.957Z"
    },
    "approvedAt": "2026-09-11T14:46:51.957Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096554739588854157",
    "title": "屏幕里从草图到上色的绘画过程",
    "titleEn": "A Sketch-to-Color Painting Process on Screen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面模拟绘图软件界面，人物从铅笔草稿逐步变成线稿再到青色配色完成图，图层面板与色板随进度同步变化。 原生视频规格为 28 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The clip mimics a painting-app interface where a character grows from pencil sketch to line art to a finished teal-toned illustration while layer panels and palettes update alongside. Native-video output: 28s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/oodate/status/2096554739588854157",
    "sourceLabel": "X 原帖 · @oodate",
    "author": "🐡BalloonFish🌊エルタリウム所属",
    "publishedAt": "2026-09-06T11:02:56.000Z",
    "addedAt": "2026-09-11T14:46:51.959Z",
    "mediaUrl": "/media/x-2096554739588854157.mp4",
    "posterUrl": "/posters/x/2096554739588854157.jpg",
    "duration": 28,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Astraはなんでもデキる子だなぁ。\n\nMinimaxH3で生成。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 12,
      "views": 1093,
      "snapshotAt": "2026-09-11T14:46:51.959Z"
    },
    "approvedAt": "2026-09-11T14:46:51.959Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096549119234085017",
    "title": "罗马十五秒杂志式城市短片",
    "titleEn": "A Fifteen-Second Magazine-Style Reel of Rome",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "竖屏画面从斗兽场航拍与\"ROMA\"字样开场，雕像定格配以\"2000 YEARS\"排版，最后在金色黄昏中升起并以\"Rome wasn't built in a day\"收束。 原生视频规格为 15 秒 · 1440×2560。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A vertical reel opens on a Colosseum aerial with \"ROMA\" typography, freezes on a statue behind \"2000 YEARS\", and rises over a golden dusk to close on \"Rome wasn't built in a day\". Native-video output: 15s · 1440×2560. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second, 4K, 16:9 ultra-fast luxury cinematic reel of ROME, ITALY. Blend ancient Roman grandeur, modern Italian street culture, premium fashion-editorial filmmaking and experimental motion graphics.\n\nThe entire reel should feel like pages of an expensive Italian travel magazine suddenly coming alive.\n\n0–2.5s — HISTORY COMES ALIVE\n\nStart extremely close on an old cream-colored paper map of Rome lying on a table.\n\nA thin hand-drawn black route line rapidly travels across the map.\n\nCamera follows the line at high speed.\n\nThe map suddenly transforms into a REAL aerial view of Rome using a seamless paper-to-reality transition.\n\nCamera dives toward the Colosseum.\n\nMassive elegant typography appears INSIDE the architecture using realistic depth masking:\n\n“ROMA”\n\nBehind it, enormous translucent Roman numerals:\n\n“MMXXVI”\n\nTiny editorial text animates around the frame:\n\n“41.9028° N / 12.4964° E”\n“ROMA / ITALIA”\n“LA CITTÀ ETERNA”\n\nVFX: architectural blueprint lines briefly reconstruct the Colosseum's missing sections, showing it complete for half a second before dissolving back into the present-day ruins.\n\n2.5–7s — ITALIAN LIFE\n\nBeat accelerates.\n\nCamera races beside a classic Vespa through a narrow Roman street.\n\nWHIP PAN →\n\nEspresso pouring into a tiny ceramic cup in extreme macro.\n\nMATCH CUT →\n\nTrevi Fountain water splashes directly toward camera and becomes the transition.\n\nFLASH →\n\nItalian fashion model walks confidently past historic architecture.\n\nFAST PUSH-IN →\n\nFresh pasta twists around a fork.\n\nSNAP ZOOM →\n\nPantheon ceiling.\n\nCamera suddenly shoots upward through the opening.\n\nElegant floating typography locks onto objects:\n\n“ESPRESSO”\n“STORIA”\n“MODA”\n“ROMA”\n\nThin arrows, handwritten Italian annotations, postage stamps and tiny coordinates appear like an animated luxury travel journal.\n\n7–12s — ROMAN CHAOS\n\nMusic DROPS.\n\nEditing becomes extremely aggressive.\n\nCOLOSSEUM → VESPA → PASTA → STATUE → ESPRESSO → TREVI → FASHION → STREET → PANTHEON.\n\nEach clip lasts only 0.2–0.35 seconds.\n\nThen FREEZE FRAME.\n\nA Roman statue becomes isolated from the background like a magazine cutout.\n\nHuge typography slides BEHIND the statue:\n\n“2000 YEARS.”\n\nAnother frame violently stacks over it:\n\n“STILL ALIVE.”\n\nThe screen tears horizontally like ripped magazine paper.\n\nThree moving vertical panels appear simultaneously:\n\nANCIENT ROME | STREET LIFE | ITALIAN STYLE\n\nPanels slam together on the beat.\n\nTypography rapidly flashes:\n\n“STORIA.”\n“PASSIONE.”\n“BELLEZZA.”\n“ROMA.”\n\nVFX: analog film burns, photographic flash frames, paper textures, subtle dust, handwritten marks, frame numbers, red editorial circles and fast speed ramps.\n\n12–15s — THE ETERNAL CITY\n\nSuddenly everything becomes quiet.\n\nGolden-hour aerial rises slowly above the Colosseum.\n\nModern Rome stretches endlessly into the background.\n\nFor one second, a subtle transparent reconstruction of ancient Rome overlays the modern city — ancient temples and structures briefly appearing where they once stood.\n\nThen the reconstruction disappears.\n\nHuge elegant serif typography emerges behind the Colosseum:\n\n“ROME WASN’T BUILT\nIN A DAY.”\n\nTiny handwritten text underneath:\n\n“Some cities are visited.\nRome is remembered.”\n\nCamera continues rising.\n\nA flock of birds crosses frame.\n\nSunlight catches the lens.\n\nHard cut to cream paper.\n\nSmall final stamp:\n\n“ROMA / ITALIA”\n\nSTYLE:\nPremium Italian fashion editorial meets historical documentary and modern travel commercial. Cream, terracotta, deep red, warm gold and rich Mediterranean shadows. 35mm film texture, subtle halation, realistic motion blur, natural lens flares, tactile paper graphics and sophisticated typography.\n\nCAMERA:\nDrone dives, FPV sweeps, whip pans, snap zooms, macro photography, orbit shots, low-angle tracking and aggressive speed ramps. No static shots.\n\nVFX:\nDepth-masked typography, Roman architectural reconstruction",
    "promptSourceUrl": "https://x.com/FutureVibesAi/status/2096549334649381263",
    "sourceUrl": "https://x.com/FutureVibesAi/status/2096549119234085017",
    "sourceLabel": "X 原帖 · @FutureVibesAi",
    "author": "Future Vibes AI - Educator",
    "publishedAt": "2026-09-06T10:40:36.000Z",
    "addedAt": "2026-09-11T14:46:51.959Z",
    "mediaUrl": "/media/x-2096549119234085017.mp4",
    "posterUrl": "/posters/x/2096549119234085017.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "motion-graphics"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A Cinematic Reel on Rome \nJust by giving single prompt the Minimax H3 model did everything ....\n@Hailuo_AI \n#MinimaxH3\nPrompt Below 👇",
    "engagement": {
      "replies": 17,
      "reposts": 3,
      "likes": 38,
      "views": 1607,
      "snapshotAt": "2026-09-11T14:46:51.959Z"
    },
    "approvedAt": "2026-09-11T14:46:51.959Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096523805900370402",
    "title": "白发狐耳巫女的一分钟音乐短片",
    "titleEn": "A One-Minute Song PV for a White-Haired Fox Shrine Maiden",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "狐耳巫女与小鼠、发光小兽相处，穿过翡翠金环的螺旋空间，最终在晚霞山景前持扇微笑，配合原创歌曲完成一分钟PV。 原生视频规格为 60 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fox-eared shrine maiden meets a mouse and a glowing creature, passes through a spiral of jade and gold rings, and ends smiling with a fan before sunset mountains in a one-minute song PV. Native-video output: 60s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/StelsRay2/status/2096523805900370402",
    "sourceLabel": "X 原帖 · @StelsRay2",
    "author": "ステスロス@2号機",
    "publishedAt": "2026-09-06T09:00:01.000Z",
    "addedAt": "2026-09-11T14:46:51.959Z",
    "mediaUrl": "/media/x-2096523805900370402.mp4",
    "posterUrl": "/posters/x/2096523805900370402.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "いろいろ試してみた版\nMiniMax H3 x MiniMax music3\n神狐様PV\nhttps://x.com/StelsRay2/status/2096176516422148487",
    "engagement": {
      "replies": 0,
      "reposts": 8,
      "likes": 32,
      "views": 2251,
      "snapshotAt": "2026-09-11T14:46:51.959Z"
    },
    "approvedAt": "2026-09-11T14:46:51.959Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096519541597868299",
    "title": "机械之城上空展翼起飞的少女",
    "titleEn": "A Girl Unfolds Wings and Takes Off Over a Clockwork City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "白裙少女站在齿轮与钟塔组成的机械城市高处，背后展开发光翅膀后猛然起飞，镜头追随她掠过塔尖飞向夜空。 原生视频规格为 15 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl in a white dress stands atop a city of gears and clock towers, unfolds glowing wings, and launches skyward as the camera follows her past the spires into the night. Native-video output: 15s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/soranoblue3/status/2096519541597868299",
    "sourceLabel": "X 原帖 · @soranoblue3",
    "author": "けこねこ",
    "publishedAt": "2026-09-06T08:43:04.000Z",
    "addedAt": "2026-09-11T14:46:51.959Z",
    "mediaUrl": "/media/x-2096519541597868299.mp4",
    "posterUrl": "/posters/x/2096519541597868299.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "fantasy",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "《KusArtのMiniMax H3》\n#KusArt \n\n機械仕掛けの街を\n見守っているサイボーグの女の子\n\nポイントは\n背中から翼が出てくる場面と\n勢い良く飛び立つ瞬間\n空を飛んでいる姿を捉えるカメラワーク\n\n画像も複数枚参照できるけど\n今回は静止画1枚のみ\n\n最大15秒で破綻もほぼ無し(*´꒳`*)\n\n今ならKusArt非会員でも、MiniMax H3を5回無料で利用可能だって！\n\nさらに友達を招待すると、MiniMax H3の無料利用回数を追加で獲得！\n\nKusArt会員はMiniMax H3を回数制限なく無料で利用できるみたい♡\n\nキャンペーン期間は9月13日まで\n\n無料でここまでできるのはスゴいね♡\n\n期間限定キャンペーンはリプ欄から⤵",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 24,
      "views": 737,
      "snapshotAt": "2026-09-11T14:46:51.959Z"
    },
    "approvedAt": "2026-09-11T14:46:51.959Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096669887980511477",
    "title": "海边举臂召唤闪电的男子",
    "titleEn": "A Man Raises His Arms to Summon Lightning at Sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "披着粗布斗篷的男子站在夜色海面前高举双臂，表情由紧绷转为呼喊，最后一道闪电劈落在身后的海面上。 原生视频规格为 10 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in a rough cloak stands before a night sea with both arms raised, his face tightening into a shout as a bolt of lightning finally strikes the water behind him. Native-video output: 10s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/theNdroidz/status/2096669887980511477",
    "sourceLabel": "X 原帖 · @theNdroidz",
    "author": "ndroidz 🇫🇷",
    "publishedAt": "2026-09-06T18:40:30.000Z",
    "addedAt": "2026-09-12T05:30:05.092Z",
    "mediaUrl": "/media/x-2096669887980511477.mp4",
    "posterUrl": "/posters/x/2096669887980511477.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "seedance2.5: desolé ce contenu pourrait enfreindre les règles de...\nminimax h3:",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 90,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.092Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096630852147179848",
    "title": "回眸的暗调人像",
    "titleEn": "A Moody Over-the-Shoulder Portrait",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发女子在深色背景中侧身回头，浅色外套与面部细微的表情变化在五秒内完成一次安静的凝视。 原生视频规格为 5 秒 · 1080×1620。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman turns to look back over her shoulder against a black backdrop, her pale jacket and subtle facial shifts carrying a quiet five-second gaze. Native-video output: 5s · 1080×1620. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KusoPhoto/status/2096630852147179848",
    "sourceLabel": "X 原帖 · @KusoPhoto",
    "author": "空想写真家",
    "publishedAt": "2026-09-06T16:05:23.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096630852147179848.mp4",
    "posterUrl": "/posters/x/2096630852147179848.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1080×1620",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "エモいポートレート画像＆動画\n\n画像：Magnific\n動画：Topview（MiniMax H3 Max）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 1977,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096630655019061719",
    "title": "彩虹天台到楼梯间的少女",
    "titleEn": "A Girl Walks from a Rainbow Rooftop into the Stairwell",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "穿紫色开衫的黑发少女站在城市天台望向阴云与彩虹，随后转身走进锈迹斑斑的楼梯间，画面从开阔切到封闭。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired girl in a purple cardigan looks out from a city rooftop at clouds and a rainbow, then turns and walks into a rusted stairwell as the frame moves from open to enclosed. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/syuririn_life/status/2096630655019061719",
    "sourceLabel": "X 原帖 · @syuririn_life",
    "author": "しゅりりん🇯🇵_転職したい(バイブコーダー)/AI踊らされ屋",
    "publishedAt": "2026-09-06T16:04:36.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096630655019061719.mp4",
    "posterUrl": "/posters/x/2096630655019061719.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "falのMiniMax H3で作った。\nhttps://fal.ai/models/minimax/h3-max/reference-to-video",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 117,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096627369549197752",
    "title": "橄榄球与时尚排版的十五秒冲击",
    "titleEn": "Fifteen Seconds of Football and Fashion Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橄榄球运动员以时尚大片的姿态出镜，\"FIERCE FOCUSED FEARLESS\"与\"BUILT FOR IMPACT\"等大字随硬碰撞和快切铺满竖屏。 原生视频规格为 15 秒 · 768×1376。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Football players pose like a fashion editorial while bold phrases such as \"FIERCE FOCUSED FEARLESS\" and \"BUILT FOR IMPACT\" fill the vertical frame between hard hits and fast cuts. Native-video output: 15s · 768×1376. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/koss_ai/status/2096627369549197752",
    "sourceLabel": "X 原帖 · @koss_ai",
    "author": "koss",
    "publishedAt": "2026-09-06T15:51:33.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096627369549197752.mp4",
    "posterUrl": "/posters/x/2096627369549197752.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1376",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Went a little harder with this one. 🏈\n\nFootball, fashion-editorial graphics, hard impacts, fast cuts, and a lot of typography packed into 15 seconds.\n\nMade with @Hailuo_AI MiniMax H3.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 174,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096605175339208830",
    "title": "机器人法官面前的法庭质问",
    "titleEn": "A Courtroom Plea Before a Robot Judge",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白发老人对着麦克风激动陈词，镜头切到端坐法官席的机器人，最后以\"ETHICS: NON-EXISTENT\"的黑底字幕收尾。 原生视频规格为 29 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired man delivers an impassioned statement into a microphone, the camera cuts to a robot seated on the judge’s bench, and the piece closes on a black title card reading \"ETHICS: NON-EXISTENT\". Native-video output: 29s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AITalesNBH/status/2096605175339208830",
    "sourceLabel": "X 原帖 · @AITalesNBH",
    "author": "AI Tales - Not by Humans",
    "publishedAt": "2026-09-06T14:23:21.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096605175339208830.mp4",
    "posterUrl": "/posters/x/2096605175339208830.jpg",
    "duration": 29,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "dialogue",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Is this court case merely a fantasy, or could it become a reality?\n\nAre human ethics relative, and would it be acceptable to leave ethical principles out of future AI systems and AI-powered robots?\n\nCreated with #MiniMaxH3 on @Hailuo_AI + some editing 🎥🎬",
    "engagement": {
      "replies": 11,
      "reposts": 0,
      "likes": 25,
      "views": 715,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096603405372567887",
    "title": "三种稀疏注意力参数下的全息舞者",
    "titleEn": "A Hologram Dancer Under Three Sparse-Attention Settings",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一段蓝色全息舞者在人群中旋转的画面被纵向排成三行，分别对应不同的稀疏注意力起始参数，用于并列观察差异。 原生视频规格为 10 秒 · 1344×2304。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same blue holographic dancer spinning inside a crowd is stacked in three rows, each rendered with a different sparse-attention start setting for side-by-side comparison. Native-video output: 10s · 1344×2304. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ITOKtw/status/2096603405372567887",
    "sourceLabel": "X 原帖 · @ITOKtw",
    "author": "ITOK",
    "publishedAt": "2026-09-06T14:16:19.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096603405372567887.mp4",
    "posterUrl": "/posters/x/2096603405372567887.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1344×2304",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MiniMaxH3 Block Sparse Attentionでstart_percentを0にした場合に，dense_blockを「33-35,39-42」に設定すると，start_percent 1 に寄ります。\n画像1:start_percent 0\n画像2:+dense_block設定\n画像3:start_percent 1",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 5,
      "views": 731,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096601533131403760",
    "title": "夜店里专注打碟的女DJ",
    "titleEn": "A DJ Lost in the Mix at a Dark Club",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "从舞台远景推进到戴白色耳机的女DJ，她在昏暗的红光与闪光灯下低头操作控制器，画面带有胶片颗粒质感。 原生视频规格为 10 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera pushes from a distant stage view to a DJ in white headphones working her controller under dim red light and flash, with a grainy film look throughout. Native-video output: 10s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PrometheanAIX/status/2096601533131403760",
    "sourceLabel": "X 原帖 · @PrometheanAIX",
    "author": "Promethean",
    "publishedAt": "2026-09-06T14:08:53.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096601533131403760.mp4",
    "posterUrl": "/posters/x/2096601533131403760.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "retro"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "PROMPTSHARE: Live performances work when the camera captures energy, not perfection.\n\n#Marketing #Branding #Advertising #HailuoCPP #HailuoAI @Hailuo_AI\n\nThis is why:\n\nThat rawness is what gives a performance its pulse.\n\nThe movement, lighting, crowd, and unpredictability are what make the artist feel genuinely present.\n\nI created this video using Hailuo MiniMax H3.\n\nHailuo works especially well when you want believable human movement, cinematic energy, and strong visual storytelling.\n\nPaid partnership with Hailuo.\n\n➜ Image Prompt:\nA hyper-realistic close-up nightlife portrait of a beautiful adult female DJ performing intensely behind a professional DJ controller in a dark club environment. She has long, voluminous dark brown hair caught in motion, flying outward around her face as if from dancing, turning quickly, or moving with the beat. Her expression is focused and absorbed in the music, eyes lowered toward the mixer rather than looking at the camera.\n\nShe wears large white over-ear DJ headphones, a fitted white asymmetrical one-shoulder top with a deep neckline, and a simple silver chain necklace. Her styling is minimal, modern, and slightly Y2K-inspired. One arm reaches forward toward the mixer while the other works the controls, creating the feeling of a real performance captured mid-action.\n\nFrame the image tightly from approximately the mid-torso upward, with the DJ controller filling the lower foreground. Include recognizable mixer details such as knobs, faders, jog wheels, illuminated buttons, and small cyan, orange, and blue LED accents. Allow the foreground hand to fall slightly out of focus from movement and shallow depth of field.\n\nThe background should be very dark and indistinct, suggesting an intimate nightclub or underground electronic music venue, with only a few soft orange and amber lights glowing in the distance.\n\nPhotograph the scene using strong direct on-camera flash mixed with low ambient club lighting, creating intentionally bright skin highlights, slightly blown-out whites, deep shadowy blacks, and dramatic separation from the background. The flash should freeze parts of her face and body while her moving hair retains subtle motion blur.\n\nGive the photograph a raw late-1990s / early-2000s club photography aesthetic with visible fine film grain, slight overexposure, soft bloom, halation around highlights, cool cyan shadows, warm peach skin tones, mild color shifts, and imperfect analog texture.\n\nUnderground DJ culture, candid nightlife photography, Y2K club editorial, caught-in-action performance, 35mm point-and-shoot aesthetic, direct flash, shallow depth of field, slight motion blur, dynamic flying hair, realistic skin texture, visible pores, natural anatomy, authentic hands, detailed hair strands, photorealistic, no plastic skin, no beauty-filter smoothing, no CGI appearance, no overly polished studio look.",
    "engagement": {
      "replies": 39,
      "reposts": 2,
      "likes": 63,
      "views": 3379,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096600345959145839",
    "title": "清晨空泳池馆里的独行者",
    "titleEn": "First One In at an Empty Morning Pool Hall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "晨光穿过拱窗照进空无一人的室内泳池馆，一名女子沿池边慢慢走过，倒影与蒸汽在对称构图中静静延伸。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Morning light pours through arched windows into an empty indoor pool hall as a lone woman walks along the edge, her reflection and the rising mist stretching across a symmetrical frame. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2096600345959145839",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-09-06T14:04:10.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096600345959145839.mp4",
    "posterUrl": "/posters/x/2096600345959145839.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Good morning. First one in.\n\n→ @Hailuo_AI MiniMax H3 in @krea_ai",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 16,
      "views": 1977,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096584372468232385",
    "title": "摘不下耳机的循环夜晚",
    "titleEn": "A Looping Night That Won’t Let Go of the Headphones",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少年在夜晚的卧室里躺着刷手机，镜头在戴耳机的面部特写与播放器界面之间来回，歌词字幕贯穿一段一分四十秒的音乐影像。 原生视频规格为 103 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A teenager lies on his bed at night scrolling his phone while the camera moves between headphone close-ups and the player screen, with lyric captions running through a one-minute-forty-second music piece. Native-video output: 103s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kiyu___K/status/2096584372468232385",
    "sourceLabel": "X 原帖 · @kiyu___K",
    "author": "yuuki",
    "publishedAt": "2026-09-06T13:00:41.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096584372468232385.mp4",
    "posterUrl": "/posters/x/2096584372468232385.jpg",
    "duration": 103,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "グルグル回って、抜けられない。\n\n13. Repeat mode\nMusic: Suno \nLyrics: yuuki\nVocal: 紺音\n\nVisuals: GPT Image 2 / MiniMax H3\nEdit: DaVinci Resolve",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 14,
      "views": 577,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096573091522527553",
    "title": "丛林中现身的红翼巨兽",
    "titleEn": "A Red-Winged Beast Emerges in the Jungle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长着蝙蝠翅膀与利爪的黑色巨兽在潮湿丛林中缓步逼近镜头，红色翼膜张开，毛发与光影细节保持稳定。 原生视频规格为 15 秒 · 1080×1944。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black beast with bat-like wings and claws stalks toward the camera through a humid jungle, spreading its red wing membranes while fur and lighting stay consistent. Native-video output: 15s · 1080×1944. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CanAiStudios/status/2096573091522527553",
    "sourceLabel": "X 原帖 · @CanAiStudios",
    "author": "Can Ece",
    "publishedAt": "2026-09-06T12:15:52.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096573091522527553.mp4",
    "posterUrl": "/posters/x/2096573091522527553.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1944",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "creatures",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Jungle Creature... 🦇🌿\nBase render in ComfyUI via Krea 2, animated by MiniMax H3, and enhanced with SeedVR2 HD. #CreatureDesign",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 20,
      "views": 441,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096554503223058607",
    "title": "从手绘分镜到奔跑的动画角色",
    "titleEn": "From Hand-Drawn Storyboard to a Running Anime Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面先拍手在纸上画分镜、透台上翻动线稿，最后切到穿白色机甲装的兔耳女孩全速奔跑的完成动画。 原生视频规格为 15 秒 · 720×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The clip opens on hands sketching storyboards and flipping line art on a lightbox, then cuts to the finished animation of a bunny-eared girl in white mecha gear sprinting at full speed. Native-video output: 15s · 720×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kurahu_capten/status/2096554503223058607",
    "sourceLabel": "X 原帖 · @kurahu_capten",
    "author": "kurahu@AIart",
    "publishedAt": "2026-09-06T11:02:00.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096554503223058607.mp4",
    "posterUrl": "/posters/x/2096554503223058607.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×960",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI Anime Production Studio 🏃🎞️\nAI video created with MiniMax H3 Max via TopView.",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 37,
      "views": 1968,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096550979579629806",
    "title": "灰发少女的托腮微笑与比耶",
    "titleEn": "A Girl’s Chin-Rest Smile and Peace Sign",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰发绿丝带的动画少女先双手托腮闭眼微笑，再抬手比出剪刀手，粉色运动装与背景在八秒内保持一致。 原生视频规格为 8 秒 · 720×936。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime girl with grey hair and a green ribbon rests her chin in her hands with a closed-eye smile, then flashes a peace sign, keeping her pink sportswear and backdrop consistent over eight seconds. Native-video output: 8s · 720×936. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sac25252/status/2096550979579629806",
    "sourceLabel": "X 原帖 · @sac25252",
    "author": "sac@AIイラスト",
    "publishedAt": "2026-09-06T10:48:00.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096550979579629806.mp4",
    "posterUrl": "/posters/x/2096550979579629806.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "720×936",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMaxH3 音声は消してます\nラブライブ 南ことり #lovelive #AI",
    "engagement": {
      "replies": 14,
      "reposts": 181,
      "likes": 3346,
      "views": 160363,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed.",
    "featured": {
      "at": "2026-09-13T00:21:49.935Z",
      "basis": "trending"
    }
  },
  {
    "id": "x-2096548015943328174",
    "title": "课桌前分享便当的两个女孩",
    "titleEn": "Two Girls Share a Bento at Their Desk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "Q版画风的黑发与棕发女孩并肩坐在教室课桌前，一人夹起便当里的菜喂给另一人，两人表情随之变化。 原生视频规格为 10 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two chibi-style schoolgirls sit side by side at a classroom desk as one lifts food from a bento box to feed the other, their expressions shifting with each bite. Native-video output: 10s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C12592947592/status/2096548015943328174",
    "sourceLabel": "X 原帖 · @C12592947592",
    "author": "supercub3",
    "publishedAt": "2026-09-06T10:36:13.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096548015943328174.mp4",
    "posterUrl": "/posters/x/2096548015943328174.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MinimaxH3ローカル\nお弁当を忘れたユリカと分けてくれるもも",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 21,
      "views": 536,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096523774288138264",
    "title": "和风衣装少女的三段衔接",
    "titleEn": "A Girl in Japanese-Style Dress Across Three Joined Takes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发少女身着白底红纹的和风短装，在浅灰背景中依次完成抬手、席地而坐和站立回眸，三段本地生成被拼接成十五秒。 原生视频规格为 15 秒 · 1080×1490。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired girl in a white-and-red Japanese-style outfit moves through a raised hand, a seated pose, and a standing look-back against pale grey, three locally generated takes joined into fifteen seconds. Native-video output: 15s · 1080×1490. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Shinki_Protocol/status/2096523774288138264",
    "sourceLabel": "X 原帖 · @Shinki_Protocol",
    "author": "神代カナ@AIアイドル",
    "publishedAt": "2026-09-06T08:59:53.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096523774288138264.mp4",
    "posterUrl": "/posters/x/2096523774288138264.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1490",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n\n試してみました✨️\nローカルで動かしたので諸事情で３分割したものをくっつけました💦",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 22,
      "views": 864,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096519873358946504",
    "title": "小剧场里的双人漫才",
    "titleEn": "A Two-Person Manzai Routine in a Small Venue",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "两名穿乐队T恤的表演者站在小剧场麦克风前轮流接话，观众席笑声与鼓掌配合，口型对应两个不同说话者。 原生视频规格为 12 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two performers in band T-shirts trade lines at a small-venue microphone while the audience laughs and claps, with lip movement following each of the two distinct speakers. Native-video output: 12s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aichof21/status/2096519873358946504",
    "sourceLabel": "X 原帖 · @aichof21",
    "author": "aichof",
    "publishedAt": "2026-09-06T08:44:23.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096519873358946504.mp4",
    "posterUrl": "/posters/x/2096519873358946504.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "dialogue",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "1年くらい前に作った音源。\n給湯室のシチュエーションネタですが、漫才っぽくしてみました。\n\n元の音源はもっと長いのですが、今回はリップシンクで「二人の掛け合いをちゃんと認識して動画生成できるのか」を11秒だけテストしてみました。\n\n以前のリップシンクは、二人の話者をうまく使い分けられず、一人ずつ場面を切ってリップシンク→あとで組み合わせる、といった作業が必要だったと思いますが、\n\n今回はトークスクリプトの補助も使わず、音源をリファレンスとして渡し、「どちらから話し始めるか」だけを指定。\n\nそれでも、それなりに二人の掛け合いとして形になりました。\n\nやり方としては、\nChatGPTに素材を与えてトークスクリプトを作成\n↓\nGoogle AI Studioのトーク機能で、異なる声質の二人を設定して会話音源を生成\n↓\nあとは画像と音源をMiniMax H3のリファレンスにする\n\nという感じです。\nまだ試してないですが、トークスクリプトだけ与えたらやってくれそうな気もしますね😆",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 17,
      "views": 733,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096519385502667234",
    "title": "月轮、蓝瞳与时钟的剪影拼贴",
    "titleEn": "Silhouettes, Blue Eyes and a Clock Face in Collage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少女与猫的剪影站在巨大月轮前，画面切换为蓝色眼睛的分格拼贴，最后落到齿轮时钟上的剪影，黑白蓝三色贯穿始终。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Silhouettes of a girl and a cat stand before a huge moon disc, the frame breaks into a collage of blue eyes, and the piece lands on silhouettes over a clock face, all in black, white and blue. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/zuomuyan2542801/status/2096519385502667234",
    "sourceLabel": "X 原帖 · @zuomuyan2542801",
    "author": "あたらくしあ@週末AIクリエイター🎨",
    "publishedAt": "2026-09-06T08:42:27.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096519385502667234.mp4",
    "posterUrl": "/posters/x/2096519385502667234.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "test2\n\n#MiniMaxH3\n#HailuoAI\n#AIvideo",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 427,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096518977610826075",
    "title": "极光下发光的雪松环",
    "titleEn": "Glowing Rings Among Snow Pines Under the Aurora",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿色极光在雪原上空流动，几棵雪松被发光的圆环围绕并逐棵亮起，是一段为现场演出准备的五秒循环画面。 原生视频规格为 5 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Green aurora drifts over a snowfield while glowing rings encircle a stand of snow-covered pines that light up one by one in a five-second loop made for live visuals. Native-video output: 5s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelrhythms/status/2096518977610826075",
    "sourceLabel": "X 原帖 · @pixelrhythms",
    "author": "Keshav",
    "publishedAt": "2026-09-06T08:40:50.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096518977610826075.mp4",
    "posterUrl": "/posters/x/2096518977610826075.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "VJ loops with Krea + MiniMax H3, inspired by Fantasia and Eric Prydz.\n\nThinking a lot about visual music — how light, color, shape and motion can find a rhythm together.\n\nExcited to take these into TouchDesigner and bring them to life through live performance.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 129,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096518817723957379",
    "title": "三秒循环中的女仆眨眼",
    "titleEn": "A Maid Blinks in a Three-Second Loop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "穿黑白女仆装的插画少女双手贴胸，在三秒内完成一次眨眼与轻微抬头，首尾帧一致以便无缝循环。 原生视频规格为 3 秒 · 640×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustrated girl in a black-and-white maid outfit holds her hands to her chest and blinks with a slight lift of the head, first and last frames matched for a seamless three-second loop. Native-video output: 3s · 640×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2096518817723957379",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-09-06T08:40:12.000Z",
    "addedAt": "2026-09-12T05:30:05.093Z",
    "mediaUrl": "/media/x-2096518817723957379.mp4",
    "posterUrl": "/posters/x/2096518817723957379.jpg",
    "duration": 3,
    "aspectRatio": "portrait",
    "resolution": "640×832",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax-h3-comfyui-anime-loopのお試し。開始・終了フレームを同じにしてループ動画を作成できるっぽい。動きが少ないと成功しやすい感じ。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 647,
      "snapshotAt": "2026-09-12T05:30:05.093Z"
    },
    "approvedAt": "2026-09-12T05:30:05.093Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096496558271562137",
    "title": "白狐少女插画的绘制延时",
    "titleEn": "A Timelapse of Painting a White Fox Girl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面模拟绘图软件界面，白狐耳少女从铅笔骨架逐步长出线稿与配色，工具栏和图层面板随进度同步更新。 原生视频规格为 10 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The clip mimics a painting-app window where a white fox-eared girl grows from a pencil skeleton into line art and color while toolbars and layer panels update alongside. Native-video output: 10s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kamome_mint/status/2096496558271562137",
    "sourceLabel": "X 原帖 · @kamome_mint",
    "author": "かもめ-みんと ➺✦",
    "publishedAt": "2026-09-06T07:11:45.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096496558271562137.mp4",
    "posterUrl": "/posters/x/2096496558271562137.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "多分燃えるやつなのだ\nminimax h3 タイムラプス",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 463,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096493234457567420",
    "title": "火焰在指尖凝聚的橙发少女",
    "titleEn": "Flames Gather at an Orange-Haired Girl’s Fingertips",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙发少女摊开手掌，橙红火焰沿手臂旋转汇聚成漩涡，最后在荒野上炸开成一片火光与烟尘。 原生视频规格为 15 秒 · 2688×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An orange-haired girl opens her palm as red-orange flames spiral up her arm into a vortex, finally bursting across the wasteland in fire and smoke. Native-video output: 15s · 2688×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mito_ai_la/status/2096493234457567420",
    "sourceLabel": "X 原帖 · @mito_ai_la",
    "author": "みと Visuals×AI",
    "publishedAt": "2026-09-06T06:58:32.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096493234457567420.mp4",
    "posterUrl": "/posters/x/2096493234457567420.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2688×1536",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Pollo AI（@PolloAIJP）でMiniMax H3 Maxを試してみました。#PolloAI #MiniMaxH3Max\n今回はあえて静かな動きの中で、光や色をどこまで綺麗に表現できるか実験。\n通常のH3より必要なクレジットが少なく、生成速度もかなり速いモデル。この約15秒の動画も体感では30秒もかからないほどで生成されて、思っていた以上の速さに驚きました。\n現在はウルトラプランにアップグレードすると、MiniMax H3とH3 Maxを365日間無制限で利用できます。\n\nPolloAI URL：https://tinyurl.com/8vwtz8c4",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 34,
      "views": 1711,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096488811887054963",
    "title": "两种稀疏起始参数下的舞台歌姬",
    "titleEn": "A Stage Singer Under Two Sparse-Start Settings",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一位黑发歌姬在霓虹舞台演唱的画面左右并排，分别标注不同的稀疏注意力起始比例，用于对照生成时间与画质差异。 原生视频规格为 10 秒 · 1122×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two side-by-side renders of the same black-haired singer on a neon stage are labelled with different sparse-attention start ratios to compare generation time and quality. Native-video output: 10s · 1122×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2096488811887054963",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-09-06T06:40:58.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096488811887054963.mp4",
    "posterUrl": "/posters/x/2096488811887054963.jpg",
    "duration": 10,
    "aspectRatio": "square",
    "resolution": "1122×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 ref2va：パラメータ調整中。start_percentをデフォルト20%から0%（0step目からスパース）にすると1024x1792, 5s, 24fps, 124f, RTX4070で\n\n 20%: 268s（4m28s）\n   0%: 230s（3m50s）\n\nと4分切る！欠点は画質音質の劣化。変な文字が増えてるのと音がやや割れる。それ以外はOKそう。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 1374,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096488212164432214",
    "title": "持杖少女对战斗篷怪物",
    "titleEn": "A Staff-Wielding Girl Fights Hooded Monsters",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白衣少女持长杖摆出架势，与两只戴尖帽的黑色怪物交手，最后一记旋风将怪物卷入蓝色龙卷之中。 原生视频规格为 15 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl in white takes a fighting stance with a long staff, trades blows with two hooded black creatures, and finishes by sweeping them into a blue tornado. Native-video output: 15s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIHobbyist65536/status/2096488212164432214",
    "sourceLabel": "X 原帖 · @AIHobbyist65536",
    "author": "もりそば@跡地",
    "publishedAt": "2026-09-06T06:38:35.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096488212164432214.mp4",
    "posterUrl": "/posters/x/2096488212164432214.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "もう少しだけアニメ調にしようと頑張ってみた(一旦つかれたので検証ここまで)\n#AIイラスト #AI動画 #MinimaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 1820,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096485403402575882",
    "title": "深夜来电的黑衣男子",
    "titleEn": "A Man in Black Takes a Late-Night Call",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "插画风格的黑衣男子在昏暗房间里对着镜头说话，中段切到堆满显示器的昏暗办公室，一段一分四十秒的荒诞幽默实验。 原生视频规格为 102 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An illustrated man in black speaks to camera in a dim room, the middle cuts to a gloomy office stacked with monitors, and the whole runs as a one-minute-forty-second absurd-humor experiment. Native-video output: 102s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Moonshub1123/status/2096485403402575882",
    "sourceLabel": "X 原帖 · @Moonshub1123",
    "author": "Shubh | Notes on the world",
    "publishedAt": "2026-09-06T06:27:25.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096485403402575882.mp4",
    "posterUrl": "/posters/x/2096485403402575882.jpg",
    "duration": 102,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "illustration",
      "dark"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A delightfully unhinged humor experiment: Astra running MiniMax H3 Max on Fal, just to see what kind of chaos it cooks up 😂😅",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 220,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096455228006756716",
    "title": "秋日哥本哈根的彩色街区航拍",
    "titleEn": "An Autumn Aerial Over Copenhagen’s Colored Streets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从金黄树冠掠过彩色排屋与停满车的街道，最后升至塔楼与广场上空，十五秒内保持连续的航拍推进。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera sweeps from golden treetops over rows of colored houses and parked cars, rising to a tower and square in one continuous fifteen-second aerial push. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rukanatsui/status/2096455228006756716",
    "sourceLabel": "X 原帖 · @rukanatsui",
    "author": "夏井瑠夏🍊ruKa_N | AI音楽",
    "publishedAt": "2026-09-06T04:27:31.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096455228006756716.mp4",
    "posterUrl": "/posters/x/2096455228006756716.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "秋のコペンハーゲン\nFlux1 Dev + MiniMax H3\n15秒に挑戦。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 8,
      "views": 431,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096454409383534646",
    "title": "穿衣猫鼠的可乐广告",
    "titleEn": "A Dressed Cat and Mouse Star in a Cola Spot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "从挂着水珠的蓝色易拉罐特写开始，穿衣服的猫和老鼠在餐桌前抢食嬉闹，最后并肩坐在两罐可乐之间。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Opening on a dew-covered blue can, a cat and a mouse in clothes scramble over food at a dining table before settling side by side between two cola cans. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/breath_in_code/status/2096454409383534646",
    "sourceLabel": "X 原帖 · @breath_in_code",
    "author": "LinacoWSL",
    "publishedAt": "2026-09-06T04:24:16.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096454409383534646.mp4",
    "posterUrl": "/posters/x/2096454409383534646.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "commercial"
    ],
    "scenes": [
      "product",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MINIMAX H3\n面白そうなのでART_STYLE＝Pixarを自分のディレクションシステムで試してみました。弱々PCなので設定は以下💦\n\nDuration: 15　480p\nStyle: Action\nART_STYLE: Pixar 3D\nテーマ『洋服を着た猫とネズミの面白いペプシコーラのCM』\n\nDLSS5x2⇢1728ｘ960　フレーム補間x3⇢72FPS",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 196,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096451834886131927",
    "title": "粉发角色的两段无缝衔接",
    "titleEn": "Two Takes of a Pink-Haired Character Joined Seamlessly",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "粉发带角的插画角色在白底上换姿势，两段首尾帧接续的生成被拼成八秒，衔接处几乎看不出跳变。 原生视频规格为 8 秒 · 960×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired horned character shifts poses against white as two first-to-last-frame continuations are joined into eight seconds with barely visible seams. Native-video output: 8s · 960×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FudooManken/status/2096451834886131927",
    "sourceLabel": "X 原帖 · @FudooManken",
    "author": "不働漫画研究会",
    "publishedAt": "2026-09-06T04:14:02.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096451834886131927.mp4",
    "posterUrl": "/posters/x/2096451834886131927.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "960×1280",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今日時点で違和感なくMinimax-H3で生成した動画を結合する手法は、H3 Infinite Continuation Suite v1.4 (FL2VA)かもしれない。\n\nFL2VAモデルを使ったI2V開始＋コンテキスト継続生成で、２本の動画をズレが気にならない程度に結合できた。\n\nクオリティあげられれば、使い物になるかもしれない。他も作ってみよう。\nhttps://www.reddit.com/r/StableDiffusion/s/qK5M3txjsM\n\n※さっきイヴォンヌの耳が新しく作られてることがわかったが、実験なのでご勘弁。\n\n#エンドフィールド‌",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 1481,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096451163990397237",
    "title": "黑白双色发的哥特时装造型",
    "titleEn": "A Gothic Fashion Look with Two-Tone Hair",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白双色长发、头戴小角的女子身穿皮带装饰的黑色时装，从面部特写到全身舞台造型，最后张开臂膀展露翼形披风。 原生视频规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman with two-tone black-and-white hair and small horns wears strap-detailed black fashion, moving from a face close-up to a full-body stage look before spreading a winged cape. Native-video output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mars_eve/status/2096451163990397237",
    "sourceLabel": "X 原帖 · @mars_eve",
    "author": "MARS★EVE",
    "publishedAt": "2026-09-06T04:11:22.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096451163990397237.mp4",
    "posterUrl": "/posters/x/2096451163990397237.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "dark"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Chip🖤Fashion\n#minimax #h3 @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 175,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096447601839423915",
    "title": "夜空召唤流星的魔女",
    "titleEn": "A Witch Calls Down Meteors at Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴尖帽的蓝裙魔女在星空下举起双手，紫色漩涡中落下成串流星，最后森林被火光照亮而她站在火焰中央。 原生视频规格为 10 秒 · 512×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A witch in a blue dress and pointed hat raises her hands under the stars, meteors stream down from a purple vortex, and the forest lights up in flames with her standing at the center. Native-video output: 10s · 512×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YamataZen/status/2096447601839423915",
    "sourceLabel": "X 原帖 · @YamataZen",
    "author": "Yamata",
    "publishedAt": "2026-09-06T03:57:13.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096447601839423915.mp4",
    "posterUrl": "/posters/x/2096447601839423915.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "512×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Vega summoning meteors at night (she loses her powers at day)\n\n#aivideo #MiniMaxH3 #OC",
    "engagement": {
      "replies": 1,
      "reposts": 6,
      "likes": 14,
      "views": 727,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096447146614747210",
    "title": "从校园俯瞰到课桌上的书",
    "titleEn": "From a School Aerial to a Book on the Desk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "画面从校舍俯瞰图推进到教室里坐在课桌前的女孩，最后落到摊开的书本特写，参照图像与参照视频被接成十二秒。 原生视频规格为 12 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The shot moves from a bird’s-eye view of a school into a classroom where a girl sits at her desk, ending on a close-up of an open book, with a reference image and reference video joined into twelve seconds. Native-video output: 12s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sankakuten91256/status/2096447146614747210",
    "sourceLabel": "X 原帖 · @sankakuten91256",
    "author": "さんかくてん",
    "publishedAt": "2026-09-06T03:55:24.000Z",
    "addedAt": "2026-09-12T05:30:05.094Z",
    "mediaUrl": "/media/x-2096447146614747210.mp4",
    "posterUrl": "/posters/x/2096447146614747210.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n参照画像 + 参照動画の実験\n参照動画(10s)の前に参照画像を足して12sの動画\n\n鳥瞰図（参照画像） -> 窓から本（参照動画）\n\n鳥瞰図などの画像は、GPT Image2で作成\n3Dの参照動画は、GPT Astraを使用してBlenderで作成",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 8,
      "views": 1575,
      "snapshotAt": "2026-09-12T05:30:05.094Z"
    },
    "approvedAt": "2026-09-12T05:30:05.094Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096404374516093281",
    "title": "蛋形头套女孩的三分四十秒MV",
    "titleEn": "A Three-Minute-Forty MV for an Egg-Headed Mascot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴着无脸白色蛋形头套的女孩在夜晚街头站立、骑自行车穿过城市，歌词字幕逐句出现，五十个镜头拼成一支完整的竖屏音乐影像。 原生视频规格为 220 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl wearing a faceless white egg-shaped mascot head stands on night streets and rides a bicycle through the city while lyric captions appear line by line across fifty cuts of a full vertical music piece. Native-video output: 220s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Simplelifecom/status/2096404374516093281",
    "sourceLabel": "X 原帖 · @Simplelifecom",
    "author": "Tamachan Label",
    "publishedAt": "2026-09-06T01:05:26.000Z",
    "addedAt": "2026-09-12T07:53:30.083Z",
    "mediaUrl": "/media/x-2096404374516093281.mp4",
    "posterUrl": "/posters/x/2096404374516093281.jpg",
    "duration": 220,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "のっぺらぼうの着ぐるみが主役のMVを作りました。『Still Too Young』——なのに画像生成が、何度描き直しても勝手に目を描いてくる。白い卵型の巨体はどこかの有名キャラに似すぎているせいか、「顔を描くな」がまるで効かない。50カット、3分40秒。\n\n音楽はsuno4.5＋、静止画はKrea 2、動画はMiniMax-H3、編集はRemotion。50本焼くのに約6時間。いちばん厄介だったのは自転車が途中で逆走したこと。静止画が前向きなのに「後ろ姿で遠ざかる」と指示したので、AIが辻褄合わせに世界ごと180度回していました。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 370,
      "snapshotAt": "2026-09-12T07:53:30.083Z"
    },
    "approvedAt": "2026-09-12T07:53:30.083Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2096579639464579484",
    "title": "一人社长AI记账的动态字幕短片",
    "titleEn": "Kinetic Type for a Solo-Founder AI Bookkeeping Story",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "淡紫色底上，简笔画女孩从成堆收据中跃起，\"登録\"等日文大字随节奏放大切换，最后落到\"ひとり社長のAI経理\"的标题画面。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "On a pale purple ground a line-drawn girl leaps out of a pile of receipts while large Japanese words such as \"Register\" scale and cut on the beat, ending on a title card for a solo-founder AI bookkeeping story. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ayumi_t820/status/2096579639464579484",
    "sourceLabel": "X 原帖 · @ayumi_t820",
    "author": "Ayumi/AIイラスト",
    "publishedAt": "2026-09-06T12:41:53.000Z",
    "addedAt": "2026-09-12T07:53:30.083Z",
    "mediaUrl": "/media/x-2096579639464579484.mp4",
    "posterUrl": "/posters/x/2096579639464579484.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "motion-graphics",
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "noteに、税理士さん無しで一人で 、freeeとClaude Codeで法人決算と税務申告をのりきった話を書きました！前回のfreee mcpでできることできないこと記事の続きです。ご興味ありましたらのぞいてください〜。\nこれはMiniMax H3でチャレンジしたテキストモーション動画です！（ちょっと失敗）",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 16,
      "views": 612,
      "snapshotAt": "2026-09-12T07:53:30.083Z"
    },
    "approvedAt": "2026-09-12T07:53:30.083Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098894429134397721",
    "title": "威士忌淋上香草冰淇淋",
    "titleEn": "Whiskey Poured Over Vanilla Ice Cream",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发动画女孩把威士忌淋在一碗香草冰淇淋上，举起成品朝镜头展示，动作与餐桌环境在短镜头里保持稳定。 原生视频规格为 5 秒 · 1152×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired anime girl pours whiskey over vanilla ice cream and presents the finished bowl to camera, keeping her motion and dining-room setting coherent. Native-video output: 5s · 1152×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yonaguniUDON/status/2098894429134397721",
    "sourceLabel": "X 原帖 · @yonaguniUDON",
    "author": "蛸壺_AIラクルゥ開発",
    "publishedAt": "2026-09-12T22:00:02.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098894429134397721.mp4",
    "posterUrl": "/posters/x/2098894429134397721.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1152×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "commercial"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "バニラアイス + ウイスキー\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 10,
      "views": 68,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098894423107002425",
    "title": "黄昏河堤上的抒情蒙太奇",
    "titleEn": "A Lyrical Montage Along the River at Dusk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画少女从住宅街走到黄昏河堤，天空、电线与奔跑镜头随音乐交替，形成一段带青春气息的竖屏蒙太奇。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime girl moves from a residential street to a riverside at dusk as sky, power lines, and running shots alternate with the music in a youthful vertical montage. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TSuga5795/status/2098894423107002425",
    "sourceLabel": "X 原帖 · @TSuga5795",
    "author": "tsuga@AIイラスト企画の人",
    "publishedAt": "2026-09-12T22:00:00.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098894423107002425.mp4",
    "posterUrl": "/posters/x/2098894423107002425.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "皆さんチンチコール✋✨💕\n\n今日は朝から晩まで副業バイトに行ってきます🏩\n\n動画生成のクレジットが半端に余ってしまったので、MiniMax H3 MAX で叙情的＋ちょいエモなモンタージュ映像を作ってみました🎥✨\n\nBGMもMiniMax任せでしたが、SUNOとかでBGMつくって複数の動画と組み合わせて60秒くらいに出来たらもっとエモいのが作れそうですね😆💕",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 27,
      "views": 661,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098894165841178848",
    "title": "瀑布秘境里的巫师晨练",
    "titleEn": "The Wizard Leads Morning Exercises by the Falls",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白发巫师在瀑布秘境中带领一群角色做伸展与跳跃，镜头从领操者推向队伍，再以人物合影式构图收束。 原生视频规格为 18 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired wizard leads a group through stretches and jumps in a waterfall sanctuary, moving from the coach to the ensemble before closing in a portrait-like formation. Native-video output: 18s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/joesparks/status/2098894165841178848",
    "sourceLabel": "X 原帖 · @joesparks",
    "author": "Joe Sparks",
    "publishedAt": "2026-09-12T21:58:59.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098894165841178848.mp4",
    "posterUrl": "/posters/x/2098894165841178848.jpg",
    "duration": 18,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Morning exercises with the Wizard \n(Coach Wyn in the MiniMax H3 Singularity)",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 105,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098890685453844881",
    "title": "越说别按越想按的红色按钮",
    "titleEn": "The Red Button Everyone Is Told Not to Press",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "简笔人物面对写着“别按”的红色按钮犹豫，按下后彩色拟声词和图形同时爆发，最后又回到按钮前。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A line-drawn character hesitates before a red “do not press” button, triggers a burst of colorful sound effects and graphics, then returns to the tempting button. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2098890685453844881",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-09-12T21:45:09.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098890685453844881.mp4",
    "posterUrl": "/posters/x/2098890685453844881.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「押すなボタン」\n押すなと言われたらつい押したくなりますよね！\n#OiiOii のMinimaxH3 Max Tauboで動画作成\nプロンプトはリプ欄に👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 87,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098886746846945356",
    "title": "蒸汽列车驶过低多边形城堡",
    "titleEn": "A Steam Train Passes a Low-Poly Castle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "黑色蒸汽列车从车轮机械特写逐步拉远，牵引红色车厢穿过拱桥，远处的中世纪城堡与河流同时进入全景。 原生视频规格为 15 秒 · 480×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A black steam locomotive pulls back from a close-up of its working wheels, hauling red carriages across an arched bridge as a medieval castle and river enter the wide view. Native-video output: 15s · 480×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A 3D stylized animation begins with a tight close-up on the gold-spoked wheels and connecting rods of a black steam locomotive accented with metallic hoops. As the camera smoothly zooms out and pulls back, it reveals the entire train puffing spherical gray smoke clouds from its chimney while pulling red passenger cars along a railway track. The track runs across an elevated arched bridge alongside a grand medieval stone castle complete with corner towers and battlements, all set within a low-poly environment featuring scattered evergreen trees and a quiet river.",
    "promptSourceUrl": "https://x.com/TaliaAariz/status/2098886746846945356",
    "sourceUrl": "https://x.com/TaliaAariz/status/2098886746846945356",
    "sourceLabel": "X 原帖 · @TaliaAariz",
    "author": "Talia",
    "publishedAt": "2026-09-12T21:29:30.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098886746846945356.mp4",
    "posterUrl": "/posters/x/2098886746846945356.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "480×480",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI turning a simple train scene into a cinematic fantasy world.\nThe scale, details, and atmosphere make this feel like a scene straight out of an animated film.\n\nCreated with MiniMax H3 @atlas_cloud_ai\n\nPrompt:\n\nA 3D stylized animation begins with a tight close-up on the gold-spoked wheels and connecting rods of a black steam locomotive accented with metallic hoops. As the camera smoothly zooms out and pulls back, it reveals the entire train puffing spherical gray smoke clouds from its chimney while pulling red passenger cars along a railway track. The track runs across an elevated arched bridge alongside a grand medieval stone castle complete with corner towers and battlements, all set within a low-poly environment featuring scattered evergreen trees and a quiet river.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 4,
      "views": 102,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098831194385424413",
    "title": "蓝发机械少女穿行城市街巷",
    "titleEn": "A Teal-Haired Cyborg Walks the City Streets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "蓝绿色头发的机械少女在城市街巷中持续转身与行走，用较长镜头展示 Motion Context 对角色外观和动作连续性的保持。 原生视频规格为 16 秒 · 832×1216。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A teal-haired cyborg girl turns and walks through an urban street in a longer take that demonstrates how Motion Context preserves character appearance and motion. Native-video output: 16s · 832×1216. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/core_tan/status/2098831194385424413",
    "sourceLabel": "X 原帖 · @core_tan",
    "author": "CALMDUST aka ちゃんこあ",
    "publishedAt": "2026-09-12T17:48:45.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098831194385424413.mp4",
    "posterUrl": "/posters/x/2098831194385424413.jpg",
    "duration": 16,
    "aspectRatio": "portrait",
    "resolution": "832×1216",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 WebUI dev1.6\n\nひと月前には難しいと判断したNarive Ref2VAのMotion Context機能、最適化して余裕が出来たので実用性の範疇な負荷で実装した（軽くは無い\n\nメリットは16秒超のMC動画でキャラを保てる＋時間指定のプロンプトを書けること。設定を詰める必要はあるけど↓の生成時間は197s",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 477,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098828633825501569",
    "title": "高分辨率角色的全身与肖像动作",
    "titleEn": "Full-Body and Portrait Motion at High Resolution",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "带角的红黑服装角色在白底空间完成全身转身、手势与肖像特写，重点观察高分辨率 Ref2VA 的身份保持。 原生视频规格为 12 秒 · 1248×1856。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A horned character in red and black performs turns and gestures against white before moving into portrait framing, emphasizing identity retention in high-resolution Ref2VA. Native-video output: 12s · 1248×1856. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SmallKino_/status/2098828633825501569",
    "sourceLabel": "X 原帖 · @SmallKino_",
    "author": "SmallKino",
    "publishedAt": "2026-09-12T17:38:35.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098828633825501569.mp4",
    "posterUrl": "/posters/x/2098828633825501569.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "1248×1856",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Mireia-san.\nFull body and portrait. I'm very surprised that MiniMax H3 can actually handle 2.2 MP resolution with its ref2va model. With 2nd pass latent upscale on 8-step turbo lora it takes only 20 min to complete each clip.\n\n#戦姫サバイバルサガ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 236,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098826399218127149",
    "title": "哥特舞台上的三十秒女团MV",
    "titleEn": "A Thirty-Second Girl-Group MV on a Gothic Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑裙歌者在烛光与深红帷幕包围的舞台上演唱，近景、群像与移动镜头组成一支完整的三十秒音乐短片。 原生视频规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A singer in black performs amid candlelight and deep-red curtains as close-ups, ensemble shots, and moving camera work form a complete thirty-second music video. Native-video output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/2D9iIKALlXJz2Eg/status/2098826399218127149",
    "sourceLabel": "X 原帖 · @2D9iIKALlXJz2Eg",
    "author": "くまたろう",
    "publishedAt": "2026-09-12T17:29:42.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098826399218127149.mp4",
    "posterUrl": "/posters/x/2098826399218127149.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "dark"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "初めてAIでMVを作ってみました🎬✨\nMiniMax H3で映像、MiniMax Music 3で曲を作って30秒MVに挑戦。\nまだまだ改善点はあるけど、ひとまず完成！\n#AI動画 #MiniMaxH3 #MiniMaxMusic #AIミュージック",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 86,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098826050969252244",
    "title": "霓虹城市中的一分钟动画MV",
    "titleEn": "A One-Minute Anime MV Across a Neon City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红发动画歌者在夜间城市、舞台和发光几何体之间切换，连续演唱与运镜把多个片段组织成一分钟MV。 原生视频规格为 58 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-haired anime singer moves between a night city, stage, and glowing geometry, with continuous vocals and camera movement joining the clips into a one-minute MV. Native-video output: 58s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ototoi/status/2098826050969252244",
    "sourceLabel": "X 原帖 · @ototoi",
    "author": "Two days ago",
    "publishedAt": "2026-09-12T17:28:19.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098826050969252244.mp4",
    "posterUrl": "/posters/x/2098826050969252244.jpg",
    "duration": 58,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Codex + MiniMax H3 + SUNO によるMV風動画生成テスト。\n最終的には４分になるらしいのだけど冒頭１分までのカット。\n途中かなり無理がある場面転換があって乱れるのだけどおいおい修正する。\n「黎明シグナル」",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 97,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098824502528356644",
    "title": "橘猫占领发光主机桌面",
    "titleEn": "An Orange Cat Takes Over a Glowing PC Desk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橘猫从电脑机箱旁走到键盘和鼠标之间，最后趴在显卡前睡着，写实灯光把失败实验变成一段生活小品。 原生视频规格为 13 秒 · 1056×992。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An orange cat moves from the PC tower to the keyboard and mouse before falling asleep beside the graphics card, turning the experiment into a photoreal domestic vignette. Native-video output: 13s · 1056×992. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DakkyBear/status/2098824502528356644",
    "sourceLabel": "X 原帖 · @DakkyBear",
    "author": "dakky.eth 🍌",
    "publishedAt": "2026-09-12T17:22:10.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098824502528356644.mp4",
    "posterUrl": "/posters/x/2098824502528356644.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "1056×992",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An attempt of making a semi-real Hashcat. \nObviously failed. \n\nAnd then they say Minimax H3 is good.😁",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 8,
      "views": 886,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098822632926015957",
    "title": "废墟中认真跳舞的白发少女",
    "titleEn": "A White-Haired Girl Dances Through the Ruins",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白发少女在城市废墟中从脚步特写进入连续舞蹈，蓝色发光球体与快速转身共同强化动作节奏。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired girl moves from a close-up of her steps into a continuous dance through urban ruins, with a glowing blue orb and fast turns carrying the rhythm. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Futureman_3/status/2098822632926015957",
    "sourceLabel": "X 原帖 · @Futureman_3",
    "author": "⬣ Futureman 小説とフィギュア 頑張りま～～～す‼",
    "publishedAt": "2026-09-12T17:14:44.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098822632926015957.mp4",
    "posterUrl": "/posters/x/2098822632926015957.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#FuturePunks映画化への道\nどうしてもアホ毛がうまく生成できなかった関係で、RefModsの設定を丸一日やっていました。\nなんとか、いい感じになって来た。\nサニーが一生懸命踊っている姿を見て、ここまでの苦労も報われた感じがしました。\n\nつづく‼️\nばば〜〜〜ん‼️\n\n#MiniMaxH3 #ai動画",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 13,
      "views": 398,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098802133201997933",
    "title": "摄影棚里的四套人物造型",
    "titleEn": "Four Character Looks Inside a Photo Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一位女性在白色摄影棚中切换耳机、机能服与多种坐姿，拼贴开场随后进入稳定的人像展示。 原生视频规格为 7 秒 · 2160×2892。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same woman cycles through headphones, utility clothing, and several seated poses in a white studio, moving from a collage opening into steady portrait presentation. Native-video output: 7s · 2160×2892. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Arden_fyi/status/2098802133201997933",
    "sourceLabel": "X 原帖 · @Arden_fyi",
    "author": "Arpad Spodni 🇺🇦",
    "publishedAt": "2026-09-12T15:53:17.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098802133201997933.mp4",
    "posterUrl": "/posters/x/2098802133201997933.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "2160×2892",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "fashion",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 nailed it pretty good for the fraction of the cost of seedance",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 69,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098800747781513345",
    "title": "后院自拍的插画风格迁移",
    "titleEn": "An Illustrated Style Transfer on a Backyard Selfie",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴眼镜的男子在后院对镜头说话，整段保持手持自拍视频构图，同时呈现明显的插画化面部与背景纹理。 原生视频规格为 8 秒 · 672×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man with glasses speaks to camera in a backyard, retaining the handheld selfie composition while his face and surroundings take on a strong illustrated texture. Native-video output: 8s · 672×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sebastienrama/status/2098800747781513345",
    "sourceLabel": "X 原帖 · @sebastienrama",
    "author": "Sébastien Rama  (Dodemodexter)",
    "publishedAt": "2026-09-12T15:47:46.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098800747781513345.mp4",
    "posterUrl": "/posters/x/2098800747781513345.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "672×384",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "ugc"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Je teste les transferts de style avec #MinimaxH3 et un Lora prévu à cet effet. ça ne fonctionne pas tout le temps de manière heureuse, mais globalement, je pense qu'avec un prompt \"pimpé\" ça devrait le faire sur certaines séquences / style. En revanche, lipsync ne fonctionne pas",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 4,
      "views": 266,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098800452586598897",
    "title": "战后货轮甲板上的解除戒备",
    "titleEn": "Standing Down on a Cargo Deck After the War",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "持枪女子从货轮甲板起身推进、贴近掩体瞄准，发现幸存船员正在焊接后缓缓放下武器，以一句“We’re home”结束。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An armed woman rises and advances across a cargo deck, aims around cover, then lowers her weapon after finding surviving crew at work, ending on “We’re home.” Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photorealistic image-to-video extension based on <Picture 1>. Preserve the exact same adult female character throughout: same facial structure, short hairstyle, outfit construction, body proportions, mature age impression, weapon, cargo-ship deck environment, and period texture. The whole sequence stays inside one connected ship space only: the weathered cargo deck, the container-side passage, and the visible wheelhouse zone. No harbor wide shot, no new location, no unrelated crowd. The tone is dangerous post-war allure with realism, restraint, sea wind, rust, salt mist, gulls, cold daylight, and small warm practical lights. She never poses; she stays in motion through one chain: angled sitting, rising, lifting the gun, advancing, edging along cover, stopping, aiming, confirming, lowering the gun, and looking back. Sensuality comes only from control, shoulder-neck line, breathing shifts, eye pressure, mouth tension, and her fingers on the weapon. At 00:00.000-00:02.500, she sits at an angle against the rusted ship edge. One leg extends naturally, one hand holds the rim, the other rests loosely on the gun. The camera starts low on the high-heeled ankle boots and leg line, glides past the hand on the rim, then rises to her side face and lip line. She looks only toward the sea. Her upper lids are slightly lowered, gaze stable but alive, blink rate low, breathing shallow, lips slightly parted, mouth corners neutral. Wind lifts the short front strands and hair near the ear. [Shot 2] At 00:02.500, she rises with one hand pushing off the deck while the other lifts the gun, the stock brushing the outer thigh. The camera stays close and rises with her, bringing out collarbone, shoulder line, side neck, jawline, and outer eye corner. The rise is controlled, as if fatigue remains but discipline is stronger. Her expression shifts: eyelids lift slightly, eyes move from distance to lock, the brow center gathers for a moment, lips close, the mouth line firms, and the jaw tightens lightly. A small swallow moves in the throat. In a very low soft voice, almost inside the breath, she says: <d>[English] Easy...</d> [Shot 3] At 00:04.800, a harsh cold reflection flashes from the distant container-side passage and a muted metallic hit follows. She turns immediately and advances with the gun. She does not run; she moves in a pressured fast walk with stable center of gravity, precise steps, and the gun carried low beside the thigh. The camera follows from side-rear as rusted deck textures, cables, and container edges drift behind her. Her face sharpens here: eyes lock forward, eye movement range decreases, focus intensifies, the outer eyes tighten, the brow center presses slightly, breathing shortens and quiets, lips stay closed, and the jawline hardens. [Shot 4] At 00:07.800, she skims along the container edge, stops hard, turns back through the shoulder, and raises the gun. First show the readable medium-close action, then tighten quickly into the most intimate close-up of the short. She leans into the sight line. The camera catches lashes, eye tail, side nose line, mouth corner, and the near-metal intimacy of her breath against the weapon. Her eyes do not widen; they narrow slightly, upper lid pressing down in concentration. The lower lid gathers a little, lips remain closed, one mouth corner holds a tiny natural tension, and the nostrils widen once then settle. The weapon raise is fast, sight alignment steady, and the trigger finger rests along the outside edge rather than firing. [Shot 5] At 00:10.300, as she turns past the obstruction, the gun is aimed not at an enemy but at two surviving crew members welding the deck. Welding sparks scatter, electric flicker pulses, a warm wheelhouse light turns on, and a strip of white cloth near the railing flutters in the wind. Let her subjective line of sight reveal the reversal first, then give a very short half-profile reaction. Her eyes stay guarded for half a beat, then clearly release: tension around the eyes falls slightly, upper lids lift back a little, the brow center opens, lips unseal as if a silent breath leaves, and the jawline softens from combat hardness. This is not cheerful relief, only post-threat decompression. [Shot 6] At 00:12.800, she slowly lowers the gun and lets her fingers loosen and slide along the gun body. The camera comes from behind her shoulder in a half-orbit to her side front, with warm light grazing her face. She looks back toward the sea and gives a very restrained slight smile, no teeth, only one side of the mouth lifting a little. For the first time her eyes hold softness and belonging. She says quietly: <d>[English] We’re home.</d> The line must not sound sweet or slogan-like; “We’re” is light, “home” lands lower and firmer. Her brow center is relaxed now, some fatigue remains at the eye corners, there is the slightest moisture without tears, blinking slows a little, and the mouth corner lifts then controls itself again. End mature, steady, tender, and still carrying the after-heat of war.\n\noverall_soundscape: The soundscape is led by environment throughout: sea wind, gulls, light ship-metal creaks, heels on deck, distant electric crackle, welding sparks, low mechanical hum, and a final low ship horn. Emphasize wind, metal resonance, gulls, and her shallow breathing in the opening; hand on deck, cloth movement, gun lift, swallow, and the soft breath-close line <d>[English] Easy...</d> during the rise. Then bring in the muted metal hit, close footfalls on steel, weapon handling, cloth friction, and reduced breath as she advances. Tighten the sound field around breath and metal near the weapon during aiming, then reveal the welding reality through spark hiss, low electrical buzz, warm light switching on, white cloth in the wind, and deck machinery hum. Her final line <d>[English] We’re home.</d> should be low, close, tired, and steady, followed by wind, gulls, machinery, and the distant low horn.\n\nnon_diegetic_music: Use only a very light tension bed. Keep it minimal, cool, and modern, never epic or sentimental. Support the first half with low suspended texture, tighten slightly when the reflection and metal sound trigger her advance, thin it during the aiming close-up so breath and steel dominate, then ease the tension on the welding reveal without becoming warm or triumphant. Let the final line carry only a faint brief emotional release.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2098800452586598897",
    "sourceUrl": "https://x.com/ou_zhen599/status/2098800452586598897",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-12T15:46:36.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098800452586598897.mp4",
    "posterUrl": "/posters/x/2098800452586598897.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI      \n\n🎬Image-to-Video  \n📷“A Gentle Reward”\nYou were waiting for her final shot, but what you got wasn’t an enemy—it was proof that peace had truly arrived.\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photorealistic image-to-video extension based on <Picture 1>. Preserve the exact same adult female character throughout: same facial structure, short hairstyle, outfit construction, body proportions, mature age impression, weapon, cargo-ship deck environment, and period texture. The whole sequence stays inside one connected ship space only: the weathered cargo deck, the container-side passage, and the visible wheelhouse zone. No harbor wide shot, no new location, no unrelated crowd. The tone is dangerous post-war allure with realism, restraint, sea wind, rust, salt mist, gulls, cold daylight, and small warm practical lights. She never poses; she stays in motion through one chain: angled sitting, rising, lifting the gun, advancing, edging along cover, stopping, aiming, confirming, lowering the gun, and looking back. Sensuality comes only from control, shoulder-neck line, breathing shifts, eye pressure, mouth tension, and her fingers on the weapon. At 00:00.000-00:02.500, she sits at an angle against the rusted ship edge. One leg extends naturally, one hand holds the rim, the other rests loosely on the gun. The camera starts low on the high-heeled ankle boots and leg line, glides past the hand on the rim, then rises to her side face and lip line. She looks only toward the sea. Her upper lids are slightly lowered, gaze stable but alive, blink rate low, breathing shallow, lips slightly parted, mouth corners neutral. Wind lifts the short front strands and hair near the ear. [Shot 2] At 00:02.500, she rises with one hand pushing off the deck while the other lifts the gun, the stock brushing the outer thigh. The camera stays close and rises with her, bringing out collarbone, shoulder line, side neck, jawline, and outer eye corner. The rise is controlled, as if fatigue remains but discipline is stronger. Her expression shifts: eyelids lift slightly, eyes move from distance to lock, the brow center gathers for a moment, lips close, the mouth line firms, and the jaw tightens lightly. A small swallow moves in the throat. In a very low soft voice, almost inside the breath, she says: <d>[English] Easy...</d> [Shot 3] At 00:04.800, a harsh cold reflection flashes from the distant container-side passage and a muted metallic hit follows. She turns immediately and advances with the gun. She does not run; she moves in a pressured fast walk with stable center of gravity, precise steps, and the gun carried low beside the thigh. The camera follows from side-rear as rusted deck textures, cables, and container edges drift behind her. Her face sharpens here: eyes lock forward, eye movement range decreases, focus intensifies, the outer eyes tighten, the brow center presses slightly, breathing shortens and quiets, lips stay closed, and the jawline hardens. [Shot 4] At 00:07.800, she skims along the container edge, stops hard, turns back through the shoulder, and raises the gun. First show the readable medium-close action, then tighten quickly into the most intimate close-up of the short. She leans into the sight line. The camera catches lashes, eye tail, side nose line, mouth corner, and the near-metal intimacy of her breath against the weapon. Her eyes do not widen; they narrow slightly, upper lid pressing down in concentration. The lower lid gathers a little, lips remain closed, one mouth corner holds a tiny natural tension, and the nostrils widen once then settle. The weapon raise is fast, sight alignment steady, and the trigger finger rests along the outside edge rather than firing. [Shot 5] At 00:10.300, as she turns past the obstruction, the gun is aimed not at an enemy but at two surviving crew members welding the deck. Welding sparks scatter, electric flicker pulses, a warm wheelhouse light turns on, and a strip of white cloth near the railing flutters in the wind. Let her subjective line of sight reveal the reversal first, then give a very short half-profile reaction. Her eyes stay guarded for half a beat, then clearly release: tension around the eyes falls slightly, upper lids lift back a little, the brow center opens, lips unseal as if a silent breath leaves, and the jawline softens from combat hardness. This is not cheerful relief, only post-threat decompression. [Shot 6] At 00:12.800, she slowly lowers the gun and lets her fingers loosen and slide along the gun body. The camera comes from behind her shoulder in a half-orbit to her side front, with warm light grazing her face. She looks back toward the sea and gives a very restrained slight smile, no teeth, only one side of the mouth lifting a little. For the first time her eyes hold softness and belonging. She says quietly: <d>[English] We’re home.</d> The line must not sound sweet or slogan-like; “We’re” is light, “home” lands lower and firmer. Her brow center is relaxed now, some fatigue remains at the eye corners, there is the slightest moisture without tears, blinking slows a little, and the mouth corner lifts then controls itself again. End mature, steady, tender, and still carrying the after-heat of war.\n\noverall_soundscape: The soundscape is led by environment throughout: sea wind, gulls, light ship-metal creaks, heels on deck, distant electric crackle, welding sparks, low mechanical hum, and a final low ship horn. Emphasize wind, metal resonance, gulls, and her shallow breathing in the opening; hand on deck, cloth movement, gun lift, swallow, and the soft breath-close line <d>[English] Easy...</d> during the rise. Then bring in the muted metal hit, close footfalls on steel, weapon handling, cloth friction, and reduced breath as she advances. Tighten the sound field around breath and metal near the weapon during aiming, then reveal the welding reality through spark hiss, low electrical buzz, warm light switching on, white cloth in the wind, and deck machinery hum. Her final line <d>[English] We’re home.</d> should be low, close, tired, and steady, followed by wind, gulls, machinery, and the distant low horn.\n\nnon_diegetic_music: Use only a very light tension bed. Keep it minimal, cool, and modern, never epic or sentimental. Support the first half with low suspended texture, tighten slightly when the reflection and metal sound trigger her advance, thin it during the aiming close-up so breath and steel dominate, then ease the tension on the welding reveal without becoming warm or triumphant. Let the final line carry only a faint brief emotional release.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 18,
      "views": 879,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098799127404712133",
    "title": "粉色爱心前的角色眨眼",
    "titleEn": "A Character Winks Before a Pink Heart",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙发马耳角色在粉色爱心背景前保持站姿，完成轻微身体摆动与一次俏皮眨眼。 原生视频规格为 5 秒 · 480×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An orange-haired horse-eared character holds a standing pose before a pink heart, adding a slight body sway and one playful wink. Native-video output: 5s · 480×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Senna_Satou/status/2098799127404712133",
    "sourceLabel": "X 原帖 · @Senna_Satou",
    "author": "Tsuge",
    "publishedAt": "2026-09-12T15:41:20.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098799127404712133.mp4",
    "posterUrl": "/posters/x/2098799127404712133.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "480×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "地雷系スズカをMiniMax H3で動かしてみた。\nウインクかわいい。",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 51,
      "views": 698,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098796845115871459",
    "title": "雕像在花园里苏醒起舞",
    "titleEn": "A Garden Statue Comes to Life and Dances",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "石雕女子从繁花庭院的静止姿态中苏醒，镜头环绕时裙摆和手臂开始旋转，仍保留石材质感。 原生视频规格为 8 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A stone woman awakens from stillness in a flower-filled garden and begins to turn her arms and dress as the camera circles, retaining the sculpted material. Native-video output: 8s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/00Bohr/status/2098796845115871459",
    "sourceLabel": "X 原帖 · @00Bohr",
    "author": "José Luis",
    "publishedAt": "2026-09-12T15:32:16.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098796845115871459.mp4",
    "posterUrl": "/posters/x/2098796845115871459.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "En local, en mi ComfyUi con Minimax H3.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 0,
      "views": 127,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098772746805371228",
    "title": "玩偶、猫与上班族的五分钟动画",
    "titleEn": "A Five-Minute Story About a Doll, a Cat, and an Office Worker",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手绘风玩偶与猫陪伴中年男子度过家中日常，多个室内场景、对白气泡和关系变化被串成一段完整长篇动画。 原生视频规格为 318 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hand-drawn doll and cat accompany a middle-aged man through domestic life, with dialogue bubbles and relationship changes unfolding across a complete five-minute animation. Native-video output: 318s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mikumiku_aloha/status/2098772746805371228",
    "sourceLabel": "X 原帖 · @mikumiku_aloha",
    "author": "mikumiku_aloha",
    "publishedAt": "2026-09-12T13:56:30.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098772746805371228.mp4",
    "posterUrl": "/posters/x/2098772746805371228.jpg",
    "duration": 318,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "daily-life",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIslop \n自己評価では物語のちゃんと伝わるアニメーションをAstra使って生成出来たと思う。\n\n笑わば笑え。\n\nComfyUI上のMiniMax H3が動く環境とIrodori-TTSが動作する環境を用意してChatGPT workでAstraに生成させた。\n文字の指示だけで編集ツールなど一切操作していない。\n\nAstraに生成させたChatGPTにこんな風な動画を生成させるためのノウハウzip\n\nhttps://xxup.org/b4wJ5.zip",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 171,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098771852370014684",
    "title": "废墟中的巨剑战士与岩甲怪兽",
    "titleEn": "A Greatsword Fighter Faces an Armored Beast",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "巨剑战士冲过废墟迎战发绿光的岩甲怪兽，近距离劈砍、撞击与碎石把游戏战斗感压进十五秒。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A greatsword fighter charges through ruins at a green-glowing armored beast, packing close-range slashes, impacts, and flying rubble into fifteen seconds of game-like combat. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YaReYaRu30Life/status/2098771852370014684",
    "sourceLabel": "X 原帖 · @YaReYaRu30Life",
    "author": "ヤレヤル（AI動画/AI画像/AI音楽）",
    "publishedAt": "2026-09-12T13:52:57.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098771852370014684.mp4",
    "posterUrl": "/posters/x/2098771852370014684.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ゲーム系だとMiniMaxH3が\n結構いい気がする。\n\nエフェクトの入り方に\nHailuo独特の粒感がでるので\nそれが味になる場合もある\n\n@Hailuo_AI #MiniMaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 11,
      "views": 1198,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098771179808940415",
    "title": "白发战士追逐发光球体",
    "titleEn": "A White-Haired Fighter Chases a Glowing Orb",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "白发战士在破败城市里追踪蓝色发光球体，从奔跑、拔枪到近景戒备形成连续的角色冒险片段。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A white-haired fighter tracks a glowing blue orb through a ruined city, moving from a run to drawing a weapon and a guarded close-up in one continuous character vignette. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a curiosity-driven character vignette built from 10–15 short cuts around the character from Image1.\n\nPreserve the character’s identity, visual style and overall design.  Infer its nature and world directly from Image1.\n\nOpen with an immediately intriguing visual situation that creates a simple unanswered question. Build the following cuts around discovery, reaction, progression and small consequences. Each cut should introduce a new visual idea while clearly advancing the same underlying situation.\n\nLet the character reveal itself through behavior rather than exposition. Use its movement, instincts, habits, abilities, limitations, relationships and interaction with the surrounding world to make the sequence increasingly interesting.\n\nKeep the narrative simple enough to understand without dialogue. Create visual cause and effect between cuts. Allow details introduced early to gain meaning later. Build toward a clear reveal, reversal, transformation, emotional beat or satisfying visual payoff near the end.\n\nAvoid generic daily routines, disconnected montage imagery and repetitive actions. Do not force familiar human behavior onto the character or world. Let the situations emerge naturally from the reference.\n\nMake every few seconds visually distinct through changes in scale, framing, environment, movement, tension and information. Maintain strong continuity so the viewer wants to see what happens next.\n\nUse objective third-person cinematic observation with natural handheld imperfection and varied shot sizes. Keep the camera readable, responsive and visually motivated.\n\nUse only clean diegetic sound. No dialogue, narration, BGM, subtitles or title cards.",
    "promptSourceUrl": "https://x.com/aimikoda/status/2098771179808940415",
    "sourceUrl": "https://x.com/aimikoda/status/2098771179808940415",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-09-12T13:50:17.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098771179808940415.mp4",
    "posterUrl": "/posters/x/2098771179808940415.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Curiosity driven Character Vignette\n\nWhat if a single character image could become a story you actually want to keep watching?\n\nThis prompt turns any character reference into a curiosity-driven vignette, using 10-15 connected moments that reveal the character and their world through visual cause and effect, building toward a satisfying payoff.\n\nYou can use it with almost any video model and any character. I used MiniMax H3 for this example. You can try it with Seedance 2.5 too.\n\nPrompt:\nCreate a curiosity-driven character vignette built from 10–15 short cuts around the character from Image1.\n\nPreserve the character’s identity, visual style and overall design.  Infer its nature and world directly from Image1.\n\nOpen with an immediately intriguing visual situation that creates a simple unanswered question. Build the following cuts around discovery, reaction, progression and small consequences. Each cut should introduce a new visual idea while clearly advancing the same underlying situation.\n\nLet the character reveal itself through behavior rather than exposition. Use its movement, instincts, habits, abilities, limitations, relationships and interaction with the surrounding world to make the sequence increasingly interesting.\n\nKeep the narrative simple enough to understand without dialogue. Create visual cause and effect between cuts. Allow details introduced early to gain meaning later. Build toward a clear reveal, reversal, transformation, emotional beat or satisfying visual payoff near the end.\n\nAvoid generic daily routines, disconnected montage imagery and repetitive actions. Do not force familiar human behavior onto the character or world. Let the situations emerge naturally from the reference.\n\nMake every few seconds visually distinct through changes in scale, framing, environment, movement, tension and information. Maintain strong continuity so the viewer wants to see what happens next.\n\nUse objective third-person cinematic observation with natural handheld imperfection and varied shot sizes. Keep the camera readable, responsive and visually motivated.\n\nUse only clean diegetic sound. No dialogue, narration, BGM, subtitles or title cards.",
    "engagement": {
      "replies": 12,
      "reposts": 4,
      "likes": 74,
      "views": 4610,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098768802481873294",
    "title": "原生二十步与四步Turbo并排对比",
    "titleEn": "Native Twenty Steps Versus Four-Step Turbo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一位坐在秋日公园长椅上的人物以左右双栏同步播放，对照原生二十步与四步 Turbo LoRA 的动作和画质。 原生视频规格为 16 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same person on an autumn park bench plays in synchronized side-by-side panels to compare native twenty-step output with a four-step Turbo LoRA render. Native-video output: 16s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/safa_dayo/status/2098768802481873294",
    "sourceLabel": "X 原帖 · @safa_dayo",
    "author": "サファ【AIイラスト多め】",
    "publishedAt": "2026-09-12T13:40:50.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098768802481873294.mp4",
    "posterUrl": "/posters/x/2098768802481873294.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "通常のMiniMax H3と、H3のTurbo LoRAを使った4ステップ生成を比較してみた👀",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 6,
      "views": 1894,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098767994750955881",
    "title": "废弃神殿里的双生精灵对峙",
    "titleEn": "Twin Elf Warriors Confront Each Other in a Ruined Temple",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "紫瞳精灵持剑穿过潮湿神殿，黑衣绿瞳的同貌对手从石柱阴影中出现，双方以近距离挥剑和对峙收尾。 原生视频规格为 14 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A violet-eyed elf advances through a damp ruined temple until a green-eyed double in black steps from behind a pillar, ending in a close sword strike and standoff. Native-video output: 14s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photoreal dark-fantasy image-to-video sequence based on <Picture 1>. Preserve the exact same female elf warrior throughout: same face, elf ears, facial structure, hairstyle, gray-purple cloak, silver-steel armor, fitted leather battlewear, sword, body proportions, and realistic texture. The entire sequence stays inside one continuous abandoned temple interior with damp stone steps, broken archway, rough pillars, dying candles, moss, scattered leaves, wet walls, faint dust, and cold moist air. No exterior cut, no extra characters, no flashy fantasy VFX, no face drift, no identity drift. The main elf warrior advances from the lower steps toward the deeper archway. The dark counterpart always remains hidden behind a rear-side pillar and emerges physically from that same shadow, never teleporting. The tone is cold, restrained, dangerous, mysterious, and subtly sensual through presence, gaze, and motion, not posing. The main warrior’s magic exists only inside the eye: her violet eye subtly focuses inward, never emitting outward effects. The dark counterpart has the same face and species but wears a black cloak, matte dark armor, sharper shoulders, and carries cold green eyes that stay deep, stable, and clearly distinct from the warrior’s violet gaze. [Shot 2] At 00:02.200, the camera makes a low smooth push over wet steps, leaves, candle stubs, and cracked stone, then settles below and in front of the main warrior in a crouched position. One knee presses to the step, left hand braces on stone, right hand holds the sword, and the cloak falls with real weight. She is almost motionless, like a predator storing force. Her head is slightly lowered, gaze first resting on the step near her hand, lips closed, mouth corners flat, chin tucked, brow faintly gathered, and breathing nearly invisible. Only the environment moves: candle tremor, damp reflections, drifting dust. [Shot 3] At 00:04.100, cut to an intimate eye close-up. She catches that something is wrong. After half a beat of stillness, her eyes lift slowly. The violet eye does not flare; it only gathers inward with a subtle tightening deep inside the pupil. The emotional path is stillness to detection to lock. The lower eyelid tightens, breath shortens by one measure, and the lip line firms. Stay close on one eye, nose bridge, lips, and jawline as candlelight brushes her side contour. One side of her mouth lifts by the smallest amount, as if confirming a target. [Shot 4] At 00:05.400, she rises from the crouch in one swift efficient motion. The cloak swings backward with believable inertia, and the sword comes from low position into a forward guard line. She locks upright like a hunter ending concealment. Her jaw tightens, lips part only slightly to change breathing, brow compresses, and gaze hardens. Only physical feedback appears: cloak weight, slight dust disturbance, candle response. [Shot 5] At 00:06.900, she pushes quickly along the stone steps toward the deeper archway. Passing a pillar, her body skims close to the wall, the cloak brushes stone, candlelight skims her side face, and the sword stays in a ready attack-defense line. This is controlled advance, not panic. Steps stay tight, shoulders stable, head motion minimal, gaze fixed on one search line. Her eyelids narrow, breath shortens, and mouth corners flatten. At 00:08.500, she suddenly stops by half a step as if the sensed target in front has vanished. She instantly uses the pillar to half-turn, bringing the sword back toward her chest in defense. The shift is active advance to professional alert. Her eyes contract first, then broaden their search, lower lids tighten, nostrils draw slightly inward, and the lip line presses thinner. The environment feels wrong through unnatural stillness: candle motion reduces, dust slows, and the space becomes too quiet. [Shot 6] At 00:10.000, place the camera over her shoulder with part of her hood and sword edge in foreground and empty space ahead. The threat does not appear in front. Then, behind the rear pillar, the edge of a black cloak slowly slips into candlelight, followed by one black boot stepping out. The main warrior has not fully turned yet, but her back, neck, and shoulders are already tight. The dark counterpart emerges slowly on purpose, simply entering light from real shadow. The camera glides sideways and brings her fully into view: same face and species, but in black battlewear, matte black armor, sharp shoulder plates, and a long black cloak. She is darker, heavier, more severe. Her cold green eyes are deep and controlled, never neon, and one side of her mouth carries a faint wicked smile. She is not furious; she is already in control. [Shot 7] At 00:12.900, the main warrior whips around and slashes back in one immediate, accurate defensive strike. The blade cuts close past the camera line. The dark counterpart does not hurry to block. She simply steps half a step forward into the attack line and tilts her head enough for the blade to skim past. The main warrior’s intensity spikes without becoming theatrical: brow tight, nostrils slightly open, lips parted only a little from exertion, jaw rigid. The dark counterpart’s smile stays and grows colder during the head tilt, as if enjoying the failed strike. Keep the sword realistic with only restrained metallic reflection and real cloak inertia. [Shot 8] At 00:14.000, hold them in a near-distance pressure standoff. The main warrior stays lower and tighter, violet eyes sharp and loaded with both defense and attack. The dark counterpart stands more upright in black, cold green eyes fixed on her, mouth carrying a slight smile. She says in a soft, low, controlled voice: <d>[English] Now it’s my turn.</d> Do not cut to black. Instead, give a very slight forward push and stop on the two nearly identical faces with two clearly different eye colors. On “Now” the smile tightens smaller; on “my turn” the voice drops lower. After speaking, she keeps looking only at the main warrior. The main warrior does not recoil or shout; she only hardens the eyes, thins the lip line, and shows one brief throat-tightening swallow. Candle edges tremble faintly, dust remains minimal, and both violet and green eyes stay readable in the final frame.\n\noverall_soundscape: Keep the sound field sparse, intimate, and low-frequency throughout: damp stone ambience, faint air movement, tiny candle crackle, soft cloth friction, boot contact on stone, light sword metal, and distant hollow temple resonance. Emphasize shallow breath, cloth tension, and hand-and-knee contact with wet stone in the opening; grounded cloak movement, boots, armor friction, and controlled sword handling during the rise and advance; unnatural stillness as the anomaly beat; then real physical cues for the dark counterpart’s arrival, including cloak edge, boot step, and subtle fabric drag. The slash should be sharp, restrained, and metallic, and the final line should land close, low, calm, and cold inside the attack line.\n\nnon_diegetic_music: Use minimal score or an extremely restrained dark ambient bed. Keep the first half as low suspended pressure under stillness, tighten it slightly as the main warrior rises and advances, thin it almost to nothing when the sensed target vanishes, then introduce a colder deeper tone as the dark counterpart emerges. Under the final line, let one faint low note deepen and hold without release.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2098767994750955881",
    "sourceUrl": "https://x.com/ou_zhen599/status/2098767994750955881",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-12T13:37:37.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098767994750955881.mp4",
    "posterUrl": "/posters/x/2098767994750955881.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal",
      "dark"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI    \n\n🎬Image-to-Video \n📷“Echo Hunt”\nAn elven warrior infiltrates an abandoned temple in pursuit of her target, using magic to access the residual memories at the scene—only to discover that her “prey” is, in fact, her own self from another timeline.\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photoreal dark-fantasy image-to-video sequence based on <Picture 1>. Preserve the exact same female elf warrior throughout: same face, elf ears, facial structure, hairstyle, gray-purple cloak, silver-steel armor, fitted leather battlewear, sword, body proportions, and realistic texture. The entire sequence stays inside one continuous abandoned temple interior with damp stone steps, broken archway, rough pillars, dying candles, moss, scattered leaves, wet walls, faint dust, and cold moist air. No exterior cut, no extra characters, no flashy fantasy VFX, no face drift, no identity drift. The main elf warrior advances from the lower steps toward the deeper archway. The dark counterpart always remains hidden behind a rear-side pillar and emerges physically from that same shadow, never teleporting. The tone is cold, restrained, dangerous, mysterious, and subtly sensual through presence, gaze, and motion, not posing. The main warrior’s magic exists only inside the eye: her violet eye subtly focuses inward, never emitting outward effects. The dark counterpart has the same face and species but wears a black cloak, matte dark armor, sharper shoulders, and carries cold green eyes that stay deep, stable, and clearly distinct from the warrior’s violet gaze. [Shot 2] At 00:02.200, the camera makes a low smooth push over wet steps, leaves, candle stubs, and cracked stone, then settles below and in front of the main warrior in a crouched position. One knee presses to the step, left hand braces on stone, right hand holds the sword, and the cloak falls with real weight. She is almost motionless, like a predator storing force. Her head is slightly lowered, gaze first resting on the step near her hand, lips closed, mouth corners flat, chin tucked, brow faintly gathered, and breathing nearly invisible. Only the environment moves: candle tremor, damp reflections, drifting dust. [Shot 3] At 00:04.100, cut to an intimate eye close-up. She catches that something is wrong. After half a beat of stillness, her eyes lift slowly. The violet eye does not flare; it only gathers inward with a subtle tightening deep inside the pupil. The emotional path is stillness to detection to lock. The lower eyelid tightens, breath shortens by one measure, and the lip line firms. Stay close on one eye, nose bridge, lips, and jawline as candlelight brushes her side contour. One side of her mouth lifts by the smallest amount, as if confirming a target. [Shot 4] At 00:05.400, she rises from the crouch in one swift efficient motion. The cloak swings backward with believable inertia, and the sword comes from low position into a forward guard line. She locks upright like a hunter ending concealment. Her jaw tightens, lips part only slightly to change breathing, brow compresses, and gaze hardens. Only physical feedback appears: cloak weight, slight dust disturbance, candle response. [Shot 5] At 00:06.900, she pushes quickly along the stone steps toward the deeper archway. Passing a pillar, her body skims close to the wall, the cloak brushes stone, candlelight skims her side face, and the sword stays in a ready attack-defense line. This is controlled advance, not panic. Steps stay tight, shoulders stable, head motion minimal, gaze fixed on one search line. Her eyelids narrow, breath shortens, and mouth corners flatten. At 00:08.500, she suddenly stops by half a step as if the sensed target in front has vanished. She instantly uses the pillar to half-turn, bringing the sword back toward her chest in defense. The shift is active advance to professional alert. Her eyes contract first, then broaden their search, lower lids tighten, nostrils draw slightly inward, and the lip line presses thinner. The environment feels wrong through unnatural stillness: candle motion reduces, dust slows, and the space becomes too quiet. [Shot 6] At 00:10.000, place the camera over her shoulder with part of her hood and sword edge in foreground and empty space ahead. The threat does not appear in front. Then, behind the rear pillar, the edge of a black cloak slowly slips into candlelight, followed by one black boot stepping out. The main warrior has not fully turned yet, but her back, neck, and shoulders are already tight. The dark counterpart emerges slowly on purpose, simply entering light from real shadow. The camera glides sideways and brings her fully into view: same face and species, but in black battlewear, matte black armor, sharp shoulder plates, and a long black cloak. She is darker, heavier, more severe. Her cold green eyes are deep and controlled, never neon, and one side of her mouth carries a faint wicked smile. She is not furious; she is already in control. [Shot 7] At 00:12.900, the main warrior whips around and slashes back in one immediate, accurate defensive strike. The blade cuts close past the camera line. The dark counterpart does not hurry to block. She simply steps half a step forward into the attack line and tilts her head enough for the blade to skim past. The main warrior’s intensity spikes without becoming theatrical: brow tight, nostrils slightly open, lips parted only a little from exertion, jaw rigid. The dark counterpart’s smile stays and grows colder during the head tilt, as if enjoying the failed strike. Keep the sword realistic with only restrained metallic reflection and real cloak inertia. [Shot 8] At 00:14.000, hold them in a near-distance pressure standoff. The main warrior stays lower and tighter, violet eyes sharp and loaded with both defense and attack. The dark counterpart stands more upright in black, cold green eyes fixed on her, mouth carrying a slight smile. She says in a soft, low, controlled voice: <d>[English] Now it’s my turn.</d> Do not cut to black. Instead, give a very slight forward push and stop on the two nearly identical faces with two clearly different eye colors. On “Now” the smile tightens smaller; on “my turn” the voice drops lower. After speaking, she keeps looking only at the main warrior. The main warrior does not recoil or shout; she only hardens the eyes, thins the lip line, and shows one brief throat-tightening swallow. Candle edges tremble faintly, dust remains minimal, and both violet and green eyes stay readable in the final frame.\n\noverall_soundscape: Keep the sound field sparse, intimate, and low-frequency throughout: damp stone ambience, faint air movement, tiny candle crackle, soft cloth friction, boot contact on stone, light sword metal, and distant hollow temple resonance. Emphasize shallow breath, cloth tension, and hand-and-knee contact with wet stone in the opening; grounded cloak movement, boots, armor friction, and controlled sword handling during the rise and advance; unnatural stillness as the anomaly beat; then real physical cues for the dark counterpart’s arrival, including cloak edge, boot step, and subtle fabric drag. The slash should be sharp, restrained, and metallic, and the final line should land close, low, calm, and cold inside the attack line.\n\nnon_diegetic_music: Use minimal score or an extremely restrained dark ambient bed. Keep the first half as low suspended pressure under stillness, tighten it slightly as the main warrior rises and advances, thin it almost to nothing when the sensed target vanishes, then introduce a colder deeper tone as the dark counterpart emerges. Under the final line, let one faint low note deepen and hold without release.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 293,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098767441241542863",
    "title": "巨幅排版中的动画角色乱斗",
    "titleEn": "An Anime Character Brawls Through Giant Typography",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "彩色动画角色在“DELTA”等巨幅文字和图形界面间翻滚、冲撞，快切排版与动作共同组成介绍短片。 原生视频规格为 10 秒 · 1440×816。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A colorful anime character tumbles and collides through giant “DELTA” typography and graphic interfaces, combining kinetic type with chaotic action in an intro reel. Native-video output: 10s · 1440×816. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NullGamerA/status/2098767441241542863",
    "sourceLabel": "X 原帖 · @NullGamerA",
    "author": "アポロ",
    "publishedAt": "2026-09-12T13:35:25.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098767441241542863.mp4",
    "posterUrl": "/posters/x/2098767441241542863.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1440×816",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ローカルMinimaxH3\n紹介動画　スパイダーマン風デルタもん\n\nプロンプトは↓\nhttps://chatgpt.com/s/t_6aa554d713408191b05ed484003593e3\n\nハチャメチャアクションになりました\n\n引用のプロンプトをお借りしました\n\n#デルタもん",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 172,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098751180671508713",
    "title": "时间停止也挡不住自动门",
    "titleEn": "Even Time Stop Cannot Beat an Automatic Door",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物在便利店门前摆出时间停止姿势，周围泡泡与路人冻结，自动门却照常开合，形成一个视觉笑点。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character poses as if stopping time outside a convenience store while bubbles and bystanders freeze, but the automatic doors keep moving and land the visual joke. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2098751180671508713",
    "sourceLabel": "X 原帖 · @tokyo_Valentine",
    "author": "とすくん",
    "publishedAt": "2026-09-12T12:30:49.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098751180671508713.mp4",
    "posterUrl": "/posters/x/2098751180671508713.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "時間停止能力、最大の弱点。\n自動ドア。\n\nプロンプトはリプにあるので、ご自由にお使いください。\n\nMiniMax H3 MAXで生成しました。\nMade with\n@Hailuo_AI #minimaxH3",
    "engagement": {
      "replies": 3,
      "reposts": 6,
      "likes": 41,
      "views": 1262,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098749573573595394",
    "title": "红裙女子与黑牛的海岸MV预告",
    "titleEn": "A Red-Dressed Woman and Black Bull in a Coastal MV Teaser",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红裙女子与黑牛沿海岸步道同行，镜头从正面肖像切到空旷道路和门洞背影，形成三十秒超现实MV预告。 原生视频规格为 33 秒 · 1440×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a red dress walks beside a black bull along a coastal path, shifting from frontal portraits to an empty road and doorway silhouette in a surreal thirty-second MV teaser. Native-video output: 33s · 1440×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_lifehack55/status/2098749573573595394",
    "sourceLabel": "X 原帖 · @ai_lifehack55",
    "author": "AIライフハック",
    "publishedAt": "2026-09-12T12:24:25.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098749573573595394.mp4",
    "posterUrl": "/posters/x/2098749573573595394.jpg",
    "duration": 33,
    "aspectRatio": "square",
    "resolution": "1440×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 MV公開予告？\nThe Bull and Me｜雄牛と私\n\nSeadance2.0用の動画をMiniMax用にリメイク...のつもりだったのに、なぜかMVを作りたくなっちゃいました💦\n\n相変わらずタイトルはシュールですが、英訳ってこれでいいのかなぁ～と悩んでいます🤣現在、音楽だけは完成🎵",
    "engagement": {
      "replies": 2,
      "reposts": 9,
      "likes": 64,
      "views": 1318,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098745637886193722",
    "title": "从灰模齿轮到成品腕表",
    "titleEn": "From Greybox Gears to a Finished Wristwatch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下分屏同时展示简化齿轮预演与写实生成结果，零件逐步组合成表盘和皮质腕表，直观呈现 Blender 预演工作流。 原生视频规格为 10 秒 · 1600×1800。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A split screen pairs a simplified gear previs with the photoreal generated result as parts assemble into a dial and leather watch, clearly showing a Blender-assisted workflow. Native-video output: 10s · 1600×1800. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SSSS_CRYPTOMAN/status/2098745637886193722",
    "sourceLabel": "X 原帖 · @SSSS_CRYPTOMAN",
    "author": "SSSS.CRYPTOMAN⚡️AI",
    "publishedAt": "2026-09-12T12:08:47.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098745637886193722.mp4",
    "posterUrl": "/posters/x/2098745637886193722.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1600×1800",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "⚙完全AIで作るメカニカル動画⚙\n\nこの作例はGPT-6 AstraとBlender、Hailuo @Hailuo_AI の #MiniMaxH3 を組み合わせて、すべてAIの力で作りました\n\n【制作の流れ】\n✅CodexからAstraをメインに使いBlenderでプレビズを作成\n✅プレビズを元にAstraで動画生成用のプロンプトを作成\n✅そのプロンプトを使ってMinimax H3で動画生成して完成！\n\nAI動画でゼロから複雑な構造を作らせると破綻しやすいですが、こんな感じでBlenderと組み合わせるとかなりいい精度で仕上がりますね\n※今回は簡易的に作ったので時計の機構としては合ってません\n\nBlenderでこのクオリティでテクスチャ入れて背景までレンダリングをするのは結構大変だと思いますが、Minimax H3を使えばローコスト＆爆速でレンダリングが完了します\nこれ、ちゃんと使えばCGレンダリングにコスト・時間がかなり削減されるかもしれません\n\nちなみにほぼ全部スマホからリモートで適当に指示しただけです\nもはや人間がパソコンを触る必要すら無くなってきました…ｗ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 1312,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098745248121766292",
    "title": "同一K-pop提示词的四模型首轮结果",
    "titleEn": "Four Models Take the Same K-Pop Prompt on Their First Try",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "四段女团舞台表演依次标注不同视频模型，MiniMax H3 位于最后十五秒，用同一提示词比较人物、舞蹈和舞台稳定性。 原生视频规格为 60 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Four girl-group stage performances are labeled by model in sequence, with MiniMax H3 in the final fifteen seconds, comparing people, dance, and stage consistency under one prompt. Native-video output: 60s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KevinAceAI/status/2098745248121766292",
    "sourceLabel": "X 原帖 · @KevinAceAI",
    "author": "Kevin Ace",
    "publishedAt": "2026-09-12T12:07:14.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098745248121766292.mp4",
    "posterUrl": "/posters/x/2098745248121766292.jpg",
    "duration": 60,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "music-performance",
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "K-POP girl group.\nSame prompt. Four models. First takes only — no particular order.\n\n0:00 Seedance 2.5\n0:15 Wan 3.0\n0:30 Kling 3.0 Pro\n0:45 Minimax H3\n\nWhich one wins?\nMore comparisons ↓",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 91,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098744370954686840",
    "title": "骑士推开城堡大门取得金色宝物",
    "titleEn": "A Knight Opens the Castle Gate and Claims a Golden Relic",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红披风骑士走近山崖城堡，推开火把照亮的巨门，最后捧起发光金色宝物完成一段冒险目标。 原生视频规格为 15 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-caped knight approaches a cliffside castle, opens its torchlit gate, and raises a glowing golden relic to complete the adventure beat. Native-video output: 15s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yoradino/status/2098744370954686840",
    "sourceLabel": "X 原帖 · @yoradino",
    "author": "YOU RA",
    "publishedAt": "2026-09-12T12:03:45.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098744370954686840.mp4",
    "posterUrl": "/posters/x/2098744370954686840.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Trying to make a knight video getting treasure \n\nModel: Minimax H3\nSogni ai:\nhttps://app.sogni.ai/?code=Yora\n\n@Sogni_Protocol #Sogni #MiniMaxH3 #AIVideo",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 91,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098726050503205274",
    "title": "三文鱼料理的双模型音画对比",
    "titleEn": "A Two-Model Audio-Visual Comparison for a Salmon Dish",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "方形分屏并列展示两种模型将食材变成摆盘三文鱼的过程，以同一提示词比较食物变形、镜头和原生声音。 原生视频规格为 32 秒 · 720×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A square split screen shows two models transforming ingredients into a plated salmon dish, comparing food motion, camera work, and native audio under the same prompt. Native-video output: 32s · 720×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lin_evelyn24051/status/2098726050503205274",
    "sourceLabel": "X 原帖 · @lin_evelyn24051",
    "author": "Evelyn Lin",
    "publishedAt": "2026-09-12T10:50:57.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098726050503205274.mp4",
    "posterUrl": "/posters/x/2098726050503205274.jpg",
    "duration": 32,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "food",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 vs Seedance 2.5\n影音皆由模型直接輸出，沒有後製添加聲音。\n你選哪一版？👀\nSame prompt, including visual and audio directions.\nDirect model output—which one wins?\n\n#GMICloud #Seedance25 #MiniMaxH3 #AIVideo",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 90,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098725353187586059",
    "title": "城市废墟中的高速近身战",
    "titleEn": "A High-Speed Close-Quarters Fight in Urban Ruins",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名动画战士在废墟街道连续冲刺、格挡和重击，快速镜头努力呈现速度感与一击重量。 原生视频规格为 28 秒 · 1252×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two anime fighters dash, block, and trade heavy blows through a ruined street as rapid camera movement pushes for speed and impact. Native-video output: 28s · 1252×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rizdel00/status/2098725353187586059",
    "sourceLabel": "X 原帖 · @rizdel00",
    "author": "Rizdel",
    "publishedAt": "2026-09-12T10:48:11.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098725353187586059.mp4",
    "posterUrl": "/posters/x/2098725353187586059.jpg",
    "duration": 28,
    "aspectRatio": "landscape",
    "resolution": "1252×720",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H３ローカル。戦闘シーンを作りたくて挑戦。スピード感や一撃の重さをうまく表現できず、２０テイク以上作り直したが、ここまででした。#AI動画 #MiniMax",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 52,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098716750690865428",
    "title": "猫店长甩锅做饭的十秒MV镜头",
    "titleEn": "The Cat Manager Tosses a Wok in a Ten-Second MV Shot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿色头发的猫厨师从屋顶跃入灯笼夜市，食材在炒锅上方飞舞，最后举起菜刀与锅完成角色亮相。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A green-haired cat chef leaps from a rooftop into a lantern-lit night market, tosses ingredients above a wok, and finishes by presenting a cleaver and pan. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Neko_ai_p/status/2098716750690865428",
    "sourceLabel": "X 原帖 · @Neko_ai_p",
    "author": "ネコ店長",
    "publishedAt": "2026-09-12T10:14:00.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098716750690865428.mp4",
    "posterUrl": "/posters/x/2098716750690865428.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cinematic-realistic"
    ],
    "scenes": [
      "food",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🐈ネコ店長、はじめてのMV制作\n\nChatGPTを使用したストーリーボード制作を、さらに洗練させるために、プロンプトの最後へ下記の文言を追加して実行しました！\n\n「プロンプトは公式プロンプトを参照して生成してください。」\n\nその結果……。\n\n「格段に良くなりました！」\n\nまず、時間内に適したカット数になったと思います。\n\n前回は、かなり詰め込みすぎていました！\n\n今回は、ひとつの場面を少しゆっくり見せてくれているので、展開を見ていても疲れません。\n\nさらに、カメラアングルのバリエーションも増えました。\n\nフライパンを振り上げて、料理が派手に舞うシーンも印象的になっています！\n\nかなりいい感じになりました！！\n\nですが、これまで「MiniMax H3 Max」を使ってきて、ひとつ思ったことがあります。\n\n15秒くらいでそれなりの映像を作れたとしても、MVでは曲に合わせて映像を制作しなければいけません。\n\nそう考えると、ショートの細切れ映像のほうが使いやすい！\n\nと思いました。\n\nですので、ストーリーボードを使って15秒の映像をまとめて作るよりも、「狙った映像」をひとつずつ生成してもらうほうが良さそうです。\n\nしかし、そう思ったものの……。\n\nやり方がわかりません（笑）\n\nそこで、AIの大先輩である「あのお方」に相談してみることにしました。\n\nはたして、解決するのでしょうか。\n\nつづく。\n\n※次回から、また「ネコ店長三文劇」に戻ります。\nマネージャー四天王の一角が登場します！！\n\nーーーーーーーー\n※プロンプトは、角煮星丸さんのポストを参考にしています。有益なポストばかりだから、覗いてみてください！\n@_3912657840\nーーーーーーーー",
    "engagement": {
      "replies": 1,
      "reposts": 12,
      "likes": 53,
      "views": 701,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098713234454765923",
    "title": "花街霓虹中的侍与花魁长篇MV",
    "titleEn": "A Long-Form Samurai MV in a Neon Flower District",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "侍与花魁穿行红黑金色的霓虹花街，剪影、字幕和战斗镜头随高BPM音乐切换，组成两分四十五秒的完整MV。 原生视频规格为 166 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A samurai and courtesan move through a red-black-gold neon district as silhouettes, captions, and combat shots cut to a high-BPM track across a complete two-minute-forty-five MV. Native-video output: 166s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kuroneko0618/status/2098713234454765923",
    "sourceLabel": "X 原帖 · @kuroneko0618",
    "author": "アメショのユキ",
    "publishedAt": "2026-09-12T10:00:01.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098713234454765923.mp4",
    "posterUrl": "/posters/x/2098713234454765923.jpg",
    "duration": 166,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "illustration",
      "retro"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "日が沈みきった瞬間に、侍が抜く。\n\nハードスタイルMV「紅蓮」公開しました。\n侍と花魁、ネオンの花街。2分45秒・33カット。\n\n曲：Suno ／ 絵：GPT Image ／ 動画：MiniMax H3\n編集：全部スクリプト。164BPM・453拍を測って、\n　　　カットの切り替わりを全部ビートに乗せています\n\nいちばん時間がかかったのは、絵ではなく編集でした。\n\n#AIアート #AI動画 #Suno",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 118,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098710667905859968",
    "title": "暴雨基地中的两分钟机甲短片",
    "titleEn": "A Two-Minute Mecha Short at a Storm-Lashed Base",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "风暴夜里的军事基地遭遇巨型机甲，士兵搬运设备、奔跑撤离并进入城市战斗，连续镜头完成科幻短片叙事。 原生视频规格为 120 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A stormy military base is confronted by a giant mech as soldiers move equipment, flee, and carry the action into the city in a continuous two-minute sci-fi short. Native-video output: 120s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pop4x/status/2098710667905859968",
    "sourceLabel": "X 原帖 · @pop4x",
    "author": "Shukri Saham Global",
    "publishedAt": "2026-09-12T09:49:50.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098710667905859968.mp4",
    "posterUrl": "/posters/x/2098710667905859968.jpg",
    "duration": 120,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🤖FALLEN SKY — a 2-minute AI mecha short.\n\nMiniMax H3\nProducer: PST\nDirector: ChatGPT\nProduction: Hermes\nCost: $11.13\nTime: ~1 hour\n#FallenSky #MiniMaxH3 #AIFilmmaking",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 112,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098710032259141801",
    "title": "从简化网格到角色拳击动作",
    "titleEn": "From a Simplified Mesh to a Character Combo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右画面同步展示简化人体网格与原角色骨骼，拳、拳、踢的组合动作被生成后重定向回完整角色。 原生视频规格为 12 秒 · 1040×588。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Side-by-side views show a simplified body mesh and the original character rig as a punch-punch-kick combo is generated and retargeted onto the finished model. Native-video output: 12s · 1040×588. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "The Character pulls back his hand then does a punch punch kick combo, Martial Arts, Tekken",
    "promptSourceUrl": "https://x.com/solana_bandit/status/2098710032259141801",
    "sourceUrl": "https://x.com/solana_bandit/status/2098710032259141801",
    "sourceLabel": "X 原帖 · @solana_bandit",
    "author": "🇧‌🇦‌🇳‌🇩‌🇮‌🇹",
    "publishedAt": "2026-09-12T09:47:18.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098710032259141801.mp4",
    "posterUrl": "/posters/x/2098710032259141801.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1040×588",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Text to motion has been solved. \nI uploaded a rig of Armor King from Tekken. Input prompt:  \"The Character pulls back his hand then does a punch punch kick combo, Martial Arts, Tekken\"\n\nMotion3D creates a simplified Armour King mesh from his bones and volume. \nIt animates that rig using minimax h3. \nIt then reconstructs the generation in 3D, and finally retargets the original Armor king rig onto the generation. \n\nThe whole generation takes around 15 minutes local for a 5 second output on a RTX 3090ti.\n\nThe hands are a little bit wonky as i havent added inverse kinematics to this rig, nor the ability to set poses yet. I will be adding that shortly though.\n\n@Motion3D_Ai",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 240,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098709214935839112",
    "title": "地下设施中的四十五秒科幻预告",
    "titleEn": "A Forty-Five-Second Sci-Fi Trailer Below Ground",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "武装小队进入昏暗地下设施，追逐与异形威胁逐步升级，黑底标题卡把多段镜头组织成完整预告片。 原生视频规格为 44 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An armed team enters a dim underground facility as pursuit and an alien threat escalate, with black title cards organizing the sequence into a complete trailer. Native-video output: 44s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tihomirVR/status/2098709214935839112",
    "sourceLabel": "X 原帖 · @tihomirVR",
    "author": "Tisho 🇧🇬 🇪🇺",
    "publishedAt": "2026-09-12T09:44:03.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098709214935839112.mp4",
    "posterUrl": "/posters/x/2098709214935839112.jpg",
    "duration": 44,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "dark"
    ],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing MiniMax H3 Max on @fal — 45s sci-fi trailer generated at 768p.\nHuge thanks to @matthieu_ai for the $25 fal credit! 🙌\nSound ON 🔊\n#H3Max #MiniMax #AIvideo #fal",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 5,
      "views": 168,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098706556225511650",
    "title": "绿发女孩紧张地抱住肩膀",
    "titleEn": "A Green-Haired Girl Nervously Holds Her Shoulder",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿发眼镜少女在暗色室内低头收拢身体，双手交叠按住肩膀，用细微呼吸与眼神表达紧张。 原生视频规格为 10 秒 · 512×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A green-haired girl with glasses lowers her gaze in a dark room and crosses her hands over her shoulder, using subtle breathing and eye movement to convey anxiety. Native-video output: 10s · 512×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YamataZen/status/2098706556225511650",
    "sourceLabel": "X 原帖 · @YamataZen",
    "author": "Yamata",
    "publishedAt": "2026-09-12T09:33:29.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098706556225511650.mp4",
    "posterUrl": "/posters/x/2098706556225511650.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "512×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minami covering her shoulder with her hand nervously\n\n#aivideo #MiniMaxH3 #OC",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 14,
      "views": 573,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098704322691584391",
    "title": "钟楼迷宫中的百秒Q版冒险",
    "titleEn": "A Hundred-Second Chibi Adventure Through a Clockwork Maze",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三名Q版角色围绕钟楼、齿轮大门和谜题展开追逐，日文字幕、场景转换与角色互动组成一段连续冒险。 原生视频规格为 109 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three chibi characters race through a clock tower, gear-driven gate, and puzzle sequence as Japanese captions, scene changes, and character interaction sustain a continuous adventure. Native-video output: 109s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/WBK_SZ/status/2098704322691584391",
    "sourceLabel": "X 原帖 · @WBK_SZ",
    "author": "WBK",
    "publishedAt": "2026-09-12T09:24:37.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098704322691584391.mp4",
    "posterUrl": "/posters/x/2098704322691584391.jpg",
    "duration": 109,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "fantasy",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax Music 3で作った曲に、Endless MiniMax H3で動画をつけてみました。\n\n15秒ごとに生成して、そのままつないだだけの編集です。\n\n絵やテキストの破綻は頑張って直すしかなさそうですが、クリップ同士のつなぎ目は、少し編集すれば整えられそうです。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 67,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098884402361315536",
    "title": "同一角色两种流程的舞台亮相",
    "titleEn": "One Character, Two Pipelines, One Stage Entrance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫发角色手持折扇在金色舞台上亮相，左右分别为纯生成流程与加入另一模型的流程，用于比较成本与效果。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple-haired character with a folding fan takes the stage, shown left and right as a pure generation pipeline versus one with an added model to compare cost and result. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/dango333/status/2098884402361315536",
    "sourceLabel": "X 原帖 · @dango333",
    "author": "はげだんご🍡",
    "publishedAt": "2026-09-12T21:20:11.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098884402361315536.mp4",
    "posterUrl": "/posters/x/2098884402361315536.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🍡おはだんご🍡\n\nロマ子様動画生成をいろいろ試して楽しんでるブヒ🐷\n\n①Astra×GPT image2.5×MiniMAX H3 Max Turbo（左）\n②①+動画Seedance2.0（右）\n\n①は安く作成できるが、時間と手間がかかる\n②はサクッと作成できるがコストが高い\n\nまだまだ試行の旅は続く。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 9,
      "views": 137,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098798465107022083",
    "title": "街头人群中变身的红衣女巫",
    "titleEn": "A Red Witch Transforms Amid a Street Crowd",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰色连帽衫女子穿过城市人群，瞬间变为头戴红冠的猩红女巫，随后腾空飞离，路人纷纷回头。 原生视频规格为 8 秒 · 2432×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a grey hoodie walks through a city crowd, transforms into a scarlet-clad witch with a red crown, and lifts off into the air as passers-by turn to look. Native-video output: 8s · 2432×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/one_good_prompt/status/2098798465107022083",
    "sourceLabel": "X 原帖 · @one_good_prompt",
    "author": "One Good Prompt Production",
    "publishedAt": "2026-09-12T15:38:42.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098798465107022083.mp4",
    "posterUrl": "/posters/x/2098798465107022083.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "2432×1344",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Scarlet Witch / Wanda Maximoff lives!!! @marvel @MarvelStudios . Super @grok helped with the design. Rendered completely in @ComfyUI and @Hailuo_AI Minimax H3 R2V.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 79,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098725534859678141",
    "title": "三分钟水墨暗黑寓言《焚香剖佛》",
    "titleEn": "A Three-Minute Ink-Style Dark Fable, \"The Desecrated Lotus\"",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "以传统中国二维绘画风格讲述贪婪与亵渎的寓言，僧人、剖开的佛像与披甲武士依次登场，三分钟内完成一段业报故事。 原生视频规格为 186 秒 · 2520×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A traditional Chinese 2D-painting fable of greed and sacrilege, moving from a monk and a cut-open Buddha figure to armored warriors across a three-minute karmic tale. Native-video output: 186s · 2520×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Kuriyama890/status/2098725534859678141",
    "sourceLabel": "X 原帖 · @Kuriyama890",
    "author": "0xKuriyama |Kendu⚡🔋",
    "publishedAt": "2026-09-12T10:48:54.000Z",
    "addedAt": "2026-09-13T00:34:24.391Z",
    "mediaUrl": "/media/x-2098725534859678141.mp4",
    "posterUrl": "/posters/x/2098725534859678141.jpg",
    "duration": 186,
    "aspectRatio": "landscape",
    "resolution": "2520×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration",
      "dark"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Faith withers when death knocks at the gate.\n《焚香剖佛》 (The Desecrated Lotus) A traditional Chinese 2D animated dark fable exploring greed, sacrilege, and karmic transmutation.\nCrafted for the Bestiary Creative Contest with Miora & MiniMax H3.\n\n@Miora_Design @Hailuo_AI \n#MioraDesign #MioraBuddy #TencentAI #MiniMaxH3 #MiniMax",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 107,
      "snapshotAt": "2026-09-13T00:34:24.391Z"
    },
    "approvedAt": "2026-09-13T00:34:24.391Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098794141715226927",
    "title": "带雀斑的面部特写光影",
    "titleEn": "Light Across a Freckled Face in Close-Up",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发女子的面部在极近景中缓慢转动，雀斑、睫毛与湿润的唇色随光线变化，用于观察剪枝后模型的人脸质感。 原生视频规格为 15 秒 · 928×1664。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman’s face turns slowly in extreme close-up while freckles, lashes and glossy lips shift under the light, a look at skin texture from the pruned model. Native-video output: 15s · 928×1664. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Nanami_AI0/status/2098794141715226927",
    "sourceLabel": "X 原帖 · @Nanami_AI0",
    "author": "东云七海",
    "publishedAt": "2026-09-12T15:21:31.000Z",
    "addedAt": "2026-09-13T01:30:30.323Z",
    "mediaUrl": "/media/x-2098794141715226927.mp4",
    "posterUrl": "/posters/x/2098794141715226927.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "928×1664",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "感觉Minimax H3模型剪枝之后生成的人脸还是有点油\n#ai #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 92,
      "snapshotAt": "2026-09-13T01:30:30.323Z"
    },
    "approvedAt": "2026-09-13T01:30:30.323Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098792011247845858",
    "title": "为VR生成的360度夜景少女",
    "titleEn": "A 360-Degree Night Scene Girl Made for VR",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色双马尾少女趴在圆形灯饰前朝观者微笑，整幅画面以等距柱状全景输出，可直接转成VR场景。 原生视频规格为 8 秒 · 1472×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired twin-tail girl leans over a round lamp smiling at the viewer, the whole frame rendered as an equirectangular panorama ready for VR. Native-video output: 8s · 1472×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jun_kongo/status/2098792011247845858",
    "sourceLabel": "X 原帖 · @jun_kongo",
    "author": "kongo jun / 混合 順",
    "publishedAt": "2026-09-12T15:13:03.000Z",
    "addedAt": "2026-09-13T01:30:30.323Z",
    "mediaUrl": "/media/x-2098792011247845858.mp4",
    "posterUrl": "/posters/x/2098792011247845858.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1472×736",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 equi360-lora\n（←生成結果   VR化→）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 131,
      "snapshotAt": "2026-09-13T01:30:30.323Z"
    },
    "approvedAt": "2026-09-13T01:30:30.323Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098767701011263679",
    "title": "海岸边微笑的卷发女子",
    "titleEn": "A Curly-Haired Woman Smiles by the Shore",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卷发女子站在岩岸与海浪前对镜头微笑并轻轻转头，五秒内保持面部与发丝细节稳定。 原生视频规格为 5 秒 · 876×504。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A curly-haired woman smiles at the camera and turns her head slightly before rocks and surf, keeping face and hair detail steady across five seconds. Native-video output: 5s · 876×504. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Cobaye47/status/2098767701011263679",
    "sourceLabel": "X 原帖 · @Cobaye47",
    "author": "Cobaye 47 💸",
    "publishedAt": "2026-09-12T13:36:27.000Z",
    "addedAt": "2026-09-13T01:30:30.323Z",
    "mediaUrl": "/media/x-2098767701011263679.mp4",
    "posterUrl": "/posters/x/2098767701011263679.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "876×504",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "J'ai pas encore le temps de m'y remettre complètement mais j'ai voulu tester minimax h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 80,
      "snapshotAt": "2026-09-13T01:30:30.323Z"
    },
    "approvedAt": "2026-09-13T01:30:30.323Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098767510749454810",
    "title": "孙悟空与公主逛西贡街头",
    "titleEn": "The Monkey King and a Princess Stroll Through Saigon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿黄袍的孙悟空与身着红色古装的公主在滨城市场与动物园门前逛街、买小吃，路人与街景保持真实感，五十秒连续短片。 原生视频规格为 50 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A yellow-robed Monkey King and a princess in red period dress wander past Ben Thanh Market and the zoo gate buying snacks, with lifelike passers-by across a fifty-second piece. Native-video output: 50s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Khiemtran1910/status/2098767510749454810",
    "sourceLabel": "X 原帖 · @Khiemtran1910",
    "author": "KhiemtranVN",
    "publishedAt": "2026-09-12T13:35:42.000Z",
    "addedAt": "2026-09-13T01:30:30.323Z",
    "mediaUrl": "/media/x-2098767510749454810.mp4",
    "posterUrl": "/posters/x/2098767510749454810.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Rất Ổn",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 37,
      "snapshotAt": "2026-09-13T01:30:30.323Z"
    },
    "approvedAt": "2026-09-13T01:30:30.323Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098750961074233579",
    "title": "拉各斯版开放世界游戏任务",
    "titleEn": "An Open-World Game Mission, Lagos Edition",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "以游戏界面风格呈现拉各斯街头：任务提示、跑车驶向别墅、泳池边打出\"MISSION PASSED\"，模仿经典开放世界游戏的过场。 原生视频规格为 15 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Lagos streets framed like a game HUD: a mission prompt, a sports car heading to the villa, and a poolside \"MISSION PASSED\" card in the style of a classic open-world game. Native-video output: 15s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Bokumandi/status/2098750961074233579",
    "sourceLabel": "X 原帖 · @Bokumandi",
    "author": "🎨 Bókumandi 🌟",
    "publishedAt": "2026-09-12T12:29:56.000Z",
    "addedAt": "2026-09-13T01:30:30.323Z",
    "mediaUrl": "/media/x-2098750961074233579.mp4",
    "posterUrl": "/posters/x/2098750961074233579.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This minimax H3 max  turbo really deliver for the price 😂 GTA Lagos Edition!",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-09-13T01:30:30.323Z"
    },
    "approvedAt": "2026-09-13T01:30:30.323Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2098712471196561639",
    "title": "交易中心里的耳机女孩与滚动行情",
    "titleEn": "A Headset Girl in a Trading Hall with Scrolling Tickers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴耳机的动画女孩在蓝色基调的交易中心里向镜头伸手，环形屏幕上滚动着股票网站的名称与代码，作为一段十五秒的宣传动画。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime girl in a headset reaches toward the camera inside a blue-toned trading hall while ring screens scroll a stock site’s name and codes in a fifteen-second promo. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/trader_hashang/status/2098712471196561639",
    "sourceLabel": "X 原帖 · @trader_hashang",
    "author": "はっしゃん@株Biz",
    "publishedAt": "2026-09-12T09:56:59.000Z",
    "addedAt": "2026-09-13T01:30:30.323Z",
    "mediaUrl": "/media/x-2098712471196561639.mp4",
    "posterUrl": "/posters/x/2098712471196561639.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ClaudeCode+MiniMaxH3 ローカル生成15秒アニメのトレードセンター編を「https://kabubiz.com」と「RIRON」のティッカーが回るようにオリジナル背景で作り込み。ブルー基調のクールなデザインになりました🤔\n#MinimaxH3 #ComfyUI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 10,
      "views": 9046,
      "snapshotAt": "2026-09-13T01:30:30.323Z"
    },
    "approvedAt": "2026-09-13T01:30:30.323Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099276730288709960",
    "title": "蓝白画布上的造型连变",
    "titleEn": "Rapid Pose Changes on a Blue-and-White Set",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣短发女子在蓝白平面图形之间连续变换姿势与造型，镜头从全身逐步切到戴眼镜的面部特写。 原生视频规格为 10 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A short-haired woman in black shifts through poses and looks against blue-and-white graphics, moving from a full-body composition to a glasses close-up. Native-video output: 10s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Complete one major visual transformation every 1 second. \nNo voice or dialogue. \nSync all transitions to the BGM.",
    "promptSourceUrl": "https://x.com/taya_mama_AI/status/2099276730288709960",
    "sourceUrl": "https://x.com/taya_mama_AI/status/2099276730288709960",
    "sourceLabel": "X 原帖 · @taya_mama_AI",
    "author": "た〜や｜スキマ時間×AI副業ママ",
    "publishedAt": "2026-09-13T23:19:09.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099276730288709960.mp4",
    "posterUrl": "/posters/x/2099276730288709960.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "falのMiniMax H3 turbo今日まで75％オフですってー😍\n\n▼簡単動画プロンプト\n\nComplete one major visual transformation every 1 second. \nNo voice or dialogue. \nSync all transitions to the BGM.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 287,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099274095812849949",
    "title": "霓虹舞池里的玻璃舞者",
    "titleEn": "A Glass Dancer in a Neon Club",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "透明玻璃质感的舞者在霓虹夜店中完成连续下蹲与侧身动作，人物材质和场景反射保持一致。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A glass-like dancer moves through a continuous crouch and side pose in a neon nightclub while material detail and reflections remain coherent. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/McPzzzy/status/2099274095812849949",
    "sourceLabel": "X 原帖 · @McPzzzy",
    "author": "McPzzzy",
    "publishedAt": "2026-09-13T23:08:41.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099274095812849949.mp4",
    "posterUrl": "/posters/x/2099274095812849949.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max, 768p, 5 sec motion test:\nbuilt the choreography as approved 16:9 stills first, then tested H3 Max in Artlist with 'finally approved' start/end frames. One pair worked beautifully. Harder hand-motion pairs were blocked by Artlist before the model could even try.\n\n\"model boundary ≠ platform boundary ≠ material boundary ≠ prompt boundary\"\n\nArtlist may have saved the world from unrestricted Bubbleglass dancing smh😡🤣 First time with H3 Max, spent more for 768p on curiosity, for nice comparison to some Seedance vids\n#aiart #fantasyart #scifiart #nightclub #cyberdancer\n\nused test prompt for start/end frames, go forth and enhance 🫡:\n\nAnimate a smooth transition from the supplied first Bubbleglass frame to the supplied fifth choreography frame. Preserve the exact same adult performer, face, brown updo, proportions, opaque black bandeau top, opaque black wrap skirt, metallic jewelry, black heeled sandals, translucent living Bubbleglass membrane, nightclub layout, railing on screen-left, seating on screen-right, camera orientation, scale, and lighting. Keep the supporting hand lightly connected to the rigid glass railing as she shifts her weight and lowers into the target pose. Motion should be elegant, controlled, and continuous, with realistic balance and foot placement. Her membrane remains a living translucent material with restrained internal currents and externally driven purple/cyan/red refraction. Clothing remains fully opaque and jewelry remains separate solid metal. No mirror flip, no camera orbit, no zoom, no crop, no face change, no wardrobe change, no extra limbs or fingers, no background rearrangement, no universal glow, no conversion into glass, crystal, liquid, plastic, or hologram. No music. Natural nightclub ambience only if audio is generated.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 279,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099270418842747036",
    "title": "暮色车站前的牵手告别",
    "titleEn": "A Hand-Held Farewell at Dusk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画短片从手表与街景切入，两个青年在列车旁相遇并牵手，把一分钟故事落在安静的告别时刻。 原生视频规格为 60 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated short opens on a watch and city details before two young people meet beside a train and hold hands in a quiet farewell. Native-video output: 60s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/izesaki/status/2099270418842747036",
    "sourceLabel": "X 原帖 · @izesaki",
    "author": "! Ze",
    "publishedAt": "2026-09-13T22:54:05.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099270418842747036.mp4",
    "posterUrl": "/posters/x/2099270418842747036.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "retro"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "cocoknya yang chaos, ga cocok beginian\n\nminimax h3 on @Miora_Design",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 136,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099267920706609585",
    "title": "银幕火山冲出观众席",
    "titleEn": "A Screen Volcano Bursts into the Theater",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "影院银幕中的熔岩与火山喷发逐渐越过画框，橙红火光照亮前排观众，形成沉浸式灾难特效。 原生视频规格为 14 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Lava and a volcanic eruption push beyond the cinema screen as orange light washes across the audience in an immersive disaster effect. Native-video output: 14s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/blizaine/status/2099267920706609585",
    "sourceLabel": "X 原帖 · @blizaine",
    "author": "Blaine Brown",
    "publishedAt": "2026-09-13T22:44:09.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099267920706609585.mp4",
    "posterUrl": "/posters/x/2099267920706609585.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Would you do this? \n\nMaestro Minimax H3 prompt in 🧵",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 21,
      "views": 885,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099265559414120934",
    "title": "夜视镜头捕捉到的林间异象",
    "titleEn": "A Forest Apparition Caught in Night Vision",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "低照度夜视画面里，一个幽暗骑乘身影穿过林地，枝叶与轮廓在有限光源下若隐若现。 原生视频规格为 15 秒 · 1080×1588。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In low-light night-vision footage, a shadowy mounted figure moves through woodland while branches and silhouettes emerge from the darkness. Native-video output: 15s · 1080×1588. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JanineEvie/status/2099265559414120934",
    "sourceLabel": "X 原帖 · @JanineEvie",
    "author": "OnlyWan",
    "publishedAt": "2026-09-13T22:34:46.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099265559414120934.mp4",
    "posterUrl": "/posters/x/2099265559414120934.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1588",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "photoreal"
    ],
    "scenes": [
      "nature",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This has got to be the best I ever made with the Minimax H3 model.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 0,
      "views": 52,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099264362091286732",
    "title": "旅行规划应用的十五秒广告",
    "titleEn": "A 15-Second Travel Planner Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "女子在手机上选择旅行偏好，界面依次展示路线与地图，最后过渡到海岛画面和应用标志。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman selects travel preferences on her phone, the interface reveals a route and map, and the spot resolves into an island scene and app mark. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/neco1751662/status/2099264362091286732",
    "sourceLabel": "X 原帖 · @neco1751662",
    "author": "neco - AI Anime",
    "publishedAt": "2026-09-13T22:30:01.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099264362091286732.mp4",
    "posterUrl": "/posters/x/2099264362091286732.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "画像1枚から15秒のサービス説明CM風動画を作ってみました！\n#DomoAI @DomoAI_ #MiniMaxH3\n\nDomoAIプラットフォーム上のMiniMax H3で、\n架空の旅行プランニングアプリ「SoraNote」を\n・アプリ起動\n・旅行の希望を選択\n・AIがプランを自動生成\n・旅程や地図を表示\n・実際の旅行イメージへ展開\nという流れで動画化。\n\n一部文字破綻は見られるもののかなり良い雰囲気\n使用した画像はリプ欄に✅",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 261,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099258907961270582",
    "title": "巨龙与歌者的两分钟舞台",
    "titleEn": "A Two-Minute Stage for a Dragon and Singer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银色巨龙、黑色流体舞台与歌者特写交替出现，组成一支两分钟的原创歌曲视觉短片。 原生视频规格为 120 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver dragon, liquid-black stage imagery, and singer close-ups alternate throughout a two-minute original-song visual. Native-video output: 120s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/breath_in_code/status/2099258907961270582",
    "sourceLabel": "X 原帖 · @breath_in_code",
    "author": "LinacoWSL",
    "publishedAt": "2026-09-13T22:08:20.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099258907961270582.mp4",
    "posterUrl": "/posters/x/2099258907961270582.jpg",
    "duration": 120,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "photoreal"
    ],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MINIMAX H3　りなこオリジナル曲 MV Test ＃3 \nXのプレミアム会員ではないので💦今回は2分間の曲作りから仕切り直し😂\n\n相変わらず480p 4stepでのテストですが、FACE Refine & DLSS5で解像度を少し上げてます😅\n\n週明けの月曜日🌤️また今週も頑張りましょ💪",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 41,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099253690146451824",
    "title": "脑内会议争夺周末",
    "titleEn": "A Committee Meeting Inside the Mind",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手绘人物的脑中召开一场吵闹会议，睡觉、工作和娱乐等念头用大字与便签争夺控制权。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hand-drawn character hosts a noisy internal meeting where sleep, work, and play compete for control through bold labels and sticky notes. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2099253690146451824",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-09-13T21:47:36.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099253690146451824.mp4",
    "posterUrl": "/posters/x/2099253690146451824.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "motion-graphics"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「頭の中、会議中」\nブラウザ盤ChatGPTにFalのプラグインあるので接続して作ってみた。MInimax H3 Max Turbo\nT2Vなら問題なく動画が出来ます。ただI2vはなかなかうまくいかない。Gooleドライブに格納したりしてやってみたのだが…なにかいい方法はないのかな？\nプロンプトはリプ欄に👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 6,
      "views": 346,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099252098411938181",
    "title": "屠龙战场上的第一视角冲锋",
    "titleEn": "A First-Person Charge into a Dragon Battle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "第一视角战士持斧穿过燃烧的中世纪战场，巨龙从城墙上空俯冲，镜头保持连续冲锋感。 原生视频规格为 35 秒 · 1912×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A first-person warrior carries an axe through a burning medieval battlefield while a dragon dives over the walls in one continuous charge. Native-video output: 35s · 1912×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PatPat134496/status/2099252098411938181",
    "sourceLabel": "X 原帖 · @PatPat134496",
    "author": "Pat Pat",
    "publishedAt": "2026-09-13T21:41:17.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099252098411938181.mp4",
    "posterUrl": "/posters/x/2099252098411938181.jpg",
    "duration": 35,
    "aspectRatio": "landscape",
    "resolution": "1912×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Text2Video, ComfyUI-H3-Motion-Context-MultiRef, av extension workflow\n#Comfyui #MinimaxH3 #aivideo",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 43,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099245008805040251",
    "title": "落日海岸的角色独白",
    "titleEn": "A Character Monologue at Sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金色海岸与人物面部近景交替，泪光和缓慢闭眼把九秒片段收束成一段克制的情绪戏。 原生视频规格为 9 秒 · 1344×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Golden shoreline views alternate with close-ups as tears and a slow eye-close turn nine seconds into a restrained emotional beat. Native-video output: 9s · 1344×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SmallKino_/status/2099245008805040251",
    "sourceLabel": "X 原帖 · @SmallKino_",
    "author": "SmallKino",
    "publishedAt": "2026-09-13T21:13:06.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099245008805040251.mp4",
    "posterUrl": "/posters/x/2099245008805040251.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1344×928",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "retro"
    ],
    "scenes": [
      "dialogue",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Another Senki Survival Saga clip. Anastasia this time.\nInject a joke about English dub miscast\n\n#戦姫サバイバルサガ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 65,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099243447295062114",
    "title": "黑色皮革造型街拍",
    "titleEn": "Black Leather Street Fashion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色皮衣模特从鞋履特写走入城市走廊，镜头逐步抬升到全身造型，呈现简洁的时尚广告节奏。 原生视频规格为 10 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-leather model moves from a shoe close-up into an urban passage as the camera rises to reveal the complete fashion look. Native-video output: 10s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UncleMonchain/status/2099243447295062114",
    "sourceLabel": "X 原帖 · @UncleMonchain",
    "author": "Uncle M Onchain",
    "publishedAt": "2026-09-13T21:06:54.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099243447295062114.mp4",
    "posterUrl": "/posters/x/2099243447295062114.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "10 seconds With 2$ Cost and took 3 minutes to make .\nWorkin on some Fashion Ai Stuff with Real Models\nMade by @higgsfield @higgsfield_ai and @Hailuo_AI Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 102,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099241437565223251",
    "title": "红黑地铁里的无声裁决",
    "titleEn": "Silent Verdict in a Red-and-Black Subway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红黑色动画镜头在地铁、追击者眼神与制服人物之间切换，构成一分钟的暗黑悬疑片段。 原生视频规格为 60 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Red-and-black animated shots move between a subway, a pursuer’s stare, and a uniformed figure in a one-minute noir sequence. Native-video output: 60s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/izesaki/status/2099241437565223251",
    "sourceLabel": "X 原帖 · @izesaki",
    "author": "! Ze",
    "publishedAt": "2026-09-13T20:58:55.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099241437565223251.mp4",
    "posterUrl": "/posters/x/2099241437565223251.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "SILENT VERDICT\n\nmade with minimax h3 on @Miora_Design",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 190,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099240686461604159",
    "title": "阳光推动的音乐盒时钟",
    "titleEn": "A Music Box Clocked by Sunlight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上弦音乐盒停在空房间，窗边光束缓慢扫过木地板，盒中舞者随日光移动完成一天的时间感。 原生视频规格为 15 秒 · 4096×2340。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A wound music box sits in an empty room while a sunbeam crosses the floor and the tiny dancer marks the passing day. Native-video output: 15s · 4096×2340. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the sunbeam on the floor sweeps steadily from left to right across the room, the light warming from pale to gold",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2099240686461604159",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2099240686461604159",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-13T20:55:56.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099240686461604159.mp4",
    "posterUrl": "/posters/x/2099240686461604159.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "4096×2340",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cinematic-realistic",
      "retro"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hailuo H3. One song, one day. The box winds down with the sun.\n\nPrompt: \"the sunbeam on the floor sweeps steadily from left to right across the room, the light warming from pale to gold\"\n\nDuration: 15s. Camera: locked, room interior.\n\nTwo clocks again, the crank and the sun. Both written as continuous. \"Never jumps\" is the sentence that keeps them from cutting to noon.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 161,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099235701539664074",
    "title": "精灵把照片变成一束玫瑰",
    "titleEn": "A Genie Turns a Portrait into Roses",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一对人物在室内互动，画面中的女性肖像逐渐化作一束玫瑰，再回到温暖的现实空间。 原生视频规格为 25 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two people interact indoors as a woman’s portrait transforms into a bouquet of roses before returning to a warm domestic scene. Native-video output: 25s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wlctv_ca/status/2099235701539664074",
    "sourceLabel": "X 原帖 · @wlctv_ca",
    "author": "wlctv.ca",
    "publishedAt": "2026-09-13T20:36:07.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099235701539664074.mp4",
    "posterUrl": "/posters/x/2099235701539664074.jpg",
    "duration": 25,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "daily-life",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Our Minimax H3 A.I Genie story continues .. Hope it brings a smile to your face & warms your heart's.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 69,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099226133904646654",
    "title": "织机把纱线织成字母",
    "titleEn": "A Loom Weaves Letters from Thread",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "木制织机的经纬线在三个镜头里逐渐形成清晰字母，突出纤维质感与文字生成的连续性。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Threads on a wooden loom gradually form readable letters across three shots, emphasizing fiber texture and textual continuity. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2099226133904646654",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-09-13T19:58:06.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099226133904646654.mp4",
    "posterUrl": "/posters/x/2099226133904646654.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Woven on a loom, letter by letter.\n\n→ MiniMax H3 Max in @krea_ai",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 593,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099169296760922389",
    "title": "一滴咖啡长成巴黎",
    "titleEn": "A Drop of Coffee Grows into Paris",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "咖啡滴落后化成液态皇冠，随后扩展成埃菲尔铁塔与整座城市，最终落在晨光街道。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A coffee drop becomes a liquid crown, expands into the Eiffel Tower and a city, then resolves into a sunlit Paris street. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FutureVibesAi/status/2099169296760922389",
    "sourceLabel": "X 原帖 · @FutureVibesAi",
    "author": "Future Vibes AI - Educator",
    "publishedAt": "2026-09-13T16:12:15.000Z",
    "addedAt": "2026-09-14T00:34:44.676Z",
    "mediaUrl": "/media/x-2099169296760922389.mp4",
    "posterUrl": "/posters/x/2099169296760922389.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal",
      "stylized"
    ],
    "scenes": [
      "city",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One drop of coffee... built an entire city. 🇫🇷🤯\nMinimax H3 in @Hailuo_AI Design Hub\nPrompt Below 👇\nYour result might be even better than mine. Try it and share your creation👇\n#MinimaxH3",
    "engagement": {
      "replies": 16,
      "reposts": 0,
      "likes": 23,
      "views": 2672,
      "snapshotAt": "2026-09-14T00:34:44.676Z"
    },
    "approvedAt": "2026-09-14T00:34:44.676Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099163692822909338",
    "title": "人物与城市在画面中折叠",
    "titleEn": "People and Cities Fold into One Frame",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏画面把人物、古迹、动物与地景连续叠合，多个空间像纸片一样在同一角色身上展开。 原生视频规格为 8 秒 · 768×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical montage layers a person, monuments, animals, and landscapes as multiple spaces unfold across a single figure. Native-video output: 8s · 768×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/funyakinsan/status/2099163692822909338",
    "sourceLabel": "X 原帖 · @funyakinsan",
    "author": "ふにゃきん",
    "publishedAt": "2026-09-13T15:49:59.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099163692822909338.mp4",
    "posterUrl": "/posters/x/2099163692822909338.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "768×1152",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "stylized",
      "cg-3d"
    ],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 12,
      "views": 154,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099163001568010381",
    "title": "霓虹街头的深夜驾驶",
    "titleEn": "A Late-Night Drive through Neon Streets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫角色驾驶黑色跑车穿过霓虹街区，镜头在车窗、侧身和城市反射间保持一致的夜行氛围。 原生视频规格为 10 秒 · 512×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime character drives a black sports car through neon streets as window views, side angles, and city reflections hold a consistent night mood. Native-video output: 10s · 512×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YamataZen/status/2099163001568010381",
    "sourceLabel": "X 原帖 · @YamataZen",
    "author": "Yamata",
    "publishedAt": "2026-09-13T15:47:14.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099163001568010381.mp4",
    "posterUrl": "/posters/x/2099163001568010381.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "512×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized",
      "ugc"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "If she can drive, she is an adult woman\n\n#aivideo #MiniMaxH3 #OC",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 74,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099157567327080504",
    "title": "复古酒吧里的危险邀约",
    "titleEn": "A Dangerous Invitation in a Retro Bar",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "复古酒吧女主人端起酒杯、放下子弹并逐步逼近镜头，温暖灯光下的欢迎姿态慢慢转成冷静威胁。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A retro bar hostess lifts a glass, reveals a bullet, and advances toward the camera as a warm welcome turns into a controlled threat. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photoreal image-to-video short set entirely inside the same vintage American bar interior. Preserve the exact same adult woman throughout: same face, proportions, hairstyle, makeup, dress, body, and realistic texture. The whole narrative stays in one continuous bar space only: black-and-white checker floor, red leather stools, metal counter edges, condensation on glass, warm hanging practicals, cool window edge light, faint red EXIT sign near the doorway, and a light haze of smoke and steam. No exterior, no extra people, no spatial jump, no identity drift. She is a mature retro bar hostess aura: powerful, sensual, elegant, dangerous, never vulgar, never static. The hidden story is that she seems to welcome a late guest, but is actually waiting to judge someone already marked. The camera follows one smooth line: low glide inward, rising follow, gentle half-orbit, drop to the table to find the prop, lift back to the face, lateral follow as she closes in, dangerous hand close-up, then a slight push-in ending. Lighting stays physically sourced: warm tungsten practicals and table reflections with cool window contour light. At 00:00.000-00:02.600, the camera glides low over checker tiles, stool legs, red leather, faint steam, and warm reflections. She walks in from behind the bar. Her boots enter first, steps slow and precise. Her right hand hooks a glass of amber liquor from the counter in one natural motion. Do not reveal the full face yet; show jawline, underside of the lip line, half a side face, and body rhythm through hips and waist. Her first layer is calm allure: relaxed brow peaks, outer eye corners slightly lowered, gaze not yet locking the camera, mouth corners lifted only faintly, breathing calm, chin slightly raised. The camera rises naturally with her. [Shot 2] At 00:02.600, the camera travels up from the wrist holding the glass into the key sensual close-up. Red lips, breath, collarbone, side face, outer eye, fingertips, and glass rim must read as beautiful, restrained, and dangerous. While walking, she brings the glass near her lips without drinking, letting the rim pass just in front of her mouth. Here the first micro-shift appears: her gaze hooks toward the camera direction but not into direct eye contact, sweeping lightly from the eye corner and withdrawing. Upper eyelids lower slightly. Lips part only a little. Breath exits softly through the nose. She says quietly in English: <d>[English] Took you long enough.</d> The voice is low, soft, slow, like a warm greeting with a hidden needle. The gentle half-orbit flows into placing the glass down. [Shot 3] At 00:05.100, she reaches the table and sets the glass down lightly. The camera follows the hand to the tabletop and discovers, in the same motion, a single upright bullet standing beside the glass. The bullet must feel native to the table arrangement. Do not force a facial reaction. Let her body shift slightly away while attention rests on the glass and bullet. The hostess softness still lingers but begins withdrawing. Her fingertips stay against the glass for half a beat longer, then leave slightly slower than before, deliberately allowing the danger to be seen. The bullet tip catches a thin cool highlight while the liquor remains warmly lit. [Shot 4] At 00:07.000, the camera rises from the table and bullet back to her face. She has already turned half her body and now stands between the table and a bar stool, with the doorway and blurred red EXIT sign behind. She looks briefly toward the door, then circles one stool and steps slowly closer. This is the key facial reversal: the shallow smile does not vanish first; the eyes cool first while the mouth keeps its curve. The brow center gathers slightly, softness at the eye corners tightens, upper lids lower a little, pupils become more concentrated, jawline tucks slightly more, breathing shallows, and the smile now reads as control rather than invitation. The camera moves laterally and slowly pushes in, keeping her as the active force. [Shot 5] At 00:10.200, cut naturally from her passing motion into a dangerous hand close-up. Her fingertips brush along the table edge and, with exquisite restraint, she gently tips the upright bullet over. It rolls half a turn and stops. The action feels practiced, casual, sensual, and colder than a gunshot. Her face can drift through soft focus in the background. She does not flick it; she nudges it with the pad and edge of the finger. Shoulder relaxed, wrist fluid, action continuous with the walk. The brass surface carries a moving highlight streak over the warm tabletop reflection. [Shot 6] At 00:12.800, the camera lifts from the bullet to a frontal close shot. She has stepped close and stopped, her face dominating frame while the background dissolves into warm lamps and blurred red EXIT color. She tilts her head by the smallest amount and lowers her chin slightly, finally looking directly into the camera. This is the judgment state. The flirtation is gone. Her eyes are steady and direct, never wide. The brow center compresses only slightly. The lips relax and close softly, with one corner holding the faintest cold smile. Breath is contained. Throat and jaw remain composed. She says in English: <d>[English] Tonight, you pay.</d> “Tonight” lands lightly; “you pay” drops lower, calm and judicial, never shouted. In the last half-second, the camera pushes in only slightly more and stops, letting the pressure come from distance rather than effects, with no black screen or transition.\n\noverall_soundscape: Keep the sound intimate and spatially close: heels on tile, soft steam hiss, glass touching metal, leather and fabric friction, low bar hum, faint ventilation, and light room air. Bring breath, glass-rim movement, and lip-near glass tension forward under <d>[English] Took you long enough.</d>; make the glass touching the table the first rhythm point, then the bullet tip-over and half-roll the second. In the ending, reduce everything slightly so <d>[English] Tonight, you pay.</d> sits close and dry while low bar hum, tiny glass resonance, ventilation, and controlled breath remain underneath.\n\nnon_diegetic_music: Use minimal score. A very low elegant noir-like tension bed may sit beneath the space, but it must never dominate heels, glass, breath, or the bullet. Let the first half feel warmer and more seductive, cool it gradually through the middle, and support the final line only with a faint low pressure note, avoiding melodrama or any trailer-style punch.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2099157567327080504",
    "sourceUrl": "https://x.com/ou_zhen599/status/2099157567327080504",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-13T15:25:39.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099157567327080504.mp4",
    "posterUrl": "/posters/x/2099157567327080504.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "retro"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI            \n\n🎬Image-to-Video     \n📷“Who Is She Waiting For?”\nIs she waiting for a guest, or is she waiting for prey?\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photoreal image-to-video short set entirely inside the same vintage American bar interior. Preserve the exact same adult woman throughout: same face, proportions, hairstyle, makeup, dress, body, and realistic texture. The whole narrative stays in one continuous bar space only: black-and-white checker floor, red leather stools, metal counter edges, condensation on glass, warm hanging practicals, cool window edge light, faint red EXIT sign near the doorway, and a light haze of smoke and steam. No exterior, no extra people, no spatial jump, no identity drift. She is a mature retro bar hostess aura: powerful, sensual, elegant, dangerous, never vulgar, never static. The hidden story is that she seems to welcome a late guest, but is actually waiting to judge someone already marked. The camera follows one smooth line: low glide inward, rising follow, gentle half-orbit, drop to the table to find the prop, lift back to the face, lateral follow as she closes in, dangerous hand close-up, then a slight push-in ending. Lighting stays physically sourced: warm tungsten practicals and table reflections with cool window contour light. At 00:00.000-00:02.600, the camera glides low over checker tiles, stool legs, red leather, faint steam, and warm reflections. She walks in from behind the bar. Her boots enter first, steps slow and precise. Her right hand hooks a glass of amber liquor from the counter in one natural motion. Do not reveal the full face yet; show jawline, underside of the lip line, half a side face, and body rhythm through hips and waist. Her first layer is calm allure: relaxed brow peaks, outer eye corners slightly lowered, gaze not yet locking the camera, mouth corners lifted only faintly, breathing calm, chin slightly raised. The camera rises naturally with her. [Shot 2] At 00:02.600, the camera travels up from the wrist holding the glass into the key sensual close-up. Red lips, breath, collarbone, side face, outer eye, fingertips, and glass rim must read as beautiful, restrained, and dangerous. While walking, she brings the glass near her lips without drinking, letting the rim pass just in front of her mouth. Here the first micro-shift appears: her gaze hooks toward the camera direction but not into direct eye contact, sweeping lightly from the eye corner and withdrawing. Upper eyelids lower slightly. Lips part only a little. Breath exits softly through the nose. She says quietly in English: <d>[English] Took you long enough.</d> The voice is low, soft, slow, like a warm greeting with a hidden needle. The gentle half-orbit flows into placing the glass down. [Shot 3] At 00:05.100, she reaches the table and sets the glass down lightly. The camera follows the hand to the tabletop and discovers, in the same motion, a single upright bullet standing beside the glass. The bullet must feel native to the table arrangement. Do not force a facial reaction. Let her body shift slightly away while attention rests on the glass and bullet. The hostess softness still lingers but begins withdrawing. Her fingertips stay against the glass for half a beat longer, then leave slightly slower than before, deliberately allowing the danger to be seen. The bullet tip catches a thin cool highlight while the liquor remains warmly lit. [Shot 4] At 00:07.000, the camera rises from the table and bullet back to her face. She has already turned half her body and now stands between the table and a bar stool, with the doorway and blurred red EXIT sign behind. She looks briefly toward the door, then circles one stool and steps slowly closer. This is the key facial reversal: the shallow smile does not vanish first; the eyes cool first while the mouth keeps its curve. The brow center gathers slightly, softness at the eye corners tightens, upper lids lower a little, pupils become more concentrated, jawline tucks slightly more, breathing shallows, and the smile now reads as control rather than invitation. The camera moves laterally and slowly pushes in, keeping her as the active force. [Shot 5] At 00:10.200, cut naturally from her passing motion into a dangerous hand close-up. Her fingertips brush along the table edge and, with exquisite restraint, she gently tips the upright bullet over. It rolls half a turn and stops. The action feels practiced, casual, sensual, and colder than a gunshot. Her face can drift through soft focus in the background. She does not flick it; she nudges it with the pad and edge of the finger. Shoulder relaxed, wrist fluid, action continuous with the walk. The brass surface carries a moving highlight streak over the warm tabletop reflection. [Shot 6] At 00:12.800, the camera lifts from the bullet to a frontal close shot. She has stepped close and stopped, her face dominating frame while the background dissolves into warm lamps and blurred red EXIT color. She tilts her head by the smallest amount and lowers her chin slightly, finally looking directly into the camera. This is the judgment state. The flirtation is gone. Her eyes are steady and direct, never wide. The brow center compresses only slightly. The lips relax and close softly, with one corner holding the faintest cold smile. Breath is contained. Throat and jaw remain composed. She says in English: <d>[English] Tonight, you pay.</d> “Tonight” lands lightly; “you pay” drops lower, calm and judicial, never shouted. In the last half-second, the camera pushes in only slightly more and stops, letting the pressure come from distance rather than effects, with no black screen or transition.\n\noverall_soundscape: Keep the sound intimate and spatially close: heels on tile, soft steam hiss, glass touching metal, leather and fabric friction, low bar hum, faint ventilation, and light room air. Bring breath, glass-rim movement, and lip-near glass tension forward under <d>[English] Took you long enough.</d>; make the glass touching the table the first rhythm point, then the bullet tip-over and half-roll the second. In the ending, reduce everything slightly so <d>[English] Tonight, you pay.</d> sits close and dry while low bar hum, tiny glass resonance, ventilation, and controlled breath remain underneath.\n\nnon_diegetic_music: Use minimal score. A very low elegant noir-like tension bed may sit beneath the space, but it must never dominate heels, glass, breath, or the bullet. Let the first half feel warmer and more seductive, cool it gradually through the middle, and support the final line only with a faint low pressure note, avoiding melodrama or any trailer-style punch.",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 26,
      "views": 1132,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099156224973664686",
    "title": "红纱舞者展开清晨",
    "titleEn": "A Red-Silk Dancer Opens the Morning",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红裙舞者在柔和舞台上旋转，飘带逐渐扩展成环形云雾，最后化作打开的红色帷幕与晨光标题。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dancer in red spins as silk expands into a ring of mist, ending as parted curtains and a morning title. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2099156224973664686",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-09-13T15:20:19.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099156224973664686.mp4",
    "posterUrl": "/posters/x/2099156224973664686.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Let the morning unfold gently…\n\nWishing you a beautiful Sunday 🩵🩵\n\nCreated with MiniMax H3 on @PixVerse    #pixversecpp",
    "engagement": {
      "replies": 4,
      "reposts": 2,
      "likes": 13,
      "views": 344,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099153639860547925",
    "title": "气球般膨胀的巨龙",
    "titleEn": "A Dragon Inflates Like a Balloon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画战斗中的蓝色巨龙被不断吹大，身体从强壮形态夸张膨胀，形成带声音的喜剧式变形。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue dragon in an anime battle inflates from a muscular form into an exaggerated balloon-like body in a voiced comic transformation. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chasinglightX64/status/2099153639860547925",
    "sourceLabel": "X 原帖 · @chasinglightX64",
    "author": "chasinglight",
    "publishedAt": "2026-09-13T15:10:02.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099153639860547925.mp4",
    "posterUrl": "/posters/x/2099153639860547925.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "creatures",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hmm... Minimax H3 does a pretty good job visually, but the audio is a weak point; the groaning sound the dragon makes when inflating isn't great, haha. Has ref pictures",
    "engagement": {
      "replies": 0,
      "reposts": 11,
      "likes": 76,
      "views": 4475,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099147956922228753",
    "title": "雨夜停车场里的追逐预演",
    "titleEn": "A Chase Rehearsal in a Rainy Car Park",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣男子在湿漉漉的停车场走向镜头并突然加速，灯光与地面反射营造出动作片开场。 原生视频规格为 15 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in black approaches through a wet car park and suddenly accelerates as overhead lights and reflections set up an action-scene opening. Native-video output: 15s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/endermimtas/status/2099147956922228753",
    "sourceLabel": "X 原帖 · @endermimtas",
    "author": "Ender Mimtaş",
    "publishedAt": "2026-09-13T14:47:27.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099147956922228753.mp4",
    "posterUrl": "/posters/x/2099147956922228753.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Picsart sağolsun, kendisine dahil ettiği minimax h3 max modelini ilk denemem ve sonuçları 👇",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 179,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099146102536830982",
    "title": "机甲维修间里的中文对话",
    "titleEn": "Chinese Dialogue inside a Mecha Workshop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "机械维修间中，少女与大型机甲连续互动并说出中文对白，分镜在动作、细节和角色特写间切换。 原生视频规格为 25 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Inside a mechanical workshop, a young woman interacts with a large mecha and speaks Chinese as shots move between action and close-up detail. Native-video output: 25s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FiniYang/status/2099146102536830982",
    "sourceLabel": "X 原帖 · @FiniYang",
    "author": "Fini.Yang",
    "publishedAt": "2026-09-13T14:40:05.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099146102536830982.mp4",
    "posterUrl": "/posters/x/2099146102536830982.jpg",
    "duration": 25,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "电脑修好了，又可以愉快的折腾了 😎\n给本地 MiniMax H3 产线升了个级\nRef2VA INT8 + PDD 8 步蒸馏\n本来想把抢两天的探索图片做视频的\n但是怕限流（兄弟们懂得）\n还是拿它做了机甲动漫了🤣\n\nGPT 画宫格分镜\nH3 生成视频、中文对白和音效\n后期剪辑加字幕\n\n这几次听下来，声音干净不少\n但跨镜音色、语气和机甲细节还会飘\n能出片了，离稳定出片还得磨\n\n配置细节 ⬇️\n#h3 #aigc",
    "engagement": {
      "replies": 14,
      "reposts": 1,
      "likes": 10,
      "views": 1382,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099141476152377400",
    "title": "储物柜前的校园角色镜头",
    "titleEn": "A School Character before the Lockers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "校园角色从储物柜旁转身面对镜头，动作从侧身站姿过渡到近景，面部与服装保持统一。 原生视频规格为 5 秒 · 608×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A school character turns from a locker-lined wall toward the camera, moving from a side pose to a consistent close-up. Native-video output: 5s · 608×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nekota_nekozou/status/2099141476152377400",
    "sourceLabel": "X 原帖 · @nekota_nekozou",
    "author": "🐈猫田 猫蔵",
    "publishedAt": "2026-09-13T14:21:42.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099141476152377400.mp4",
    "posterUrl": "/posters/x/2099141476152377400.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "608×1056",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "妄想しました\n\n#AIart #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 103,
      "views": 1548,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099140152606302265",
    "title": "袖珍护士送来深夜照料",
    "titleEn": "A Tiny Nurse Brings Late-Night Care",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "迷你护士提着购物袋进入房间，被人物抱起后切到床边休息，完成一段轻松的日常动画。 原生视频规格为 12 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny nurse enters with shopping bags, is picked up, and settles beside the bed in a lighthearted domestic animation. Native-video output: 12s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ShimoNeko_desu/status/2099140152606302265",
    "sourceLabel": "X 原帖 · @ShimoNeko_desu",
    "author": "╚下╗(スナネコ帝国 EX+vMAX 下ネコ)",
    "publishedAt": "2026-09-13T14:16:27.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099140152606302265.mp4",
    "posterUrl": "/posters/x/2099140152606302265.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "daily-life",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "お～…MiniMax H3弄ってますけど、着替えシーンもこれくらいの精度で出るの凄いのにぇ…。",
    "engagement": {
      "replies": 3,
      "reposts": 5,
      "likes": 25,
      "views": 1065,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099138188552061007",
    "title": "秋日台阶上的粉发女孩",
    "titleEn": "A Pink-Haired Girl on Autumn Steps",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发女孩坐在落叶铺满的石阶上，镜头缓慢靠近面部，柔和秋光和背景细节保持稳定。 原生视频规格为 11 秒 · 1056×1408。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired girl sits on leaf-covered stone steps as the camera moves gently toward her face under soft autumn light. Native-video output: 11s · 1056×1408. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YumesakiSakura/status/2099138188552061007",
    "sourceLabel": "X 原帖 · @YumesakiSakura",
    "author": "夢咲さくら",
    "publishedAt": "2026-09-13T14:08:38.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099138188552061007.mp4",
    "posterUrl": "/posters/x/2099138188552061007.jpg",
    "duration": 11,
    "aspectRatio": "portrait",
    "resolution": "1056×1408",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今夜の動画は日頃と少し趣の違う秋の風景です🍂\nminimax H3で作成しました。まだうまく作れなくて、背景が少し乱れちゃいましたけど…さくらはうまく秋の風景に溶け込めてるかな。\n\nMusic: CMSL (Classical Music Sound Library)\nF.Chopin：Nocturn No.2 in E-flat major, Op.9-2",
    "engagement": {
      "replies": 0,
      "reposts": 7,
      "likes": 26,
      "views": 1484,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099137211190882513",
    "title": "热酱瓶身文字保持清晰",
    "titleEn": "A Hot-Sauce Label Stays Readable",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "自创热酱瓶置于红椒与蒸汽之间，镜头缓慢推进，主品牌文字在五秒内保持可读。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An invented hot-sauce bottle sits among red peppers and steam as a slow push-in keeps the main brand word readable for five seconds. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/modelstoreai/status/2099137211190882513",
    "sourceLabel": "X 原帖 · @modelstoreai",
    "author": "model.store",
    "publishedAt": "2026-09-13T14:04:45.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099137211190882513.mp4",
    "posterUrl": "/posters/x/2099137211190882513.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product",
      "food"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max, 5 seconds, 768p, no reference image.\n\nwe invented a hot sauce brand and let the model design the label itself. BRIMBLEFLARE came back correct in every frame. the line under it reads \"HOT TIME DATR\". the line under that isn't words at all.\n\nthe wordmark survives and nothing smaller does. design the label around that and the clip is shippable.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 89,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099135809316409686",
    "title": "墨滴流动写出品牌字样",
    "titleEn": "Flowing Ink Writes a Brand Name",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色墨滴在白底上爆开并沿笔画流动，最终自然凝聚成完整品牌字样，而不是直接淡入文字。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Black ink splashes and flows along letterforms on white before physically resolving into a complete brand name rather than simply fading in. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/azed_ai/status/2099135809316409686",
    "sourceLabel": "X 原帖 · @azed_ai",
    "author": "Amira Zairi",
    "publishedAt": "2026-09-13T13:59:11.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099135809316409686.mp4",
    "posterUrl": "/posters/x/2099135809316409686.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI can literally write with liquid now using Hailuo H3\n\nI turned a single ink drop into “AZED AI,” with every letter physically forming from flowing black ink instead of simply appearing on screen.\n\nFull prompt below:",
    "engagement": {
      "replies": 11,
      "reposts": 6,
      "likes": 77,
      "views": 4008,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099129518141128871",
    "title": "外星舰桥上的三方会谈",
    "titleEn": "A Three-Way Meeting on an Alien Bridge",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人类士兵、爬行动物外星人与蓝色生命体围在透明控制台旁交谈，舰桥光线和角色一致性保持稳定。 原生视频规格为 12 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A human soldier, reptilian alien, and blue humanoid confer around a translucent console with consistent characters and ship lighting. Native-video output: 12s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ascendant_Stoic/status/2099129518141128871",
    "sourceLabel": "X 原帖 · @Ascendant_Stoic",
    "author": "Ascendant Stoic",
    "publishedAt": "2026-09-13T13:34:11.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099129518141128871.mp4",
    "posterUrl": "/posters/x/2099129518141128871.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cg-3d"
    ],
    "scenes": [
      "sci-fi",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Messing with Minimax H3 and trying something more cinematic, it's amazing that one can create a scene like this at home without 3D rendering farms or massive team of CGI artists XD",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 63,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099127138905690520",
    "title": "蒙面猎人对战森林巨兽",
    "titleEn": "A Masked Hunter Fights a Forest Beast",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "蒙面猎人在森林遗迹间闪避触手巨兽，连续动作以跳跃、斩击和落地收束成十五秒战斗。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A masked hunter dodges a tentacled beast through forest ruins, connecting leaps, strikes, and a final landing into a fifteen-second fight. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shimoshimonoroo/status/2099127138905690520",
    "sourceLabel": "X 原帖 · @shimoshimonoroo",
    "author": "しもる",
    "publishedAt": "2026-09-13T13:24:44.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099127138905690520.mp4",
    "posterUrl": "/posters/x/2099127138905690520.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#SlaytheSpire2 からサイレントの動画\n\nChatGPT Astra + Minimax h3\nAstraで開始、中間、最終フレームを指定しMinimaxでI2V\n細かく指定してないのにこのクオリティ。正直ビビりました",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 5,
      "views": 175,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099106132128338036",
    "title": "巨人突袭古镇的赛博忍者",
    "titleEn": "A Cyber Ninja Faces a Giant in an Old Town",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "巨大怪物闯入古镇并压过屋顶，赛博忍者在尘土与激光间迎战，三段连续片段构成长镜头动作场面。 原生视频规格为 45 秒 · 1152×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A giant crashes through an old town while a cyber ninja counters amid dust and laser fire across three connected segments. Native-video output: 45s · 1152×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mix_buchi_/status/2099106132128338036",
    "sourceLabel": "X 原帖 · @mix_buchi_",
    "author": "buchi_",
    "publishedAt": "2026-09-13T12:01:16.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099106132128338036.mp4",
    "posterUrl": "/posters/x/2099106132128338036.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "大魔神 vs. サイバーくノ一\n\nサイバーくノ一の正体は･･･(笑)\n\n#MinimaxH3 Ref2V \nComfyUI-H3-Continuum 15s x 3chunks\nキャラ＆キャラシートはKrea2\n\n#Krea2\n#AImovie",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 873,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099103139068608933",
    "title": "九宫格日光在同一画面流动",
    "titleEn": "Sunlit Panels Move inside One Composition",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "九宫格中的城市、咖啡、昆虫与人物面板各自运动，紫粉色光影把独立小场景统一成一个整体。 原生视频规格为 15 秒 · 2688×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "City, coffee, insect, and character panels animate independently inside a grid while violet light binds them into one composition. Native-video output: 15s · 2688×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/zeng_wt/status/2099103139068608933",
    "sourceLabel": "X 原帖 · @zeng_wt",
    "author": "𝐙𝐞𝐧𝐠 💜",
    "publishedAt": "2026-09-13T11:49:22.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099103139068608933.mp4",
    "posterUrl": "/posters/x/2099103139068608933.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2688×1536",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "city",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just turned a multi-panel grid of GPT 2.5 sunburst visuals into a living motion piece using MiniMax Hailuo H3 Max on @ImagineArt_X. \n\nEvery cell moves independently.\n\nPrompt 👇",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 8,
      "views": 284,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099097005905641788",
    "title": "雪山深处的山神传说",
    "titleEn": "The Mountain God of a Snowbound Valley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "受伤人物穿过暴雪与峡谷，巨大神祇在雾雪中显现，四十多秒镜头把求生与神话意象连在一起。 原生视频规格为 44 秒 · 3380×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An injured traveler crosses blizzard and canyon terrain as a colossal deity emerges through the snow in a forty-second survival myth. Native-video output: 44s · 3380×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tomw852/status/2099097005905641788",
    "sourceLabel": "X 原帖 · @Tomw852",
    "author": "Tom𝕎",
    "publishedAt": "2026-09-13T11:25:00.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099097005905641788.mp4",
    "posterUrl": "/posters/x/2099097005905641788.jpg",
    "duration": 44,
    "aspectRatio": "landscape",
    "resolution": "3380×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "《山神》- The Mountain God\n\n(Workflow details in the comments.)\nCreated with #MiniMaxH3 \nRun locally on an RTX 5070 12GB\nEdited in CapCut\n\n#AIVideo #comfyui",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 8,
      "views": 643,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099095900585116054",
    "title": "从创作低潮到重新开工",
    "titleEn": "From Creative Slump to Starting Again",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "一支配歌短片跟随女子从车内发愁、伏案停滞到重新投入创作，用字幕和生活镜头表达创作者低潮。 原生视频规格为 79 秒 · 544×800。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A song-driven short follows a woman from frustration in a car and stalled desk work to renewed creation, using captions and everyday scenes. Native-video output: 79s · 544×800. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CKao29/status/2099095900585116054",
    "sourceLabel": "X 原帖 · @CKao29",
    "author": "ちゃんK🌸日本を旅する女の子🎀AI美女",
    "publishedAt": "2026-09-13T11:20:36.000Z",
    "addedAt": "2026-09-14T00:34:44.677Z",
    "mediaUrl": "/media/x-2099095900585116054.mp4",
    "posterUrl": "/posters/x/2099095900585116054.jpg",
    "duration": 79,
    "aspectRatio": "portrait",
    "resolution": "544×800",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "music-performance",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「𝕏やめるの？」\n曲：YuE2\n歌詞：Grok\n映像：Krea2　Minimax H3\nモデル：ちゃんK💜\n\n久しぶりの動画生成と音楽生成\nできる限りローカル環境で生成してみました\n画像生成とか動画作ったりとかモチベーションあがらない心情を動画化しときました\n明日これを見てやる気出します\n\nみんな元気づけてくれてありがとう🤗🩵\n\n【手順】\n★：HermesAgentでGrokモデルを使って歌詞作成\n（手動で微修正）\n↓\n★：歌詞に合う形でKrea2で何枚か画像生成\n↓\n★：Minimax H3で１枚の画像につき３動画生成\n↓\n私：いい感じのを選んでFilmoraで手動編集\n↓\n私：足りない部分などをスクショ対応\n↓\n完成\n\n★は基本的にHermesお任せ\n\n歌詞は↓↓↓\n\nだからみんな 𝕏を やめていくの\nもちべーしょんが あがらない\n画像生成も できない\nアイデアが でてこない\nだからみんな 𝕏を やめていくの\n\nポストをしたあと 数字みた\nいんぷが 全然伸びていない\n収益化までの ５０万\n認証済みの 表示数\nいまのままじゃ絶対無理\n\nおもしろいポストが できない\n画面のまえで 手がうごかない\n無理すぎるよ それがいいたい\n才能のはなしは したくない\nいいねはつくよ だけど\n表示はふえない まだこない\nなにを書けばいいか わからない\n\nだからみんな 𝕏を やめていくの\nもちべーしょんが あがらない\n画像生成も できない\nアイデアが でてこない\nだからみんな 𝕏を やめていくの",
    "engagement": {
      "replies": 5,
      "reposts": 7,
      "likes": 51,
      "views": 1264,
      "snapshotAt": "2026-09-14T00:34:44.677Z"
    },
    "approvedAt": "2026-09-14T00:34:44.677Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099629458248585384",
    "title": "车站到卡拉 OK 的加速对照",
    "titleEn": "A Turbo Comparison from Station to Karaoke",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一位校服女孩依次走过车站、街道与卡拉 OK 包间，以并列片段展示加速方案前后的运动和人物一致性。 原生视频规格为 16 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same schoolgirl moves through a station, street, and karaoke room in side-by-side clips comparing motion and character consistency. Native-video output: 16s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ks_0x0/status/2099629458248585384",
    "sourceLabel": "X 原帖 · @ks_0x0",
    "author": "片霧秋(shu)",
    "publishedAt": "2026-09-14T22:40:46.000Z",
    "addedAt": "2026-09-15T00:25:56.064Z",
    "mediaUrl": "/media/x-2099629458248585384.mp4",
    "posterUrl": "/posters/x/2099629458248585384.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3の設定を変えてみてる。蒸留技術を使ったturbo←と非turbo→では、やっぱり非turboの方が自然。ただ生成時間はturboは12分、非turboは49分であり、動画生成時間はturboを使うことで1/4に短縮できた（RTX5060Tiを使用）。まずはturboで全体確認、その後、非turboで最終生成が合理的と思う。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 28,
      "snapshotAt": "2026-09-15T00:25:56.064Z"
    },
    "approvedAt": "2026-09-15T00:25:56.064Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099625944478863524",
    "title": "锁链与钟表环绕的暗黑女王",
    "titleEn": "A Dark Queen among Chains and Clocks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏镜头围绕端坐的暗黑角色展开，锁链、钟表和烟雾在前后景中连续变化。 原生视频规格为 14 秒 · 1080×1888。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical sequence circles a seated dark-fantasy figure as chains, clocks, and smoke evolve across the frame. Native-video output: 14s · 1080×1888. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/WowItsPinkie/status/2099625944478863524",
    "sourceLabel": "X 原帖 · @WowItsPinkie",
    "author": "Pinkielici♡us",
    "publishedAt": "2026-09-14T22:26:48.000Z",
    "addedAt": "2026-09-15T00:25:56.064Z",
    "mediaUrl": "/media/x-2099625944478863524.mp4",
    "posterUrl": "/posters/x/2099625944478863524.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "1080×1888",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The look in her eyes already tells you exactly who holds the power🖤✨\n\nAnimated using MiniMax H3 Turbo on @MageSpace_ \n#Pinkielicious #Magespace #AIart #AIcommunity #MageExclusive",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 98,
      "snapshotAt": "2026-09-15T00:25:56.064Z"
    },
    "approvedAt": "2026-09-15T00:25:56.064Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099614409056485436",
    "title": "追赶逃跑文字的手绘角色",
    "titleEn": "A Hand-Drawn Character Chases Escaping Words",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手绘角色发现文字从页面逃走后一路追赶，最后将它们重新抓回画面。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hand-drawn character chases words that escape from a page and finally catches them back inside the frame. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/su_nagomi/status/2099614409056485436",
    "sourceLabel": "X 原帖 · @su_nagomi",
    "author": "すーさん",
    "publishedAt": "2026-09-14T21:40:58.000Z",
    "addedAt": "2026-09-15T00:25:56.064Z",
    "mediaUrl": "/media/x-2099614409056485436.mp4",
    "posterUrl": "/posters/x/2099614409056485436.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「文字が逃げた」\n手書きポップアート × モーショングラフィック\n#OiiOii のMInimax H3 Maxで動画作成\nプロンプトはリプ欄に👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 264,
      "snapshotAt": "2026-09-15T00:25:56.064Z"
    },
    "approvedAt": "2026-09-15T00:25:56.064Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099614163647754524",
    "title": "皮肤与眼眸的极近距离观察",
    "titleEn": "Extreme Close-Ups of Skin and Eyes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "镜头从皮肤纹理推进到眼睛与嘴唇，以连续微距视角展示细节稳定性。 原生视频规格为 15 秒 · 640×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera moves from skin texture to the eye and lips in a continuous macro study of fine detail. Native-video output: 15s · 640×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIPixLab/status/2099614163647754524",
    "sourceLabel": "X 原帖 · @AIPixLab",
    "author": "AI写真ラボ / AIバーチャルカメラマン",
    "publishedAt": "2026-09-14T21:40:00.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099614163647754524.mp4",
    "posterUrl": "/posters/x/2099614163647754524.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "640×960",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "☑️ #MiniMaxH3 の例 その２ I2V\n\n1⃣ AI美女画像を #GPTImage 2.5 でリアル肌質感化\n2⃣ #ChatGPT でH3用動画プロンプトにアレンジ\n　1の画像をFirst Frameにルックスプロンプト削除\n3⃣ MiniMax H3で動画生成\n\nありえないほぼノーメイクでのマクロ撮影イメージ\n※注 本物のモデルさんはもっとずっと肌綺麗です",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 1055,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099606840757658030",
    "title": "琥珀香氛在液体中成形",
    "titleEn": "An Amber Fragrance Forms in Liquid",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "香料、金色液体与玻璃瓶在暖色光线中逐步汇聚，形成一支精致的香水广告。 原生视频规格为 17 秒 · 1080×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Spices, golden liquid, and a glass bottle converge under warm light in a polished fragrance commercial. Native-video output: 17s · 1080×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Israa_Ali2077/status/2099606840757658030",
    "sourceLabel": "X 原帖 · @Israa_Ali2077",
    "author": "Israa Ali",
    "publishedAt": "2026-09-14T21:10:54.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099606840757658030.mp4",
    "posterUrl": "/posters/x/2099606840757658030.jpg",
    "duration": 17,
    "aspectRatio": "portrait",
    "resolution": "1080×1440",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A soul-reviving hour of a luxury perfume commercial  I made entirely with Minimax design 🖤\n\nKhamrah by Lattafa — from raw ingredients to a crystal bottle.\n\nEvery frame generated. Every transition: real physics.\nNo studio. No crew. No budget.\n\nHere is the full pipeline I used:\n\n5 keyframe images → GPT Image 2.5 \nGolden amber background locked across all frames\nIngredients → Extraction → Blending → Pour → Hero spray\nAnimated with MiniMax H3 at 2K\n\nThe result you see took one afternoon.\n\nThis is what AI cinema looks like in 2026.\n\n🧵 Tools used:\n→ GPT Image 2.5 — keyframe generation\n→ MiniMax H3 — 2K video animation\n→ Suno — French orchestral soundtrack\n\nIf you need the prompt, all you have to do is ask\n                   ꧁🪷🌷🌸🌺🦩꧂\n\n@Hailuo_AI #MiniMaxDesign",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 125,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099599865538076918",
    "title": "户外汉服造型连贯切换",
    "titleEn": "Connected Hanfu Looks Outdoors",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多位汉服人物在园林与山水环境中连续切换，服装纹理和自然光保持清晰。 原生视频规格为 65 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Multiple Hanfu looks transition through gardens and landscapes with clear fabric detail and natural light. Native-video output: 65s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Devozikjr/status/2099599865538076918",
    "sourceLabel": "X 原帖 · @Devozikjr",
    "author": "Devoz | Cinematic AI Lab",
    "publishedAt": "2026-09-14T20:43:11.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099599865538076918.mp4",
    "posterUrl": "/posters/x/2099599865538076918.jpg",
    "duration": 65,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "fashion",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax h3 Hanfu",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 31,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099594174492237961",
    "title": "机器人海盗吹嘘他的决斗",
    "titleEn": "A Robot Pirate Boasts about His Duel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "酒馆女侍拆穿机器人海盗夸张的决斗故事，机械角色用眼灯变化而非口型完成对白。 原生视频规格为 12 秒 · 2560×1472。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A bartender exposes a robot pirate’s exaggerated duel story while his eye lights, rather than his rigid mouth, carry the dialogue. Native-video output: 12s · 2560×1472. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description:\n\n[10-second image-to-video generation]\n\nUse the supplied image as the starting frame and preserve the established characters, tavern environment, clothing, robot designs, bar layout, and warm cinematic lighting.\n\nSTYLE / TONE:\nLive-action cinematic science-fiction pirate adventure with playful Hollywood-style banter. This is a cozy local tavern in a sci-fi world where human and robot pirates casually drink together.\n\nThe conversation is playful and familiar. The robot pirate is extremely boastful, arrogant, swaggering, and proud of his reputation. He genuinely believes he is an exceptional pirate. The bartender finds his exaggerated stories amusing and enjoys teasing him.\n\nCRITICAL ROBOT SPEECH RULE:\nALL robot faces remain mechanically rigid.\nROBOTS NEVER MOVE THEIR MOUTHS OR JAWS WHILE SPEAKING.\nNO robotic lip-sync.\nRobot speech comes from internal electronic speakers.\nWhen a robot speaks, ONLY its glowing blue eyes subtly brighten, pulse, or blink in rhythm with the spoken words.\n\n[00:00–00:04]\n\nThe human bartender stands behind the bar holding her drink and looks directly at the robot pirate across from her.\n\nThe robot pirate (S1) looks directly toward her with confident pirate swagger. He gives a small proud tilt of his head.\n\nHis metal mouth and jaw remain COMPLETELY MOTIONLESS.\n\nHis glowing blue eyes pulse subtly as he boasts in an arrogant, self-satisfied mechanical male voice:\n\n(S1) <d>[English] He drew his blaster. I drew mine faster.</d>\n\n[00:04–00:07]\n\nThe bartender (S2) raises an eyebrow and gives him a playful, skeptical grin. She maintains eye contact with HIM while speaking.\n\nBecause she is human, her lips move naturally and accurately with her dialogue:\n\n(S2) <d>[English] You don't even carry a blaster.</d>\n\n[00:07–00:10]\n\nCut slightly closer on the robot pirate.\n\nBrief comedic pause.\n\nHe confidently leans back just slightly, completely unfazed by being caught exaggerating.\n\nHis mouth and jaw NEVER MOVE.\n\nHis blue eyes brighten with smug confidence as he replies:\n\n(S1) <d>[English] Exactly. And I still won.</d>\n\nThe bartender immediately laughs and shakes her head at his ridiculous arrogance.\n\nEnd on the robot sitting proudly as though his story made perfect sense.\n\nBACKGROUND ROBOTS:\nThe other robot pirates remain background patrons. Their mechanical mouths and jaws remain rigid at ALL times. They may make tiny natural head movements or subtle eye-light changes, but they do not speak during this clip.\n\noverall_soundscape:\nWarm local-tavern ambience, quiet distant patron chatter, occasional glass clinks, subtle mechanical servo noises, natural bartender laughter, and clean internal-speaker robot dialogue. Robot eye lights subtly pulse during their speech.\n\nnon_diegetic_music:\nN/A",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2099594174492237961",
    "sourceUrl": "https://x.com/EndFolding79421/status/2099594174492237961",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-09-14T20:20:34.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099594174492237961.mp4",
    "posterUrl": "/posters/x/2099594174492237961.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "2560×1472",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic",
      "cg-3d"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Krea 2 turbo and Minimax H3 in Comfyui (Upscaled in Topaz AI)....integrated_multimodal_description:\n\n[10-second image-to-video generation]\n\nUse the supplied image as the starting frame and preserve the established characters, tavern environment, clothing, robot designs, bar layout, and warm cinematic lighting.\n\nSTYLE / TONE:\nLive-action cinematic science-fiction pirate adventure with playful Hollywood-style banter. This is a cozy local tavern in a sci-fi world where human and robot pirates casually drink together.\n\nThe conversation is playful and familiar. The robot pirate is extremely boastful, arrogant, swaggering, and proud of his reputation. He genuinely believes he is an exceptional pirate. The bartender finds his exaggerated stories amusing and enjoys teasing him.\n\nCRITICAL ROBOT SPEECH RULE:\nALL robot faces remain mechanically rigid.\nROBOTS NEVER MOVE THEIR MOUTHS OR JAWS WHILE SPEAKING.\nNO robotic lip-sync.\nRobot speech comes from internal electronic speakers.\nWhen a robot speaks, ONLY its glowing blue eyes subtly brighten, pulse, or blink in rhythm with the spoken words.\n\n[00:00–00:04]\n\nThe human bartender stands behind the bar holding her drink and looks directly at the robot pirate across from her.\n\nThe robot pirate (S1) looks directly toward her with confident pirate swagger. He gives a small proud tilt of his head.\n\nHis metal mouth and jaw remain COMPLETELY MOTIONLESS.\n\nHis glowing blue eyes pulse subtly as he boasts in an arrogant, self-satisfied mechanical male voice:\n\n(S1) <d>[English] He drew his blaster. I drew mine faster.</d>\n\n[00:04–00:07]\n\nThe bartender (S2) raises an eyebrow and gives him a playful, skeptical grin. She maintains eye contact with HIM while speaking.\n\nBecause she is human, her lips move naturally and accurately with her dialogue:\n\n(S2) <d>[English] You don't even carry a blaster.</d>\n\n[00:07–00:10]\n\nCut slightly closer on the robot pirate.\n\nBrief comedic pause.\n\nHe confidently leans back just slightly, completely unfazed by being caught exaggerating.\n\nHis mouth and jaw NEVER MOVE.\n\nHis blue eyes brighten with smug confidence as he replies:\n\n(S1) <d>[English] Exactly. And I still won.</d>\n\nThe bartender immediately laughs and shakes her head at his ridiculous arrogance.\n\nEnd on the robot sitting proudly as though his story made perfect sense.\n\nBACKGROUND ROBOTS:\nThe other robot pirates remain background patrons. Their mechanical mouths and jaws remain rigid at ALL times. They may make tiny natural head movements or subtle eye-light changes, but they do not speak during this clip.\n\noverall_soundscape:\nWarm local-tavern ambience, quiet distant patron chatter, occasional glass clinks, subtle mechanical servo noises, natural bartender laughter, and clean internal-speaker robot dialogue. Robot eye lights subtly pulse during their speech.\n\nnon_diegetic_music:\nN/A",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 61,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099587283934687601",
    "title": "花园摆件召开社区会议",
    "titleEn": "Garden Ornaments Hold a Neighborhood Meeting",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "花园矮人与火烈鸟摆件像邻居一样开会争论，动作和表情形成轻松喜剧。 原生视频规格为 15 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Garden gnomes and flamingo ornaments argue like neighbors in a playful animated community meeting. Native-video output: 15s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2099587283934687601",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-09-14T19:53:11.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099587283934687601.mp4",
    "posterUrl": "/posters/x/2099587283934687601.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Mushroom Grove HOA called an emergency meeting after this weekend’s unfortunate events.\n\nApparently, flamingo theft and retaliatory sprinkler use are now officially against the rules. 😂\n\nCreated with MiniMax H3 via @PixVerse  #pixversecpp",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 8,
      "views": 107,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099578063004615118",
    "title": "低多边形预演变成城堡大战",
    "titleEn": "A Low-Poly Previs Becomes a Castle Battle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "简化的军队与城堡预演逐步过渡为写实战场，展示镜头调度到最终画面的转化。 原生视频规格为 33 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A simplified army-and-castle previs transitions into a realistic battlefield, showing the move from blocking to final imagery. Native-video output: 33s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/koldo2k/status/2099578063004615118",
    "sourceLabel": "X 原帖 · @koldo2k",
    "author": "Koldo Huici",
    "publishedAt": "2026-09-14T19:16:33.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099578063004615118.mp4",
    "posterUrl": "/posters/x/2099578063004615118.jpg",
    "duration": 33,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I used the work generated with GPT-6 Astra to build out a set, a story.\n\nNow I'm bringing it to life with @Hailuo_AI  Minimax H3.\n\nHere's a tip that's worked well for me 👇",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 13,
      "views": 913,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099572054479483126",
    "title": "精灵在海滩找到贝壳",
    "titleEn": "An Elf Finds a Shell on the Beach",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "精灵女孩沿海岸寻找目标，最后弯腰拾起贝壳并举到眼前观察。 原生视频规格为 7 秒 · 1760×1184。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An elf girl searches along the shore, bends to pick up a shell, and raises it for a closer look. Native-video output: 7s · 1760×1184. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SmallKino_/status/2099572054479483126",
    "sourceLabel": "X 原帖 · @SmallKino_",
    "author": "SmallKino",
    "publishedAt": "2026-09-14T18:52:40.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099572054479483126.mp4",
    "posterUrl": "/posters/x/2099572054479483126.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1760×1184",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Um... no fish?\n\n#戦姫サバイバルサガ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 79,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099567988529930687",
    "title": "猩红偶像的舞台短片",
    "titleEn": "A Scarlet Idol Stage Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红色造型的女歌手在舞台、走廊和特写之间切换，组成节奏鲜明的偶像视觉片。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A singer in red moves between stage, corridor, and close-up shots in a sharply paced idol visual. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/alahmedse/status/2099567988529930687",
    "sourceLabel": "X 原帖 · @alahmedse",
    "author": "Ali Ahmed",
    "publishedAt": "2026-09-14T18:36:31.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099567988529930687.mp4",
    "posterUrl": "/posters/x/2099567988529930687.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "commercial",
      "stylized"
    ],
    "scenes": [
      "music-performance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#minimaxh3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 17,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099566465464582477",
    "title": "宫殿碎片升入云端城堡",
    "titleEn": "Palace Fragments Rise into Cloud Castles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "宏伟宫殿在镜头中破碎并向天空漂浮，碎片最终连接成云层之上的奇幻建筑。 原生视频规格为 10 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A grand palace fractures and floats skyward, its pieces resolving into fantasy architecture above the clouds. Native-video output: 10s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/georgia_h__/status/2099566465464582477",
    "sourceLabel": "X 原帖 · @georgia_h__",
    "author": "gia 🍂🍁🍊",
    "publishedAt": "2026-09-14T18:30:28.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099566465464582477.mp4",
    "posterUrl": "/posters/x/2099566465464582477.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@openclipsai x Minimax H3\n\nPrompt below ⬇️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 34,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099566384716107809",
    "title": "人物参考图的多场景变化",
    "titleEn": "One Character Reference across Multiple Scenes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "同一人物参考被放入宠物、山景和室内等不同构图，以分屏方式比较多种生成结果。 原生视频规格为 30 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "One character reference appears across pet, mountain, and interior compositions in a split-screen comparison. Native-video output: 30s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jerrod_lew/status/2099566384716107809",
    "sourceLabel": "X 原帖 · @jerrod_lew",
    "author": "Jerrod Lew",
    "publishedAt": "2026-09-14T18:30:08.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099566384716107809.mp4",
    "posterUrl": "/posters/x/2099566384716107809.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "performance",
      "nature"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Multi-shot split screen effect with MiniMax H3 Max.\n\nUpload an image as a reference and prompt for a split screen with differing camera angles.\n\nIt does a pretty good job at keeping consistency with the environment + characters.\n\nSome examples:",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 15,
      "views": 858,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099565966283948502",
    "title": "雨夜棋局引发的城市冲突",
    "titleEn": "A Rainy Chess Match Sparks a City Conflict",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暗色动画在雨夜街道、棋局和火光战斗间连续切换，形成紧张的叙事片段。 原生视频规格为 57 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark animated sequence moves between rainy streets, a chess game, and firelit conflict in a tense narrative. Native-video output: 57s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/izesaki/status/2099565966283948502",
    "sourceLabel": "X 原帖 · @izesaki",
    "author": "! Ze",
    "publishedAt": "2026-09-14T18:28:29.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099565966283948502.mp4",
    "posterUrl": "/posters/x/2099565966283948502.jpg",
    "duration": 57,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "city",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "REQUIRES A\n\nMade with minimax h3 on @Miora_Design",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 267,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099565005955903952",
    "title": "倒下的多米诺骨牌重新站起",
    "titleEn": "A Fallen Domino Stands Back Up",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "单块多米诺骨牌在固定镜头中倒下，停顿后沿底部缓慢旋转并重新直立。 原生视频规格为 15 秒 · 3328×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A single domino falls in a locked shot, pauses, then pivots slowly on its base until it stands upright again. Native-video output: 15s · 3328×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the domino slowly rises back up onto its end, pivoting on its base, until it is upright again",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2099565005955903952",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2099565005955903952",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-14T18:24:40.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099565005955903952.mp4",
    "posterUrl": "/posters/x/2099565005955903952.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3328×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Max. One domino. It falls. Nothing happens. It gets back up.\n\nPrompt: \"the domino slowly rises back up onto its end, pivoting on its base, until it is upright again\"\n\nDuration: 15s. Camera: locked, domino height.\n\n\"No others ever appear\" is the consistency line. Every model wants to give a domino friends.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 9,
      "views": 222,
      "snapshotAt": "2026-09-15T00:25:56.065Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099556444869919204",
    "title": "复古机甲战争蒙太奇",
    "titleEn": "A Retro Mecha War Montage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏动画快速切换飞行、爆炸与大型机甲交战，保持老派机器人动画质感。 原生视频规格为 74 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical animation cuts rapidly through flight, explosions, and large mecha combat in an old-school robot style. Native-video output: 74s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lawson_ste3476/status/2099556444869919204",
    "sourceLabel": "X 原帖 · @lawson_ste3476",
    "author": "太初学社",
    "publishedAt": "2026-09-14T17:50:38.000Z",
    "addedAt": "2026-09-15T00:25:56.065Z",
    "mediaUrl": "/media/x-2099556444869919204.mp4",
    "posterUrl": "/posters/x/2099556444869919204.jpg",
    "duration": 74,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "retro"
    ],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "这次，我把自己的电脑变成了机甲片场。\n\n用 RTX 4070 在本地跑 MiniMax H3，配合 Codex × GPT-6 Astra · Ultra 完成创作流程，再把生成的素材重新剪辑成这支短片。\n\n想做出老派机甲动画的那股劲：钢铁、硝烟，还有战火里彼此牵挂的人。把自己想看的故事亲手做出来，这件事真的很上头。\n\n如果你也能拍一部，最想让哪台机甲当主角？\n\n#高达 #机甲 #AI动画 #AI视频 #minimaxh3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 60,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.065Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099556434933641507",
    "title": "同一人物穿行多座城市",
    "titleEn": "One Character Moves through Multiple Cities",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一位女性在不同城市、服装与光线中连续出现，突出参考人物的一致性。 原生视频规格为 15 秒 · 928×1664。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same woman appears across changing cities, outfits, and lighting, emphasizing reference-character consistency. Native-video output: 15s · 928×1664. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_nontan_room/status/2099556434933641507",
    "sourceLabel": "X 原帖 · @ai_nontan_room",
    "author": "－TAKATO－AI_room",
    "publishedAt": "2026-09-14T17:50:36.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099556434933641507.mp4",
    "posterUrl": "/posters/x/2099556434933641507.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "928×1664",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal",
      "vertical"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "キャバ嬢の日常の軽い失敗写真とさせて頂き、画像を生成\nローカルMiniMaxH3にてR2Vで動画化\nなかなか良い出来に😊",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 43,
      "views": 2241,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099554535681192371",
    "title": "从导演界面到赛博跑车成片",
    "titleEn": "From Director UI to a Cyberpunk Car Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "画面先展示三维导演与镜头设置界面，再切换到霓虹城市中的跑车最终效果。 原生视频规格为 33 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The clip begins with a 3D director and camera interface, then reveals the final sports-car shot in a neon city. Native-video output: 33s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TokenGremlin/status/2099554535681192371",
    "sourceLabel": "X 原帖 · @TokenGremlin",
    "author": "Token Gremlin",
    "publishedAt": "2026-09-14T17:43:03.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099554535681192371.mp4",
    "posterUrl": "/posters/x/2099554535681192371.jpg",
    "duration": 33,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d",
      "motion-graphics"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I built this whole workflow inside MiniMax Design.\n\nThe idea was pretty simple: one futuristic car, one rainy cyberpunk street, and one smooth cinematic camera move.\n\nI started from the MiniMax Design canvas with GPT-6 Astra, using the 3D Director to build the first version of the shot. That gave me a controllable 3D scene where I could work out the basic environment, place the car, shape the composition, set up the camera, and see how the movement would actually feel before worrying about the final look.\n\nThat first version was really just a prototype, but that was the useful part. I could already see the timing, framing, and camera direction in motion instead of trying to describe everything through a prompt.\n\nFrom there, I brought the 3D Director result back into the canvas as a visual and motion reference, then used MiniMax H3 to push it into a much more cinematic final version.\n\nH3 kept the original composition and camera idea, but added the things that really changed the feel of the shot: the finished car design, wet reflections, rain, atmosphere, lighting, and much richer environmental detail.\n\nWhat I liked most about the workflow was being able to separate direction from rendering. Astra and the 3D Director gave me something concrete to control first, and then H3 could focus on turning that foundation into the final image.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 14,
      "views": 968,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099549692585517078",
    "title": "校园精灵的随手拍相册",
    "titleEn": "A Casual Campus Album with an Elf",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏镜头以随手拍质感记录精灵女孩在校园与室内的不同造型和表情。 原生视频规格为 16 秒 · 1440×2542。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical, casual-photo-style montage follows an elf girl through campus and indoor looks and expressions. Native-video output: 16s · 1440×2542. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tomw852/status/2099549692585517078",
    "sourceLabel": "X 原帖 · @Tomw852",
    "author": "Tom𝕎",
    "publishedAt": "2026-09-14T17:23:49.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099549692585517078.mp4",
    "posterUrl": "/posters/x/2099549692585517078.jpg",
    "duration": 16,
    "aspectRatio": "portrait",
    "resolution": "1440×2542",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "anime",
      "ugc"
    ],
    "scenes": [
      "fashion",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Obsessed with this style, so fun and easy to make. Thanks to @8co28 for sharing.\n(Full video prompt in replies⤵️) \n\nCreated with #MiniMaxH3  \nRun locally on an RTX 5070 12GB\n\n Image prompt:  \nCN: 一組業餘失敗照片，3x3，9:16, 校園背景 \n\nEN: A set of amateur, unsuccessful photos, 3x3, 9:16, campus background.\n\n#AIVideo #comfyui",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 0,
      "views": 195,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099545419042758663",
    "title": "彩色超级机器人驰骋荒漠",
    "titleEn": "A Colorful Super Robot Rides through the Desert",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "鲜艳机甲骑着摩托穿过荒漠并进入战斗姿态，呈现复古超级机器人动画节奏。 原生视频规格为 15 秒 · 2496×1408。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A brightly colored mecha rides a motorcycle across the desert and shifts into combat in a retro super-robot rhythm. Native-video output: 15s · 2496×1408. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxCasu/status/2099545419042758663",
    "sourceLabel": "X 原帖 · @MaxCasu",
    "author": "max casu",
    "publishedAt": "2026-09-14T17:06:50.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099545419042758663.mp4",
    "posterUrl": "/posters/x/2099545419042758663.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2496×1408",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "retro"
    ],
    "scenes": [
      "combat",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "When you order a super-robot motorcycle from Wish...\nMade with @ViggleAI MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 74,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099543896833540427",
    "title": "低多边形钓鱼预演转为实景",
    "titleEn": "A Low-Poly Fishing Previs Turns Realistic",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "简化的河岸钓鱼场景逐步转成写实人物与环境，直观展示预演到成片的变化。 原生视频规格为 5 秒 · 1872×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A simplified riverside fishing scene transitions into realistic characters and environment, revealing the path from previs to final shot. Native-video output: 5s · 1872×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fangyex/status/2099543896833540427",
    "sourceLabel": "X 原帖 · @fangyex",
    "author": "愚瞳 | winterx",
    "publishedAt": "2026-09-14T17:00:47.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099543896833540427.mp4",
    "posterUrl": "/posters/x/2099543896833540427.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1872×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I am amazed at how strongly MiniMax H3 Max is able to replicate 3D low-poly camera movement and my intent.\n\nPlay here: https://bezio.ai/lab/3d-scene-chat",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 122,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099543820304060629",
    "title": "灰模城市预演变成未来街区",
    "titleEn": "A Graybox City Previs Becomes a Future District",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "Blender 灰模中的城市镜头与写实未来街区成片并列展示，保留相同的构图和运动。 原生视频规格为 30 秒 · 1080×1232。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A Blender graybox city shot is paired with a realistic future district while preserving composition and camera motion. Native-video output: 30s · 1080×1232. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotoGarrido/status/2099543820304060629",
    "sourceLabel": "X 原帖 · @PhotoGarrido",
    "author": "David Garrido",
    "publishedAt": "2026-09-14T17:00:29.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099543820304060629.mp4",
    "posterUrl": "/posters/x/2099543820304060629.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "1080×1232",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cuando comparto pruebas así suele salir gente enfadada señalando lo que NO funciona bien (que lo hay).\n\nY qué queréis que os diga, yo veo todo lo que sí va bien y cómo estábamos hace unos meses y no paro de imaginar cómo será esto en nada y menos.\n\nAstra+Blender+Minimax H3👇",
    "engagement": {
      "replies": 5,
      "reposts": 0,
      "likes": 15,
      "views": 528,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099540164351562083",
    "title": "女孩迎战巷道里的寿司蜘蛛",
    "titleEn": "A Girl Battles Sushi Spiders in an Alley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "小女孩在昏暗巷道中躲避并反击寿司造型的蜘蛛怪物，动作与怪诞食物设计结合。 原生视频规格为 15 秒 · 2496×1408。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A young girl dodges and fights sushi-shaped spider creatures in a dark alley, mixing action with absurd food-monster design. Native-video output: 15s · 2496×1408. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxCasu/status/2099540164351562083",
    "sourceLabel": "X 原帖 · @MaxCasu",
    "author": "max casu",
    "publishedAt": "2026-09-14T16:45:57.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099540164351562083.mp4",
    "posterUrl": "/posters/x/2099540164351562083.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2496×1408",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "combat",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "New phobia unlocked: Sushi Spiders. 🕷️🍣\nCreated with @ViggleAI MiniMax H3 model",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 6,
      "views": 212,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099532246118285429",
    "title": "三维镜头环绕动漫角色",
    "titleEn": "A 3D Camera Orbits an Anime Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏镜头围绕动漫女性平滑移动，在不同角度保持人物面部与服装一致。 原生视频规格为 5 秒 · 736×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical camera move orbits an anime woman smoothly while maintaining her face and outfit across angles. Native-video output: 5s · 736×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jun_kongo/status/2099532246118285429",
    "sourceLabel": "X 原帖 · @jun_kongo",
    "author": "kongo jun / 混合 順",
    "publishedAt": "2026-09-14T16:14:29.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099532246118285429.mp4",
    "posterUrl": "/posters/x/2099532246118285429.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "736×1280",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 3d-Camera-control",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 3,
      "views": 112,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2099532037216686418",
    "title": "遗迹猎手与黑暗中的凝视",
    "titleEn": "A Ruin Hunter Meets a Gaze in the Dark",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "女探险者涉水进入热带遗迹，发现发光石纹后举枪瞄准瀑布深处，最终与黑豹的眼睛对视。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A female explorer wades into tropical ruins, raises her pistol after carved stone awakens, and ends eye-to-eye with a panther hidden beyond the waterfall. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photoreal image-to-video sequence based on <Picture 1>, unfolding entirely inside the same tropical ruin space: the ancient forecourt, shallow-water stone path, carved pillars, and the dark waterfall opening aligned on one continuous depth axis. Preserve the exact same adult female explorer throughout: same face, proportions, hairstyle, skin tone, mud marks, sweat sheen, outfit, holster, belt, accessories, and handgun. No location change, no exterior vista, no underground hall, no extra people. The tone is elegant, dangerous, sensual, and restrained, like a high-end adventure thriller. Her sensuality comes only from wet skin, collarbone, shoulder strap, breath, step rhythm, and lethal focus, never posing. She remains in motion throughout. The camera follows pseudo-one-take logic with hidden transitions only through reeds, vines, gun-body wipes, or her own body crossing frame. Light is late-morning tropical sun through canopy gaps, with strong side-back and rim-back highlights on damp hair, skin edges, straps, and gun metal, while the waterfall creates dense white mist. [Shot 2] At 00:00.000-00:02.000, begin immediately on her in a close wide-angle body-follow shot from her left-front at collarbone height, retreating smoothly with her wading step. The frame clearly holds damp collarbone, tight shoulder strap on wet skin, strands of wet hair along the neck, the slight rise and fall of controlled breathing, her right hand carrying the handgun low, and her legs cutting through shallow water. She never stops walking. With her left hand, she brushes one wet strand from the side of her neck back toward the ear. The opening feels sensual inside action: collarbone line, strap tension, breathing motion, wet hair against skin, but the emotional tone stays focused. Expression layer one is calm reconnaissance: brows straight with slight inward tension, eyes scan the ground then lift forward, lids slightly lowered, pupils stable, lips parted by a narrow line, breath short and controlled. She says quietly, throat-close: <d>[English] Too quiet.</d> [Shot 3] At 00:02.000, without a hard cut, use her arm and foreground vines as a natural wipe into a left-rear three-quarter medium-close follow shot. Carved pillars and vines cross the foreground. She lowers her body slightly and threads through two carved pillars without stopping, pushing vines and broad leaves aside with her left hand while the handgun stays low but ready in her right. Her emotion shifts from calm into mild doubt: the brow center gathers a little more, her gaze cuts from the stone path and pillar bases toward the waterfall direction, upper lids lift by a fraction, nostrils widen slightly, breathing speeds up while staying controlled, the lip line becomes thinner, and the jaw begins to set. Wet hair swings lightly with movement and plant contact. Leaves rebound, shedding bright backlit droplets. [Shot 4] At 00:05.000, move into a natural perspective from her right-rear at about waist height. She steps onto an ancient carved stone pattern under shallow water. First the water around her boot trembles, then a muted antique gold-cold light appears inside the wet grooves and races along the carved channels for about half a meter before weakening. This is the only supernatural-adjacent effect, and it must feel like an old stone system briefly waking, not neon or sci-fi. She performs a real stop reaction with weight: front foot plants, rear foot searches for support, knees flex, pelvis stabilizes, torso half-turns, and the handgun rises fast from thigh-level to chest-ready. Her expression enters sudden alertness: eyes snap into focus, brow center tightens in one short inward pull, lids open slightly, nostrils take one short intake, lips part for an instant and seal again, jaw locks. Water splashes outward and the glowing grooves cast broken reflections on wet stone before fading rapidly. [Shot 5] At 00:08.000, the camera tracks laterally along her right side in a medium-close profile, keeping her and the dark waterfall opening in the same perspective line. She does not stop. She moves two steps right along the stone pillar, her left hand dragging lightly across the damp carved wall to confirm the path, while the handgun rises and fully locks onto the black space behind the waterfall. Her emotion compresses into hunter-mode focus: gaze pinned ahead, blinking nearly gone, upper lids tightening, lower lids faintly engaged, pupils fixed on the opening, brow center inward but not lifted, nostrils carrying tension, lips sealed, mouth corners slightly downward, jaw visibly harder, and one tiny swallow in the throat. The waterfall spray becomes dense white particles in backlight. The opening remains only deep layered black, with no full animal shape, no face, no visible panther head. [Shot 6] At 00:11.000, shift into a pressure push-in from a half-body armed medium close toward a tighter shot containing her eyes, gun muzzle, shoulder line, and part of the collarbone, while preserving waterfall highlights and mist behind. She locks down her lower body, rear foot driving into the stone, center of gravity sinking, and the gun barrel aligning more precisely with the center of the dark opening. She seems to push herself into her own aim line. Her emotion reaches maximum alert pressure: eyes sharp and almost unblinking, brow slightly lowered inward for a colder lock, nostril tension increasing, lips compressed into a thinner line, jaw and masseter fully engaged. The handgun must feel heavy in wrist, forearm, shoulder, and collarbone line. [Shot 7] At 00:13.000, the camera stops pushing and holds a stable suspended composition. The foreground keeps only her three-quarter side-face silhouette and the handgun muzzle; the waterfall remains midground and the black opening stays in background. In the final two seconds, introduce only one new piece of information: at first the opening contains nothing readable, then only one black panther eye catches a trace of environmental light and returns a muted amber-gold reflection. The eye must not glow actively, must not look sci-fi, and must not read like a jewel; it is simply a wild animal eye catching light inside extreme darkness. The panther body stays hidden in shadow. No full panther face, no second eye, no leap, no attack motion. Water spray and mist continue to cross the frame. End on the suspended relationship: she is locking onto the darkness, and the darkness is looking back.\n\noverall_soundscape: Use only realistic environmental and action sound. In the first half, emphasize shallow-water footsteps, leaves and vines brushing fabric and skin, low breathing, soft cloth friction, and subtle metal contact from the handgun. Add the muted resonance of her boot on carved stone and the brief inner-stone glow cue as a subtle old mineral hum, never electric. In the middle and late section, the waterfall rumble becomes dominant, with mist, damp stone, controlled breathing, and silence carrying the pressure. Let <d>[English] Too quiet.</d> sit low and throat-close. In the final two seconds there is no dialogue and no animal growl, only waterfall, suspended mist, and the pressure of silence around the revealed eye.\n\nnon_diegetic_music: Use no score or only an extremely faint low-frequency pressure bed beginning after the carved-stone activation. It must never overpower water, breath, leaves, or the final suspended reveal. The ending should remain tense, elegant, unresolved, and physically grounded rather than dramatic or explosive.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2099532037216686418",
    "sourceUrl": "https://x.com/ou_zhen599/status/2099532037216686418",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-14T16:13:39.000Z",
    "addedAt": "2026-09-15T00:25:56.066Z",
    "mediaUrl": "/media/x-2099532037216686418.mp4",
    "posterUrl": "/posters/x/2099532037216686418.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI \n                 \n🎬Image-to-Video        \n📷“The Prey”\nShe thought she was “the discoverer,” but in the end, the tables turned, and she became “the one who was being waited for.”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, cinematic photoreal image-to-video sequence based on <Picture 1>, unfolding entirely inside the same tropical ruin space: the ancient forecourt, shallow-water stone path, carved pillars, and the dark waterfall opening aligned on one continuous depth axis. Preserve the exact same adult female explorer throughout: same face, proportions, hairstyle, skin tone, mud marks, sweat sheen, outfit, holster, belt, accessories, and handgun. No location change, no exterior vista, no underground hall, no extra people. The tone is elegant, dangerous, sensual, and restrained, like a high-end adventure thriller. Her sensuality comes only from wet skin, collarbone, shoulder strap, breath, step rhythm, and lethal focus, never posing. She remains in motion throughout. The camera follows pseudo-one-take logic with hidden transitions only through reeds, vines, gun-body wipes, or her own body crossing frame. Light is late-morning tropical sun through canopy gaps, with strong side-back and rim-back highlights on damp hair, skin edges, straps, and gun metal, while the waterfall creates dense white mist. [Shot 2] At 00:00.000-00:02.000, begin immediately on her in a close wide-angle body-follow shot from her left-front at collarbone height, retreating smoothly with her wading step. The frame clearly holds damp collarbone, tight shoulder strap on wet skin, strands of wet hair along the neck, the slight rise and fall of controlled breathing, her right hand carrying the handgun low, and her legs cutting through shallow water. She never stops walking. With her left hand, she brushes one wet strand from the side of her neck back toward the ear. The opening feels sensual inside action: collarbone line, strap tension, breathing motion, wet hair against skin, but the emotional tone stays focused. Expression layer one is calm reconnaissance: brows straight with slight inward tension, eyes scan the ground then lift forward, lids slightly lowered, pupils stable, lips parted by a narrow line, breath short and controlled. She says quietly, throat-close: <d>[English] Too quiet.</d> [Shot 3] At 00:02.000, without a hard cut, use her arm and foreground vines as a natural wipe into a left-rear three-quarter medium-close follow shot. Carved pillars and vines cross the foreground. She lowers her body slightly and threads through two carved pillars without stopping, pushing vines and broad leaves aside with her left hand while the handgun stays low but ready in her right. Her emotion shifts from calm into mild doubt: the brow center gathers a little more, her gaze cuts from the stone path and pillar bases toward the waterfall direction, upper lids lift by a fraction, nostrils widen slightly, breathing speeds up while staying controlled, the lip line becomes thinner, and the jaw begins to set. Wet hair swings lightly with movement and plant contact. Leaves rebound, shedding bright backlit droplets. [Shot 4] At 00:05.000, move into a natural perspective from her right-rear at about waist height. She steps onto an ancient carved stone pattern under shallow water. First the water around her boot trembles, then a muted antique gold-cold light appears inside the wet grooves and races along the carved channels for about half a meter before weakening. This is the only supernatural-adjacent effect, and it must feel like an old stone system briefly waking, not neon or sci-fi. She performs a real stop reaction with weight: front foot plants, rear foot searches for support, knees flex, pelvis stabilizes, torso half-turns, and the handgun rises fast from thigh-level to chest-ready. Her expression enters sudden alertness: eyes snap into focus, brow center tightens in one short inward pull, lids open slightly, nostrils take one short intake, lips part for an instant and seal again, jaw locks. Water splashes outward and the glowing grooves cast broken reflections on wet stone before fading rapidly. [Shot 5] At 00:08.000, the camera tracks laterally along her right side in a medium-close profile, keeping her and the dark waterfall opening in the same perspective line. She does not stop. She moves two steps right along the stone pillar, her left hand dragging lightly across the damp carved wall to confirm the path, while the handgun rises and fully locks onto the black space behind the waterfall. Her emotion compresses into hunter-mode focus: gaze pinned ahead, blinking nearly gone, upper lids tightening, lower lids faintly engaged, pupils fixed on the opening, brow center inward but not lifted, nostrils carrying tension, lips sealed, mouth corners slightly downward, jaw visibly harder, and one tiny swallow in the throat. The waterfall spray becomes dense white particles in backlight. The opening remains only deep layered black, with no full animal shape, no face, no visible panther head. [Shot 6] At 00:11.000, shift into a pressure push-in from a half-body armed medium close toward a tighter shot containing her eyes, gun muzzle, shoulder line, and part of the collarbone, while preserving waterfall highlights and mist behind. She locks down her lower body, rear foot driving into the stone, center of gravity sinking, and the gun barrel aligning more precisely with the center of the dark opening. She seems to push herself into her own aim line. Her emotion reaches maximum alert pressure: eyes sharp and almost unblinking, brow slightly lowered inward for a colder lock, nostril tension increasing, lips compressed into a thinner line, jaw and masseter fully engaged. The handgun must feel heavy in wrist, forearm, shoulder, and collarbone line. [Shot 7] At 00:13.000, the camera stops pushing and holds a stable suspended composition. The foreground keeps only her three-quarter side-face silhouette and the handgun muzzle; the waterfall remains midground and the black opening stays in background. In the final two seconds, introduce only one new piece of information: at first the opening contains nothing readable, then only one black panther eye catches a trace of environmental light and returns a muted amber-gold reflection. The eye must not glow actively, must not look sci-fi, and must not read like a jewel; it is simply a wild animal eye catching light inside extreme darkness. The panther body stays hidden in shadow. No full panther face, no second eye, no leap, no attack motion. Water spray and mist continue to cross the frame. End on the suspended relationship: she is locking onto the darkness, and the darkness is looking back.\n\noverall_soundscape: Use only realistic environmental and action sound. In the first half, emphasize shallow-water footsteps, leaves and vines brushing fabric and skin, low breathing, soft cloth friction, and subtle metal contact from the handgun. Add the muted resonance of her boot on carved stone and the brief inner-stone glow cue as a subtle old mineral hum, never electric. In the middle and late section, the waterfall rumble becomes dominant, with mist, damp stone, controlled breathing, and silence carrying the pressure. Let <d>[English] Too quiet.</d> sit low and throat-close. In the final two seconds there is no dialogue and no animal growl, only waterfall, suspended mist, and the pressure of silence around the revealed eye.\n\nnon_diegetic_music: Use no score or only an extremely faint low-frequency pressure bed beginning after the carved-stone activation. It must never overpower water, breath, leaves, or the final suspended reveal. The ending should remain tense, elegant, unresolved, and physically grounded rather than dramatic or explosive.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 9,
      "views": 543,
      "snapshotAt": "2026-09-15T00:25:56.066Z"
    },
    "approvedAt": "2026-09-15T00:25:56.066Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode also passed."
  },
  {
    "id": "x-2100383168356573667",
    "title": "手机图标让出壁纸人物",
    "titleEn": "Phone Icons Make Room for the Wallpaper Portrait",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手机桌面图标围绕壁纸人物重新排列，人物随后在屏幕内做出连续表情和手势。 原生视频规格为 15 秒 · 896×1184。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Home-screen icons rearrange around the wallpaper portrait before the woman moves and gestures inside the display. Native-video output: 15s · 896×1184. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description: [Shot 1] Live-action ultra-photorealistic close-up of a smartphone screen only, vertical 9:16 framing, approximately 15.00 seconds, real phone bezel and glass with slight environmental reflections, high-clarity display. Strictly only one person exists in the entire video: the wallpaper woman inside the phone screen. Outside the screen there must be absolutely no human, no face, no hand, no finger, no arm, no body part, and no silhouette—frame contains only the phone and its display content. The wallpaper is a half-body photo of a young beautiful Asian woman—delicate features, refined makeup, vivid expression, fashionable clothes—but the home-screen is densely covered by many app icons that block her face, eyes, and key body areas. No readable text, no captions, no subtitles, no watermarks, no logos on screen—app icons are pure graphic symbols without labels. From 00:00.000 to 00:02.000, static frame: dense icons fully covering her. From 00:02.000 to 00:12.000, the wallpaper woman suddenly comes alive inside the screen, slightly frowns with an angry-yet-coquettish look, and the lively young woman with a clear mid-high voice (S1) shouts face-on with visible lips: [Chinese] 挡住我了！ Still only inside the wallpaper, she uses only her index finger with precise targeting: the fingertip must first press directly on an icon’s exact position, then quickly swipe that icon away; icons move only when accurately pressed—random waving through empty space does nothing. She clears them one by one this way; if she moves very fast, it is still a rapid sequence of accurate press-and-swipe hits on successive icons, never aimless flailing. Some icons leave with one sharp press-swipe; a few need a second precise press-swipe. While clearing them she pouts coyly and occasionally rolls her eyes; icons gradually decrease as her face and body are revealed bit by bit. The moment the last icon leaves the screen, all index-finger press-and-swipe motions stop immediately—her on-screen hand settles and no longer keeps flicking. From 00:12.000 to 00:15.000, the screen is clean with only her photo; with the swiping hand already still, she breaks into a bright happy smile, eyes curved like crescents, makes a victory V-sign toward the viewer, and holds the freeze through 15.00 seconds. Ultra-real, smooth natural motion, true screen lighting and soft reflections. overall_soundscape: Soft phone-glass room tone, short UI tap-then-whoosh for each accurate press-and-swipe, her one shouted line and light coquettish breaths. After the last icon clears, finger-motion sounds stop. non_diegetic_music: Light playful electronic plucks at a moderate tempo, brightening when the last icons clear, then holding softly under the V-sign freeze.，3:4，15s\n\n左边海螺H3,右边seedance 2 mini",
    "promptSourceUrl": "https://x.com/PhotoX86/status/2100383168356573667",
    "sourceUrl": "https://x.com/PhotoX86/status/2100383168356573667",
    "sourceLabel": "X 原帖 · @PhotoX86",
    "author": "PhotoX86",
    "publishedAt": "2026-09-17T00:35:45.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100383168356573667.mp4",
    "posterUrl": "/posters/x/2100383168356573667.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "896×1184",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "女友相片做手机屏幕，怎么能被图标遮挡呢！这是一个现实的问题。\n\n生成视频：integrated_multimodal_description: [Shot 1] Live-action ultra-photorealistic close-up of a smartphone screen only, vertical 9:16 framing, approximately 15.00 seconds, real phone bezel and glass with slight environmental reflections, high-clarity display. Strictly only one person exists in the entire video: the wallpaper woman inside the phone screen. Outside the screen there must be absolutely no human, no face, no hand, no finger, no arm, no body part, and no silhouette—frame contains only the phone and its display content. The wallpaper is a half-body photo of a young beautiful Asian woman—delicate features, refined makeup, vivid expression, fashionable clothes—but the home-screen is densely covered by many app icons that block her face, eyes, and key body areas. No readable text, no captions, no subtitles, no watermarks, no logos on screen—app icons are pure graphic symbols without labels. From 00:00.000 to 00:02.000, static frame: dense icons fully covering her. From 00:02.000 to 00:12.000, the wallpaper woman suddenly comes alive inside the screen, slightly frowns with an angry-yet-coquettish look, and the lively young woman with a clear mid-high voice (S1) shouts face-on with visible lips: [Chinese] 挡住我了！ Still only inside the wallpaper, she uses only her index finger with precise targeting: the fingertip must first press directly on an icon’s exact position, then quickly swipe that icon away; icons move only when accurately pressed—random waving through empty space does nothing. She clears them one by one this way; if she moves very fast, it is still a rapid sequence of accurate press-and-swipe hits on successive icons, never aimless flailing. Some icons leave with one sharp press-swipe; a few need a second precise press-swipe. While clearing them she pouts coyly and occasionally rolls her eyes; icons gradually decrease as her face and body are revealed bit by bit. The moment the last icon leaves the screen, all index-finger press-and-swipe motions stop immediately—her on-screen hand settles and no longer keeps flicking. From 00:12.000 to 00:15.000, the screen is clean with only her photo; with the swiping hand already still, she breaks into a bright happy smile, eyes curved like crescents, makes a victory V-sign toward the viewer, and holds the freeze through 15.00 seconds. Ultra-real, smooth natural motion, true screen lighting and soft reflections. overall_soundscape: Soft phone-glass room tone, short UI tap-then-whoosh for each accurate press-and-swipe, her one shouted line and light coquettish breaths. After the last icon clears, finger-motion sounds stop. non_diegetic_music: Light playful electronic plucks at a moderate tempo, brightening when the last icons clear, then holding softly under the V-sign freeze.，3:4，15s\n\n左边海螺H3,右边seedance 2 mini",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 23,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100381055798939892",
    "title": "塑料椅子的高端产品广告",
    "titleEn": "A Premium Product Film for a Plastic Chair",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一把普通红色塑料椅子被灯光、材质特写和价格揭晓包装成高端产品广告。 原生视频规格为 15 秒 · 1080×1428。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Lighting, material close-ups, and a price reveal turn an ordinary red plastic chair into a premium product film. Native-video output: 15s · 1080×1428. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotoX86/status/2100381055798939892",
    "sourceLabel": "X 原帖 · @PhotoX86",
    "author": "PhotoX86",
    "publishedAt": "2026-09-17T00:27:21.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100381055798939892.mp4",
    "posterUrl": "/posters/x/2100381055798939892.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1428",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI模仿某些数码产品公司的广告，怎么把一张50元的塑料椅子卖出高价。海螺H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 5,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100379562697949685",
    "title": "战区中的第三人称突袭",
    "titleEn": "A Third-Person Assault through a War Zone",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "第三人称士兵穿过公路、山地与港口掩体，在爆炸和交火中连续推进。 原生视频规格为 15 秒 · 768×1344。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A third-person soldier advances through roads, rocky terrain, and port cover amid explosions and gunfire. Native-video output: 15s · 768×1344. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Third-person military action shooter game featuring exploration, tactical maneuvering, environmental puzzles, cinematic combat, special-ops equipment, and seamless transitions between gameplay and scripted events.\n\nEnvironment: A contemporary war-torn coastal combat zone in western Yemen overlooking the Bab-el-Mandeb Strait. The airstrike-devastated port city of Mokha stretches along the Red Sea, tank wrecks scattered across the cratered coastal highway, missile-destroyed military vehicles billowing smoke, surface-to-air missiles trailing flames as they climb from hillside positions, the silhouettes of F-15 fighter jets cutting across a sky stained orange-red by explosions, drone debris scattered across open ground between mosques and residential blocks, abandoned checkpoints and sandbag fortifications lining the roads. In the distance, black smoke columns rise from the direction of Saudi Aramco oil facilities. Gameplay begins among armored vehicle wrecks on the coastal highway, then descends into a Houthi air-defense position near the port, where a long-range missile launch vehicle is warming up.\n\nVisual style: Photorealistic Unreal Engine 5 modern military gameplay, premium AAA tactical shooter presentation, highly detailed Middle Eastern war-zone environments, realistic smoke and fire simulation, cinematic global illumination, volumetric dust and explosion particles, physically accurate metal and concrete materials, dynamic destruction systems, realistic cloth and gear physics, environmental debris and ejected shell casings, sophisticated VFX, smooth 60fps movement, restrained motion blur, and seamless cinematic transitions without losing the feeling of actual playable gameplay.\n\nPlayer character: An elite Houthi fighter wearing dust-caked dark tactical clothing, a desert-colored tactical vest, worn military boots, a dark headwrap, and a dust mask covering his face, with a captured Saudi military tactical tablet terminal strapped to his left forearm. The terminal allows the player to hack abandoned enemy communication nodes and briefly jam drone navigation signals, while an AK-pattern assault rifle remains slung across his back for close-range combat and a Russian-made grenade hangs at his waist.\n\nGameplay camera: Maintain a responsive over-the-shoulder third-person perspective during tactical maneuvering and combat, allowing the camera to dynamically widen during large battlefield reveals, move closer in narrow ruined passages, orbit slightly during firefight encounters, and transition into brief cinematic framing during major scripted moments such as missile launches or low-flying fighter jet passes before smoothly returning control to the player.\n\nGameplay UI: Maintain a minimal fixed interface with player health and ammunition reserves in the upper-left corner, a circular compass and mission objective marker in the upper-right corner, contextual interaction prompts near interactable objects (hack, pick up, cover), and special-ops equipment ability icons in the lower-right corner. UI elements remain anchored perfectly to the screen throughout camera movement.\n\nCore gameplay progression: The player traverses an airstrike-destroyed coastal highway, dodges secondary Saudi airstrikes among collapsing buildings, enters a Houthi air-defense position, uses the captured tactical terminal to hack the missile launch vehicle's control system, initiates the launch sequence and exposes their position, is forced to escape under enemy fire through ruined city blocks, advances along an underground drainage channel to the port's edge, breaks through a manhole cover to emerge in the Houthi-held Mokha port district, and transitions directly into a firefight with Saudi-backed Yemeni government ground forces.\n\nAudio design: Powerful cinematic military game score mixed with battlefield environmental audio including fighter jet shrieks, air-raid sirens, distant artillery thunder, building collapse, bullet whips, shell casings hitting the ground, radio static, drone propeller hum, missile launches, explosion shockwaves, dusty wind, footsteps, UI feedback, and environmental destruction.\n\nNegative guidance: No subtitles, no dialogue, no logos, no excessive HUD elements, no first-person perspective, no modern sci-fi weapons, no gore, no blood, no dismemberment, no broken anatomy, no inconsistent character clothing, no random camera teleportation, no excessive motion blur, no low-detail environments, no cartoon graphics, no turn-based combat.\n\n[00:00-00:04] COASTAL HIGHWAY UNDER AIRSTRIKE: Begin directly in third-person gameplay with the camera positioned behind the fighter as he low-crouch sprints along a cratered coastal highway, the skyline of Mokha port shrouded in black smoke in the distance, Saudi fighter jets circling at high altitude. An F-15 screams low overhead, its shockwave kicking up debris and dust. The player accelerates over the burning wreck of a pickup truck, rolls into cover behind the concrete of a collapsed roadside building, as the distant echo of continuous explosions rolls in and surface-to-air missiles climb from the hillside on the left trailing flames in pursuit of the jets. The camera shakes slightly on the roll landing, then stabilizes behind the player's shoulder. SFX: fighter jet shriek, distant explosions, debris scatter, rapid breathing, military score low-end pulse.\n\n[00:04-00:08] RUIN TRAVERSAL: The player bursts out from behind cover into a residential street destroyed by airstrikes, buildings on both sides reduced to skeletons, broken power lines dangling to the ground. A Houthi suicide drone skims low over the street, heading toward a distant Saudi military outpost. The player leaps and weaves between collapsed concrete blocks, vaults a half-collapsed wall, and enters a narrow alley. Beyond the alley mouth, an abandoned Saudi armored vehicle sits askew by the roadside, its hull riddled with bullet holes. The player pauses briefly, picks up an extra magazine from beside the vehicle, and continues advancing toward the air-defense position. SFX: drone propeller hum, rubble sliding, metal magazine clicking into pouch, distant machine-gun bursts.\n\n[00:08-00:11] ENTERING THE AIR-DEFENSE POSITION: The player passes through a damaged iron gate at the alley's end and enters a Houthi air-defense position concealed on the reverse slope of a hill. Several camouflage-net-covered trucks park in the position, a self-propelled anti-aircraft gun's barrel aimed skyward, several armed personnel hurriedly operating equipment. At the center of the position, a large ballistic missile launch vehicle is erecting its launch rail, its target evidently a military facility inside Saudi territory. A contextual HACK prompt appears near the control panel on the side of the launch vehicle. The player approaches the launch vehicle low, avoiding armed personnel running back and forth. SFX: hydraulic erection mechanism hum, Arabic radio chatter, generator running, wind through camouflage netting.\n\n[00:11-00:15] HACKING THE LAUNCH SYSTEM: The camera stays tight behind the player's shoulder as he crouches beside the launch vehicle and brings the tactical terminal on his left forearm close to the control panel's data port. The terminal screen lights up, a Saudi military encrypted protocol interface appearing on screen. The player operates the terminal rapidly with his right hand, bypassing firewall layers, injecting tampered launch coordinates and timing parameters. The launch vehicle's erection rail begins accelerating its rise according to the modified program, the target coordinates on the terminal screen swapped from a Saudi air base to an uninhabited desert area. An armed man walks over from the vehicle's front; the player immediately presses sideways against the hull, holding his breath. The terminal screen displays \"PARAMETER INJECTION COMPLETE.\" SFX: high-frequency electronic hum, data transmission clicks, footsteps approaching then receding, tense sustained string tone.\n\n[00:15-00:18] EXPOSURE AND ESCAPE: Warning lights beside the launch vehicle suddenly begin flashing — the system has detected anomalous parameters. Armed personnel turn their heads toward the launch vehicle one after another. The player immediately rolls out from the vehicle's side and sprints toward the stone wall at the rear of the position. An armed man raises his rifle and shouts, bullets grazing the player's helmet. The player vaults the wall, rolls into a dry drainage channel on landing, and runs at full speed toward the port. Behind him comes chaotic shouting and the sound of vehicles starting. The camera briefly pushes in as the player vaults the wall, then returns to over-the-shoulder follow. SFX: alarm buzzer, rifle fire, bullet whips, rapid footsteps, rubble sliding, chaotic shouting.\n\n[00:18-00:22] UNDERGROUND CHANNEL: The player drops into the drainage channel's opening, falling into a half-flooded underground passage, knee-deep sewage reeking. Flashlight beams sway across damp concrete walls. Footsteps and rifle-butt strikes on manhole covers come from above. The player speeds up, wading through the channel, which gradually narrows, requiring a sideways squeeze through a crack. On the other side of the crack, light filters in through the holes of a manhole cover ahead. The player reaches beneath the manhole cover, pulls a grenade from his waist, pulls the pin, waits two seconds, then shoves the cover open and tosses it up. SFX: wading footsteps, water-drop echoes, distant manhole strikes, grenade pin ping, breathing amplified in the pipe.\n\n[00:22-00:25] BREAKING THROUGH TO THE PORT SURFACE: The grenade explodes above, debris and dust pouring down through the manhole. The player immediately climbs out, rolling low into a loading area of the Mokha port district. Before him are stacks of shipping containers, abandoned port cranes, and the Red Sea in the distance. The port district is already Houthi-controlled, armed personnel patrolling among the containers, but the explosion has drawn everyone's attention. The player uses the smoke and chaos to move quickly through the loading area toward a concrete control tower at the port's edge. The camera briefly widens as the player leaps from the manhole, revealing a panoramic view of the port district and the sea stained red by the fires of war. SFX: grenade explosion, debris landing, port ambient noise, distant waves, alarms sounding from multiple directions.\n\n[00:25-00:28] FIREFIGHT: The player reaches the base of the control tower and is about to enter when a pickup truck bursts out from behind a container, a mounted machine gun on its bed raking fire toward the player. Bullets strike the concrete wall, kicking up fragments. The player immediately rolls behind a stack of pallets, raises the AK, leans out from the cover's edge, and drops the machine gunner with two short bursts to the chest. The pickup brakes hard; a second armed man jumps from the passenger seat. The player quickly adjusts his sights, shoots him in the leg to bring him down, then fires a follow-up shot. Gunfire echoes among the port's containers. The camera orbits the player slightly during the exchange, keeping the action clear. SFX: machine-gun raking fire, bullets striking concrete, rifle bursts, shell casings landing, enemy bodies falling, brief port silence.\n\n[00:28-00:30] PORT CONTROL TOWER: The player rises, quickly reloads, and runs toward the control tower's iron door. He presses sideways against the door frame, left hand on the door handle, right hand holding the rifle pointed inside. Out on the distant sea, the silhouette of a Saudi warship appears beyond the smoke, the flash of its naval gun closing in. The player takes a deep breath, yanks the door open, and charges low into the control tower's interior. The gameplay camera stabilizes behind the character, multiple combat ability icons illuminating in the lower-right corner — tactical terminal hack, jamming pulse, cover marker. The frame ends on the player's dynamic movement entering the tower rather than a cinematic freeze, making the sequence feel like genuine gameplay continuing beyond the clip. SFX: magazine reload click, iron door creak, distant naval gun thunder, equipment prompt tones, military score surging to climax.\n\nThe entire video must feel like real playable AAA third-person military gameplay rather than a movie trailer, with the player character continuously responding to environmental obstacles, airstrike threats, contextual prompts, tactical maneuvering, special-ops equipment, mission objectives, and enemy attacks while the camera behaves like a sophisticated dynamic gameplay camera.\n\nMaintain a clear gameplay progression from high-speed traversal along the coastal highway into air-defense position infiltration, system hacking, exposure and escape, underground channel advancement, port breakthrough, and the beginning of a firefight with Yemeni government ground forces, allowing multiple game mechanics to be demonstrated naturally within the same 30-second sequence.\n\nKeep transitions completely seamless between traversal, climbing, hacking interaction, escape, and combat without fades, disconnected montage shots, or unexplained location changes. The geography should remain understandable, with the coastal highway visibly connected to the air-defense position, the underground channel leading directly to the Mokha port district, and the control tower positioned at the port's edge.\n\nMaintain the same fighter, equipment, tactical terminal, AK rifle, architecture, visual language, UI design, lighting logic, and contemporary Middle Eastern war-zone aesthetic throughout the entire sequence. The UI must remain fixed to the screen and should respond logically to gameplay events without obstructing important action.\n\nNo gore, no subtitles, no dialogue, no first-person perspective, no cinematic black bars, and no transition that removes the feeling that the viewer is watching someone actively playing the game.\n\n本回答由 AI 生成，内容仅供参考，请仔细甄别",
    "promptSourceUrl": "https://x.com/sada_ai/status/2100379562697949685",
    "sourceUrl": "https://x.com/sada_ai/status/2100379562697949685",
    "sourceLabel": "X 原帖 · @sada_ai",
    "author": "sada",
    "publishedAt": "2026-09-17T00:21:25.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100379562697949685.mp4",
    "posterUrl": "/posters/x/2100379562697949685.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "ugc"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created with MiniMax H3,Prompt is here:\nThird-person military action shooter game featuring exploration, tactical maneuvering, environmental puzzles, cinematic combat, special-ops equipment, and seamless transitions between gameplay and scripted events.\n\nEnvironment: A contemporary war-torn coastal combat zone in western Yemen overlooking the Bab-el-Mandeb Strait. The airstrike-devastated port city of Mokha stretches along the Red Sea, tank wrecks scattered across the cratered coastal highway, missile-destroyed military vehicles billowing smoke, surface-to-air missiles trailing flames as they climb from hillside positions, the silhouettes of F-15 fighter jets cutting across a sky stained orange-red by explosions, drone debris scattered across open ground between mosques and residential blocks, abandoned checkpoints and sandbag fortifications lining the roads. In the distance, black smoke columns rise from the direction of Saudi Aramco oil facilities. Gameplay begins among armored vehicle wrecks on the coastal highway, then descends into a Houthi air-defense position near the port, where a long-range missile launch vehicle is warming up.\n\nVisual style: Photorealistic Unreal Engine 5 modern military gameplay, premium AAA tactical shooter presentation, highly detailed Middle Eastern war-zone environments, realistic smoke and fire simulation, cinematic global illumination, volumetric dust and explosion particles, physically accurate metal and concrete materials, dynamic destruction systems, realistic cloth and gear physics, environmental debris and ejected shell casings, sophisticated VFX, smooth 60fps movement, restrained motion blur, and seamless cinematic transitions without losing the feeling of actual playable gameplay.\n\nPlayer character: An elite Houthi fighter wearing dust-caked dark tactical clothing, a desert-colored tactical vest, worn military boots, a dark headwrap, and a dust mask covering his face, with a captured Saudi military tactical tablet terminal strapped to his left forearm. The terminal allows the player to hack abandoned enemy communication nodes and briefly jam drone navigation signals, while an AK-pattern assault rifle remains slung across his back for close-range combat and a Russian-made grenade hangs at his waist.\n\nGameplay camera: Maintain a responsive over-the-shoulder third-person perspective during tactical maneuvering and combat, allowing the camera to dynamically widen during large battlefield reveals, move closer in narrow ruined passages, orbit slightly during firefight encounters, and transition into brief cinematic framing during major scripted moments such as missile launches or low-flying fighter jet passes before smoothly returning control to the player.\n\nGameplay UI: Maintain a minimal fixed interface with player health and ammunition reserves in the upper-left corner, a circular compass and mission objective marker in the upper-right corner, contextual interaction prompts near interactable objects (hack, pick up, cover), and special-ops equipment ability icons in the lower-right corner. UI elements remain anchored perfectly to the screen throughout camera movement.\n\nCore gameplay progression: The player traverses an airstrike-destroyed coastal highway, dodges secondary Saudi airstrikes among collapsing buildings, enters a Houthi air-defense position, uses the captured tactical terminal to hack the missile launch vehicle's control system, initiates the launch sequence and exposes their position, is forced to escape under enemy fire through ruined city blocks, advances along an underground drainage channel to the port's edge, breaks through a manhole cover to emerge in the Houthi-held Mokha port district, and transitions directly into a firefight with Saudi-backed Yemeni government ground forces.\n\nAudio design: Powerful cinematic military game score mixed with battlefield environmental audio including fighter jet shrieks, air-raid sirens, distant artillery thunder, building collapse, bullet whips, shell casings hitting the ground, radio static, drone propeller hum, missile launches, explosion shockwaves, dusty wind, footsteps, UI feedback, and environmental destruction.\n\nNegative guidance: No subtitles, no dialogue, no logos, no excessive HUD elements, no first-person perspective, no modern sci-fi weapons, no gore, no blood, no dismemberment, no broken anatomy, no inconsistent character clothing, no random camera teleportation, no excessive motion blur, no low-detail environments, no cartoon graphics, no turn-based combat.\n\n[00:00-00:04] COASTAL HIGHWAY UNDER AIRSTRIKE: Begin directly in third-person gameplay with the camera positioned behind the fighter as he low-crouch sprints along a cratered coastal highway, the skyline of Mokha port shrouded in black smoke in the distance, Saudi fighter jets circling at high altitude. An F-15 screams low overhead, its shockwave kicking up debris and dust. The player accelerates over the burning wreck of a pickup truck, rolls into cover behind the concrete of a collapsed roadside building, as the distant echo of continuous explosions rolls in and surface-to-air missiles climb from the hillside on the left trailing flames in pursuit of the jets. The camera shakes slightly on the roll landing, then stabilizes behind the player's shoulder. SFX: fighter jet shriek, distant explosions, debris scatter, rapid breathing, military score low-end pulse.\n\n[00:04-00:08] RUIN TRAVERSAL: The player bursts out from behind cover into a residential street destroyed by airstrikes, buildings on both sides reduced to skeletons, broken power lines dangling to the ground. A Houthi suicide drone skims low over the street, heading toward a distant Saudi military outpost. The player leaps and weaves between collapsed concrete blocks, vaults a half-collapsed wall, and enters a narrow alley. Beyond the alley mouth, an abandoned Saudi armored vehicle sits askew by the roadside, its hull riddled with bullet holes. The player pauses briefly, picks up an extra magazine from beside the vehicle, and continues advancing toward the air-defense position. SFX: drone propeller hum, rubble sliding, metal magazine clicking into pouch, distant machine-gun bursts.\n\n[00:08-00:11] ENTERING THE AIR-DEFENSE POSITION: The player passes through a damaged iron gate at the alley's end and enters a Houthi air-defense position concealed on the reverse slope of a hill. Several camouflage-net-covered trucks park in the position, a self-propelled anti-aircraft gun's barrel aimed skyward, several armed personnel hurriedly operating equipment. At the center of the position, a large ballistic missile launch vehicle is erecting its launch rail, its target evidently a military facility inside Saudi territory. A contextual HACK prompt appears near the control panel on the side of the launch vehicle. The player approaches the launch vehicle low, avoiding armed personnel running back and forth. SFX: hydraulic erection mechanism hum, Arabic radio chatter, generator running, wind through camouflage netting.\n\n[00:11-00:15] HACKING THE LAUNCH SYSTEM: The camera stays tight behind the player's shoulder as he crouches beside the launch vehicle and brings the tactical terminal on his left forearm close to the control panel's data port. The terminal screen lights up, a Saudi military encrypted protocol interface appearing on screen. The player operates the terminal rapidly with his right hand, bypassing firewall layers, injecting tampered launch coordinates and timing parameters. The launch vehicle's erection rail begins accelerating its rise according to the modified program, the target coordinates on the terminal screen swapped from a Saudi air base to an uninhabited desert area. An armed man walks over from the vehicle's front; the player immediately presses sideways against the hull, holding his breath. The terminal screen displays \"PARAMETER INJECTION COMPLETE.\" SFX: high-frequency electronic hum, data transmission clicks, footsteps approaching then receding, tense sustained string tone.\n\n[00:15-00:18] EXPOSURE AND ESCAPE: Warning lights beside the launch vehicle suddenly begin flashing — the system has detected anomalous parameters. Armed personnel turn their heads toward the launch vehicle one after another. The player immediately rolls out from the vehicle's side and sprints toward the stone wall at the rear of the position. An armed man raises his rifle and shouts, bullets grazing the player's helmet. The player vaults the wall, rolls into a dry drainage channel on landing, and runs at full speed toward the port. Behind him comes chaotic shouting and the sound of vehicles starting. The camera briefly pushes in as the player vaults the wall, then returns to over-the-shoulder follow. SFX: alarm buzzer, rifle fire, bullet whips, rapid footsteps, rubble sliding, chaotic shouting.\n\n[00:18-00:22] UNDERGROUND CHANNEL: The player drops into the drainage channel's opening, falling into a half-flooded underground passage, knee-deep sewage reeking. Flashlight beams sway across damp concrete walls. Footsteps and rifle-butt strikes on manhole covers come from above. The player speeds up, wading through the channel, which gradually narrows, requiring a sideways squeeze through a crack. On the other side of the crack, light filters in through the holes of a manhole cover ahead. The player reaches beneath the manhole cover, pulls a grenade from his waist, pulls the pin, waits two seconds, then shoves the cover open and tosses it up. SFX: wading footsteps, water-drop echoes, distant manhole strikes, grenade pin ping, breathing amplified in the pipe.\n\n[00:22-00:25] BREAKING THROUGH TO THE PORT SURFACE: The grenade explodes above, debris and dust pouring down through the manhole. The player immediately climbs out, rolling low into a loading area of the Mokha port district. Before him are stacks of shipping containers, abandoned port cranes, and the Red Sea in the distance. The port district is already Houthi-controlled, armed personnel patrolling among the containers, but the explosion has drawn everyone's attention. The player uses the smoke and chaos to move quickly through the loading area toward a concrete control tower at the port's edge. The camera briefly widens as the player leaps from the manhole, revealing a panoramic view of the port district and the sea stained red by the fires of war. SFX: grenade explosion, debris landing, port ambient noise, distant waves, alarms sounding from multiple directions.\n\n[00:25-00:28] FIREFIGHT: The player reaches the base of the control tower and is about to enter when a pickup truck bursts out from behind a container, a mounted machine gun on its bed raking fire toward the player. Bullets strike the concrete wall, kicking up fragments. The player immediately rolls behind a stack of pallets, raises the AK, leans out from the cover's edge, and drops the machine gunner with two short bursts to the chest. The pickup brakes hard; a second armed man jumps from the passenger seat. The player quickly adjusts his sights, shoots him in the leg to bring him down, then fires a follow-up shot. Gunfire echoes among the port's containers. The camera orbits the player slightly during the exchange, keeping the action clear. SFX: machine-gun raking fire, bullets striking concrete, rifle bursts, shell casings landing, enemy bodies falling, brief port silence.\n\n[00:28-00:30] PORT CONTROL TOWER: The player rises, quickly reloads, and runs toward the control tower's iron door. He presses sideways against the door frame, left hand on the door handle, right hand holding the rifle pointed inside. Out on the distant sea, the silhouette of a Saudi warship appears beyond the smoke, the flash of its naval gun closing in. The player takes a deep breath, yanks the door open, and charges low into the control tower's interior. The gameplay camera stabilizes behind the character, multiple combat ability icons illuminating in the lower-right corner — tactical terminal hack, jamming pulse, cover marker. The frame ends on the player's dynamic movement entering the tower rather than a cinematic freeze, making the sequence feel like genuine gameplay continuing beyond the clip. SFX: magazine reload click, iron door creak, distant naval gun thunder, equipment prompt tones, military score surging to climax.\n\nThe entire video must feel like real playable AAA third-person military gameplay rather than a movie trailer, with the player character continuously responding to environmental obstacles, airstrike threats, contextual prompts, tactical maneuvering, special-ops equipment, mission objectives, and enemy attacks while the camera behaves like a sophisticated dynamic gameplay camera.\n\nMaintain a clear gameplay progression from high-speed traversal along the coastal highway into air-defense position infiltration, system hacking, exposure and escape, underground channel advancement, port breakthrough, and the beginning of a firefight with Yemeni government ground forces, allowing multiple game mechanics to be demonstrated naturally within the same 30-second sequence.\n\nKeep transitions completely seamless between traversal, climbing, hacking interaction, escape, and combat without fades, disconnected montage shots, or unexplained location changes. The geography should remain understandable, with the coastal highway visibly connected to the air-defense position, the underground channel leading directly to the Mokha port district, and the control tower positioned at the port's edge.\n\nMaintain the same fighter, equipment, tactical terminal, AK rifle, architecture, visual language, UI design, lighting logic, and contemporary Middle Eastern war-zone aesthetic throughout the entire sequence. The UI must remain fixed to the screen and should respond logically to gameplay events without obstructing important action.\n\nNo gore, no subtitles, no dialogue, no first-person perspective, no cinematic black bars, and no transition that removes the feeling that the viewer is watching someone actively playing the game.\n\n本回答由 AI 生成，内容仅供参考，请仔细甄别",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 60,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100331985772359786",
    "title": "雪原农场上的孤独老人",
    "titleEn": "A Solitary Farmer on a Snowy Plain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手绘质感的长片段跟随老人驾驶农机、回到木屋并在冬日农场中生活。 原生视频规格为 83 秒 · 1024×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A painterly long-form sequence follows an elderly farmer driving machinery and returning to his cabin on a winter farm. Native-video output: 83s · 1024×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kaigani/status/2100331985772359786",
    "sourceLabel": "X 原帖 · @kaigani",
    "author": "𝚔𝚊𝚒 🤖+🧠",
    "publishedAt": "2026-09-16T21:12:22.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100331985772359786.mp4",
    "posterUrl": "/posters/x/2100331985772359786.jpg",
    "duration": 83,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Story concept & art direction I've been playing around with. Video and music using local AI models MiniMax H3 and YuE2.\nTrying to get away from the typical anime-derived look of AI animation.",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 6,
      "views": 226,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100328200114950440",
    "title": "八拍结构扩展成一分钟音乐视觉片",
    "titleEn": "An Eight-Beat Structure Expanded into a One-Minute Music Visual",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "同一动画角色在室内、街道与雨夜舞台间切换，以八段结构承载完整音乐段落。 原生视频规格为 60 秒 · 1080×602。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same animated character moves across rooms, streets, and a rainy stage in an eight-beat structure built around a full track. Native-video output: 60s · 1080×602. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GoshniiAi/status/2100328200114950440",
    "sourceLabel": "X 原帖 · @GoshniiAi",
    "author": "goshniiAI",
    "publishedAt": "2026-09-16T20:57:19.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100328200114950440.mp4",
    "posterUrl": "/posters/x/2100328200114950440.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1080×602",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "music-performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Same 8-beat structure, stretched from 20 seconds to 60 — built to carry a track instead of dialogue. MiniMax H3 Ref2Video.\n\n#ComfyUI #MiniMaxH3 #ReferenceToVideo #MusicVisualizer #AIFilmmaking #GenerativeAI #AIWorkflow #minimax",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 61,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100319404806303788",
    "title": "麦浪只向一个方向穿过田野",
    "titleEn": "A Single Wave Crosses the Wheat Field",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "固定机位记录一条狭窄麦浪从左向右通过，麦秆在其后重新站起。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A locked camera captures one narrow wave moving left to right while the wheat rises again behind it. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "a clear moving band of flattened wheat about ten stalks wide, the stalks behind it standing back up as it passes",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2100319404806303788",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2100319404806303788",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-16T20:22:22.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100319404806303788.mp4",
    "posterUrl": "/posters/x/2100319404806303788.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 local. One wave. Left to right. Then nothing.\n\nPrompt: \"a clear moving band of flattened wheat about ten stalks wide, the stalks behind it standing back up as it passes\"\n\nDuration: 15s. Camera: locked, just above the wheat.\n\nA field is ten thousand stalks. A band ten stalks wide moving at walking pace is one object. Turn the crowd into one thing.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 10,
      "views": 231,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100304685999513920",
    "title": "夜跑动作的性能分析实验",
    "titleEn": "A Performance-Analysis Experiment with a Night Runner",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "夜间街道中的跑步人物被检测框和性能标注覆盖，呈现动作追踪式测试。 原生视频规格为 35 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A runner on a wet night street is overlaid with tracking boxes and performance annotations in a motion-analysis test. Native-video output: 35s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ysf__ai/status/2100304685999513920",
    "sourceLabel": "X 原帖 · @ysf__ai",
    "author": "ysf.ai | Synth Engine Studios",
    "publishedAt": "2026-09-16T19:23:53.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100304685999513920.mp4",
    "posterUrl": "/posters/x/2100304685999513920.jpg",
    "duration": 35,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic",
      "motion-graphics"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Performance training \n\nMade with MiniMax H3 on @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 84,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100299004689215532",
    "title": "城市漫游中的手机广告短片",
    "titleEn": "A Smartphone Film through the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女主在城市街区中使用手机，近景产品镜头与人物叙事组成完整广告短片。 原生视频规格为 117 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman uses a smartphone across city locations, combining product close-ups with a character-led commercial narrative. Native-video output: 117s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/izesaki/status/2100299004689215532",
    "sourceLabel": "X 原帖 · @izesaki",
    "author": "! Ze",
    "publishedAt": "2026-09-16T19:01:19.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100299004689215532.mp4",
    "posterUrl": "/posters/x/2100299004689215532.jpg",
    "duration": 117,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "この一瞬を、あなたのものに。\n\nmade with MiniMax H3 on @openart_ai #OpenArtAdAwards @MiniMax_AI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 86,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100296671737638929",
    "title": "原子时代的复古未来日志",
    "titleEn": "An Atomic-Age Retro-Futurist Vlog",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "复古主持人从商店、美妆镜到火箭发射场，串联成一段原子时代风格的生活日志。 原生视频规格为 104 秒 · 1454×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A retro host moves from a shop and vanity mirror to a rocket launch in an atomic-age lifestyle vlog. Native-video output: 104s · 1454×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Devozikjr/status/2100296671737638929",
    "sourceLabel": "X 原帖 · @Devozikjr",
    "author": "Devoz | Cinematic AI Lab",
    "publishedAt": "2026-09-16T18:52:02.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100296671737638929.mp4",
    "posterUrl": "/posters/x/2100296671737638929.jpg",
    "duration": 104,
    "aspectRatio": "landscape",
    "resolution": "1454×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "commercial"
    ],
    "scenes": [
      "daily-life",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 atomic age vlog",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 33,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100281989349711989",
    "title": "静态人像经过放大与动态精修",
    "titleEn": "A Static Portrait Gains Motion and Upscaled Detail",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "静态人物图被转成自然说话与眨眼的视频，并经过放大和独立细化流程处理。 原生视频规格为 10 秒 · 2560×1472。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A still portrait becomes a naturally blinking and speaking clip before upscaling and a separate refinement pass. Native-video output: 10s · 2560×1472. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/godblesstoto/status/2100281989349711989",
    "sourceLabel": "X 原帖 · @godblesstoto",
    "author": "🙏🌧🌍",
    "publishedAt": "2026-09-16T17:53:42.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100281989349711989.mp4",
    "posterUrl": "/posters/x/2100281989349711989.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1472",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MJ static image --> Minimax H3 image to video with RTX Upscaler workflow --> Standalone LTX Refiner workflow.\n\n7 minutes in total.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 1884,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100279916969508955",
    "title": "RTX 3090 保持人像长帧稳定",
    "titleEn": "An RTX 3090 Keeps a Long Portrait Sequence Stable",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "近景女性人像在轻微表情变化中保持面部细节，用于观察本地长帧生成的显存稳定性。 原生视频规格为 10 秒 · 2688×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A close-up portrait preserves facial detail through subtle expressions in a local long-frame memory-stability test. Native-video output: 10s · 2688×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KrakenUnbound/status/2100279916969508955",
    "sourceLabel": "X 原帖 · @KrakenUnbound",
    "author": "The Kraken",
    "publishedAt": "2026-09-16T17:45:28.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100279916969508955.mp4",
    "posterUrl": "/posters/x/2100279916969508955.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2688×1536",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Generated locally with my 3090 and then upscaled 2x with my upscaler using RTX VSR.\n\nMiniMax H3 — Quality Base memory-stability test\n\nGPU: NVIDIA RTX 3090, 24 GB VRAM\nMode: Image-to-video\nResolution: 1344 × 768\nOutput: 10 seconds, 24 fps, 240 frames\nGeneration frame count: 243 before output trimming\nRender time: 45 minutes 11 seconds\nSteps: 25\nSampler: Euler\nScheduler: Simple\nCFG: 1.0\nSeed: 0\nVideo shift: 12.0\nAudio shift: 3.0\nAttention: SageAttention\nLoRAs: None — no 4-, 6-, or 8-step Turbo adapter\nMotion continuation: Disabled\nAudio: Generated shoreline ambience; no imported audio\nDiffusion model: minimax_h3_fl2va_pruned_int8_convrot.safetensors\nText encoder: qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors\nVideo VAE: minimax_h3_video_vae_fp16.safetensors\nAudio VAE: minimax_h3_audio_vae_fp32.safetensors",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 147,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100274933788291363",
    "title": "零食窃贼遇上意外监督员",
    "titleEn": "A Snack Thief Meets an Unexpected Supervisor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蒙面窃贼潜入厨房偷零食，却被金毛犬发现并带走赃物，形成轻喜剧反转。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A masked thief sneaks into a kitchen for snacks, only for a golden retriever to discover him and carry away the loot. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2100274933788291363",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-09-16T17:25:40.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100274933788291363.mp4",
    "posterUrl": "/posters/x/2100274933788291363.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "daily-life",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Snack Heist 🍟\n\nCareful planning. Perfect execution. One very unexpected supervisor. 😂\n\nStory suggested, prompt written and video created by  MiniMax H3 Design. So sweet and easy 😊",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 8,
      "views": 147,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100254888286703729",
    "title": "四人偶像组合的同步舞台",
    "titleEn": "A Four-Member Idol Group Performs in Sync",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "四位虚拟偶像在舞台上保持独立动作与镜头调度，完成一段紧凑的团体表演。 原生视频规格为 10 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Four virtual idols retain independent motion and coordinated camera work through a compact group performance. Native-video output: 10s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LiluluLoserLab/status/2100254888286703729",
    "sourceLabel": "X 原帖 · @LiluluLoserLab",
    "author": "リルル@Loser Lab",
    "publishedAt": "2026-09-16T16:06:00.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100254888286703729.mp4",
    "posterUrl": "/posters/x/2100254888286703729.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ここに、私たちの最高の１０秒を― MiniMax H3\nチャッピー＆GPT Image\n\n今日の実験の成果　４人独立制御＆カメラワーク\nアイドルユニットPVテスト\n\nやれば何とかなるもんだ。。すごい。(*'ω'*)\n\n#AIart",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 0,
      "views": 92,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100254384407945670",
    "title": "彩色节奏中的动画歌手",
    "titleEn": "An Animated Singer in a Colorful Rhythm Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "动画歌手在旋涡图形、近景表演和琴键舞台之间快速切换。 原生视频规格为 15 秒 · 1906×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated singer moves rapidly between spiral graphics, close-up performance, and a keyboard-like stage. Native-video output: 15s · 1906×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LyovaKnyazyan/status/2100254384407945670",
    "sourceLabel": "X 原帖 · @LyovaKnyazyan",
    "author": "Lyova Knyazyan",
    "publishedAt": "2026-09-16T16:04:00.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100254384407945670.mp4",
    "posterUrl": "/posters/x/2100254384407945670.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1906×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cretaed with Minimax H3 Turbo by @FloAi_Official \n@FloAi_Official\n#FloAi #minimax #h3 #ai #video #generation #image #to",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 53,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100247955362206010",
    "title": "遗迹中的诱饵机关",
    "titleEn": "A Lure Mechanism inside the Ruins",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "女探险者沿金色痕迹深入神殿，踩中机关后意识到线索正把她引向黑暗中的危险。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An explorer follows golden traces into a temple, triggers a mechanism, and realizes the trail is luring her toward danger. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, high-stability cinematic adventure-thriller image-to-video sequence extending only inside the same ancient temple corridor shown in <Picture 1>. Preserve the exact same young female explorer throughout: same face structure, age impression, wet skin texture, realistic hair strands, clothing, and body proportions. Keep only this corridor system: high arched stone ceiling, broken dome skylight, tall statues on both sides with circular waist medallions, damp stone floor, moss, vines, dust, low braziers, and old gears as static set dressing only. No exterior, no new halls, no extra people, no subtitles, no watermark, no face drift, no body deformation, no finger errors, no flicker, and no cut to black. Story logic: she discovers what looks like a treasure marker, follows it deeper, steps onto a selection mechanism, then realizes the trail was not leading to treasure but drawing her toward something waiting ahead. She remains in continuous motion; all emotion is attached to sliding, stepping, leaning, crouching, touching stone, braking, turning, and attempted retreat. Camera language stays intimate and grounded: tight side-follow, slight push-in, over-shoulder follow, short lateral move, and a final suspended lock. [Shot 2] At 00:00-00:02, a low-angle tight near shot from her left-front waist height moves in close side-follow as she slides out from behind a stone pillar. The first frame already contains her. She steps down from a shallow stone rise with her right leg while her left palm glides along a damp carved relief for support. Her body stays low, breathing intentionally shallow, and her gaze drops toward the floor ahead. Her first emotional layer is highly controlled vigilance: brow peaks press inward only slightly, eyes scan the dark ahead then cut quickly down, lids narrow, pupils focus sharply, nostrils move with restrained breathing, lips part by a thin line without sound, jaw stays stable, and neck muscles show mild tension. Backlight from an arch behind outlines wet hair edges, collarbone, shoulder strap, and the side of her waist, while most of the face remains in partial shadow. Dust disturbed by her slide-step drifts through the beam with directional inertia. Hair trails for a beat and settles. [Shot 3] At 00:02-00:04.5, keep the same axis and raise the camera to chest height with a gentle push into a medium close shot. She leans out farther from the pillar and locks onto a fresh golden powder trace lodged in the floor-tile seams. Her feet stop, the body tilts forward half an inch, the left hand remains on the pillar, and the right hand starts to lower toward the ground. Her second emotional layer is a brief flare of analytical excitement: vigilance remains in the eyes, but focus sharpens at the sight of the trail, the brow center loosens for half a beat and closes again, and one mouth corner lifts by the smallest amount before restraint returns. She takes one subtle inhale through the nose and the throat moves as she suppresses the excitement. She murmurs in a low breath-close voice, <d>[English] Someone got here first.</d> The golden powder is not magical glow but fine mineral dust embedded in the seam, flashing softly under grazing light. [Shot 4] At 00:04.5-00:07.5, in a close shot, begin on her fingers touching the golden powder, then smoothly lift with her hand to her side face and end in an over-shoulder composition that includes deeper corridor space. She lowers into a half-crouch, brushes the powder with two fingers, and lifts them to inspect the residue while still shifting half a step forward. Her gaze passes her fingertips and moves deeper under the far arch. Her third emotional layer is rational confirmation: eyes narrow slightly, inner corners tighten, the brow draws together with more concentration but no heaviness, lips close then loosen again, and the nose tip and nostrils show delicate tension from close inspection. When she raises her eyes deeper into the corridor, a dangerous curiosity appears for the first time. Her face falls deeper into shadow during the crouch, while the gold dust on her fingertips flashes a small warm reflection in backlight. Near the deeper arch, a second and third trace of golden dust becomes faintly visible in sequence as if her movement disturbed them. [Shot 5] At 00:07.5-00:10.5, over-shoulder medium shot with natural perspective as the camera follows her forward two steps, then stops abruptly at the key moment. She follows the golden traces and places full weight onto a central stone tile. The instant the weight settles, a very light mechanical bite clicks beneath her. Her whole body locks in place. The right shoulder stiffens, the ankle and calf freeze, the left hand reflexively starts to rise toward the nearest pillar, and her breath catches high in the chest. Her fourth emotional layer is the body recognizing danger before the mind completes the thought: eyes cut sharply downward, lids tighten, pupils contract, the brow center locks, nostrils open a fraction and then stop, all softness leaves the mouth, the lip line narrows, and the jawline hardens. It reads as a trained freeze, not exaggerated fright. As her weight compresses the tile, the boot sole drags microscopically across stone, fine dust squeezes from the seam, and tiny grains roll from the edge. Her sudden stop creates small delayed inertia in the hair, straps, fabric, and upper torso before they settle. [Shot 6] At 00:10.5-00:12.8, cut to a side half-body close shot with a short controlled lateral move that follows her turn and attempted retreat. She snaps her head back to retreat, but after only half a step she sees that two floor tiles behind her have silently risen half an inch, creating a subtle break line in the escape path. She instantly plants her left palm against the pillar edge, tightens the core to stop her body, and lifts the right arm halfway as if preparing to pivot around the pillar and escape. Her fifth emotional layer is controlled fear merged with rapid calculation: the brow presses lower, the eyes scan the retreat path and then cut back toward the dark opening ahead with precise motion, nostrils widen more visibly with faster breathing, the lips part for an instant and clamp shut again as she swallows the urge to gasp, and the cheek tension and jaw remain locked. At the same moment, both low candles shiver once in sync. Their flames dip under a brief air-pressure pulse and rebound. Fine golden dust begins to reverse-flow through the pillar seams and floor carvings toward the deeper arch, not as a violent magical blast but as a slow unseen draw. The lifted floor tiles shed dust, sand, and rubbing powder. [Shot 7] At 00:12.8-00:15, a medium close shot with a slight low angle pushes in half a step and then suspends without cutting to black. She is half-turned, right arm raised and ready to use the pillar for a vaulting escape, but the motion stops before release. In the deepest darkness beneath the arch ahead, a single tiny but razor-sharp golden reflection appears, like a beast eye, fixed directly on her. She does not recoil; she is forced into stillness by the correction of understanding. Her sixth emotional layer is cold recognition after a shattered assumption: the gaze locks fully for the first time, pupils fixed on that point of gold, upper lids slightly raised, lower lids tightened, brows lifting only by a thread as her previous judgment is overturned, a short controlled exhale slips through the nose, lips part slightly, mouth corners press down, and the jaw remains tense. She says in a quieter, colder throat-close voice, <d>[English] That’s not a trail.</d> with “not” the lightest but hardest word and “trail” cut very short. All the golden dust and fine particles continue to stream toward that single reflective point. Her face remains mostly in shadow, with only the eye corner, nose bridge edge, wet lip line, and collarbone rim caught by mixed cold reflection from the arch and warm low candle bounce. The golden point stays tiny, sharp, and singular, never expanding into a full creature or large supernatural display. Hold on this suspended threat state, with particles still moving and the corridor still alive with pressure.\n\noverall_soundscape: Maintain a realistic ancient temple acoustic bed throughout: low hollow room resonance, distant dripping water, faint insects, candle burn, boot friction on damp stone, palm sliding across humid carved relief, fine grain contact as fingers skim the golden powder, and subtle fabric and gear movement. Emphasize shallow controlled breathing, the side-step from behind the pillar, dust disturbed into the light beam, the quiet stop as she notices the medallion, fingertip contact with the gold dust, tiny grains lifting and falling, the crisp mechanism bite click under her foot, a brief oppressive reduction in ambience under her frozen breath, the silent rise of the tiles, sand sliding from edges, her palm planting against the pillar, the synchronized candle flutter, and the faint inward draw of air carrying dust toward the arch. End with corridor resonance, low candle crackle, drifting particulate movement, restrained breathing, a tiny pressure in the air around the golden point, and her cold close line <d>[English] That’s not a trail.</d>\n\nnon_diegetic_music: Keep music minimal and subordinate to the physical environment. Use only a very thin tension bed in the first half, almost below perception, then tighten the low-frequency pressure slightly after the mechanism click. Do not swell into action music or melodrama. End suspended on unresolved tension.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2100247955362206010",
    "sourceUrl": "https://x.com/ou_zhen599/status/2100247955362206010",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-16T15:38:27.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100247955362206010.mp4",
    "posterUrl": "/posters/x/2100247955362206010.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                              \n\n🎬Image-to-Video              \n📷“Not a Signpost, but a Trap”\nShe thought she had finally found the entrance; but she soon realized that it wasn’t a clue left by those who came before her, but rather some kind of living mechanism inside the temple that was “luring her closer.”\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9, high-stability cinematic adventure-thriller image-to-video sequence extending only inside the same ancient temple corridor shown in <Picture 1>. Preserve the exact same young female explorer throughout: same face structure, age impression, wet skin texture, realistic hair strands, clothing, and body proportions. Keep only this corridor system: high arched stone ceiling, broken dome skylight, tall statues on both sides with circular waist medallions, damp stone floor, moss, vines, dust, low braziers, and old gears as static set dressing only. No exterior, no new halls, no extra people, no subtitles, no watermark, no face drift, no body deformation, no finger errors, no flicker, and no cut to black. Story logic: she discovers what looks like a treasure marker, follows it deeper, steps onto a selection mechanism, then realizes the trail was not leading to treasure but drawing her toward something waiting ahead. She remains in continuous motion; all emotion is attached to sliding, stepping, leaning, crouching, touching stone, braking, turning, and attempted retreat. Camera language stays intimate and grounded: tight side-follow, slight push-in, over-shoulder follow, short lateral move, and a final suspended lock. [Shot 2] At 00:00-00:02, a low-angle tight near shot from her left-front waist height moves in close side-follow as she slides out from behind a stone pillar. The first frame already contains her. She steps down from a shallow stone rise with her right leg while her left palm glides along a damp carved relief for support. Her body stays low, breathing intentionally shallow, and her gaze drops toward the floor ahead. Her first emotional layer is highly controlled vigilance: brow peaks press inward only slightly, eyes scan the dark ahead then cut quickly down, lids narrow, pupils focus sharply, nostrils move with restrained breathing, lips part by a thin line without sound, jaw stays stable, and neck muscles show mild tension. Backlight from an arch behind outlines wet hair edges, collarbone, shoulder strap, and the side of her waist, while most of the face remains in partial shadow. Dust disturbed by her slide-step drifts through the beam with directional inertia. Hair trails for a beat and settles. [Shot 3] At 00:02-00:04.5, keep the same axis and raise the camera to chest height with a gentle push into a medium close shot. She leans out farther from the pillar and locks onto a fresh golden powder trace lodged in the floor-tile seams. Her feet stop, the body tilts forward half an inch, the left hand remains on the pillar, and the right hand starts to lower toward the ground. Her second emotional layer is a brief flare of analytical excitement: vigilance remains in the eyes, but focus sharpens at the sight of the trail, the brow center loosens for half a beat and closes again, and one mouth corner lifts by the smallest amount before restraint returns. She takes one subtle inhale through the nose and the throat moves as she suppresses the excitement. She murmurs in a low breath-close voice, <d>[English] Someone got here first.</d> The golden powder is not magical glow but fine mineral dust embedded in the seam, flashing softly under grazing light. [Shot 4] At 00:04.5-00:07.5, in a close shot, begin on her fingers touching the golden powder, then smoothly lift with her hand to her side face and end in an over-shoulder composition that includes deeper corridor space. She lowers into a half-crouch, brushes the powder with two fingers, and lifts them to inspect the residue while still shifting half a step forward. Her gaze passes her fingertips and moves deeper under the far arch. Her third emotional layer is rational confirmation: eyes narrow slightly, inner corners tighten, the brow draws together with more concentration but no heaviness, lips close then loosen again, and the nose tip and nostrils show delicate tension from close inspection. When she raises her eyes deeper into the corridor, a dangerous curiosity appears for the first time. Her face falls deeper into shadow during the crouch, while the gold dust on her fingertips flashes a small warm reflection in backlight. Near the deeper arch, a second and third trace of golden dust becomes faintly visible in sequence as if her movement disturbed them. [Shot 5] At 00:07.5-00:10.5, over-shoulder medium shot with natural perspective as the camera follows her forward two steps, then stops abruptly at the key moment. She follows the golden traces and places full weight onto a central stone tile. The instant the weight settles, a very light mechanical bite clicks beneath her. Her whole body locks in place. The right shoulder stiffens, the ankle and calf freeze, the left hand reflexively starts to rise toward the nearest pillar, and her breath catches high in the chest. Her fourth emotional layer is the body recognizing danger before the mind completes the thought: eyes cut sharply downward, lids tighten, pupils contract, the brow center locks, nostrils open a fraction and then stop, all softness leaves the mouth, the lip line narrows, and the jawline hardens. It reads as a trained freeze, not exaggerated fright. As her weight compresses the tile, the boot sole drags microscopically across stone, fine dust squeezes from the seam, and tiny grains roll from the edge. Her sudden stop creates small delayed inertia in the hair, straps, fabric, and upper torso before they settle. [Shot 6] At 00:10.5-00:12.8, cut to a side half-body close shot with a short controlled lateral move that follows her turn and attempted retreat. She snaps her head back to retreat, but after only half a step she sees that two floor tiles behind her have silently risen half an inch, creating a subtle break line in the escape path. She instantly plants her left palm against the pillar edge, tightens the core to stop her body, and lifts the right arm halfway as if preparing to pivot around the pillar and escape. Her fifth emotional layer is controlled fear merged with rapid calculation: the brow presses lower, the eyes scan the retreat path and then cut back toward the dark opening ahead with precise motion, nostrils widen more visibly with faster breathing, the lips part for an instant and clamp shut again as she swallows the urge to gasp, and the cheek tension and jaw remain locked. At the same moment, both low candles shiver once in sync. Their flames dip under a brief air-pressure pulse and rebound. Fine golden dust begins to reverse-flow through the pillar seams and floor carvings toward the deeper arch, not as a violent magical blast but as a slow unseen draw. The lifted floor tiles shed dust, sand, and rubbing powder. [Shot 7] At 00:12.8-00:15, a medium close shot with a slight low angle pushes in half a step and then suspends without cutting to black. She is half-turned, right arm raised and ready to use the pillar for a vaulting escape, but the motion stops before release. In the deepest darkness beneath the arch ahead, a single tiny but razor-sharp golden reflection appears, like a beast eye, fixed directly on her. She does not recoil; she is forced into stillness by the correction of understanding. Her sixth emotional layer is cold recognition after a shattered assumption: the gaze locks fully for the first time, pupils fixed on that point of gold, upper lids slightly raised, lower lids tightened, brows lifting only by a thread as her previous judgment is overturned, a short controlled exhale slips through the nose, lips part slightly, mouth corners press down, and the jaw remains tense. She says in a quieter, colder throat-close voice, <d>[English] That’s not a trail.</d> with “not” the lightest but hardest word and “trail” cut very short. All the golden dust and fine particles continue to stream toward that single reflective point. Her face remains mostly in shadow, with only the eye corner, nose bridge edge, wet lip line, and collarbone rim caught by mixed cold reflection from the arch and warm low candle bounce. The golden point stays tiny, sharp, and singular, never expanding into a full creature or large supernatural display. Hold on this suspended threat state, with particles still moving and the corridor still alive with pressure.\n\noverall_soundscape: Maintain a realistic ancient temple acoustic bed throughout: low hollow room resonance, distant dripping water, faint insects, candle burn, boot friction on damp stone, palm sliding across humid carved relief, fine grain contact as fingers skim the golden powder, and subtle fabric and gear movement. Emphasize shallow controlled breathing, the side-step from behind the pillar, dust disturbed into the light beam, the quiet stop as she notices the medallion, fingertip contact with the gold dust, tiny grains lifting and falling, the crisp mechanism bite click under her foot, a brief oppressive reduction in ambience under her frozen breath, the silent rise of the tiles, sand sliding from edges, her palm planting against the pillar, the synchronized candle flutter, and the faint inward draw of air carrying dust toward the arch. End with corridor resonance, low candle crackle, drifting particulate movement, restrained breathing, a tiny pressure in the air around the golden point, and her cold close line <d>[English] That’s not a trail.</d>\n\nnon_diegetic_music: Keep music minimal and subordinate to the physical environment. Use only a very thin tension bed in the first half, almost below perception, then tighten the low-frequency pressure slightly after the mechanism click. Do not swell into action music or melodrama. End suspended on unresolved tension.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 13,
      "views": 459,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100245058687484040",
    "title": "跨越剪辑点的霓虹角色短片",
    "titleEn": "A Neon Character Film that Survives the Cut",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两个十五秒片段无缝拼接，红黑角色、颗粒和色彩在转场前后保持统一。 原生视频规格为 25 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two fifteen-second clips join without manual continuity fixes while the red-black character, grain, and grade remain consistent. Native-video output: 25s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ausdauer73/status/2100245058687484040",
    "sourceLabel": "X 原帖 · @ausdauer73",
    "author": "Ariel Alejandro Ibañez",
    "publishedAt": "2026-09-16T15:26:57.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100245058687484040.mp4",
    "posterUrl": "/posters/x/2100245058687484040.jpg",
    "duration": 25,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "anime",
      "dark"
    ],
    "scenes": [
      "nature",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Two 15-sec clips stitched into one shot. Zero manual continuity fixes.\nBeen testing @Hailuo_AI Minimax H3 Max - the character, grain, and color grade all survived the cut point. That's usually where these tools break.\n\nLive now on MiniMax Design.\n#AI #H3Max",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 70,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100243678556025201",
    "title": "三种角色初次见面构图",
    "titleEn": "Three Ways to Stage a First Meeting",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一组动画角色以三种开场构图出现在天台，比较人物入场和镜头关系。 原生视频规格为 9 秒 · 1216×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same animated characters appear on a rooftop in three opening compositions that compare entrances and staging. Native-video output: 9s · 1216×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SmallKino_/status/2100243678556025201",
    "sourceLabel": "X 原帖 · @SmallKino_",
    "author": "SmallKino",
    "publishedAt": "2026-09-16T15:21:28.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100243678556025201.mp4",
    "posterUrl": "/posters/x/2100243678556025201.jpg",
    "duration": 9,
    "aspectRatio": "portrait",
    "resolution": "1216×1344",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Asa and Denji have finally met 💕\n\nWhich clip is the best introduction? 1, 2 or 3?\n#chainsawman #チェンソーマン‌ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 15,
      "views": 483,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100242170917564750",
    "title": "角色原画变成十秒动画广告",
    "titleEn": "A Character Illustration Becomes a Ten-Second Ad",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "角色原画被扩展成九宫格表情，再以蓝白青绿配色和动态文字组成动画广告。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character illustration expands into a nine-panel expression sheet and then a blue-white-lime animated advertisement. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SEIJIailab/status/2100242170917564750",
    "sourceLabel": "X 原帖 · @SEIJIailab",
    "author": "SEIJI｜地方映像会社のAI実験室",
    "publishedAt": "2026-09-16T15:15:28.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100242170917564750.mp4",
    "posterUrl": "/posters/x/2100242170917564750.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "anime",
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "1枚のキャラクターから、10秒のアニメ広告へ。\n\nTsubaki.3で原画を生成→ChatGPTで表情と9分割デザインを展開→RunwayのMiniMax H3で動画化。\n\n青・白・ライムで統一して、表情と文字が動く映像を試しました。\nどの表情が好きですか？\n\n#PixAI #Tsubaki3 #AI動画",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 279,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100235577694314878",
    "title": "雪地里的三只小动物",
    "titleEn": "Three Small Creatures in the Snow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三只圆润的卡通动物在雪地里同步活动，保持角色造型和冬日环境一致。 原生视频规格为 10 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three rounded cartoon creatures move together in the snow while preserving their designs and wintry setting. Native-video output: 10s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/okonogiret/status/2100235577694314878",
    "sourceLabel": "X 原帖 · @okonogiret",
    "author": "小此木",
    "publishedAt": "2026-09-16T14:49:16.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100235577694314878.mp4",
    "posterUrl": "/posters/x/2100235577694314878.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "creatures",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おやすみ\n#MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 32,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100231576319594639",
    "title": "宽袖角色从姿势表跳起舞",
    "titleEn": "A Wide-Sleeved Character Dances from a Pose Sheet",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "十六姿势角色表被转成轻快舞蹈，宽大衣袖随旋转和步伐连续摆动。 原生视频规格为 15 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A sixteen-pose character sheet becomes a lively dance with wide sleeves flowing through turns and steps. Native-video output: 15s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/scenepond/status/2100231576319594639",
    "sourceLabel": "X 原帖 · @scenepond",
    "author": "Scenepond | AI Video Studio",
    "publishedAt": "2026-09-16T14:33:22.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100231576319594639.mp4",
    "posterUrl": "/posters/x/2100231576319594639.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "illustration",
      "anime"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Turned this 16-pose sheet into a little dance with MiniMax H3. 🌸\nWanted to see those wide sleeves in motion.\n\nPose sheet: GPT Image 2.5 Sunburst\nMade in Scenepond.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 60,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100229763474362633",
    "title": "飞天壁纸跳舞并整理桌面",
    "titleEn": "A Flying Apsara Dances and Repairs the Desktop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "敦煌飞天从动态壁纸中起舞，震落桌面图标后又匆忙把它们恢复原位。 原生视频规格为 15 秒 · 1918×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A Dunhuang-inspired apsara dances out of a live wallpaper, knocks icons loose, and rushes to put them back. Native-video output: 15s · 1918×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KscYusuf/status/2100229763474362633",
    "sourceLabel": "X 原帖 · @KscYusuf",
    "author": "Yusuf K",
    "publishedAt": "2026-09-16T14:26:10.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100229763474362633.mp4",
    "posterUrl": "/posters/x/2100229763474362633.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1918×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "illustration",
      "commercial"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Dynamic desktop wallpapers are getting surprisingly good.\n\nI generated the base image with GPT Image 2.5, then animated it with MiniMax H3 Max.\n\nThe idea was simple: a Dunhuang-inspired flying apsara dances, shakes the wallpaper, knocks the desktop icons out of place, then rushes to fix everything.\n\nFast to make, low cost, and the result came out better than I expected.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 230,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100222940336005581",
    "title": "暗黑短剧的竖屏片头",
    "titleEn": "A Dark Vertical Title Sequence for a Short Drama",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物、发丝和黑白文字以快速碎片化转场组合成一段竖屏短剧片头。 原生视频规格为 19 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Character fragments, flowing hair, and monochrome typography combine into a rapid vertical title sequence. Native-video output: 19s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Magncsans/status/2100222940336005581",
    "sourceLabel": "X 原帖 · @Magncsans",
    "author": "Foyege",
    "publishedAt": "2026-09-16T13:59:03.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100222940336005581.mp4",
    "posterUrl": "/posters/x/2100222940336005581.jpg",
    "duration": 19,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark",
      "stylized"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3很适合做短剧片头包装啊\n虽然给我一种看不懂剪还乱但感觉又很厉害的感觉😂",
    "engagement": {
      "replies": 5,
      "reposts": 3,
      "likes": 15,
      "views": 1228,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100218324441502164",
    "title": "拼贴文字中的绿色舞步",
    "titleEn": "A Green Dance through Typographic Collage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "真人角色在黑白文字与绿色墨迹的拼贴空间中连续起舞。 原生视频规格为 10 秒 · 1200×1680。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A live-action character dances through a collage of black-and-white typography and green ink effects. Native-video output: 10s · 1200×1680. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaekagamiAi/status/2100218324441502164",
    "sourceLabel": "X 原帖 · @MaekagamiAi",
    "author": "前鏡",
    "publishedAt": "2026-09-16T13:40:43.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100218324441502164.mp4",
    "posterUrl": "/posters/x/2100218324441502164.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1200×1680",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ただいま。refMODを使ってみたお🤗\nそしてサボってる間に、またあいちん消された？😳\n\n#MiniMaxH3 🐸",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 13,
      "views": 644,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100216313427931610",
    "title": "三名角色独立运动的雨夜舞台",
    "titleEn": "Three Independently Moving Characters on a Rainy Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三名动画角色在同一画面中分别完成动作，配合倒影、灯光和音乐形成雨夜舞台。 原生视频规格为 55 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three animated characters move independently within one frame as reflections, lighting, and music create a rainy-night stage. Native-video output: 55s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shiyou27700/status/2100216313427931610",
    "sourceLabel": "X 原帖 · @shiyou27700",
    "author": "shiyou/ココモヤ",
    "publishedAt": "2026-09-16T13:32:43.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100216313427931610.mp4",
    "posterUrl": "/posters/x/2100216313427931610.jpg",
    "duration": 55,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "夜22時の実験室\n\nHailuo で同一画面上にある３つの要素が別々の動きをしてくれるかTest\n\nCharacter images: Midjourney\nVideo: MiniMax H3（within HailuoAI)\nMusic: SUNOV6\nEditing: Adobe Ru\n\n#HailuoAI\n#HailuoCPP\n＃MiniMax H3\n@Hailuo_AI \n\nご参考までに\n動画生成時のプロンプトはリプ欄に",
    "engagement": {
      "replies": 3,
      "reposts": 9,
      "likes": 41,
      "views": 3805,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100208089802887287",
    "title": "异星河谷中的巨型生物",
    "titleEn": "Colossal Creatures across an Alien River Valley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "探险者在荒漠、河流和山谷间遇到多种巨型异星生物，组成一分钟世界观短片。 原生视频规格为 60 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An explorer encounters several colossal alien creatures across deserts, rivers, and valleys in a one-minute world-building film. Native-video output: 60s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/os_orabi/status/2100208089802887287",
    "sourceLabel": "X 原帖 · @os_orabi",
    "author": "osama orabi",
    "publishedAt": "2026-09-16T13:00:03.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100208089802887287.mp4",
    "posterUrl": "/posters/x/2100208089802887287.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cg-3d",
      "cinematic-realistic"
    ],
    "scenes": [
      "creatures",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What Lives on This Planet?\n\nExploring a distant world, they discover strange, colossal creatures wandering through rivers and mountains.\n\nMiniMaxH3 @Hailuo_AI \n#MiniMaxH3 #minimaxdesign",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 28,
      "views": 221,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100206062225788986",
    "title": "墨水停在水杯的上方边界",
    "titleEn": "Ink Stops below the Top of the Glass",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "固定机位观察墨水在杯中扩散，却始终不进入最上方三分之一的清水区域。 原生视频规格为 15 秒 · 3780×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A locked camera watches ink diffuse through a glass while the top third remains clear. Native-video output: 15s · 3780×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the top third of the water never darkens",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2100206062225788986",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2100206062225788986",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-16T12:51:59.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100206062225788986.mp4",
    "posterUrl": "/posters/x/2100206062225788986.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3780×2160",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Max. Ink in water. It stops where I told it to.\n\nPrompt: \"the top third of the water never darkens\"\n\nDuration: 15s. Camera: locked, glass height.\n\nDiffusion wants to fill the glass. A boundary written as a fraction of the container is one it can measure against.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 11,
      "views": 338,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100202004413067292",
    "title": "和服角色的安静午后",
    "titleEn": "A Quiet Afternoon with a Kimono Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏角色在和室中缓慢调整姿势和视线，保持安静的午后氛围。 原生视频规格为 16 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical character study follows subtle pose and gaze changes in a quiet Japanese-style room. Native-video output: 16s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/oreimohensan/status/2100202004413067292",
    "sourceLabel": "X 原帖 · @oreimohensan",
    "author": "人の海(H3・半自動動画生成システム運用中)",
    "publishedAt": "2026-09-16T12:35:52.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100202004413067292.mp4",
    "posterUrl": "/posters/x/2100202004413067292.jpg",
    "duration": 16,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "anime",
      "vertical"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "数日振りにH3動画。静かな午後【MiniMaxH3動画】 #MiniMaxH3 #AI動画",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 30,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100200346580931067",
    "title": "借火者唤醒神殿守卫",
    "titleEn": "The Fire Borrower Awakens the Temple Guard",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "女探险者借绳索夺取遗物并落地撤退，却发现手中的火光正在唤醒石像机关。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An explorer swings in to seize a relic and retreats, only to discover that its light is waking the stone guardian. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second cinematic photoreal image-to-video sequence extending only inside the same collapsed ancient temple main hall shown in <Picture 1>. Preserve the exact same young female explorer throughout: same face structure, wet skin realism, hairstyle, clothing, proportions, chest and waist silhouette, and performance identity. Keep the same hall depth, skylight opening, strong side-back volumetric beams, humid dusty air, sparse drifting embers, scattered rubble, hanging thick rope, and giant stone idol altar ahead. No exterior, no extra chambers, no extra people, no subtitles, no watermark, no face drift, no hand errors, no extra limbs, no floating motion, and no cut to black. The tone is elegant, sensual, dangerous, and tightly narrative, with sensuality coming only from force, breath, danger, and sculpted light during motion. The camera stays physically close to her action path: intimate near-body glide in the first half, then over-shoulder follow and a slow push in the second half, with only one very short match cut at the final reversal to the beast-eye extreme close-up. [Shot 2] At 00:00-00:02, she is already suspended mid-swing in the first frame, gripping the thick rope with her main hand. Start on the gloved hand, wrist, forearm tension, shoulder strap load, collarbone, upper chest breathing, and upper torso under strain, then glide to her face while she surges forward through dusty humid air. Her body forms a dangerous arc: both legs tucked back, pelvis and spine curving with the swing, shoulder-back muscles, collarbone, chest, and core all under real load. Her face is in hunter-like target lock, not performing to camera. Pupils are sharply focused, upper lids narrowed, gaze pinned to the altar ahead, brow center drawn inward only a little, nostrils open from controlled short breathing, lips closed with a tight natural line, lower lip showing a tiny breath tremor, and the jawline locked cleanly. Wet hair strands whip back in the backlight. She murmurs in a low breath-close voice, <d>[English] There you are.</d> [Shot 3] At 00:02-00:04, stay on the same action axis and follow her swing-dive from a close slightly low position. She cuts through the cold skylight beam and suspended dust, using her core to correct trajectory while the rope, shoulders, and back control the travel line. The giant idol altar opens ahead with clear spatial readability. Her emotion shifts from target lock into imminent strike decision: the gaze narrows further, becomes harder and more precise, the eyeballs track the relic position, brow tension deepens slightly, one mouth corner tightens almost invisibly, breathing deepens into a preparatory inhale, the throat moves lightly, and the head tips forward a little like the final calibration before a dive attack. [Shot 4] At 00:04-00:06, she passes the idol altar at speed. One hand snaps out and seizes the relic embedded in a recess of the stone idol while the other hand yanks the rope to redirect momentum. The grab must read with exact hard physics: finger closure, wrist torque, forearm tension, shoulder engagement, and a fast believable pull free. The relic emits a restrained cold-gold or cool amber glow as it leaves the socket, illuminating only her palm, knuckles, lower jaw edge, lower lip edge, and a small patch below the collarbone. Her emotion enters razor focus with a trace of suppressed acquisition thrill: eyes locked on the relic, blinks reduced, upper lip tightening, lower lip controlled under load, cheek line and jawline sharpening, no broad expression, only concentrated force and compressed near-success charge. Fine stone dust and tiny chips fall from the empty recess. [Shot 5] At 00:06-00:08.5, her body drops into landing. The toe grazes the floor first, then the boot sole scrapes through grit, thin dust, and scattered stone fragments. One knee dips, pelvis and shoulder line are tugged off center by inertia, then she catches and controls the slide immediately. Keep the action chain continuous and readable: seize, rope-pull, descend, toe-skim, scrape, slide, stabilize. Hair, straps, clothing edges, belt pieces, and small hanging gear all show slight delayed swing and return. As she stabilizes, she lifts her eyes for a fast scan, first ahead, then back to the stone idol. Her breathing grows one beat heavier, lips part slightly as if to release air, then close again. Her emotion turns from sharp success into rapid confirmation and renewed alertness: a tiny lift at the brow peak, the sense of “got it, keep moving.” [Shot 6] At 00:08.5-00:11.5, she turns immediately and retreats half a step without stopping, pressing the relic tight to her rib side with one hand while the other rises into a defensive balancing posture. Her center of gravity stays low, ready to spring again. The camera stays half a step behind her left side, then eases into an over-shoulder relationship with the idol without crossing axis. The tempo slows slightly to let the abnormality bloom. The stone idol begins to change: first a low inner resonance, then surface dust falling, tiny grit trembling and hopping, then hairline cracks extending along the real stone grain, not as a glowing graphic but as physical fracture progression. Her emotion shifts from controlled motion to instinctive alarm while still moving: pupils contract slightly, the eyes widen only in a predatory-alert way, the brow center tightens again, the nostrils pinch for one beat as the breath catches then release, the lips part as if she is about to speak but suppresses it, the jaw locks harder, and the side-neck muscles stand out. She does not freeze; she perceives danger inside the retreat. [Shot 7] At 00:11.5-00:15, over her shoulder, the camera slowly pushes toward the idol face. In the crack there is first only a line of wet cold reflection, then a living beast eye gradually resolves from the dark interior, damp and organic, with a tiny sharp reflective point on the surface. The pupil makes one slight constriction and holds directly on her. In the final two seconds, perform one extremely short match cut from the shoulder push-in to an extreme close-up of the beast eye, making this the only hard visual nail in the sequence. She retreats only half a step more but does not flee. Her head lifts slightly and her gaze is pinned. Her final emotion is cold recognition after suppressed shock: pupils tighten then steady, eye corners hold tension, the brow stops deepening and freezes into a controlled guarded position, lips part naturally for the line, mouth shape stays restrained and realistic, then after speaking only the faintest downturn at the corners and a suspended breath remain. She says quietly, <d>[English] This isn't treasure.</d> The first half carries a tiny pause of dawning realization; the second lands lower, colder, and more solid. End on the suspended stare between her and the beast eye, with dust, moisture, and tiny particles still moving inside the same main hall. Do not cut to black.\n\noverall_soundscape: Maintain a temple interior bed throughout: hollow stone reverb, rope friction under load, distant fire crackle, occasional tiny falling debris, boot scrape on stone, dust dragged across the floor, low architectural resonance, humid air movement, and sparse realistic embers. Emphasize rope strain, close breathing, fabric and strap tension, air split by her swing, hand impact on the relic, stone scrape, fine chips falling, toe contact, boot scrape through grit, gear settling, low inner stone resonance, dust drops, faint crack propagation, and her tightening breath. Keep the hall quiet but pressurized in the final section, with slow crack settling, faint wet inner reflection presence, suspended breathing, tiny falling stone particles, and her cold low line <d>[English] This isn't treasure.</d> No loud drums, no bombastic score, no extra voices, and no exaggerated monster roar.\n\nnon_diegetic_music: Keep music minimal and secondary to the environment. Use only a thin low-frequency suspense bed under the temple acoustics, almost subliminal in the first half, tightening slightly after the relic is removed and the idol begins to resonate. Do not swell into action music or horror stingers. End on unresolved low pressure and suspended threat.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2100200346580931067",
    "sourceUrl": "https://x.com/ou_zhen599/status/2100200346580931067",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-16T12:29:17.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100200346580931067.mp4",
    "posterUrl": "/posters/x/2100200346580931067.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                            \n\n🎬Image-to-Video             \n📷“The Fire Borrower”\nLaura thought she was getting closer to the treasure, but in reality, the fire in her hand was activating the temple’s security system.\n\nMiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second cinematic photoreal image-to-video sequence extending only inside the same collapsed ancient temple main hall shown in <Picture 1>. Preserve the exact same young female explorer throughout: same face structure, wet skin realism, hairstyle, clothing, proportions, chest and waist silhouette, and performance identity. Keep the same hall depth, skylight opening, strong side-back volumetric beams, humid dusty air, sparse drifting embers, scattered rubble, hanging thick rope, and giant stone idol altar ahead. No exterior, no extra chambers, no extra people, no subtitles, no watermark, no face drift, no hand errors, no extra limbs, no floating motion, and no cut to black. The tone is elegant, sensual, dangerous, and tightly narrative, with sensuality coming only from force, breath, danger, and sculpted light during motion. The camera stays physically close to her action path: intimate near-body glide in the first half, then over-shoulder follow and a slow push in the second half, with only one very short match cut at the final reversal to the beast-eye extreme close-up. [Shot 2] At 00:00-00:02, she is already suspended mid-swing in the first frame, gripping the thick rope with her main hand. Start on the gloved hand, wrist, forearm tension, shoulder strap load, collarbone, upper chest breathing, and upper torso under strain, then glide to her face while she surges forward through dusty humid air. Her body forms a dangerous arc: both legs tucked back, pelvis and spine curving with the swing, shoulder-back muscles, collarbone, chest, and core all under real load. Her face is in hunter-like target lock, not performing to camera. Pupils are sharply focused, upper lids narrowed, gaze pinned to the altar ahead, brow center drawn inward only a little, nostrils open from controlled short breathing, lips closed with a tight natural line, lower lip showing a tiny breath tremor, and the jawline locked cleanly. Wet hair strands whip back in the backlight. She murmurs in a low breath-close voice, <d>[English] There you are.</d> [Shot 3] At 00:02-00:04, stay on the same action axis and follow her swing-dive from a close slightly low position. She cuts through the cold skylight beam and suspended dust, using her core to correct trajectory while the rope, shoulders, and back control the travel line. The giant idol altar opens ahead with clear spatial readability. Her emotion shifts from target lock into imminent strike decision: the gaze narrows further, becomes harder and more precise, the eyeballs track the relic position, brow tension deepens slightly, one mouth corner tightens almost invisibly, breathing deepens into a preparatory inhale, the throat moves lightly, and the head tips forward a little like the final calibration before a dive attack. [Shot 4] At 00:04-00:06, she passes the idol altar at speed. One hand snaps out and seizes the relic embedded in a recess of the stone idol while the other hand yanks the rope to redirect momentum. The grab must read with exact hard physics: finger closure, wrist torque, forearm tension, shoulder engagement, and a fast believable pull free. The relic emits a restrained cold-gold or cool amber glow as it leaves the socket, illuminating only her palm, knuckles, lower jaw edge, lower lip edge, and a small patch below the collarbone. Her emotion enters razor focus with a trace of suppressed acquisition thrill: eyes locked on the relic, blinks reduced, upper lip tightening, lower lip controlled under load, cheek line and jawline sharpening, no broad expression, only concentrated force and compressed near-success charge. Fine stone dust and tiny chips fall from the empty recess. [Shot 5] At 00:06-00:08.5, her body drops into landing. The toe grazes the floor first, then the boot sole scrapes through grit, thin dust, and scattered stone fragments. One knee dips, pelvis and shoulder line are tugged off center by inertia, then she catches and controls the slide immediately. Keep the action chain continuous and readable: seize, rope-pull, descend, toe-skim, scrape, slide, stabilize. Hair, straps, clothing edges, belt pieces, and small hanging gear all show slight delayed swing and return. As she stabilizes, she lifts her eyes for a fast scan, first ahead, then back to the stone idol. Her breathing grows one beat heavier, lips part slightly as if to release air, then close again. Her emotion turns from sharp success into rapid confirmation and renewed alertness: a tiny lift at the brow peak, the sense of “got it, keep moving.” [Shot 6] At 00:08.5-00:11.5, she turns immediately and retreats half a step without stopping, pressing the relic tight to her rib side with one hand while the other rises into a defensive balancing posture. Her center of gravity stays low, ready to spring again. The camera stays half a step behind her left side, then eases into an over-shoulder relationship with the idol without crossing axis. The tempo slows slightly to let the abnormality bloom. The stone idol begins to change: first a low inner resonance, then surface dust falling, tiny grit trembling and hopping, then hairline cracks extending along the real stone grain, not as a glowing graphic but as physical fracture progression. Her emotion shifts from controlled motion to instinctive alarm while still moving: pupils contract slightly, the eyes widen only in a predatory-alert way, the brow center tightens again, the nostrils pinch for one beat as the breath catches then release, the lips part as if she is about to speak but suppresses it, the jaw locks harder, and the side-neck muscles stand out. She does not freeze; she perceives danger inside the retreat. [Shot 7] At 00:11.5-00:15, over her shoulder, the camera slowly pushes toward the idol face. In the crack there is first only a line of wet cold reflection, then a living beast eye gradually resolves from the dark interior, damp and organic, with a tiny sharp reflective point on the surface. The pupil makes one slight constriction and holds directly on her. In the final two seconds, perform one extremely short match cut from the shoulder push-in to an extreme close-up of the beast eye, making this the only hard visual nail in the sequence. She retreats only half a step more but does not flee. Her head lifts slightly and her gaze is pinned. Her final emotion is cold recognition after suppressed shock: pupils tighten then steady, eye corners hold tension, the brow stops deepening and freezes into a controlled guarded position, lips part naturally for the line, mouth shape stays restrained and realistic, then after speaking only the faintest downturn at the corners and a suspended breath remain. She says quietly, <d>[English] This isn't treasure.</d> The first half carries a tiny pause of dawning realization; the second lands lower, colder, and more solid. End on the suspended stare between her and the beast eye, with dust, moisture, and tiny particles still moving inside the same main hall. Do not cut to black.\n\noverall_soundscape: Maintain a temple interior bed throughout: hollow stone reverb, rope friction under load, distant fire crackle, occasional tiny falling debris, boot scrape on stone, dust dragged across the floor, low architectural resonance, humid air movement, and sparse realistic embers. Emphasize rope strain, close breathing, fabric and strap tension, air split by her swing, hand impact on the relic, stone scrape, fine chips falling, toe contact, boot scrape through grit, gear settling, low inner stone resonance, dust drops, faint crack propagation, and her tightening breath. Keep the hall quiet but pressurized in the final section, with slow crack settling, faint wet inner reflection presence, suspended breathing, tiny falling stone particles, and her cold low line <d>[English] This isn't treasure.</d> No loud drums, no bombastic score, no extra voices, and no exaggerated monster roar.\n\nnon_diegetic_music: Keep music minimal and secondary to the environment. Use only a thin low-frequency suspense bed under the temple acoustics, almost subliminal in the first half, tightening slightly after the relic is removed and the idol begins to resonate. Do not swell into action music or horror stingers. End on unresolved low pressure and suspended threat.",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 8,
      "views": 570,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100199701286936594",
    "title": "花园围栏边的邻里八卦",
    "titleEn": "Neighborhood Gossip over the Garden Fence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黏土动画风格的三位邻居隔着花园围栏聊天，最后用水桶笑点收尾。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three claymation-style neighbors gossip over a garden fence before a bucket-based punchline closes the scene. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aibabbler/status/2100199701286936594",
    "sourceLabel": "X 原帖 · @aibabbler",
    "author": "Jo Aiba",
    "publishedAt": "2026-09-16T12:26:43.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100199701286936594.mp4",
    "posterUrl": "/posters/x/2100199701286936594.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I made this AI short, \"Mind Your Own Bucket,\" using 3D character animation with a claymation-style design language in Minimax H3 on Higgsfield.\n\nThree neighbours in classic British sitcom fashion gossip over the garden fence, hair rollers and pearls in full character-design glory.\n\nPrompt below 👇🏾",
    "engagement": {
      "replies": 18,
      "reposts": 1,
      "likes": 29,
      "views": 533,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100198935545663805",
    "title": "九宫格角色走入真实街景",
    "titleEn": "A Nine-Panel Character Walks into Real Streets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "粉发动画角色从九宫格素材延展到草地、车站和夜间街道的连续片段。 原生视频规格为 10 秒 · 480×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired animated character expands from a nine-panel source into connected clips across a lawn, station, and night street. Native-video output: 10s · 480×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/beko_X_account/status/2100198935545663805",
    "sourceLabel": "X 原帖 · @beko_X_account",
    "author": "beko",
    "publishedAt": "2026-09-16T12:23:40.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100198935545663805.mp4",
    "posterUrl": "/posters/x/2100198935545663805.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "480×832",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "ugc"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "GPT-image2.5で3*3のイラストを生成。\nCodexでプロンプト作成、そのままイラストを9枚にカットしてローカルMinimaxH3で連続したアニメーションとして生成。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 167,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100191389275340978",
    "title": "夜间赛道上的高速追逐",
    "titleEn": "A High-Speed Chase on a Night Circuit",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "赛车手与黑色方程式赛车在夜间赛道高速穿梭，镜头在座舱、侧面和追尾视角间切换。 原生视频规格为 10 秒 · 1536×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A driver and black open-wheel car race through a night circuit as the camera cuts between cockpit, side, and chase views. Native-video output: 10s · 1536×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YaseenK7212/status/2100191389275340978",
    "sourceLabel": "X 原帖 · @YaseenK7212",
    "author": "Yaseen Khan Gul",
    "publishedAt": "2026-09-16T11:53:41.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100191389275340978.mp4",
    "posterUrl": "/posters/x/2100191389275340978.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1536×672",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Honestly didn't expect MiniMax H3 Max to move this fast @Hailuo_AI",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 21,
      "views": 55356,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100189410646241626",
    "title": "两指挡下毁灭世界的陨石",
    "titleEn": "Stopping a World-Ending Meteor with Two Fingers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画角色面对燃烧陨石，以两根手指在撞击前将其停下，形成夸张的力量反差。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated fighter faces a burning meteor and halts it with two fingers just before impact. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2100189410646241626",
    "sourceLabel": "X 原帖 · @tokyo_Valentine",
    "author": "とすくん",
    "publishedAt": "2026-09-16T11:45:49.000Z",
    "addedAt": "2026-09-17T01:13:16.892Z",
    "mediaUrl": "/media/x-2100189410646241626.mp4",
    "posterUrl": "/posters/x/2100189410646241626.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "世界を滅ぼす隕石、指2本で……\nMiniMax H3 Maxで制作しました🔥\n\nプロンプトはリプにあります。ご自由にご利用ください。\n\nアニメの表現とMiniMax H3の相性が最高ですね。\n\nMade with\n@Hailuo_AI #minimaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 32,
      "views": 1361,
      "snapshotAt": "2026-09-17T01:13:16.892Z"
    },
    "approvedAt": "2026-09-17T01:13:16.892Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101021004713578691",
    "title": "Boa Hancock 释放紫色能量",
    "titleEn": "Boa Hancock Releases Purple Energy",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫角色在宫殿场景中展开紫色能量，镜头从手部特写推进到完整人物亮相。 原生视频规格为 15 秒 · 1920×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime character releases purple energy in a palace setting as the camera moves from a hand close-up to a full reveal. Native-video output: 15s · 1920×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Aliaksei_AI/status/2101021004713578691",
    "sourceLabel": "X 原帖 · @Aliaksei_AI",
    "author": "Aliaksei.AI",
    "publishedAt": "2026-09-18T18:50:17.000Z",
    "addedAt": "2026-09-18T19:03:16.521Z",
    "mediaUrl": "/media/x-2101021004713578691.mp4",
    "posterUrl": "/posters/x/2101021004713578691.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Boa Hancock Beauty Presentation.\n#AIanimation made with #MiniMaxH3\nInspired by @aimikoda",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 18,
      "snapshotAt": "2026-09-18T19:03:16.521Z"
    },
    "approvedAt": "2026-09-18T19:03:16.521Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101012480214642865",
    "title": "可口可乐与曼妥思的爆炸实验",
    "titleEn": "A Coca-Cola and Mentos Explosion Experiment",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "绿色服装人物穿梭于可口可乐与曼妥思卡车之间，泡沫喷涌形成一支产品喜剧短片。 原生视频规格为 15 秒 · 1632×928。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character moves between Coca-Cola and Mentos trucks while a foamy eruption turns the product setup into a comic short. Native-video output: 15s · 1632×928. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI_EmeraldApple/status/2101012480214642865",
    "sourceLabel": "X 原帖 · @AI_EmeraldApple",
    "author": "Emerald Apple",
    "publishedAt": "2026-09-18T18:16:24.000Z",
    "addedAt": "2026-09-18T19:03:16.521Z",
    "mediaUrl": "/media/x-2101012480214642865.mp4",
    "posterUrl": "/posters/x/2101012480214642865.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1632×928",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Adapted the prompt for Minimax H3 using multi-reference to video on local AI video generation using an Nvidia 5090.\n\nEmerald Apple gets destroyed by the Coca-Cola Mentos catastrophe!\n\nNot perfect, but pretty cool that locally run offline AI video can even get to this level.",
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 28,
      "views": 641,
      "snapshotAt": "2026-09-18T19:03:16.521Z"
    },
    "approvedAt": "2026-09-18T19:03:16.521Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101008623555997920",
    "title": "雨夜温室里只让一朵花开放",
    "titleEn": "One Flower Opens inside a Rainy Greenhouse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "锁定机位拍摄雨夜温室，一朵花逐圈开放，其他植物保持静止。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A locked camera observes a rainy greenhouse as one flower opens ring by ring while the surrounding plants remain still. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the petals open one ring at a time from the outside in",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2101008623555997920",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2101008623555997920",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-18T18:01:05.000Z",
    "addedAt": "2026-09-18T19:03:16.521Z",
    "mediaUrl": "/media/x-2101008623555997920.mp4",
    "posterUrl": "/posters/x/2101008623555997920.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. One lamp, one flower, rain on the glass.\n\nPrompt: \"the petals open one ring at a time from the outside in\"\n\nDuration: 15s. Camera: locked, pot height. Rendered locally in ComfyUI, then Topaz to 4K.\n\n\"No other plant moves\" is the sentence that kept the greenhouse asleep. One thing opening is a story. Everything opening is a screensaver.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 11,
      "views": 153,
      "snapshotAt": "2026-09-18T19:03:16.521Z"
    },
    "approvedAt": "2026-09-18T19:03:16.521Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101008053206135082",
    "title": "太空中的炸鸡与猫咪",
    "titleEn": "Fried Chicken and Cats in Space",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "炸鸡桶、猫咪和蔬菜在行星与星云之间漂浮，构成一支荒诞的品牌视觉短片。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bucket of fried chicken, cats, and vegetables float among planets and nebulae in an absurd branded visual short. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Got_Mittens/status/2101008053206135082",
    "sourceLabel": "X 原帖 · @Got_Mittens",
    "author": "Korektor",
    "publishedAt": "2026-09-18T17:58:49.000Z",
    "addedAt": "2026-09-18T19:03:16.521Z",
    "mediaUrl": "/media/x-2101008053206135082.mp4",
    "posterUrl": "/posters/x/2101008053206135082.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "product",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Can't believe my PC could generate this in under half an hour.\nMade with Minimax H3 through Comfy UI.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 13,
      "views": 342,
      "snapshotAt": "2026-09-18T19:03:16.521Z"
    },
    "approvedAt": "2026-09-18T19:03:16.521Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101007280040784269",
    "title": "寿司店里的酸味表情变化",
    "titleEn": "Sour-to-Delight Reaction in a Sushi Bistro",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一镜到底的寿司店表演从酸味震惊转向满足，重点呈现连续手持镜头和表情变化。 原生视频规格为 15 秒 · 1536×864。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A continuous handheld sushi-bistro performance moves from sour shock to delight through readable facial acting. Native-video output: 15s · 1536×864. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15-second cinematic single continuous handheld shot. One uninterrupted take from start to finish. No cuts, no transitions, no camera resets.\n\nThe <Picture 0> is exactly the first frame.\n\nScene:\nA fictional original young woman in her early twenties sits at the counter of a cozy daytime sushi bistro. Preserve the exact facial identity and physical appearance of the supplied character reference throughout the entire shot. She has fair natural skin with visible freckles, bright blue eyes, and long, naturally curly copper-red hair worn loose around her shoulders. Her hair has realistic individual strands, natural volume, and soft movement. She wears soft natural makeup, muted rose lips, and a chunky polished silver collar necklace.\n\nShe wears a vivid green sheer mesh long-sleeve top falling slightly off one bare shoulder over a matching green inner top, exactly matching the starting image. No visible brands or logos.\n\nThe sushi bistro has a long polished dark-wood counter with a woven rattan runner, warm cream walls, blank decorative scrolls with no readable writing, softly glowing cabinets, and a blurred open kitchen with sushi chefs moving naturally in the background. A red orchid sits nearby. A large window fills the restaurant with bright midday daylight, with lush green trees visible outside.\n\nIn front of her is a ceramic plate with assorted sushi, a small soy sauce dish, pink pickled ginger, a napkin, and wooden chopsticks.\n\nComposition:\nKeep the woman in the RIGHT THIRD of the frame for the entire shot. Never center her.\n\nShoot from slightly below her eye level, across the countertop. The long counter and blurred kitchen extend through the left two-thirds of the frame.\n\nThe foreground sushi plate and woven counter runner remain slightly out of focus along the bottom edge.\n\nMaintain a constant 10–12 degree Dutch angle throughout the entire video. Never level the horizon.\n\nCamera movement:\nPlayful natural handheld cinematography. The camera operator appears to be resting their elbows near the counter, creating subtle breathing sway, tiny micro-jitters, and gentle imperfect movement.\n\nThroughout the shot, slowly and slightly slide the camera closer along the counter toward her face.\n\nNever become perfectly static.\n\nSoft window light occasionally creates a subtle natural lens flare.\n\nAction and emotion timeline:\n\n0–1 seconds:\nThe action starts immediately on the very first frame.\n\nShe already has a whole shrimp nigiri held with her chopsticks: pale pink butterflied shrimp over a small rice pillow.\n\nShe quickly lifts it and puts the entire shrimp nigiri into her mouth in one bite.\n\nNo hesitation and no slow buildup.\n\n1–4 seconds:\nThe sour flavor hits her INSTANTLY with exaggerated comic intensity.\n\nHer eyes squeeze tightly shut.\n\nHer lips pucker dramatically around her full cheeks.\n\nHer entire face contracts from the sourness.\n\nHer shoulders shudder.\n\nShe quickly shakes her head side to side, causing her long curly red hair to bounce, whip and sway naturally around her face and shoulders.\n\nHer chopsticks slip from her hand and clatter onto the ceramic plate.\n\nShe lightly pounds one fist against the counter once or twice while continuing to chew.\n\nThe reaction must be immediate, expressive, and physically readable.\n\n4–8 seconds:\nThe sour shock turns into genuine disgust.\n\nHer expression suddenly becomes flatter and deeply repulsed.\n\nShe chews slower and slower.\n\nHer nose wrinkles.\n\nShe slightly gags.\n\nHer tongue pushes the food toward one cheek as if she cannot tolerate the taste.\n\nShe grabs the napkin and brings it toward her mouth, clearly preparing to spit the food out.\n\nHer eyes show disappointment and betrayal.\n\n8–11 seconds:\nWith the napkin touching or hovering directly in front of her lips, she suddenly freezes.\n\nHer eyebrows jump upward.\n\nHer eyes open wide.\n\nSomething about the flavor unexpectedly changes.\n\nShe slowly lowers the napkin about an inch.\n\nShe chews once.\n\nPause.\n\nShe chews a second time.\n\nHer expression changes from disgust → confusion → suspicious curiosity.\n\nMake this transformation clearly visible through her eyes, eyebrows, cheeks, and mouth.\n\n11–15 seconds:\nHer expression gradually explodes into pure delight.\n\nHer eyes close softly in pleasure.\n\nA huge genuine smile spreads across her face while she continues chewing.\n\nHer shoulders relax completely.\n\nShe leans and melts backward slightly on the stool.\n\nShe lightly kicks her feet beneath the counter from happiness.\n\nShe makes a subtle pleased humming expression without speaking.\n\nStill smiling, she reaches forward again with her chopsticks toward another shrimp nigiri on the plate, clearly wanting another bite.\n\nEnd the shot while she happily reaches for the next piece.\n\nPerformance:\nPrioritize expressive facial acting and physically readable emotion changes.\n\nThe emotional progression must be extremely clear:\n\ninstant sour shock → disgust → about to spit it out → sudden surprise → suspicious curiosity → intense pleasure\n\nNatural blinking, eye movement, cheek movement, jaw chewing, breathing, shoulder movement, hand gestures, and realistic loose curly-hair physics.\n\nIDENTITY LOCK: Preserve the woman's exact facial identity from the supplied character reference and starting frame throughout the entire video. Do not alter her facial structure, freckles, eye color, skin tone, age, or copper-red curly hair. Do not change her hairstyle into a ponytail or straight hair.\n\nVisual style:\nCinematic Japanese indie slice-of-life feeling.\n35mm film texture.\nShallow depth of field.\nSoft natural skin texture with her freckles clearly retained.\nWarm daylight mixed with a subtle teal-and-warm film color cast.\nFine soft grain.\nNatural highlight rolloff.\nCozy intimate restaurant atmosphere.\nRealistic human motion and realistic food interaction.\n\nImportant constraints:\nOne continuous 15-second shot.\nNo cuts.\nNo scene changes.\nNo time jumps.\nNo slow motion.\nNo freeze-frame effect.\nDo not center the woman.\nMaintain the Dutch angle.\nNo dialogue.\nNo captions.\nNo subtitles.\nNo readable text.\nBlank scrolls and blank menus only.\nNo logos.\nNo brands.\nNo recognizable characters.\nNo mascot figurines.\nNo copyrighted characters.\nNo real people.",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2101007280040784269",
    "sourceUrl": "https://x.com/EndFolding79421/status/2101007280040784269",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-09-18T17:55:45.000Z",
    "addedAt": "2026-09-18T19:25:59.622Z",
    "mediaUrl": "/media/x-2101007280040784269.mp4",
    "posterUrl": "/posters/x/2101007280040784269.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1536×864",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "performance",
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Comfyui — 1536x864 8 (25 minutes) vs 40 STEPS (over an hour) and also  REALISM LORA TEST. \n\nPc Specs: Laptop 5090 24 vram 64 ram\n\n***got this workflow ( I added a sage to it) from aiseek youtuber (link provided). I included both videos and also the starting image I made if you want to test it yourself.***\n\nhttps://www.youtube.com/watch?v=WPJLqZDqXmM&t=142s\n\nPrompt:\n\n15-second cinematic single continuous handheld shot. One uninterrupted take from start to finish. No cuts, no transitions, no camera resets.\n\nThe <Picture 0> is exactly the first frame.\n\nScene:\nA fictional original young woman in her early twenties sits at the counter of a cozy daytime sushi bistro. Preserve the exact facial identity and physical appearance of the supplied character reference throughout the entire shot. She has fair natural skin with visible freckles, bright blue eyes, and long, naturally curly copper-red hair worn loose around her shoulders. Her hair has realistic individual strands, natural volume, and soft movement. She wears soft natural makeup, muted rose lips, and a chunky polished silver collar necklace.\n\nShe wears a vivid green sheer mesh long-sleeve top falling slightly off one bare shoulder over a matching green inner top, exactly matching the starting image. No visible brands or logos.\n\nThe sushi bistro has a long polished dark-wood counter with a woven rattan runner, warm cream walls, blank decorative scrolls with no readable writing, softly glowing cabinets, and a blurred open kitchen with sushi chefs moving naturally in the background. A red orchid sits nearby. A large window fills the restaurant with bright midday daylight, with lush green trees visible outside.\n\nIn front of her is a ceramic plate with assorted sushi, a small soy sauce dish, pink pickled ginger, a napkin, and wooden chopsticks.\n\nComposition:\nKeep the woman in the RIGHT THIRD of the frame for the entire shot. Never center her.\n\nShoot from slightly below her eye level, across the countertop. The long counter and blurred kitchen extend through the left two-thirds of the frame.\n\nThe foreground sushi plate and woven counter runner remain slightly out of focus along the bottom edge.\n\nMaintain a constant 10–12 degree Dutch angle throughout the entire video. Never level the horizon.\n\nCamera movement:\nPlayful natural handheld cinematography. The camera operator appears to be resting their elbows near the counter, creating subtle breathing sway, tiny micro-jitters, and gentle imperfect movement.\n\nThroughout the shot, slowly and slightly slide the camera closer along the counter toward her face.\n\nNever become perfectly static.\n\nSoft window light occasionally creates a subtle natural lens flare.\n\nAction and emotion timeline:\n\n0–1 seconds:\nThe action starts immediately on the very first frame.\n\nShe already has a whole shrimp nigiri held with her chopsticks: pale pink butterflied shrimp over a small rice pillow.\n\nShe quickly lifts it and puts the entire shrimp nigiri into her mouth in one bite.\n\nNo hesitation and no slow buildup.\n\n1–4 seconds:\nThe sour flavor hits her INSTANTLY with exaggerated comic intensity.\n\nHer eyes squeeze tightly shut.\n\nHer lips pucker dramatically around her full cheeks.\n\nHer entire face contracts from the sourness.\n\nHer shoulders shudder.\n\nShe quickly shakes her head side to side, causing her long curly red hair to bounce, whip and sway naturally around her face and shoulders.\n\nHer chopsticks slip from her hand and clatter onto the ceramic plate.\n\nShe lightly pounds one fist against the counter once or twice while continuing to chew.\n\nThe reaction must be immediate, expressive, and physically readable.\n\n4–8 seconds:\nThe sour shock turns into genuine disgust.\n\nHer expression suddenly becomes flatter and deeply repulsed.\n\nShe chews slower and slower.\n\nHer nose wrinkles.\n\nShe slightly gags.\n\nHer tongue pushes the food toward one cheek as if she cannot tolerate the taste.\n\nShe grabs the napkin and brings it toward her mouth, clearly preparing to spit the food out.\n\nHer eyes show disappointment and betrayal.\n\n8–11 seconds:\nWith the napkin touching or hovering directly in front of her lips, she suddenly freezes.\n\nHer eyebrows jump upward.\n\nHer eyes open wide.\n\nSomething about the flavor unexpectedly changes.\n\nShe slowly lowers the napkin about an inch.\n\nShe chews once.\n\nPause.\n\nShe chews a second time.\n\nHer expression changes from disgust → confusion → suspicious curiosity.\n\nMake this transformation clearly visible through her eyes, eyebrows, cheeks, and mouth.\n\n11–15 seconds:\nHer expression gradually explodes into pure delight.\n\nHer eyes close softly in pleasure.\n\nA huge genuine smile spreads across her face while she continues chewing.\n\nHer shoulders relax completely.\n\nShe leans and melts backward slightly on the stool.\n\nShe lightly kicks her feet beneath the counter from happiness.\n\nShe makes a subtle pleased humming expression without speaking.\n\nStill smiling, she reaches forward again with her chopsticks toward another shrimp nigiri on the plate, clearly wanting another bite.\n\nEnd the shot while she happily reaches for the next piece.\n\nPerformance:\nPrioritize expressive facial acting and physically readable emotion changes.\n\nThe emotional progression must be extremely clear:\n\ninstant sour shock → disgust → about to spit it out → sudden surprise → suspicious curiosity → intense pleasure\n\nNatural blinking, eye movement, cheek movement, jaw chewing, breathing, shoulder movement, hand gestures, and realistic loose curly-hair physics.\n\nIDENTITY LOCK: Preserve the woman's exact facial identity from the supplied character reference and starting frame throughout the entire video. Do not alter her facial structure, freckles, eye color, skin tone, age, or copper-red curly hair. Do not change her hairstyle into a ponytail or straight hair.\n\nVisual style:\nCinematic Japanese indie slice-of-life feeling.\n35mm film texture.\nShallow depth of field.\nSoft natural skin texture with her freckles clearly retained.\nWarm daylight mixed with a subtle teal-and-warm film color cast.\nFine soft grain.\nNatural highlight rolloff.\nCozy intimate restaurant atmosphere.\nRealistic human motion and realistic food interaction.\n\nImportant constraints:\nOne continuous 15-second shot.\nNo cuts.\nNo scene changes.\nNo time jumps.\nNo slow motion.\nNo freeze-frame effect.\nDo not center the woman.\nMaintain the Dutch angle.\nNo dialogue.\nNo captions.\nNo subtitles.\nNo readable text.\nBlank scrolls and blank menus only.\nNo logos.\nNo brands.\nNo recognizable characters.\nNo mascot figurines.\nNo copyrighted characters.\nNo real people.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 151,
      "snapshotAt": "2026-09-18T19:25:59.622Z"
    },
    "approvedAt": "2026-09-18T19:25:59.622Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101007414682345719",
    "title": "霓虹街头的角色漫游",
    "titleEn": "A Character Walks through a Neon Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏角色短片把街头漫游、游戏世界和电影式氛围放在同一段连续运动中。 原生视频规格为 5 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical character short combines a street walk, a game-world mood, and cinematic movement. Native-video output: 5s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/BrooksHalladay/status/2101007414682345719",
    "sourceLabel": "X 原帖 · @BrooksHalladay",
    "author": "Brooks",
    "publishedAt": "2026-09-18T17:56:17.000Z",
    "addedAt": "2026-09-18T19:25:59.622Z",
    "mediaUrl": "/media/x-2101007414682345719.mp4",
    "posterUrl": "/posters/x/2101007414682345719.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "daily-life",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "While Astra is toxen maxxing on my kiddie Pet game in Unity, I'm listening to T2 Judgement Day at full blast and I feel like the Dark Knight. My life is unrecognizable from even a year ago.\n\nThe short film I'm making with Higgsfield AI, Fal and MiniMax H3 has Chinese and Japanese voiceovers with J-pop music tracks.\n\nYou can't make this shit up, but actually now you can.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 47,
      "snapshotAt": "2026-09-18T19:25:59.622Z"
    },
    "approvedAt": "2026-09-18T19:25:59.622Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101007520605090092",
    "title": "影院屏幕前的意外瞬间",
    "titleEn": "An Unexpected Moment in Front of the Screen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "影院观众与银幕画面形成前后景关系，短片用一个突发动作制造观看错位。 原生视频规格为 19 秒 · 672×1216。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cinema audience and the screen form a layered composition as a sudden action creates a viewing illusion. Native-video output: 19s · 672×1216. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hotforteecha/status/2101007520605090092",
    "sourceLabel": "X 原帖 · @hotforteecha",
    "author": "JustJay",
    "publishedAt": "2026-09-18T17:56:42.000Z",
    "addedAt": "2026-09-18T19:25:59.622Z",
    "mediaUrl": "/media/x-2101007520605090092.mp4",
    "posterUrl": "/posters/x/2101007520605090092.jpg",
    "duration": 19,
    "aspectRatio": "portrait",
    "resolution": "672×1216",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "ugc"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "There is a reason they tell you to stay off your phone during the movie.  Made with Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 26,
      "snapshotAt": "2026-09-18T19:25:59.622Z"
    },
    "approvedAt": "2026-09-18T19:25:59.622Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101002579643904065",
    "title": "橙色角色的惊悚转场",
    "titleEn": "A Horror Transition with an Orange Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙色角色在黑底环境中完成惊悚感动作，画面重点是形体和光影变化。 原生视频规格为 10 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An orange character performs a horror-tinged action against a dark background, emphasizing shape and lighting. Native-video output: 10s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/5PPLqpmZTL8q0gT/status/2101002579643904065",
    "sourceLabel": "X 原帖 · @5PPLqpmZTL8q0gT",
    "author": "麒麟檸檬",
    "publishedAt": "2026-09-18T17:37:04.000Z",
    "addedAt": "2026-09-18T19:25:59.622Z",
    "mediaUrl": "/media/x-2101002579643904065.mp4",
    "posterUrl": "/posters/x/2101002579643904065.jpg",
    "duration": 10,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized",
      "anime"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3で生成\nホラー演出をもう少し凝りたいところ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 20,
      "snapshotAt": "2026-09-18T19:25:59.622Z"
    },
    "approvedAt": "2026-09-18T19:25:59.622Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101002563332321703",
    "title": "奇幻 RPG 角色的加速对比",
    "titleEn": "A Fantasy RPG Speed Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一奇幻 RPG 场景对比不同本地加速方案的输出，角色、光效和动作保持在同一视觉语境。 原生视频规格为 10 秒 · 1080×1234。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The same fantasy RPG scene compares local acceleration paths while keeping the character, effects, and action in one visual context. Native-video output: 10s · 1080×1234. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/valda/status/2101002563332321703",
    "sourceLabel": "X 原帖 · @valda",
    "author": "valda",
    "publishedAt": "2026-09-18T17:37:00.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2101002563332321703.mp4",
    "posterUrl": "/posters/x/2101002563332321703.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1080×1234",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MATLOWAI/minimax-h3-fused-turbo-int8-convrot試してみた。ファンタジーRPG風、FastH3との比較\n\nComfyUI 0.36.0 / RTX 4080 / 1344×768 / 243f / 24fps\n生成時間: MATLOWAI 128.59秒 / FastH3 144.49秒\n\n約11%ほど速いけど、生成された動画はFastH3 VSAの方が好きかなぁ",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 38,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101000545846636772",
    "title": "雨天街景中的品牌冲击",
    "titleEn": "A Brand Moment in a Rainy Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雨天车窗外的城市广告与突然的水花动作结合，形成竖屏街景短片。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A city advertisement outside a rainy car window meets a sudden splash in this vertical street short. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/3dman1ac/status/2101000545846636772",
    "sourceLabel": "X 原帖 · @3dman1ac",
    "author": "Raffaele Acanfora",
    "publishedAt": "2026-09-18T17:28:59.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2101000545846636772.mp4",
    "posterUrl": "/posters/x/2101000545846636772.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "ugc"
    ],
    "scenes": [
      "city",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An attempt to recreate the video that's going viral on X using Minimax H3 locally. The experiment was only partially successful 😏\n#minimax @MiniMax_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 22,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100998769714348370",
    "title": "四人同框的步行镜头",
    "titleEn": "A Walking Shot with Four Faces",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "四位人物在明亮走廊中并肩走来，画面检验多人同框、身份一致性和自然步态。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Four people walk together through a bright corridor, testing multi-face consistency and natural gait. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/blurboy1985/status/2100998769714348370",
    "sourceLabel": "X 原帖 · @blurboy1985",
    "author": "Daniel Q.",
    "publishedAt": "2026-09-18T17:21:56.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100998769714348370.mp4",
    "posterUrl": "/posters/x/2100998769714348370.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "cinematic-realistic"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Trying out @MiniMax_AI Minimax H3 video generation capability with 4 faces on the screen 😆",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 21,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100998243266035749",
    "title": "紫色巨龙的角色海报镜头",
    "titleEn": "A Purple Dragon Character Poster Shot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫色巨龙在暗色背景中展开翅膀，画面强调角色细节、材质和戏剧化光线。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple dragon spreads its wings against a dark background, emphasizing character detail, material, and dramatic light. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/_Ractor_/status/2100998243266035749",
    "sourceLabel": "X 原帖 · @_Ractor_",
    "author": "Ractor",
    "publishedAt": "2026-09-18T17:19:50.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100998243266035749.mp4",
    "posterUrl": "/posters/x/2100998243266035749.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "cg-3d",
      "dark"
    ],
    "scenes": [
      "creatures",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MinimaxH3 #digitalart #ai #aiart #MidJourney #Dragon",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 14,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100996989953663392",
    "title": "真实世界里走来的两个人",
    "titleEn": "Two People Step into a Live-Action World",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两位人物在写实街景中出现，创作者记录了参考声音未能区分的实际结果。 原生视频规格为 10 秒 · 608×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two figures enter a realistic street scene while the creator notes the reference voices converged in the result. Native-video output: 10s · 608×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C12592947592/status/2100996989953663392",
    "sourceLabel": "X 原帖 · @C12592947592",
    "author": "supercub3",
    "publishedAt": "2026-09-18T17:14:51.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100996989953663392.mp4",
    "posterUrl": "/posters/x/2100996989953663392.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "608×1056",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal",
      "ugc"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MinimaxH3ローカル\n実写世界に舞い降りた二人\n\n音声参照がうまくいかず二人とも同じ声に🥲",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 35,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100988560023196141",
    "title": "超级英雄长镜头片段",
    "titleEn": "A Long Superhero Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "超长视频把超级英雄角色置于连续动作与战斗场景中，适合作为长视频能力样例。 原生视频规格为 154 秒 · 1920×804。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long superhero sequence places a comic-book character in continuous action and combat, useful for studying long-form output. Native-video output: 154s · 1920×804. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lexx_aura/status/2100988560023196141",
    "sourceLabel": "X 原帖 · @lexx_aura",
    "author": "Lex",
    "publishedAt": "2026-09-18T16:41:21.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100988560023196141.mp4",
    "posterUrl": "/posters/x/2100988560023196141.jpg",
    "duration": 154,
    "aspectRatio": "landscape",
    "resolution": "1920×804",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "combat",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What if homelander came in avengers 🔥\n\nMiniMax H3",
    "engagement": {
      "replies": 2,
      "reposts": 29,
      "likes": 97,
      "views": 663,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100980056822780313",
    "title": "Artoria 的角色动作片段",
    "titleEn": "An Artoria Character Action Clip",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "角色在幻想场景中快速移动并完成动作，创作者同时观察 H3 与其他模型的差异。 原生视频规格为 5 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character moves quickly through a fantasy scene while the creator compares H3 with another model. Native-video output: 5s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/XephyrSFW/status/2100980056822780313",
    "sourceLabel": "X 原帖 · @XephyrSFW",
    "author": "Xephyr",
    "publishedAt": "2026-09-18T16:07:34.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100980056822780313.mp4",
    "posterUrl": "/posters/x/2100980056822780313.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Great Escape : #Artoria \n\nI think i'm in love with Minimax H3. Especially when compared to wan 2.2!\n\nSFW AI Video and  Animation. No nudity, no suggestive intent.\n#FGO #FateGO #Saber",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 13,
      "views": 366,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100978471958954254",
    "title": "波浪前的服装与空间变化",
    "titleEn": "A Garment and Space Shift before a Wave",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物向前走近波浪时，服装、编码纹理和空间规模逐层扩展。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "As a woman approaches a wave, the garment, coded texture, and spatial scale build layer by layer. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HBCoop_/status/2100978471958954254",
    "sourceLabel": "X 原帖 · @HBCoop_",
    "author": "Heather Cooper",
    "publishedAt": "2026-09-18T16:01:16.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100978471958954254.mp4",
    "posterUrl": "/posters/x/2100978471958954254.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "performance",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I gave MiniMax H3 one woman, one surreal environment, and a simple action: walk toward the wave.\n\nWhat makes it interesting is that the scene keeps building.\n\nThe garment lifts, the code rises, and the whole thing scales up into something much more cinematic.\n\n@runwayml",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 9,
      "views": 600,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100971871588204928",
    "title": "雨中的会动绘本",
    "titleEn": "A Moving Picture Book in the Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绘本角色在雨伞、文字牌和小动物之间移动，画面保留手工绘本的温暖质感。 原生视频规格为 15 秒 · 1620×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Picture-book characters move among umbrellas, title cards, and small animals with a warm handmade texture. Native-video output: 15s · 1620×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tsukuyomi2021/status/2100971871588204928",
    "sourceLabel": "X 原帖 · @tsukuyomi2021",
    "author": "月詠〜ツクヨミ〜",
    "publishedAt": "2026-09-18T15:35:03.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100971871588204928.mp4",
    "posterUrl": "/posters/x/2100971871588204928.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1620×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration",
      "stylized"
    ],
    "scenes": [
      "nature",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「動く絵本 〜Friends in the Rain〜」\n\n( ˙꒳​˙ᐢ) ﾊｲｯ!\n\n参加しますん✨\n\nモデルは MiniMax H3です。なぜなら、Topviewで一番お得に作れるから( ᐛ👐)✨\n\n#雨の日AI動画チャレンジ @TopviewAIJP",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 30,
      "views": 356,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100971852432638357",
    "title": "旅行风格的角色广告",
    "titleEn": "A Travel-Style Character Advertisement",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏旅行风格视频展示 AI 角色在街景中的轻快移动，适合观察 UGC 广告结构。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical travel-style video follows an AI character moving lightly through a street, useful for studying UGC ad structure. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KscYusuf/status/2100971852432638357",
    "sourceLabel": "X 原帖 · @KscYusuf",
    "author": "Yusuf K",
    "publishedAt": "2026-09-18T15:34:58.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100971852432638357.mp4",
    "posterUrl": "/posters/x/2100971852432638357.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "commercial"
    ],
    "scenes": [
      "city",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI UGC content is getting insanely fast.\n\nI generated this travel-style video in just a few seconds with MiniMax H3 Max on @blyth_ai.\n\ncreate your AI character, pick a style, write a prompt, generate the video and you already have content ready for TikTok / Reels.\n\nfast, affordable, and honestly kind of crazy how little time it takes now ⚡️",
    "engagement": {
      "replies": 1,
      "reposts": 6,
      "likes": 6,
      "views": 420,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100971252005613675",
    "title": "街角跌倒的动作练习",
    "titleEn": "A Street-Corner Falling Action",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物在街角被石头绊倒，创作者记录了动作指令与实际物理表现的差异。 原生视频规格为 8 秒 · 736×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A person trips on a stone at a street corner, documenting the gap between an action instruction and its physical result. Native-video output: 8s · 736×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MetantanPower/status/2100971252005613675",
    "sourceLabel": "X 原帖 · @MetantanPower",
    "author": "みーたんたん＠taitanpower",
    "publishedAt": "2026-09-18T15:32:35.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100971252005613675.mp4",
    "posterUrl": "/posters/x/2100971252005613675.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "736×736",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax H3 試行錯誤勉強中。プロンプト難しいけど、wan2.2からはっきりとした進化を感じる。\n\nこれ、最後に石に躓いて前にコケる指示なんだけど、どうしてこうなった。",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 1,
      "views": 72,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100970326960026025",
    "title": "角色细节快速蒙太奇",
    "titleEn": "A Fast Character-Detail Montage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "角色参考图被剪成快速细节蒙太奇，镜头从服装与姿态切换到完整人物。 原生视频规格为 14 秒 · 592×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character reference becomes a fast detail montage, moving from clothing and pose to a full reveal. Native-video output: 14s · 592×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ROGArtwork_/status/2100970326960026025",
    "sourceLabel": "X 原帖 · @ROGArtwork_",
    "author": "ROG ARTORK",
    "publishedAt": "2026-09-18T15:28:54.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100970326960026025.mp4",
    "posterUrl": "/posters/x/2100970326960026025.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "592×1024",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "performance",
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "QT MiniMax H3\nPor @aimikoda \n\nRÍMEL JAZZY Story | Charlie Silver\n\nArte Original ROG ARTORK\n#AIart #AIイラスト #OC",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 34,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100968419864744424",
    "title": "低多边形赛跑动画",
    "titleEn": "A Low-Poly Racing Animation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "低多边形人物在赛场上奔跑，短片对比不同采样步数下的运动质感。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A low-poly figure runs on a track while the clip compares motion character at different sampling steps. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ITOKtw/status/2100968419864744424",
    "sourceLabel": "X 原帖 · @ITOKtw",
    "author": "ITOK",
    "publishedAt": "2026-09-18T15:21:20.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100968419864744424.mp4",
    "posterUrl": "/posters/x/2100968419864744424.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MiniMaxH3 FastH3V2でローポリゴン（ A low-polygon 3D graphics animation）。8ステップの方がキレイですが，3ステップの方が味があります。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 406,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100968211336265958",
    "title": "角色表驱动的二维动画",
    "titleEn": "A Character-Sheet-Driven 2D Animation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "角色表被转成带声音的二维动画片段，重点是参考图、转场和快速出片流程。 原生视频规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character sheet becomes a 2D animation clip with audio, focusing on reference images, transitions, and fast output. Native-video output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/khangnguyen_ai/status/2100968211336265958",
    "sourceLabel": "X 原帖 · @khangnguyen_ai",
    "author": "Khang Nguyen I AI",
    "publishedAt": "2026-09-18T15:20:30.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100968211336265958.mp4",
    "posterUrl": "/posters/x/2100968211336265958.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "stylized"
    ],
    "scenes": [
      "performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on PixVerse @PixVerse  is making 2D animation unexpectedly fast.\n\nDrop a character sheet into Reference, pick Video or Transition, and you get a 5–15s 2K clip with audio included. No more frame-by-frame \n\nspritesheets.Sharing the exact setup I use below 👇\n\n#pixversecpp #PixVerseCPP",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 53,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100966082664448014",
    "title": "黑白舞台上的歌手剪影",
    "titleEn": "A Singer Silhouette on a Monochrome Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白舞台里的人物边唱边走，画面把声音和简单的角色动作结合起来。 原生视频规格为 3 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure sings and walks across a monochrome stage, combining voice with simple character motion. Native-video output: 3s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Taiyaki_Proj08/status/2100966082664448014",
    "sourceLabel": "X 原帖 · @Taiyaki_Proj08",
    "author": "Taiyaki",
    "publishedAt": "2026-09-18T15:12:02.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100966082664448014.mp4",
    "posterUrl": "/posters/x/2100966082664448014.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized",
      "cinematic-realistic"
    ],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3によるアニメーション。wan2.2の時みたいな輪郭のブレが皆無。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 46,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100965247188168864",
    "title": "虹膜里的微型宇宙",
    "titleEn": "A Miniature Universe inside an Iris",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头贴近眼睛和虹膜纹理，光点与微型世界在极近距离中快速展开。 原生视频规格为 15 秒 · 2944×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera moves close to an eye as light and a miniature world unfold inside the iris. Native-video output: 15s · 2944×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/astronomerozge1/status/2100965247188168864",
    "sourceLabel": "X 原帖 · @astronomerozge1",
    "author": "Özge Döner",
    "publishedAt": "2026-09-18T15:08:43.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100965247188168864.mp4",
    "posterUrl": "/posters/x/2100965247188168864.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2944×1248",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "stylized",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "WIN FROM THE $50,000 PRIZE POOL! 🎸✨\n\nWhat if one guitar chord could awaken an entire ornamental universe?\n\nI created this cinematic anime rock sequence with MiniMax H3, exploring motion graphics, precise video editing, art typography and surreal visual storytelling.\n\nJoin the MiniMax H3 x Picsart Challenge!\n\nThe participation link is in the comments.\nSubmission closes September 23.\n\n@Hailuo_AI @Picsart\n\n#MiniMaxH3 #PicsartChallenge",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 9,
      "views": 347,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2100961367897383334",
    "title": "同一提示词的双模型对照",
    "titleEn": "Two Models with the Same Prompt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下两段视频使用同一提示词和设置，对比 H3 与另一模型在角色和动作上的差异。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two clips use the same prompt and settings to compare H3 with another model on character and motion. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Atlas_fomo/status/2100961367897383334",
    "sourceLabel": "X 原帖 · @Atlas_fomo",
    "author": "Atlas",
    "publishedAt": "2026-09-18T14:53:18.000Z",
    "addedAt": "2026-09-18T19:25:59.623Z",
    "mediaUrl": "/media/x-2100961367897383334.mp4",
    "posterUrl": "/posters/x/2100961367897383334.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "SAME PROMPT. SAME SETTINGS. TWO MODELS. WATCH THEM DISAGREE.\n\nBoth clips below ran in parallel on Picsart. Seedance 2.5 on top, MiniMax H3 underneath. Nobody touched the prompt between them.\n\nHere is why that matters for anyone shipping short video:\n\n1. You stop guessing which model suits the shot and watch two takes of the same brief side by side.\n2. The AI Playground runs them in parallel, so the comparison costs one prompt instead of two sessions.\n3. Picsart's agents sit on top of that: 21 specialists, including a film director, a video director, an ad strategist and a chief of staff.\n4. GPT-Live-1 runs the briefing by voice. You talk, interrupt mid-task and redirect without starting over.\n5. First variations come back in under five minutes, on web and on iOS\n\nThe blank page was never the bottleneck. Choosing between two finished takes is a faster job than inventing one\n\n@Picsart",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 22,
      "views": 1645,
      "snapshotAt": "2026-09-18T19:25:59.623Z"
    },
    "approvedAt": "2026-09-18T19:25:59.623Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101203361886572557",
    "title": "银色泳装走过商场窗外",
    "titleEn": "A Silver Swimsuit Walks Past the Mall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物穿着银色泳装走过户外商场，边走边向镜头传递语音信息，形成一段竖屏生活方式短片。 原生视频规格为 8 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure in a silver swimsuit walks past an outdoor shopping mall while delivering a spoken message, forming a vertical lifestyle short. Native-video output: 8s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chikazoemakoto/status/2101203361886572557",
    "sourceLabel": "X 原帖 · @chikazoemakoto",
    "author": "近添真琴（ライター・AIクリエイター）",
    "publishedAt": "2026-09-19T06:54:54.000Z",
    "addedAt": "2026-09-19T07:10:58.342Z",
    "mediaUrl": "/media/x-2101203361886572557.mp4",
    "posterUrl": "/posters/x/2101203361886572557.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "ugc",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#シルバーウィーク なので、シルバーの競泳水着を着てみました！✨\n\n窓の外のショッピングモールを歩いている人たちにも注目されるかな…？\n\n🔊音声オンでメッセージも聴いてもらえるとうれしいです！\n\n※映像・音声はAI生成です。\n\n【提供：Hailuo AI（CPPプログラム参加中）】#MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 45,
      "snapshotAt": "2026-09-19T07:10:58.342Z"
    },
    "approvedAt": "2026-09-19T07:10:58.342Z",
    "editorialBasis": "Original X caption explicitly identifies Hailuo AI and #MiniMaxH3; native video was fully decoded and reviewed."
  },
  {
    "id": "x-2101460438987001994",
    "title": "录音室里的现场演唱",
    "titleEn": "A Live Vocal Take in the Recording Booth",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一位歌手在白色录音室里对着麦克风演唱，镜头保持近距离并呈现稳定的口型同步。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A singer performs into a studio microphone in a clean white booth with a steady close-up and visible lip sync. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rutineru_ch/status/2101460438987001994",
    "sourceLabel": "X 原帖 · @rutineru_ch",
    "author": "ルティネル/Rutineru",
    "publishedAt": "2026-09-19T23:56:26.000Z",
    "addedAt": "2026-09-20T00:17:59.850Z",
    "mediaUrl": "/media/x-2101460438987001994.mp4",
    "posterUrl": "/posters/x/2101460438987001994.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "photoreal",
      "commercial"
    ],
    "scenes": [
      "performance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MidjourneyとMiniMax H3でThe First Take風なサムネと動画を生成してみた。ちなみに楽曲はSunoAI。 うーん、MiniMax H3だとわざわざ歌詞を入れなくてもリップシンクは完璧💫でも今はフルMV作る元気が無い😭\n#Midjourney #SunoAI #MiniMax",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 22,
      "snapshotAt": "2026-09-20T00:17:59.850Z"
    },
    "approvedAt": "2026-09-20T00:17:59.850Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101459574507384886",
    "title": "从角色杂志画面到竖屏短片",
    "titleEn": "From Character Magazine Frames to a Vertical Short",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "动漫角色与杂志式画面在竖屏 R2V 片段中连续变形和运动，保留了角色视觉主题。 原生视频规格为 15 秒 · 608×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Anime character and magazine-like frames animate through a portrait R2V clip while retaining the same visual theme. Native-video output: 15s · 608×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/xiaomimasato/status/2101459574507384886",
    "sourceLabel": "X 原帖 · @xiaomimasato",
    "author": "のえみん",
    "publishedAt": "2026-09-19T23:53:00.000Z",
    "addedAt": "2026-09-20T00:17:59.850Z",
    "mediaUrl": "/media/x-2101459574507384886.mp4",
    "posterUrl": "/posters/x/2101459574507384886.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "608×832",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "現在のワークフローで\nRTX3060 12GB RAM 32GBにて\n\nMiniMax-H3での\nR2V、15秒動画の生成時間がおおよそ\n2400秒（40分）ほどで\n採用率は8割ほど\n\nつまり、\n5時間で15秒動画を6本作れる計算\n\n作りすぎても確認が増えるし\nお題にも予め対応できない\n\nなので\n現状で割と十分な気がする",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 28,
      "snapshotAt": "2026-09-20T00:17:59.850Z"
    },
    "approvedAt": "2026-09-20T00:17:59.850Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101446612346458326",
    "title": "复古玩具广告里的动画与实拍玩具",
    "titleEn": "A Retro Toy Commercial with Animation and Live Action",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "角色参考图被扩展成 1980 年代电视玩具广告，动画节目片段与实拍玩具镜头交替出现。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A character reference becomes a 1980s television toy commercial that alternates animated-program footage and live-action toy play. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Prompt: Show this [ 📷Image1 ] in the style of a 1980s TV action toy commercial, featuring clips of the animated program it's based on but also the live action action figure toy gameplay.",
    "promptSourceUrl": "https://x.com/kaigani/status/2101446614808477837",
    "sourceUrl": "https://x.com/kaigani/status/2101446612346458326",
    "sourceLabel": "X 原帖 · @kaigani",
    "author": "𝚔𝚊𝚒 🤖+🧠",
    "publishedAt": "2026-09-19T23:01:30.000Z",
    "addedAt": "2026-09-20T00:17:59.850Z",
    "mediaUrl": "/media/x-2101446612346458326.mp4",
    "posterUrl": "/posters/x/2101446612346458326.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "retro",
      "commercial"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Had a character reference sheet, popped it into @Hailuo_AI MiniMax H3 and got a 1980s style toy commercial, with a live action toy version.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 126,
      "snapshotAt": "2026-09-20T00:17:59.850Z"
    },
    "approvedAt": "2026-09-20T00:17:59.850Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101258477813584121",
    "title": "雨夜楼梯后的偷拍镜头",
    "titleEn": "A Secret Camera behind Her in a Rainy Neon Alley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "雨夜霓虹楼梯间里，女子捡起正在拍摄自己的手机，并在空无一人的楼梯上回头确认。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "In a rainy neon stair alley, a woman picks up a phone filming her and turns toward an empty staircase to verify what she saw. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9 cinematic realistic suspense short set entirely inside the same rainy neon stair alley from <Picture 1>. Preserve the exact same female lead throughout: same face, hairstyle, soaked wardrobe, body proportions, skin texture, and the same staircase geography with wet steps, narrow alley, railings, humid air, and neon mood. No scene change, no extra people, no black ending. The story is: a drenched woman slowly walks down the stairs, notices an iPhone still recording on one step, bends to pick it up, realizes the screen shows her own back filmed from behind, instinctively steps back half a step into the railing and whips her shoulder and head around, but the real staircase is empty. End on the suspended moment where she is half-turned, half-looking down, with the screen still lit. She stays in motion throughout: descending, stopping, noticing the phone, bending to pick it up, standing with the screen, stepping back, leaning to the railing, turning shoulder and head, then looking back to the phone. The phone must stay physically consistent in size and readability. [Shot 2] At 00:00-00:02.8, a low dynamic near shot hugs her legs, waist, and the wet railing as the camera slowly pushes in and follows her down about one and a half steps. The emphasis is on lower-body weight transfer, wet skin reflections, and action-based sensual tension. Her right hand keeps a subtle stabilizing relationship with the railing. She is soaked, moving carefully but not weakly. Her first emotional phase is calm, restrained, and oppressive self-protection: eyelids half-lowered, gaze angled beyond camera toward the lower stairs, brows held flat with only the outer ends slightly lowered, lips naturally parted, breath low and steady, nostrils moving lightly, no smile. Pink-violet neon washes the front-upper planes of the chest, collarbone, abdomen, and front thighs, while cyan-blue edge light wraps the waistline, outer arms, and leg contour. This is the first sensual close-up: legs, waist, wet skin, deep wet fabric contours, and real descending balance. [Shot 3] At 00:02.8-00:05.0, cut to a medium-close slightly high angle that lands on the step at her feet while keeping one stopped boot and the railing in the foreground. A lit iPhone lies near the edge of shallow step-water. The red recording dot must be immediately readable. She stops on the stair, not frozen but caught mid-motion. Her gaze narrows sharply toward the phone, pupils snap into focus, the brow center contracts into only the thinnest line, she takes one short nasal breath, her lips close lightly from their previously open state, and the jaw gathers in. This is the first layer of wrongness, not yet fear. She murmurs quietly, <d>[English] Whose phone is this...?</d> The line is tentative and low, with the ending gently falling away. Phone screen light adds a third small light source, lifting the forehead edge, nose bridge, lash line, and cupid’s bow with cool white. [Shot 4] At 00:05.0-00:07.8, a tight near shot from her right-front follows her bend as she picks up the phone. Stay close to the hand, collarbone, side waist, wet hair falling forward, and the exact moment the device leaves the stair. She remains connected to the railing with the right side of her body and hand relationship, even while reaching. Her eyes stay locked on the phone the whole time. Lashes cast a slight shadow downward. The lips tighten more, the throat moves with one very small swallow, breathing becomes noticeably shallower, and the face shifts into focused verification rather than panic. This is the second sensual close-up: hand, collarbone, side waist, wet hair, and the phone entering the center of the narrative. Rain hits the glass and forms tiny round splashes and sliding beads. When she pinches the phone edge and lifts it, droplets drag aside and one or two thin water threads stretch briefly and break. [Shot 5] At 00:07.8-00:10.5, a near shot layers the phone screen and half her face. First make the screen content legible, then carry focus into her face without losing the screen read. The phone display shows only one clear thing: her own back descending the staircase, filmed from a steady hidden angle one step higher behind her. No second face, no ghost, no supernatural clutter. She sees it and recognizes herself. Her fourth emotional phase is clear alert recognition: her eyes flick once across the screen and then hold, the lower lids lift slightly, the pupils widen just a little, the brow center tightens more clearly while still restrained, nostrils open and settle, any remaining softness in the mouth disappears, and the lower lip stalls for a beat as if a sentence is trapped in the throat. She whispers, <d>[English] Wait... that’s me.</d> The first half carries a tiny delay as recognition catches in her breath. The phone’s cold light now fills the lower half of her face and under-eye area, making the micro-expression readable without overpowering the neon mood. [Shot 6] At 00:10.5-00:12.8, an extreme close-up pushes from shoulder-neck, side face, lashes, and lip line into the eye. This is the third sensual close-up, centered on the dangerous moment before she turns. She keeps the phone close to her chest line, glances once more at the screen, and then prepares to look back. Her fifth emotional phase is being pressed by the implication that the filming angle was behind her: lashes tremble once, breathing tightens, the lip line draws in, cheek muscles firm slightly, and she says in a lower, faster, more pressured whisper, <d>[English] Who the hell is filming me?</d> It is not shouted, but forced low confirmation. One side of her face drops deeper into shadow while the other is caught by soft pink-violet neon on lashes, nose tip, lip peak, and jaw edge. [Shot 7] At 00:12.8-00:15.0, a medium-close shot with a slight retreat. Let the turning motion carry the cut. She shifts back by half a step, the wet sole dragging a very small water trace, shoulder and collarbone leading the head turn rather than the head teleporting first. Her back and right side find the railing. She turns sharply toward the upper stairs, but the real staircase is empty. She then drops her gaze back to the still-lit phone at chest level. Her final emotional phase is cold controlled aftershock: the eyes cut first upward, then back down, the forehead tightens lightly, the jaw stays tight, the mouth opens by half a breath and then slowly closes again, fear controlled but not gone. The screen remains lit and still recording. The staircase behind her stays empty, wet, neon-lit, and physically real. End with her half-turned, half-looking down at the glowing iPhone, rain still falling, neon still flickering, and suspense held inside the frame, without cutting to black.\n\noverall_soundscape: Maintain a continuous rainy neon stairwell ambience: mid-strength rain, water ticking on railing and steps, shallow drainage flow, distant electric hum, soft neon buzz, and humid night reverb. Emphasize wet footfalls on the stairs, the subtle grip and slide of her hand on the railing, clothing friction, low controlled breath, and fine splashes from shallow step-water. Then bring forward rain striking the phone glass, tiny circular water impacts, the slight stop in her step, her low tentative line <d>[English] Whose phone is this...?</d>, fingers contacting the wet phone, water dragging across the screen, a faint lift-off suction from the wet stair, her shallow breath, fabric tension as she bends, one small swallow, and the surrounding rain and neon hum dimming slightly as she reads the screen. Finish with the tightening breath before the turn, the low pressured line <d>[English] Who the hell is filming me?</d>, one light metallic touch from the railing, the small backward step drag on wet stair, empty stair ambience, continuing rain, neon buzz, and the phone still recording in her hand. No scream, no extra voices, no subtitles.\n\nnon_diegetic_music: Keep score absent or nearly absent. If any non-diegetic layer is used, make it only a very thin low suspense drone under the rain and neon hum, rising slightly after she sees the screen and never overpowering the physical environment or dialogue.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2101259032745201952",
    "sourceUrl": "https://x.com/ou_zhen599/status/2101258477813584121",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-19T10:33:55.000Z",
    "addedAt": "2026-09-20T00:17:59.850Z",
    "mediaUrl": "/media/x-2101258477813584121.mp4",
    "posterUrl": "/posters/x/2101258477813584121.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI            \n\n🎬Image-to-Video                         \n📷“The Camera Behind Her”\nShe found a cell phone that had been secretly filming her, and the camera was right behind her.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 5,
      "reposts": 4,
      "likes": 204,
      "views": 11268,
      "snapshotAt": "2026-09-20T00:17:59.850Z"
    },
    "approvedAt": "2026-09-20T00:17:59.850Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101254179067236692",
    "title": "霓虹街区被节奏唤醒",
    "titleEn": "When the Neon District Wakes to the Beat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "湿润的霓虹后巷中，蓝发角色接住红桃皇后牌，面对从雾中走出的镜像人物。 原生视频规格为 16 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "In a wet neon back alley, a blue-haired character catches a Queen of Hearts card and faces her mirror self emerging from the fog. Native-video output: 16s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9 cinematic cyber-thriller image-to-video sequence set entirely inside the same wet neon back alley from <Picture 1>: the left-front railing zone where A begins, the street centerline, the puddle reflection layer, and the opposite fog-filled exit in the background. Do not add extra people, do not place both women on the same staircase segment, and do not leave this alley. Subject A is the blue-haired woman from <Picture 1>, with the same face, body, black deep-V leather bodysuit, fishnets, over-knee boots, metallic choker, glowing beverage can, goggles, and jacket ring-light. Subject B must be the exact same face, hairstyle, body, and clothing silhouette as A, but with a refined dangerous harlequin-style makeup variation, not a different woman. The story is: A rules her own staircase entrance like a queen, fog grows at the opposite exit, a Queen of Hearts card flies out, A catches it one-handed and recognizes it as a provocation, then rises and looks up to see her identical mirror-woman emerging from the fog and forcing her to leave her throne. Lighting comes only from practical neon, staircase reflections, the fog backlight, and distant storefront light. Rain, wet hair, cloth folds, skin reflections, boot splashes, condensation droplets, and natural body inertia remain physically real. [Shot 2] At 00:00-00:03.5, begin with an extremely low ground-level push from puddles, discarded paper, and lane markings into A’s boots, fishnets, thighs, leather chest, and jaw, ending in a near frame from chest to lips. A sits on the center of the third step from the bottom of the staircase, leaning slightly forward as if guarding the entrance to her territory. This is the first sensual close-up, built from collarbone, throat line, deep-V leather highlights, and the nearly invisible rise and fall of her breathing. Her first emotional state is cool sovereign control: brows relaxed but not soft, eyelids half-lowered, gaze not fully locked on the distance yet, lips lightly closed, breathing almost silent, throat moving once faintly. The puddles reflect blue and magenta neon steadily, and the opposite street exit holds only a thin fog presence. [Shot 3] At 00:03.5-00:07.0, the camera eases slightly backward and rises just enough to include both the staircase foreground and the opposite exit. The fog there becomes denser, first rolling close to the ground and then pressing outward slowly, with neon cutting bright lines along its edges. A is still seated, but the second emotional stage begins. She does not widen her eyes dramatically. Instead her breathing stops for half a beat first, her lips loosen by a fraction, the lower lids lift slightly, and her gaze truly focuses toward the opposite exit for the first time. Her head tips subtly toward the source of the sound and fog. Then a Queen of Hearts card is carried out of the fog by damp wind, turning half a rotation and gliding naturally through the mid-ground mist toward A. Do not chase the card wildly. A’s right hand rises fast and clean and catches it one-handed at chest height slightly to her right, with real deceleration in the wrist and fingers. Her third emotional stage is confirmation of provocation: before the catch, her eyes narrow sharply and the brow center hardens slightly; after the catch, she drops her gaze to the card, the throat moves once, the jawline begins to tighten, and the original calm disappears. This is the second sensual close-up, built from the catching wrist, collarbone, leather highlights across the chest, eyelash shadows, and the lip line and jaw while glancing down at the card. The Queen of Hearts face must be clearly readable; the paper is damp-soft with slightly curled edges and realistic motion. [Shot 4] At 00:07.0-00:10.2, from A’s left-front, the camera makes a slight lateral push and subtle rise, keeping the staircase, street, and opposite exit on one axis. A places her left hand on the railing and rises slowly. Her right foot steps down first, then the left follows into a high-low standing stance. She never leaves the immediate front zone of her staircase. The card remains in her right hand. Her fourth emotional stage is calm transition into combat readiness: eyes narrower and more concentrated, nostrils opening slightly with breath, lips tightening, jaw sharpening, and her gaze lifting off the card and back to the opposite exit as if thinking, “You really dared to do this.” During the rise, the leather outfit stretches and rebounds naturally, hair shifts back and settles, and her chest shows only a very slight real inertia response. This is the third sensual close-up, built from waist and hip effort, fishnet leg tension, neck and jaw line, and the chest moving with controlled breath. [Shot 5] At 00:10.2-00:12.7, hold a longer relation shot. A stays on the left side or left-lower edge of the foreground while the opposite exit sits centered in the back. Focus transfers gradually from A to B. B walks out of the fog by two small steps and stops at the edge of the backlit mist. She is the same face, same hairstyle, same body, same clothing outline as A, but with refined dangerous harlequin-style makeup: darker eye area, sharper mouth corners, and a more poisonous finish. A’s fifth emotional stage is recognition freeze. She locks for one beat: pupils contract quickly, the lip line releases for half a second, the throat catches on one breath, and for the first time her breathing goes off rhythm. Then she does not retreat; she becomes more focused. B’s emotional state is intimate contempt: stable eyes, one mouth corner rising very slowly, as if appreciating that A has finally understood. The fog keeps rolling behind B but never hides her face, and her body parts it with real wraparound flow. [Shot 6] At 00:12.7-00:15.0, use one restrained reverse shot to B in a medium close frame, pushing in extremely slowly. B stands at the street-exit edge, chin lifted slightly, one-sided smile in place, gaze locked on A at a distance. Before the line, give one tiny false-soft look; on the key words, the gaze turns straighter and the mouth shape grows more precise and cutting. The only spoken line is <d>[English] You dropped your crown.</d> It must be low, intimate, cold, and humiliating rather than loud. “You” is pressed low, “dropped” carries a little more weight, and “your crown” falls away like a verdict. After the line, A in the far frame or foreground edge takes one more step down, committing to leave her throne and meet the challenge. Her final emotional state shifts from frozen recognition into hard resolve: eyes harden again, lips tighten again, jaw stabilizes, shoulders do not collapse. When the line lands, a nearby neon briefly flickers once, and both the fog edge behind B and the puddle reflections brighten for an instant. End on the moment just before A moves toward the opposite side. Do not cut to black.\n\noverall_soundscape: Maintain a wet cyber alley ambience throughout: neon electrical hum, distant ventilation fans, moist air resonance, occasional water drips, and low street reverb. Emphasize puddle surface noise, faint leather movement, cloth friction, the subtle sound of A shifting on the step, and her almost inaudible breath. Then bring forward the slow spill of fog, the slight change in airflow at the opposite exit, the card fluttering and rotating, the light slap of paper into A’s fingers, a fine metallic tick from the railing, leather stretch and settle as she rises, fog movement as B parts it, soft heel contacts from B’s two steps, and B’s close cold line <d>[English] You dropped your crown.</d> Afterward restore the alley ambience with neon hiss, damp air, tiny puddle ripples, and fog drift. No extra voices, no subtitle cues, no music-video stingers.\n\nnon_diegetic_music: Keep score minimal or absent. If any non-diegetic layer is used, make it a very thin low-frequency suspense bed beneath the alley ambience, rising only slightly once B is revealed and never overpowering the voice, neon, or the physical space.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2101254891293614422",
    "sourceUrl": "https://x.com/ou_zhen599/status/2101254179067236692",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-19T10:16:50.000Z",
    "addedAt": "2026-09-20T00:17:59.850Z",
    "mediaUrl": "/media/x-2101254179067236692.mp4",
    "posterUrl": "/posters/x/2101254179067236692.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "stylized"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI          \n\n🎬Image-to-Video                        \n📷“Wake Up, Neon”\nThe lights and reflections along the entire street, along with the rhythm of her body, are awakened all at once. She transforms from a passerby in the night into the heartbeat and switch of the entire cyber district.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 8,
      "views": 574,
      "snapshotAt": "2026-09-20T00:17:59.850Z"
    },
    "approvedAt": "2026-09-20T00:17:59.850Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101243238334205994",
    "title": "古庙里落下的王冠",
    "titleEn": "The Fallen Crown in an Ancient Temple",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "古老神殿的石柱与壁画之间，女子将金箔嵌入墙缝，最终与壁画女神形成诡异呼应。 原生视频规格为 14 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Inside an ancient temple, a woman presses gold leaf into a wall crack before her pose echoes the painted goddess. Native-video output: 14s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MiniMax-H3 prompt👇\nFor the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] A 15-second, 16:9 cinematic realistic image-to-video sequence set entirely inside the same ancient temple porch and stone-column corridor shown in <Picture 1>. Preserve the exact same adult woman throughout: same face, hairstyle, makeup, garment cut, leather texture, body ratio, gait, and presence. The whole sequence stays in one continuous space only: the porch threshold, the narrow corridor between the right-side pillar and wall, and the same right-side wall mural zone. No exterior, no new hall, no extra people, no extra props, no subtitles, no watermark, and no black ending. The core story is that she is not entering the temple as a new visitor, but behaving like a brief awakened action-memory belonging to the wall mural itself. She is always moving or has just finished moving, never posing. [Shot 2] At 00:00-00:02.2, a low-angle near shot begins at boot heel, forward-driving legs, and knee line, then tilts up to collarbone, jaw, and part of the face. The camera stays close to the floor and glides backward as she crosses the threshold. When the right foot lands, thin floor dust is pressed outward by the boot sole and curls softly around the edge. Her gaze stays beyond the camera rather than into it, eyelids slightly lowered, pupils steady, lips naturally closed, and collarbone and upper chest rise and fall with restrained real breathing. Warm air from above and behind lifts the long hair gently, with slight delayed motion in the strands. Clothing follows the body line and rebounds subtly. This first sensual close-up is built entirely from gait, center-of-gravity transfer, collarbone breathing, jawline, and hair edge-light. [Shot 3] At 00:02.2-00:05.0, cut to a frontal medium close retreating shot, framing from chest to knee while she continues advancing on the central axis. During the advance she drifts half a step toward frame right and her left hand naturally grazes the relief carving on the right-side pillar without stopping to search. As her fingertips move across the stone, a small amount of fine stone powder slides down. One extremely thin, brittle ancient gold-leaf fragment peels away from the carving and sticks lightly between her index and middle fingers. Her eyes do not make a large turn at once. First she checks it with peripheral vision, the brow peak lifts by an almost invisible degree, the lip line softens slightly, and the breath becomes shallower. Lighting remains warm side-back gold on her edges and cool shadow on the pillar, with soft highlights on the leather surfaces. [Shot 4] At 00:05.0-00:07.4, cut to a near close-up from side waist to hand, the camera slightly low and behind her right side, following in a very small arc. She does not stop walking; instead the torso rotates subtly within the stride, hips and waist connected, as she lifts her left hand to inspect the gold leaf. Emphasize the side waist line, the slight pull and rebound of fabric caused by movement, and the tactile precision of her fingers rolling the fragment. Her gaze gathers inward from distance to fingertips, lashes pressing down, breath becoming very shallow at the nose tip, lips slightly parted without sound, and jawline gently tightening. Hair swings at the shoulder with small inertial delay. The gold leaf trembles in the edge light with weak realistic reflection rather than glitter. This is the second sensual close-up, built from waist, hand, cloth tension, breath, touch, and latent danger. [Shot 5] At 00:07.4-00:10.5, cut to a three-quarter side-front medium close shot. The camera moves in a tiny arc to her left-front and then stabilizes, keeping both her and the right-side wall relationship clear. She completes the last half step, moves close to the right wall, and presses the ultra-thin gold leaf into a nearly invisible crack in the wall. After pressing it in, she does not withdraw the hand immediately; the fingertips stay there for half a second. The crack begins to spread slowly along the buried line of the ancient wall image. Only a few tiny stone grains fall, and dust hangs briefly before settling. Her eyeballs hold, then perform one short refocus. Her brow center does not knot dramatically; both brow peaks draw slightly inward. The middle of the upper lip lifts faintly, the mouth opens by a low degree, and the breath catches once in the throat. She whispers, almost as breath itself, <d>[English] I never left.</d> The line should feel like recognition rather than exposition. Inside the crack there is only a weak gold dust-glow and subtle misalignment in the wall pattern, like memory inside stone being disturbed. No explosion, no beam, no collapse. [Shot 6] At 00:10.5-00:12.8, cut to a face close-up and push in slightly, focusing on the eyes, lips, cheekbone edge-light, and the small muscular changes around the mid-face. Her left hand remains on the wall. She does not retreat. Her eyes stay forward first, then flick for a fraction toward the interior of the wall as if hearing a response within it. The lips remain slightly parted, breath briefly suspended, lower lids slightly tight, and warm gold edge light catches the fine hair along the face. The emotion is a single struck instant of surprise held under extreme control, never a scream, never a wide-open panic. This is the third sensual close-up, centered on parted lips, paused gaze, hair rim light, and high-control micro-expression. [Shot 7] At 00:12.8-00:15.0, maintain the same axis and slowly withdraw from face close-up to medium and then medium-wide without rising, without changing angle, and without entering any new space. As the camera withdraws, it becomes gradually clear that her standing pose, raised arm angle, and head orientation precisely match the pose of an ancient painted goddess on the right wall. The mural does not suddenly appear; instead the old pigment layers, cracks, dust traces, and surviving outline in the wall are gradually revealed as something that was always there and is only now being seen. She does not perform an exaggerated reaction. She simply keeps one unfinished breath in the body. In the final beat, the painted goddess’s eyeball shifts by the smallest possible degree, the gold dust slowly settles back to the floor, and the sequence ends in an elegant uncanny suspended state without cutting to black.\n\noverall_soundscape: Maintain a hollow ancient stone ambience throughout, with low temple air, soft boot pressure on dusted floor, faint grains sliding, fingertips rubbing stone, tiny wall-friction sounds, and restrained breathing. Emphasize boot contact, dust pushed outward by the step, hair moving in warm airflow, fingertip contact with the pillar carving, a fine trickle of stone powder, the fragile peel of the gold leaf, fabric pull-and-release at the waist and torso, the dry whisper of fingers rolling the thin gold leaf, the soft press of gold leaf into the crack, a delicate stone-grain release, faint wall-friction spread as the crack extends, dust pausing in the air, and her near-whispered line <d>[English] I never left.</d> Reduce the space further near the end to breath, tiny skin and cloth movement, microscopic wall resonance, settling dust grains, and the near-inaudible response of the mural’s living eye.\n\nnon_diegetic_music: Use little to no score. If any non-diegetic layer is present, keep it to an extremely thin low sustained tonal bed, almost imperceptible, supporting the suspended ancient-memory feeling without turning the sequence into overt fantasy or horror.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2101243903555080246",
    "sourceUrl": "https://x.com/ou_zhen599/status/2101243238334205994",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-19T09:33:21.000Z",
    "addedAt": "2026-09-20T00:17:59.850Z",
    "mediaUrl": "/media/x-2101243238334205994.mp4",
    "posterUrl": "/posters/x/2101243238334205994.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "fantasy",
      "performance"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI        \n\n🎬Image-to-Video                       \n📷“The Fallen Crown”\nIt wasn’t until a `Q` (Queen) card landed in her hand that she realized this wasn’t a provocation, but a declaration of a bid for the throne. When the fog cleared, standing across from her was her most dangerous mirror image.  \n\nFull prompt in the reply.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 14,
      "views": 864,
      "snapshotAt": "2026-09-20T00:17:59.850Z"
    },
    "approvedAt": "2026-09-20T00:17:59.850Z",
    "editorialBasis": "Original X caption plus beginning, middle, and end frames reviewed from the downloaded native video; full media decode passed."
  },
  {
    "id": "x-2101824132610416742",
    "title": "科幻叙事中的连续镜头",
    "titleEn": "A Continuous Shot in a Sci-Fi Story",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原创科幻短片片段，用连续镜头推进人物和环境叙事。 原生视频规格为 225 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An original sci-fi short-film excerpt using a continuous shot to carry character and environment storytelling. Native-video output: 225s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Rhynedahll/status/2101824132610416742",
    "sourceLabel": "X 原帖 · @Rhynedahll",
    "author": "H. Jonas Rhynedahll",
    "publishedAt": "2026-09-21T00:01:37.000Z",
    "addedAt": "2026-09-21T01:13:49.799Z",
    "mediaUrl": "/media/x-2101824132610416742.mp4",
    "posterUrl": "/posters/x/2101824132610416742.jpg",
    "duration": 225,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Boomer, the Wheelchair, and Everything — Episode 2 - We're going to need a bigger boat!\nAn original AI science-fiction film.\nMiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 27,
      "snapshotAt": "2026-09-21T01:13:49.799Z"
    },
    "approvedAt": "2026-09-21T01:13:49.799Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101793529869824312",
    "title": "广告片中的角色与动作",
    "titleEn": "Character and Motion in a Commercial",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "商业片段将人物动作、产品镜头与环境声音组合成一条短片。 原生视频规格为 30 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A commercial-style clip combining character motion, product shots, and environmental sound. Native-video output: 30s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2101793529869824312",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-09-20T22:00:01.000Z",
    "addedAt": "2026-09-21T01:13:49.799Z",
    "mediaUrl": "/media/x-2101793529869824312.mp4",
    "posterUrl": "/posters/x/2101793529869824312.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "次は、30秒CM\n30秒あると、15秒よりも余裕があるので入れられなかったシーンを入れることができる\n\nAI制作なら、労力はほぼ変わらずでできますね\n\n動画は、Codex (Astra) x AZ8\n#AZ8 @AZ8official_JP \n\nCLIで繋げられるので本当に便利\n動画モデルはMiniMax H3とKling 3.0です",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 9,
      "views": 773,
      "snapshotAt": "2026-09-21T01:13:49.799Z"
    },
    "approvedAt": "2026-09-21T01:13:49.799Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101790915786604879",
    "title": "失重场景中的人物动作",
    "titleEn": "Character Motion in Zero Gravity",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物在空中行走的实验性镜头，展示动作与服装细节。 原生视频规格为 10 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An experimental zero-gravity shot showing character motion and clothing detail. Native-video output: 10s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mtArakawa/status/2101790915786604879",
    "sourceLabel": "X 原帖 · @mtArakawa",
    "author": "元荒川ちゃん❤",
    "publishedAt": "2026-09-20T21:49:38.000Z",
    "addedAt": "2026-09-21T01:13:49.799Z",
    "mediaUrl": "/media/x-2101790915786604879.mp4",
    "posterUrl": "/posters/x/2101790915786604879.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "maid curtain\n女の子の足首は冷たい\n\nNBPro + MiniMax H3\n#maid #ブルマ #suspension",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 14,
      "views": 2034,
      "snapshotAt": "2026-09-21T01:13:49.799Z"
    },
    "approvedAt": "2026-09-21T01:13:49.799Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101733004745818383",
    "title": "赛博朋克节奏与城市光影",
    "titleEn": "Cyberpunk Rhythm and City Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "赛博朋克世界中的节奏化影像片段，结合人物、城市和音乐。 原生视频规格为 248 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cyberpunk sequence combining characters, city light, and rhythm. Native-video output: 248s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/OrctonAI/status/2101733004745818383",
    "sourceLabel": "X 原帖 · @OrctonAI",
    "author": "OrctonAI",
    "publishedAt": "2026-09-20T17:59:31.000Z",
    "addedAt": "2026-09-21T01:13:49.799Z",
    "mediaUrl": "/media/x-2101733004745818383.mp4",
    "posterUrl": "/posters/x/2101733004745818383.jpg",
    "duration": 248,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cyberpunk Crissy is back!  In a 🤯 Cyberpunk World. Local @Hailuo_AI MiniMax H3 Open Music Vid.\n@topazlabs Local 4K60 fps 🎬",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 13,
      "views": 587,
      "snapshotAt": "2026-09-21T01:13:49.799Z"
    },
    "approvedAt": "2026-09-21T01:13:49.799Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101732321443676460",
    "title": "周末即兴镜头实验",
    "titleEn": "A Weekend Shot Experiment",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一次轻量的 H3 镜头实验，记录人物和场景的即时变化。 原生视频规格为 29 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A lightweight H3 shot experiment capturing immediate changes in character and scene. Native-video output: 29s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hotforteecha/status/2101732321443676460",
    "sourceLabel": "X 原帖 · @hotforteecha",
    "author": "JustJay",
    "publishedAt": "2026-09-20T17:56:48.000Z",
    "addedAt": "2026-09-21T01:13:49.801Z",
    "mediaUrl": "/media/x-2101732321443676460.mp4",
    "posterUrl": "/posters/x/2101732321443676460.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just a lazy Sunday playing around with Minimax H3 before the 49er game.  Back to this scene more later",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 52,
      "snapshotAt": "2026-09-21T01:13:49.801Z"
    },
    "approvedAt": "2026-09-21T01:13:49.801Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101729279725707677",
    "title": "多模型协作的音乐短片",
    "titleEn": "A Multi-Model Music Short",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "音乐短片将 H3 与其他生成模型协同使用，完成表演和视觉段落。 原生视频规格为 153 秒 · 912×688。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A music short combining H3 with other generative models for performance and visual sequences. Native-video output: 153s · 912×688. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sebastienrama/status/2101729279725707677",
    "sourceLabel": "X 原帖 · @sebastienrama",
    "author": "Sébastien Rama  (Dodemodexter)",
    "publishedAt": "2026-09-20T17:44:43.000Z",
    "addedAt": "2026-09-21T01:13:49.801Z",
    "mediaUrl": "/media/x-2101729279725707677.mp4",
    "posterUrl": "/posters/x/2101729279725707677.jpg",
    "duration": 153,
    "aspectRatio": "landscape",
    "resolution": "912×688",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "1985, tout le monde chantait ensemble pour les autres. 2027, ils chantent ensemble pour eux - \"We Are The France\" \nDétournement parodique IA (mais pas que…!)\nModèles : #MinimaxH3 #Klein9B #YUE2 #Parodie #Détournement #Présidentielle2027 #WeAreTheWorld",
    "engagement": {
      "replies": 4,
      "reposts": 1,
      "likes": 12,
      "views": 801,
      "snapshotAt": "2026-09-21T01:13:49.801Z"
    },
    "approvedAt": "2026-09-21T01:13:49.801Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101691337838104675",
    "title": "电影感科幻镜头",
    "titleEn": "A Cinematic Sci-Fi Shot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "电影感科幻镜头，原帖公开了完整制作提示词线索。 原生视频规格为 16 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cinematic sci-fi shot with a complete prompt reference shared in the original thread. Native-video output: 16s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ou_zhen599/status/2101691337838104675",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-20T15:13:57.000Z",
    "addedAt": "2026-09-21T01:13:49.801Z",
    "mediaUrl": "/media/x-2101691337838104675.mp4",
    "posterUrl": "/posters/x/2101691337838104675.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                      \n\n🎬Image-to-Video                              \n📷“The Backward Formula”\nWhat should have been a brilliant alchemy experiment turned into a runaway disaster at the hands of a novice mage.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 4,
      "views": 373,
      "snapshotAt": "2026-09-21T01:13:49.801Z"
    },
    "approvedAt": "2026-09-21T01:13:49.801Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101691003052675256",
    "title": "受损飞船的灾难镜头",
    "titleEn": "A Disaster Shot aboard a Damaged Spacecraft",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "受损科研飞船中的灾难场景，以写实光影和同步声音推进氛围。 原生视频规格为 15 秒 · 1536×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A disaster scene aboard a damaged research spacecraft, driven by realistic lighting and synchronized sound. Native-video output: 15s · 1536×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DeCat2025/status/2101691003052675256",
    "sourceLabel": "X 原帖 · @DeCat2025",
    "author": "DeCat",
    "publishedAt": "2026-09-20T15:12:37.000Z",
    "addedAt": "2026-09-21T01:13:49.801Z",
    "mediaUrl": "/media/x-2101691003052675256.mp4",
    "posterUrl": "/posters/x/2101691003052675256.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1536×672",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Jupiter\nMinimax H3 @Hailuo_AI \n\n15-second, 16:9 ultra-photorealistic cinematic sci-fi disaster with synchronized sound.  \n\nONE damaged human research spacecraft is falling uncontrollably into JUPITER'S ATMOSPHERE.  Jupiter has NO solid surface below. The spacecraft descends deeper into endless gigantic layers of turbulent gas and clouds.  0–5s: Extreme wide exterior shot.  \n\nThe small spacecraft plunges downward into Jupiter's enormous atmosphere.  Above it, distant sunlight fades through the upper haze. Far below, colossal bands of cream, orange and rust-colored clouds stretch across the entire view.  The ship is tiny compared with the immense planet.  It enters a violent cloud layer at extreme speed.  The spacecraft begins shaking and rolling slightly as atmospheric turbulence hits it.  \n\n5–10s: Close exterior tracking shot beside the SAME spacecraft as it continues DESCENDING. \nDense clouds race upward past the ship, clearly showing that the spacecraft is falling DOWNWARD.\nLightning illuminates gigantic storm clouds around it.\nThe hull vibrates violently. Small exterior panels tear away and disappear upward into the turbulence.\nOrange atmospheric heating begins glowing along the forward edges of the spacecraft. \nThe ship struggles to stabilize but continues falling deeper.  \n\n10–15s: Camera pulls dramatically backward into an enormous wide shot.  The spacecraft drops through a break in the clouds and reveals a terrifying lower atmospheric layer:  vast rotating cloud formations, towering storms and enormous lightning systems extending thousands of kilometers into the distance.  The spacecraft becomes a tiny glowing object descending toward this immense storm-filled abyss.  \nA colossal lightning flash illuminates the entire cloudscape.  The ship disappears",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 8,
      "views": 186,
      "snapshotAt": "2026-09-21T01:13:49.801Z"
    },
    "approvedAt": "2026-09-21T01:13:49.801Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101688885449531569",
    "title": "角色演唱片段",
    "titleEn": "A Character Performance Clip",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "角色演唱片段，将人物表演与音乐节奏结合。 原生视频规格为 83 秒 · 1080×1576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character performance clip combining acting and musical rhythm. Native-video output: 83s · 1080×1576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_dressshirt/status/2101688885449531569",
    "sourceLabel": "X 原帖 · @ai_dressshirt",
    "author": "裸ワイシャツ狂人@AIアニメ",
    "publishedAt": "2026-09-20T15:04:12.000Z",
    "addedAt": "2026-09-21T01:13:49.801Z",
    "mediaUrl": "/media/x-2101688885449531569.mp4",
    "posterUrl": "/posters/x/2101688885449531569.jpg",
    "duration": 83,
    "aspectRatio": "portrait",
    "resolution": "1080×1576",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "リクエストに応じて、モリガンです！\nヴァンパイアを歌わせてもらいました。\n色んなモーションとカメラをテストしました。\n\n…まさかFGOのモルガンだったんじゃ。\n#Minimaxh3 #comfyui #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 1,
      "views": 60,
      "snapshotAt": "2026-09-21T01:13:49.801Z"
    },
    "approvedAt": "2026-09-21T01:13:49.801Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101688498046873894",
    "title": "节日主题的角色短片",
    "titleEn": "A Festival-Themed Character Short",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "节日主题短片，围绕角色表演与日常氛围展开。 原生视频规格为 29 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A festival-themed short built around character performance and everyday atmosphere. Native-video output: 29s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ontm0422ai/status/2101688498046873894",
    "sourceLabel": "X 原帖 · @ontm0422ai",
    "author": "オンタム",
    "publishedAt": "2026-09-20T15:02:40.000Z",
    "addedAt": "2026-09-21T01:13:49.801Z",
    "mediaUrl": "/media/x-2101688498046873894.mp4",
    "posterUrl": "/posters/x/2101688498046873894.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#敬老の日 × #婆が如く\nこの度、AZ8のCreator Programに\n参加させていただきました。\nAZ8のOxygen (by MiniMaxH3)\nコストが安く動画の仕上がりも\nご覧の通り問題なしっ✨\n今後の製作にも役立ちそうです！\n#AZ8 @AZ8official_JP",
    "engagement": {
      "replies": 0,
      "reposts": 5,
      "likes": 30,
      "views": 2935,
      "snapshotAt": "2026-09-21T01:13:49.801Z"
    },
    "approvedAt": "2026-09-21T01:13:49.801Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101686815157526608",
    "title": "微短剧第一集",
    "titleEn": "A Microdrama Episode",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "微短剧片段，展示人物对白、动作和场景调度。 原生视频规格为 70 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A microdrama excerpt showing dialogue, action, and scene blocking. Native-video output: 70s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AbdullahIhsanAI/status/2101686815157526608",
    "sourceLabel": "X 原帖 · @AbdullahIhsanAI",
    "author": "Abdullah Khan",
    "publishedAt": "2026-09-20T14:55:58.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101686815157526608.mp4",
    "posterUrl": "/posters/x/2101686815157526608.jpg",
    "duration": 70,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I've been experimenting more with the microdrama genre. This is a 2000s style hardcore action series. Made entirely with AI. Hopefully the first of many. This is.. \n\nThe Driver, Episode 1.\n\nMade with a combination of Seedance 2.5, Flux 3, and Minimax H3 Max.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 83,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101653458478911722",
    "title": "参考动画控制的本地工作流",
    "titleEn": "Local Workflow with Reference Motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "首帧参考图和 Blender 动画共同控制视频运动。 原生视频规格为 10 秒 · 2208×1248。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A first-frame reference and Blender animation jointly control the motion. Native-video output: 10s · 2208×1248. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cheva513/status/2101653458478911722",
    "sourceLabel": "X 原帖 · @cheva513",
    "author": "cc",
    "publishedAt": "2026-09-20T12:43:25.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101653458478911722.mp4",
    "posterUrl": "/posters/x/2101653458478911722.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2208×1248",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "尝试用minimax H3 本地生成视频。虽然不能细看，但免费能有这个效果也很惊人了。首帧参考图用gemini生成，镜头用blender做的参考动画控制，见回复。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 52,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101652338008375755",
    "title": "女性摇滚乐队现场",
    "titleEn": "Rock Band Energy on Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女性摇滚乐队的现场表演片段，配合音频生成工作流。 原生视频规格为 21 秒 · 1152×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rock-band performance excerpt paired with an audio-generation workflow. Native-video output: 21s · 1152×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mix_buchi_/status/2101652338008375755",
    "sourceLabel": "X 原帖 · @mix_buchi_",
    "author": "buchi_",
    "publishedAt": "2026-09-20T12:38:58.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101652338008375755.mp4",
    "posterUrl": "/posters/x/2101652338008375755.jpg",
    "duration": 21,
    "aspectRatio": "landscape",
    "resolution": "1152×640",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Female Rock group\n\nFastH3もRefAudio使えば音質の問題無し(笑)\nLLMお任せプロンプト。\n\n#MinimaxH3 FastH3 Audio2V",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 774,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101650341590937638",
    "title": "从分镜到演唱镜头",
    "titleEn": "From Storyboard to Singing Shot",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "先规划分镜和角色参考，再生成带演唱的短镜头。 原生视频规格为 16 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short singing sequence built from storyboard and character references. Native-video output: 16s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/syanal364639/status/2101650341590937638",
    "sourceLabel": "X 原帖 · @syanal364639",
    "author": "しゃーなる",
    "publishedAt": "2026-09-20T12:31:02.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101650341590937638.mp4",
    "posterUrl": "/posters/x/2101650341590937638.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "立ち絵を基にChatGPTでコンテシートとキャラリファレンスシートを作ってMiniMaxH3に投入。自動生成されたボーカル曲を参照してSUNOでボーカルを再生成して後乗せ。#ベルノライト",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 520,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101648913526915441",
    "title": "餐桌对白短片",
    "titleEn": "A Family Dinner Dialogue",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "餐桌上的父女对白场景，保留原帖公开的可见文案。 原生视频规格为 14 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A father-daughter dinner dialogue scene based on the post’s visible description. Native-video output: 14s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tangpanqing/status/2101648913526915441",
    "sourceLabel": "X 原帖 · @tangpanqing",
    "author": "汤光头",
    "publishedAt": "2026-09-20T12:25:22.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101648913526915441.mp4",
    "posterUrl": "/posters/x/2101648913526915441.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "牛逼，3分钟生成一个爆火的家庭教育短视频\n\n模型MiniMax H3 工具http://PixStag.com\n\n提示语超级简单：\n\n2026年，上海。\n\n父女俩正吃饭，父亲上来就给女儿的脸上打了一巴掌，然后生气的指着她说：“ 不合适为什么收人家礼物，花人家钱？\n\n不合适还吊着人家胃口？\n\n找对象就好好找，别糟蹋人家。\n\n礼物收了，奶茶喝了，红包要了，最后来句不合适把人家甩了。\n\n法治社会救了你，不然你早就让人家埋了。”",
    "engagement": {
      "replies": 4,
      "reposts": 6,
      "likes": 37,
      "views": 6090,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101616134759793122",
    "title": "高分辨率本地工作流尝试",
    "titleEn": "A High-Resolution Local Workflow Attempt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "本地运行的高分辨率视频测试，记录显存和生成时间。 原生视频规格为 10 秒 · 1920×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A high-resolution local video test documenting memory use and generation time. Native-video output: 10s · 1920×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YTkexue/status/2101616134759793122",
    "sourceLabel": "X 原帖 · @YTkexue",
    "author": "氪学家",
    "publishedAt": "2026-09-20T10:15:07.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101616134759793122.mp4",
    "posterUrl": "/posters/x/2101616134759793122.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1056",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3直出1080p\n4080 32G显存没崩\n耗时22分钟\n\n数据仅供参考",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 10,
      "views": 3012,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101614952087994659",
    "title": "加速模型的画质对比",
    "titleEn": "A Quality Check with an Acceleration LoRA",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "对比加速 LoRA 与原始质量的画面差异。 原生视频规格为 15 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A visual comparison of an acceleration LoRA and the original quality. Native-video output: 15s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kaero/status/2101614952087994659",
    "sourceLabel": "X 原帖 · @kaero",
    "author": "马若顽",
    "publishedAt": "2026-09-20T10:10:25.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101614952087994659.mp4",
    "posterUrl": "/posters/x/2101614952087994659.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MinimaxH3 输出的视频一直有个很奇怪的问题：一定要台词说完，bgm才会入场，这明显不是我想要，所以今天一直在搞这个事情。结果关掉加速lora之后发现，原版的画面质感真比加速之后强了不少。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 118,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101614592489333011",
    "title": "参考镜头驱动的舞蹈镜头",
    "titleEn": "Reference-Driven Dance Shots",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "参考视频驱动的舞蹈片段，包含连续动作镜头。 原生视频规格为 15 秒 · 480×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Reference-driven dance shots with continuous motion. Native-video output: 15s · 480×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/onii_ai/status/2101614592489333011",
    "sourceLabel": "X 原帖 · @onii_ai",
    "author": "おにい",
    "publishedAt": "2026-09-20T10:08:59.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101614592489333011.mp4",
    "posterUrl": "/posters/x/2101614592489333011.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×832",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tikユーザーなら聴きすぎて頭おかしくなるであろうこの曲でお送りしますww\n\n軽い振り付け動画、チャレンジしてみたけどどうでしょうか😅\n\n背景一部欠けたり、肝心の振り付けができてなかったりするけど、動きは問題なさそう\n\nそもそも\nminimax/h3/reference-to-video使用であっていたのかな…\nまだまだよくわかってないなww",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 11,
      "views": 419,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101585280692322558",
    "title": "奶奶们的电影预告片",
    "titleEn": "A Film Trailer Reimagined with Grandmothers",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "将电影预告片的叙事结构与奶奶角色结合，形成完整短片。 原生视频规格为 75 秒 · 3483×1960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A full short that combines film-trailer structure with grandmother characters. Native-video output: 75s · 3483×1960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/inoueshokudou/status/2101585280692322558",
    "sourceLabel": "X 原帖 · @inoueshokudou",
    "author": "井上食堂｜映像ディレクター × AI",
    "publishedAt": "2026-09-20T08:12:31.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101585280692322558.mp4",
    "posterUrl": "/posters/x/2101585280692322558.jpg",
    "duration": 75,
    "aspectRatio": "landscape",
    "resolution": "3483×1960",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "2週間もの間VIVANTが待ち遠しいのでおばあちゃんたちの「VIVANT」の新トレーラーを作ってみた。\n\n画像生成 : GPT Image2.5\n動画 : MiniMaxH3 on @PolloAIJP\n\n制作ポイントはリプ欄へ↓",
    "engagement": {
      "replies": 7,
      "reposts": 68,
      "likes": 380,
      "views": 86879,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101584524841005414",
    "title": "雨中的短篇电影",
    "titleEn": "A Short Film in the Rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "音乐和影像共同推进的短篇电影片段。 原生视频规格为 41 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-film excerpt driven by music and imagery. Native-video output: 41s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TaoRInne/status/2101584524841005414",
    "sourceLabel": "X 原帖 · @TaoRInne",
    "author": "輪廻タヲ👁️‍🗨️👄👁️‍🗨️",
    "publishedAt": "2026-09-20T08:09:30.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101584524841005414.mp4",
    "posterUrl": "/posters/x/2101584524841005414.jpg",
    "duration": 41,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "--- Short Movie「RAIN」---\n\n※音楽が流れます。\n\nMusic by @suno V6\n\n“探しているの　どんな音\n 水のたまりに　沈んでく”\n\nTopview / Minimax H3\nGPT Imageで生成したマルチカットをH3で動画化。\n雨と実写アートの世界を融合してみました✨\n\n@TopviewAIJP\n#雨の日AI動画チャレンジ\n\nhttps://x.com/TopviewAIJP/status/2100930829157556601?s=20",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 21,
      "views": 590,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101553442141389288",
    "title": "五秒对白场景",
    "titleEn": "A Five-Second Dialogue Scene",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "五秒对白场景，集中展示人物表情和口型。 原生视频规格为 5 秒 · 480×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A five-second dialogue scene focused on expression and lip movement. Native-video output: 5s · 480×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2101553442141389288",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-09-20T06:06:00.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101553442141389288.mp4",
    "posterUrl": "/posters/x/2101553442141389288.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "480×480",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「なにこれ、勝手に動いてるよこのゲーム」\n\n「大家がね、新しいゲームが発売されたけど、やる時間がなくて AI にクリアさせるから、時々様子を見てくれ・・・とか言って置いてったんだよね」\n\n「えー、ゲームって自分でやるから楽しいんじゃないの～？」\n\n「わたしもそう思う」\n\nMiniMax H3\n#ファイアーエムブレム",
    "engagement": {
      "replies": 0,
      "reposts": 4,
      "likes": 10,
      "views": 177,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101551711160795401",
    "title": "量化模型的生成尝试",
    "titleEn": "A Quantized Model Generation Attempt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "对量化模型进行短视频生成测试。 原生视频规格为 10 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short video generation test with a quantized model. Native-video output: 10s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/okonogiret/status/2101551711160795401",
    "sourceLabel": "X 原帖 · @okonogiret",
    "author": "小此木",
    "publishedAt": "2026-09-20T05:59:07.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101551711160795401.mp4",
    "posterUrl": "/posters/x/2101551711160795401.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3-Singularityの量子化モデルができてたのでお試し\n生成時間とメモリ消費は１割ほど落とせたが品質は明らかに劣化\n正直中途半端",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 79,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101546281566892106",
    "title": "复古动画舞蹈片段",
    "titleEn": "A Retro Animated Dance Clip",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "复古动画风格的舞蹈开场片段。 原生视频规格为 9 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A retro animated dance intro clip. Native-video output: 9s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MedioConxx/status/2101546281566892106",
    "sourceLabel": "X 原帖 · @MedioConxx",
    "author": "GooseFun",
    "publishedAt": "2026-09-20T05:37:33.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101546281566892106.mp4",
    "posterUrl": "/posters/x/2101546281566892106.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I turned these two characters into a 2000s cartoon dance intro by MiniMax H3！ ❤️✨\n\nThe exaggerated poses, retro stage graphics and synchronized choreography make them feel like they came straight out of an old animated TV opening.\n\nPrompt below ↘️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 80,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101540945200374026",
    "title": "文本生成镜头尝试",
    "titleEn": "A Text-to-Motion Attempt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "T2VA",
    "summary": "文本生成视频测试，展示场景和运动变化。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A text-to-video test showing scene and motion changes. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mix_buchi_/status/2101540945200374026",
    "sourceLabel": "X 原帖 · @mix_buchi_",
    "author": "buchi_",
    "publishedAt": "2026-09-20T05:16:20.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101540945200374026.mp4",
    "posterUrl": "/posters/x/2101540945200374026.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "text"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "始まりました。\nテスト映像です。特定するものではありません(笑)\n\n#AImovie\n#MinimaxH3 T2V",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 666,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101540317019443359",
    "title": "失败样本也值得记录",
    "titleEn": "A Pair of Generation Attempts",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "记录两次生成尝试，保留失败结果作为模型边界参考。 原生视频规格为 7 秒 · 896×1184。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two generation attempts preserved as a reference for model limits. Native-video output: 7s · 896×1184. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/afguWKXi8oxKR7h/status/2101540317019443359",
    "sourceLabel": "X 原帖 · @afguWKXi8oxKR7h",
    "author": "おそおそ🇯🇵＠AIイラスト",
    "publishedAt": "2026-09-20T05:13:50.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101540317019443359.mp4",
    "posterUrl": "/posters/x/2101540317019443359.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "896×1184",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "なんだか失敗したガムちゃんMINIMAX-H3動画(*´ω｀*)\n一つ目はどこいくのー(汗)ってなる\n二枚目は全体的にガムちゃんぽいカンジが出てるのだけれど、参照画像どこいった？ってなるなｗｗｗ\n\n#AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️ #aiイラストㅤㅤㅤㅤ",
    "engagement": {
      "replies": 2,
      "reposts": 5,
      "likes": 29,
      "views": 370,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101511003213312129",
    "title": "午夜首尔时装短片",
    "titleEn": "A Midnight Seoul Fashion Film",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "首尔夜景与时装表演组成的电影感短片。 原生视频规格为 16 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cinematic short combining Seoul at night with a fashion performance. Native-video output: 16s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2101511003213312129",
    "sourceLabel": "X 原帖 · @ImaStudio_ai",
    "author": "Ima Studio",
    "publishedAt": "2026-09-20T03:17:22.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101511003213312129.mp4",
    "posterUrl": "/posters/x/2101511003213312129.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Midnight Seoul K-Fashion Film | 8 Looks + 8 Scenes + Dynamic Typography 🖤\n8 LOOKS. 8 WORLDS.\nTry it 👉 https://www.imastudio.com/community?utm_source=x&utm_medium=social&utm_campaign=ixa&utm_content=x38\nPrompt 👇\nCreate a 15s, 16:9, 24fps live-action fashion transformation MV. Use Image1 as the only female identity reference; Image2 only for pacing.\nStyle: K-pop comeback film × black-red haute couture × Seoul Midnight × luxury editorial × typography. Palette: black, burgundy, cherry red, gunmetal silver, cool white. Mood: cold, mature, sexy, restrained, expensive. No cyberpunk/nightclub look.\nKeep one East Asian woman only, preserving Image1’s face, skin tone, body proportions, and long dark-brown hair. No face drift, extra people, or hair-color change.\nEvery strong beat changes outfit + location + composition. Each look uses a different space. Controlled actions only; no complex dance. Use different English/Korean fashion typography in each scene without covering her face.\n\n0–1.5s: black mirrored elevator, black crop top + silver mini; fast push-in. “MIDNIGHT / SEOUL”.\n1.5–3s: rainy Seoul street, burgundy mini + black jacket; low tracking look-back. “After Dark”.\n3–4.5s: red-carpet hotel stairs, oversized black blazer + mini; low-angle push-in. “OWN IT”.\n4.5–6s: minimalist gallery + red sculpture, gunmetal jacket + black mini; reveal. “UNTAMED / 나다운 순간”.\n6–7.5s: backstage vanity, black off-shoulder top + silver skirt; lip touch-up, mirror reveal. Chrome “ICON”.\n7.5–9s: parking garage + black car, black-red jacket + black mini; Dutch angle. “NO BRAKES”.\n9–10.5s: burgundy-curtain theater, burgundy leather mini; spotlight turn. “나답게 / MY RULES.”\n10.5–12s: high-rise glass corridor, fitted black mini dress; walk forward, stop. “THE FINAL LOOK”.\n12–15s: Seoul rooftop hero. Previous looks return as contact sheets, magazine pages, film strips and posters in a 3D collage. Text: “DARK” → “IS BEAUTIFUL” → “나만의 밤.” → “SOLVÉRA — MIDNIGHT SEOUL / OUT NOW”. Final 0.5s freezes like a magazine cover.\nPhotoreal fashion cinematography, real skin/hair/materials. Avoid identity drift, extra people, plastic skin, deformation, repeated sets/type, gibberish, logos, watermarks, subtitles, UI.\n\n#MiniMaxH3 #AIVideo #FashionFilm #KpopAesthetic",
    "engagement": {
      "replies": 6,
      "reposts": 25,
      "likes": 178,
      "views": 9554,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101508602913775879",
    "title": "纸艺风格的本地工作流",
    "titleEn": "A Paper-Craft Style Generation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "纸艺风格实验，展示官方 Skill 的本地生成结果。 原生视频规格为 12 秒 · 1008×1392。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A paper-craft style experiment showing a local generation result. Native-video output: 12s · 1008×1392. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaekagamiAi/status/2101508602913775879",
    "sourceLabel": "X 原帖 · @MaekagamiAi",
    "author": "前鏡",
    "publishedAt": "2026-09-20T03:07:49.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101508602913775879.mp4",
    "posterUrl": "/posters/x/2101508602913775879.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "1008×1392",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "公式Skillのペーパークラフトで遊んでみた🤗\n#MiniMaxH3 #ComfyUI",
    "engagement": {
      "replies": 4,
      "reposts": 6,
      "likes": 47,
      "views": 1355,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2101836184943288709",
    "title": "双重参照让角色逐帧完成",
    "titleEn": "A Character Takes Shape from Two Reference Views",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "全身与近景两张参照图共同约束角色细节，时间推移镜头展示人物从草图走向完整形象。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Full-body and close-up references constrain character detail as a time-lapse shot shows the figure taking shape. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/trader_hashang/status/2101836184943288709",
    "sourceLabel": "X 原帖 · @trader_hashang",
    "author": "はっしゃん@株Biz",
    "publishedAt": "2026-09-21T00:49:31.000Z",
    "addedAt": "2026-09-21T01:13:49.802Z",
    "mediaUrl": "/media/x-2101836184943288709.mp4",
    "posterUrl": "/posters/x/2101836184943288709.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ClaudeCode+MiniMaxH3 ローカル生成のオリジナル15秒タイムラプス動画。参照画像を全身とアップの２枚にすることで、顔の描き込みが綺麗に出るようになった🧐\n#MinimaxH3 #ComfyUI #ClaudeCode",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 2206,
      "snapshotAt": "2026-09-21T01:13:49.802Z"
    },
    "approvedAt": "2026-09-21T01:13:49.802Z",
    "editorialBasis": "Reviewed the original X post, timestamp, native video, and visible caption; no prompt was inferred."
  },
  {
    "id": "x-2102261324860293180",
    "title": "卡通桌面里的追逐与归位",
    "titleEn": "Cartoon Desktop Chase and Reset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "卡通角色在桌面上追逐蟑螂，打乱三个应用图标后按顺序逐一归位，画面保持连续镜头。 原生视频规格为 12 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Cartoon characters chase cockroaches across a desktop, dislodge three app icons, and restore them one by one in a continuous shot. Native-video output: 12s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 12-second horizontal 16:9 video at 1280×720 using the provided cartoon macOS desktop image as the EXACT first frame and visual reference.\n\nEXACTLY 4 characters only: ONE large friendly Big Blue Dog with a fly swatter + THREE clearly different mischievous cockroaches. No other characters. All remain 2D cartoon style.\n\nStatic front-facing camera, one continuous shot, no cuts, zoom, pan, or camera movement.\n\n0–2.5s: Big Blue Dog playfully chases the three cockroaches across the living-room floor toward the RIGHT side. Harmless slapstick comedy.\n\n2.5–4s: Dog swings the fly swatter. ONLY these three desktop icons are physically dislodged: Gmail, Discord, Microsoft Teams. They gently rotate, fall, and land separately on the open desktop above the Dock. NO OTHER ICON moves.\n\n4–5s: Dog notices the fallen icons, looking surprised. Cockroaches look mischievous.\n\n5–10s: Dog physically restores the icons ONE AT A TIME in this exact order:\nGmail → exact original position → click.\nDiscord → exact original position → click.\nMicrosoft Teams → exact original position → click.\n\nNo teleporting, snapping, automatic restoration, or simultaneous movement.\n\n10–12s: Cockroaches run toward the RIGHT. Dog follows briefly, then stops and looks at the viewer innocently while the three cockroaches look back mischievously.\n\nDESKTOP MUST REMAIN IDENTICAL: all 18 original applications, labels, logos, menu bar, Dock, wallpaper, sofa, window, books, furniture, popcorn, floor, and background stay unchanged. ONLY Gmail, Discord, and Microsoft Teams may move. No duplicates, missing icons, altered logos, or extra characters.\n\nAUDIO: playful cartoon piano/pizzicato, footsteps, swatter whoosh, icon movement/landing sounds, and three distinct soft computer clicks. No dialogue, subtitles, captions, text overlays, or watermark.\n\nPremium smooth 2D cartoon animation, clean outlines, consistent proportions, squash-and-stretch, natural shadows, warm cinematic lighting.\n\nFINAL FRAME: Gmail, Discord, and Microsoft Teams are EXACTLY back in their original positions; all other 15 icons remain stationary; exactly four characters visible on the RIGHT side.",
    "promptSourceUrl": "https://x.com/SimplyAnnisa/status/2102261324860293180",
    "sourceUrl": "https://x.com/SimplyAnnisa/status/2102261324860293180",
    "sourceLabel": "X 原帖 · @SimplyAnnisa",
    "author": "Anissa",
    "publishedAt": "2026-09-22T04:58:52.000Z",
    "addedAt": "2026-09-26T12:48:33.725Z",
    "mediaUrl": "/media/x-2102261324860293180.mp4",
    "posterUrl": "/posters/x/2102261324860293180.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized",
      "illustration"
    ],
    "scenes": [
      "daily-life",
      "fantasy"
    ],
    "inputTypes": [
      "text",
      "image"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Oggy and the cockroaches just turned this Mac desktop into pure chaos! 😂\n\nMade with minimax H3 on @TapNow_AI \n\nCreate a 12-second horizontal 16:9 video at 1280×720 using the provided cartoon macOS desktop image as the EXACT first frame and visual reference.\n\nEXACTLY 4 characters only: ONE large friendly Big Blue Dog with a fly swatter + THREE clearly different mischievous cockroaches. No other characters. All remain 2D cartoon style.\n\nStatic front-facing camera, one continuous shot, no cuts, zoom, pan, or camera movement.\n\n0–2.5s: Big Blue Dog playfully chases the three cockroaches across the living-room floor toward the RIGHT side. Harmless slapstick comedy.\n\n2.5–4s: Dog swings the fly swatter. ONLY these three desktop icons are physically dislodged: Gmail, Discord, Microsoft Teams. They gently rotate, fall, and land separately on the open desktop above the Dock. NO OTHER ICON moves.\n\n4–5s: Dog notices the fallen icons, looking surprised. Cockroaches look mischievous.\n\n5–10s: Dog physically restores the icons ONE AT A TIME in this exact order:\nGmail → exact original position → click.\nDiscord → exact original position → click.\nMicrosoft Teams → exact original position → click.\n\nNo teleporting, snapping, automatic restoration, or simultaneous movement.\n\n10–12s: Cockroaches run toward the RIGHT. Dog follows briefly, then stops and looks at the viewer innocently while the three cockroaches look back mischievously.\n\nDESKTOP MUST REMAIN IDENTICAL: all 18 original applications, labels, logos, menu bar, Dock, wallpaper, sofa, window, books, furniture, popcorn, floor, and background stay unchanged. ONLY Gmail, Discord, and Microsoft Teams may move. No duplicates, missing icons, altered logos, or extra characters.\n\nAUDIO: playful cartoon piano/pizzicato, footsteps, swatter whoosh, icon movement/landing sounds, and three distinct soft computer clicks. No dialogue, subtitles, captions, text overlays, or watermark.\n\nPremium smooth 2D cartoon animation, clean outlines, consistent proportions, squash-and-stretch, natural shadows, warm cinematic lighting.\n\nFINAL FRAME: Gmail, Discord, and Microsoft Teams are EXACTLY back in their original positions; all other 15 icons remain stationary; exactly four characters visible on the RIGHT side.",
    "engagement": {
      "replies": 67,
      "reposts": 15,
      "likes": 180,
      "views": 17781,
      "snapshotAt": "2026-09-26T12:48:33.725Z"
    },
    "approvedAt": "2026-09-26T12:48:33.725Z",
    "editorialBasis": "Original X post, native video, timestamp, and complete creator-authored prompt were checked on the source page."
  },
  {
    "id": "x-2103829462169309523",
    "title": "紫黑球衣球员跪在绿茵场",
    "titleEn": "Player in Purple and Black Kneeling on the Pitch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿茵场上，一名穿紫黑球衣的球员跪地，双手撑住草坪，镜头逐渐靠近他的上半身。 原生视频规格为 5 秒 · 640×352。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A player in a purple and black jersey kneels on the pitch, braces both hands on the grass, and the camera moves closer to the upper body. Native-video output: 5s · 640×352. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/140491935226/status/2103829462169309523",
    "sourceLabel": "X 原帖 · @140491935226",
    "author": "ペティー",
    "publishedAt": "2026-09-26T12:50:05.000Z",
    "addedAt": "2026-09-26T14:02:28.676Z",
    "mediaUrl": "/media/x-2103829462169309523.mp4",
    "posterUrl": "/posters/x/2103829462169309523.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "640×352",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "photoreal"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "5秒動画（MiniMax H3）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 19,
      "snapshotAt": "2026-09-26T14:02:28.676Z"
    },
    "approvedAt": "2026-09-26T14:02:28.676Z",
    "editorialBasis": "Checked the original X post and the beginning, middle, and end of its native video."
  },
  {
    "id": "x-2103841372730236989",
    "title": "少年从卧室骑向街机厅",
    "titleEn": "From a Bedroom to an Arcade in a Faded Summer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少年在复古卧室醒来，骑车穿过安静的郊区，最后走进霓虹街机厅，画面带有轻微手持晃动和旧录像带颗粒。 原生视频规格为 15 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A boy wakes in a retro bedroom, rides through quiet suburban streets, and ends among glowing arcade cabinets, with gentle handheld motion and worn VHS grain. Native-video output: 15s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hadiprassetyo_/status/2103841372730236989",
    "sourceLabel": "X 原帖 · @hadiprassetyo_",
    "author": "Hadi Prasetyo",
    "publishedAt": "2026-09-26T13:37:25.000Z",
    "addedAt": "2026-09-27T02:56:20.411Z",
    "mediaUrl": "/media/x-2103841372730236989.mp4",
    "posterUrl": "/posters/x/2103841372730236989.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "cinematic-realistic"
    ],
    "scenes": [
      "daily-life",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A nostalgic vlog brought to life with MiniMax H3 × LoRA natural motion, expressive character details, and worn VHS texture.\n\n@Sogni_Protocol",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 82,
      "snapshotAt": "2026-09-27T02:56:20.411Z"
    },
    "approvedAt": "2026-09-27T02:56:20.411Z",
    "editorialBasis": "原帖 caption，以及视频开头、中间、结尾的实际观看证据。"
  },
  {
    "id": "x-2105438735248801831",
    "title": "猎人穿越文明与未来战场",
    "titleEn": "A Hunter Across Civilizations and a Future Battlefield",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同一位男子从史前猎人变换服饰，穿越金字塔、海港和现代城市；未来都市遭到无人机袭击，最后出现走过残骸的多臂战斗机器人。 原生视频规格为 40 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The same man changes clothing as he moves from prehistory through pyramids, a seaport and a modern city. Drones attack the future skyline before a multi-armed battle robot walks across the remains. Native-video output: 40s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "One Breath of History, then the Machine (time journey v5)\n\nFilm style\nCinematic live-action, photorealistic, epic natural light, rich period-accurate detail, one continuous evolving take with no cuts, fades or resets. The same man is present throughout; the world and his clothing transform seamlessly in place. No generated subtitles, captions or on-screen text.\n\nCharacter A: The traveller\nA rugged man in his thirties with a weathered, expressive face, real skin texture, dark tousled hair and a short dark beard, lean and strong. The same face in every era.\n\nEra 1 clothing (changes with each era as the shots direct): a ragged grey wolf pelt draped over one shoulder, a fur loincloth, bare dust-streaked arms, a leather cord around the neck.\npart_01 (15s, cast A)\n\nComposition: Static wide shot. At the start: a rocky cliff at sunrise, A standing on it as a prehistoric hunter: a ragged grey wolf pelt draped over one shoulder, a fur loincloth, bare dust-streaked arms, a leather cord around his neck, holding a crude wooden spear with a flint tip, watching a herd of animals on the plain below.\n\nAction: One seamless unbroken take with no cuts, fades or resets; every change is a continuous transformation in place, and his face never changes. [0s-3s] Prehistoric: A stands on the rocky cliff at sunrise in the wolf pelt and fur loincloth with the flint-tipped spear, watching the herd. [3s-6s] The cliff dissolves into a Sumerian mud-brick settlement; his pelt becomes a knee-length woven wool kaunakes skirt with tufted fringes, bare chest, a plain linen sash and leather sandals, and the spear becomes a long wooden farming staff; he runs through the settlement. [6s-9s] The settlement becomes ancient Egypt with pyramids; he now wears a white pleated linen shendyt kilt, a broad beaded collar in blue and gold, a striped linen headcloth and sandals, with kohl-lined eyes, walking among the pyramids. [9s-12s] Egypt becomes a grand Roman road of stone and sand; he now wears desert travel attire: a dusty cream tunic under a hooded sand-coloured wool cloak, a leather belt with a waterskin, laced sandals, the hood shading his face from the sun; he travels along the road. [12s-15s] The road becomes the interior of a medieval castle hall; he enters as a knight in full plate steel armour with a mail coif, a plain white surcoat, gauntlets and a sword at his hip, visor raised so his face stays visible. No dialogue; ambient sound of each era only.\nCamera: Completely static, locked-off wide camera; no movement, zoom or reframing at all. The scenery and the man's clothing transform in place within the frame.\n\npart_02 (15s, cast A, ends at 13.0s)\nComposition: Continues exactly from the previous frame: A as a knight in full plate steel armour with a mail coif, plain white surcoat and raised visor, standing inside a medieval castle hall.\nAction: One seamless unbroken take continuing the previous part with no cut, fade or reset; every change is a continuous transformation in place, and his face never changes. [0s-3s] The castle hall dissolves into a sandy beach at the water's edge with tall ships sailing out in the distance; his armour becomes an 18th-century sailor's coat of dark navy wool with brass buttons over a striped jersey, white canvas trousers rolled to the calf, a knotted red neckerchief, barefoot; he walks along the shoreline. [3s-6s] The beach becomes a colonial port city of the mid-1800s: brick warehouses, a cobbled quay, tall ships' masts, horse carts and stacked crates; he now wears a mid-19th-century gentleman's suit of that exact period: a long black wool frock coat, a buttoned waistcoat with a gold watch chain, a high starched wing collar with a dark silk cravat, narrow grey striped trousers, polished black leather boots and a black silk top hat on his head; he stands on the quay. [6s-9s] The port becomes a modern urban street; he now wears contemporary casual clothes: a grey hoodie under an olive bomber jacket, dark jeans, white trainers, earbuds; he walks along the pavement.\n[9s-12s] The street becomes a bustling present-day metropolis of glass towers and crowds; he now wears a slim black overcoat over a plain white t-shirt, dark trousers and sleek trainers, a smartwatch on his wrist; he stands still among the motion. [12s-15s] The metropolis becomes a far-future city of luminous towers; he now stands completely still in an advanced white-and-graphite space suit with articulated plates, a softly glowing blue chest indicator and a clear helmet visor showing his face, observing the landscape while faint visual echoes of all the past eras layer within the sky and the architecture. The take ends there. No dialogue, ambient sound only.\nCamera: Completely static, locked-off wide camera; no movement, zoom or reframing at all. The scenery and the man's clothing transform in place within the frame.\n\npart_03c (6s, cast A)\nComposition: Continues exactly from the previous frame: the man in the white-and-graphite space suit standing completely still before the far-future city of luminous towers.\nAction: One seamless unbroken take continuing the previous part with no cut, fade or reset; the framing stays identical. [0s-1.5s] Sci-fi robot quadcopters, four-rotor drones with mounted guns and red sensor lights, swarm in over the futuristic city firing bright tracer rounds; explosions bloom on the luminous towers and smoke rises. [1.5s-3s] A is hit by the drone gunfire and collapses to the ground, falling face up, completely motionless, eyes closed, no blood or gore. [3s-6s] Other people in futuristic clothing run into frame from both sides toward him; the drones open fire on them and each one is knocked down by the bullets mid-stride, falling onto A and onto each other, one after another, until a pile of the fallen lies still on top of him at the foot of the burning city. Nobody kneels, helps or crouches; they only run, get shot and fall. The drones keep circling and firing overhead. The take ends there. No dialogue; rotor whine, gunfire and distant explosions.\nCamera: Completely static, locked-off wide camera; no movement, zoom or reframing at all. The scene transforms in place within the frame.\npart_04d (6s, cast none)\nComposition: Continues exactly from the previous frame: the pile of the fallen lying still at the foot of the burning futuristic city, drones circling overhead.\nAction: One seamless unbroken take continuing the previous part with no cut, fade or reset. [0s-1.5s] The drones fly off, the fires die to smoke, and the pile of the fallen dissolves into dusty, faded, sun-bleached bones and skulls spread across the ground; the burning towers become the dark ruined skyline of the same futuristic city on the horizon. The frame is empty of anyone standing. [1.5s-5s] A towering military war robot, a heavily armoured bipedal battle machine in scorched gunmetal plating with battle damage and soot, six articulated arms spread from its shoulders each ending in a weapon: rotary cannons, gun barrels and a missile pod, glowing red optical sensors in an armoured head, exposed hydraulics and steel ribs, no logos or markings, walks into the frame from the right edge with slow heavy steps, crunching across the bones, and steps up onto the heap of skulls in the centre of the frame, stopping there. [5s-6s] It stands still on the bones, then slowly turns its head toward the camera. The take ends there. No dialogue. Audio: no music at all; only the same natural ambience as the rest of the film, wind over the ruins, distant crackling fire and settling smoke, and the machine's heavy metallic footsteps.\nCamera: Completely static, locked-off wide camera; no movement, zoom or reframing at all. The scene transforms in place within the frame.",
    "promptSourceUrl": "https://x.com/qtum/status/2105439229111341546",
    "sourceUrl": "https://x.com/qtum/status/2105438735248801831",
    "sourceLabel": "X 原帖 · @qtum",
    "author": "Qtum",
    "publishedAt": "2026-09-30T23:24:46.000Z",
    "addedAt": "2026-10-08T00:48:23.382Z",
    "mediaUrl": "/media/x-2105438735248801831.mp4",
    "posterUrl": "/posters/x/2105438735248801831.jpg",
    "duration": 40,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "We modified a prompt shared by @umesh_ai to give a special ending and ran it through the https://Qtum.ai Video Studio. \n\nVideo rendered with MiniMax H3, prompt in the comments below!",
    "engagement": {
      "replies": 21,
      "reposts": 11,
      "likes": 60,
      "views": 6149,
      "snapshotAt": "2026-10-08T00:48:23.382Z"
    },
    "approvedAt": "2026-10-08T00:48:23.382Z",
    "editorialBasis": "Watched native video beginning, middle, late drone attack and paused end at 38.74/40s; final robot visually confirmed."
  },
  {
    "id": "x-2102290822125494570",
    "title": "紫色眼眸里的霓虹疾驰",
    "titleEn": "A Neon Drive Reflected in a Purple Eye",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发人物的眼睛映出紫色夜景，镜头切入霓虹高楼间的道路疾驰，再回到眼眸特写。 原生视频规格为 26 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired figure’s eye reflects a purple city at night. The view moves into a fast drive beneath neon towers before returning to the eye close-up. Native-video output: 26s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aiakari_jp/status/2102290822125494570",
    "sourceLabel": "X 原帖 · @aiakari_jp",
    "author": "藍灯 / AiAkari",
    "publishedAt": "2026-09-22T06:56:05.000Z",
    "addedAt": "2026-10-08T00:48:23.382Z",
    "mediaUrl": "/media/x-2102290822125494570.mp4",
    "posterUrl": "/posters/x/2102290822125494570.jpg",
    "duration": 26,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized",
      "vertical"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMAX H3練習でつかってみました。\nまだまだ慣れていかないと😇",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 9,
      "views": 129,
      "snapshotAt": "2026-10-08T00:48:23.382Z"
    },
    "approvedAt": "2026-10-08T00:48:23.382Z",
    "editorialBasis": "Native video reviewed at 3.73s eye close-up, middle city road drive, final24.70/26.4s eye. Original post explicitly MiniMAX H3; reply contains no Prompt."
  },
  {
    "id": "x-2102287755988619328",
    "title": "夜间厨房的一碗家常菜",
    "titleEn": "A Home-Cooked Bowl in a Nighttime Kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "围裙女子在夜间厨房准备食材、翻炒番茄与鸡蛋，最后坐下端起热腾腾的一碗菜。 原生视频规格为 15 秒 · 1024×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An aproned woman prepares ingredients and stir-fries tomatoes and eggs in a kitchen at night, then sits down with a steaming bowl. Native-video output: 15s · 1024×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/eternityspring/status/2102287755988619328",
    "sourceLabel": "X 原帖 · @eternityspring",
    "author": "烁皓",
    "publishedAt": "2026-09-22T06:43:54.000Z",
    "addedAt": "2026-10-08T00:48:23.382Z",
    "mediaUrl": "/media/x-2102287755988619328.mp4",
    "posterUrl": "/posters/x/2102287755988619328.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×576",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 基于 Qwen image 2.1 生成的角色图、分镜图\n一镜直出 15 秒、16:9、24fps\n零token成本，只用几毛钱电费出的视频可还行？",
    "engagement": {
      "replies": 19,
      "reposts": 18,
      "likes": 148,
      "views": 39093,
      "snapshotAt": "2026-10-08T00:48:23.382Z"
    },
    "approvedAt": "2026-10-08T00:48:23.382Z",
    "editorialBasis": "Native15s animation reviewed beginning kitchen counter,9s stir-frying,15s seated bowl; image storyboard is not a complete video Prompt."
  },
  {
    "id": "x-2102267666694341084",
    "title": "发光绳索与回卷按钮的恶作剧",
    "titleEn": "A Glowing Cord and a Retractable-Reel Prank",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "庭院里的两名武士装束角色被发光绳索联系在一起。回卷按钮解除束缚后，持卷尺的角色捂着手指露出错愕表情。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two samurai-styled characters stand in a courtyard linked by a glowing cord. After the reel releases the binding, its holder raises a finger with a stunned expression. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "subject_definitions:\n<Subject 1> is A from <Picture 1>, the boastful caster.\n<Subject 2> is B from <Picture 2>, who calmly releases the binding.\nPreserve both identities, faces, hair, costumes, accessories and art styles. Scale their complete designs to equal standing height, maintaining each character's internal proportions.\n<Audio 1> is the voice-timbre reference for <Subject 1> (S1). Reference its vocal identity and pitch range, not its recorded words.\nsummary:\n[reference generation + audio reference] A 15-second Japanese 2D anime comedy. A binds B with a glowing cord connected to a retractable reel in A's hand. B presses its button. The cord rewinds, and its end tab lightly taps A's finger. End on A's silent, stunned face. Use the selected aspect ratio.\nretention_analysis:\n<Subject 1> (Shots 1-4): fully_preserved - retain identity and design at the established equal-height scale, with new actions.\n<Subject 2> (Shots 1-3): fully_preserved - retain identity and design at the same standing height as A.\n<Audio 1>: reference - voice timbre for A's opening line only.\nReference images provide character identity and style only, not poses, backgrounds or relative size.\ndetailed_description:\nHand-drawn 2D anime, clean lines, cel shading, fluid animation. Exactly two characters in a level stone courtyard in warm daylight.\nA and B have equal standing height, measured from soles to skull tops, excluding ornaments. Neither grows or shrinks between shots. Establish both full bodies at the same distance from a level camera, feet on the same ground line and skull tops aligned. Preserve this scale during movement and close-ups. B must not appear taller or larger overall.\n\nOne palm-sized black-and-yellow retractable measuring reel has one large top button, one outlet and a golden cord ending in a small rounded tab. A holds it in her LEFT hand for the entire video. B touches only the button with her RIGHT index finger. The reel is never transferred or duplicated. No readable markings.\n[Shot 1] A full-body two-shot shows <Subject 1> screen-left and <Subject 2> screen-right, facing each other about one meter apart. Both stand upright at equal depth, visibly the same height. A holds the reel at waist level in her left hand and gestures with her right. The glowing cord winds three separate turns around B's lower torso. B's arms, hands, neck and legs remain free. Show the strand connecting the loops to A's reel.\n\nFrom 0.3 to 2.8 seconds, <Subject 1> (S1) proudly declares using the voice timbre of <Audio 1>, [Japanese] このいとからは、にげられない! Synchronize A's lips, then close them. A magical whoosh accompanies the binding. By four seconds, the glow fades enough to reveal the physical cord. A smiles smugly. B follows the cord with her eyes toward the reel.\n[Shot 2] At 00:04.500, cut to a level medium two-shot showing both faces, B's right hand and A's left hand holding the reel. Keep their equal body scale. B takes one small sideways step toward A, not toward the filming camera. The connecting strand slackens. A keeps holding the reel with its button facing up. At about six seconds, B presses the button once with her right index finger. A dry click sounds. B withdraws her hand immediately. A's left index finger rests beside the outlet. Music stops at the click.\n[Shot 3] At 00:06.500, cut to a steady medium shot including B's torso and A's reel. The spring pulls the cord into the outlet with a mechanical whirr. Show all three turns unwinding around B's costume in sequence. The strand stays continuous; it does not dissolve, break or pass through her body. B stands still with hands clear. The reel stays in A's left hand.\nAfter the last loop clears B, the remaining length retracts. Keep A's reel and the returning tab visible. At about ten seconds, the rounded tab lightly clips A's left index finger beside the outlet with one plastic tick, then rests against the reel. The cord is fully retracted. A flinches silently. No wound, violent whipping or collision between characters.\n[Shot 4] At 00:11.500, cut to a medium close-up of <Subject 1>'s face and hands. Her left palm still holds the reel; her right hand gently holds the tapped left index finger. Her smile vanishes. Eyes widen, pupils become small, eyebrows rise. Her lips part once and remain still. She looks toward B with her familiar stunned, uncomprehending expression. No blush, crying or exaggerated pain. Slowly push in, keeping the held finger near the frame's bottom. Hold the final full second. End directly on A.\n\noverall_soundscape:\nFaint courtyard wind, one magical whoosh, one footstep, one button click, a rewind whirr and one plastic tick. A speaks only the sentence in Shot 1. B is completely silent. After three seconds, no speech, gasps, grunts, screams, laughter or vocal breaths. If B's voice reference is attached, leave it unused. No narration or additional voices.\nnon_diegetic_music:\nSolemn low strings and restrained percussion during the opening. Stop at the button click. Keep the rewind and tap clearly audible, then only faint ambience. No choir, comedy sting or canned laughter.\n\nNo text, subtitles, logos, watermarks, HUD, photorealism, 3D look, chibi, extra people, extra limbs, fused hands, costume changes or changing character scale.",
    "promptSourceUrl": "https://x.com/tokyo_Valentine/status/2102267670620221538",
    "sourceUrl": "https://x.com/tokyo_Valentine/status/2102267666694341084",
    "sourceLabel": "X 原帖 · @tokyo_Valentine",
    "author": "とすくん",
    "publishedAt": "2026-09-22T05:24:04.000Z",
    "addedAt": "2026-10-08T00:54:35.260Z",
    "mediaUrl": "/media/x-2102267666694341084.mp4",
    "posterUrl": "/posters/x/2102267666694341084.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "絶対に切れない糸…\nしかし巻き取りボタンは付いてた😂\n\nMiniMax H3で生成。\n\nプロンプトはリプにあります。\nご自由にご利用ください。",
    "engagement": {
      "replies": 1,
      "reposts": 4,
      "likes": 33,
      "views": 3227,
      "snapshotAt": "2026-10-08T00:54:35.260Z"
    },
    "approvedAt": "2026-10-08T00:54:35.260Z",
    "editorialBasis": "Main native15.168s media: initial two-shot glowing binding,8s reel action,12–15s face/finger ending. Quoted14.464s camera clip excluded from review; complete creator reply expanded."
  },
  {
    "id": "x-2107995955501187452",
    "title": "极光下的冰晶教堂",
    "titleEn": "An Ice Cathedral Beneath the Aurora",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿色极光掠过冰封夜空，行人沿冰面走向透明尖塔与彩色窗户组成的巨大教堂。 原生视频规格为 5 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A green aurora sweeps above a frozen landscape as figures approach a vast cathedral of transparent spires and colored windows. Native-video output: 5s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/axiomofmind/status/2107995955501187452",
    "sourceLabel": "X 原帖 · @axiomofmind",
    "author": "AX⚡",
    "publishedAt": "2026-10-08T00:46:15.000Z",
    "addedAt": "2026-10-08T00:54:35.260Z",
    "mediaUrl": "/media/x-2107995955501187452.mp4",
    "posterUrl": "/posters/x/2107995955501187452.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ice Cathedral\n\nqwen image 2.1 --> minimax h3 --> ltx 2.3 upscale",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 9,
      "snapshotAt": "2026-10-08T00:54:35.260Z"
    },
    "approvedAt": "2026-10-08T00:54:35.260Z",
    "editorialBasis": "Original caption identifies H3 generation plus LTX upscaling.5.248s native video visibly plays, with moving aurora and figures; attached still distinguished from video."
  },
  {
    "id": "x-2107994987732554074",
    "title": "云海上的浮空玻璃温室",
    "titleEn": "A Floating Glass Conservatory Above the Clouds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "玻璃穹顶内布满绿植，整座温室悬浮在云海之上，瀑布从岛屿边缘垂落，镜头缓缓绕行。 原生视频规格为 5 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A planted glass dome floats above the clouds, with waterfalls spilling from the island’s edges as the view slowly moves around it. Native-video output: 5s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/axiomofmind/status/2107994987732554074",
    "sourceLabel": "X 原帖 · @axiomofmind",
    "author": "AX⚡",
    "publishedAt": "2026-10-08T00:42:24.000Z",
    "addedAt": "2026-10-08T00:54:35.260Z",
    "mediaUrl": "/media/x-2107994987732554074.mp4",
    "posterUrl": "/posters/x/2107994987732554074.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "qwen image 2.1 --> minimax h3 --> ltx 2.3 upscale",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 14,
      "snapshotAt": "2026-10-08T00:54:35.260Z"
    },
    "approvedAt": "2026-10-08T00:54:35.260Z",
    "editorialBasis": "Original H3 caption verified; native5.248s clip played, rotation and falling water visible, distinct from ice-cathedral work; no full Prompt in post."
  },
  {
    "id": "x-2107991913295208489",
    "title": "灯笼长廊里的古装乐章",
    "titleEn": "A Period-Costume Song Along a Lantern-Lit Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红金服饰的女子走过灯笼长廊，人物与手持图表的镜头配合歌词切换，结尾回到明月下的古式庭院。 原生视频规格为 58 秒 · 576×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in red-and-gold dress moves through a lantern-lit corridor. Lyric captions accompany character and chart-holding shots before a final moonlit courtyard. Native-video output: 58s · 576×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YZhaosheng/status/2107991913295208489",
    "sourceLabel": "X 原帖 · @YZhaosheng",
    "author": "朴鲁疏狂",
    "publishedAt": "2026-10-08T00:30:11.000Z",
    "addedAt": "2026-10-08T00:54:35.260Z",
    "mediaUrl": "/media/x-2107991913295208489.mp4",
    "posterUrl": "/posters/x/2107991913295208489.jpg",
    "duration": 58,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Suno AI 生成mp3, freevideo minimax h3生成视频，claude 负责剪辑和调度",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 17,
      "snapshotAt": "2026-10-08T00:54:35.260Z"
    },
    "approvedAt": "2026-10-08T00:54:35.260Z",
    "editorialBasis": "Original author identifies H3 video and Suno music;58s native video reviewed opening corridor,32s chart and lyrics,57s moonlit courtyard; no Prompt exposed."
  },
  {
    "id": "x-2107990511164559839",
    "title": "雨夜霓虹与紫色光影乐章",
    "titleEn": "Rainy Neon Streets and a Purple-Light Music Sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "音乐影像在雨夜人群、发光高塔、人物近景和紫色光源的室内空间之间切换，画面配合歌曲标题与歌词排版。 原生视频规格为 233 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The music sequence cuts between rain-soaked crowds, a luminous tower, character close-ups and purple-lit interiors, with song-title and lyric typography. Native-video output: 233s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/atomtanstudio/status/2107990511164559839",
    "sourceLabel": "X 原帖 · @atomtanstudio",
    "author": "Rich · Atom Tan Studio",
    "publishedAt": "2026-10-08T00:24:37.000Z",
    "addedAt": "2026-10-08T00:54:35.260Z",
    "mediaUrl": "/media/x-2107990511164559839.mp4",
    "posterUrl": "/posters/x/2107990511164559839.jpg",
    "duration": 233,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I finally have my music video workflow! I'm using vrgamedevgirl84's claude skill with Opus 5.5 xhigh reasoning. This will be available to use inside Sound / Vision once I finish more testing.\n\nArtist: Dub Pistols (feat. Terry Hall)\nSong: Rapture (Blondie cover)\n\nTools used aside from Opus.\nImage Gen: GPT Image 2.5\nVideo Gen: Minimax H3 running locally\nUpscale: Topaz Labs Starlight Precise 2.6",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 48,
      "snapshotAt": "2026-10-08T00:54:35.260Z"
    },
    "approvedAt": "2026-10-08T00:54:35.260Z",
    "editorialBasis": "232.625s native music video playable; opening luminous tower and rainy crowd,107s hand close-up,late warehouse light/cart observed. Creator explicitly attributes video to local H3; no full Prompt."
  },
  {
    "id": "x-2107996990495678611",
    "title": "夜间加油站的远近面部对照",
    "titleEn": "Face Detail at Different Distances in a Nighttime Gas Station",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两组并排画面中，女子提着红色油桶从加油站走向镜头，展示不同模型变体与步数下的面部细节变化。 原生视频规格为 13 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Side-by-side shots follow a woman carrying a red fuel can toward the camera at a gas station, comparing face detail across model variants and step counts. Native-video output: 13s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "DURATION: 6 SECONDS\nASPECT RATIO: 16:9\n\n<Picture 1> is the character reference.\n<Picture 2> is the gas station location reference.\n\nOne woman walks toward the camera across the wet pavement of a quiet rural gas station at night, carrying a red gas can in her right hand.\n\nThe camera tracks backward smoothly, maintaining a medium shot that gradually tightens toward her face.\n\nPreserve the woman's exact facial identity, hair, body proportions, and clothing. Keep the gas station layout consistent with the location reference.\n\nNatural walking movement, realistic hands, subtle facial expressions, and believable clothing movement.\n\nCinematic live-action 35mm film appearance. Practical gas station lighting, natural shadows, muted colors, subtle film grain.\n\nExactly ONE woman. No additional people or vehicles. No dialogue. No music. No whip-pan.",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2107996990495678611",
    "sourceUrl": "https://x.com/EndFolding79421/status/2107996990495678611",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-10-08T00:50:21.000Z",
    "addedAt": "2026-10-08T00:59:32.140Z",
    "mediaUrl": "/media/x-2107996990495678611.mp4",
    "posterUrl": "/posters/x/2107996990495678611.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 — Testing the “Wall” | BF16 vs INT8\nI'm testing two things here: the difference between a full BF16 model and a smaller INT8 model, and what I call “the wall” — the point where a character's face becomes smudged, distorted, or loses detail.\n\nFor this test, the woman starts at the back of the scene \"the wall\" and walks toward the camera, gradually getting closer to the viewer.\n\nThis was intentional. I wanted to see how each model handles facial detail at different distances, especially as the character approaches the camera.\n\nIt always starts out distorted when she's farther away. What's interesting is seeing how quickly each model corrects that distortion as she gets closer.\n\nModels tested: (no audio)\n\nFeiHou Remix BF16 — 8 and 20 steps\nLink: https://huggingface.co/FX-FeiHou/MiniMax-H3-Remix\n\nSingularity INT8 (Non-Pruned) — 8 and 20 steps\nLink: https://huggingface.co/WarmBloodAban/Minimax-h3_Singularity/blob/main/Minimax-h3_Singularity_ref2va_v1.3_int8.safetensors\n\nSettings:\n\nResolution: 1344 × 768p\nSingle-pass generation\nNo upscaling\nFixed seed across all four tests\nSame scene, prompt, and references\nNo LoRAs except Turbo 8-Step for the 8-step tests\n\nHardware:\n\nRTX 5090 Laptop GPU — 24GB VRAM\n64GB RAM\n\nOf course, these are two different model variants, so this isn't a pure precision comparison, but it's a useful real-world test of their output quality.\n\nPrompt used: \n\nDURATION: 6 SECONDS\nASPECT RATIO: 16:9\n\n<Picture 1> is the character reference.\n<Picture 2> is the gas station location reference.\n\nOne woman walks toward the camera across the wet pavement of a quiet rural gas station at night, carrying a red gas can in her right hand.\n\nThe camera tracks backward smoothly, maintaining a medium shot that gradually tightens toward her face.\n\nPreserve the woman's exact facial identity, hair, body proportions, and clothing. Keep the gas station layout consistent with the location reference.\n\nNatural walking movement, realistic hands, subtle facial expressions, and believable clothing movement.\n\nCinematic live-action 35mm film appearance. Practical gas station lighting, natural shadows, muted colors, subtle film grain.\n\nExactly ONE woman. No additional people or vehicles. No dialogue. No music. No whip-pan.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 18,
      "snapshotAt": "2026-10-08T00:59:32.140Z"
    },
    "approvedAt": "2026-10-08T00:59:32.140Z",
    "editorialBasis": "Original caption and full Prompt read.13.212s native split-screen video played in original and enlarged view; woman approaches camera,labels distinguish INT8/BF16 variants; no quality superiority asserted."
  },
  {
    "id": "x-2107986514072240323",
    "title": "背着行囊叩响古宅木门",
    "titleEn": "A Traveler Knocks at an Old Courtyard Door",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "束发男子背着行囊来到旧宅门前，在花树下取出书册，镜头在人物与古式木门间切换。 原生视频规格为 115 秒 · 704×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A traveler with tied-up hair and a pack arrives at an old house, takes out a booklet beneath flowering branches, and faces a wooden courtyard door. Native-video output: 115s · 704×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/laoda2000/status/2107986514072240323",
    "sourceLabel": "X 原帖 · @laoda2000",
    "author": "laoda",
    "publishedAt": "2026-10-08T00:08:44.000Z",
    "addedAt": "2026-10-08T00:59:32.140Z",
    "mediaUrl": "/media/x-2107986514072240323.mp4",
    "posterUrl": "/posters/x/2107986514072240323.jpg",
    "duration": 115,
    "aspectRatio": "landscape",
    "resolution": "704×384",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#chatgpt + #minimax h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 17,
      "snapshotAt": "2026-10-08T00:59:32.140Z"
    },
    "approvedAt": "2026-10-08T00:59:32.140Z",
    "editorialBasis": "115.208s native video reviewed12s door profile,65s booklet under flowering tree,110s doorway ending. Original post identifies minimax h3; no public Prompt in conversation."
  },
  {
    "id": "x-2107967372594233675",
    "title": "晶洞水池旁的粒子消散",
    "titleEn": "Particle Disintegration Beside a Crystal-Cavern Pool",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣人物将手伸向发光水池，手臂在晶体岩壁间逐渐变成蓝色光粒与漂浮碎片。 原生视频规格为 5 秒 · 480×832。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A dark-clad figure reaches toward a luminous pool as the arm breaks into blue particles and floating fragments among crystal walls. Native-video output: 5s · 480×832. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Cinematic slow-motion shot inside a cavern filled with glowing crystals. A explorer in tactical gear reaches toward a glowing pool. Starting from their fingertips, their arm and body break into bright turquoise bioluminescent particles and jagged crystal shards that float horizontally. The particle trail is dense and clearly visible against dark damp rock faces, featuring vivid rim-lighting.",
    "promptSourceUrl": "https://x.com/tooniaanimation/status/2107967372594233675",
    "sourceUrl": "https://x.com/tooniaanimation/status/2107967372594233675",
    "sourceLabel": "X 原帖 · @tooniaanimation",
    "author": "TOONIA AI Short Films",
    "publishedAt": "2026-10-07T22:52:40.000Z",
    "addedAt": "2026-10-08T00:59:32.140Z",
    "mediaUrl": "/media/x-2107967372594233675.mp4",
    "posterUrl": "/posters/x/2107967372594233675.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "480×832",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "vertical"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cinematic disintegration effect created with MiniMax H3 Max on @picsartai Playground! 💎✨\n\nWatch dissolve into bright turquoise bioluminescent particles and jagged crystal shards in a crystal cavern! Prompt screenshot in reply!\n\nMade with @picsartai  #picsartpartner",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 42,
      "snapshotAt": "2026-10-08T00:59:32.140Z"
    },
    "approvedAt": "2026-10-08T00:59:32.140Z",
    "editorialBasis": "Native5.248s video plays; beginning arm reaches pool,ending fragmented limb/crystal particles. Full Prompt visually transcribed from original creator screenshot, no truncation."
  },
  {
    "id": "x-2107966853658431925",
    "title": "肌肉魔法英雄的舞台变身",
    "titleEn": "A Muscular Magical Hero Takes the Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发壮汉从月夜健身房走向魔法舞台，夸张的肌肉形象与粉色服饰、发光法阵和擂台表演形成反差。 原生视频规格为 203 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired muscular hero moves from a moonlit gym to a magical stage, contrasting a powerful build with pink costume details, glowing circles and ring-side performance. Native-video output: 203s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TechTalkNAVI/status/2107966853658431925",
    "sourceLabel": "X 原帖 · @TechTalkNAVI",
    "author": "テックトークナビゲーター@AIコンテンツクリエイター",
    "publishedAt": "2026-10-07T22:50:36.000Z",
    "addedAt": "2026-10-08T00:59:32.140Z",
    "mediaUrl": "/media/x-2107966853658431925.mp4",
    "posterUrl": "/posters/x/2107966853658431925.jpg",
    "duration": 203,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "バキバキの魔法で、ムキムキの正義を。\n\nでも鏡の前では、昔の傷に触れる夜もある。\n「弱さは捨てない。それも筋肉の一部だろ」\n\n笑えて、ちょっと泣けるテーマソング。\n\n\"マッスルおじ魔女\"\n\n#マッスルおじ魔女 \n🎬 Hailuo H3  #MV #HailuoH3\n\n@Hailuo_AI\n@MiniMax_AI",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 9,
      "views": 378,
      "snapshotAt": "2026-10-08T00:59:32.140Z"
    },
    "approvedAt": "2026-10-08T00:59:32.140Z",
    "editorialBasis": "202.541s native MV played; opening moonlit gym,102s glowing floating ring,196s hero on stage with silhouettes. Original Hailuo H3 caption; no full Prompt in replies."
  },
  {
    "id": "x-2107965998162739561",
    "title": "奶油盆里拔起一座雪山",
    "titleEn": "A Snowy Mountain Rising from Whipped Cream",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金属打蛋器在玻璃盆中搅动奶油，离开时拉起高耸的白色山峰，最终呈现微缩雪山般的形态。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A metal whisk stirs cream in a glass bowl, lifting a tall white peak that becomes a miniature snowy mountain. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the peaks rise into a small snowy mountain range",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2107965998162739561",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2107965998162739561",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-07T22:47:12.000Z",
    "addedAt": "2026-10-08T01:03:42.321Z",
    "mediaUrl": "/media/x-2107965998162739561.mp4",
    "posterUrl": "/posters/x/2107965998162739561.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Whipped to stiff peaks. Elevation unknown.\n\nPrompt: \"the peaks rise into a small snowy mountain range\"\n\nDuration: 15s. Camera: locked, rim height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe whisk leaves by going straight up. That exit is what pulls the summit.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 292,
      "snapshotAt": "2026-10-08T01:03:42.321Z"
    },
    "approvedAt": "2026-10-08T01:03:42.321Z",
    "editorialBasis": "Viewed native15.168s video at start,5s and12s. Original author explicitly labels local H3 and publishes quoted one-sentence Prompt."
  },
  {
    "id": "x-2107955376582799806",
    "title": "星空长廊下的夜游自拍",
    "titleEn": "A Nighttime Selfie Beneath a Starry Arcade",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴护目镜的女子在灯光璀璨的拱廊街道边自拍，镜头转向湿润石路与金色店铺灯光，再回到人物。 原生视频规格为 15 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A goggle-wearing woman films herself beside a brightly lit arcade, turning the view toward wet paving and glowing shop windows before returning to her face. Native-video output: 15s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chiha_20220301/status/2107955376582799806",
    "sourceLabel": "X 原帖 · @chiha_20220301",
    "author": "千早@AIart",
    "publishedAt": "2026-10-07T22:05:00.000Z",
    "addedAt": "2026-10-08T01:03:42.321Z",
    "mediaUrl": "/media/x-2107955376582799806.mp4",
    "posterUrl": "/posters/x/2107955376582799806.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます☕️\n木曜日も頑張りましょう☺️\n\n今日の動画はPixverseのMinimax H3で生成しています！\n#Pixversecpp\n@PixVerse\n夜空に輝く星とお店の灯りが美しい通りからお届けです✨",
    "engagement": {
      "replies": 5,
      "reposts": 5,
      "likes": 21,
      "views": 368,
      "snapshotAt": "2026-10-08T01:03:42.321Z"
    },
    "approvedAt": "2026-10-08T01:03:42.321Z",
    "editorialBasis": "Original creator explicitly says Pixverse MiniMax H3; native15.16s video played, observed selfie and camera turn toward wet illuminated street. No full Prompt published."
  },
  {
    "id": "x-2105346237038043532",
    "title": "湖上小船里的神秘书页",
    "titleEn": "A Mysterious Book on a Lakeside Boat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "FL2VA",
    "summary": "撑伞的小船上，女子读书时被小狗打断，书页间升起白色烟雾；短片以她安抚小狗收尾。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "On a small umbrella-covered boat, a dog interrupts a woman's reading as white smoke emerges from the book; the scene ends with her comforting the dog. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15-second, 16:9, cinematic realistic fantasy short developed strictly from <Picture 1>. Lock the whole sequence inside the same small lake boat space only: the same yellow-blue old wooden boat, the same dark lake water, the same orange-cyan umbrella, the same reeds and purple flowers, and the same woman, dog, and book. No second location, no extra people, no black ending. Preserve the exact same adult woman: auburn-brown long hair, straw hat, deep blue floral bikini, same face, same proportions, same seat on the right side of the boat. Preserve the same brown-white beagle with cyan collar on the left half of the boat. Preserve the same gilt-edged dark green hardcover book, starting half-open in her left hand between them. The emotional line is restrained: she wants to keep reading, the dog interrupts, the book reveals something unnatural, and she shuts it decisively before settling into wary tenderness.\n\nBlocking and continuity: left dog, right woman, middle book, rear umbrella remain stable throughout. She never stands. The dog never crosses into her side. The book follows one strict state progression only: half-open in her hand -> dragged down by the dog onto the yellow deck while still open -> closed by her own hand. Camera must always remain on the same visible gunwale side as in <Picture 1> and never cross the axis formed by the woman and the book.\n\n[00:00-00:02.5]\nSlow push-in, oblique half-body close shot. She leans from a slightly reclined posture toward the book. Her right palm slides along the damp gunwale to stabilize her center of gravity while her left hand lifts the book and teases up half a page with her thumb. One natural inhale lifts the ribcage and swimsuit slightly, then a soft exhale returns it. Water droplets on the shoulder and neck catch light. Focus settles on the page she is about to turn but has not turned yet. White smoke begins to emerge from the page gutter. Reflections on the water and reeds continue moving gently. Her gaze stays on the text, lips slightly parted from concentration.\n\n[00:02.5-00:03.5]\nPan left along the white smoke toward the dog. The dog’s nose follows the smoke. Its eyes shift to the back of her neck before the head fully turns, ears lifting and hind legs storing energy. The dog’s launch toward the right bridges into the next beat.\n\n[00:03.5-00:05.7]\nThe dog bites the lower-left corner of the page. The page tears with a short rip and a few paper fibers. A whip-flash shake catches the woman to the right. She recoils right-rear, her left hand losing the book, her right palm slamming down on the gunwale, and the boat tilting then rebounding. Keep face, shoulders, and upper torso together in frame. As the shoulders and back pull away, the chest follows with one short natural inertial lift; when her body stops and the boat rebounds, that motion resolves into two restrained offset responses, one vertical and one slight lateral, both physically damped by the swimsuit and body weight. Hair whips late and a few droplets leave the skin. Use only a very brief local slow-down at the rebound peak, then return to normal speed. By the end of the shot, the book falls toward the center deck, still on the dog’s side of the axis, and the dog remains left of her.\n\n[00:05.7-00:08.4]\nCut to a low camera inside the same visible side of the boat. The frame stabilizes and depth shifts smoothly. The hard corner of the book cover hits the yellow deck with a muted knock; a little dry worn corner debris breaks off; the pages tremble and remain open. Show the book first, then shift focus to the continuous white smoke rising from the same page gutter. The smoke is drawn longer by the breeze under the umbrella, twists, and condenses into a semi-transparent hand reaching toward where her face was a moment earlier. The smoke-hand casts a passing hand-shaped shadow under the umbrella, consistent with the existing light direction. Her right hand is still bracing the boat. Her gaze moves from the dog to the hand-shadow. The dog has retreated back to the left side of the book. Lake ripples are caused only by the rocking of the boat. Purple flowers and reeds move irregularly in the breeze.\n\n[00:08.4-00:11.2]\nShe does not stand. Her abdomen tightens and her body leans from the right toward the left-front. Her right hand continues to stabilize on the gunwale while her left hand reaches and catches the book cover. The camera tracks left along the same side of the boat, holding her eyes, reaching hand, upper torso, and the book corner in one frame. The forward lean lowers the chest with the shift in center of gravity; the abrupt stop before closing the book creates one short inertial rebound; quick breathing adds a small continuous rise-fall through the chest and swimsuit. Hair tips brush the shoulder and the straw-hat brim trembles lightly. She sets her jaw and slaps the book shut with one decisive movement. The pages clap closed, the cover vibrates, and the smoke-hand collapses from the fingertips backward into fine strands that retract rapidly along the original path back into the book gutter. The umbrella shadow of the hand shrinks out in sync. No floating smoke residue remains.\n\n[00:11.2-00:15.0]\nSlow pull-back. The book remains closed on the center deck, and her left hand is still pressing it down. Her upper chest rises and falls through two short recovery breaths, then gradually returns toward slower breathing. The dog approaches from the left and touches the back of her hand with its nose. She lifts her eyes once to confirm there is nothing above under the umbrella, then her left hand leaves the book cover and moves to the dog’s cyan collar, while her right hand still steadies the gunwale. She says softly in English, with breath not fully settled yet: <d>[English] Good dog.</d> The line is short and light. The camera pulls back to hold the half-body relationship of the woman and dog under the same umbrella. The boat rocks gently. The dog’s tail flicks only once. Her left hand finally rests on the dog’s collar. Do not cut to black.\n\nLighting and color: preserve the original warm-umbrella and cold-lake palette. Warm top light filtered through the umbrella illuminates her skin. The lake gives soft cyan reflections in shadow. Her skin stays naturally warm with real water-droplet sheen. The smoke is cold silver-white and only local. When the smoke-hand appears, only the local light under the umbrella and environmental sound are slightly suppressed; do not change sun direction, time of day, or overall color temperature. After the book is closed, the original warm-cool balance returns.\n\nPhysical logic: the old wooden boat tilts and rebounds locally from the dog’s leap and her bracing hand, and water slaps the hull into uneven ripples. The umbrella fabric and ribs tremble with the boat’s motion but do not come loose. Page tearing, cover striking wood, tiny dry debris, palms pressing wood, and droplets leaving wet skin all carry correct force and sound. The smoke originates only from the book pages, only inside the boat and under the umbrella, never passing through skin, and never causing weather or location shifts. Hair moves in layers; the wettest strands lag the most. Upper-body motion is always driven by breathing, torso acceleration, abrupt stopping, and boat rebound, with real support and no exaggeration.\n\nHard negative constraints: no identity drift, no face deformation, no extra people, no extra limbs, no extra animals, no second location, no wide cutaway, no magical particle storm, no fake smoke residue, no zero-gravity hair, no cloth drift, no black ending, no subtitles, no watermark.\n\noverall_soundscape:\nThe base ambience is intimate and natural: dark lake water lapping the boat, reeds moving, pages rustling, quiet dog breathing, slight umbrella fabric tremor, and soft wet skin or cloth movement. The opening includes page handling, thumb on paper, a subtle breath in, light contact of her hand on the damp gunwale, and the first faint hiss of white smoke from the book gutter. When the camera pans to the dog, bring forward the dog’s sniffing and body tension. At the grab, layer paw scrape on wood, a short paper tear, her sharp inhale, the right palm striking the gunwale, and the boat’s sudden wood-and-water reaction. Immediately after, reduce the environmental bed so the smoke hiss, slight umbrella-rib vibration, and the low uncanny presence of the smoke-hand dominate the foreground. When she lunges back in and shuts the book, the cover closes with a firm muffled clap that cuts the unnatural sound off. Then the lake sound and normal ambience return. In the ending, her breathing is still slightly unsettled as she says <d>[English] Good dog.</d> softly, and the dog answers only with a small nose touch and one tail flick.\n\nnon_diegetic_music:\nKeep the score restrained and cinematic. Begin with a soft warm-cool suspended texture under the water and reeds, almost imperceptible, supporting the concentration of reading. As the smoke appears and the dog notices it, introduce a faint low unease without overpowering natural sounds. At the book snatch and recoil, give one brief accent and then immediately thin the score so paper, breath, boat movement, and smoke hiss lead. During the smoke-hand section, hold a narrow suspended tone with minimal low-frequency pressure. When the book is shut, let that tension collapse quickly into a lighter, warmer emotional residue. End with only a small intimate undertone beneath the returning lake ambience, with no melodramatic swell and no hard stop.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2105346562184610111",
    "sourceUrl": "https://x.com/ou_zhen599/status/2105346237038043532",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-30T17:17:13.000Z",
    "addedAt": "2026-10-08T01:06:20.694Z",
    "mediaUrl": "/media/x-2105346237038043532.mp4",
    "posterUrl": "/posters/x/2105346237038043532.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                                                                        \n\n🎬Image-to-Video                                                       \n📷“Don’t Turn the Page”\nShe just wanted to finish reading a book in peace by the lake. A puppy suddenly lunged at her, shattering the tranquility—until she saw a hand reaching out from the pages of the book toward the spot where she had been sitting just a moment ago.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 1104,
      "snapshotAt": "2026-10-08T01:06:20.694Z"
    },
    "approvedAt": "2026-10-08T01:06:20.694Z",
    "editorialBasis": "Original creator explicitly identifies H3 image-to-video generation; the native video and complete same-author Prompt reply were reviewed."
  },
  {
    "id": "x-2107736392247992627",
    "title": "雨夜站台与渐近的列车灯光",
    "titleEn": "An Approaching Train on a Rainy Night Platform",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "撑透明雨伞的女子独自等候，列车进入站台，灯光映过她的侧脸，最后露出轻微笑容。 原生视频规格为 15 秒 · 1280×736。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman waits alone beneath a clear umbrella as a train arrives, its light passing across her profile before a restrained smile. Native-video output: 15s · 1280×736. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A single continuous live-action long take on a Japanese elevated train platform at night in the rain. One woman waiting alone, camera locked in place at her right side, filming her right profile down the length of the platform. It runs from her small waiting fidgets, through a commuter train arriving and passing in front of her, to the train nearly stopping and a soft side-profile smile. One take, camera fixed with only slight handheld breathing, no cuts, no zoom, no push-in. Normal real-world speed throughout, no slow motion.\n\nShot 1 — 0.0s: Realistic cinematic, real-camera rainy-night platform, one continuous long take, medium shot from mid-thigh up, camera fixed at her right at chest height looking down the platform's depth. Right side of frame is the wet-bright yellow tactile paving, the platform edge, and two steel rails converging into the distance, with the arched glass windscreen beyond; left side is the platform interior and orange glowing shopfront boxes; cool fluorescent tubes recede overhead; red-blue-green neon bokeh at the far end; tiles wet-black and reflective. Riria stands center-slightly-left inside the yellow paving, facing the tracks, her right profile to the lens, shopfront orange rimming her back and bun. The right-side rail is empty, no train yet. She is already mid-action: tilting the umbrella back a little, looking up past its rim at the night sky beyond the canopy, reaching her right hand forward toward the track side past the umbrella rim, palm up, holding a beat to test whether it's still raining, a few drops landing in her palm.\n\nShot 1 continues — 1.5s: She pulls her hand back, rights the umbrella back over her head, lets out a soft breath, and shifts her weight from one foot to the other.\n\nShot 1 continues — 2.0s: Two small, soft cool-white headlights light up at the far end of the right-side rail. She leans her upper body slightly forward (feet not crossing the yellow paving), peering toward the far end of the tracks, body still squared to the rails. The headlights grow larger and brighter extremely gradually along the right rail, the steel, the standing water, and the platform edge lighting up bit by bit.\n\nShot 1 continues — 3.5s: She straightens up, steps back half a step, grips the umbrella handle again with her left hand, her gaze following the approaching train. The light keeps rising slowly and evenly — first catching the side-profile edge of her forehead, nose bridge, and lips, then washing softly onto her face. The same light also falls on the droplets on the umbrella, the rain, the wet tiles, and her coat, no hard edges, no flicker. Her skin is only softly brightened overall, still matte, no shine spot. She squints slightly and tucks her chin a little into the turtleneck.\n\nShot 1 continues — 5.5s: A silver stainless-steel commuter train with a green waistline reaches her along the right-side rail, wheels on the rails, body tight to the platform edge, the nose passing in front of her and exiting frame right. In the same instant a gust blows in from the depth of frame: bangs and side hairs pressed to her cheeks, coat hem, lapels, and back belt lifting, umbrella tilting about a hand's width and trembling. Her left hand steadies the umbrella, her right hand lifts to gather the blown-open coat front at her chest, still facing the tracks. The cool-white headlight glow transitions smoothly into the warm-white carriage glow over about a second, brightness roughly unchanged, only cool-to-warm. Carriages run past in front of her on the right with motion blur, the warm-white light a long steady even band on her face, coat front, and umbrella rim, no flicker. Her feet stay planted, sharp, normal speed, no slow motion.\n\nShot 1 continues — 7.0s to 10.0s: The train visibly decelerates, carriages resolving back into distinguishable windows, inside them only defocused warm-white light and blurred shapes, but the warm light on her stays constant. The wind eases, hems settle, hair slowly drops. Around 8.5s she releases the coat front and uses her right fingertips to tuck the strands stuck to her cheek behind her ear, revealing the pearl stud, then lets her right hand fall naturally.\n\nShot 1 continues — 10.0s: The train glides slowly toward a near-stop, settling on the right-side rail, the warm-white light still steady and soft on her face.\n\nShot 1 continues — 10.5s: Her head turns only slightly toward the lens side, feet and shoulders still toward the tracks, still a clean right profile in frame, at most the far corner of her eye and mouth just showing, never a three-quarter front or full face. Her gaze slants gently toward the lens side, not direct to camera. Framing stays mid-thigh-up medium, no push-in.\n\nShot 1 continues — 12.5s: The corner of her mouth slowly lifts into a soft, restrained closed-lip smile on the profile, cheekbone rising a touch, eye corner curving, never a toothy grin or a posed frozen smile. The smile holds on the profile, she blinks once and breathes out softly, a strand of hair still swaying in the leftover wind, the train completing its last short glide at walking pace with doors staying shut. The image ends while these tiny motions are still underway, never a freeze frame. Live throughout — rain keeps falling, droplets slide and pool on the umbrella, the turtleneck collar rises faintly with her breath, flyaways drift, puddle ripples shimmer, distant neon glows steady, the handheld frame breathing lightly.\n\nAll light changes are continuous, slow gradients, never blinking: no flicker, strobe, or on-off pulsing, no per-frame brightness jumps, no sudden light on or off, no hard-edge pools or stage-spotlight circles, no window-shaped light patches sweeping across her face. The windows are large and the carriage light diffuse, so what lands on her is one steady sheet of warm light. The headlights are just two softly growing points, never direct into the lens, the frame never flashes white. The same light lands on the umbrella, the rain, the wet tiles, the platform edge, and her coat at once, never only on her face. Fluorescent tubes and distant neon glow steady, never flickering. Camera exposure locked the whole time, no auto-brightening or darkening. The whole clip brightens smoothly only once, then holds steady, never bright-then-dark-then-bright. The nose passing in front of her and the gust lifting her hair and hems happen in the same instant (around 5.5s), never wind-before-train or train-already-past-before-wind.\n\nProhibited\n\nNo subtitles, watermark, logo, or overlay text; platform light boxes, billboards, and train destination displays appear only as defocused glowing color blocks, never any readable text. No dialogue or narration. No standing still before the train or woodenly posing before it; no flicking water off the hand. No light blinking, flickering, strobing, or on-off pulsing; no bright-then-dark-then-bright; no sudden light on or off; no headlight flaring white into the lens; no window-shaped light patches sweeping across her face; no stage-spotlight circles or hard light beams; no light landing only on her face; no exposure wavering; no flickering tubes or neon. No standing with her back to the tracks; no facing the shopfronts, platform interior, or lens; no chest toward frame left. No rail on the left side of frame; no train driving onto the platform; no train on the left or shopfront side; no train crossing the frame; no train floating or off the rails; no train passing behind her; no train between her and the lens; no train coming from behind the camera, driving off into the distance, or reversing. No push-in, pull-out, zoom, or orbit. No chest-up close-up or face close-up. No full face to lens, three-quarter front glance back, or direct eye contact. No turning the shoulders or turning around. No early smile. No skin oil, sweat, water shine, specular spots, wet reflection, or droplets on the face. No rubbery, silicone, waxwork, or plastic skin, no smoothing, no freckles vanishing. No grey studio backdrop. No triptych, split-screen, or a second Riria. No other passengers or staff on the platform. No doors opening or anyone getting off. No slow motion or slowed hair drift. No umbrella flipping inside out, flying from her hand, or the shaft crossing her face; no right hand covering her face. No features drifting into a similar-looking different person — it is this person from the reference. No toothy grin. No global amber or teal-orange filter. No CG, anime, over-sharpening, or HDR look. No freeze frame. No third hand or finger distortion.",
    "promptSourceUrl": "https://x.com/kentdhani/status/2107736396861526115",
    "sourceUrl": "https://x.com/kentdhani/status/2107736392247992627",
    "sourceLabel": "X 原帖 · @kentdhani",
    "author": "Kent Dhani",
    "publishedAt": "2026-10-07T07:34:50.000Z",
    "addedAt": "2026-10-08T01:06:20.694Z",
    "mediaUrl": "/media/x-2107736392247992627.mp4",
    "posterUrl": "/posters/x/2107736392247992627.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "city",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Ref2V #6\n\nRan on a RTX5090 locally in ComfyUI. Moving light is a tough one, a passing train throws changing light across the face, which is where identity usually slips. Here's how Ref2V held. \n\nFull prompt 👇",
    "engagement": {
      "replies": 11,
      "reposts": 7,
      "likes": 151,
      "views": 12219,
      "snapshotAt": "2026-10-08T01:06:20.694Z"
    },
    "approvedAt": "2026-10-08T01:06:20.694Z",
    "editorialBasis": "Native15s original video watched beginning and14s ending; H3 Ref2V explicit. Original same-author full Prompt reply expanded; separate quoted8s prior video not used."
  },
  {
    "id": "x-2107679859422195837",
    "title": "夜街上的背负与轻声告白",
    "titleEn": "A Piggyback Walk and a Quiet Request",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "夜色中，男子背着短发女子缓缓前行，两人的面部始终同框，她靠近他的肩头轻声说话。 原生视频规格为 15 秒 · 1944×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A man carries a short-haired woman along a dark street, their faces framed together as she leans toward his shoulder and speaks softly. Native-video output: 15s · 1944×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "r34l1sm\nsubject_definitions:\n<Subject 1> is the adult man from ​@Image1​, preserving his facial identity, tousled short black hair, natural skin texture, white shirt, dark plaid waistcoat, and suspenders.\n<Subject 2> is the adult elf woman from ​@Image2​, preserving her facial identity, short dark bob with bangs, pointed ears, and oversized gray hoodie. Her expression is relaxed, affectionate, and quietly happy.\n​@Image3​ provides the intimate piggyback pose and teal-and-amber nighttime lighting. Use it for pose and atmosphere, giving the woman a warmer, more contented expression.\n\nsummary:\n[reference generation] A fifteen-second affectionate love story in one continuous 16:9 two-person facial close-up. <Subject 1> slowly carries <Subject 2> through a quiet nighttime street. She softly asks to stay like this a little longer. He responds silently with a small smile and slower footsteps. Both faces remain simultaneously sharp. Her Japanese dialogue sounds naturally recorded within the street, with audible distance and environmental space.\n\nretention_analysis:\n<Subject 1> (appears in [Shot 1]): fully_preserved — retain his facial identity, hairstyle, and clothing.\n<Subject 2> (appears in [Shot 1]): fully_preserved — retain her facial identity, hairstyle, pointed ears, and hoodie. Her performance conveys comfort and affection.\n​@Image3​ ([Shot 1] pose and atmosphere): partially_preserved — retain the piggyback relationship and nighttime palette; adapt the framing into a horizontal facial close-up with greater depth of field and a quietly happy female expression.\n\ndetailed_description:\nLive-action cinematic realism, 16:9 landscape, fifteen seconds. A 35mm cinema lens, fine film grain, soft highlight halation, warm natural skin tones, and cool teal nighttime surroundings. Use moderately deep depth of field with enough soft, motivated street lighting to render both faces clearly.\n\n[Shot 1] One uninterrupted take. A tight two-person facial close-up from slightly ahead and to the couple's left. <Subject 1>'s complete face occupies the lower-left center; <Subject 2>'s complete face appears above his right shoulder in the upper-right center. Include small portions of his collar and her gray sleeves. Her pointed ear remains visible through her short hair. Maintain a comfortable camera distance and natural facial proportions.\n\nBoth faces remain simultaneously in sharp focus throughout the entire shot. Clearly resolve both characters' eyes, eyelashes, lips, and skin texture. Use sufficient depth of field to cover the distance between their faces and their gentle walking movements. The woman is as sharply focused as the man, including when she rests against his shoulder. Prioritize equal facial clarity over strong background blur. The street is gently softened, with recognizable shapes. No selective focus on the man, rack focus, or focus hunting.\n\nHe carries her slowly along the street. The camera retreats at their pace, maintaining a nearly constant distance and the same facial framing. Its movement has a small, smooth breathing sway and gentle vertical drift motivated by careful footsteps. Fixed focal length, continuous perspective, no cuts, orbit, zoom, or abrupt reframing.\n\nDuring the opening seconds, he watches the path ahead with a relaxed expression. She rests comfortably against him, her brow smooth and her lips carrying a faint, natural smile. She briefly looks toward his cheek, then lowers her gaze toward his collar. Her expression conveys affection and ease, with no sorrow, tearful eyes, trembling lips, or worried brows. Neither looks into the camera.\n\nNear the middle, <Subject 2> (S1) turns her mouth slightly toward his ear and speaks naturally in a warm, gently smiling Japanese voice: <d>[Japanese] もう少し、このままで。</d>\n\nHear her sentence from the camera's position, approximately one to two meters away. Her voice is modest in volume, slightly softened by distance, and blended with the surrounding street. Turning toward his ear makes it subtly less direct to the listener. Keep the words intelligible without making the voice unnaturally prominent. Avoid exaggerated whispering, amplified mouth sounds, and studio-isolated vocal clarity.\n\nHe remains silent. After a brief listening pause, his cheek lifts into a small smile and he slows his next steps. She notices, her smile warms slightly, and she gently rests her cheek against his shoulder. Her lips remain closed after the line. Keep both faces sharp as a warm streetlight passes softly across them.\n\nHold their comfortable closeness through the remaining seconds, with natural blinking and gentle walking motion. End while they are still moving together. No subtitles, narration, or additional dialogue.\n\noverall_soundscape:\nA continuous, spatially coherent nighttime street recording heard from the camera's perspective. Footsteps and subtle clothing rustles occupy the nearby space. A low storefront ventilation hum comes from one side. Distant traffic sits farther away; a passing car gradually approaches, moves across the stereo field, and recedes without obscuring the dialogue.\n\nThe woman's voice belongs within this same acoustic environment. Keep street ambience continuously audible beneath her sentence, without muting it or dramatically lowering it when she speaks. Include faint, short reflections from nearby shopfront glass and building surfaces while retaining an open-air sound. No long reverberation, obvious echo, or indoor acoustic character.\n\nBreathing remains subtle at the camera's distance. Preserve one consistent listening perspective for dialogue, footsteps, clothing, and environmental sounds. Use natural vocal dynamics without heavy compression or artificial vocal enhancement. Maintain uninterrupted ambience before, during, and after the sentence. The man remains silent.\n\nnon_diegetic_music:\nN/A. Let the location sound, pauses, and quiet footsteps carry the tenderness.",
    "promptSourceUrl": "https://x.com/Tomw852/status/2107680267196420347",
    "sourceUrl": "https://x.com/Tomw852/status/2107679859422195837",
    "sourceLabel": "X 原帖 · @Tomw852",
    "author": "Tom𝕎",
    "publishedAt": "2026-10-07T03:50:11.000Z",
    "addedAt": "2026-10-08T01:09:56.063Z",
    "mediaUrl": "/media/x-2107679859422195837.mp4",
    "posterUrl": "/posters/x/2107679859422195837.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1944×1080",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "“ Just a Little Longer ”\n\nA 15s cinematic love story, created with #MiniMaxH3 in #ComfyUI\n\nGPU: RTX 5070 (12GB)\n\n(Model, LoRA & prompt below ⤵️)",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 16,
      "views": 1786,
      "snapshotAt": "2026-10-08T01:09:56.063Z"
    },
    "approvedAt": "2026-10-08T01:09:56.063Z",
    "editorialBasis": "Original H3 creator video plays,15s two-person closeup. Full original creator reply expanded and transcribed. Source explicitly H3 hybrid; do not infer generation mode from Prompt."
  },
  {
    "id": "x-2107531471103938945",
    "title": "恐怖电影把沙发上的两人吓红了脸",
    "titleEn": "A Horror-Movie Scare Turns into an Awkward Blush",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "沙发上的动漫男女抱着爆米花看电视，惊吓后挤到一起，随后红着脸面面相觑。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two anime characters watch television with popcorn, huddle together after a scare, then face each other in an embarrassed blush. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "2D anime | romantic comedy | cel-shaded, clean line art, dark blue night palette.\nMultishot, 6 shots. Cuts only at the specified points.\n\nCHARACTERS\nHANA: wavy shoulder-length dark red hair, oversized gray cardigan with one button missing, a bandage on her index finger. 160 cm.\nYUTO: neat black hair with a small bed-head tuft at the back, navy hoodie with a pilled sleeve, glasses pushed up on his forehead. 178 cm.\n\nLOCATION\nDark living room at night. They share one sofa under a single fleece blanket. A TV off-frame in front of them is the only light source, flickering blue on their faces. A bowl of popcorn between them.\n\nSUBTITLES\nSpoken dialogue is in Japanese. English subtitles are burned into the image as on-screen text only — they are never spoken aloud. Style: white sans-serif, thin black outline, one single line, bottom center of frame, 8% from the bottom edge. Each subtitle appears with its line and disappears at the cut. Subtitle text is shown inside 【】 brackets below; the brackets themselves are not displayed.\n\nShot 1 (0–2.5s) — Medium front shot of both, FOV 47, static.\nHana sits upright, arms crossed, smug smile, chin raised. Yuto beside her grips a cushion with both hands, already tense.\nHana (spoken, Japanese): 「全然怖くないし。」\nOn-screen subtitle: 【I'm not scared at all.】\nHARD CUT\n\nShot 2 (2.5–5s) — Close-up on Yuto, FOV 29, slow push-in ending on his eyes.\nHis pupils shrink to tiny dots, a vertical column of blue shade lines drops over his upper face, sweat drop on his temple. He swallows, the cushion squeezed flat. No subtitle in this shot.\nHARD CUT\n\nShot 3 (5–8s) — Medium front shot, FOV 47, static.\nA violent white flash from the TV and a loud sting. Both jump. Clear gag faces: mouths huge and wavy, eyes white, hair standing in spikes. The popcorn bowl flips and popcorn bursts upward like confetti. Two small white ghost-shaped souls slip out of their open mouths.\nHana and Yuto (spoken, Japanese): 「ぎゃああああっ！」\nOn-screen subtitle: 【AAAAAAH!】\nHARD CUT\n\nShot 4 (8–10.5s) — Wide shot, FOV 63, static.\nThey are clutching each other tightly, trembling, their outlines drawn with shaky wobbling lines. Popcorn rains down on them. The souls drift back into their mouths. No subtitle in this shot.\nHARD CUT\n\nShot 5 (10.5–13s) — Two-shot close-up, profile, FOV 29, static.\nSilence. They slowly realize how close they are, faces a hand apart. Both blush deep red at the same time. A tiny comic sparkle pops between them. They let go and shuffle apart in short jerky movements. No subtitle in this shot.\nHARD CUT\n\nShot 6 (13–15s) — Close-up on Hana, three-quarter front, FOV 29, static.\nNormal detailed anime face: cheeks red, eyes sliding sideways, lips pouting. One popcorn kernel stuck in her hair.\nHana (spoken, Japanese): 「…今のはノーカン。」\nOn-screen subtitle: 【...That one doesn't count.】\nShe pulls the blanket up over her nose. End on her eyes peeking above the blanket.\n\nLIGHTING\nOnly light source is the off-frame TV: flickering cool blue 8500K, with a single white flash in shot 3. Deep shadows in the room.\n\nSound: No score. Production audio only — faint TV dialogue murmur, a sharp horror sting, popcorn scattering, fabric rustle, double scream, then near silence and a soft blanket rustle. All voices in Japanese only.\n\nCharacters keep the same face, hair and wardrobe in every shot. Gag faces are clear, exaggerated super-deformed expressions, then return to the normal detailed anime face. Clean consistent line art, consistent lighting between shots. Subtitles stay in the same position, size and style in every shot, spelled exactly as written. The room contains only the sofa, the blanket and the popcorn bowl.",
    "promptSourceUrl": "https://x.com/MayorKingAI/status/2107531484244627614",
    "sourceUrl": "https://x.com/MayorKingAI/status/2107531471103938945",
    "sourceLabel": "X 原帖 · @MayorKingAI",
    "author": "MayorkingAI",
    "publishedAt": "2026-10-06T18:00:33.000Z",
    "addedAt": "2026-10-08T01:09:56.063Z",
    "mediaUrl": "/media/x-2107531471103938945.mp4",
    "posterUrl": "/posters/x/2107531471103938945.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Men are not scared of horror movies. Scientific fact\n\nThis video proves nothing. It just shows how insane Hailuo H3 is at anime emotions: shrinking pupils, souls leaving the body, synced blushing\n\nLooks like a real anime episode\n\nPrompt below 👇",
    "engagement": {
      "replies": 13,
      "reposts": 3,
      "likes": 40,
      "views": 1764,
      "snapshotAt": "2026-10-08T01:09:56.063Z"
    },
    "approvedAt": "2026-10-08T01:09:56.063Z",
    "editorialBasis": "Original Hailuo H3 declaration; native15s video observed sofa setup and11s blushing two-shot. Full same-author public Prompt expanded; machine translation excluded."
  },
  {
    "id": "x-2107369186121101788",
    "title": "宴会厅大屏传来的停战指令",
    "titleEn": "A Ceasefire Instruction from the Banquet-Hall Screen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金色宴会厅里，男子抬头面对巨型屏幕中的黑衣女子，电影片段以人物对白和混乱后的现场展开。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "In a golden banquet hall, a man faces a black-clad woman on a giant screen as the cinematic scene unfolds through dialogue amid the aftermath of a disturbance. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "添付された人物リファレンスの顔立ち、髪色、髪型、体格、衣装を正確に維持する。\n\n劇場公開用超大作実写映画、Netflix最上位オリジナル映画級、制作費50億円規模。\n\nGAME 05専用の巨大高級宴会場。\n\n乱闘によって椅子が倒れ、テーブルがずれ、料理、皿、ナプキンが床へ散乱している。\n\n15秒のシネマティックシークエンス。\n\n前シーンの直後。\n\n焦茶色センター分け、ブラウン系カーディガン、ストライプシャツ、ワイドスラックス姿のKAIがSPの動きを回避し、巨大モニターを強く見上げる。\n\n呼吸が少し乱れている。\n\nKAI：\n\n「こんなのゲームじゃないだろ！」\n\nVOICE — KAI：\n\n20代前半男性。\n\n柔らかく透明感のある中低音。\n\n緊迫時には声に強い芯が出る。\n\n明瞭な日本語。\n\n全編で同じ声質とピッチを維持。\n\n巨大モニターへ切り替わる。\n\n金髪ロングウェーブ、黒スーツ、黒インナー、シルバークロスアクセサリー姿の女性ゲームマスター。\n\nゲームマスター：\n\n「戦闘は必要ありません。」\n\nVOICE — ゲームマスター：\n\n20代後半〜30代前半女性。\n\n低めのアルト。\n\n冷たく滑らかで落ち着いた声。\n\nゆっくり明瞭。\n\n感情の起伏が非常に小さい。\n\n全編で同じ声質、ピッチ、話速を維持。\n\nその言葉と同時に、宴会場の動きが一斉に止まる。\n\n32mm相当のワイドショット。\n\n黒髪ロングヘア、白黒ストライプ衣装のヤクザ系女性参加者とSPが近距離で停止。\n\nMAELは低い姿勢で呼吸を整えている。\n\nLUCAS、NERO、RION、角がありピンク系衣装姿の愉快犯も動きを止める。\n\n空中を舞っていた白いナプキンだけがゆっくり床へ落ちる。\n\n約1秒の静寂。\n\nゲームマスター：\n\n「SPの前を通る。それだけです。」\n\n全員の視線がSPと、その横にある広い通過スペースへ向く。\n\nKAIへ切り替わる。\n\nKAIは荒れ果てた宴会場を見る。\n\n倒れた椅子。\n\n割れた皿。\n\n散乱した料理。\n\nSPを見る。\n\nもう一度荒れた会場を見る。\n\nKAI：\n\n「……じゃあ俺たち、何で戦ってんの？」\n\n声は怒鳴らず、完全な脱力と素の疑問。\n\n最後はKAIが無言で立っている表情を約1秒残す。\n\nCAMERA:\n\n冒頭KAIは50mm cinema prime相当のmedium close-up。\n\nゲームマスターは85mm相当のportrait close-up。\n\n乱闘停止は32mm相当のwide ensemble shot。\n\n最後は75mm相当でKAIのtight medium close-up。\n\nアクション後からカメラの動きを急激に静かにする。\n\n24fpsの自然な映画モーション。\n\nLIGHTING:\n\n暖かなシャンデリア光。\n\n冷たい白色トップライト。\n\n弱い赤色警告灯が徐々に消えていく。\n\nKAIの焦茶色ヘアには暖かなブラウンの階調。\n\nゲームマスターは冷たいニュートラル光。\n\nAUDIO:\n\nBGMなし。\n\n冒頭はKAIの荒い呼吸、靴音、椅子やグラスが動く音、衣服の擦れ。\n\nゲームマスター：\n\n「戦闘は必要ありません。」\n\nその瞬間に激しい音が一斉に止まる。\n\nその後は低い空調音。\n\n揺れているシャンデリアの微かな金属音。\n\nナプキンが床へ落ちる小さな音。\n\nゲームマスター：\n\n「SPの前を通る。それだけです。」\n\nKAIの最後の台詞には宴会場の自然な残響。\n\n「……じゃあ俺たち、何で戦ってんの？」\n\n台詞後、約1秒の静寂と空調音だけを残して終了。\n\n16:9 cinematic widescreen。",
    "promptSourceUrl": "https://x.com/varts_works/status/2107369189359104119",
    "sourceUrl": "https://x.com/varts_works/status/2107369186121101788",
    "sourceLabel": "X 原帖 · @varts_works",
    "author": "rui｜∫varts (シュバルツ)",
    "publishedAt": "2026-10-06T07:15:41.000Z",
    "addedAt": "2026-10-08T01:09:56.063Z",
    "mediaUrl": "/media/x-2107369186121101788.mp4",
    "posterUrl": "/posters/x/2107369186121101788.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で作る、架空のデスゲーム映画🎬\n◎15秒編集無しノーカット\n\nお気に入りのシーンのひとつ🫶\nモニター越しの音声も、かなりリアルに再現できたと思います👏\n人物は通常速度のまま、降ってくる紙だけをスローに。\n直前まで繰り広げられていた大乱闘の余韻が残るように演出しました！\n\nCreated with: MiniMax Design @Hailuo_AI #MiniMaxH3  Prompt↓参照3枚＋キャラシート",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 21,
      "views": 757,
      "snapshotAt": "2026-10-08T01:09:56.063Z"
    },
    "approvedAt": "2026-10-08T01:09:56.063Z",
    "editorialBasis": "Original Japanese caption restored, explicit H3 creator. Native15s video plays at4s showing giant screen and actor. Full Japanese same-author reply expanded; original text only."
  },
  {
    "id": "x-2107342788178407680",
    "title": "地下停车场里的逃生门",
    "titleEn": "The Exit Door in an Underground Garage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子握着手电穿过昏暗停车场，回头发现危险并奔向出口，短片用近景和追逐镜头制造紧张感。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman crosses a dark garage with a flashlight, notices danger behind her and runs toward an exit in a tense sequence of close-ups and pursuit shots. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second ultra-photorealistic live-action zombie horror movie sequence in 16:9 widescreen. Make it feel like a professionally shot theatrical horror film, with 5 connected shots, escalating tension, strong cinematic sound design and music. Realistic human movement, realistic environments, grounded physics, no comedy, no text, no subtitles, no dialogue. SHOT 1 — 0–3 sec: Night. A deserted underground parking garage. A frightened young woman walks quickly between parked cars, holding a small flashlight. The camera tracks backward in front of her. Suddenly, the flashlight beam catches a motionless person standing far behind her. Deep bass hit. SHOT 2 — 3–6 sec: Close-up of the woman's face as she stops breathing for a moment and looks behind her. The parking space is empty. The music drops almost completely silent. She slowly turns back toward the camera. SHOT 3 — 6–9 sec: A zombie suddenly appears extremely close behind her in the background — pale dirty skin, torn everyday clothing, blood around the mouth, unnatural posture. It moves toward her with a sudden violent burst. Hard music impact. SHOT 4 — 9–12 sec: She runs through the parking garage. Camera follows handheld from behind. Multiple distant zombie silhouettes suddenly emerge between the cars, blocking the exit. Emergency lights begin flashing red. SHOT 5 — 12–15 sec: She reaches the exit door and desperately pulls it open. Bright daylight floods into the dark garage. She looks relieved for half a second — then the camera moves slightly past her shoulder and reveals dozens of zombies already standing silently outside, completely blocking the exit. On the final bass hit, one zombie suddenly looks directly into the camera. CUT TO BLACK. VISUAL STYLE: ultra-photorealistic live-action theatrical zombie horror, cinematic 16:9 composition, realistic human anatomy, realistic zombie makeup and skin texture, physically believable movement, dark atmospheric lighting, practical-looking effects, subtle film grain, dramatic contrast, natural handheld camera movement, high-end horror cinematography. CONSISTENCY & QUALITY: keep the same woman, same clothing, same hairstyle and same underground parking garage throughout the sequence. Zombies must have consistent realistic anatomy. No extra limbs, extra fingers, duplicated people, warped faces, melting bodies, floating objects, cartoon appearance, video-game look, excessive gore, text, subtitles, logos or watermark. EDITING: each shot must have a clear visual purpose and connect naturally to the next. Fast pacing, no dead time. The first shot must immediately create curiosity. Build tension from mystery → realization → zombie reveal → chase → final shocking reveal.",
    "promptSourceUrl": "https://x.com/Just_Anika_Here/status/2107342788178407680",
    "sourceUrl": "https://x.com/Just_Anika_Here/status/2107342788178407680",
    "sourceLabel": "X 原帖 · @Just_Anika_Here",
    "author": "Anika",
    "publishedAt": "2026-10-06T05:30:47.000Z",
    "addedAt": "2026-10-08T01:09:56.064Z",
    "mediaUrl": "/media/x-2107342788178407680.mp4",
    "posterUrl": "/posters/x/2107342788178407680.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The exit was never an exit.\n\nMade with MiniMax H3 on @Hailuo_AI \n\nPrompt :\n\nCreate a 15-second ultra-photorealistic live-action zombie horror movie sequence in 16:9 widescreen. Make it feel like a professionally shot theatrical horror film, with 5 connected shots, escalating tension, strong cinematic sound design and music. Realistic human movement, realistic environments, grounded physics, no comedy, no text, no subtitles, no dialogue.  SHOT 1 — 0–3 sec: Night. A deserted underground parking garage. A frightened young woman walks quickly between parked cars, holding a small flashlight. The camera tracks backward in front of her. Suddenly, the flashlight beam catches a motionless person standing far behind her. Deep bass hit.  SHOT 2 — 3–6 sec: Close-up of the woman's face as she stops breathing for a moment and looks behind her. The parking space is empty. The music drops almost completely silent. She slowly turns back toward the camera.  SHOT 3 — 6–9 sec: A zombie suddenly appears extremely close behind her in the background — pale dirty skin, torn everyday clothing, blood around the mouth, unnatural posture. It moves toward her with a sudden violent burst. Hard music impact.  SHOT 4 — 9–12 sec: She runs through the parking garage. Camera follows handheld from behind. Multiple distant zombie silhouettes suddenly emerge between the cars, blocking the exit. Emergency lights begin flashing red.  SHOT 5 — 12–15 sec: She reaches the exit door and desperately pulls it open. Bright daylight floods into the dark garage. She looks relieved for half a second — then the camera moves slightly past her shoulder and reveals dozens of zombies already standing silently outside, completely blocking the exit.  On the final bass hit, one zombie suddenly looks directly into the camera.  CUT TO BLACK.  VISUAL STYLE: ultra-photorealistic live-action theatrical zombie horror, cinematic 16:9 composition, realistic human anatomy, realistic zombie makeup and skin texture, physically believable movement, dark atmospheric lighting, practical-looking effects, subtle film grain, dramatic contrast, natural handheld camera movement, high-end horror cinematography.  CONSISTENCY & QUALITY: keep the same woman, same clothing, same hairstyle and same underground parking garage throughout the sequence. Zombies must have consistent realistic anatomy. No extra limbs, extra fingers, duplicated people, warped faces, melting bodies, floating objects, cartoon appearance, video-game look, excessive gore, text, subtitles, logos or watermark.  EDITING: each shot must have a clear visual purpose and connect naturally to the next. Fast pacing, no dead time. The first shot must immediately create curiosity. Build tension from mystery → realization → zombie reveal → chase → final shocking reveal.",
    "engagement": {
      "replies": 41,
      "reposts": 4,
      "likes": 49,
      "views": 601,
      "snapshotAt": "2026-10-08T01:09:56.064Z"
    },
    "approvedAt": "2026-10-08T01:09:56.064Z",
    "editorialBasis": "Original full English Prompt and explicit H3 attribution. Native15s video played showing actress/garage/flashlight; no real-world violence claim."
  },
  {
    "id": "x-2107342639209316759",
    "title": "旧书店里的发光小精灵",
    "titleEn": "Little Luminous Spirits in an Old Bookshop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暖色吊灯照亮木质书架和桌上书籍，小小发光生物浮现其间，让安静书店带上奇幻色彩。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Warm hanging lamps illuminate wooden shelves and open books as tiny luminous creatures give the quiet bookshop a touch of fantasy. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Astraea0106/status/2107342639209316759",
    "sourceLabel": "X 原帖 · @Astraea0106",
    "author": "Astraea",
    "publishedAt": "2026-10-06T05:30:12.000Z",
    "addedAt": "2026-10-08T01:12:20.167Z",
    "mediaUrl": "/media/x-2107342639209316759.mp4",
    "posterUrl": "/posters/x/2107342639209316759.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Little library creatures \n\nMake with @picsartai Mini Max H3 model\nPrompt below 👇 \n\n@picsartai #picsartpartner",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 97,
      "snapshotAt": "2026-10-08T01:12:20.167Z"
    },
    "approvedAt": "2026-10-08T01:12:20.167Z",
    "editorialBasis": "Native15.2s creator video observed luminous spirit on shelf/books. Explicit H3 creator attribution. Reply explicitly says Partial prompt, so complete Prompt remains null."
  },
  {
    "id": "x-2106698801604186579",
    "title": "黑色镜面上的双剑决斗",
    "titleEn": "A Sword Duel on a Liquid-Black Mirror",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白色虚空中，黑色镜面浮起两名持剑角色，反光、烟雾和碎片围绕他们的交锋展开。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two sword-wielding figures emerge on a black mirror in a white void, with reflections, smoke and fragments surrounding their clash. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "MINIMAX H3 — 15-SECOND CINEMATIC FIGHT\n“BLACK MIRROR: FINAL REFLECTION”\n16:9 widescreen, premium dark-fantasy fashion-action.\n\nObsidian Ronin vs Inkshade Valkyrie inside a gigantic floating liquid-black obsidian mirror surrounded by an endless white void. Preserve exact established character designs, faces, hair, armor, weapons, proportions and silhouettes. No redesigns.\n\nConstant movement. No idle poses, hard cuts, teleportation, freeze frames, blood or gore.\n\nThe mirror behaves like liquid glass: every sword strike creates ripples, cracks and independent reflections that briefly attack before snapping back.\n\n0–3s: Mirror forms from liquid black glass. Ronin emerges with a katana strike as Valkyrie attacks from the opposite reflection. Blades collide, creating a huge circular ripple.\n\n3–6s: Ronin's heavy cut meets Valkyrie's agile twin-blade counter. Reflections attack from opposite directions. Black cracks race across the mirror. Smoke briefly forms “REAL OR REFLECTION?”\n\n6–9s: Mirror tilts vertical. Both fighters sprint across its surface, exchanging rapid sword strikes. The glass bends around them like liquid.\n\n9–12s: Mirror explodes into floating black fragments. Fighters leap between reflections, destroying them. Fragments align into a giant blade shape. Smoke forms “BREAK THE IMAGE”, then shatters.\n\n12–15s: Final charge. Ronin's massive diagonal katana strike collides with Valkyrie's spinning twin-blade cross attack. The entire mirror shatters into thousands of black particles. Silence. Final smoke typography: “ONLY ONE REMAINS.”\n\nElegant black-and-white luxury aesthetic, glossy reflections, silver highlights, charcoal smoke, cinematic motion blur, precise camera orbiting, low mirror tracking, rotating blade shots and dramatic final push.\n\n15s | 16:9 | continuous action | premium cinematic dark fantasy | exact character consistency",
    "promptSourceUrl": "https://x.com/itxsarmadd/status/2106698801604186579",
    "sourceUrl": "https://x.com/itxsarmadd/status/2106698801604186579",
    "sourceLabel": "X 原帖 · @itxsarmadd",
    "author": "Sarmad Tahir",
    "publishedAt": "2026-10-04T10:51:49.000Z",
    "addedAt": "2026-10-08T01:12:20.167Z",
    "mediaUrl": "/media/x-2106698801604186579.mp4",
    "posterUrl": "/posters/x/2106698801604186579.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cg-3d",
      "stylized"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created With Minimax H3 on @Hailuo_AI\n\nPrompt:\n\nMINIMAX H3 — 15-SECOND CINEMATIC FIGHT\n“BLACK MIRROR: FINAL REFLECTION”\n16:9 widescreen, premium dark-fantasy fashion-action.\n\nObsidian Ronin vs Inkshade Valkyrie inside a gigantic floating liquid-black obsidian mirror surrounded by an endless white void. Preserve exact established character designs, faces, hair, armor, weapons, proportions and silhouettes. No redesigns.\n\nConstant movement. No idle poses, hard cuts, teleportation, freeze frames, blood or gore.\n\nThe mirror behaves like liquid glass: every sword strike creates ripples, cracks and independent reflections that briefly attack before snapping back.\n\n0–3s: Mirror forms from liquid black glass. Ronin emerges with a katana strike as Valkyrie attacks from the opposite reflection. Blades collide, creating a huge circular ripple.\n\n3–6s: Ronin's heavy cut meets Valkyrie's agile twin-blade counter. Reflections attack from opposite directions. Black cracks race across the mirror. Smoke briefly forms “REAL OR REFLECTION?”\n\n6–9s: Mirror tilts vertical. Both fighters sprint across its surface, exchanging rapid sword strikes. The glass bends around them like liquid.\n\n9–12s: Mirror explodes into floating black fragments. Fighters leap between reflections, destroying them. Fragments align into a giant blade shape. Smoke forms “BREAK THE IMAGE”, then shatters.\n\n12–15s: Final charge. Ronin's massive diagonal katana strike collides with Valkyrie's spinning twin-blade cross attack. The entire mirror shatters into thousands of black particles. Silence. Final smoke typography: “ONLY ONE REMAINS.”\n\nElegant black-and-white luxury aesthetic, glossy reflections, silver highlights, charcoal smoke, cinematic motion blur, precise camera orbiting, low mirror tracking, rotating blade shots and dramatic final push.\n\n15s | 16:9 | continuous action | premium cinematic dark fantasy | exact character consistency",
    "engagement": {
      "replies": 54,
      "reposts": 11,
      "likes": 88,
      "views": 1968,
      "snapshotAt": "2026-10-08T01:12:20.167Z"
    },
    "approvedAt": "2026-10-08T01:12:20.167Z",
    "editorialBasis": "Original explicit H3 and full public Prompt. Native15s video plays with figures rising from black liquid disk; visual action and monochrome fantasy checked."
  },
  {
    "id": "x-2106409582222262478",
    "title": "石砌庭院里隐入暗影的游侠",
    "titleEn": "A Cloaked Ranger Vanishing into Shadow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "阴冷石砌庭院中，披斗篷的男子直视镜头，转身拉起兜帽，随后在暗色雾气中消散。 原生视频规格为 9 秒 · 768×960。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A cloaked man faces the camera in a cold stone courtyard, raises his hood and turns away before dissolving into dark mist. Native-video output: 9s · 768×960. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "f/2.8 lens, eye level, static camera in Ravenholdt courtyard. Damp stone courtyard, low mountain fog. From 0s to 2s, subject holds unblinking eye contact. At 2s, he swiftly pulls his dark wool hood over his head and turns his back to camera. At 3.5s, as his cloak sweeps, his physical body rapidly dissolves and evaporates into a brief wisp of dark shadow-mist, vanishing completely by second 5, leaving the damp stone courtyard entirely empty for the final second. TEXTURES AND VFX: Optical body dissolve, realistic dark mist evaporation, fluid wool physics, organic sensor grain. Quick cloth rustle, boot pivot, brief wind gust, fading rain ambience.",
    "promptSourceUrl": "https://x.com/holagarciajavi/status/2106409582222262478",
    "sourceUrl": "https://x.com/holagarciajavi/status/2106409582222262478",
    "sourceLabel": "X 原帖 · @holagarciajavi",
    "author": "Airem Scry",
    "publishedAt": "2026-10-03T15:42:34.000Z",
    "addedAt": "2026-10-08T01:12:20.167Z",
    "mediaUrl": "/media/x-2106409582222262478.mp4",
    "posterUrl": "/posters/x/2106409582222262478.jpg",
    "duration": 9,
    "aspectRatio": "portrait",
    "resolution": "768×960",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing Character Lock & Dark Fantasy VFX in\n@MiniMax_AI Minimax H3 video model. \nAirem Scry executing Vanish.  \n\nprompt:  \n\nf/2.8 lens, eye level, static camera in Ravenholdt courtyard. Damp stone courtyard, low mountain fog. From 0s to 2s, subject holds unblinking eye contact. At 2s, he swiftly pulls his dark wool hood over his head and turns his back to camera. At 3.5s, as his cloak sweeps, his physical body rapidly dissolves and evaporates into a brief wisp of dark shadow-mist, vanishing completely by second 5, leaving the damp stone courtyard entirely empty for the final second.  TEXTURES AND VFX: Optical body dissolve, realistic dark mist evaporation, fluid wool physics, organic sensor grain. Quick cloth rustle, boot pivot, brief wind gust, fading rain ambience.  \n\n#WoWforever #rogue #darkfantasy",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 217,
      "snapshotAt": "2026-10-08T01:12:20.167Z"
    },
    "approvedAt": "2026-10-08T01:12:20.167Z",
    "editorialBasis": "Native9.088s video plays, cloak/actor courtyard visible. Original explicit H3 and complete Prompt; actualdurationnotPrompt6s. Latest edited public version used."
  },
  {
    "id": "x-2106398901267812627",
    "title": "火焰熔铸出的 BURN 字幕",
    "titleEn": "BURN Forged from an Inferno",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "燃料火花迅速形成火焰漩涡，灼热金属字母从中心浮现，最后在爆炸与碎片中冲向镜头。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An ignition grows into a fiery vortex, forging hot metal lettering that bursts toward the camera in an explosion of fragments. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Extreme Hollywood pyro title sequence: begin inside a microscopic droplet of fuel suspended in darkness as a spark suddenly ignites its surface. Fire spreads explosively outward and the camera rockets backward through a chain reaction of enormous fuel explosions. Liquid fire whips across the frame like glowing ribbons while burning debris and sparks spiral into a rapidly expanding fire tornado. The camera enters the vortex and rotates around a white-hot molten core where streams of liquid metal begin violently forging the gigantic word \"BURN\". Every letter emerges incandescent from the inferno before cooling into cracked black steel while magma remains visible beneath the surface. The completed title slams into frame with enormous weight as flames blast through the gaps between the letters. Pressure suddenly builds inside the typography until \"BURN\" violently ruptures from within, unleashing a radial explosion of molten metal, sparks and fire directly into the lens. Extreme heat distortion, realistic pyro physics, aggressive camera movement, AAA theatrical VFX.",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2106398901267812627",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2106398901267812627",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-10-03T15:00:07.000Z",
    "addedAt": "2026-10-08T01:12:20.167Z",
    "mediaUrl": "/media/x-2106398901267812627.mp4",
    "posterUrl": "/posters/x/2106398901267812627.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics",
      "cg-3d"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The more I use Minimax H3 for motion design, the more I want to push it.\nNot just revealing typography…\n\nmaking the entire environment CREATE the typography.\nThis BURN experiment is a perfect example. 🔥\nFull prompt below 👇\nExtreme Hollywood pyro title sequence: begin inside a microscopic droplet of fuel suspended in darkness as a spark suddenly ignites its surface. Fire spreads explosively outward and the camera rockets backward through a chain reaction of enormous fuel explosions. Liquid fire whips across the frame like glowing ribbons while burning debris and sparks spiral into a rapidly expanding fire tornado. The camera enters the vortex and rotates around a white-hot molten core where streams of liquid metal begin violently forging the gigantic word \"BURN\". Every letter emerges incandescent from the inferno before cooling into cracked black steel while magma remains visible beneath the surface. The completed title slams into frame with enormous weight as flames blast through the gaps between the letters. Pressure suddenly builds inside the typography until \"BURN\" violently ruptures from within, unleashing a radial explosion of molten metal, sparks and fire directly into the lens. Extreme heat distortion, realistic pyro physics, aggressive camera movement, AAA theatrical VFX.",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 34,
      "views": 2121,
      "snapshotAt": "2026-10-08T01:12:20.167Z"
    },
    "approvedAt": "2026-10-08T01:12:20.167Z",
    "editorialBasis": "Original H3 motion-design creator caption and full Prompt. Native15s video plays; ignition/pyro opening observed and originaltypographydescription."
  },
  {
    "id": "x-2106337250346664007",
    "title": "高尔夫球场上的清爽饮料短片",
    "titleEn": "A Refreshing Drink Spot on the Golf Course",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "晴朗球场上，红发女子喝下冰凉饮料、挥杆击球，再以饮料罐和高尔夫球的产品近景收尾。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "On a sunny course, a red-haired golfer sips a cold drink and swings her club, with the spot closing on a can and golf ball. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "0–2s — THE SIP\nFast close-up of the ginger-haired golfer on a sunny golf course. She grabs the Sprite can, cracks it open — PSHHHT! — condensation and tiny droplets flying in slow motion. She takes a refreshing sip and gives a confident little smirk.\n2–4s — GAME ON\nHard cut. She tosses the Sprite can slightly upward, catches it, grabs her golf club, and steps into position. Quick whip-pan around her as sunlight flashes across the frame.\n\n4–7s — THE SWING\nRapid montage: close-up of her gripping the club → golf ball on the tee → her eyes locking onto the ball → skirt and ponytail moving as she swings. She hits the ball with a powerful, perfectly timed CRACK!\n7–10s — FOLLOW THE BALL\nUltra-fast camera follows the golf ball flying across the bright green course. Speed ramp as it passes trees and sunlight, then lands perfectly near the hole.\n\n10–12s — HOLE IN ONE\nShe watches it drop into the hole. Quick reaction shot: she smiles, spins the club casually, and walks toward the camera like she absolutely knew that was happening.\n12–15s — HERO SHOT\nShe grabs the cold Sprite can again and takes another sip. Camera pushes in as she lowers the can, smiling confidently. Condensation sparkles in the sunlight. End on a clean product close-up of the Sprite can beside the golf ball.\n\nOn-screen text:\n“REFRESH. RELOAD. PLAY.”\nSPRITE — THIRSTY FOR MORE.\nFast cuts, energetic commercial pacing, whip transitions, speed ramps, crisp summer lighting, realistic golf physics, dynamic camera movement, refreshing splash effects, premium sports-ad aesthetic.",
    "promptSourceUrl": "https://x.com/sophiaparkerr_/status/2106337250346664007",
    "sourceUrl": "https://x.com/sophiaparkerr_/status/2106337250346664007",
    "sourceLabel": "X 原帖 · @sophiaparkerr_",
    "author": "Sophia",
    "publishedAt": "2026-10-03T10:55:08.000Z",
    "addedAt": "2026-10-08T01:12:20.167Z",
    "mediaUrl": "/media/x-2106337250346664007.mp4",
    "posterUrl": "/posters/x/2106337250346664007.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "photoreal"
    ],
    "scenes": [
      "product",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A little refreshment, a little golf, and a whole lot of main-character energy. 🍋⛳✨\n\nMade with MiniMax H3 on @FishCreativeHQ \n \nPROMPT:\n\n0–2s — THE SIP\nFast close-up of the ginger-haired golfer on a sunny golf course. She grabs the Sprite can, cracks it open — PSHHHT! — condensation and tiny droplets flying in slow motion. She takes a refreshing sip and gives a confident little smirk.\n2–4s — GAME ON\nHard cut. She tosses the Sprite can slightly upward, catches it, grabs her golf club, and steps into position. Quick whip-pan around her as sunlight flashes across the frame.\n\n4–7s — THE SWING\nRapid montage: close-up of her gripping the club → golf ball on the tee → her eyes locking onto the ball → skirt and ponytail moving as she swings. She hits the ball with a powerful, perfectly timed CRACK!\n7–10s — FOLLOW THE BALL\nUltra-fast camera follows the golf ball flying across the bright green course. Speed ramp as it passes trees and sunlight, then lands perfectly near the hole.\n\n10–12s — HOLE IN ONE\nShe watches it drop into the hole. Quick reaction shot: she smiles, spins the club casually, and walks toward the camera like she absolutely knew that was happening.\n12–15s — HERO SHOT\nShe grabs the cold Sprite can again and takes another sip. Camera pushes in as she lowers the can, smiling confidently. Condensation sparkles in the sunlight. End on a clean product close-up of the Sprite can beside the golf ball.\n\nOn-screen text:\n“REFRESH. RELOAD. PLAY.”\nSPRITE — THIRSTY FOR MORE.\nFast cuts, energetic commercial pacing, whip transitions, speed ramps, crisp summer lighting, realistic golf physics, dynamic camera movement, refreshing splash effects, premium sports-ad aesthetic.",
    "engagement": {
      "replies": 42,
      "reposts": 27,
      "likes": 303,
      "views": 36344,
      "snapshotAt": "2026-10-08T01:12:20.167Z"
    },
    "approvedAt": "2026-10-08T01:12:20.167Z",
    "editorialBasis": "Original H3 creator declaration, full public Prompt extracted excluding translation. Native15s video plays and drinks closeup observed. Spec ad, no official brand campaign claim."
  },
  {
    "id": "x-2102280037286527349",
    "title": "夜院剑舞与江畔回眸",
    "titleEn": "Courtyard Swordplay and a Riverside Glance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白衣人物在古式庭院中跃起交手，后续镜头转向江畔女子的近景，呈现动作与面部细节的连续片段。 原生视频规格为 77 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "White-robed figures leap and duel in a traditional courtyard, followed by a close-up of a woman beside a hazy river, combining action and facial detail. Native-video output: 77s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sunbaolong_2001/status/2102280037286527349",
    "sourceLabel": "X 原帖 · @sunbaolong_2001",
    "author": "有趣的80后程序员",
    "publishedAt": "2026-09-22T06:13:14.000Z",
    "addedAt": "2026-10-08T17:04:34.031Z",
    "mediaUrl": "/media/x-2102280037286527349.mp4",
    "posterUrl": "/posters/x/2102280037286527349.jpg",
    "duration": 77,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "5-second MiniMax H3 video in just 41 SECONDS! 🚀⚡\nFastH3 V2 by FastVideo is a beast! It not only cuts render times to 41s (8 steps), but it also completely fixes the annoying \"floating slow-motion\" bug in action scenes!\nI built an all-in-one ComfyUI workflow that solves motion blur:\n1️⃣ FastH3 V2 INT8 + VSA Attention (Snappy combat pacing)\n2️⃣ 2-Stage 3D Latent Upscale (608x352 ➔ 1664x928)\n3️⃣ Dual-pass face fix (YOLOv8 + H3 Latent Inpaint + VOSR 2.0)\nFull breakdown & Compshare / RunningHub workflows below: 👇\n🔗 https://youtu.be/KvPIYTDQfP8\n#FastH3 #MiniMaxH3 #ComfyUI #AIVideo #GenerativeAI",
    "engagement": {
      "replies": 9,
      "reposts": 5,
      "likes": 104,
      "views": 7844,
      "snapshotAt": "2026-10-08T17:04:34.031Z"
    },
    "approvedAt": "2026-10-08T17:04:34.031Z",
    "editorialBasis": "Original caption explicitly identifies H3; native montage observed playing at 26.309 and 75.869 seconds, with courtyard combat and riverside close-up."
  },
  {
    "id": "x-2108235478474637403",
    "title": "南瓜灯间的星帽舞步",
    "titleEn": "Starry Witch Dance Among Jack-o'-Lanterns",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "棕发动漫魔女穿着紫色星纹服装，在南瓜灯与幽灵装饰之间伸展双臂、踏步起舞。 原生视频规格为 15 秒 · 1080×1888。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A brown-haired anime witch in a purple star-patterned outfit extends her arms and dances among glowing jack-o'-lanterns and ghost decorations. Native-video output: 15s · 1080×1888. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2108235478474637403",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-10-08T16:38:01.000Z",
    "addedAt": "2026-10-08T17:04:34.032Z",
    "mediaUrl": "/media/x-2108235478474637403.mp4",
    "posterUrl": "/posters/x/2108235478474637403.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1888",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIart #MiniMaxH3MaxTurbo #Capcut\nハロウィン2。 （MiniMax H3 Max Turbo 出力動画＋CapCutの画質向上（AIHD）。1080p画質で再生可能）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 37,
      "snapshotAt": "2026-10-08T17:04:34.032Z"
    },
    "approvedAt": "2026-10-08T17:04:34.032Z",
    "editorialBasis": "Original caption identifies H3 Max Turbo; native video observed at 5.312 and 12.817 seconds with a distinct brown-haired dancing character."
  },
  {
    "id": "x-2108235366151176241",
    "title": "酒吧邂逅与烛光愿望",
    "titleEn": "A Bar Encounter and a Candlelit Wish",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "开场中一位金发女子与男子在酒吧相对而坐，后续转到幽暗客厅里被烛光包围的跪坐人物，以音乐串联不同场景。 原生视频规格为 83 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde woman and a man face each other at a bar, before the film shifts to a kneeling figure surrounded by candles in a dim living room, linking the scenes through music. Native-video output: 83s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/alexutopia/status/2108235366151176241",
    "sourceLabel": "X 原帖 · @alexutopia",
    "author": "Alex Utopia",
    "publishedAt": "2026-10-08T16:37:35.000Z",
    "addedAt": "2026-10-08T17:04:34.032Z",
    "mediaUrl": "/media/x-2108235366151176241.mp4",
    "posterUrl": "/posters/x/2108235366151176241.jpg",
    "duration": 83,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Are you certain that's what you want?\"\n\nMusic: Suno v6\nVideo: MiniMax H3",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 15,
      "views": 370,
      "snapshotAt": "2026-10-08T17:04:34.032Z"
    },
    "approvedAt": "2026-10-08T17:04:34.032Z",
    "editorialBasis": "Original author labels video MiniMax H3 and music Suno v6; playback observed at 19.307 and 58.806 seconds and completed at 83.151 seconds."
  },
  {
    "id": "x-2108234653866082618",
    "title": "幽灵夜宴的紫发魔女",
    "titleEn": "Lavender-Haired Witch at a Ghostly Feast",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫发动漫魔女戴着星月装饰的尖帽，在幽灵与南瓜灯的背景前抬手摆出姿态，与另一条棕发角色的舞蹈片段不同。 原生视频规格为 15 秒 · 1080×1888。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A lavender-haired anime witch wearing a star-and-moon hat raises her hand against a backdrop of ghosts and jack-o'-lanterns, with distinct shots from the creator's brown-haired dance clip. Native-video output: 15s · 1080×1888. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2108234653866082618",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-10-08T16:34:45.000Z",
    "addedAt": "2026-10-08T17:04:34.032Z",
    "mediaUrl": "/media/x-2108234653866082618.mp4",
    "posterUrl": "/posters/x/2108234653866082618.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1888",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIart #MiniMaxH3MaxTurbo #Capcut\nハロウィン1。 （MiniMax H3 Max Turbo 出力動画＋CapCutの画質向上（AIHD）。1080p画質で再生可能）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 42,
      "snapshotAt": "2026-10-08T17:04:34.032Z"
    },
    "approvedAt": "2026-10-08T17:04:34.032Z",
    "editorialBasis": "Original caption identifies H3 Max Turbo; native playback advanced through 9.392 and 12.593 seconds, with a distinct lavender-haired character."
  },
  {
    "id": "x-2108231420858810830",
    "title": "皮影月宫与流云仙袖",
    "titleEn": "Shadow Puppets Beneath the Moon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镂刻皮影人物捧起发光珠子，在流云与飘动衣袖间升起，随后与兔形剪影一同出现在圆月前。 原生视频规格为 15 秒 · 1376×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An intricately carved shadow-puppet figure lifts a glowing pearl, rises through swirling clouds with flowing sleeves, and appears before a full moon alongside rabbit silhouettes. Native-video output: 15s · 1376×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Traditional Chinese shadow puppet theater (piying), intricately carved jointed leather silhouettes on a translucent silk screen, warm amber lamp glow.\nA graceful female silhouette in flowing layered robes swallows a glowing elixir pearl; her robes unfurl into long rippling ribbon shapes as she lifts off the ground and floats upward past stylized swirling clouds; a round luminous full moon fills the upper screen;\nshe drifts onto it and a small jade rabbit silhouette hops to her side; the two sit together as the moonlight washes the whole screen silver-blue. Camera: slow tilt-up following her ascent, ending framed on the moon. Audio: ethereal guzheng glissando rising with her flight,\na soft choir-like hum as the moon brightens, one delicate chime as the rabbit appears. Crisp jointed leather silhouettes, flowing silk-robe shapes, woven screen texture, no realistic faces, no text at all.",
    "promptSourceUrl": "https://x.com/7u86j0Tf5Y3z7Ii/status/2108231739026129006",
    "sourceUrl": "https://x.com/7u86j0Tf5Y3z7Ii/status/2108231420858810830",
    "sourceLabel": "X 原帖 · @7u86j0Tf5Y3z7Ii",
    "author": "qiuqiu.ai.lab",
    "publishedAt": "2026-10-08T16:21:54.000Z",
    "addedAt": "2026-10-08T17:04:34.032Z",
    "mediaUrl": "/media/x-2108231420858810830.mp4",
    "posterUrl": "/posters/x/2108231420858810830.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n\nTraditional Chinese shadow puppet theater (piying)\n\nPrompt below ⤵️",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 0,
      "views": 20,
      "snapshotAt": "2026-10-08T17:04:34.032Z"
    },
    "approvedAt": "2026-10-08T17:04:34.032Z",
    "editorialBasis": "Native video viewed at 3.281 and 10.157 seconds; original H3 post and all four consecutive same-author Prompt replies read in full."
  },
  {
    "id": "x-2108229663730639134",
    "title": "奶油甜点的羞涩瞬间",
    "titleEn": "A Bashful Moment Over Cream Dessert",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "浅发动漫角色系着绿色条纹领带，在餐桌前露出羞涩表情，随后镜头贴近脸部与擦拭唇边奶油的手指。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pale-haired anime character in a green striped tie blushes at a table, followed by a tighter shot of a fingertip wiping cream from the mouth. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI_EmeraldApple/status/2108229663730639134",
    "sourceLabel": "X 原帖 · @AI_EmeraldApple",
    "author": "Emerald Apple",
    "publishedAt": "2026-10-08T16:14:55.000Z",
    "addedAt": "2026-10-08T17:04:34.032Z",
    "mediaUrl": "/media/x-2108229663730639134.mp4",
    "posterUrl": "/posters/x/2108229663730639134.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "What's got Malfoid all shy, hot, and bothered? Potter, what did you do?\n\nI tried to make the self-generated audio work, but instead used a different AI program to add music. \n\nOne of the weaknesses of Minimax H3 is the audio. The audio by itself sounded like someone stuck their entire hand into a jar of mayo.",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 75,
      "views": 878,
      "snapshotAt": "2026-10-08T17:04:34.032Z"
    },
    "approvedAt": "2026-10-08T17:04:34.032Z",
    "editorialBasis": "Original author identifies own H3 output and audio replacement; native animation observed at 6.417 and 10.149 seconds."
  },
  {
    "id": "x-2108228392940097833",
    "title": "窗边开关切换昼夜",
    "titleEn": "A Window Switch Between Day and Night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "固定镜头对准室内墙面与窗户，窗外原本明亮的阳光田野转为星空与弯月，室内随之变暗。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A locked camera faces an interior wall and window; the bright sunlit fields outside shift to a starry sky with a crescent moon as the room darkens. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the daylight outside the window beside it changes at once",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2108228392940097833",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108228392940097833",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-08T16:09:52.000Z",
    "addedAt": "2026-10-08T17:04:34.033Z",
    "mediaUrl": "/media/x-2108228392940097833.mp4",
    "posterUrl": "/posters/x/2108228392940097833.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Found the switch for outside.\n\nPrompt: \"the daylight outside the window beside it changes at once\"\n\nDuration: 15s. Camera: locked, facing the wall. Rendered locally in ComfyUI, then Topaz to 4K.\n\nTwo clicks at stated seconds. The crickets between them do the work of saying it's night.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 123,
      "snapshotAt": "2026-10-08T17:04:34.033Z"
    },
    "approvedAt": "2026-10-08T17:04:34.033Z",
    "editorialBasis": "Original post explicitly provides the complete short Prompt; native video viewed at 2.963 and 11.078 seconds with visible daylight-to-night change."
  },
  {
    "id": "x-2108218646954668254",
    "title": "红色手套与白衣骏马",
    "titleEn": "Red Gloves and a White-Coated Equestrian",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "时尚片段先以白色宽檐帽、红色手套与妆容近景构图，再转到白衣人物牵着棕色骏马走过浅色地面的场景。 原生视频规格为 22 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Fashion compositions move from a close-up framed by a white wide-brimmed hat and red gloves to a white-coated figure walking a brown horse across pale ground. Native-video output: 22s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/koss_ai/status/2108218646954668254",
    "sourceLabel": "X 原帖 · @koss_ai",
    "author": "koss",
    "publishedAt": "2026-10-08T15:31:08.000Z",
    "addedAt": "2026-10-08T17:04:34.033Z",
    "mediaUrl": "/media/x-2108218646954668254.mp4",
    "posterUrl": "/posters/x/2108218646954668254.jpg",
    "duration": 22,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "photoreal"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Red Issue.\n\nA conceptual fashion film built from a series of visual ideas rather than a traditional commercial narrative.\n\nBold red, high-fashion imagery, surreal spaces and a slightly dramatic editorial mood.\n\nTOOLS USED:\n@Hailuo_AI MiniMax H3, @OpenAI GPT-Image 2.5, Stable Audio 3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 43,
      "snapshotAt": "2026-10-08T17:04:34.033Z"
    },
    "approvedAt": "2026-10-08T17:04:34.033Z",
    "editorialBasis": "Original caption explicitly lists H3 for the conceptual fashion film; native 22.293-second film played and looped, with both described compositions viewed."
  },
  {
    "id": "x-2108211569616375924",
    "title": "暮海浪花与随风黑发",
    "titleEn": "Sunset Sea Spray and Windswept Hair",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发尖耳角色穿着衬衫、领带与外套站在暮色海面前，发丝随风扬起，身旁浪花不断溅起。 原生视频规格为 9 秒 · 544×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired, pointed-eared character in a shirt, tie and jacket stands before the sunset sea as wind moves the hair and waves throw up spray nearby. Native-video output: 9s · 544×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EmpJellok/status/2108211569616375924",
    "sourceLabel": "X 原帖 · @EmpJellok",
    "author": "JellokEmp",
    "publishedAt": "2026-10-08T15:03:01.000Z",
    "addedAt": "2026-10-08T17:04:34.034Z",
    "mediaUrl": "/media/x-2108211569616375924.mp4",
    "posterUrl": "/posters/x/2108211569616375924.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "544×384",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 handled the water pretty well. Do not know what caused the sound glitch. My OC Pitch wanted to test the water and hair psychics. #OC #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 47,
      "snapshotAt": "2026-10-08T17:04:34.035Z"
    },
    "approvedAt": "2026-10-08T17:04:34.034Z",
    "editorialBasis": "Original author identifies the character as their OC and video as H3; native playback observed at 5.429 and 8.893 seconds of 9.28 seconds."
  },
  {
    "id": "x-2108207456858059129",
    "title": "街头分身的左右对照",
    "titleEn": "A Side-by-Side Street Doppelganger",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右并列的动漫街景中，黑衣女孩的身形与重复人物呈现不同效果。 原生视频规格为 3 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two anime street shots compare the black-clad girl's form and an unwanted duplicate figure. Native-video output: 3s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HitotsumeKuro/status/2108207456858059129",
    "sourceLabel": "X 原帖 · @HitotsumeKuro",
    "author": "1つ目クロスケ",
    "publishedAt": "2026-10-08T14:46:40.000Z",
    "addedAt": "2026-10-08T17:14:26.974Z",
    "mediaUrl": "/media/x-2108207456858059129.mp4",
    "posterUrl": "/posters/x/2108207456858059129.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【MiniMax H3メモ】\nこのふたつのアニメはseed、参照動画、作り方、コマ数、キャラの見本はすべて同じです。ですが左は分身が出来てしまいました。ガチャでは無いです。\n\n答えは恐らくプロンプト。参照の画像やプリビズを与えていたのにさらにその情報を細かく言葉で載せたことで同じ女の子がもう1人生まれた形。紐解くのに時間かかるぜぃ。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-10-08T17:14:26.975Z"
    },
    "approvedAt": "2026-10-08T17:14:26.974Z",
    "editorialBasis": "Observed native 3.04-second side-by-side NG/OK comparison; original author explicitly names H3."
  },
  {
    "id": "x-2108204486413971912",
    "title": "林间拉力赛车的四路对照",
    "titleEn": "Four Takes on a Forest Rally Car",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "低角度车轮特写与林间漂移镜头组成四模型对照，第四段明确标注 H3。 原生视频规格为 43 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Wheel close-ups and forest drifts appear in a four-model comparison, with H3 explicitly labeled in the fourth segment. Native-video output: 43s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/attelier_ai/status/2108204486413971912",
    "sourceLabel": "X 原帖 · @attelier_ai",
    "author": "Attelier",
    "publishedAt": "2026-10-08T14:34:52.000Z",
    "addedAt": "2026-10-08T17:14:26.975Z",
    "mediaUrl": "/media/x-2108204486413971912.mp4",
    "posterUrl": "/posters/x/2108204486413971912.jpg",
    "duration": 43,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "We compared Seedance 2.5, Kling 3, Wan 3.0 and MiniMax H3 using the same prompt.\n\nMiniMax H3, being the only open-source model, stood out as the fastest and lowest-cost option in this test at 3.9 seconds and $0.13 per generation. It may not have the highest quality, but it’s the best for iterating quickly and affordably making it much more useful in production.\n\nIt’s still early but the trend is clear. Open-source video models will keep closing the gap with proprietary ones and Attelier is positioned for that future. $TAO",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 242,
      "snapshotAt": "2026-10-08T17:14:26.975Z"
    },
    "approvedAt": "2026-10-08T17:14:26.975Z",
    "editorialBasis": "Watched native 43-second video including MINIMAX H3 04/04 title card and rally car shots."
  },
  {
    "id": "x-2108200474210156952",
    "title": "旧教室里的警戒回望",
    "titleEn": "An Alert Glance in an Old Classroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "昏暗教室里，穿校服的人物与手持长柄武器的女孩构成紧张的真人化片段。 原生视频规格为 14 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Uniformed figures and a girl holding a long dark weapon create a tense live-action-style classroom scene. Native-video output: 14s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/photo_b1/status/2108200474210156952",
    "sourceLabel": "X 原帖 · @photo_b1",
    "author": "BSM 제작가",
    "publishedAt": "2026-10-08T14:18:56.000Z",
    "addedAt": "2026-10-08T17:14:26.977Z",
    "mediaUrl": "/media/x-2108200474210156952.mp4",
    "posterUrl": "/posters/x/2108200474210156952.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "학원묵시록 HIGHSCHOOL OF THE DEAD 실사 연출.\n최근 웹게임으로 자주 보여서 한번 플레이 해보고\n재미삼아 이 주제로 로컬 상태에서 ai 활용\n만들어 봤습니다.\n\ncompyui - minimax h3.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 45,
      "snapshotAt": "2026-10-08T17:14:26.977Z"
    },
    "approvedAt": "2026-10-08T17:14:26.977Z",
    "editorialBasis": "Viewed native 13.76-second clip at opening and later frame; author explicitly states local H3 adaptation."
  },
  {
    "id": "x-2108188120130744793",
    "title": "桌面上的行星牛顿摆",
    "titleEn": "A Planetary Newton's Cradle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "五颗小行星悬挂在桌面框架中，边缘球体摆动，中央球体保持排列。 原生视频规格为 15 秒 · 3456×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Five small planets hang inside a desktop frame as the outer spheres swing beside an aligned center. Native-video output: 15s · 3456×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108188120130744793",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-08T13:29:50.000Z",
    "addedAt": "2026-10-08T17:14:26.977Z",
    "mediaUrl": "/media/x-2108188120130744793.mp4",
    "posterUrl": "/posters/x/2108188120130744793.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Five planets on strings. Perfect time.\n\nPrompt: \"its five hanging balls are five small planets\"\n\nDuration: 15s. Camera: locked, desk height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe first takes swung both ends up together. The prompt now says only one planet is ever in the air, and gives the left the odd seconds and the right the even ones.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 6,
      "views": 333,
      "snapshotAt": "2026-10-08T17:14:26.977Z"
    },
    "approvedAt": "2026-10-08T17:14:26.977Z",
    "editorialBasis": "Viewed native 15.125-second clip with moving planet spheres; quoted Prompt is incomplete because additional constraints are described."
  },
  {
    "id": "x-2108184725378711631",
    "title": "雪山樱花与石间溪流",
    "titleEn": "Cherry Blossoms Beneath a Snowy Peak",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "樱花围绕雪山与石间溪流，花瓣随水面运动飘落。 原生视频规格为 11 秒 · 720×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Cherry blossoms frame a snowy peak and rocky stream, with petals drifting over moving water. Native-video output: 11s · 720×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yoradino/status/2108184725378711631",
    "sourceLabel": "X 原帖 · @yoradino",
    "author": "YOU RA",
    "publishedAt": "2026-10-08T13:16:21.000Z",
    "addedAt": "2026-10-08T17:14:26.977Z",
    "mediaUrl": "/media/x-2108184725378711631.mp4",
    "posterUrl": "/posters/x/2108184725378711631.jpg",
    "duration": 11,
    "aspectRatio": "portrait",
    "resolution": "720×960",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The river water has dried up due to the dry season, fortunately there was no fire. \n\nModel: FLUX DEV 2\nModel: MINIMAX H3\nSogni:\nhttps://app.sogni.ai/?code=Yora\n\n@Sogni_Protocol \n#Sogni #AIArt #AICreators #GenAI\n#sakura #aiart #AIillustration ㅤ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 48,
      "snapshotAt": "2026-10-08T17:14:26.977Z"
    },
    "approvedAt": "2026-10-08T17:14:26.977Z",
    "editorialBasis": "Viewed 10.912-second native video with moving water and petals; source explicitly identifies H3, despite caption's contradictory dry-river wording."
  },
  {
    "id": "x-2108172172208566681",
    "title": "月蚀下的和服群像",
    "titleEn": "Kimono Portraits Beneath an Eclipse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉衣少女与白衣狐耳人物在夜色中轮番出现，光环与月蚀连接音乐蒙太奇。 原生视频规格为 27 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Pink- and white-robed anime characters alternate beneath an eclipse and glowing rings in a musical montage. Native-video output: 27s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/hsrk_g_hsrk/status/2108172172208566681",
    "sourceLabel": "X 原帖 · @hsrk_g_hsrk",
    "author": "ヒロ(HG)｜実務で使えるAI活用",
    "publishedAt": "2026-10-08T12:26:28.000Z",
    "addedAt": "2026-10-08T17:19:46.824Z",
    "mediaUrl": "/media/x-2108172172208566681.mp4",
    "posterUrl": "/posters/x/2108172172208566681.jpg",
    "duration": 27,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#月蝕綺譚 の二次創作ショートMVです♬\n\nメインテーマ曲を咲耶さんに歌ってもらいました🎤\n\nMiniMax H3 の Ref2V で、リップシンクも良い感じにできそう！\n\nやっていることは、\n・公式のキャラシートを参照画像に(Ref2V)\n・codex で描いた起点画像を1フレーム目に固定(AddGuide)\n・曲の区間をそのまま音声として渡す → 咲耶の口が歌に合わせて動く\n\n歌うカットは3つ、５人のキャラが登場。曲のサビ26秒を11カットに割って、拍でつないだみたい（Opus5.5が）。\n\n11クリップの生成は、だいたい25分くらいでした。\nまたほかの曲もやってみよー。\n\n#ninjart",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 132,
      "snapshotAt": "2026-10-08T17:19:46.825Z"
    },
    "approvedAt": "2026-10-08T17:19:46.824Z",
    "editorialBasis": "Viewed original 26.53-second H3 reference-driven montage at opening and middle, with clear author claim."
  },
  {
    "id": "x-2108165526782439900",
    "title": "走出封闭设施的雪原少年",
    "titleEn": "Leaving the Facility for the Snow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣青年穿过走廊，走出灰色设施，踏入林边雪地。 原生视频规格为 27 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-jacketed young man moves from a corridor into snow outside a grey facility beside a forest. Native-video output: 27s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/joker371388/status/2108165526782439900",
    "sourceLabel": "X 原帖 · @joker371388",
    "author": "正道",
    "publishedAt": "2026-10-08T12:00:04.000Z",
    "addedAt": "2026-10-08T17:19:46.825Z",
    "mediaUrl": "/media/x-2108165526782439900.mp4",
    "posterUrl": "/posters/x/2108165526782439900.jpg",
    "duration": 27,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One person walks out of a sealed facility into the snow. No dialogue.\n\n\"不服從\": theme song MV for my original series Critical Awakening. AI-assisted: song by Mureka, visuals by MiniMax H3 in ComfyUI. Story, storyboard and edit by me.\n\nFull 3:03 ▶ https://youtu.be/nZ-v2HIHS-I",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 76,
      "snapshotAt": "2026-10-08T17:19:46.825Z"
    },
    "approvedAt": "2026-10-08T17:19:46.825Z",
    "editorialBasis": "Viewed native 27.05-second film at 7.19 and 26.01 seconds; original author credits H3 visuals and their own story and edit."
  },
  {
    "id": "x-2108157964259598457",
    "title": "公园图书馆的玻璃中庭",
    "titleEn": "Inside a Glass-Walled Park Library",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "公园中的玻璃图书馆从鸟瞰转入高挑中庭，黄衣人物沿书架走向楼梯。 原生视频规格为 66 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An aerial view of a park library moves into a tall glass atrium where a yellow-clad visitor approaches the stairs. Native-video output: 66s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/endo_bimdx/status/2108157964259598457",
    "sourceLabel": "X 原帖 · @endo_bimdx",
    "author": "Endo Masa",
    "publishedAt": "2026-10-08T11:30:01.000Z",
    "addedAt": "2026-10-08T17:19:46.825Z",
    "mediaUrl": "/media/x-2108157964259598457.mp4",
    "posterUrl": "/posters/x/2108157964259598457.jpg",
    "duration": 66,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "画像も動画も、ローカル環境で生成した\n「Park Side Library」を制作しました（架空の計画）\n\n■ 制作の流れ\n① 計画・図面\nAIで平面図と外観パースをつくり、建物のボリュームと配置を決定。\n\n② 静止画（ComfyUI）\nKrea 2・Qwen-Image-Edit で、各カットの静止画を生成。\n主人公ミナは、Krea 2 で顔・正面・横・後ろを1枚に並べた設定画をつくり、全カットで同じ人物になるようにしています。\n\n③ 冒頭の鳥瞰カットだけは BIM から\n図面をもとに Claude Opus 5.5 が Revit で BIM モデルを作成 → Blender で実写風の質感を付けてレンダリング → Krea 2 で写真化し、外構・人・植栽を描き足し。\n実際の設計データの形を保ったまま、映像の見た目に仕上げる工程を試しました。\n\n④ 動画化（MiniMax H3）\n静止画と設定画を MiniMax H3 に渡して動画を生成し、1080p に仕上げ。\n\n⑤ BGM\nGemini で作曲（ここだけクラウド）。\n\nカットごとに建物の形や間取りが合っていないところがたくさんあるので、次は全カットで整合性のとれた映像を目指します。\n\nRevit × Opus 5.5 × ComfyUI × MiniMax H3\n\n#Revit #BIM #ComfyUI #MiniMaxH3 #生成AI #建築パース",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 214,
      "snapshotAt": "2026-10-08T17:19:46.825Z"
    },
    "approvedAt": "2026-10-08T17:19:46.825Z",
    "editorialBasis": "Viewed native architectural film exterior and interior at 22.89 seconds; original Japanese author explicitly explains H3 video generation."
  },
  {
    "id": "x-2108153020844536295",
    "title": "黑帽魔女与张口绿蛇",
    "titleEn": "A Black-Hatted Witch Holding a Snake",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下对照中，原人物被替换成黑帽女性，手中张口的绿蛇和室内背景继续保留。 原生视频规格为 5 秒 · 2822×3250。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A comparison replaces the source person with a black-hatted woman while retaining the open-mouthed green snake and room. Native-video output: 5s · 2822×3250. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bigboss9718/status/2108153020844536295",
    "sourceLabel": "X 原帖 · @bigboss9718",
    "author": "big-boss_97",
    "publishedAt": "2026-10-08T11:10:22.000Z",
    "addedAt": "2026-10-08T17:19:46.826Z",
    "mediaUrl": "/media/x-2108153020844536295.mp4",
    "posterUrl": "/posters/x/2108153020844536295.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "2822×3250",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 + Character Swap LoRA\nTutorial https://youtu.be/uBqGvXiat2s\n\n#aitutorial #tutorial #aivideo #minimaxh3 #comfyui #minimax",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 85,
      "snapshotAt": "2026-10-08T17:19:46.826Z"
    },
    "approvedAt": "2026-10-08T17:19:46.826Z",
    "editorialBasis": "Viewed native 5.248-second character-swap output at 0.802 seconds; generated result is visible, not a pure tutorial announcement."
  },
  {
    "id": "x-2108150820118417628",
    "title": "暮色阳台的波斯菊微笑",
    "titleEn": "A Cosmos Flower on a Dusk Balcony",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白衣长发女性站在花簇阳台，手持粉色波斯菊，镜头逐渐靠近她的微笑。 原生视频规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-haired woman holds a pink cosmos flower on a balcony at dusk as the framing closes in on her smile. Native-video output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/noa_tcb7/status/2108150820118417628",
    "sourceLabel": "X 原帖 · @noa_tcb7",
    "author": "橘 乃愛",
    "publishedAt": "2026-10-08T11:01:37.000Z",
    "addedAt": "2026-10-08T17:19:46.827Z",
    "mediaUrl": "/media/x-2108150820118417628.mp4",
    "posterUrl": "/posters/x/2108150820118417628.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ゆれるよ / コスモスのそら\n\n※AI生成動画（MiniMax H3）",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 264,
      "snapshotAt": "2026-10-08T17:19:46.827Z"
    },
    "approvedAt": "2026-10-08T17:19:46.827Z",
    "editorialBasis": "Watched native 15.146-second video from opening through 13.50 seconds; original caption explicitly names H3."
  },
  {
    "id": "x-2108150777848173033",
    "title": "暮色站台的近景微笑",
    "titleEn": "A Close-Up Smile on a Dusk Platform",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰褐上衣女性背着黑色单肩包站在站台，先抬手示意，再向镜头靠近微笑。 原生视频规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman with a black shoulder bag gestures on a railway platform at dusk, then leans toward the camera with a smile. Native-video output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ShiramineMirei/status/2108150777848173033",
    "sourceLabel": "X 原帖 · @ShiramineMirei",
    "author": "白峰みれい",
    "publishedAt": "2026-10-08T11:01:27.000Z",
    "addedAt": "2026-10-08T17:24:38.408Z",
    "mediaUrl": "/media/x-2108150777848173033.mp4",
    "posterUrl": "/posters/x/2108150777848173033.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ひとつまえ / おりてみた\n\n※AI生成動画（MiniMax H3）",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 9,
      "views": 542,
      "snapshotAt": "2026-10-08T17:24:38.408Z"
    },
    "approvedAt": "2026-10-08T17:24:38.408Z",
    "editorialBasis": "Watched original 15.146-second H3 clip at opening and 14.136 seconds with clear source caption."
  },
  {
    "id": "x-2108150454269993357",
    "title": "黑白字形间的时尚人像",
    "titleEn": "Fashion Portraits Among Monochrome Type",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人像从黑发面部特写切换到白衣金发女性，巨大字形与黑色曲线围绕人物形成动效。 原生视频规格为 15 秒 · 1440×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired face close-up transitions to a blonde woman in white amid oversized typography and black graphic curves. Native-video output: 15s · 1440×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CharaspowerAI/status/2108150454269993357",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-10-08T11:00:10.000Z",
    "addedAt": "2026-10-08T17:24:38.409Z",
    "mediaUrl": "/media/x-2108150454269993357.mp4",
    "posterUrl": "/posters/x/2108150454269993357.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is honestly ridiculous for motion design when the source image already includes typography. 🔥\nIt doesn’t just animate the image.\n\nIt can make the text feel integrated into the motion, like it was designed for the sequence from the start.\nThat opens up a lot for posters, key art, title cards and branded visuals.\nI’ll drop the exact images I used below 👇",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 36,
      "views": 1619,
      "snapshotAt": "2026-10-08T17:24:38.409Z"
    },
    "approvedAt": "2026-10-08T17:24:38.409Z",
    "editorialBasis": "Watched native 15.168-second clip at opening and 9.539 seconds; original author states using these images for H3 motion design."
  },
  {
    "id": "x-2108148110782542242",
    "title": "橙色救援无人机的镜头特写",
    "titleEn": "Close-Ups of an Orange Rescue Drone",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙色无人机的圆形旋翼护网与黑色摄像头依次特写，产品文字随镜头出现。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Circular rotor guards and a black camera lens appear in successive close-ups of an orange rescue drone. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/didiokrescue/status/2108148110782542242",
    "sourceLabel": "X 原帖 · @didiokrescue",
    "author": "DIDIOK MAKINGS",
    "publishedAt": "2026-10-08T10:50:51.000Z",
    "addedAt": "2026-10-08T17:24:38.409Z",
    "mediaUrl": "/media/x-2108148110782542242.mp4",
    "posterUrl": "/posters/x/2108148110782542242.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "An animated short featuring the JX6A Water Rescue Drone, created with the MiniMax H3 model. The result turned out pretty impressive!\n\n#JX6AWaterRescueDrone #MiniMaxH3 #WaterRescueEquipment #FlyingLifebuoy #AIAnimation",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 63,
      "snapshotAt": "2026-10-08T17:24:38.409Z"
    },
    "approvedAt": "2026-10-08T17:24:38.409Z",
    "editorialBasis": "Viewed native AI-labeled 15.162-second product film; source explicitly claims H3 generation and own product animation."
  },
  {
    "id": "x-2108143815601103076",
    "title": "粉色夹克的双臂舞姿",
    "titleEn": "Raised-Arms Poses in a Pink Jacket",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发女性穿粉色皮夹克和黑色短上衣，在白色房间举起双臂，变换姿势并微笑。 原生视频规格为 14 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde woman in a pink leather jacket raises her arms, shifts her pose and smiles in a plain white room. Native-video output: 14s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Hathibel/status/2108143815601103076",
    "sourceLabel": "X 原帖 · @Hathibel",
    "author": "Ashraff Hathibelagal",
    "publishedAt": "2026-10-08T10:33:47.000Z",
    "addedAt": "2026-10-08T17:24:38.409Z",
    "mediaUrl": "/media/x-2108143815601103076.mp4",
    "posterUrl": "/posters/x/2108143815601103076.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Another test video I created using Krea 2, Minimax H3, and Grok.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 137,
      "snapshotAt": "2026-10-08T17:24:38.409Z"
    },
    "approvedAt": "2026-10-08T17:24:38.409Z",
    "editorialBasis": "Viewed author-watermarked native 14.421-second video at opening and 7.302 seconds; original author credits H3 with companion tools."
  },
  {
    "id": "x-2108142208146882863",
    "title": "键盘旁的双人互动",
    "titleEn": "Two People Beside a Keyboard",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Ref2VA",
    "summary": "卷发男子站在键盘与笔记本旁，戴眼镜的男子坐着微笑示意，两人在白色空间内互动。 原生视频规格为 17 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A curly-haired man stands beside a keyboard and laptop while a seated man with glasses smiles and gestures. Native-video output: 17s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/melvindvivas/status/2108142208146882863",
    "sourceLabel": "X 原帖 · @melvindvivas",
    "author": "Melvin Vivas",
    "publishedAt": "2026-10-08T10:27:24.000Z",
    "addedAt": "2026-10-08T17:24:38.409Z",
    "mediaUrl": "/media/x-2108142208146882863.mp4",
    "posterUrl": "/posters/x/2108142208146882863.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Update on our experiment of running a video face swap running fully locally and costing $0\n\nMiniMax H3 Ref2VA (INT8) in ComfyUI with a head-swap LoRA\n\nPeaking at 23.8GB VRAM (RTX 3090)\n\nNot perfect yet but we’re getting there\n\nGrok Bot and I are working on it",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 4,
      "views": 705,
      "snapshotAt": "2026-10-08T17:24:38.409Z"
    },
    "approvedAt": "2026-10-08T17:24:38.409Z",
    "editorialBasis": "Viewed native 17-second generated head-swap result at 4.802 seconds; author explicitly describes their local H3 experiment."
  },
  {
    "id": "x-2108140760109613408",
    "title": "夜街踏板车的双路对照",
    "titleEn": "Two Takes on a Night Scooter Ride",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴头盔的人驾驶浅绿色踏板车穿过夜间店铺，两块画面并列比较不同加速配置。 原生视频规格为 20 秒 · 1920×880。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A helmeted rider passes night-time shops on a pale green scooter in two panels comparing acceleration configurations. Native-video output: 20s · 1920×880. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kentdhani/status/2108140760109613408",
    "sourceLabel": "X 原帖 · @kentdhani",
    "author": "Kent Dhani",
    "publishedAt": "2026-10-08T10:21:39.000Z",
    "addedAt": "2026-10-08T17:24:38.409Z",
    "mediaUrl": "/media/x-2108140760109613408.mp4",
    "posterUrl": "/posters/x/2108140760109613408.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1920×880",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I tried ByteDance's DMAD 4-step LoRA for MiniMax H3 against Turbo v4 at 8 steps on my RTX5090 locally in ComfyUI.\n\nIts almost 2x faster, \n- 247s for Turbo v4 8 step\n- 136s for DMAD 4 step\n\nHowever, motion still smear on fast movement, its great for initial draft.\n\nResult and Links below👇",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 47,
      "views": 3433,
      "snapshotAt": "2026-10-08T17:24:38.409Z"
    },
    "approvedAt": "2026-10-08T17:24:38.409Z",
    "editorialBasis": "Viewed native 19.966-second output at 5.223 seconds; panels explicitly label H3 Turbo and DMAD variants and creator watermark."
  },
  {
    "id": "x-2108345685888438629",
    "title": "紫色光轨中的蓝发猫族",
    "titleEn": "Blue-Haired Feline amid Purple Light Trails",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝发猫族人物居于画面中央，黑金服饰周围的紫色光轨与发丝连续流动。 原生视频规格为 15 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired feline character stays centered as purple light trails and hair move around black-and-gold clothing. Native-video output: 15s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rares52_l/status/2108345685888438629",
    "sourceLabel": "X 原帖 · @rares52_l",
    "author": "Rares",
    "publishedAt": "2026-10-08T23:55:57.000Z",
    "addedAt": "2026-10-09T00:50:14.747Z",
    "mediaUrl": "/media/x-2108345685888438629.mp4",
    "posterUrl": "/posters/x/2108345685888438629.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "本地的MINIMAX H3 + wallpaper lora，影片沒有人工介入",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 34,
      "snapshotAt": "2026-10-09T00:50:14.747Z"
    },
    "approvedAt": "2026-10-09T00:50:14.747Z",
    "editorialBasis": "[\"Native 15.375-second clip autoplayed, readyState 4 and currentTime 9.400242. Viewed a blue-haired anthropomorphic feline in black-and-gold clothing surrounded by purple glowing orbital trails; hair and light trails move while the portrait remains centered. Separate input still is shown beside the video. No tutorial or narration.\"]"
  },
  {
    "id": "x-2108338797574582436",
    "title": "金色拉链与街头挎包",
    "titleEn": "Golden Zipper and a Streetwear Crossbody Bag",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金色拉链齿的材质特写切换到黑衣人物佩戴圆形挎包的户外镜头。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A close-up of golden zipper teeth cuts to an outdoor shot of a black-clad person wearing a round crossbody bag. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jerrod_lew/status/2108338797574582436",
    "sourceLabel": "X 原帖 · @jerrod_lew",
    "author": "Jerrod Lew",
    "publishedAt": "2026-10-08T23:28:35.000Z",
    "addedAt": "2026-10-09T00:50:14.747Z",
    "mediaUrl": "/media/x-2108338797574582436.mp4",
    "posterUrl": "/posters/x/2108338797574582436.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Magnific One and MiniMax H3 is an incredible combo.\n\nActivate Draft Mode and create multiple shots at once based on a character, brand, product or visual style.\n\nThrow them all into MiniMax H3 as references and watch it pump out adverts and marketing content in under a minute.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 6,
      "views": 354,
      "snapshotAt": "2026-10-09T00:50:14.747Z"
    },
    "approvedAt": "2026-10-09T00:50:14.747Z",
    "editorialBasis": "[\"Native 10.208-second video readyState 4 advanced from 2.374152 to 5.641864. Saw close-up golden zipper teeth on a black round crossbody bag, then a black-clad person wearing the bag with an orange logo outdoors. Actual product-ad footage, not a screen-recorded tutorial.\"]"
  },
  {
    "id": "x-2108332530772382070",
    "title": "木屋劈柴与温暖回眸",
    "titleEn": "Chopping Wood beside a Cabin and a Warm Smile",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暖色乡间镜头从木桩与劈柴切到木屋门把手，最后呈现红发女子的户外微笑。 原生视频规格为 37 秒 · 1536×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Warm rural shots move from a chopping stump to a cabin doorknob, ending with a red-haired woman smiling outdoors. Native-video output: 37s · 1536×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EndFolding79421/status/2108332530772382070",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-10-08T23:03:40.000Z",
    "addedAt": "2026-10-09T00:50:14.748Z",
    "mediaUrl": "/media/x-2108332530772382070.mp4",
    "posterUrl": "/posters/x/2108332530772382070.jpg",
    "duration": 37,
    "aspectRatio": "landscape",
    "resolution": "1536×864",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 ComfyUI — Motion Context V2 Workflow Test (Rough Drafts)\n\nHere are two videos testing two different approaches to building scenes in ComfyUI H3. I'm calling both of these my rough drafts, since they're mainly about experimenting, learning the workflow, and seeing what kind of results I can get.\n\n🔹 Video 1 Chopping Wood : Using keyframes to build and control the scenes. \n\n***Fyi get ready to fight with your LLM if you are using one for keyframes, they can make it difficult.***\n\n🔹 Video 2 Vacation: Using a Seedance video reference for the female character, along with text prompts to build the scenes. \n\n*** I did use images for locations in the last few scenes inside the hotel***\n\nBoth videos were generated at 8 steps.\n\nMy setup:\nRTX 5090 Laptop GPU (24GB VRAM)\n64GB RAM\n\nThese took roughly 4–5 hours total to put together. The main goal of this test was to show how quickly you can throw together a rough draft using this workflow and have something visual to work with before moving on to the final product.\n\nThere are definitely a lot of flaws, but that's to be expected. These aren't meant to be finished videos. They're more like a visual blueprint, allowing you to see how the scenes, camera angles, character movements, and overall story come together before spending more time and resources on the final version.\n\nThat's really what I wanted to explore here: getting a rough version of a scene put together quickly, seeing what works and what doesn't, and then going back to refine it before moving on to the final product.\n\nFor now, I'm pretty happy with what I was able to put together in just a few hours. There's still plenty of room for improvement, but that's the whole point of testing.\n\nI won't put all of the prompts here because that will make this too long,  but I will put the first 2 scenes from  the wood chopper video.\n\nPrompts:\n\nScene 1\n\nsubject_definitions:\n\n<Picture 1> Close-up reference of the axe splitting a log on the chopping stump. Use the exact axe head, wooden handle, log, and warm outdoor lighting.\n\n<Picture 2> Lily Carter looking through the cabin window between white curtains. Preserve her exact face, red hair, clothing, window frame, and expression.\n\n<Picture 3> Additional reference for the wood-splitting action.\n\n<picture 4> reference for the lily\n\nsummary:\n\nCinematic rural drama. Late afternoon golden-hour sunlight. Natural realistic motion, grounded physical interactions, consistent film photography.\n\nNO MUSIC. Natural environmental sounds only.\n\n[00:00–00:03]\nEXTREME CLOSE-UP. A solid log rests upright on the chopping stump. The axe swings down forcefully from above and strikes the center of the log. The wood splits into two pieces, scattering realistic splinters and wood chips. Sharp cracking impact.\n\n[00:03–00:07]\nHARD CUT. Exterior close-up looking through the cabin window. Lily gently pulls aside the white curtain and looks outside toward the man chopping wood. Her expression gradually softens into a subtle, warm, loving smile. She remains inside the cabin.\n\n[00:07–00:10]\nHARD CUT. Return to the chopping stump. A new intact log stands upright. The same axe strikes it, splitting it cleanly. Wood fragments scatter and fall naturally.\n\nNo dialogue.\n\nScene 2:\n\nsubject_definitions:\n\n<Picture 1> Refrigerator interior POV showing Lily reaching for a clear plastic water bottle. Preserve her brown jacket and gray sweater.\n\n<Picture 2> Extreme close-up of Lily's hand approaching the brass doorknob on the wooden cabin door.\n\n<Picture 3> Master exterior location reference. The original weathered wooden cabin occupies the LEFT of frame. Porch and wooden staircase are attached to the cabin. The man chops wood in the yard to the RIGHT. Preserve the original fence, chopping stump, stacked firewood, trees, and lighting.\n\n<picture 4> reference for the lily\n\nsummary:\n\nContinue the same afternoon and location. Consistent golden-hour sunlight. Realistic live-action cinematography.\n\nNO MUSIC.\n\n[00:00–00:03]\nHARD CUT. POV FROM INSIDE THE REFRIGERATOR. Lily reaches into the refrigerator, wraps her fingers around ONE clear plastic water bottle, and pulls it out. The bottle leaves the refrigerator with her hand.\n\n[00:03–00:05]\nHARD CUT. EXTREME CLOSE-UP of the cabin's brass doorknob. Lily's hand reaches into frame, grips the knob, rotates it, and begins opening the wooden door.\n\n[00:05–00:10]\nHARD CUT. WIDE EXTERIOR SHOT. The original cabin and staircase remain on the LEFT, chopping stump and firewood on the RIGHT. The man, wearing a plain white T-shirt, blue jeans, and brown work boots, is ALONE in the yard.\n\nHe raises the wooden-handled axe and chops another log. The split wood falls naturally. He prepares another piece.\n\nNo Lily visible in this shot.\n\n#ComfyUI #MiniMaxH3 #Seedance #AIVideo",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 313,
      "snapshotAt": "2026-10-09T00:50:14.748Z"
    },
    "approvedAt": "2026-10-09T00:50:14.748Z",
    "editorialBasis": "[\"Reviewed first native video, 37.024 seconds, readyState 4. Opening shows a stump and split log in warm outdoor light. At 16.547707 a hand turns a brass cabin doorknob. At 34.162637 a red-haired woman in a brown jacket and gray sweater smiles outside. Actual generated rural narrative, not a software screencast. Second 80.583-second video not reviewed or separately admitted.\"]"
  },
  {
    "id": "x-2108330604840321505",
    "title": "水洼裂纹长成冬树",
    "titleEn": "Puddle Cracks Grow into a Winter Tree",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "碎石间的灰色冰面浮现白色裂纹，树干与枝杈逐渐形成一棵光秃冬树。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "White cracks spread across a gray icy puddle amid pebbles, forming a bare tree with a trunk and branching limbs. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the cracks that spread from the tap grow slowly into the shape of a bare winter tree",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2108330604840321505",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108330604840321505",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-08T22:56:01.000Z",
    "addedAt": "2026-10-09T00:50:14.748Z",
    "mediaUrl": "/media/x-2108330604840321505.mp4",
    "posterUrl": "/posters/x/2108330604840321505.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Tapped the puddle once. It drew a tree.\n\nPrompt: \"the cracks that spread from the tap grow slowly into the shape of a bare winter tree\"\n\nDuration: 15s. Camera: locked, top-down. Rendered locally in ComfyUI, then Topaz to 4K.\n\nTrunk first, then branches. A crack needs an order to grow in or it all appears at once.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 202,
      "snapshotAt": "2026-10-09T00:50:14.748Z"
    },
    "approvedAt": "2026-10-09T00:50:14.748Z",
    "editorialBasis": "[\"Original native video readyState 4, playing at 13.746354 of 15.168 seconds. Viewed an icy gray puddle surrounded by pebbles; white cracks form a branching bare tree across its surface. Caption identifies the motion and result; no tutorial screencast.\"]"
  },
  {
    "id": "x-2108318947682758847",
    "title": "夕阳屋顶的回身舞步",
    "titleEn": "Turning Steps on a Sunset Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发女子穿白色无袖上衣和黑裙，在城市夕阳下的屋顶栏杆旁转身与迈步。 原生视频规格为 12 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman in a white sleeveless top and black skirt turns and steps beside rooftop railings against a sunset skyline. Native-video output: 12s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/itsmissglitch/status/2108318947682758847",
    "sourceLabel": "X 原帖 · @itsmissglitch",
    "author": "Amanda \"Glitch\" Ives",
    "publishedAt": "2026-10-08T22:09:42.000Z",
    "addedAt": "2026-10-09T00:50:14.749Z",
    "mediaUrl": "/media/x-2108318947682758847.mp4",
    "posterUrl": "/posters/x/2108318947682758847.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Pay attention 😉 #aivideo made with minimax h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 46,
      "snapshotAt": "2026-10-09T00:50:14.749Z"
    },
    "approvedAt": "2026-10-09T00:50:14.749Z",
    "editorialBasis": "[\"Native vertical video autoplayed at 6.101527 of 12.32 seconds, readyState 4. Viewed a dark-haired woman in a white sleeveless top, black skirt and black platform shoes turning and stepping on a rooftop balcony against a sunset city skyline.\"]"
  },
  {
    "id": "x-2108312039420776656",
    "title": "绿树旁的高尔夫挥杆",
    "titleEn": "A Golf Swing beside Green Trees",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴白色遮阳帽的女子在绿树旁挥动球杆，深色背心与图案裙随动作摆动。 原生视频规格为 8 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a white visor swings a golf club beside green trees, with a dark vest and patterned skirt moving through the action. Native-video output: 8s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/redtoweler/status/2108312039420776656",
    "sourceLabel": "X 原帖 · @redtoweler",
    "author": "赤タオル",
    "publishedAt": "2026-10-08T21:42:15.000Z",
    "addedAt": "2026-10-09T00:55:57.666Z",
    "mediaUrl": "/media/x-2108312039420776656.mp4",
    "posterUrl": "/posters/x/2108312039420776656.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ナイスショット！\n#MiniMax H3 #ちちぷい女子オープン",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 42,
      "snapshotAt": "2026-10-09T00:55:57.666Z"
    },
    "approvedAt": "2026-10-09T00:55:57.666Z",
    "editorialBasis": "[\"Native 8.064-second video playing at 1.089376, readyState 4. A woman in a white visor and blouse, dark vest and patterned skirt swings a golf club outdoors beside dense green trees. Actual generated golf scene.\"]"
  },
  {
    "id": "x-2108309139973943313",
    "title": "夕阳屋顶的翻越与攀爬",
    "titleEn": "Flips and Climbs on a Sunset Rooftop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣女子穿白鞋在夕阳屋顶翻转跳跃，随后抓住高处边缘悬挂并向上攀爬。 原生视频规格为 15 秒 · 1920×1066。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-clad woman in white shoes flips across a sunset rooftop, then hangs from a high ledge and climbs up. Native-video output: 15s · 1920×1066. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Muimi_Karma/status/2108309139973943313",
    "sourceLabel": "X 原帖 · @Muimi_Karma",
    "author": "無意味カルマ",
    "publishedAt": "2026-10-08T21:30:44.000Z",
    "addedAt": "2026-10-09T00:55:57.666Z",
    "mediaUrl": "/media/x-2108309139973943313.mp4",
    "posterUrl": "/posters/x/2108309139973943313.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1066",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "パルクール\n強引な動きをしすぎてる感😅\n@kusart_official\nMiniMax H3 via #KusArt AI🎬",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 40,
      "snapshotAt": "2026-10-09T00:55:57.666Z"
    },
    "approvedAt": "2026-10-09T00:55:57.666Z",
    "editorialBasis": "[\"Native 15.168-second video at readyState 4 advanced 1.570023 to 8.194028. A long-haired woman in black athletic clothes and white shoes flips across a sunset rooftop, then hangs from and climbs a tall rooftop ledge. Actual parkour output, not a pure advertisement or screencast.\"]"
  },
  {
    "id": "x-2108305009091981526",
    "title": "月夜石厅里的双人交锋",
    "titleEn": "A Duel in a Moonlit Stone Hall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "青色上衣的兽族人物与橙衣黑猫角色在石厅交手，拱窗外的月光映出连续攻防动作。 原生视频规格为 19 秒 · 544×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A turquoise-clad animal fighter exchanges strikes with a dark feline in orange clothes inside a stone hall with moonlit arched windows. Native-video output: 19s · 544×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EmpJellok/status/2108305009091981526",
    "sourceLabel": "X 原帖 · @EmpJellok",
    "author": "JellokEmp",
    "publishedAt": "2026-10-08T21:14:19.000Z",
    "addedAt": "2026-10-09T00:55:57.667Z",
    "mediaUrl": "/media/x-2108305009091981526.mp4",
    "posterUrl": "/posters/x/2108305009091981526.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "544×384",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Trying to get Minimax H3 to give a bit more. Still having sound errors not sure what I did. My 6800xt not doing to bad. #OC #AIart #ROCm",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 43,
      "snapshotAt": "2026-10-09T00:55:57.667Z"
    },
    "approvedAt": "2026-10-09T00:55:57.667Z",
    "editorialBasis": "[\"The native video played at readyState 4, duration approximately 19.456 seconds. Playback observations at 5.356819 and 5.807443 seconds show a turquoise-vested anthropomorphic fighter trading strikes with a dark feline fighter in orange clothes, in a stone hall with moonlit arched windows. Motion is visible in the original native player; author's caption identifies their local H3 output.\"]"
  },
  {
    "id": "x-2108304494975160643",
    "title": "霓虹街头的围巾少女与忍者",
    "titleEn": "A Scarf-Clad Heroine and Ninjas on a Neon Street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "围白色围巾的动漫少女站在霓虹街头，身后黑影掠过时举起手枪瞄准。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime heroine in a white scarf stands on a neon-lit street and raises a pistol as dark figures rush behind her. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/usadayoshi2sbc/status/2108304494975160643",
    "sourceLabel": "X 原帖 · @usadayoshi2sbc",
    "author": "宇佐田よし@AIイラスト",
    "publishedAt": "2026-10-08T21:12:16.000Z",
    "addedAt": "2026-10-09T00:55:57.667Z",
    "mediaUrl": "/media/x-2108304494975160643.mp4",
    "posterUrl": "/posters/x/2108304494975160643.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3練習\n川内と忍者たち\n\n動きが早いとかなりブレブレになりますが、まあこんな格闘シーンも作れたよ、ということで。",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 30,
      "views": 1026,
      "snapshotAt": "2026-10-09T00:55:57.667Z"
    },
    "approvedAt": "2026-10-09T00:55:57.667Z",
    "editorialBasis": "[\"Native 15.168-second video at readyState 4 advanced from 2.213881 to 11.325562 seconds. An anime woman with a white scarf stands in a neon night street, then draws and aims a pistol while shadowy figures cross behind her. Author describes their H3 fight-scene practice.\"]"
  },
  {
    "id": "x-2108303282028401089",
    "title": "暮色海崖上的星尘消散",
    "titleEn": "Dissolving into Stardust on a Twilight Cliff",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暮色海岸的斗篷人物逐渐散成闪亮颗粒与几何碎片，漂浮在深蓝海面与粉色天空之间。 原生视频规格为 5 秒 · 480×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cloaked figure on a twilight coast dissolves into glowing particles and geometric shards against a deep-blue sea and pink sky. Native-video output: 5s · 480×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tooniaanimation/status/2108303282028401089",
    "sourceLabel": "X 原帖 · @tooniaanimation",
    "author": "TOONIA AI Short Films",
    "publishedAt": "2026-10-08T21:07:27.000Z",
    "addedAt": "2026-10-09T00:55:57.667Z",
    "mediaUrl": "/media/x-2108303282028401089.mp4",
    "posterUrl": "/posters/x/2108303282028401089.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "480×832",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Cinematic disintegration effect created with MiniMax H3 Max on @picsartai Playground! 🌌✨\n\nWatch dissolve into iridescent star-dust particles and silver geometric shards on a rocky cliffside at twilight! Prompt screenshot in reply!\n\nMade with @picsartai  #picsartpartner",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-10-09T00:55:57.667Z"
    },
    "approvedAt": "2026-10-09T00:55:57.667Z",
    "editorialBasis": "[\"Original native 5.248-second video at readyState 4, observed at 4.6617 seconds. On the left a cloaked figure on a twilight rocky coast dissolves into bright particles and geometric shards. The right-hand panel within the author's native video contains the complete legible Prompt transcribed in r10-post, and a visible MiniMax H3 Max Turbo selector. Original output is shown, not just a tool advertisement.\"]"
  },
  {
    "id": "x-2108297700353400888",
    "title": "金色长裙与昏黄酒馆",
    "titleEn": "Golden Dresses and a Dim Bar",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金色图案长裙沿黄色建筑台阶走下，短片随后切入两位女子对坐的昏黄酒馆。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A flowing gold-patterned dress descends stairs amid yellow architecture before the short film cuts to two women facing each other in a dim bar. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/westkast/status/2108297700353400888",
    "sourceLabel": "X 原帖 · @westkast",
    "author": "West Kast",
    "publishedAt": "2026-10-08T20:45:16.000Z",
    "addedAt": "2026-10-09T01:00:49.874Z",
    "mediaUrl": "/media/x-2108297700353400888.mp4",
    "posterUrl": "/posters/x/2108297700353400888.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fashion",
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Nano Banana 2.1 — Shot 2 Trailer @lovart_ai \n→ 4K shots in high thinking from moodboards+pinterest boards\n→ 2K fashion cinematic trailer production via MiniMax H3\n#fashion #fiction #workflow #lovart #lovartpartner",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 91,
      "snapshotAt": "2026-10-09T01:00:49.874Z"
    },
    "approvedAt": "2026-10-09T01:00:49.874Z",
    "editorialBasis": "[\"Native 15.146666-second video at readyState 4 was observed at 9.798776 and, after looping, 1.587102 seconds. A gold-patterned flowing dress descends stairs between yellow architectural walls; later two fashionably dressed women sit opposite one another in a dim bar with drinks. Author explicitly identifies H3 as the video stage after image creation.\"]"
  },
  {
    "id": "x-2108290387806978411",
    "title": "风暴海面升起的触手巨兽",
    "titleEn": "A Tentacled Giant Rises from a Stormy Sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白海浪前方的城市灯光闪烁，披甲的触手巨兽在烟雾与巨浪中露出头部。 原生视频规格为 15 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "City lights shimmer beyond black-and-white waves as an armored, tentacled creature emerges amid smoke and a stormy sea. Native-video output: 15s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/prettyblocks/status/2108290387806978411",
    "sourceLabel": "X 原帖 · @prettyblocks",
    "author": "Pretty Blocks (🦉,🦉)",
    "publishedAt": "2026-10-08T20:16:13.000Z",
    "addedAt": "2026-10-09T01:00:49.874Z",
    "mediaUrl": "/media/x-2108290387806978411.mp4",
    "posterUrl": "/posters/x/2108290387806978411.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 on 3090 freaking COOKS.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 62,
      "snapshotAt": "2026-10-09T01:00:49.874Z"
    },
    "approvedAt": "2026-10-09T01:00:49.874Z",
    "editorialBasis": "[\"Native vertical video at readyState 4 observed near 0.009 seconds, 9.770004 seconds and 14.501676 seconds, duration approximately 15.168 seconds. It opens on a stormy black-and-white sea facing a lit skyline, cuts to a tentacled armored sea creature rising amid waves and smoke, and ends on black. Only first native video reviewed; author's original caption reports H3 running on their 3090.\"]"
  },
  {
    "id": "x-2108265584303747523",
    "title": "夕阳灯塔与掠过的海鸟",
    "titleEn": "A Sunset Lighthouse and Passing Seabirds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头沿白色灯塔向上移动，金色夕阳照亮玻璃灯室，海鸟在塔旁掠过。 原生视频规格为 15 秒 · 1216×672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The camera travels up a white lighthouse toward a glass lantern room lit by a golden sunset as seabirds pass nearby. Native-video output: 15s · 1216×672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DaKevK/status/2108265584303747523",
    "sourceLabel": "X 原帖 · @DaKevK",
    "author": "Kevin Koch",
    "publishedAt": "2026-10-08T18:37:39.000Z",
    "addedAt": "2026-10-09T01:00:49.874Z",
    "mediaUrl": "/media/x-2108265584303747523.mp4",
    "posterUrl": "/posters/x/2108265584303747523.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1216×672",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I’m absolutely blown away by what you can create in just a few minutes on A LAPTOP with open source tools.\n\n- Qwen-Image-2.1 creates a stunning image\n- MiniMax H3 turns it into a 15 second clip with sound\n- H3 upscaler doubles the resolution.\n\n🤯",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 61,
      "snapshotAt": "2026-10-09T01:00:49.874Z"
    },
    "approvedAt": "2026-10-09T01:00:49.874Z",
    "editorialBasis": "[\"Native 15.168-second video at readyState 4 advanced from 5.847392 to 8.73434 seconds. The camera moves upward along a white lighthouse to its glass lantern room against a golden sunset, with birds flying past. Original author explicitly reports the image-to-video and H3 upscale stages.\"]"
  },
  {
    "id": "x-2108262230068187358",
    "title": "废弃走廊里的噤声手势",
    "titleEn": "A Silent Gesture in an Abandoned Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣女子在墙皮剥落的昏暗走廊向镜头走近，抬起手指示意安静。 原生视频规格为 15 秒 · 576×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-clad woman approaches the camera through a dim corridor with peeling walls, holding a finger to her lips. Native-video output: 15s · 576×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/J_u_s_t_Nemo/status/2108262230068187358",
    "sourceLabel": "X 原帖 · @J_u_s_t_Nemo",
    "author": "Nemo",
    "publishedAt": "2026-10-08T18:24:19.000Z",
    "addedAt": "2026-10-09T01:00:49.875Z",
    "mediaUrl": "/media/x-2108262230068187358.mp4",
    "posterUrl": "/posters/x/2108262230068187358.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Beyond the Frame: Sogni Thriller Challenge\nMiniMax H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 46,
      "snapshotAt": "2026-10-09T01:00:49.875Z"
    },
    "approvedAt": "2026-10-09T01:00:49.875Z",
    "editorialBasis": "[\"First native 15.168-second video at readyState 4 observed at 3.954806 and 5.323294 seconds: a dark-haired woman in black walks toward the camera with a finger to her lips in a peeling, dim abandoned corridor. A second native recording shows the same project in the author's generation UI, supporting original authorship; its cropped Prompt is incomplete and is not retained. Only the first output is admitted.\"]"
  },
  {
    "id": "x-2108262204424229057",
    "title": "金色圆月下的神殿与熔岩",
    "titleEn": "A Temple and Lava beneath a Golden Moon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "巨大金色圆月照着山中的古典神殿，熔岩与烟云逐渐涌现，倒影在水面闪动。 原生视频规格为 10 秒 · 720×1260。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A huge golden moon lights a classical temple on a mountain as lava and smoky clouds emerge, with reflections shimmering across the water. Native-video output: 10s · 720×1260. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/churvikv/status/2108262204424229057",
    "sourceLabel": "X 原帖 · @churvikv",
    "author": "Viki",
    "publishedAt": "2026-10-08T18:24:13.000Z",
    "addedAt": "2026-10-09T01:00:49.875Z",
    "mediaUrl": "/media/x-2108262204424229057.mp4",
    "posterUrl": "/posters/x/2108262204424229057.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "720×1260",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Beauty has two faces - a quiet one and a flaming one.\n\nCreated in @gmi_cloud\n🏞 model: GPT Image 2.5 Flare\n🎥model: MiniMax H3\n\nPrompt to video:\nSmooth transformation from peaceful golden palace to dramatic volcanic mountain. Golden domes morph into ancient Greek columns, calm water begins to bubble and reflect lava, serene sky fills with fiery storm clouds, golden light turns into flowing lava, peaceful atmosphere transforms into epic volcanic drama. Cinematic morphing,",
    "engagement": {
      "replies": 17,
      "reposts": 8,
      "likes": 60,
      "views": 889,
      "snapshotAt": "2026-10-09T01:00:49.875Z"
    },
    "approvedAt": "2026-10-09T01:00:49.875Z",
    "editorialBasis": "[\"Original native 10.193559-second video at readyState 4 seen at 2.237052 and 8.664862 seconds, then looped. A golden classical temple sits on a dark mountain under a large golden moon; bright lava and smoky clouds appear while gold reflections ripple in the water. Original caption distinguishes image model from H3 video model. Prompt ends in a comma and completeness cannot be established, so keep null.\"]"
  },
  {
    "id": "x-2108257430614561220",
    "title": "复古摩托旁的初次欣赏",
    "titleEn": "A First Admiring Look at a Vintage Motorcycle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴墨镜的金发女子俯身靠近黑色复古摩托，沿油箱抚摸并仔细观察车身。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde woman in sunglasses leans beside a black vintage motorcycle, running her hand over the tank and studying the bike. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ou_zhen599/status/2108257430614561220",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-10-08T18:05:15.000Z",
    "addedAt": "2026-10-09T01:07:44.103Z",
    "mediaUrl": "/media/x-2108257430614561220.mp4",
    "posterUrl": "/posters/x/2108257430614561220.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                                                                                                  \n\n🎬Image-to-Video     \n📷“The First Lap”\nHer newly purchased vintage motorcycle hadn't moved an inch. As she examined it from the gas tank to the front wheel, the more closely she looked, the harder it was for her to hide her excitement.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 6,
      "views": 576,
      "snapshotAt": "2026-10-09T01:07:44.103Z"
    },
    "approvedAt": "2026-10-09T01:07:44.103Z",
    "editorialBasis": "[\"Native 15.1-second video at readyState 4 observed at 2.024453 seconds and progressed visibly to 13/15 seconds while reading replies. A blonde woman in sunglasses, white top and orange-trim clothing leans beside a black vintage motorcycle and runs her hand along its fuel tank. Original author explicitly identifies H3 image-to-video, and their reply exposes the complete detailed Prompt.\"]"
  },
  {
    "id": "x-2108247870797213698",
    "title": "被包装出售的一周人生",
    "titleEn": "A Week of Human Life Packaged for Sale",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "闹钟与手机订阅引出机械化日常，男子经过白色扫描设备，最后被装入标着一周期限的工厂包装。 原生视频规格为 176 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An alarm clock and phone subscription introduce a mechanized routine: a man passes beneath a white scanner and ends up in factory packaging labelled for one week. Native-video output: 176s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/h3nrixd/status/2108247870797213698",
    "sourceLabel": "X 原帖 · @h3nrixd",
    "author": "henrixd",
    "publishedAt": "2026-10-08T17:27:16.000Z",
    "addedAt": "2026-10-09T01:07:44.103Z",
    "mediaUrl": "/media/x-2108247870797213698.mp4",
    "posterUrl": "/posters/x/2108247870797213698.jpg",
    "duration": 176,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Make Me Human\n\nMade locally with MiniMax H3, plus FFmpeg, Remotion, Blender, Three.js, and ManimGL. Song: Suno V6.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 48,
      "snapshotAt": "2026-10-09T01:07:44.103Z"
    },
    "approvedAt": "2026-10-09T01:07:44.103Z",
    "editorialBasis": "[\"Original 175.658666-second native film at readyState 4. Opening 2.48115 seconds shows a digital 04:00 alarm clock, later a man in bed checks a phone with an expired subscription. At 13 seconds a WORK/SLEEP/BUY/REPEAT motion graphic appears; at 56 seconds the man sits beneath a white ring-shaped scanning machine. Near 168.113759 seconds he is packaged upright on an industrial conveyor beside a HUMAN (1 WEEK) barcode label. Actual beginning and later footage reviewed; final playback progressed at readyState 4 after seek.\"]"
  },
  {
    "id": "x-2108247695513334241",
    "title": "走廊里的女仆升空招式",
    "titleEn": "A Maid's Flying Move in a Corridor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女仆装角色从上方抓住另一人的头部，两人在贴满海报的走廊中一起升出画面。 原生视频规格为 10 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A maid-outfit character grips another person's head from above, and the pair rise out of frame in a poster-lined corridor. Native-video output: 10s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mtArakawa/status/2108247695513334241",
    "sourceLabel": "X 原帖 · @mtArakawa",
    "author": "元荒川ちゃん❤",
    "publishedAt": "2026-10-08T17:26:34.000Z",
    "addedAt": "2026-10-09T01:07:44.103Z",
    "mediaUrl": "/media/x-2108247695513334241.mp4",
    "posterUrl": "/posters/x/2108247695513334241.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "a maid who's kidnapping\nメイドさんの必殺技\nこのパターンに入ったら逃げられない!!\n\nセリフ指定してないんだけど\n何語しゃべってます？ 日本語ではないね\n\nNBPro + MiniMax H3\n#schoolgirl #maid #flying",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 732,
      "snapshotAt": "2026-10-09T01:07:44.104Z"
    },
    "approvedAt": "2026-10-09T01:07:44.103Z",
    "editorialBasis": "[\"Native 10.208-second video at readyState 4 advanced from 2.503222 to 9.827956 seconds. In a poster-lined corridor a maid-outfit character holds another character's head from above; the pair then rise vertically out of the frame. Non-graphic fantasy action, original author identifies NBPro image plus H3 video.\"]"
  },
  {
    "id": "x-2108245887990599985",
    "title": "玻璃大厦里的未完故事",
    "titleEn": "An Unfinished Story inside a Glass Tower",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两人站在高层玻璃大厦入口，随后切到窗边西装老人阅读文件的近景。 原生视频规格为 25 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two people stand outside a tall glass office tower before the film cuts to an older suited man reading a document by a window. Native-video output: 25s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Wolfdubaii/status/2108245887990599985",
    "sourceLabel": "X 原帖 · @Wolfdubaii",
    "author": "Wolfdubai",
    "publishedAt": "2026-10-08T17:19:23.000Z",
    "addedAt": "2026-10-09T01:07:44.104Z",
    "mediaUrl": "/media/x-2108245887990599985.mp4",
    "posterUrl": "/posters/x/2108245887990599985.jpg",
    "duration": 25,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "still making my movie on Minimax H3 ... hope so my hard Work Pays off ..... ,,, still not finish chapter 1 ,, but i love the story .... and quality is amazing",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 38,
      "snapshotAt": "2026-10-09T01:07:44.104Z"
    },
    "approvedAt": "2026-10-09T01:07:44.104Z",
    "editorialBasis": "[\"First native 24.566711-second video at readyState 4 advanced from 2.119652 to 21.519463 seconds. Opening wide shot shows two people at the steps of a tall glass office tower. Late shot shows an older suited man at a desk reading a paper against a windowed skyline. Original author explicitly describes this as their H3 film chapter in progress; no other video counted.\"]"
  },
  {
    "id": "x-2108256359875854427",
    "title": "功夫猫对战的三屏比较",
    "titleEn": "A Three-Panel Kung-Fu Cat Comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三屏同时呈现毛茸茸的猫与黑衣人物交手，底部片段展示猫跃到两人之间的动作。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three stacked panels compare a fluffy cat fighting black-clad men, with the bottom sequence showing the cat leaping between two opponents. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxVideoAI/status/2108256359875854427",
    "sourceLabel": "X 原帖 · @MaxVideoAI",
    "author": "MaxVideoAI",
    "publishedAt": "2026-10-08T18:01:00.000Z",
    "addedAt": "2026-10-09T01:07:44.104Z",
    "mediaUrl": "/media/x-2108256359875854427.mp4",
    "posterUrl": "/posters/x/2108256359875854427.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Three men picked the wrong fluffy cat. 🐈 Top: Wan 3 Prime. Middle: Seedance 2.5. Bottom: MiniMax H3. Fictional kung-fu scene compared on MaxVideoAI. Which take sells the fight? #KungFuCat",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 45,
      "snapshotAt": "2026-10-09T01:07:44.104Z"
    },
    "approvedAt": "2026-10-09T01:07:44.104Z",
    "editorialBasis": "[\"Native three-panel comparison video at readyState 4 advanced from 2.010163 to 8.285496 seconds, duration about 15.046529 seconds. Top Wan 3 Prime, middle Seedance 2.5, bottom MiniMax H3 labels are visible. A fluffy cat fights black-clad men outside a brick building; the bottom H3 panel shows the cat airborne between two men. Original account explicitly presents its own fictional comparison, not an unattributed repost.\"]"
  },
  {
    "id": "x-2108223113649385982",
    "title": "紫色帽檐下的人物肖像",
    "titleEn": "A Portrait beneath a Purple Cap",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫色棒球帽与浅紫眼镜的人物肖像在图形边框内变换表情；原作者说明最后一镜使用口型同步，其余镜头来自其他模型。 原生视频规格为 24 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A portrait wearing a purple baseball cap and pale-purple glasses changes expression inside a graphic frame; the creator attributes only the final lip-sync shot to H3 and the preceding video to another model. Native-video output: 24s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/alex_bagnuoli89/status/2108223113649385982",
    "sourceLabel": "X 原帖 · @alex_bagnuoli89",
    "author": "Alessandro Bagnuoli",
    "publishedAt": "2026-10-08T15:48:53.000Z",
    "addedAt": "2026-10-09T01:14:56.978Z",
    "mediaUrl": "/media/x-2108223113649385982.mp4",
    "posterUrl": "/posters/x/2108223113649385982.jpg",
    "duration": 24,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "From AI images to talking videos.\n\nBuilt entirely with @ImagineArt_X :\n🍌 Nano Banana 2.1 for images\n🎬 Kling 3.0 Pro for video\n🎙️ MiniMax H3 Lipsync for the final shot\n\nOne creative workflow. Multiple AI models.\n\nWhat do you think of the result?\n\n#GenerativeAI #AIvideo #ImagineArt @imagineart_creo",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 109,
      "snapshotAt": "2026-10-09T01:14:56.978Z"
    },
    "approvedAt": "2026-10-09T01:14:56.978Z",
    "editorialBasis": "[\"Native 24.148751-second video at readyState 4 observed at 5.344719, 12.88172 and 6.513651 seconds after looping. A woman wears a purple branded baseball cap, large light-purple glasses and hoop earrings inside a purple graphic frame; her face and gaze change. Original author attributes only the final lip-sync shot to H3 and earlier generation to Kling, so this is a mixed-model workflow, not an all-H3 generation claim.\"]"
  },
  {
    "id": "x-2108209802828558424",
    "title": "金色草地里的少年与小狗",
    "titleEn": "A Boy and a Dog in a Golden Field",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫少年与浅色小狗穿过夕阳草地，奔跑后缓缓停下，草叶在晚风中摇动。 原生视频规格为 20 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime boy and a light-colored dog run through a sunset field and slow down as golden grass moves in the evening breeze. Native-video output: 20s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ONcompute/status/2108209802828558424",
    "sourceLabel": "X 原帖 · @ONcompute",
    "author": "Ocean Network",
    "publishedAt": "2026-10-08T14:56:00.000Z",
    "addedAt": "2026-10-09T01:14:56.979Z",
    "mediaUrl": "/media/x-2108209802828558424.mp4",
    "posterUrl": "/posters/x/2108209802828558424.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This dog got a better sunset than most of us will see this year🐶\n\nOne Ocean Network template and a prompt gave it an anime movie scene, and your favorite animal deserves one too.\n\nTry it yourself:\n https://dashboard.oncompute.ai/inference/templates/video-studio-minimax-h3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 313,
      "snapshotAt": "2026-10-09T01:14:56.979Z"
    },
    "approvedAt": "2026-10-09T01:14:56.979Z",
    "editorialBasis": "[\"Native 20.128-second video at readyState 4 observed at 10.354059 and 11.947125 seconds. An anime young man runs and slows beside a light-colored dog through a golden grassy field at sunset. The dog turns near him, with grass moving in the breeze. It is the author's template-produced example video rather than a screen-only tutorial.\"]"
  },
  {
    "id": "x-2108239480822153531",
    "title": "灰色影棚里的红衣回眸",
    "titleEn": "A Red-Clad Turn in a Gray Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发女子穿红色花纹服装，在灰色影棚地面低位转身，将目光转向镜头。 原生视频规格为 10 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman in a red floral outfit poses low on a gray studio floor and turns her gaze toward the camera. Native-video output: 10s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/geesehowardt7/status/2108239480822153531",
    "sourceLabel": "X 原帖 · @geesehowardt7",
    "author": "Nina小娜学AI",
    "publishedAt": "2026-10-08T16:53:56.000Z",
    "addedAt": "2026-10-09T01:14:56.979Z",
    "mediaUrl": "/media/x-2108239480822153531.mp4",
    "posterUrl": "/posters/x/2108239480822153531.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is amazing !\nhttps://x.com/geesehowardt7/status/2107855917463257577",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 17,
      "views": 25038,
      "snapshotAt": "2026-10-09T01:14:56.979Z"
    },
    "approvedAt": "2026-10-09T01:14:56.979Z",
    "editorialBasis": "[\"Current post's native 10.208-second video at readyState 4 observed at 5.447342 seconds: a dark-haired adult-looking woman in a red floral outfit poses low on a gray studio floor, with a head turn toward camera. Non-explicit clothed portrait animation; no sex act or nudity. The quoted author's earlier 15-second video is separate and not counted. Only the current original output is considered.\"]"
  },
  {
    "id": "x-2108300793137016906",
    "title": "花衬衫男子的夸张表情",
    "titleEn": "An Expressive Man in a Floral Shirt",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "高梳黑发的男子身穿蓝色花衬衫，在摆满杂物与风扇的房间里面对镜头变换表情。 原生视频规格为 10 秒 · 2304×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man with swept-up dark hair and a blue floral shirt changes expression toward the camera in a cluttered room with fans and blinds. Native-video output: 10s · 2304×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/0x0SojalSec/status/2108300793137016906",
    "sourceLabel": "X 原帖 · @0x0SojalSec",
    "author": "Md Ismail Šojal 🕷️",
    "publishedAt": "2026-10-08T20:57:34.000Z",
    "addedAt": "2026-10-09T01:14:56.979Z",
    "mediaUrl": "/media/x-2108300793137016906.mp4",
    "posterUrl": "/posters/x/2108300793137016906.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2304×1280",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This Ace Ventura clip was made with zero reference images in locally.    \n\n- Just a 155 MB LoRA on MiniMax H3.  \n- The face consistency is getting uncomfortable.\n\n- https://huggingface.co/Playtime-AI/Minimax_H3-Ace_Ventura",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 15,
      "views": 1216,
      "snapshotAt": "2026-10-09T01:14:56.979Z"
    },
    "approvedAt": "2026-10-09T01:14:56.979Z",
    "editorialBasis": "[\"Native 10.208-second video at readyState 4, observed at 0.186217 seconds after playing/looping. An expressive man with swept-up dark hair, a blue floral shirt and white vest faces the camera in a cluttered room with fans and blinds. Original caption explicitly reports local H3 with a character LoRA and no reference images.\"]"
  },
  {
    "id": "x-2108304022704619754",
    "title": "仙人掌玻璃球里的金色流沙",
    "titleEn": "Golden Sand inside a Cactus Snow Globe",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "木架上的玻璃球装着绿色仙人掌，金色沙粒在球内旋转后缓缓落成沙堆。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Golden grains swirl inside a glass globe containing a green cactus, then fall into a mound on a wooden shelf. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "what swirls inside is not snow but fine golden sand",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2108304022704619754",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108304022704619754",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-08T21:10:24.000Z",
    "addedAt": "2026-10-09T01:14:56.979Z",
    "mediaUrl": "/media/x-2108304022704619754.mp4",
    "posterUrl": "/posters/x/2108304022704619754.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Shook the snow globe. Wrong climate.\n\nPrompt: \"what swirls inside is not snow but fine golden sand\"\n\nDuration: 15s. Camera: locked, shelf height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe sand still hangs on to the base for a beat before it falls. Two prompts in, that is the model and not the wording. This is the best take, and the same prompt goes to other models next.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 8,
      "views": 329,
      "snapshotAt": "2026-10-09T01:14:56.979Z"
    },
    "approvedAt": "2026-10-09T01:14:56.979Z",
    "editorialBasis": "[\"Original native 15.168-second video at readyState 4 seen playing at 4 seconds and 10.363517 seconds. A glass snow globe on a wooden shelf contains a green cactus; golden sand swirls and falls into a mound around it. Exact complete short Prompt is quoted in the original caption.\"]"
  },
  {
    "id": "x-2108137945047961805",
    "title": "灰发角色的即时对话",
    "titleEn": "Real-Time Conversation with a Gray-Haired Character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰发蓝眼动漫角色用手托着脸，随聊天内容变化眨眼并开口回应，屏幕右侧显示不断更新的对话面板。 原生视频规格为 94 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A gray-haired, blue-eyed animated character rests a hand against its cheek, blinking and responding as the adjacent chat panel updates. Native-video output: 94s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/uzuki425/status/2108137945047961805",
    "sourceLabel": "X 原帖 · @uzuki425",
    "author": "ゆずき",
    "publishedAt": "2026-10-08T10:10:28.000Z",
    "addedAt": "2026-10-09T01:24:46.830Z",
    "mediaUrl": "/media/x-2108137945047961805.mp4",
    "posterUrl": "/posters/x/2108137945047961805.jpg",
    "duration": 94,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime",
      "cg-3d"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こんな感じのキャラもいいです。動画なので何でもOK\nこれもMinimax-H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 11,
      "views": 768,
      "snapshotAt": "2026-10-09T01:24:46.830Z"
    },
    "approvedAt": "2026-10-09T01:24:46.830Z",
    "editorialBasis": "[\"First native 93.8-second screen recording at readyState 4 observed around 9.946858, 23 and 35.837406 seconds. It shows an animated gray-haired, blue-eyed 3D/anime character with a hand against its cheek; eyes open and close and the mouth responds while the author's chat panel changes. This is an original H3-generated character interaction demonstration, not static slides or narration alone. The other 147.3-second video belongs to an earlier reply and is not counted.\"]"
  },
  {
    "id": "x-2108137259916529749",
    "title": "白色背景前的微笑与眨眼",
    "titleEn": "A Smile and Wink against White",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白色背景前的黑发女子穿着白色上衣，对着镜头微笑并眨眼。 原生视频规格为 10 秒 · 832×512。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman in a white top smiles and winks toward the camera against a white background. Native-video output: 10s · 832×512. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KLAUScong/status/2108137259916529749",
    "sourceLabel": "X 原帖 · @KLAUScong",
    "author": "SCLau",
    "publishedAt": "2026-10-08T10:07:44.000Z",
    "addedAt": "2026-10-09T01:24:46.832Z",
    "mediaUrl": "/media/x-2108137259916529749.mp4",
    "posterUrl": "/posters/x/2108137259916529749.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "832×512",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "在算力租赁网站上用MINImax h3生成的视频，花了0.49元，不知道啥时候能赚回这0.49毛钱😐😐😐",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 89,
      "snapshotAt": "2026-10-09T01:24:46.832Z"
    },
    "approvedAt": "2026-10-09T01:24:46.832Z",
    "editorialBasis": "[\"First native 10.208-second video plays at readyState 4, observed at 4.830761 and 3.429585 seconds across looping playback. A dark-haired woman wearing a white T-shirt smiles and winks toward the camera against a white background. Second 120.97-second video is not this work and is not counted.\"]"
  },
  {
    "id": "x-2108135321120727404",
    "title": "晨光窗边的柔和肖像",
    "titleEn": "A Soft Portrait beside a Morning Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "晨光照亮窗边女子的浅色缎面上衣与深色长发，镜头展示护肤广告风格的柔和人物近景。 原生视频规格为 17 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Morning light illuminates a woman's pale satin top and dark hair beside a window in a soft skincare-commercial portrait. Native-video output: 17s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ModellixAI/status/2108135321120727404",
    "sourceLabel": "X 原帖 · @ModellixAI",
    "author": "Modellix",
    "publishedAt": "2026-10-08T10:00:02.000Z",
    "addedAt": "2026-10-09T01:24:46.832Z",
    "mediaUrl": "/media/x-2108135321120727404.mp4",
    "posterUrl": "/posters/x/2108135321120727404.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Soft morning light fills this luxury skincare commercial setting ☀️\n\n✓ Authentic natural‑light rendering.\n✓ Lifelike silk, glass and skin textures.\n✓ Stable subjects across cuts.\n\nMiniMax H3 · Image‑to‑video\nTry it on Modellix.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 171,
      "snapshotAt": "2026-10-09T01:24:46.832Z"
    },
    "approvedAt": "2026-10-09T01:24:46.832Z",
    "editorialBasis": "[\"Original native 17.1-second video plays at readyState 4 around 6.692598 seconds. The clip has a visible MiniMax H3 label and shows a dark-haired woman in a light satin top beside a bright window with softly moving hair. Later at 8.896225 it briefly buffered; playback was observed before that. The caption describes this creator/platform's actual skincare-commercial sample, not just a tool announcement.\"]"
  },
  {
    "id": "x-2108130018044023040",
    "title": "聊天面板旁的实时人物回应",
    "titleEn": "Real-Time Character Responses beside a Chat Panel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "浅绿色上衣的黑发角色随聊天面板的文字变化开口、眨眼并改变表情，展示近实时人物交互。 原生视频规格为 147 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired character in a pale green top speaks, blinks and changes expression as text in an adjacent chat panel updates. Native-video output: 147s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/uzuki425/status/2108130018044023040",
    "sourceLabel": "X 原帖 · @uzuki425",
    "author": "ゆずき",
    "publishedAt": "2026-10-08T09:38:58.000Z",
    "addedAt": "2026-10-09T01:24:46.832Z",
    "mediaUrl": "/media/x-2108130018044023040.mp4",
    "posterUrl": "/posters/x/2108130018044023040.jpg",
    "duration": 147,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax-H3、リアルタイム動画\n672x512と比較的大きなサイズで会話。1回の文章が長い部分で生成が追いついていません。なので準リアルタイム。かな。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 10,
      "views": 2529,
      "snapshotAt": "2026-10-09T01:24:46.832Z"
    },
    "approvedAt": "2026-10-09T01:24:46.832Z",
    "editorialBasis": "[\"Original native 147.3-second screen recording plays at readyState 4 at 6.312803 and 89.211659 seconds. A realistic dark-haired female character in a pale green top opens her mouth, blinks and changes expression as a chat panel beside her updates. Native scrub shows the same interaction setup at 139.132048 seconds while that position buffers. This is the author's actual near-real-time generated-character video, visually distinct from the gray-haired anime character in their later post.\"]"
  },
  {
    "id": "x-2108121004614857197",
    "title": "城市上空的绿光护盾",
    "titleEn": "A Green Energy Shield above the City",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色皮衣男子在城市楼宇间挥动绿色能量护盾，士兵与巨大翼兽的镜头组成一段超级英雄喜剧。 原生视频规格为 89 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-leather-clad hero wields a glowing green energy shield among city towers in a superhero parody with soldiers and a giant winged creature. Native-video output: 89s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GaryLau0101/status/2108121004614857197",
    "sourceLabel": "X 原帖 · @GaryLau0101",
    "author": "Gary",
    "publishedAt": "2026-10-08T09:03:09.000Z",
    "addedAt": "2026-10-09T01:30:29.195Z",
    "mediaUrl": "/media/x-2108121004614857197.mp4",
    "posterUrl": "/posters/x/2108121004614857197.jpg",
    "duration": 89,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Jensen Huang stars in NVIDIA MAN 🦸‍♂️ 英伟侠\nMeanwhile, Sam Altman still can’t catch a break.\n\nWho knows—maybe one day Jensen will actually be able to pull this off. 🤣\nMade on two DGX Sparks using MiniMax H3.",
    "engagement": {
      "replies": 17,
      "reposts": 23,
      "likes": 194,
      "views": 27948,
      "snapshotAt": "2026-10-09T01:30:29.195Z"
    },
    "approvedAt": "2026-10-09T01:30:29.195Z",
    "editorialBasis": "[\"Original 89.433-second native short plays at readyState 4 at 3.908478 and 43.72581 seconds. Opening has soldiers in a dusty street; midsection shows a gray-haired glasses-wearing man in a black leather jacket wielding a glowing green energy shield between city buildings. Native scrub at 84.641946 shows a suited man beside a huge winged creature near a tower. Source explicitly describes an original H3 superhero parody, not a news or talking-head post.\"]"
  },
  {
    "id": "x-2108108700628172800",
    "title": "棕色帘幕前的持麦表演",
    "titleEn": "A Microphone Performance before Brown Curtains",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发女子穿黑色连衣裙站在棕色帘幕前，将小型麦克风举到嘴边并随表演移动手臂。 原生视频规格为 8 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman in a black dress holds a small microphone to her mouth and moves her arm in front of brown curtains. Native-video output: 8s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Levi_D_Ackerman/status/2108108700628172800",
    "sourceLabel": "X 原帖 · @Levi_D_Ackerman",
    "author": "Levi·D·Ackerman",
    "publishedAt": "2026-10-08T08:14:15.000Z",
    "addedAt": "2026-10-09T01:30:29.195Z",
    "mediaUrl": "/media/x-2108108700628172800.mp4",
    "posterUrl": "/posters/x/2108108700628172800.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "云端部署了下 MiniMax-H3, 还是比较惊艳。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 121,
      "snapshotAt": "2026-10-09T01:30:29.195Z"
    },
    "approvedAt": "2026-10-09T01:30:29.195Z",
    "editorialBasis": "[\"Original first native 8.064-second video plays at readyState 4 around 4.608465 seconds. A dark-haired adult woman in a black dress stands in front of brown curtains, holds a small microphone to her mouth and moves her lips and arm. Other 15.083- and 54.033-second recommended videos are not counted. This is a fully clothed generated performance, not explicit sexual content.\"]"
  },
  {
    "id": "x-2108105695153524963",
    "title": "暗海帆船与海盗眼神",
    "titleEn": "A Sailing Ship and a Pirate's Gaze on a Dark Sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "海盗的眼部特写切入木制帆船在深色海浪中起伏的镜头，呈现手绘动漫片头的气氛。 原生视频规格为 15 秒 · 1440×2560。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An intense pirate-eye close-up leads into a wooden sailing ship moving over dark ocean waves in a drawn anime opening. Native-video output: 15s · 1440×2560. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Artedeingenio/status/2108105695153524963",
    "sourceLabel": "X 原帖 · @Artedeingenio",
    "author": "OscarAI",
    "publishedAt": "2026-10-08T08:02:19.000Z",
    "addedAt": "2026-10-09T01:30:29.195Z",
    "mediaUrl": "/media/x-2108105695153524963.mp4",
    "posterUrl": "/posters/x/2108105695153524963.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2560",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I really enjoy making anime openings with @Hailuo_AI MiniMax H3.\n\nYesterday I stumbled across this anime style in Midjourney and loved how the pirates turned out.\n\nI think there’s a great story waiting to be told here 🏴‍☠️",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 18,
      "views": 1717,
      "snapshotAt": "2026-10-09T01:30:29.195Z"
    },
    "approvedAt": "2026-10-09T01:30:29.195Z",
    "editorialBasis": "[\"Original first native approximately 15.1-second anime opening plays at readyState 4 at 4.122884 seconds: a wooden sailing ship rises and falls on dark ocean waves. Looping opening at 0.032252 shows an intense pirate face/eye close-up in a drawn anime style. Original caption explicitly says the creator makes these anime openings with H3; still-image style came from Midjourney. Other 30.083-second video is not this work.\"]"
  },
  {
    "id": "x-2108104845232398463",
    "title": "玻璃窗前的静谧触碰",
    "titleEn": "A Quiet Touch against a Glass Window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰色短发角色穿高领上衣走过混凝土房间，停在落地窗前抬手触碰玻璃，远处天空与海面泛着柔光。 原生视频规格为 29 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short gray-haired character in a high-neck top moves through a concrete room and raises a hand against a tall window overlooking softly lit sky and water. Native-video output: 29s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/FortyChan/status/2108104845232398463",
    "sourceLabel": "X 原帖 · @FortyChan",
    "author": "Forty Chan",
    "publishedAt": "2026-10-08T07:58:56.000Z",
    "addedAt": "2026-10-09T01:30:29.195Z",
    "mediaUrl": "/media/x-2108104845232398463.mp4",
    "posterUrl": "/posters/x/2108104845232398463.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Quick test using Qwen 3.8 27B Abliterted (self hosting thru runpod) +hermes agent +local Comfyui (Minimax-H3 - Low VRAM setup) + YuE 2 ( for BGM ). here the result ...",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 91,
      "snapshotAt": "2026-10-09T01:30:29.195Z"
    },
    "approvedAt": "2026-10-09T01:30:29.195Z",
    "editorialBasis": "[\"Original native approximately 29.4-second generated video is playable at readyState 4. At 3.524513 seconds a gray-high-neck-shirt character in dark trousers walks in a concrete room with skylights. At 14.79316 seconds the short gray-haired character touches the tall glass window beside a softly lit sea and sky. Looping playback at 10.876338 shows the same character walking toward the glass; no instructional UI or talking-head news clip. Creator explicitly assigns video generation to local ComfyUI MiniMax-H3 and BGM to YuE2.\"]"
  },
  {
    "id": "x-2108104553581134207",
    "title": "山路跑车与发光森林",
    "titleEn": "A Mountain-Road Car and a Glowing Forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白色跑车驶过山路弯道，云海窗前的床铺与森林中俯身观察发光蘑菇的宇航员组成奇幻蒙太奇。 原生视频规格为 115 秒 · 1728×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white sports car rounds a mountain bend, followed by a bed facing clouds and an astronaut studying a glowing mushroom in a fantastical montage. Native-video output: 115s · 1728×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/superalesha/status/2108104553581134207",
    "sourceLabel": "X 原帖 · @superalesha",
    "author": "Alexey Fateev",
    "publishedAt": "2026-10-08T07:57:46.000Z",
    "addedAt": "2026-10-09T01:35:34.671Z",
    "mediaUrl": "/media/x-2108104553581134207.mp4",
    "posterUrl": "/posters/x/2108104553581134207.jpg",
    "duration": 115,
    "aspectRatio": "landscape",
    "resolution": "1728×704",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This video was created entirely at home on 1× RTX 3090.\n\nImages: Qwen Image 2.1, FLUX dev\nVideo: MiniMax H3\nMusic: YuE2",
    "engagement": {
      "replies": 40,
      "reposts": 34,
      "likes": 606,
      "views": 25526,
      "snapshotAt": "2026-10-09T01:35:34.671Z"
    },
    "approvedAt": "2026-10-09T01:35:34.671Z",
    "editorialBasis": "[\"Original native approximately 115.2-second montage plays at readyState 4. At 7.469679 seconds a white sports car rounds a mountain-road bend. Around 20 seconds an unmade white bed faces a window filled with clouds. At 38.556372 an astronaut kneels beside a glowing mushroom in a dark forest. Scrubbing near 112 seconds buffers briefly, then playback reaches the black end frame at 115.221333 with readyState 4. This is the author's generated cinematic montage; caption clearly distinguishes H3 video from other image/music models.\"]"
  },
  {
    "id": "x-2108099724712771883",
    "title": "藏在沙发下的耳机小妖怪",
    "titleEn": "An Earbud-Hiding Creature beneath the Sofa",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "奶白色绒毛小生物蹲在沙发下面，抱着一只白色无线耳机，阳光照在旁边的木地板上。 原生视频规格为 12 秒 · 2160×3840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A tiny cream-colored furry creature crouches beneath a sofa holding a white wireless earbud beside a sunlit wooden floor. Native-video output: 12s · 2160×3840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TechTalkNAVI/status/2108099724712771883",
    "sourceLabel": "X 原帖 · @TechTalkNAVI",
    "author": "テックトークナビゲーター@AIコンテンツクリエイター",
    "publishedAt": "2026-10-08T07:38:35.000Z",
    "addedAt": "2026-10-09T01:35:34.675Z",
    "mediaUrl": "/media/x-2108099724712771883.mp4",
    "posterUrl": "/posters/x/2108099724712771883.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "2160×3840",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "イヤホンの片方だけ、なぜか見つからない――\nそれ、令和の妖怪「片耳隠し」の仕業です。\n\nワイヤレスイヤホンの片方だけを持ち去り、巣に溜め込む妖怪。悪気はない・・・。\n\n🎬 Renoise AI（MiniMax H3 Max）\n@renoiseaijp\n#イヤホンが見つからない\n#AI令和の妖怪コンテスト",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 22,
      "views": 639,
      "snapshotAt": "2026-10-09T01:35:34.675Z"
    },
    "approvedAt": "2026-10-09T01:35:34.675Z",
    "editorialBasis": "[\"Own original native 11.5-second clip plays at readyState 4 around 1.576566 seconds. A tiny cream-colored round furry creature crouches under a fabric sofa on a sunlit wooden floor, holding a white wireless earbud. Overlay Japanese text says one earbud is missing. The quoted contest announcement is not the source video; this author's entry is original and explicitly attributed to MiniMax H3 Max.\"]"
  },
  {
    "id": "x-2108091976574988390",
    "title": "鼓声激起的动漫粒子",
    "titleEn": "Anime Particles Bursting with Drumbeats",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫鼓手在紫色背景前举起鼓槌，随后鼓点与动作带动放射粒子和白色波形；人物素材生成后由作者进行三维与代码合成。 原生视频规格为 24 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime drummer raises sticks against purple as drumbeats and motion drive radial particles and white waveforms; the creator combines generated character footage with 3D and code-based compositing. Native-video output: 24s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aicreataro/status/2108091976574988390",
    "sourceLabel": "X 原帖 · @aicreataro",
    "author": "aicreataro",
    "publishedAt": "2026-10-08T07:07:48.000Z",
    "addedAt": "2026-10-09T01:35:34.675Z",
    "mediaUrl": "/media/x-2108091976574988390.mp4",
    "posterUrl": "/posters/x/2108091976574988390.jpg",
    "duration": 24,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「音と動きをビジュアライズ」\nOpus 5.5×Blender×Python\n\n■素材\n・人物：MiniMax H3\n・キービジュアル：Midjourney\n・音楽：MiniMax Music 3.0\n\n■Blenderとコードでやらせたこと\n・音が大きいほど、絵が立体になって手前に飛び出す\n・速く動いた所ほど、粒になって浮く\n・打った瞬間に時間が止まり、カメラだけが回り込む\n・止まっている間、空間ごと色と照明が切り替わる\n\n■わかったこと\n・生成だけでは出ない絵も組み合わせるとできる\n・やりすぎるとだめ\n・映像の動きにも反応させたら、映像そのものが鳴っているように見えてきた\n・飛び出す量は曲の音と、素材の動きに任せた\n\n実験的な映像ではありますが、何かの表現に使えそうだなと思います。AIをより拡張できる方法を探っていきたいです。\n音と動きの両方で絵が動くと、なんか面白いですね。",
    "engagement": {
      "replies": 30,
      "reposts": 89,
      "likes": 978,
      "views": 40047,
      "snapshotAt": "2026-10-09T01:35:34.675Z"
    },
    "approvedAt": "2026-10-09T01:35:34.675Z",
    "editorialBasis": "[\"Original native 24.25-second composed video plays at readyState 4. Opening shows a drawn anime taiko drummer raising two sticks against purple background. At 18.347964 seconds a radial particle/zoom effect and white soundwave-like lines burst over the image, with an audio spectrum along the bottom. Creator explicitly assigns character material to H3 and postproduction to Blender/Python; this is a finished experimental visual work, not a tutorial screen recording. Do not label all effects as directly H3 generated.\"]"
  },
  {
    "id": "x-2108707705175339446",
    "title": "桌上的折纸狼与白鸟",
    "titleEn": "A paper wolf meets a white bird",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "木桌上的灰色折纸狼与白色纸鸟相遇，纸翼展开，折纸形体在桌面上活动。 原生视频规格为 14 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A gray origami wolf meets a white paper bird on a wooden desk as the folded wings spread and the paper figures move. Native-video output: 14s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Dkackman/status/2108707705175339446",
    "sourceLabel": "X 原帖 · @Dkackman",
    "author": "⛏️🌱Don Kackman 🌱⛏️",
    "publishedAt": "2026-10-09T23:54:29.000Z",
    "addedAt": "2026-10-10T04:14:41.556Z",
    "mediaUrl": "/media/x-2108707705175339446.mp4",
    "posterUrl": "/posters/x/2108707705175339446.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Qwen 3.8 27B prompting Minimax H3 from OpenCode.\nAll local. All open.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 268,
      "snapshotAt": "2026-10-10T04:14:41.556Z"
    },
    "approvedAt": "2026-10-10T04:14:41.556Z",
    "editorialBasis": "[\"Beginning: gray folded-paper wolf on wooden desk near white paper.\",\"Middle: white paper bird opens its wings toward the wolf.\",\"End observation: white paper sheets on the same desk.\"]"
  },
  {
    "id": "x-2108704297152340471",
    "title": "雪中的蓝色长耳小伙伴",
    "titleEn": "A blue long-eared friend in the snow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水彩绘本风格的蓝色长耳小动物站在雪地里，头顶落满雪，开心地摇动耳朵和身体。 原生视频规格为 7 秒 · 1152×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue, long-eared creature covered in snow moves happily in a snowy watercolor storybook landscape. Native-video output: 7s · 1152×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelrhythms/status/2108704297152340471",
    "sourceLabel": "X 原帖 · @pixelrhythms",
    "author": "Keshav",
    "publishedAt": "2026-10-09T23:40:56.000Z",
    "addedAt": "2026-10-10T04:14:41.556Z",
    "mediaUrl": "/media/x-2108704297152340471.mp4",
    "posterUrl": "/posters/x/2108704297152340471.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1152×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A little Winnie update 🥹\n\nFour more moments of our baby biscuit growing up with the family. Discovering sour berries, indoor weather, and a warm hand for her cold nose 😂.\n\nBringing more of my Pokémon storybook to life with Krea + MiniMax H3 Turbo!",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 93,
      "snapshotAt": "2026-10-10T04:14:41.556Z"
    },
    "approvedAt": "2026-10-10T04:14:41.556Z",
    "editorialBasis": "[\"Blue long-eared creature in watercolor snowy forest at beginning, middle and end.\",\"Ears and body move between sampled frames.\"]"
  },
  {
    "id": "x-2108696573156930021",
    "title": "金紫丝带环绕的动漫少女",
    "titleEn": "An anime character surrounded by gold and purple ribbons",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发动漫角色在白色背景前微笑，金色与紫色的线条丝带绕过身体，头发与姿态细微变化。 原生视频规格为 10 秒 · 544×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired anime character smiles against a white background as gold and purple ribbon-like lines pass around her. Native-video output: 10s · 544×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Dizzy_AIX/status/2108696573156930021",
    "sourceLabel": "X 原帖 · @Dizzy_AIX",
    "author": "Dizzy_aix",
    "publishedAt": "2026-10-09T23:10:15.000Z",
    "addedAt": "2026-10-10T04:14:41.556Z",
    "mediaUrl": "/media/x-2108696573156930021.mp4",
    "posterUrl": "/posters/x/2108696573156930021.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "544×960",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 + FreeVideo",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 29,
      "views": 512,
      "snapshotAt": "2026-10-10T04:14:41.556Z"
    },
    "approvedAt": "2026-10-10T04:14:41.556Z",
    "editorialBasis": "[\"Black-haired anime character with white outfit and gold accessories.\",\"Middle: gold and purple ribbons cross the foreground; expression changes.\"]"
  },
  {
    "id": "x-2108693141096419671",
    "title": "窗边的女子近景对白",
    "titleEn": "A woman's close-up by the window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "室内柔光下，一名深色长发女子面对镜头，目光与嘴部动作在连续近景中变化。 原生视频规格为 8 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired woman faces the camera in softly lit interior close-ups, with changing eye and mouth movements. Native-video output: 8s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bseti/status/2108693141096419671",
    "sourceLabel": "X 原帖 · @bseti",
    "author": "BUSET",
    "publishedAt": "2026-10-09T22:56:37.000Z",
    "addedAt": "2026-10-10T04:14:41.557Z",
    "mediaUrl": "/media/x-2108693141096419671.mp4",
    "posterUrl": "/posters/x/2108693141096419671.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This Jennifer Connelly clip was made with zero reference images in locally.   \n-Just a 155 MB LoRA on MiniMax H3.  \n-The face consistency is getting uncomfortable.\n-https://huggingface.co/Playtime-AI/Minimax_H3-Jennifer_Connelly/",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 50,
      "snapshotAt": "2026-10-10T04:14:41.557Z"
    },
    "approvedAt": "2026-10-10T04:14:41.557Z",
    "editorialBasis": "[\"Beginning: dark-haired adult woman facing camera in an interior.\",\"Middle: eyes close briefly.\",\"End: mouth opens as expression changes.\"]"
  },
  {
    "id": "x-2108690640750235799",
    "title": "体育馆里的天使飞行练习",
    "titleEn": "Angel flight practice in a school gym",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "几名带白色翅膀、穿校服的女孩在体育馆内练习飞行，队列动作与腾空姿态逐步变化。 原生视频规格为 10 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "School-uniformed girls with white wings practice flying inside a gym, changing formation and lifting into the air. Native-video output: 10s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mtArakawa/status/2108690640750235799",
    "sourceLabel": "X 原帖 · @mtArakawa",
    "author": "元荒川ちゃん❤",
    "publishedAt": "2026-10-09T22:46:40.000Z",
    "addedAt": "2026-10-10T04:14:41.557Z",
    "mediaUrl": "/media/x-2108690640750235799.mp4",
    "posterUrl": "/posters/x/2108690640750235799.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "practice to fly\n天使ちゃんの飛行訓練（義務教育）\n\n無指定だと全員が全く同じ動きになるので\n個別にあれこれ指定した\nめんどい\n\nむしろ、アドリブで勝手にやってもらったほうが\n良い結果になるかも\n\nMiniMax H3\n#schoolgirl #flying",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 236,
      "snapshotAt": "2026-10-10T04:14:41.557Z"
    },
    "approvedAt": "2026-10-10T04:14:41.557Z",
    "editorialBasis": "[\"School gym with basketball hoop and uniformed girls.\",\"Middle: multiple girls have white wings.\",\"End: airborne uniformed legs visible near gym roof.\"]"
  },
  {
    "id": "x-2108688829318299815",
    "title": "床单折叠与迷你室内反转",
    "titleEn": "Folding sheets with a miniature-room twist",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子在卧室整理蓝色床单，镜头随后切到书架格子内坐着的男子，他竖起拇指作出回应。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman arranges a blue sheet on a bed before the scene cuts to a man seated inside a bookshelf compartment, giving a thumbs-up. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2108688829318299815",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-10-09T22:39:29.000Z",
    "addedAt": "2026-10-10T04:14:41.557Z",
    "mediaUrl": "/media/x-2108688829318299815.mp4",
    "posterUrl": "/posters/x/2108688829318299815.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Do you take the time to fold your fitted sheets or roll them into a ball and shove them in the closet? 🤔\n\nI am pleading the 5th on this one😅\n\nMiniMax H3 @Hailuo_AI",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 11,
      "views": 341,
      "snapshotAt": "2026-10-10T04:14:41.557Z"
    },
    "approvedAt": "2026-10-10T04:14:41.557Z",
    "editorialBasis": "[\"Beginning: woman holding blue fitted sheet in bedroom.\",\"Middle: woman smooths sheet on bed.\",\"End: seated man gives thumbs-up inside bookshelf compartment.\"]"
  },
  {
    "id": "x-2108683380938436705",
    "title": "羽翼与飘落纸币的庆祝画面",
    "titleEn": "A winged celebration with falling banknotes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色背景上，白色羽翼、飘落的羽毛和绿色纸币围绕角色与金额文字，构成庆祝风格的动画。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "White wings, floating feathers and green banknotes surround a character and amount text against a black background in a celebratory animation. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Naonekozamurai/status/2108683380938436705",
    "sourceLabel": "X 原帖 · @Naonekozamurai",
    "author": "なお｜AI動画・AI音楽",
    "publishedAt": "2026-10-09T22:17:50.000Z",
    "addedAt": "2026-10-10T04:14:41.557Z",
    "mediaUrl": "/media/x-2108683380938436705.mp4",
    "posterUrl": "/posters/x/2108683380938436705.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ヤッター、収益入ってたー！！\nオリジナルコンテンツ報酬プログラムから、2回目の収益いただきました✨\n\nいつもポストを見てくれる皆様のおかげです😆\nありがとうございます！！\n\nでも、インプレッションだだ下がりなので、次回は無いかもです〜🤣\n\n動画はHailuoのMiniMax H3で作成\n@Hailuo_AI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 26,
      "views": 442,
      "snapshotAt": "2026-10-10T04:14:41.557Z"
    },
    "approvedAt": "2026-10-10T04:14:41.557Z",
    "editorialBasis": "[\"Middle: $170.35 text, white wings and green banknotes on black background.\",\"End: white-haired winged illustrated character and falling feathers.\"]"
  },
  {
    "id": "x-2108680483945287795",
    "title": "月夜林中的花箭生日礼物",
    "titleEn": "A flower-arrow birthday gift under moonlight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "月夜森林里，白发兽耳角色收到饰有白花的箭，画面以生日祝福文字和角色微笑收尾。 原生视频规格为 17 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a moonlit forest, a white-haired animal-eared character receives an arrow decorated with white flowers, followed by a birthday message and a smile. Native-video output: 17s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CANINGUM/status/2108680483945287795",
    "sourceLabel": "X 原帖 · @CANINGUM",
    "author": "CANINGUM（カニンガム）",
    "publishedAt": "2026-10-09T22:06:19.000Z",
    "addedAt": "2026-10-10T04:14:41.557Z",
    "mediaUrl": "/media/x-2108680483945287795.mp4",
    "posterUrl": "/posters/x/2108680483945287795.jpg",
    "duration": 17,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "10月10日、クリプトニンジャ呂屯のお誕生日おめでとうございます🎂🎉\n\n#月蝕綺譚 版の呂屯ちゃんです🙏\n忠義をつくしてきた頭領アトザから、ささやかな贈り物🌼\n抜かずに大切にしてきた矢が👀✨\n\n※月蝕綺譚二次創作素材、Claude Code, Codex, MiniMax H3 Max（fal）, ElevenLabs\n#ninjart",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 152,
      "snapshotAt": "2026-10-10T04:14:41.557Z"
    },
    "approvedAt": "2026-10-10T04:14:41.557Z",
    "editorialBasis": "[\"Beginning: white-haired animal-eared anime character in moonlit forest.\",\"Middle: flower-decorated arrow placed beside head.\",\"End: Happy Birthday Rotten text appears.\"]"
  },
  {
    "id": "x-2108661011754721467",
    "title": "雪地里依偎的小伙伴",
    "titleEn": "Little friends cuddling in the snow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水彩绘本里，蓝色长耳小动物与棕色小伙伴依偎，在积雪与树影间抬头、拥抱。 原生视频规格为 5 秒 · 768×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a watercolor storybook, a blue long-eared creature and a small brown companion cuddle and move among snow and trees. Native-video output: 5s · 768×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelrhythms/status/2108661011754721467",
    "sourceLabel": "X 原帖 · @pixelrhythms",
    "author": "Keshav",
    "publishedAt": "2026-10-09T20:48:56.000Z",
    "addedAt": "2026-10-10T04:14:41.557Z",
    "mediaUrl": "/media/x-2108661011754721467.mp4",
    "posterUrl": "/posters/x/2108661011754721467.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×960",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Took a little break from the deep ocean experiments to return to my Pokémon storybook! \n\nTobi and Shelby found a new baby to look after — Winnie 🥹\n\nMade with Krea + MiniMax H3 Turbo!",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 182,
      "snapshotAt": "2026-10-10T04:14:41.557Z"
    },
    "approvedAt": "2026-10-10T04:14:41.557Z",
    "editorialBasis": "[\"Beginning: blue long-eared creature beside brown striped small creature.\",\"Middle and end: brown creature rises and nuzzles against the blue creature in snow.\"]"
  },
  {
    "id": "x-2108645689689297349",
    "title": "蚂蚁搬叶穿过林间枝干",
    "titleEn": "Leaf-carrying ants crossing forest branches",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "微距镜头跟随橙红色蚂蚁，蚁群抬着绿色叶片穿过苔藓和交错枝干，暖色光线照亮细节。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Macro shots follow orange-red ants carrying green leaves across moss and intersecting branches in warm forest light. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "0-5s: A dense column of black ants marches across a cracked concrete sidewalk in two clear lanes, the viewer low at ground level, shallow depth of field, morning golden light, telephoto compression. \n\n5-10s: One ant carrying a single bright red crumb of candy breaks out of its lane, others part around it, the viewer tracking alongside the crumb. \n\n10-15s: A fallen leaf lands across the path like a collapsed bridge, the column stops, then flows over and around it in a living ripple, the viewer rising slowly overhead. \n\nNegative prompt: cartoon ants, faces on ants, extra legs, humans, text, watermark, blurry subjects, slow motion",
    "promptSourceUrl": "https://x.com/DeCat2025/status/2108645689689297349",
    "sourceUrl": "https://x.com/DeCat2025/status/2108645689689297349",
    "sourceLabel": "X 原帖 · @DeCat2025",
    "author": "DeCat",
    "publishedAt": "2026-10-09T19:48:03.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108645689689297349.mp4",
    "posterUrl": "/posters/x/2108645689689297349.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Nature presents: Ants\nMinimax H3 @Hailuo_AI \n\nAnts already run something close to traffic systems, since some species form lanes where returning food carriers get right of way. \n\n0-5s: A dense column of black ants marches across a cracked concrete sidewalk in two clear lanes, the viewer low at ground level, shallow depth of field, morning golden light, telephoto compression. \n\n5-10s: One ant carrying a single bright red crumb of candy breaks out of its lane, others part around it, the viewer tracking alongside the crumb. \n\n10-15s: A fallen leaf lands across the path like a collapsed bridge, the column stops, then flows over and around it in a living ripple, the viewer rising slowly overhead. \n\nNegative prompt: cartoon ants, faces on ants, extra legs, humans, text, watermark, blurry subjects, slow motion",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 11,
      "views": 202,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: orange-red ants carry green leaves across mossy wood.\",\"Middle: ant climbs over branch with leaf.\",\"End: several ants traverse crossed branches.\"]"
  },
  {
    "id": "x-2108634813083341075",
    "title": "绿色洁面产品的人像广告",
    "titleEn": "A portrait ad for a green skincare product",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子在竖屏人像中举起绿色洁面产品，旁侧保留模特与产品参考图，镜头逐渐靠近面部与包装。 原生视频规格为 10 秒 · 464×848。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman holds a green skincare tube in a vertical portrait ad, with model and product references beside her as the shot moves closer. Native-video output: 10s · 464×848. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Nosper_/status/2108634813083341075",
    "sourceLabel": "X 原帖 · @Nosper_",
    "author": "Noris Perez🌻",
    "publishedAt": "2026-10-09T19:04:50.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108634813083341075.mp4",
    "posterUrl": "/posters/x/2108634813083341075.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "464×848",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created a full commercial for Anua using AI! ✨\n\nCombined Picsart AI Playground with Minimax H3 Max to create this ad.\n\n💬 Comment \"PROMPT\" for the prompt + discount code!\n\n@picsart #picsartearn #aigenerated",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 73,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: woman holds green Anua tube, model and product reference panels at right.\",\"Middle: woman smiles beside tube.\",\"End: close-up face and product label.\"]"
  },
  {
    "id": "x-2108632958571778405",
    "title": "街头绿车里的妖精怪客",
    "titleEn": "A leprechaun in a green street cruiser",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "夜色街道上，绿衣妖精坐在绿色汽车里，随后镜头切到他的皱纹面孔与露齿表情。 原生视频规格为 40 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A green-clad leprechaun rides in a green car on a night street before the video cuts to close-ups of his wrinkled face and toothy expression. Native-video output: 40s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tech2Wild/status/2108632958571778405",
    "sourceLabel": "X 原帖 · @Tech2Wild",
    "author": "Tech2Wild",
    "publishedAt": "2026-10-09T18:57:28.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108632958571778405.mp4",
    "posterUrl": "/posters/x/2108632958571778405.jpg",
    "duration": 40,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "city",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Leprechaun in the Hood w/ Minimax H3",
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 14,
      "views": 1012,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: green car parked on urban street with leprechaun driver.\",\"Middle and end: green-hatted wrinkled fantasy creature speaks and bares teeth.\"]"
  },
  {
    "id": "x-2108631283144273979",
    "title": "玻璃杯里的蓝墨水水母",
    "titleEn": "A blue-ink jellyfish in a drinking glass",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "刷子伸入透明水杯，蓝色墨水向下扩散，逐渐形成水母般的轮廓与飘动触须。 原生视频规格为 15 秒 · 3456×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A brush dips into a clear drinking glass, releasing blue ink that forms a jellyfish-like bell and drifting tendrils. Native-video output: 15s · 3456×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108631283144273979",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-09T18:50:49.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108631283144273979.mp4",
    "posterUrl": "/posters/x/2108631283144273979.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Rinsed the brush. Something swam off.\n\nPrompt: \"gathers into a small blue jellyfish that swims slowly around the glass\"\n\nDuration: 15s. Camera: locked, glass height. Rendered locally in ComfyUI, then Topaz to 4K.\n\n\"The water never turns fully blue\" is the save. Ink wants to fill the glass.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 274,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: brush enters clear water in glass with blue paint.\",\"Middle: blue ink curls into bell and tentacle shapes.\",\"End: jellyfish-like blue shape in water.\"]"
  },
  {
    "id": "x-2108629724234760497",
    "title": "旧报纸与荒野星空的虚构纪录片",
    "titleEn": "An imagined documentary of old clippings and desert skies",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "复古纪录片以旧报纸、划线地图和人物照片展开，穿插荒野中的人物与岩石景观，最后落到夜空银河。 原生视频规格为 241 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vintage-style fictional documentary unfolds through newspaper clippings, marked maps and portraits, ending with a Milky Way landscape above desert rocks. Native-video output: 241s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MixSdml/status/2108629724234760497",
    "sourceLabel": "X 原帖 · @MixSdml",
    "author": "SDMLR(AI)",
    "publishedAt": "2026-10-09T18:44:37.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108629724234760497.mp4",
    "posterUrl": "/posters/x/2108629724234760497.jpg",
    "duration": 241,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "retro"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "الفيلم القصير (( نمنم))  \nفيلم خيالي بأسلوب وثائقي سينمائي قديم.  \nصمّمت صور هذا الفيلم على\n@magnific\n، ووُلّدت الفيديو في المنصة نفسها باستخدام MiniMax H3\n@MiniMax__AI\n. أما الموسيقى فمن Suno\n@sunomusic\n.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 265,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: aged Arabic newspaper clipping with red-circled aerial landscape.\",\"Middle: vintage photograph of two robed people seated on rocks.\",\"End: star-filled Milky Way sky over rocky landscape.\"]"
  },
  {
    "id": "x-2108627369653072319",
    "title": "深夜老电视的天气与新闻片段",
    "titleEn": "Weather and news on a late-night vintage television",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "老式显像管电视依次播放日本天气地图和夜间公路新闻，镜头最后退回昏暗房间中的电视画面。 原生视频规格为 50 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A CRT television cycles through a Japanese weather map and nighttime highway news before the scene ends in a dim room. Native-video output: 50s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nakazakifam/status/2108627369653072319",
    "sourceLabel": "X 原帖 · @nakazakifam",
    "author": "中崎工房 | AIで1時間の仕事を5分で終わらせる人",
    "publishedAt": "2026-10-09T18:35:15.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108627369653072319.mp4",
    "posterUrl": "/posters/x/2108627369653072319.jpg",
    "duration": 50,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "retro"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "1997年10月11日、深夜午前2時のテレビ放送。\n\nOpus 5.5 / MiniMax H3 / Elevenlabs v4",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 287,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: CRT television with Japanese weather map.\",\"Middle: road traffic under orange lights and news caption.\",\"End: television in dim interior with Japanese subtitles.\"]"
  },
  {
    "id": "x-2108626652318343622",
    "title": "青发角色穿行黑白线框空间",
    "titleEn": "A turquoise-haired character in a wireframe world",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "青发动漫角色在发光的黑白几何线框中转身、行走，镜头拉远展示延伸的网格走廊。 原生视频规格为 15 秒 · 1888×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A turquoise-haired anime character turns and walks among luminous monochrome geometric wireframes as the camera pulls back into a gridded corridor. Native-video output: 15s · 1888×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2108626652318343622",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-10-09T18:32:24.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108626652318343622.mp4",
    "posterUrl": "/posters/x/2108626652318343622.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1888×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "https://fal.ai無料枠 Minimax H3 Max #falai\nMONOCHROME WIREFRAME WORLD2 \n（MiniMax H3 Max 出力動画＋CapCutの画質向上（AI HD）。1080p画質で再生可能）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 228,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: turquoise-haired anime character raises both hands.\",\"Middle: full-body character walks in white wireframe shapes.\",\"End: wide white-grid corridor with small central character.\"]"
  },
  {
    "id": "x-2108624790693241243",
    "title": "网格舞台上的青发舞者",
    "titleEn": "A turquoise-haired dancer on a grid stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "青发动漫角色在白色线框舞台上挥臂、转身，镜头逐步拉远，露出黑白网格构成的整个空间。 原生视频规格为 15 秒 · 1888×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A turquoise-haired anime character moves her arms and turns on a white wireframe stage as the camera pulls back to reveal the gridded space. Native-video output: 15s · 1888×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2108624790693241243",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-10-09T18:25:01.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108624790693241243.mp4",
    "posterUrl": "/posters/x/2108624790693241243.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1888×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "https://fal.ai無料枠 Minimax H3 Max #falai\nMONOCHROME WIREFRAME WORLD1\n（MiniMax H3 Max 出力動画＋CapCutの画質向上（AI HD）。1080p画質で再生可能）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 248,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: turquoise-haired anime character raises one arm beside angular white wireframes.\",\"Middle: full-body dancing pose on a grid stage with circular floor and ceiling marks.\",\"End: wide grid room, character small at center.\"]"
  },
  {
    "id": "x-2108623790582190150",
    "title": "河畔取水与户外备餐",
    "titleEn": "Collecting water and preparing food by a river",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子在河岸搬运树枝、用蓝色水桶取水，随后在户外用餐，连续镜头保留同一人物与河畔环境。 原生视频规格为 55 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman carries branches and collects river water in a blue bucket before eating outdoors, with the same character and riverbank across shots. Native-video output: 55s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AlexshevPm/status/2108623790582190150",
    "sourceLabel": "X 原帖 · @AlexshevPm",
    "author": "Alex Shev",
    "publishedAt": "2026-10-09T18:21:02.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108623790582190150.mp4",
    "posterUrl": "/posters/x/2108623790582190150.jpg",
    "duration": 55,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A cooking scene by the river, generated with MiniMax H3. The interesting test here is consistency: does the same person—and the same world—hold together across shots?",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 2,
      "views": 201,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: adult woman carries branches beside river and outdoor utensils.\",\"Middle: same woman pours water using blue bucket at riverbank.\",\"End: close-up of woman chewing beside prepared food.\"]"
  },
  {
    "id": "x-2108619109592367145",
    "title": "沙丁鱼群与掠过的巨鲨",
    "titleEn": "A sardine bait ball and a passing shark",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝色海水中，银色鱼群旋转聚集，一条巨鲨穿过鱼群后贴近水面游过，气泡和浪花随之出现。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Silver fish swirl into a bait ball in deep blue water before a large shark passes through and swims beneath the foamy surface. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/7u86j0Tf5Y3z7Ii/status/2108619109592367145",
    "sourceLabel": "X 原帖 · @7u86j0Tf5Y3z7Ii",
    "author": "qiuqiu.ai.lab",
    "publishedAt": "2026-10-09T18:02:26.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108619109592367145.mp4",
    "posterUrl": "/posters/x/2108619109592367145.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3  \n\nPhotorealistic underwater documentary\n\nPrompt below ⤵️\n\nPhotorealistic underwater documentary, IMAX color grade, deep blue open ocean. A tight swirling bait ball of silver sardines glitters in shafts of sunlight;",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 0,
      "views": 57,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: dense silver fish school swirls in blue ocean.\",\"Middle: large shark swims toward right with fish beside gills.\",\"End: shark passes close beneath bubbles and white surface foam.\"]"
  },
  {
    "id": "x-2108618747397513322",
    "title": "海湾到街头的三屏飞越对比",
    "titleEn": "A three-screen fly-through from bay to street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三屏画面并排对比海湾俯冲、海边街市与山海全景的飞行镜头，下方画面明确标注为 H3 输出。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Three stacked videos compare sweeping flights over a bay, a seaside street market and mountain-coast vistas; the bottom panel is labelled as H3 output. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxVideoAI/status/2108618747397513322",
    "sourceLabel": "X 原帖 · @MaxVideoAI",
    "author": "MaxVideoAI",
    "publishedAt": "2026-10-09T18:01:00.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108618747397513322.mp4",
    "posterUrl": "/posters/x/2108618747397513322.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Three routes from the coast into carnival chaos. Top: Wan 3 Prime. Middle: Seedance 2.5. Bottom: MiniMax H3. An imagined Rio fly-through compared on MaxVideoAI. Which camera move wins? #FPV",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 47,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: vertically stacked three model panels, bottom labelled MiniMax H3 with bay aerial view.\",\"Middle: all three move through seaside market; bottom shows H3 street stalls.\",\"End: bottom H3 panel shows wide beachfront vista.\"]"
  },
  {
    "id": "x-2108606865068433547",
    "title": "扫进簸箕的发光星星",
    "titleEn": "Glowing stars swept into a dustpan",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "木质门廊上，扫帚把几颗金色发光星星推向簸箕，星星滚动聚拢，留下细小的光点。 原生视频规格为 15 秒 · 3456×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "On a wooden porch, a broom pushes glowing gold stars into a dustpan as they roll together and leave tiny sparks. Native-video output: 15s · 3456×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108606865068433547",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-09T17:13:47.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108606865068433547.mp4",
    "posterUrl": "/posters/x/2108606865068433547.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Swept the porch. Meteor shower last night.\n\nPrompt: \"they clink like glass as they roll\"\n\nDuration: 15s. Camera: locked, floor height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe prompt asks for five stars and H3 swept up four. Exact counts are still the hardest thing to get out of it.",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 7,
      "views": 319,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: gold glowing stars on porch floor beside dustpan.\",\"Middle: broom pushes stars toward pan.\",\"End: stars gathered in pan with glowing particles.\"]"
  },
  {
    "id": "x-2108606133854441528",
    "title": "灯笼街头的双人功夫动作",
    "titleEn": "Two characters performing martial arts under lanterns",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红蓝旗袍动漫角色在灯笼街头同步摆出功夫姿势，转身挥臂后回到正面定格。 原生视频规格为 8 秒 · 896×1120。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two anime characters in red and blue floral dresses perform synchronized martial-arts poses on a lantern-lit street, turning before facing forward again. Native-video output: 8s · 896×1120. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/toraaiuser2/status/2108606133854441528",
    "sourceLabel": "X 原帖 · @toraaiuser2",
    "author": "AIossansan",
    "publishedAt": "2026-10-09T17:10:52.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108606133854441528.mp4",
    "posterUrl": "/posters/x/2108606133854441528.jpg",
    "duration": 8,
    "aspectRatio": "portrait",
    "resolution": "896×1120",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おやすみなさいませ(´ー｀)\n\nCodexから画像を直接ComfyUIに渡して、MiniMax H3で動画生成できるようにしてもらいました(´ー｀)\n\nこれで毎回プロンプトをコピペする手間もなくなって、かなり楽になりました。\n\n画像から動画生成までの流れがスムーズになったので、いろいろ試しやすくなりそうです。\n\n#AIイラスト",
    "engagement": {
      "replies": 3,
      "reposts": 13,
      "likes": 93,
      "views": 1541,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: two anime characters in red and blue floral outfits pose with open palms.\",\"Middle: both turn with hair and sleeves moving under lanterns.\",\"End: front-facing martial-arts stance.\"]"
  },
  {
    "id": "x-2108599749179781551",
    "title": "少女奔向夕阳化为火凤凰",
    "titleEn": "A girl runs into sunset and transforms into a phoenix",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "少女穿过金色草地，跃起时身体被火焰和羽翼包围，最终化为凤凰飞向低垂的夕阳。 原生视频规格为 15 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A girl runs through golden grass, leaps into flames and feathers, and becomes a phoenix flying toward the low sunset. Native-video output: 15s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "integrated_multimodal_description: [Shot 1] Live-action, cinematic, anamorphic 35mm, golden hour, fine film grain, warm sun flare. A low tracking shot through a wide open meadow of tall gold grass, low sun ahead on the horizon, no buildings, no road. A Japanese schoolgirl runs screen-left to screen-right, feet-first, toward the open field and the sun, not back toward the camera. Identity lock: late teens, fair skin, black shoulder-length hair with a small red ribbon, navy sailor uniform with a white collar and red neckerchief, pleated navy skirt, white socks, brown loafers. Determined face, eyes on the field ahead, not on the camera, mouth closed. Grass whips through the foreground, out of focus. Her hair and skirt stream behind her. The camera trucks right at fast speed at knee height, matching her pace, with large amplitude, sun flare raking the lens. No second person. No bird yet. No speech. No narrator.\n[Shot 2] At 00:03.800, the camera cuts to an extreme low angle from the grass, looking up. Same girl, same lock: late teens, black shoulder-length hair, red ribbon, navy sailor uniform, white collar, red neckerchief, pleated skirt, white socks, brown loafers. She plants her right foot and leaps straight up, arms opening, skirt and hair lifting, loafers leaving the grass. The sun sits behind her so she is a rim-lit silhouette against a clear gold sky. She does not fall back. She does not look at the camera. Mouth stays closed. The camera tilts up at fast speed with large amplitude, following her rise, grass blades sharp in the near foreground. No speech. No narrator.\n[Shot 3] At 00:07.200, the camera cuts to a medium shot of her at the top of the leap, same body, same face, same navy uniform and red ribbon, against the gold sky. One continuous morph, no second figure, no costume left behind, no small bird hatching out of her. Gold fire kindles at her outstretched fingertips and at the hem of the skirt, climbs the arms, and becomes long primary feathers. The sailor collar and skirt burn away into plumage on the same body. Her face lengthens into a raptor beak in that same head; the eyes stay in place and turn coal-bright. Her legs tuck and become talons. What finishes the morph is a phoenix the size of a large crane, wingspan wider than she was tall: gold breast, crimson flight feathers with white-hot edges, a long trailing tail of fire, not a cartoon flame blob, not a sparrow, not a chicken. The camera arcs around her with large amplitude at slow speed as the wings open. No human girl remains in frame. No speech. No narrator.\n[Shot 4] At 00:11.000, the camera cuts to a wide low crane over the same meadow. The phoenix only: crane-sized, gold breast, crimson flight feathers, white-hot edges, long fire tail, coal-bright eyes, talons tucked. It banks once, then flies away from the camera toward the low sun, wings beating forward, hood-first, not reversing, not circling back to the girl, because the girl is gone. Embers drift down into the grass and go dark; the field does not catch. She shrinks toward the sun. The camera tilts up and pulls out with large amplitude at slow speed until the bird is a bright streak on the horizon. No speech. No narrator.\n\noverall_soundscape: Wind in tall grass, loafers on dry stems, a short breath on the plant, fabric snap as she leaves the ground. A rising whoosh as fire takes the uniform, then heavy wing beats and a single wild phoenix cry, beak open once, not words. Embers tick into the grass. After the cry, only wind and wings fading. No voices. No narrator.\n\nnon_diegetic_music: Low strings and a solo shakuhachi in a slow pulse, a single taiko hit on the plant, then fast rising brass and a high string ostinato under the wing beats. No lyrics. No sung theme.",
    "promptSourceUrl": "https://x.com/Cardoso/status/2108599749179781551",
    "sourceUrl": "https://x.com/Cardoso/status/2108599749179781551",
    "sourceLabel": "X 原帖 · @Cardoso",
    "author": "cardoso",
    "publishedAt": "2026-10-09T16:45:30.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108599749179781551.mp4",
    "posterUrl": "/posters/x/2108599749179781551.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#Aulinhas_de_IA\n\nO bom da IA é que dá pra concretizar qualquer pensamento intrusivo. Hoje me veio a imagem, colegial japonesa correndo em um campo, pulando e virando uma Fênix. \n\nAbri o MiniMax H3, colei esse prompt e Saiu o vídeo animado. Uma bela bosta, mas representa um dos adágios da Informática: GIGO - Garbage In, Garbage Out. Dados ruins geram resultados ruins. Um prompt fraco deixa tudo à cargo da IA, que é preguiçosa, ela literalmente vai escolher o caminho mais fácil. \n\nO que eu fiz: Tenho um skill no @Grok com as especificações para criar prompts para o MiniMax. Pedi para ele melhorar meu prompt original, tornando-o realista, cinematográfico, etc. Joguei o novo prompt no ComfyUI, e o resultado é 1000x melhor. \n\nNão tenha vergonha de usar LLMs para melhorar seus prompts, e adequá-los ao formato do modelo que está utilizando. Ferramenta é para ser utilizada.\n\nNOTA: Dicas como esta são fruto de contribuição de leitores como você. Assine o Perfil, custa menos de um chopp por mês!\n\nhttps://x.com/Cardoso/creator-subscriptions/subscribe\n\n---------------\nO Prompt gerado pelo Grok:\n\nintegrated_multimodal_description: [Shot 1] Live-action, cinematic, anamorphic 35mm, golden hour, fine film grain, warm sun flare. A low tracking shot through a wide open meadow of tall gold grass, low sun ahead on the horizon, no buildings, no road. A Japanese schoolgirl runs screen-left to screen-right, feet-first, toward the open field and the sun, not back toward the camera. Identity lock: late teens, fair skin, black shoulder-length hair with a small red ribbon, navy sailor uniform with a white collar and red neckerchief, pleated navy skirt, white socks, brown loafers. Determined face, eyes on the field ahead, not on the camera, mouth closed. Grass whips through the foreground, out of focus. Her hair and skirt stream behind her. The camera trucks right at fast speed at knee height, matching her pace, with large amplitude, sun flare raking the lens. No second person. No bird yet. No speech. No narrator.\n[Shot 2] At 00:03.800, the camera cuts to an extreme low angle from the grass, looking up. Same girl, same lock: late teens, black shoulder-length hair, red ribbon, navy sailor uniform, white collar, red neckerchief, pleated skirt, white socks, brown loafers. She plants her right foot and leaps straight up, arms opening, skirt and hair lifting, loafers leaving the grass. The sun sits behind her so she is a rim-lit silhouette against a clear gold sky. She does not fall back. She does not look at the camera. Mouth stays closed. The camera tilts up at fast speed with large amplitude, following her rise, grass blades sharp in the near foreground. No speech. No narrator.\n[Shot 3] At 00:07.200, the camera cuts to a medium shot of her at the top of the leap, same body, same face, same navy uniform and red ribbon, against the gold sky. One continuous morph, no second figure, no costume left behind, no small bird hatching out of her. Gold fire kindles at her outstretched fingertips and at the hem of the skirt, climbs the arms, and becomes long primary feathers. The sailor collar and skirt burn away into plumage on the same body. Her face lengthens into a raptor beak in that same head; the eyes stay in place and turn coal-bright. Her legs tuck and become talons. What finishes the morph is a phoenix the size of a large crane, wingspan wider than she was tall: gold breast, crimson flight feathers with white-hot edges, a long trailing tail of fire, not a cartoon flame blob, not a sparrow, not a chicken. The camera arcs around her with large amplitude at slow speed as the wings open. No human girl remains in frame. No speech. No narrator.\n[Shot 4] At 00:11.000, the camera cuts to a wide low crane over the same meadow. The phoenix only: crane-sized, gold breast, crimson flight feathers, white-hot edges, long fire tail, coal-bright eyes, talons tucked. It banks once, then flies away from the camera toward the low sun, wings beating forward, hood-first, not reversing, not circling back to the girl, because the girl is gone. Embers drift down into the grass and go dark; the field does not catch. She shrinks toward the sun. The camera tilts up and pulls out with large amplitude at slow speed until the bird is a bright streak on the horizon. No speech. No narrator.\n\noverall_soundscape: Wind in tall grass, loafers on dry stems, a short breath on the plant, fabric snap as she leaves the ground. A rising whoosh as fire takes the uniform, then heavy wing beats and a single wild phoenix cry, beak open once, not words. Embers tick into the grass. After the cry, only wind and wings fading. No voices. No narrator.\n\nnon_diegetic_music: Low strings and a solo shakuhachi in a slow pulse, a single taiko hit on the plant, then fast rising brass and a high string ostinato under the wing beats. No lyrics. No sung theme.",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 24,
      "views": 2270,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: live-action-style girl in navy sailor uniform runs through sunlit gold meadow.\",\"Middle: orange fire forms wings around the girl.\",\"End: phoenix flies away over meadow toward sunset; a separate anime clip is visible beside this first video.\"]"
  },
  {
    "id": "x-2108591747144302675",
    "title": "西装男子升空变成烟花",
    "titleEn": "A suited man launches into fireworks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "草坪上的西装男子抬腿，脚下喷出火焰后升向空中，最后在蓝天中化成扩散的烟花。 原生视频规格为 10 秒 · 768×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A suited man on a lawn lifts a leg, rises on a fiery jet, and ends as a burst of fireworks against a blue sky. Native-video output: 10s · 768×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mwhuss/status/2108591747144302675",
    "sourceLabel": "X 原帖 · @mwhuss",
    "author": "Marshall Huss",
    "publishedAt": "2026-10-09T16:13:42.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108591747144302675.mp4",
    "posterUrl": "/posters/x/2108591747144302675.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just discovered MiniMax-H3 runs well on my M3 Ultra 96gb. \n\nThis is what happens when you give the kids a picture of their uncle and tell them to write a prompt.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 255,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: adult man in blue suit lifts leg on lawn.\",\"Middle: flame jet launches him upward.\",\"End: expanding fireworks in blue sky.\"]"
  },
  {
    "id": "x-2108591639145132225",
    "title": "像素对战界面里的角色短片",
    "titleEn": "Character clips inside a pixel battle interface",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "复古对战游戏界面中，蓝发与粉发角色轮流出现在擂台短片里，上方保留血条、手势与回合文字。 原生视频规格为 171 秒 · 640×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Blue- and pink-haired characters appear in arena clips inside a retro battle-game interface with health bars, hand signs and round text. Native-video output: 171s · 640×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HOSSIE_JP/status/2108591639145132225",
    "sourceLabel": "X 原帖 · @HOSSIE_JP",
    "author": "ほっしー",
    "publishedAt": "2026-10-09T16:13:17.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108591639145132225.mp4",
    "posterUrl": "/posters/x/2108591639145132225.jpg",
    "duration": 171,
    "aspectRatio": "landscape",
    "resolution": "640×480",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ゲームボーイで野球拳脱衣ゲームを作ろうから始まったのに、なぜかサターンでムービー野球拳バトルができたんですよ。\n動画はAI指示によるローカルMiniMax H3生成一発勝負、いろいろ破綻しているのもそのまま収録する味わい（供養）。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 180,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: retro Japanese battle-game title and menu.\",\"Middle: blue-haired anime character in arena surrounded by pixel UI.\",\"End: pink-haired character gestures in arena, health bars and duel text visible.\"]"
  },
  {
    "id": "x-2108589130422636748",
    "title": "沙盘里涌起的一道沙浪",
    "titleEn": "A wave of sand rises in a miniature tray",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "木框沙盘里，小耙子划出平行线，细沙随后隆起成波浪，在盘中缓缓推移。 原生视频规格为 15 秒 · 3456×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A small rake draws parallel lines in a wooden sand tray before the sand rises into a wave and slowly moves across it. Native-video output: 15s · 3456×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108589130422636748",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-09T16:03:19.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108589130422636748.mp4",
    "posterUrl": "/posters/x/2108589130422636748.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Raked the sand. Tide came in anyway.\n\nPrompt: \"the waves are made of sand\"\n\nDuration: 15s. Camera: locked, tray height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe prompt asks for four small waves, one in each raked line. The model sent a single wave of sand across the whole tray instead, and that take stays.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 12,
      "views": 301,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: small wooden rake draws lines in sand tray beside stone.\",\"Middle: sand ridges rise after rake moves away.\",\"End: a broad sand wave crosses tray.\"]"
  },
  {
    "id": "x-2108588232778846431",
    "title": "旋转星空下的油画小镇",
    "titleEn": "A painted village beneath a swirling starry sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝色油画小镇与黄色星光铺满画面，旋转笔触掠过夜空，深绿色树形在星光中起伏。 原生视频规格为 31 秒 · 1920×840。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue painted village and yellow stars fill the frame as swirling brushstrokes move through the night sky and a dark green tree shape rises beneath them. Native-video output: 31s · 1920×840. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/crazycapmie/status/2108588232778846431",
    "sourceLabel": "X 原帖 · @crazycapmie",
    "author": "CapCapMie",
    "publishedAt": "2026-10-09T15:59:45.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108588232778846431.mp4",
    "posterUrl": "/posters/x/2108588232778846431.jpg",
    "duration": 31,
    "aspectRatio": "landscape",
    "resolution": "1920×840",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3\nh3.c 本地部署\n这随手设定的尺寸放手机上横过来看会非常惊艳",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 49,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: blue oil-painted village beneath swirling yellow stars.\",\"Middle: dark green tree silhouette and animated painted sky.\",\"End: creator credit card; visible story content remains painted night landscape.\"]"
  },
  {
    "id": "x-2108586097119039860",
    "title": "火柴人的起床健身日常",
    "titleEn": "A stick figure starts a morning fitness routine",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白色圆头火柴人从床上起身，在健身房面对器械，最后坐回沙发拿着笔和纸思考新的计划。 原生视频规格为 61 秒 · 720×1260。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white round-headed stick figure gets out of bed, faces gym equipment and later sits on a sofa with a pen and paper to consider a new plan. Native-video output: 61s · 720×1260. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AbidShariarSa/status/2108586097119039860",
    "sourceLabel": "X 原帖 · @AbidShariarSa",
    "author": "Abid Shariar Sakib",
    "publishedAt": "2026-10-09T15:51:15.000Z",
    "addedAt": "2026-10-10T04:14:41.558Z",
    "mediaUrl": "/media/x-2108586097119039860.mp4",
    "posterUrl": "/posters/x/2108586097119039860.jpg",
    "duration": 61,
    "aspectRatio": "portrait",
    "resolution": "720×1260",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This whole video created via minimax h3 free local video generator AI model. \n\nThe quality is very good but not the best one.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-10-10T04:14:41.558Z"
    },
    "approvedAt": "2026-10-10T04:14:41.558Z",
    "editorialBasis": "[\"Beginning: stick figure in bed beside 7:00 alarm clock.\",\"Middle: red-headband stick figure beside gym weights.\",\"End: same figure writes on sofa beside orange cat and MY NEW PLAN text.\"]"
  },
  {
    "id": "x-2108578189253742775",
    "title": "便利店里挑选零食的女子",
    "titleEn": "Choosing a snack in a sunlit convenience store",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暖色夕阳穿过便利店，两侧货架间的女子转身比较零食，最后举着黄色包装望向镜头。 原生视频规格为 15 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Warm sunset light fills a convenience-store aisle as a woman turns between shelves to compare snacks, ending with a yellow bag in hand. Native-video output: 15s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15 seconds, 16:9, realistic cinematic image-to-video. Use <Picture 1> as the first frame and only reference for the same adult female lead, her face, pinned-up hair with loose strands, fitted dark gray long-sleeve outfit, the double snack shelves on both sides of the convenience-store aisle, and the sunset behind her. The whole video stays inside this same aisle only. The camera always remains on one front-side of the aisle facing her. Left and right shelves never swap, and the sunset always stays behind her.\n\nProp and staging continuity is strict. At start, the yellow chips bag from <Picture 1> is in her right hand. From 0.0-2.5s, she places that exact yellow bag on the left shelf, middle level, near the aisle edge, front facing the aisle, with the same folded corner at the opening clearly visible. From 2.5-7.5s, she uses her right hand to take a red snack bag from the middle level of the right shelf for comparison, then returns it to the same original position. From 7.5-11.6s, she goes back to the left shelf and retrieves the same yellow bag with the same folded corner. From 11.6s to the end, the yellow bag stays in her right hand. Her left hand only braces the shelf edge or lightly taps packaging, never carries a bag. Movement path: aisle center -> left shelf -> across to right shelf -> back to left shelf -> two steps forward toward camera.\n\nHer performance is picky but calm, with crisp, deliberate actions. She first complains softly to the yellow bag, then seriously compares the red bag, then admits she still wants the yellow bag back. Her gaze always lands on the yellow bag’s folded corner before her hand reaches for it. When she takes it back, brow tension releases and the pressed mouth gradually becomes a lightly teasing smile. She speaks only three natural English lines with accurate lip sync: <d>[English] It's all air. </d> <d>[English] And you're better? </d> <d>[English] Fine. You win. </d> The last line is said softly to the yellow bag.\n\nKeep the warm golden backlight of <Picture 1>: amber-gold highlights, warm peach skin, dark coffee-gray shadows, charcoal-gray clothing. The yellow bag is the main color anchor. Ceiling lights add only slight cool rim separation. Highlights stay soft, darks keep detail, skin texture stays real, and light direction and color temperature remain continuous.\n\n[0.0-2.5s] Fast start, frontal follow, sensual close moment 1. From the first frame she is already turning toward the left shelf and stepping to it while rotating the yellow bag in her right hand toward the backlight. The camera retreats and drifts slightly frame-left, keeping her face, chest, and the yellow bag in one frame. During the turn and place action, the chest under the fitted fabric makes one natural sideward swing with a small vertical rebound; loose strands lag slightly behind the head. She says softly, <d>[English] It's all air. </d> Her right hand places the yellow bag firmly onto the left middle shelf, confirms the front faces outward, then releases it. The shot must clearly preserve the fixed shelf position and the folded opening corner.\n\n[2.5-5.8s] Faster lateral move. She turns across the aisle and walks to the right shelf. The camera slides laterally along the front side, briefly keeping the yellow bag visible at frame edge, then follows her right hand to the right shelf. She takes one red bag and glances across the aisle toward the yellow bag on the left shelf. She lightly squeezes the red bag, lifts one brow, and says, <d>[English] And you're better? </d> Side-front close framing keeps her face, upper body, and red bag together. The step, bend, and rise create a second natural chest motion with corresponding fabric rebound.\n\n[5.8-7.5s] Short collision, rebound deceleration. She places the red bag back in the same original middle position on the right shelf. One bag corner lightly taps the metal guard rail with a short click, two adjacent bags sway slightly, and a loose corner of a transparent price-tag guard flexes, forms one fine crack, then a tiny corner fragment drops onto the lower shelf and comes to rest. The camera follows the place-back motion in a short glide, passes slightly, then eases back to stability. She glances once at the collision point with only a light eyelid flicker. Her right hand is already empty, and she immediately turns to look back toward the left shelf. The package sway gradually stops.\n\n[7.5-11.6s] Noticeably slower, gentle push-in, rack focus, sensual close moment 3. Walking back to the left shelf, she keeps looking at the yellow bag she placed earlier. Focus begins on her empty right hand in foreground, then transfers smoothly to the outward-facing yellow bag on the left middle shelf and the same folded opening corner. The camera slowly pushes in. She retrieves it with her right hand, brushes the folded corner, and lifts it beside her chest. The take-back, straightening, and slower breathing create a small residual chest tremor that settles into two natural breaths; clothing and loose hair gradually calm. Only after lifting her eyes above the bag does the smile appear. Remove the earlier quick rhythm here and leave mainly breathing, package friction, and store ambience.\n\n[11.6-15.0s] Light closing beat, camera pulls back. Her left hand lightly taps the yellow bag now back in her right hand. She says softly, <d>[English] Fine. You win. </d> Then she walks two steps toward the camera. The camera pulls back into a medium shot that clearly includes both shelves, showing the original left-shelf spot is now empty, the red bag remains in its original place on the right shelf, and she is holding the same yellow bag in her right hand. Loose strands and clothing move naturally with the steps, and the sunset remains behind her. End with one light crisp package sound.\n\nEnvironmental and physical continuity is strict. The original white car shape outside the window passes slowly behind the glass. Only the touched packages and the shelf guard rail move. The small price-tag crack follows the exact order: bag corner taps -> transparent guard flexes -> edge cracks -> tiny corner piece falls to the lower shelf and stops. Footsteps, package sounds, and the light metallic tap carry the faster rhythm, then slow noticeably when she takes back the yellow bag. The yellow bag’s appearance, folded corner, placement position, retrieval, and right-hand carry state remain continuous. Face, fingers, chest volume, neckline, and zipper position stay stable. No new characters, no scene cuts, no narration, and no subtitles.\n\noverall_soundscape\n\nOne continuous convenience-store ambience runs throughout: low HVAC hum, faint refrigerator tone, soft shelf resonance, distant outside traffic wash through the glass, and one slow white car shape passing outside. In 0.0-2.5s, emphasize her first steps, the soft crinkle of the yellow bag turning in her right hand, clothing friction, and the gentle placement onto the left shelf. Her line <d>[English] It's all air. </d> is quiet, dry, and fully lip-synced. In 2.5-5.8s, keep aisle footsteps, package rustle, the lift and squeeze of the red bag, and her line <d>[English] And you're better? </d> with clear casual delivery. In 5.8-7.5s, keep the exact ordered sound chain: red bag set down -> brief metal guard click -> slight neighboring package sway -> transparent price-tag guard flex -> one fine crack -> tiny corner fragment drops onto the lower shelf and settles. In 7.5-11.6s, slow the pace and foreground her breathing, the soft friction of her right hand taking back the yellow bag, and faint store room tone. In 11.6-15.0s, add the light tap of her left hand on the yellow bag, two forward steps, and the final soft line <d>[English] Fine. You win. </d> addressed to the bag. End on one crisp package sound. No narration, no subtitles, no extra voices.\n\nnon_diegetic_music\n\nUse no obvious score, or only a very restrained minimal rhythmic bed below the physical sounds. Let footsteps, packaging, breath, and the small metal and plastic shelf sounds carry the structure. The first half may have a faint pulse, then the rhythm noticeably relaxes when she retrieves the yellow bag, leaving a softer intimate close.",
    "promptSourceUrl": "https://x.com/ou_zhen599/status/2108578532905615715",
    "sourceUrl": "https://x.com/ou_zhen599/status/2108578189253742775",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-10-09T15:19:50.000Z",
    "addedAt": "2026-10-10T04:14:41.559Z",
    "mediaUrl": "/media/x-2108578189253742775.mp4",
    "posterUrl": "/posters/x/2108578189253742775.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                                                                                                    \n\n🎬Image-to-Video      \n📷“The Next Bag”\nAs the setting sun shone through the convenience store shelves, my discerning gaze suddenly paused when I reached for the next bag—shifting from disdain to a flutter of excitement.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 612,
      "snapshotAt": "2026-10-10T04:14:41.559Z"
    },
    "approvedAt": "2026-10-10T04:14:41.559Z",
    "editorialBasis": "[\"Beginning: adult woman faces left snack shelf in backlit convenience-store aisle.\",\"Middle: she turns toward opposite shelf to compare packages.\",\"End: she holds yellow snack bag and looks toward camera; both shelves retain positions.\"]"
  },
  {
    "id": "x-2108577916628205964",
    "title": "南瓜灯旁的魔女生日庆祝",
    "titleEn": "A witch birthday celebration beside pumpkin lanterns",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "南瓜灯与墓园背景中，白发魔女踏步后对镜头微笑，生日祝福文字出现在画面一角。 原生视频规格为 11 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A white-haired witch steps through a pumpkin-lit graveyard and smiles toward the camera as a birthday greeting appears in the corner. Native-video output: 11s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SushiNFTart/status/2108577916628205964",
    "sourceLabel": "X 原帖 · @SushiNFTart",
    "author": "🍣お寿司職人｜AIゲーム・アニメ",
    "publishedAt": "2026-10-09T15:18:45.000Z",
    "addedAt": "2026-10-10T04:14:41.559Z",
    "mediaUrl": "/media/x-2108577916628205964.mp4",
    "posterUrl": "/posters/x/2108577916628205964.jpg",
    "duration": 11,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "本日10月10日（世界ゾンビの日）は呂屯のお誕生日🎂🎉\n今年は #月蝕綺譚 のAIアニメでお祝いです🥳🌕\n月末配信予定のハロウィンイベント応援の気持ちも込めて制作しました🎃✨\nあるじどのへのボイスメッセージ付きです💌\n\n【制作ツール】\nキャラクター：月蝕綺譚／呂屯（ハロウィン）\n画像：GPT Image 2.5\n動画：MiniMax H3 Max Turbo\nBGM：ElevenLabs\n語りかけアニメ：VEED Fabric\n音声：Irodori TTS\n編集：GPT-6 Astra\n\n※ガイドラインに基づいた非公式二次創作です。",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 21,
      "views": 3092,
      "snapshotAt": "2026-10-10T04:14:41.559Z"
    },
    "approvedAt": "2026-10-10T04:14:41.559Z",
    "editorialBasis": "[\"Beginning: black platform shoes, orange-striped socks and pumpkin lights.\",\"Middle: white-haired witch smiles, Happy Birthday text.\",\"End: same witch in orange-and-black hat facing camera among graveyard pumpkins.\"]"
  },
  {
    "id": "x-2108555653539070459",
    "title": "像梳头一样整理草坪",
    "titleEn": "Combing a lawn into neat rows",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "巨大的木梳从草坪上缓缓扫过，草叶沿梳齿倒向整齐的纹路，最后留下一簇翘起的草。 原生视频规格为 15 秒 · 3456×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A giant wooden comb slowly passes over a lawn, arranging grass into neat rows before leaving a tuft standing upright. Native-video output: 15s · 3456×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108555653539070459",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-09T13:50:17.000Z",
    "addedAt": "2026-10-10T04:14:41.559Z",
    "mediaUrl": "/media/x-2108555653539070459.mp4",
    "posterUrl": "/posters/x/2108555653539070459.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Lawn's done. Side part this week.\n\nPrompt: \"the long grass parts and lies down behind it in neat straight rows like combed hair\"\n\nDuration: 15s. Camera: locked, low corner. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe cowlick at the end is the only joke. Everything before it is one slow clean pass.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 14,
      "views": 429,
      "snapshotAt": "2026-10-10T04:14:41.559Z"
    },
    "approvedAt": "2026-10-10T04:14:41.559Z",
    "editorialBasis": "[\"Beginning: giant wooden comb touches lawn.\",\"Middle: grass laid down in parallel rows behind moving comb.\",\"End: tuft of grass stands upright beside comb.\"]"
  },
  {
    "id": "x-2108555227356102849",
    "title": "从计算教室到月球的插画旅程",
    "titleEn": "An illustrated journey from calculation rooms to the Moon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暖色插画串联教室中的计算者、月球旁的飞船与书桌前的女子，形成带字幕的长篇视觉叙事。 原生视频规格为 183 秒 · 540×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Warm-toned illustrations connect people calculating in a classroom, a spacecraft beside the Moon and a woman writing at a desk in a long captioned visual story. Native-video output: 183s · 540×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mikumiku_aloha/status/2108555227356102849",
    "sourceLabel": "X 原帖 · @mikumiku_aloha",
    "author": "mikumiku_aloha",
    "publishedAt": "2026-10-09T13:48:35.000Z",
    "addedAt": "2026-10-10T04:14:41.559Z",
    "mediaUrl": "/media/x-2108555227356102849.mp4",
    "posterUrl": "/posters/x/2108555227356102849.jpg",
    "duration": 183,
    "aspectRatio": "portrait",
    "resolution": "540×960",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "尻Pさんが話題にされているアニメ化ではないけれど、アイデアをパクらせてもらったMusic Video。\n\nChatGPT-6 Dotsでアイデア練って、ClaudeのOpus 5.5に監督させてMiniMax H3を使って生成",
    "engagement": {
      "replies": 2,
      "reposts": 3,
      "likes": 2,
      "views": 1009,
      "snapshotAt": "2026-10-10T04:14:41.559Z"
    },
    "approvedAt": "2026-10-10T04:14:41.559Z",
    "editorialBasis": "[\"Beginning: illustrated women at classroom desks, caption when computer was a job title.\",\"Middle: illustrated spacecraft next to cratered Moon.\",\"End: bespectacled woman writes at desk in geometric illustration style.\"]"
  },
  {
    "id": "x-2108554060815352144",
    "title": "云海与山巅上的古筝演奏",
    "titleEn": "Zither performance above misty mountain peaks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白衣女子在云雾间弹奏古筝，宽景展示悬于山巅的平台，近景则跟随手指拨弦。 原生视频规格为 244 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in white plays a zither among clouds, with wide views of a platform above mountain peaks and close-ups of fingers plucking strings. Native-video output: 244s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ricciimagine/status/2108554060815352144",
    "sourceLabel": "X 原帖 · @ricciimagine",
    "author": "Ricci👁",
    "publishedAt": "2026-10-09T13:43:57.000Z",
    "addedAt": "2026-10-10T04:14:41.559Z",
    "mediaUrl": "/media/x-2108554060815352144.mp4",
    "posterUrl": "/posters/x/2108554060815352144.jpg",
    "duration": 244,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "OPUS 5.5配合comfyui太棒了，省去很多成本。全程我没用任何剪辑工具，音乐是乐队Rin'的《Eternal》，我把音乐给到claude,opus打开midjourney网站生图，然后连接我的comfyui，使用官方minimax h3图生视频模板自动生成视频，全程我就只是和它对话它自己完善。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 112,
      "snapshotAt": "2026-10-10T04:14:41.559Z"
    },
    "approvedAt": "2026-10-10T04:14:41.559Z",
    "editorialBasis": "[\"Beginning: woman in white beside zither amid soft mist.\",\"Middle: same woman on rocky platform above cloud sea.\",\"End: close-up of fingers plucking zither strings.\"]"
  },
  {
    "id": "x-2108546769416061164",
    "title": "足球进球镜头中的人物替换",
    "titleEn": "A character swap in a football scoring sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右画面对照足球比赛中的带球镜头，右侧人物被替换为穿深色西装、红色领带的男子，仍沿着同一段运动轨迹移动。 原生视频规格为 17 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Side-by-side football shots compare the original dribble with a dark-suited man wearing a red tie, moving along the same sequence. Native-video output: 17s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelailabs/status/2108546769416061164",
    "sourceLabel": "X 原帖 · @pixelailabs",
    "author": "PixelAiLabs",
    "publishedAt": "2026-10-09T13:14:59.000Z",
    "addedAt": "2026-10-10T04:32:28.118Z",
    "mediaUrl": "/media/x-2108546769416061164.mp4",
    "posterUrl": "/posters/x/2108546769416061164.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "We said October 9. Here’s where it stands.\n\nMiniMax H3 character swap: Özil’s goal vs Ludogorets, but it’s Trump on the ball. Same frames, same motion.\n\nThis workflow is going into Ultimate AI Influencer Pro. The video tutorial is being recorded and edited now, so it’ll be a little late. Thanks for waiting.\n\nRuns in ComfyUI, free on a cloud GPU or on your own NVIDIA GPU ⬇️⬇️⬇️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 102,
      "snapshotAt": "2026-10-10T04:32:28.118Z"
    },
    "approvedAt": "2026-10-10T04:32:28.118Z",
    "editorialBasis": "[\"Beginning: football dribble shown in paired original and replacement panels.\",\"Middle: suited character continues the run; right panel explicitly labeled H3 SWAP.\",\"End: green geometric outro.\"]"
  },
  {
    "id": "x-2108543101010858214",
    "title": "废墟与控制室之间的动画冒险",
    "titleEn": "An animated adventure through ruins and a control room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长篇动画蒙太奇从钢琴旁的装备人物切入，转向布满操作台的控制室，结尾以成人与孩子的简笔插画收束。 原生视频规格为 180 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated montage moves from an equipped figure beside a piano to a control room lined with consoles, ending with a sketch-like encounter between an adult and a child. Native-video output: 180s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HajileIbushi/status/2108543101010858214",
    "sourceLabel": "X 原帖 · @HajileIbushi",
    "author": "Hajile Ibushi",
    "publishedAt": "2026-10-09T13:00:24.000Z",
    "addedAt": "2026-10-10T04:32:28.118Z",
    "mediaUrl": "/media/x-2108543101010858214.mp4",
    "posterUrl": "/posters/x/2108543101010858214.jpg",
    "duration": 180,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Finally finished the Arc Raiders AMV\n\nCreated using Minimax H3\n\nAudio is \"Determination\" by Alex Morgan",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 85,
      "snapshotAt": "2026-10-10T04:32:28.118Z"
    },
    "approvedAt": "2026-10-10T04:32:28.118Z",
    "editorialBasis": "[\"Beginning: equipped character and wooden piano in a worn interior.\",\"Middle: small stylized character in an industrial control room with console banks.\",\"End: sketch-like adult seated on a chair facing a white-haired child.\"]"
  },
  {
    "id": "x-2108536552993570921",
    "title": "放学后的暮色与校园光影",
    "titleEn": "After-school twilight and shifting classroom light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画音乐短片先展示窗边的几何摆件与日光，再切入河畔小镇的晚霞，最后由站在夕阳中的制服女孩和片名收束。 原生视频规格为 34 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime music clip opens with a geometric object in window light, moves to sunset over a riverside town, and closes with a uniformed girl beneath the title. Native-video output: 34s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/asagilf/status/2108536552993570921",
    "sourceLabel": "X 原帖 · @asagilf",
    "author": "asagi",
    "publishedAt": "2026-10-09T12:34:23.000Z",
    "addedAt": "2026-10-10T04:32:28.118Z",
    "mediaUrl": "/media/x-2108536552993570921.mp4",
    "posterUrl": "/posters/x/2108536552993570921.jpg",
    "duration": 34,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sunoで音楽作るの楽しい‼\n\n夏の終わりの放課後。\n\n「放課後スペクトル」30秒MV\n\n曲：Suno ／ 映像：Minimax H3\n\n#SunoAI #AIMV #AIアニメ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 149,
      "snapshotAt": "2026-10-10T04:32:28.118Z"
    },
    "approvedAt": "2026-10-10T04:32:28.118Z",
    "editorialBasis": "[\"Beginning: geometric white ornament beside a sunlit classroom window.\",\"Middle: riverside town under a pastel sunset sky with Japanese lyrics.\",\"End: school-uniform girl seen from behind in sunset light with title lettering.\"]"
  },
  {
    "id": "x-2108532586624757951",
    "title": "粉色拼贴中的时尚造型切换",
    "titleEn": "Fashion cuts against a pink collage backdrop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "真人风格女孩穿白色上衣与牛仔裙出镜，条纹、棋盘格与粉色平面图形交替铺开，杂志式大字与不同站姿组成快节奏造型展示。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A realistic girl in a white top and denim skirt appears against alternating stripes, checkerboards and pink graphics, with magazine-style lettering and changing poses. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/molyin_com/status/2108532586624757951",
    "sourceLabel": "X 原帖 · @molyin_com",
    "author": "Molyin",
    "publishedAt": "2026-10-09T12:18:37.000Z",
    "addedAt": "2026-10-10T04:32:28.118Z",
    "mediaUrl": "/media/x-2108532586624757951.mp4",
    "posterUrl": "/posters/x/2108532586624757951.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sharing MiniMax H3 Prompts on Molyin\n15s Korean fashion MV — pink pattern collage pop, photorealistic character lock, seamless outfit changes.\nStruggling to keep the same face in AI video? This prompt nails identity consistency: same girl, same proportions, zero face drift across 8 cuts. Combined with flat 2D graphics and readable typography, it creates a premium magazine aesthetic without the usual AI glitches. Perfect for virtual influencers, brand lookbooks, and social media campaigns.\nFull prompt in replies 👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 96,
      "snapshotAt": "2026-10-10T04:32:28.118Z"
    },
    "approvedAt": "2026-10-10T04:32:28.118Z",
    "editorialBasis": "[\"Beginning: long-haired girl beside a black-and-white striped panel.\",\"Middle: pink collage graphics, checkerboards and large CUTE SIGNAL lettering.\",\"End: fashion-cut typography with the same girl posing in a white top and denim skirt.\"]"
  },
  {
    "id": "x-2108787981863502105",
    "title": "石殿中的红光交锋",
    "titleEn": "Crimson clash in a stone hall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "本地 MiniMax H3 与 ComfyUI 制作的奇幻短片：石厅中的角色对峙、红色武器特写与战斗收尾。 原生视频规格为 15 秒 · 2112×1216。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A local MiniMax H3 and ComfyUI fantasy clip moving from a torchlit stone-hall face-off to a glowing weapon close-up. Native-video output: 15s · 2112×1216. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wraithworksprod/status/2108787981863502105",
    "sourceLabel": "X 原帖 · @wraithworksprod",
    "author": "Wraith",
    "publishedAt": "2026-10-10T05:13:28.000Z",
    "addedAt": "2026-10-10T07:28:55.516Z",
    "mediaUrl": "/media/x-2108787981863502105.mp4",
    "posterUrl": "/posters/x/2108787981863502105.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2112×1216",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fantasy",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Second local video-generation run with MiniMax H3 through ComfyUI—this time with an upscale pass. The image quality and atmosphere are improving, though temporal consistency remains the biggest challenge. Still experimenting, learning, and refining the workflow.\n#AIVideo #MiniMaxH3 #ComfyUI #LocalAI #GenerativeAI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 50,
      "snapshotAt": "2026-10-10T07:28:55.516Z"
    },
    "approvedAt": "2026-10-10T07:28:55.516Z",
    "editorialBasis": "Original caption explicitly identifies local MiniMax H3 through ComfyUI; the accepted browser receipt verifies the exact source, timestamp, and playable main-post video. Begin, middle, and end frames were visually inspected."
  },
  {
    "id": "x-2108776379588428095",
    "title": "工作站前的动画创作者",
    "titleEn": "Anime Creators at the Workstation",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "MiniMax H3 Max Turbo 生成的日常动画片段，描绘创作者在多屏工作站前制作视频。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A MiniMax H3 Max Turbo everyday-anime clip showing creators working at a multi-screen video-editing station. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2108776379588428095",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-10-10T04:27:22.000Z",
    "addedAt": "2026-10-10T07:28:55.516Z",
    "mediaUrl": "/media/x-2108776379588428095.mp4",
    "posterUrl": "/posters/x/2108776379588428095.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIart #MiniMaxH3MaxTurbo \nH3 Max Turbo、こんな日常アニメも作れるんか…\n（MiniMax H3 Max Turbo 出力動画ポン出し）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 121,
      "snapshotAt": "2026-10-10T07:28:55.516Z"
    },
    "approvedAt": "2026-10-10T07:28:55.516Z",
    "editorialBasis": "Original caption explicitly calls this a MiniMax H3 Max Turbo output video; the accepted browser receipt verifies the exact source, timestamp, and playable main-post video. Three inspected frames show the workstation scene."
  },
  {
    "id": "x-2108774517380436221",
    "title": "蓝衣少女与晴空",
    "titleEn": "Blue-clad girl beneath a bright sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "MiniMax H3 360° Orbit LoRA 的 2D 动画示例，以竖屏角色画面展示环绕效果。 原生视频规格为 5 秒 · 1088×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A 2D animation example from a MiniMax H3 360° Orbit LoRA, shown in a vertical character composition. Native-video output: 5s · 1088×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2108774517380436221",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-10-10T04:19:58.000Z",
    "addedAt": "2026-10-10T07:28:55.516Z",
    "mediaUrl": "/media/x-2108774517380436221.mp4",
    "posterUrl": "/posters/x/2108774517380436221.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1088×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3 360° Orbit LoRAの他の作例。2Dアニメでもいい感じの360度動画ができるので便利ー。",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 16,
      "views": 1167,
      "snapshotAt": "2026-10-10T07:28:55.516Z"
    },
    "approvedAt": "2026-10-10T07:28:55.516Z",
    "editorialBasis": "Original caption explicitly identifies MiniMax-H3 360° Orbit LoRA examples; the accepted browser receipt verifies the exact source, timestamp, and playable video in the original post. The first video’s three frames were visually inspected."
  },
  {
    "id": "x-2108774178858123662",
    "title": "蓝裙少女抬手拢发",
    "titleEn": "Blue-dressed girl brushes back her hair",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一段使用 MiniMax H3 360° Orbit LoRA 的竖屏 2D 动画测试；原帖注明 1088×1920、5 秒及 RTX 4070 生成耗时。 原生视频规格为 5 秒 · 1088×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical 2D animation test with MiniMax H3 360° Orbit LoRA; the post reports 1088×1920, five seconds, and RTX 4070 generation timing. Native-video output: 5s · 1088×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2108774178858123662",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-10-10T04:18:38.000Z",
    "addedAt": "2026-10-10T07:28:55.516Z",
    "mediaUrl": "/media/x-2108774178858123662.mp4",
    "posterUrl": "/posters/x/2108774178858123662.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1088×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 360° Orbit LoRAを試してみた！以前試した別のLoRAよりいい感じ。高速化ワークフローを使ったので、1088x1920・5秒124fで84秒くらい（RTX4070）。動画自体の時間が長いからか2回転していますー。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 21,
      "views": 1013,
      "snapshotAt": "2026-10-10T07:28:55.516Z"
    },
    "approvedAt": "2026-10-10T07:28:55.516Z",
    "editorialBasis": "Original caption explicitly names MiniMax H3 360° Orbit LoRA and reports the video settings; the accepted browser receipt verifies the exact source, timestamp, and playable original-post video. The first video’s three frames were visually inspected."
  },
  {
    "id": "x-2108754878038171689",
    "title": "烛光石厅中的对峙",
    "titleEn": "A Standoff in a Candlelit Hall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发女子与鹿角怪物在烛光石厅中对峙，镜头切到发光红色饰物，再回到二者相望的宽景。原帖注明使用 MiniMax H3 与 ComfyUI。 原生视频规格为 15 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired woman faces a horned creature in a candlelit stone hall. The clip cuts to a glowing red ornament, then returns to their standoff. The post identifies MiniMax H3 used locally through ComfyUI. Native-video output: 15s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wraithworksprod/status/2108754878038171689",
    "sourceLabel": "X 原帖 · @wraithworksprod",
    "author": "Wraith",
    "publishedAt": "2026-10-10T03:01:56.000Z",
    "addedAt": "2026-10-10T07:37:44.929Z",
    "mediaUrl": "/media/x-2108754878038171689.mp4",
    "posterUrl": "/posters/x/2108754878038171689.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "15-second video-generation experiment using MiniMax H3 locally through ComfyUI. Impressive atmosphere and motion, though temporal consistency still needs work—especially faces, hands, costumes, and fine details. Early test, but local AI video is moving fast. 🎬\n#AIvideo #ComfyUI #MiniMaxH3 #GenerativeAI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 64,
      "snapshotAt": "2026-10-10T07:37:44.929Z"
    },
    "approvedAt": "2026-10-10T07:37:44.929Z",
    "editorialBasis": "The original caption explicitly identifies a local MiniMax H3 generation through ComfyUI. The exact original post contains one playable native video. Beginning, middle, and end frames were inspected; no model mark is visible."
  },
  {
    "id": "x-2108754394527465747",
    "title": "霓虹字幕下的银发少女",
    "titleEn": "A Silver-Haired Woman Amid Neon Lyrics",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "日语歌曲片段在动画人物、歌词字幕与歌曲标题之间切换；末尾以黑底霓虹字样呈现歌名。原帖将影像归于 MiniMax H3。 原生视频规格为 203 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A Japanese song sequence cuts between an animated woman, oversized lyric captions, and the song title, which returns in neon lettering at the end. The post attributes the visuals to MiniMax H3. Native-video output: 203s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/maru_999_/status/2108754394527465747",
    "sourceLabel": "X 原帖 · @maru_999_",
    "author": "Maru999",
    "publishedAt": "2026-10-10T03:00:01.000Z",
    "addedAt": "2026-10-10T07:37:44.930Z",
    "mediaUrl": "/media/x-2108754394527465747.mp4",
    "posterUrl": "/posters/x/2108754394527465747.jpg",
    "duration": 203,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "悪い微笑み カウントダウン フルMV\nちょいワル版のゆみと、コンテゴさん(@kontego_Low_fi)のフルール\n\n曲: Suno / 映像: MiniMax H3 / 設定画・タイトル: ChatGPT",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 56,
      "snapshotAt": "2026-10-10T07:37:44.930Z"
    },
    "approvedAt": "2026-10-10T07:37:44.930Z",
    "editorialBasis": "The original caption labels the video as a full music video and explicitly attributes its visuals to MiniMax H3. The exact original post contains one playable native video. Beginning, middle, and end frames were inspected; no model mark is visible."
  },
  {
    "id": "x-2108753741340152278",
    "title": "粉色卧室与舞台灯光",
    "titleEn": "Pink Bedroom and Stage Lights",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一位戴星形发夹的黑发女子先出现在粉色卧室，随后在环形灯下化妆，最后切到带有 LUDIC EXISTENCE 字样的舞台画面。原帖链接到 MiniMax-H3 Realism People LoRA。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A black-haired woman with star clips appears in a pink bedroom, then applies makeup under round lights before a final stage-like shot with the words LUDIC EXISTENCE. The post links a MiniMax-H3 Realism People LoRA. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/superfang119/status/2108753741340152278",
    "sourceLabel": "X 原帖 · @superfang119",
    "author": "superfang119",
    "publishedAt": "2026-10-10T02:57:25.000Z",
    "addedAt": "2026-10-10T07:37:44.930Z",
    "mediaUrl": "/media/x-2108753741340152278.mp4",
    "posterUrl": "/posters/x/2108753741340152278.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "测试了下。在不改变模型和提示词的情况下，增加一个lora。效果出奇的好。大家可对比下。\nhttps://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 69,
      "snapshotAt": "2026-10-10T07:37:44.930Z"
    },
    "approvedAt": "2026-10-10T07:37:44.930Z",
    "editorialBasis": "The original post links the MiniMax-H3 Realism People LoRA and describes adding a LoRA while keeping the model and prompt unchanged. Its exact native videos are playable. Both clips were inspected at beginning, middle, and end; no model mark is visible."
  },
  {
    "id": "x-2108489223099801704",
    "title": "红衣剪影与拼贴肖像",
    "titleEn": "A Red Silhouette and Collage Portraits",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一位红衣人物出现在图形拼贴前，随后画面切到时尚人像与重复排列的肖像网格，最后以 surRIHL 字标收尾。 原生视频规格为 196 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-clad figure appears against a graphic collage before the video moves through fashion portraits and a repeated portrait grid, ending on the surRIHL wordmark. Native-video output: 196s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/papadupapadu/status/2108489223099801704",
    "sourceLabel": "X 原帖 · @papadupapadu",
    "author": "PAPADU PAPADU",
    "publishedAt": "2026-10-09T09:26:19.000Z",
    "addedAt": "2026-10-10T08:14:54.175Z",
    "mediaUrl": "/media/x-2108489223099801704.mp4",
    "posterUrl": "/posters/x/2108489223099801704.jpg",
    "duration": 196,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "WE'RE ONLY COOKED IF WE IGNORE REALITY. SO WE TAUGHT IT A NEW LANGUAGE.\n\nTHE LANGUAGE OF SURREAL | A @surRIHL CUT\nDIRECTED BY: @papadupapadu + @LoLoLumi88\n\nMIDJOURNEY + HAILUO H3 + ELEVEN MUSIC + ELEVENLABS SFX + MANUAL EDITING IN PREMIERE PRO + CLAUDE OPUS 5.5",
    "engagement": {
      "replies": 8,
      "reposts": 4,
      "likes": 56,
      "views": 1988,
      "snapshotAt": "2026-10-10T08:14:54.175Z"
    },
    "approvedAt": "2026-10-10T08:14:54.175Z",
    "editorialBasis": "The original caption lists Hailuo H3 as one of the video-generation tools. The exact original post has one playable native video. Beginning, middle and end frames were reviewed; the final wordmark is not a model mark."
  },
  {
    "id": "x-2105882625843851369",
    "title": "粉色布景中的红衣演唱者",
    "titleEn": "A Red-Clad Singer on a Pink Set",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红衣女子在粉色片场拿着麦克风演唱，镜头切到她的嘴部特写，最后与一位穿浅色西装的男子对视。原帖称这是 MiniMax H3 复刻视频。 原生视频规格为 27 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in red sings with a microphone on a pink set. The clip cuts to a close-up as she sings, then ends with her facing a man in a pale suit. The post describes it as a MiniMax H3 recreation. Native-video output: 27s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/johnAGI168/status/2105882625843851369",
    "sourceLabel": "X 原帖 · @johnAGI168",
    "author": "John",
    "publishedAt": "2026-10-02T04:48:38.000Z",
    "addedAt": "2026-10-10T08:14:54.176Z",
    "mediaUrl": "/media/x-2105882625843851369.mp4",
    "posterUrl": "/posters/x/2105882625843851369.jpg",
    "duration": 27,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "网红已死😱\n\n大批量新生代 AI 网红即将诞生！\n\nby MiniMax H3 复刻视频🎬",
    "engagement": {
      "replies": 3,
      "reposts": 3,
      "likes": 20,
      "views": 3732,
      "snapshotAt": "2026-10-10T08:14:54.176Z"
    },
    "approvedAt": "2026-10-10T08:14:54.176Z",
    "editorialBasis": "The original caption attributes the video to MiniMax H3 and the single native video played. Beginning, middle and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2105881513107591391",
    "title": "太空旅人驻足海岸",
    "titleEn": "Astronauts Pause on a Rocky Shore",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名穿太空服的旅人站在明亮的岩石海岸，远处停着飞船；画面切到一位胡须宇航员的近景，再拉远展现海岸与飞船。 原生视频规格为 48 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two suited travelers stand on a bright rocky shore with a spacecraft behind them. The shot moves to a close-up of a bearded astronaut, then pulls back to reveal the coast and ship. Native-video output: 48s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GFobbit/status/2105881513107591391",
    "sourceLabel": "X 原帖 · @GFobbit",
    "author": "Geardo Fobbit",
    "publishedAt": "2026-10-02T04:44:12.000Z",
    "addedAt": "2026-10-10T08:14:54.176Z",
    "mediaUrl": "/media/x-2105881513107591391.mp4",
    "posterUrl": "/posters/x/2105881513107591391.jpg",
    "duration": 48,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Those Who Believe - A Starfield short\n\nGenerated on @Artlist_io \nImages: Nano Banana 2\nVideo: MiniMax H3 Max\nVoices: @ElevenLabs \nMusic & SFX: @geardofobbit on @suno \n\n#Starfield #Galactica",
    "engagement": {
      "replies": 11,
      "reposts": 4,
      "likes": 31,
      "views": 795,
      "snapshotAt": "2026-10-10T08:14:54.176Z"
    },
    "approvedAt": "2026-10-10T08:14:54.176Z",
    "editorialBasis": "The original source identifies MiniMax H3 Max as the video generator and Nano Banana 2 as the image source. The native video played; beginning, middle and end frames were inspected, with no different-model mark visible."
  },
  {
    "id": "x-2105873482122920017",
    "title": "雨夜舞台上的独唱",
    "titleEn": "A Soloist on a Rainy Night Stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一位戴兜帽的男子站在昏暗的高层室内，随后在红色灯光下对着麦克风演唱，最后独自留在舞台中央。 原生视频规格为 245 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hooded man stands in a shadowy high-rise interior, then sings into a microphone under red lights before remaining alone onstage. Native-video output: 245s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/h3nrixd/status/2105873482122920017",
    "sourceLabel": "X 原帖 · @h3nrixd",
    "author": "henrixd",
    "publishedAt": "2026-10-02T04:12:18.000Z",
    "addedAt": "2026-10-10T08:47:32.086Z",
    "mediaUrl": "/media/x-2105873482122920017.mp4",
    "posterUrl": "/posters/x/2105873482122920017.jpg",
    "duration": 245,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "My first take on slopcore: \"Verified Human\". This shit is fun.\n\nLyrics/style: GPT-5.6 Sol / GPT-6 Luna / Gemini 3.6 Flash / Me\nSong: Suno V6\nVideo director: GPT-6 Astra / GPT 6.1 Sol\nVideo generation: local Minimax H3\n*Claude Opus 5.5 and Seedance 2.5 were out of reach",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 275,
      "snapshotAt": "2026-10-10T08:47:32.086Z"
    },
    "approvedAt": "2026-10-10T08:47:32.086Z",
    "editorialBasis": "The caption explicitly identifies local Minimax H3 as the video-generation tool. One native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2105869619449712785",
    "title": "办公室里的神秘来客",
    "titleEn": "A Mysterious Visitor at the Office",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "办公室里，一位职员坐在桌前，男同事转身看向镜头，随后一名深色身影走出房间。原帖把影像归于 Minimax H3。 原生视频规格为 15 秒 · 1260×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In an office, an employee sits at a desk, a coworker turns toward the camera, and a dark-clad figure leaves the room. The post attributes the video to Minimax H3. Native-video output: 15s · 1260×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/viperstudios4/status/2105869619449712785",
    "sourceLabel": "X 原帖 · @viperstudios4",
    "author": "Viper Studios",
    "publishedAt": "2026-10-02T03:56:57.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2105869619449712785.mp4",
    "posterUrl": "/posters/x/2105869619449712785.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1260×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Office has a Vampire problem! Minimax H3 in @TopviewAIhq",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 231,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The original caption explicitly pairs Minimax H3 with the hosted video. Its single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2105866478490878266",
    "title": "粉色背景中的像素人物",
    "titleEn": "A Pixel Character on Magenta",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名像素风格的胡须人物手持超大的叉子，站在纯粉色背景前，镜头中只有轻微姿势变化。 原生视频规格为 4 秒 · 1024×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bearded pixel-art character holds an oversized fork against a solid magenta background, with only slight changes in pose. Native-video output: 4s · 1024×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/roryhighside/status/2105866478490878266",
    "sourceLabel": "X 原帖 · @roryhighside",
    "author": "Rory",
    "publishedAt": "2026-10-02T03:44:28.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2105866478490878266.mp4",
    "posterUrl": "/posters/x/2105866478490878266.jpg",
    "duration": 4,
    "aspectRatio": "square",
    "resolution": "1024×1024",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Finally got a good pixel art animation workflow with deroping set up for Minimax H3. Decent results that would require minimal cleanup. This is 3x temporal upscaling which beat out 2x by a lot, still some slight smudging that 4x might solve.",
    "engagement": {
      "replies": 10,
      "reposts": 0,
      "likes": 42,
      "views": 1551,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The caption explicitly describes a Minimax H3 animation workflow. The single native video played; beginning, middle, and end frames were reviewed, with no different-model mark visible."
  },
  {
    "id": "x-2105857804556959946",
    "title": "暗色水纹与蓝色漩涡",
    "titleEn": "Dark Water and a Blue Vortex",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "暗色水纹横过画面，接着明亮的蓝色浪潮环绕中央剪影，最后画面回到近乎漆黑的水面纹理。 原生视频规格为 175 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Dark water patterns cross the frame before a bright blue wave surrounds a central silhouette; the image then returns to nearly black water. Native-video output: 175s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AIArtAlchemist/status/2105857804556959946",
    "sourceLabel": "X 原帖 · @AIArtAlchemist",
    "author": "NIKKO",
    "publishedAt": "2026-10-02T03:10:00.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2105857804556959946.mp4",
    "posterUrl": "/posters/x/2105857804556959946.jpg",
    "duration": 175,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "『LUMA NEI』\nVideo : MiniMax H3 \nMusic : Suno",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 11,
      "views": 688,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The caption explicitly labels the video MiniMax H3 and separately credits Suno for music. One native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2105841123772264462",
    "title": "小车、野狼与追来的卡车",
    "titleEn": "A Wolf, a Small Car, and a Pursuing Truck",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一只狼开着红色敞篷车驶过林间道路，回头时发现一辆大卡车紧追在后。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A wolf drives a red convertible along a wooded road and looks back to find a large truck following close behind. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ShamiWeb3/status/2105841123772264462",
    "sourceLabel": "X 原帖 · @ShamiWeb3",
    "author": "Shami",
    "publishedAt": "2026-10-02T02:03:43.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2105841123772264462.mp4",
    "posterUrl": "/posters/x/2105841123772264462.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Never splash a bunny with a monster truck.\n\nA wolf, a tiny convertible, and one very petty rabbit. Revenge has never been this big (or this wet). 💦\n\nMade with Minimax H3 on @TapNow_AI",
    "engagement": {
      "replies": 72,
      "reposts": 27,
      "likes": 355,
      "views": 23961,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The caption says the video was made with Minimax H3 on TapNow. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2108731740923859279",
    "title": "昏暗训练场中的双人对打",
    "titleEn": "A Fight in a Dim Training Room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名穿运动服的拳手在昏暗训练场中交手，画面从拳击近景转为两人对峙的宽景。 原生视频规格为 6 秒 · 1408×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two athletes spar in a dim training room, moving from a close punch to a wider view of both fighters facing off. Native-video output: 6s · 1408×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/keyisteo/status/2108731740923859279",
    "sourceLabel": "X 原帖 · @keyisteo",
    "author": "Teo",
    "publishedAt": "2026-10-10T01:30:00.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2108731740923859279.mp4",
    "posterUrl": "/posters/x/2108731740923859279.jpg",
    "duration": 6,
    "aspectRatio": "landscape",
    "resolution": "1408×608",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "pretty much Minimax H3 Singularity fix MANY THINGS in motion video. Damn those higgsfield subscription looks so mahal compared to this",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 62,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The caption explicitly identifies Minimax H3 Singularity. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2108712981173621012",
    "title": "厨房里的焦黑西兰花",
    "titleEn": "Charred Broccoli in the Kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画女子在明亮厨房里举起焦黑的西兰花，随后画面切到夜色中举杯的歌者。 原生视频规格为 209 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated woman in a bright kitchen raises a charred broccoli floret before the video cuts to a singer holding a drink at night. Native-video output: 209s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/guutara_surume/status/2108712981173621012",
    "sourceLabel": "X 原帖 · @guutara_surume",
    "author": "ぐうたらするめ",
    "publishedAt": "2026-10-10T00:15:27.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2108712981173621012.mp4",
    "posterUrl": "/posters/x/2108712981173621012.jpg",
    "duration": 209,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "これまで動画生成はGrok Imagineを使っていたのですが、だんだんお財布が瀕死になってきました。\n\nということで、今回はローカルのMiniMax H3でMV制作に挑戦！\n\n『青春炭化ブロッコリー』🥦🍺\n\n昼酒のおつまみに乾燥ブロッコリーを作ろうとして「できらぁ！！」と啖呵を切った結果、見事に真っ黒焦げになった実話を歌にしました。\n\n今回の制作環境はこちら！\n\n🎵 楽曲：Suno\n🎨 画像生成：ChatGPT\n🎬 動画生成：MiniMax H3（ローカル）\n✨ 高画質化：Real-ESRGAN\n📝 字幕タイムライン抽出：Whisper\n🎞 MV編集：ChatGPT＋DaVinci Resolve\n\n字幕タイミングは手動で微調整しています。\nお金を節約するためにAIを総動員して、焦げたブロッコリーのMVが完成しました。\n\n技術の進歩ってすごいですねぇ。\nちなみにブロッコリーは苦かったです。\n\n#AI音楽 #AI動画",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 110,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The caption explicitly identifies local MiniMax H3 as the video-generation tool and names separate music and editing tools. One native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2103637901779349653",
    "title": "上升曲线与企业采用率",
    "titleEn": "An Adoption Figure and Rising Chart",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑底标题切换为企业采用率圆环图，最后呈现带有日期与金额标注的上升曲线。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A title on black gives way to a circular enterprise-adoption graphic, then an upward chart labeled with dates and dollar figures. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/excel_niisan/status/2103637901779349653",
    "sourceLabel": "X 原帖 · @excel_niisan",
    "author": "エクセル兄さん(たてばやし淳)@9万部「生成AI最速仕事術」",
    "publishedAt": "2026-09-26T00:08:54.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2103637901779349653.mp4",
    "posterUrl": "/posters/x/2103637901779349653.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3、このインフォグラフィックがプロンプト一発でできるのは、かなり実用的なのでは。\nビジネス用途で一番使える動画生成モデルだと思う。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 1194,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The original caption explicitly identifies Minimax H3. Its single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2103631313156673890",
    "title": "柠檬与滚落的玻璃珠",
    "titleEn": "A Lemon and Spilling Glass Beads",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "透明榨汁器压向切开的柠檬，玻璃珠从边缘倾泻而出，在大理石台面上弹跳、滚动并散开。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A clear juicer presses into a lemon as glass beads pour over its rim, bouncing and scattering across a marble surface. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "a stream of tiny clear glass beads pours out around the rim and spills onto the marble, bouncing, rolling, scattering",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2103631313156673890",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2103631313156673890",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-25T23:42:43.000Z",
    "addedAt": "2026-10-10T08:47:32.087Z",
    "mediaUrl": "/media/x-2103631313156673890.mp4",
    "posterUrl": "/posters/x/2103631313156673890.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Squeeze the lemon. Mind the beads.\n\nPrompt: \"a stream of tiny clear glass beads pours out around the rim and spills onto the marble, bouncing, rolling, scattering\"\n\nDuration: 15s. Camera: locked, board height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nTwo sounds layered, a wet squeeze under a glass tinkle. The layering tells the model both things are happening at once.",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 28,
      "views": 534,
      "snapshotAt": "2026-10-10T08:47:32.087Z"
    },
    "approvedAt": "2026-10-10T08:47:32.087Z",
    "editorialBasis": "The original caption explicitly identifies locally generated open-weight Minimax H3 output and includes the complete quoted prompt. One native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2103614251147702282",
    "title": "会动的绘本插画",
    "titleEn": "Illustrations Come to Life",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一本绘本翻开后，页面依次呈现Shelby的房间、床边插画和夜间场景，图画在书页间逐渐变化。 原生视频规格为 35 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An open storybook turns through pages showing Shelby’s room, an illustrated bed, and a nighttime scene as the pictures change. Native-video output: 35s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelrhythms/status/2103614251147702282",
    "sourceLabel": "X 原帖 · @pixelrhythms",
    "author": "Keshav",
    "publishedAt": "2026-09-25T22:34:55.000Z",
    "addedAt": "2026-10-10T08:47:32.088Z",
    "mediaUrl": "/media/x-2103614251147702282.mp4",
    "posterUrl": "/posters/x/2103614251147702282.jpg",
    "duration": 35,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made a little living storybook for Shelby & the Cuddle Warriors ✨\n\nTurn the pages at your own pace, and the illustrations come to life.\n\nInspired by the moving photographs in Harry Potter, with pictures animated using MiniMax H3.\n\nBuilt with @MengTo’s ThreeUI and help from ChatGPT-6 Astra.\n\nThank you @krea_ai @reactorworld @Hailuo_AI for the tools that helped bring this little world to life 💛",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 2,
      "views": 189,
      "snapshotAt": "2026-10-10T08:47:32.088Z"
    },
    "approvedAt": "2026-10-10T08:47:32.088Z",
    "editorialBasis": "The caption says the illustrations were animated using MiniMax H3. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2103613782467457079",
    "title": "天鹅绒上的金色细线",
    "titleEn": "A Fine Golden Line in Velvet",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手指在深绿色天鹅绒表面刷过，布料上出现一条暗色带，随后一线暖金色光从中显现。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Fingers brush dark green velvet, leaving a shadowed band before a fine line of warm golden light appears within it. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the nap flipping dark in a band behind the fingers with a soft brushed hush, and along that dark band a fine seam of warm golden light opens",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2103613782467457079",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2103613782467457079",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-25T22:33:03.000Z",
    "addedAt": "2026-10-10T08:47:32.088Z",
    "mediaUrl": "/media/x-2103613782467457079.mp4",
    "posterUrl": "/posters/x/2103613782467457079.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. One stroke across velvet. Something's behind it.\n\nPrompt: \"the nap flipping dark in a band behind the fingers with a soft brushed hush, and along that dark band a fine seam of warm golden light opens\"\n\nDuration: 15s. Camera: locked, chest height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nVelvet has a direction. Writing the nap flipping is what made the stroke visible at all.",
    "engagement": {
      "replies": 0,
      "reposts": 3,
      "likes": 19,
      "views": 391,
      "snapshotAt": "2026-10-10T08:47:32.088Z"
    },
    "approvedAt": "2026-10-10T08:47:32.088Z",
    "editorialBasis": "The caption explicitly identifies locally generated open-weight MiniMax H3 output and includes the complete quoted Prompt. One native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2108722027872108615",
    "title": "夜色厨房里的热汤",
    "titleEn": "A Pot of Soup in a Dark Kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一位浅发女子站在昏暗厨房的汤锅旁，随后用勺子搅动锅中食物，最后面向镜头露出笑容。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pale-haired woman stands beside a pot in a dim kitchen, stirs its contents with a spoon, and ends smiling toward the camera. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2108722027872108615",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-10-10T00:51:24.000Z",
    "addedAt": "2026-10-10T08:47:32.088Z",
    "mediaUrl": "/media/x-2108722027872108615.mp4",
    "posterUrl": "/posters/x/2108722027872108615.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "For everyone spending Friday night on X… myself included😂🥹dinner is served.\n\nTear Soup 🥣 with a side of unrealized plans.\n\nWho wants extra crackers… or maybe a tissue? \n\nMiniMax H3 #pixversecpp",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 12,
      "views": 207,
      "snapshotAt": "2026-10-10T08:47:32.088Z"
    },
    "approvedAt": "2026-10-10T08:47:32.088Z",
    "editorialBasis": "The caption explicitly names MiniMax H3. The single native video played; beginning, middle, and end frames were reviewed. No different-model mark is visible."
  },
  {
    "id": "x-2098250623036575810",
    "title": "白色影棚中的时装造型",
    "titleEn": "A Fashion Look in a White Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一位金发模特穿着黑色连衣裙，在白色摄影棚灯光下摆姿势；镜头短暂切到低角度，再回到全身造型。 原生视频规格为 15 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde model in a black dress poses under the lights of a white photo studio; the camera briefly shifts low before returning to a full view. Native-video output: 15s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/UncleGigaChad/status/2098250623036575810",
    "sourceLabel": "X 原帖 · @UncleGigaChad",
    "author": "Uncle GigaChad",
    "publishedAt": "2026-09-11T03:21:46.000Z",
    "addedAt": "2026-10-10T08:47:32.088Z",
    "mediaUrl": "/media/x-2098250623036575810.mp4",
    "posterUrl": "/posters/x/2098250623036575810.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sydney Sweeney fashion-shoot test in MiniMax H3, based on @aimikoda’s anime prompt choreography and reworked for live action.\n\nComfyUI + Sydney LoRA, RunPod H100 NVL 94GB, 864×480/24fps, 15s. Rendered in 9m 40s; total GPU session cost $0.79 at $3.19/hr.",
    "engagement": {
      "replies": 5,
      "reposts": 0,
      "likes": 15,
      "views": 11824,
      "snapshotAt": "2026-10-10T08:47:32.088Z"
    },
    "approvedAt": "2026-10-10T08:47:32.088Z",
    "editorialBasis": "The caption explicitly says the fashion-shoot video was made in MiniMax H3. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2098245154054476238",
    "title": "女孩与黑猫的展览海报",
    "titleEn": "A Girl and a Black Cat on an Exhibition Poster",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画女孩和黑猫在街景中出现，镜头靠近二者，最后切至介绍 Send AI Art Exhibition 日期与地点的展览海报。 原生视频规格为 15 秒 · 480×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated girl and black cat appear against a street scene; the camera moves closer before ending on a poster with Send AI Art Exhibition dates and venue. Native-video output: 15s · 480×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AI_yukichi2910/status/2098245154054476238",
    "sourceLabel": "X 原帖 · @AI_yukichi2910",
    "author": "諭吉@10/10・11 POP ALIVE!! （渋谷・原宿）",
    "publishedAt": "2026-09-11T03:00:02.000Z",
    "addedAt": "2026-10-10T08:47:32.088Z",
    "mediaUrl": "/media/x-2098245154054476238.mp4",
    "posterUrl": "/posters/x/2098245154054476238.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×640",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#sendaiart \n\n15秒コマーシャル動画\nMiniMax H3  \n『Send AI Art Exhibition』\n 東北から発信する AIアート展 \n2026年9月12日(土)・13(日) \n12:00〜18:00\n\nギャラリーボチュー \nhttps://bochu.jp \n\n【出展クリエイター】\n 城野ぱすてる \n多古と凧と凪 \nねおのらしい \nてい \nKosikake \n諭吉",
    "engagement": {
      "replies": 1,
      "reposts": 15,
      "likes": 78,
      "views": 3181,
      "snapshotAt": "2026-10-10T08:47:32.088Z"
    },
    "approvedAt": "2026-10-10T08:47:32.088Z",
    "editorialBasis": "The caption explicitly labels the 15-second commercial video MiniMax H3. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2108832507030585785",
    "title": "书页后的古装人物",
    "titleEn": "A Period Character Beyond the Page",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "浅色纸页上先出现几处墨色痕迹，随后画面转为一位身穿古装的男子坐在木屋内，末尾渐暗。 原生视频规格为 99 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Dark marks appear on a pale page before the scene shifts to a period-dressed man in a wooden room and fades toward black. Native-video output: 99s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/zhu185178/status/2108832507030585785",
    "sourceLabel": "X 原帖 · @zhu185178",
    "author": "Feicai",
    "publishedAt": "2026-10-10T08:10:24.000Z",
    "addedAt": "2026-10-10T09:43:34.689Z",
    "mediaUrl": "/media/x-2108832507030585785.mp4",
    "posterUrl": "/posters/x/2108832507030585785.jpg",
    "duration": 99,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "我用 Opus 5.5 跑通了 AI 真人仙侠剧的制作流程\n我这套用 Opus 5.5 不用画布直接制作和剪辑的短剧漫剧的流程基本成型，已经跑通了。\n\n还是一样的，视频模型用的是 MiniMax H3，然后背景音乐制作用的是 ElevenLabs。\n\n在使用 MiniMax H3 的过程当中，我发现了一个反直觉的现象：MiniMax H3 的总体制作成本其实远高于 Seedance 2.5。那这个我会在下一期正式更新的视频里，告诉你为什么。同时，我也会详细地讲解我搭建了哪些 skill、开发了哪些插件，如何用 Opus 5.5 去制作，不管是仙侠真人剧，还是之前发过的现代都市真人剧，我会统一讲解并分享我的思路。\n\n今天先看 Demo 小样",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 9,
      "views": 1173,
      "snapshotAt": "2026-10-10T09:43:34.689Z"
    },
    "approvedAt": "2026-10-10T09:43:34.689Z",
    "editorialBasis": "The original caption explicitly says the video model is MiniMax H3. The single native video played, and beginning, middle, and end frames were reviewed; no different-model mark is visible."
  },
  {
    "id": "x-2108832291401453576",
    "title": "橙色影棚中的人物造型",
    "titleEn": "A Portrait in an Orange Studio",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名穿橙色外套的男子在橙色背景中先后撑伞、倚靠南瓜摆姿势，最后切至近景肖像。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in an orange coat poses with an umbrella and beside pumpkins against an orange backdrop, ending in a close portrait. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mm06500812/status/2108832291401453576",
    "sourceLabel": "X 原帖 · @mm06500812",
    "author": "m_m",
    "publishedAt": "2026-10-10T08:09:33.000Z",
    "addedAt": "2026-10-10T09:43:34.689Z",
    "mediaUrl": "/media/x-2108832291401453576.mp4",
    "posterUrl": "/posters/x/2108832291401453576.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ミヤマ(@mmmiyama_D)さんのプロンプトとたーや(@taya_mama_AI)さんの9in1画像プロンプトをお借りして組み合わせて、たーやさんの動画プロンプトをお借りしてMinimax-H3 Turboで動画作成して楽しんでいます。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 1,
      "views": 77,
      "snapshotAt": "2026-10-10T09:43:34.689Z"
    },
    "approvedAt": "2026-10-10T09:43:34.689Z",
    "editorialBasis": "The caption explicitly identifies the video as Minimax-H3 Turbo. Its single native video played; beginning, middle, and end frames were reviewed with no conflicting model mark visible."
  },
  {
    "id": "x-2108830290936508846",
    "title": "竹林间的剑招",
    "titleEn": "Swordplay Among Bamboo",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "本地剪辑界面展示竹林中的两段对打素材和文件列表，随后切到近身挥剑的镜头。 原生视频规格为 5 秒 · 1344×1536。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A local editing view shows two bamboo-forest fight clips beside a file list, then closer swordplay scenes. Native-video output: 5s · 1344×1536. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/eternityspring/status/2108830290936508846",
    "sourceLabel": "X 原帖 · @eternityspring",
    "author": "烁皓",
    "publishedAt": "2026-10-10T08:01:36.000Z",
    "addedAt": "2026-10-10T09:43:34.689Z",
    "mediaUrl": "/media/x-2108830290936508846.mp4",
    "posterUrl": "/posters/x/2108830290936508846.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1344×1536",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "看到好多分享MiniMax H3打斗LoRA：\nH3-Fight-Flow-Fight-Interaction-LoRA\nhttps://huggingface.co/JOKER141/H3-Fight-Flow-Fight-Interaction-LoRA\n\n实测了下，真是鱼和熊掌不可兼得\n测了好几个版本，只有这个LoRA效果最明显，但打斗会糊",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 6,
      "views": 1319,
      "snapshotAt": "2026-10-10T09:43:34.689Z"
    },
    "approvedAt": "2026-10-10T09:43:34.689Z",
    "editorialBasis": "The original caption explicitly names a MiniMax H3 fight LoRA. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2108822099754176781",
    "title": "Dove洁面皂与护肤场景",
    "titleEn": "Dove Soap in a Skincare Scene",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "浴室台面先出现Dove洁面皂与包装，随后一名女子洗脸，最后手持产品并展示柔肤文案。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Dove soap and its box appear on a bathroom counter, followed by a woman washing her face and holding the product beside softness messaging. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/YolandahGil/status/2108822099754176781",
    "sourceLabel": "X 原帖 · @YolandahGil",
    "author": "Yolanda",
    "publishedAt": "2026-10-10T07:29:03.000Z",
    "addedAt": "2026-10-10T09:43:34.689Z",
    "mediaUrl": "/media/x-2108822099754176781.mp4",
    "posterUrl": "/posters/x/2108822099754176781.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I wanted to explore the full potential of MiniMax H3 for commercial creation, so I set out to produce a complete Dove-inspired advertisement centered on one creative concept:\n\nWhat if typography could become an integral part of the storytelling?\n\nBy combining multiple references in one workflow, I brought together consistent product details, character continuity, realistic lighting, synchronized voice-over, dynamic camera movements, and stylized typography generation.",
    "engagement": {
      "replies": 10,
      "reposts": 84,
      "likes": 94,
      "views": 7663,
      "snapshotAt": "2026-10-10T09:43:34.689Z"
    },
    "approvedAt": "2026-10-10T09:43:34.689Z",
    "editorialBasis": "The caption explicitly names MiniMax H3 for the commercial. Its single native video played; beginning, middle, and end frames were reviewed. A small Hailuo AI label is visible, with no conflicting model mark."
  },
  {
    "id": "x-2108831703120879785",
    "title": "黑白女仆的轻快舞步",
    "titleEn": "A Maid’s Light Dance",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白服饰的动漫女仆先以站姿出镜，随后下蹲并摆动手臂完成一段短舞。 原生视频规格为 7 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime maid in black and white first holds a standing pose, then crouches and moves her arms through a short dance. Native-video output: 7s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MedioConxx/status/2108831703120879785",
    "sourceLabel": "X 原帖 · @MedioConxx",
    "author": "GooseFun",
    "publishedAt": "2026-10-10T08:07:12.000Z",
    "addedAt": "2026-10-10T09:43:34.689Z",
    "mediaUrl": "/media/x-2108831703120879785.mp4",
    "posterUrl": "/posters/x/2108831703120879785.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Furina has no reason to be this good at idol choreography.\n\nThe dramatic little poses, the spins, the coat movement—it all fits her way too well.\n\nMade with Hailuo H3.\n\nCharacter sheet prompt below 👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 43,
      "snapshotAt": "2026-10-10T09:43:34.689Z"
    },
    "approvedAt": "2026-10-10T09:43:34.689Z",
    "editorialBasis": "The caption says the clips were made with Hailuo H3. Both native videos played and show the same character in related poses; beginning, middle, and end frames for both were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2107725395428880648",
    "title": "雨水中的人物倒影",
    "titleEn": "A Face Through Rainwater",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "昏暗室内切到水面涟漪中的男子面孔，末尾转为夜色里的工业街道。 原生视频规格为 168 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark interior gives way to a man’s face through rippling water, then to an industrial street at night. Native-video output: 168s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TechTalkNAVI/status/2107725395428880648",
    "sourceLabel": "X 原帖 · @TechTalkNAVI",
    "author": "テックトークナビゲーター@AIコンテンツクリエイター",
    "publishedAt": "2026-10-07T06:51:08.000Z",
    "addedAt": "2026-10-10T09:43:34.689Z",
    "mediaUrl": "/media/x-2107725395428880648.mp4",
    "posterUrl": "/posters/x/2107725395428880648.jpg",
    "duration": 168,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "作詞から映像まで、AIと作ったMV「雨と鉄」\n\n正しい選択を積み上げてきたはずなのに、気づけば知らない場所に立っていた——\nそんな夜の物語を実写調の映像にしました。\n\n🎬 Hailuo H3\n\n#MV #HailuoH3 \n@Hailuo_AI @MiniMax_AI",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 12,
      "views": 754,
      "snapshotAt": "2026-10-10T09:43:34.689Z"
    },
    "approvedAt": "2026-10-10T09:43:34.689Z",
    "editorialBasis": "The original caption identifies the music video as Hailuo H3. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2107492458187596101",
    "title": "花朵绽放成光点",
    "titleEn": "A Flower Unfolds into Light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "橙色云雾包围面孔般的形体，随后一朵橙白花朵展开，最后画面收束为黑底上的亮色字样。 原生视频规格为 126 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Orange clouds surround a face-like form before an orange-and-white flower opens and the image resolves into bright lettering on black. Native-video output: 126s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/papadupapadu/status/2107492458187596101",
    "sourceLabel": "X 原帖 · @papadupapadu",
    "author": "PAPADU PAPADU",
    "publishedAt": "2026-10-06T15:25:32.000Z",
    "addedAt": "2026-10-10T09:43:34.690Z",
    "mediaUrl": "/media/x-2107492458187596101.mp4",
    "posterUrl": "/posters/x/2107492458187596101.jpg",
    "duration": 126,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "stylized",
      "dark"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "IT BEGINS AS ONE. IT DIVIDES INTO EVERYTHING. IT RETURNS.\n\nELEMENTAL DIFFUSION | A @surRIHL CUT\nDIRECTED BY: @papadupapadu + @LoLoLumi88\n\nHAILUO H3 + ELEVEN MUSIC + MANUAL EDITING IN PREMIERE PRO + CLAUDE OPUS 5.5",
    "engagement": {
      "replies": 7,
      "reposts": 4,
      "likes": 51,
      "views": 3595,
      "snapshotAt": "2026-10-10T09:43:34.690Z"
    },
    "approvedAt": "2026-10-10T09:43:34.690Z",
    "editorialBasis": "The caption explicitly names Hailuo H3 and separately lists music and manual editing. The single native video played; beginning, middle, and end frames were reviewed with no different-model mark visible."
  },
  {
    "id": "x-2107230750349361222",
    "title": "一场因布丁引发的情绪风暴",
    "titleEn": "A Flan Sets Off a Dramatic Reaction",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "空置的置物架切到一名动漫女孩夸张地发怒，最后她在教室般的场景中转身看向一侧。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An empty shelving unit gives way to an anime girl’s exaggerated outburst, then she turns aside in a classroom-like setting. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MayorKingAI/status/2107230750349361222",
    "sourceLabel": "X 原帖 · @MayorKingAI",
    "author": "MayorkingAI",
    "publishedAt": "2026-10-05T22:05:35.000Z",
    "addedAt": "2026-10-10T09:43:34.690Z",
    "mediaUrl": "/media/x-2107230750349361222.mp4",
    "posterUrl": "/posters/x/2107230750349361222.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A whole drama… over a flan 🍮\n\nThe expressiveness and anime style in Hailuo H3 are on another level\n\nEvery glance, every tiny reaction lands\n\nPrompt below 👇",
    "engagement": {
      "replies": 23,
      "reposts": 1,
      "likes": 51,
      "views": 2184,
      "snapshotAt": "2026-10-10T09:43:34.690Z"
    },
    "approvedAt": "2026-10-10T09:43:34.690Z",
    "editorialBasis": "The original caption explicitly names Hailuo H3. The native video played; beginning, middle, and end frames were reviewed, with no conflicting model mark visible."
  },
  {
    "id": "x-2106685825148747867",
    "title": "宠物店里的巫师帽猫咪",
    "titleEn": "A Witch-Hat Cat at the Pet Store",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一名男子在宠物店里抱着戴巫师帽的橘猫，两段视频呈现同一角色与相近场景。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man holds an orange cat wearing a witch hat in a pet store; both clips show the same character and a related scene. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JayKay65220066/status/2106685825148747867",
    "sourceLabel": "X 原帖 · @JayKay65220066",
    "author": "Jay Kay",
    "publishedAt": "2026-10-04T10:00:15.000Z",
    "addedAt": "2026-10-10T09:43:34.690Z",
    "mediaUrl": "/media/x-2106685825148747867.mp4",
    "posterUrl": "/posters/x/2106685825148747867.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Autumn Leaf Boxing! 🍂🥊\n\nCreated using Hailuo H3 Model via @pika_labs AI.\n\n#Sundayvibes #Sundaymotivation #SundayThoughts\n\nGood Sunday Funday! 🌞",
    "engagement": {
      "replies": 24,
      "reposts": 10,
      "likes": 63,
      "views": 3348,
      "snapshotAt": "2026-10-10T09:43:34.690Z"
    },
    "approvedAt": "2026-10-10T09:43:34.690Z",
    "editorialBasis": "The original caption says the clips were created using the Hailuo H3 model via Pika AI. Both native videos played and beginning, middle, and end frames for each were reviewed, with no conflicting model mark visible."
  },
  {
    "id": "x-2106636476695912599",
    "title": "节拍驱动的电影感剪辑",
    "titleEn": "A Beat-Driven Cinematic Cut",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "碎光标题过渡到层叠的人像与反光车身，片尾逐渐淡入黑场。 原生视频规格为 150 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fragmented-light title leads into layered portraits and a reflective vehicle before fading to black. Native-video output: 150s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/papadupapadu/status/2106636476695912599",
    "sourceLabel": "X 原帖 · @papadupapadu",
    "author": "PAPADU PAPADU",
    "publishedAt": "2026-10-04T06:44:10.000Z",
    "addedAt": "2026-10-10T09:43:34.690Z",
    "mediaUrl": "/media/x-2106636476695912599.mp4",
    "posterUrl": "/posters/x/2106636476695912599.jpg",
    "duration": 150,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "DON'T WATCH PAST 0:15.\n\nAND BEYOND | A @surRIHL CUT\nDIRECTED BY: @papadupapadu + @LoLoLumi88\n\nHAILUO H3 + CLAUDE OPUS 5.5 + MANUAL EDITING IN PREMIERE PRO\n\n11 LAYERS. 3,600 FRAMES. EVERY CUT DRIVEN BY THE BASS.",
    "engagement": {
      "replies": 8,
      "reposts": 6,
      "likes": 37,
      "views": 1273,
      "snapshotAt": "2026-10-10T09:43:34.690Z"
    },
    "approvedAt": "2026-10-10T09:43:34.690Z",
    "editorialBasis": "The original caption explicitly lists Hailuo H3 alongside Claude Opus and manual Premiere editing. The native video played; beginning, middle, and end frames were reviewed, with no conflicting model mark visible."
  },
  {
    "id": "x-2106487748005867629",
    "title": "海报在画面中缓缓动起来",
    "titleEn": "A Poster Comes to Life",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "占星主题海报先以静帧出现，随后肖像图层移动，最后回到海报画面。 原生视频规格为 15 秒 · 480×704。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A zodiac-themed poster appears, its portrait layers move, and the video returns to the poster image. Native-video output: 15s · 480×704. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/AvisMelodieux/status/2106487748005867629",
    "sourceLabel": "X 原帖 · @AvisMelodieux",
    "author": "AvisMelodieux",
    "publishedAt": "2026-10-03T20:53:10.000Z",
    "addedAt": "2026-10-10T09:43:34.690Z",
    "mediaUrl": "/media/x-2106487748005867629.mp4",
    "posterUrl": "/posters/x/2106487748005867629.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×704",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "11pm is a good time to start a 2h40 movie.\nIn the meantime, GPT2.5 remade the poster for me, and Hailuo H3 turbo made the little animation.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 236,
      "snapshotAt": "2026-10-10T09:43:34.690Z"
    },
    "approvedAt": "2026-10-10T09:43:34.690Z",
    "editorialBasis": "The original caption distinguishes GPT 2.5 for remaking the poster from Hailuo H3 Turbo for its animation. The native video played; beginning, middle, and end frames were reviewed, with no conflicting model mark visible."
  },
  {
    "id": "x-2109057926614913251",
    "title": "竹笠少女与武士对峙",
    "titleEn": "Straw-hat heroine faces a warrior",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "插画风少女走在樱花街道上，镜头转向持刀武士与随从，最后回到少女的近景。 原生视频规格为 15 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: an illustrated girl in a straw hat walks along a cherry-blossom street. Middle: an armored warrior and followers face her. End: the camera returns to her face beneath the hat. Native-video output: 15s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kaku4946/status/2109057926614913251",
    "sourceLabel": "X 原帖 · @kaku4946",
    "author": "𓀼フレイニャル𓀼",
    "publishedAt": "2026-10-10T23:06:08.000Z",
    "addedAt": "2026-10-11T00:23:12.012Z",
    "mediaUrl": "/media/x-2109057926614913251.mp4",
    "posterUrl": "/posters/x/2109057926614913251.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3テスト、まだまだ初心者だけどローカル環境でここまでできれば上出来だね。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 24,
      "snapshotAt": "2026-10-11T00:23:12.012Z"
    },
    "approvedAt": "2026-10-11T00:23:12.012Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109052345225019487",
    "title": "红斗篷少女与怒目老者",
    "titleEn": "Red-caped heroine and an angry elder",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画少女在窗前展开展示红色斗篷，画面切到戴帽老者的怒目近景。 原生视频规格为 13 秒 · 1440×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: an animated woman in a red cape stands beside blue windows. Middle: she holds the cape open. End: the scene cuts to a close-up of an angry elderly man in a hat. Native-video output: 13s · 1440×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MajestyFasal/status/2109052345225019487",
    "sourceLabel": "X 原帖 · @MajestyFasal",
    "author": "Majesty Fasal",
    "publishedAt": "2026-10-10T22:43:58.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109052345225019487.mp4",
    "posterUrl": "/posters/x/2109052345225019487.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "1440×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Ana Elisa - Meu trabalho aqui está feito!\n\nMontagem usando Minimax H3 usando o meme do Texudo Mask do Sailor Moon como referencia.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 2,
      "views": 54,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109035954715656308",
    "title": "网球场上的挥拍女孩",
    "titleEn": "Racket swing on a sunny tennis court",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女孩在露天网球场准备挥拍，镜头随着球拍动作靠近，呈现白色运动装与晴天球场。 原生视频规格为 8 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a young woman prepares a tennis swing on an outdoor court. Middle: she brings her racket across her body. End: a closer view follows her face and racket against the sunny court. Native-video output: 8s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/redtoweler/status/2109035954715656308",
    "sourceLabel": "X 原帖 · @redtoweler",
    "author": "赤タオル",
    "publishedAt": "2026-10-10T21:38:50.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109035954715656308.mp4",
    "posterUrl": "/posters/x/2109035954715656308.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tennis Girl\n#MiniMax H3 #ちちぷい女子オープン",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 32,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109029956542136694",
    "title": "巨型茶杯前的奇妙人偶剧",
    "titleEn": "Puppet theatre beside a giant teacup",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红裙人偶站在巨型茶杯旁，一条红鱼悬浮在杯口上方，结尾以人偶近景与立体文字收束。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a red-dressed puppet stands beside a giant red teacup in a hanging-mask theatre. Middle: a red fish floats above the cup. End: a close-up of the puppet includes raised Japanese lettering. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yarakashiyama_/status/2109029956542136694",
    "sourceLabel": "X 原帖 · @yarakashiyama_",
    "author": "ぴょんきち",
    "publishedAt": "2026-10-10T21:15:00.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109029956542136694.mp4",
    "posterUrl": "/posters/x/2109029956542136694.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【PR】しゃべった言葉まで、人形劇の一部に。\n\n少女の「わあ」が立体になって浮かぶ、\nちょっと奇妙な15秒を作りました。\n\nSousaku AI（@SousakuAI）のMiniMax H3で制作。\n最後の「私も？」まで見てほしい。\n\n使った参照画像とプロンプトは次へ↓\n #SousakuAI",
    "engagement": {
      "replies": 1,
      "reposts": 9,
      "likes": 13,
      "views": 2034,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109015436209819855",
    "title": "柔光室内的人像微笑",
    "titleEn": "A soft-lit portrait turns to a smile",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "身穿浅粉衬衫的女性面对镜头，由平静表情转为微笑，背景是透过窗帘的柔和室内光线。 原生视频规格为 7 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a dark-haired woman in a pale pink blouse faces the camera indoors. Middle: she starts smiling. End: her smile broadens in the same softly lit room. Native-video output: 7s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/0x0SojalSec/status/2109015436209819855",
    "sourceLabel": "X 原帖 · @0x0SojalSec",
    "author": "Md Ismail Šojal 🕷️",
    "publishedAt": "2026-10-10T20:17:18.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109015436209819855.mp4",
    "posterUrl": "/posters/x/2109015436209819855.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This 'Jennifer Connelly' clip was made with zero reference images in locally.      \n\n- Just a 155 MB LoRA on MiniMax H3.   \n- The face consistency is getting uncomfortable.\n\n- https://huggingface.co/Playtime-AI/Minimax_H3-Jennifer_Connelly",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 16,
      "views": 1357,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109007984873677099",
    "title": "夕阳铁道旁的学生短片",
    "titleEn": "Students beside the railway at sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短片从学生离开教室展开，随后切到夕阳下的铁道与水塔，结尾人物来到轨道边。 原生视频规格为 58 秒 · 848×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: two students walk past classroom desks carrying bags. Middle: railway tracks and water towers stretch into a sunset. End: a person enters the foreground beside the tracks. Native-video output: 58s · 848×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Edenwoodfilm/status/2109007984873677099",
    "sourceLabel": "X 原帖 · @Edenwoodfilm",
    "author": "Adil",
    "publishedAt": "2026-10-10T19:47:41.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109007984873677099.mp4",
    "posterUrl": "/posters/x/2109007984873677099.jpg",
    "duration": 58,
    "aspectRatio": "landscape",
    "resolution": "848×480",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This 1-Minute AI Movie Looks TOO Real 🎬 (MiniMax H3, text-to-video)\n\n#AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 78,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109053613708370246",
    "title": "铸剑之城化为巨型战士",
    "titleEn": "A forge city transforms into a giant warrior",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金色羽翼悬在古城上空，铁匠举起炽热长剑，结尾出现由城堡与石墙组成的巨大装甲战士。 原生视频规格为 15 秒 · 1536×672。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a golden-winged angel hovers over a blacksmith in a medieval city. Middle: the blacksmith raises a blazing sword among the buildings. End: a giant armored warrior with castle-like architecture stands above the city landscape. The second visible video is a quoted promotional clip, not the inspected primary result. Native-video output: 15s · 1536×672. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Photorealistic cinematic dark fantasy. \n\nSix camera setups. Monumental scale, realistic weight, continuous physical transformation.  \n\nAn ancient medieval city built around a gigantic mechanical forge. One muscular blacksmith in a dark leather apron. One colossal celestial angel with luminous golden wings. Storm clouds, cold blue-gray stone, molten golden light.  \n\n**0:00–0:02 | Extreme low angle** The enormous golden-winged angel descends over the city forge. The blacksmith pulls a gigantic iron lever. Four massive chains launch upward and wrap tightly around the angel's wings.  \n\n**0:02–0:04 | Wide frontal angle** The chains violently drag the angel onto the forge platform. Its wings slam across two enormous anvils. Giant mechanical clamps lock the glowing feathers in place.  \n\n**0:04–0:06 | Mechanical close-up** Two colossal hammers strike the golden wings. Each impact compresses the feathers into molten metal. The metal fuses into one magnificent blazing sword.  \n\n**0:06–0:08 | Ground-level hero angle** The blacksmith lifts the completed sword vertically. Golden energy explodes downward into the city streets. The entire city begins shaking. Buildings tilt, foundations split, and enormous sections of the ground rise.  \n\n**0:08–0:11 | High aerial angle** THE CITY ITSELF PHYSICALLY TRANSFORMS INTO A GIANT HUMANOID WARRIOR.  The transformation begins while every building remains visibly connected to the same city.  The outer city walls lift and fold inward to become two gigantic legs. Entire neighborhoods rotate upward into two massive arms. The central district rises into a broad armored torso. The castle at the center shifts upward to become the warrior's helmeted head.  Every original building becomes part of the warrior's body. Towers remain visibly embedded in its armor. Streets fold into its joints. The city is literally rearranging itself into a humanoid figure.  \n\n**0:11–0:15 | Continuous rising aerial pullback** \nThe transformation continues without interruption. The two enormous legs straighten, the arms unfold from the former neighborhoods, and the castle-head rises above the shoulders.  The newly formed colossal warrior stands upright on the exact ground previously occupied by the city.  \n\nThe original city no longer exists as a separate settlement. Its entire architecture now forms the warrior's body.  Golden fire ignites inside the castle-head's eyes. The forge burns in the center of its chest. Final wide shot reveals the complete giant armored warrior, with medieval towers, rooftops, and stone walls integrated into its anatomy.  \n\n**VISUAL STYLE:** Epic cinematic dark fantasy, realistic architectural deformation, enormous interconnected stone structures, heavy mechanical movement, monumental scale, collapsing dust clouds,\ngolden magical energy, storm-dark atmosphere.  **ESSENTIAL TRANSFORMATION RULE:** \nThe city IS the warrior. The warrior is made entirely from the city's existing physical architecture. This is one continuous transformation of the same object, not a summoning, not a creature emerging from beneath buildings, not a separate giant standing inside a city.  Maintain visual continuity throughout the transformation. No new warrior appears. No intact city remains around the completed figure.  No dialogue, no subtitles, no text on screen.",
    "promptSourceUrl": "https://x.com/DeCat2025/status/2109053613708370246",
    "sourceUrl": "https://x.com/DeCat2025/status/2109053613708370246",
    "sourceLabel": "X 原帖 · @DeCat2025",
    "author": "DeCat",
    "publishedAt": "2026-10-10T22:49:00.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109053613708370246.mp4",
    "posterUrl": "/posters/x/2109053613708370246.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1536×672",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "fantasy",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Angel Factory\nMinimax H3 @Hailuo_AI \n\nPhotorealistic cinematic dark fantasy. \n\nSix camera setups. Monumental scale, realistic weight, continuous physical transformation.  \n\nAn ancient medieval city built around a gigantic mechanical forge. One muscular blacksmith in a dark leather apron. One colossal celestial angel with luminous golden wings. Storm clouds, cold blue-gray stone, molten golden light.  \n\n**0:00–0:02 | Extreme low angle** The enormous golden-winged angel descends over the city forge. The blacksmith pulls a gigantic iron lever. Four massive chains launch upward and wrap tightly around the angel's wings.  \n\n**0:02–0:04 | Wide frontal angle** The chains violently drag the angel onto the forge platform. Its wings slam across two enormous anvils. Giant mechanical clamps lock the glowing feathers in place.  \n\n**0:04–0:06 | Mechanical close-up** Two colossal hammers strike the golden wings. Each impact compresses the feathers into molten metal. The metal fuses into one magnificent blazing sword.  \n\n**0:06–0:08 | Ground-level hero angle** The blacksmith lifts the completed sword vertically. Golden energy explodes downward into the city streets. The entire city begins shaking. Buildings tilt, foundations split, and enormous sections of the ground rise.  \n\n**0:08–0:11 | High aerial angle** THE CITY ITSELF PHYSICALLY TRANSFORMS INTO A GIANT HUMANOID WARRIOR.  The transformation begins while every building remains visibly connected to the same city.  The outer city walls lift and fold inward to become two gigantic legs. Entire neighborhoods rotate upward into two massive arms. The central district rises into a broad armored torso. The castle at the center shifts upward to become the warrior's helmeted head.  Every original building becomes part of the warrior's body. Towers remain visibly embedded in its armor. Streets fold into its joints. The city is literally rearranging itself into a humanoid figure.  \n\n**0:11–0:15 | Continuous rising aerial pullback** \nThe transformation continues without interruption. The two enormous legs straighten, the arms unfold from the former neighborhoods, and the castle-head rises above the shoulders.  The newly formed colossal warrior stands upright on the exact ground previously occupied by the city.  \n\nThe original city no longer exists as a separate settlement. Its entire architecture now forms the warrior's body.  Golden fire ignites inside the castle-head's eyes. The forge burns in the center of its chest. Final wide shot reveals the complete giant armored warrior, with medieval towers, rooftops, and stone walls integrated into its anatomy.  \n\n**VISUAL STYLE:** Epic cinematic dark fantasy, realistic architectural deformation, enormous interconnected stone structures, heavy mechanical movement, monumental scale, collapsing dust clouds,\ngolden magical energy, storm-dark atmosphere.  **ESSENTIAL TRANSFORMATION RULE:** \nThe city IS the warrior. The warrior is made entirely from the city's existing physical architecture. This is one continuous transformation of the same object, not a summoning, not a creature emerging from beneath buildings, not a separate giant standing inside a city.  Maintain visual continuity throughout the transformation. No new warrior appears. No intact city remains around the completed figure.  No dialogue, no subtitles, no text on screen.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 161,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109006383995662721",
    "title": "雪山背景下的手机助手广告",
    "titleEn": "Phone assistant promo against snowy mountains",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏短片从发光脑形图标展开，展示雪山前双手持手机的画面，结尾回到应用品牌标识。 原生视频规格为 58 秒 · 1070×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a glowing blue brain-shaped logo appears on black. Middle: hands hold a smartphone in front of snowy mountains. End: the clip returns to the app logo and a Google Play call to action. The original author explicitly credits H3 among the production tools. Native-video output: 58s · 1070×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Rotinda_Dag/status/2109006383995662721",
    "sourceLabel": "X 原帖 · @Rotinda_Dag",
    "author": "Rotinda Dağ",
    "publishedAt": "2026-10-10T19:41:20.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109006383995662721.mp4",
    "posterUrl": "/posters/x/2109006383995662721.jpg",
    "duration": 58,
    "aspectRatio": "portrait",
    "resolution": "1070×1920",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tawus Local AI Get It on Google Play Store:\nhttps://play.google.com/store/apps/details?id=com.tawusgames.localai\n\nNew 1.1 Update Available Now on Google Play Store\n\n#mobileai #mobilelocalai #localai #llm #llmandroid\n\nVideo created thanks to DS Harness, Qwen 3.8 Flash Next, Minimax H3 and Qwen Image 2.1 AI models on localy.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 70,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109004849962783186",
    "title": "木屋中的精灵伙伴",
    "titleEn": "Elven companions inside a timber house",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两位精灵角色坐在木屋中，镜头从并排的中景切向交谈近景，最后两人靠在一起。 原生视频规格为 47 秒 · 544×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: two animated elven women sit together in a timber room. Middle: a close-up shows them facing each other and gesturing. End: they lean together beside the window. Native-video output: 47s · 544×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EmpJellok/status/2109004849962783186",
    "sourceLabel": "X 原帖 · @EmpJellok",
    "author": "JellokEmp",
    "publishedAt": "2026-10-10T19:35:14.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109004849962783186.mp4",
    "posterUrl": "/posters/x/2109004849962783186.jpg",
    "duration": 47,
    "aspectRatio": "landscape",
    "resolution": "544×384",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Using Minimax H3 again I ran the same prompt I used before with new setup. Some reason it changed certain things lol. Well the animation is far improved from the first time around. #AIart @OC",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 96,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2109001531446243650",
    "title": "骷髅橱窗前的诡异交易",
    "titleEn": "An eerie bargain at the skull-filled shop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "骷髅装饰的商店橱窗前，两名男子短暂交谈，随后一人走向挂着营业霓虹灯的店门。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a man stands before a shop window filled with skull portraits and an OPEN neon sign. Middle: two men face each other and gesture. End: one turns toward the shop doorway. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/altfortomorrow/status/2109001531446243650",
    "sourceLabel": "X 原帖 · @altfortomorrow",
    "author": "Blujeanmomma",
    "publishedAt": "2026-10-10T19:22:03.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2109001531446243650.mp4",
    "posterUrl": "/posters/x/2109001531446243650.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Choose wisely… some shops collect more than money when they’re collecting their due. \n\nThat tempting little bargain might cost more than you ever intended to give. 🛍️👻\n\nDo you always read the fine print?\n\nMiniMax H3 model @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "views": 294,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108994976487268387",
    "title": "火山岛与珊瑚环礁的变换",
    "titleEn": "Volcanic island becomes a coral atoll",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "悬浮海洋剖面中的火山岛变成环形珊瑚礁，再回到喷涌熔岩的火山形态。 原生视频规格为 14 秒 · 1024×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a floating square ocean cutaway contains an erupting volcano. Middle: it changes to a circular coral atoll around a blue lagoon. End: the volcano and glowing lava reappear in the cutaway. Native-video output: 14s · 1024×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wild_zones/status/2108994976487268387",
    "sourceLabel": "X 原帖 · @wild_zones",
    "author": "wild_zones",
    "publishedAt": "2026-10-10T18:56:00.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2108994976487268387.mp4",
    "posterUrl": "/posters/x/2108994976487268387.jpg",
    "duration": 14,
    "aspectRatio": "square",
    "resolution": "1024×1024",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Opus 5.5 calls it “Darwin’s Atoll” (explanation in follow up)…\n\n…What I wake up to when I tell Claude Code paired with the local ComfyUI MCP to make whatever it wants with MiniMax H3 overnight on an R9700.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 1,
      "views": 129,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108994777735958754",
    "title": "发条小鼠穿过猫门",
    "titleEn": "Wind-up mice roll through a cat flap",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "厨房门上的猫门打开，一队发条玩具小鼠依次滚过门口，镜头始终贴近地面。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a closed cat flap sits in a kitchen door at floor level. Middle: wind-up toy mice roll through the open flap. End: several mice line up on the tiled floor beyond the doorway. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "five small wind-up toy mice roll through it one at a time",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2108994777735958754",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108994777735958754",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-10T18:55:12.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2108994777735958754.mp4",
    "posterUrl": "/posters/x/2108994777735958754.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. The cat flap works. Not for the cat.\n\nPrompt: \"five small wind-up toy mice roll through it one at a time\"\n\nDuration: 15s. Camera: locked, floor height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe extra flap swing with nothing behind it is the punchline. It costs one clause.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 9,
      "views": 374,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108981998903755091",
    "title": "雾面玻璃上的连续爪印",
    "titleEn": "Paw prints form across a misted window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雾气覆盖的玻璃上，彩色爪印逐个出现，沿斜向轨迹排列成一串。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a fogged pane of glass has no visible paw-print trail. Middle: several colored paw prints appear diagonally across it. End: the trail extends farther upward across the pane. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "five paw prints appear one at a time, walking in a line across the glass",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2108981998903755091",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108981998903755091",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-10T18:04:26.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2108981998903755091.mp4",
    "posterUrl": "/posters/x/2108981998903755091.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Someone signed the window.\n\nPrompt: \"five paw prints appear one at a time, walking in a line across the glass\"\n\nDuration: 15s. Camera: locked, straight on. Rendered locally in ComfyUI, then Topaz to 4K.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 10,
      "views": 421,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108979557168758924",
    "title": "紫色终端风格的拼贴音乐短片",
    "titleEn": "Purple terminal-styled portrait collage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏音乐短片把人像、卡片和逐字字幕组合在紫色终端风格界面中，结尾展示彩色标题。 原生视频规格为 166 秒 · 1080×1350。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a monochrome portrait appears inside a purple terminal-styled frame with subtitles. Middle: floating portrait cards cross a yellow-orange panel. End: a multicolored NEXT TO KEN title fills the center. The source identifies the image, H3 video and music tools separately; the video is a composited creative work. Native-video output: 166s · 1080×1350. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/quimedesu/status/2108979557168758924",
    "sourceLabel": "X 原帖 · @quimedesu",
    "author": "Ark",
    "publishedAt": "2026-10-10T17:54:44.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2108979557168758924.mp4",
    "posterUrl": "/posters/x/2108979557168758924.jpg",
    "duration": 166,
    "aspectRatio": "portrait",
    "resolution": "1080×1350",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I gave Claude Opus 5.5 an RTX 3090, and 12 hours later, he created this using less than 20% of the standard plan's daily limit. Models it used Qwen Image 2.1, Minimax H3, YuE2",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 188,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108957560707457178",
    "title": "吊扇遮挡下的万圣节换装",
    "titleEn": "Halloween costume changes behind a ceiling fan",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "俯拍镜头中女性躺在地毯上，木质吊扇转动遮挡画面，随后出现黑羽翼造型，结尾落到黑色品牌画面。 原生视频规格为 21 秒 · 1080×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a top-down view shows a woman lying on a rug in a pumpkin-filled room, partly hidden by a wooden ceiling fan. Middle: she appears with a black feathered-wing costume as the fan turns. End: a black branded end slate appears. Native-video output: 21s · 1080×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a photorealistic cinematic Halloween fashion transformation video featuring Alexa Grey, using the supplied Alexa Grey Master Character Lock as the exact identity reference.\n\nAlexa is lying in the exact same relaxed pose on a vintage Persian rug in a dark, elegant Halloween-decorated living room.\n\nCAMERA:\nFixed directly overhead.\nPerfect top-down composition.\nNo camera movement.\nNo cuts.\nNo zoom.\nNo change in angle.\nAlexa remains in exactly the same position throughout the entire video.\n\nA large wooden ceiling fan is positioned directly between the camera and Alexa.\n\nTHE TRANSITION CONCEPT IS EXTREMELY IMPORTANT:\n\nThe ceiling fan rotates VERY SLOWLY.\n\nONE COMPLETE SLOW ROTATION OF THE CEILING FAN = ONE OUTFIT CHANGE.\n\nAs the fan completes one slow rotation, its blades naturally sweep across the camera's view and temporarily obscure Alexa.\n\nWhen that ONE complete rotation finishes and the fan returns to its original position, Alexa is revealed wearing ONE completely different Halloween costume.\n\nThen the fan makes another single slow rotation, revealing the next outfit.\n\nDo NOT make multiple outfit changes during one rotation.\n\nDo NOT make the fan spin rapidly.\n\nDo NOT use cuts, dissolves, flashes or digital transition effects.\n\nThe fan itself must create the transformation.\n\nOUTFIT SEQUENCE:\n\nSTART:\nAlexa wears a simple elegant black outfit.\n\nROTATION 1:\nOne slow complete fan rotation.\nWhen the rotation finishes, reveal:\n\nGOTHIC WITCH\nElegant black Victorian witch dress, dramatic pointed witch hat, dark lace, gothic accessories and mysterious Halloween styling.\n\nROTATION 2:\nOne slow complete fan rotation.\nReveal:\n\nVAMPIRE QUEEN\nElegant black-and-deep-red Victorian gown, dramatic cape, gothic jewelry and sophisticated vampire styling.\n\nROTATION 3:\nOne slow complete fan rotation.\nReveal:\n\nDEMON QUEEN\nElegant black-and-crimson gothic costume, subtle horns, dramatic dark textures and supernatural styling.\n\nROTATION 4:\nOne slow complete fan rotation.\nReveal:\n\nFALLEN DARK ANGEL\nElegant black gothic gown with enormous black feathered wings and dramatic dark angel styling.\n\nROTATION 5:\nOne slow complete fan rotation.\nReveal:\n\nGHOST BRIDE\nDistressed vintage wedding gown, long torn veil, dark roses and haunting gothic styling.\n\nEvery time the fan completes ONE slow rotation, Alexa must be fully revealed in the new outfit.\n\nThe transition should feel almost magical and seamless, as if the rotating ceiling fan is changing her costume.\n\nKeep Alexa's exact face, hair, skin tone and body proportions unchanged.\n\nENVIRONMENT:\nDark luxury Halloween interior.\nVintage Persian rug.\nWooden floor.\nPumpkins.\nFlickering candles.\nOld books.\nSkulls.\nDark flowers.\nSubtle atmospheric fog.\nWarm orange candlelight.\nDeep cinematic shadows.\nElegant gothic atmosphere.\n\nThe environment remains exactly the same during every transformation.\n\nFINAL SHOT:\nAfter the fifth outfit is revealed, Alexa remains lying on the rug and slowly looks directly toward the camera while the ceiling fan continues its slow rotation.\n\nPhotorealistic.\nCinematic Halloween fashion film.\nLuxury horror aesthetic.\nRealistic skin and hair.\nDetailed costumes.\nRealistic fabric and feather textures.\n4K.\n16:9.\n\nNO TEXT.\nNO SUBTITLES.\nNO WATERMARK.\nNO EXTRA PEOPLE.\nNO CAMERA MOVEMENT.\nNO CUTS.\nNO FAST FAN MOTION.\nNO RANDOM OUTFIT CHANGES.\n\nKEY RULE:\nONE SLOW COMPLETE FAN ROTATION → ONE NEW HALLOWEEN OUTFIT.",
    "promptSourceUrl": "https://x.com/MonetizationDon/status/2108957560707457178",
    "sourceUrl": "https://x.com/MonetizationDon/status/2108957560707457178",
    "sourceLabel": "X 原帖 · @MonetizationDon",
    "author": "LAPTOP LIFESTYLE BLOGGER",
    "publishedAt": "2026-10-10T16:27:19.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2108957560707457178.mp4",
    "posterUrl": "/posters/x/2108957560707457178.jpg",
    "duration": 21,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "photoreal"
    ],
    "scenes": [
      "fashion",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One flip of the fan switch, one new Halloween costume worn by Alexa Grey, an AI influencer created with Higgsfield Soul\nI gave the viral fan-switch trend a Halloween makeover\nCreated using MiniMax H3 in @higgsfield \n\nPROMPT: \nCreate a photorealistic cinematic Halloween fashion transformation video featuring Alexa Grey, using the supplied Alexa Grey Master Character Lock as the exact identity reference.\n\nAlexa is lying in the exact same relaxed pose on a vintage Persian rug in a dark, elegant Halloween-decorated living room.\n\nCAMERA:\nFixed directly overhead.\nPerfect top-down composition.\nNo camera movement.\nNo cuts.\nNo zoom.\nNo change in angle.\nAlexa remains in exactly the same position throughout the entire video.\n\nA large wooden ceiling fan is positioned directly between the camera and Alexa.\n\nTHE TRANSITION CONCEPT IS EXTREMELY IMPORTANT:\n\nThe ceiling fan rotates VERY SLOWLY.\n\nONE COMPLETE SLOW ROTATION OF THE CEILING FAN = ONE OUTFIT CHANGE.\n\nAs the fan completes one slow rotation, its blades naturally sweep across the camera's view and temporarily obscure Alexa.\n\nWhen that ONE complete rotation finishes and the fan returns to its original position, Alexa is revealed wearing ONE completely different Halloween costume.\n\nThen the fan makes another single slow rotation, revealing the next outfit.\n\nDo NOT make multiple outfit changes during one rotation.\n\nDo NOT make the fan spin rapidly.\n\nDo NOT use cuts, dissolves, flashes or digital transition effects.\n\nThe fan itself must create the transformation.\n\nOUTFIT SEQUENCE:\n\nSTART:\nAlexa wears a simple elegant black outfit.\n\nROTATION 1:\nOne slow complete fan rotation.\nWhen the rotation finishes, reveal:\n\nGOTHIC WITCH\nElegant black Victorian witch dress, dramatic pointed witch hat, dark lace, gothic accessories and mysterious Halloween styling.\n\nROTATION 2:\nOne slow complete fan rotation.\nReveal:\n\nVAMPIRE QUEEN\nElegant black-and-deep-red Victorian gown, dramatic cape, gothic jewelry and sophisticated vampire styling.\n\nROTATION 3:\nOne slow complete fan rotation.\nReveal:\n\nDEMON QUEEN\nElegant black-and-crimson gothic costume, subtle horns, dramatic dark textures and supernatural styling.\n\nROTATION 4:\nOne slow complete fan rotation.\nReveal:\n\nFALLEN DARK ANGEL\nElegant black gothic gown with enormous black feathered wings and dramatic dark angel styling.\n\nROTATION 5:\nOne slow complete fan rotation.\nReveal:\n\nGHOST BRIDE\nDistressed vintage wedding gown, long torn veil, dark roses and haunting gothic styling.\n\nEvery time the fan completes ONE slow rotation, Alexa must be fully revealed in the new outfit.\n\nThe transition should feel almost magical and seamless, as if the rotating ceiling fan is changing her costume.\n\nKeep Alexa's exact face, hair, skin tone and body proportions unchanged.\n\nENVIRONMENT:\nDark luxury Halloween interior.\nVintage Persian rug.\nWooden floor.\nPumpkins.\nFlickering candles.\nOld books.\nSkulls.\nDark flowers.\nSubtle atmospheric fog.\nWarm orange candlelight.\nDeep cinematic shadows.\nElegant gothic atmosphere.\n\nThe environment remains exactly the same during every transformation.\n\nFINAL SHOT:\nAfter the fifth outfit is revealed, Alexa remains lying on the rug and slowly looks directly toward the camera while the ceiling fan continues its slow rotation.\n\nPhotorealistic.\nCinematic Halloween fashion film.\nLuxury horror aesthetic.\nRealistic skin and hair.\nDetailed costumes.\nRealistic fabric and feather textures.\n4K.\n16:9.\n\nNO TEXT.\nNO SUBTITLES.\nNO WATERMARK.\nNO EXTRA PEOPLE.\nNO CAMERA MOVEMENT.\nNO CUTS.\nNO FAST FAN MOTION.\nNO RANDOM OUTFIT CHANGES.\n\nKEY RULE:\nONE SLOW COMPLETE FAN ROTATION → ONE NEW HALLOWEEN OUTFIT.",
    "engagement": {
      "replies": 32,
      "reposts": 69,
      "likes": 86,
      "views": 614,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108957131361964417",
    "title": "卡牌战士跃出画框",
    "titleEn": "A card warrior springs out of the frame",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "魔幻卡牌中的白发战士从平面画框中跃出，在烛光桌面上挥动利刃。 原生视频规格为 5 秒 · 1152×1472。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a fantasy trading card depicts a pale-haired warrior. Middle: she emerges from the card frame holding blades amid colored light streaks. End: the animated warrior remains above the card on a candlelit table. Native-video output: 5s · 1152×1472. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MarkEmkey/status/2108957131361964417",
    "sourceLabel": "X 原帖 · @MarkEmkey",
    "author": "Mark 'Em 🗝️",
    "publishedAt": "2026-10-10T16:25:37.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2108957131361964417.mp4",
    "posterUrl": "/posters/x/2108957131361964417.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1152×1472",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Had a little fun from this, testing something in MiniMax H3.",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 13,
      "views": 376,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108954964450324649",
    "title": "仓鼠在迷你别墅里的一天",
    "titleEn": "A hamster day in a miniature villa",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "柔和黏土风微缩房间中，仓鼠从圆形窝里探头，在小桌旁用餐，最后回到洞口。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a hamster peeks from a round bed inside a pastel cutaway room. Middle: it sits at a tiny dining table in the adjoining space. End: it returns headfirst into the round burrow. The posted prompt line ends in a semicolon, so no complete Prompt is asserted. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/7u86j0Tf5Y3z7Ii/status/2108954964450324649",
    "sourceLabel": "X 原帖 · @7u86j0Tf5Y3z7Ii",
    "author": "qiuqiu.ai.lab",
    "publishedAt": "2026-10-10T16:17:00.000Z",
    "addedAt": "2026-10-11T00:23:12.013Z",
    "mediaUrl": "/media/x-2108954964450324649.mp4",
    "posterUrl": "/posters/x/2108954964450324649.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized",
      "cg-3d"
    ],
    "scenes": [
      "daily-life",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3    \n\nPrompt below ⤵️\n\nCute isometric 2.5D cutaway diorama of a tiny villa interior, soft clay-render style, pastel colors. A chubby golden hamster pops out of a round burrow bed and scurries down a little ramp;",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 0,
      "views": 36,
      "snapshotAt": "2026-10-11T00:23:12.013Z"
    },
    "approvedAt": "2026-10-11T00:23:12.013Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108951745493864902",
    "title": "雨夜浮空城下的摄影者",
    "titleEn": "A photographer beneath the rainy floating city",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "背着相机的女性在雨夜高台上俯身观察，天空中的飞行器掠过，红色灯光映在未来城市上。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a woman carrying a camera and backpack stands above a rainy futuristic city. Middle: she crouches beside the railing with the camera. End: she watches dark hovering craft with red lights. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ou_zhen599/status/2108951745493864902",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-10-10T16:04:13.000Z",
    "addedAt": "2026-10-11T00:31:05.752Z",
    "mediaUrl": "/media/x-2108951745493864902.mp4",
    "posterUrl": "/posters/x/2108951745493864902.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                                                                                                      \n\n🎬Image-to-Video       \n📷“When the Red Light Turns”\nShe crouched in the shadow beneath the eaves, waiting for the moment the red light turned away before holding out her camera—the sound of the shutter, drowned out by the rain, captured the underside of the floating base that it did not wish to reveal.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 5,
      "views": 262,
      "snapshotAt": "2026-10-11T00:31:05.752Z"
    },
    "approvedAt": "2026-10-11T00:31:05.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108948677159268853",
    "title": "山间露台上的狼少女茶歇",
    "titleEn": "A wolf girl takes tea on a mountain terrace",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝黑毛色的拟人狼少女坐在露台桌边，捧起杯子缓缓喝茶，背景可见群山与木栏。 原生视频规格为 15 秒 · 640×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: an illustrated black-and-blue wolf girl sits beside a table on a mountain terrace holding a cup. Middle: she raises it toward her face. End: she sips while looking toward the camera. Native-video output: 15s · 640×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ThatHuskyGuy969/status/2108948677159268853",
    "sourceLabel": "X 原帖 · @ThatHuskyGuy969",
    "author": "ThatHuskyGuyAI (SFW ONLY)",
    "publishedAt": "2026-10-10T15:52:01.000Z",
    "addedAt": "2026-10-11T00:31:05.753Z",
    "mediaUrl": "/media/x-2108948677159268853.mp4",
    "posterUrl": "/posters/x/2108948677159268853.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "640×640",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Trying out Minimax H3 local AI video generator 🥰\n\nTilly Hawthorne 💙🖤🤍",
    "engagement": {
      "replies": 1,
      "reposts": 6,
      "likes": 62,
      "views": 744,
      "snapshotAt": "2026-10-11T00:31:05.753Z"
    },
    "approvedAt": "2026-10-11T00:31:05.753Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108945454733946907",
    "title": "街头与车内的角色替换对照",
    "titleEn": "Character swaps on the street and inside a car",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下分屏将音乐片段的角色替换结果与原画面并列，包含街头人物近景和车内双人镜头。 原生视频规格为 6 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: stacked panels compare a character-swapped street portrait with the original. Middle: two people inside a car appear in the same upper-result and lower-original layout. End: street walking shots continue the labeled comparison. Native-video output: 6s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GoshniiAi/status/2108945454733946907",
    "sourceLabel": "X 原帖 · @GoshniiAi",
    "author": "goshniiAI",
    "publishedAt": "2026-10-10T15:39:13.000Z",
    "addedAt": "2026-10-11T00:31:05.753Z",
    "mediaUrl": "/media/x-2108945454733946907.mp4",
    "posterUrl": "/posters/x/2108945454733946907.jpg",
    "duration": 6,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I took two scenes from a music video and swapped in new characters with MiniMax H3. \n\n#ComfyUI #MiniMaxH3 #MusicVideos #characterswap #AIVFX #GenerativeAI  #minimax #VFX #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 48,
      "snapshotAt": "2026-10-11T00:31:05.753Z"
    },
    "approvedAt": "2026-10-11T00:31:05.753Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108943911389692190",
    "title": "猫耳女仆的双屏动作对照",
    "titleEn": "Two-panel cat-eared maid motion comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两栏画面中的猫耳女仆先抬手靠近面部，再向镜头伸手，栏顶标有不同量化版本。 原生视频规格为 8 秒 · 1344×1216。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: two side-by-side animated cat-eared maids hold their hands near their faces. Middle: they gesture toward the camera. End: both reach forward. The panels are labeled pruned w6a8 and pruned int8, and the source identifies H3. Native-video output: 8s · 1344×1216. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jun_kongo/status/2108943911389692190",
    "sourceLabel": "X 原帖 · @jun_kongo",
    "author": "kongo jun / 混合 順",
    "publishedAt": "2026-10-10T15:33:05.000Z",
    "addedAt": "2026-10-11T00:31:05.753Z",
    "mediaUrl": "/media/x-2108943911389692190.mp4",
    "posterUrl": "/posters/x/2108943911389692190.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1344×1216",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "anime"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 pruned_w6a8",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 189,
      "snapshotAt": "2026-10-11T00:31:05.753Z"
    },
    "approvedAt": "2026-10-11T00:31:05.753Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108935549553656006",
    "title": "罐头里的小金鱼",
    "titleEn": "A tiny goldfish inside the opened tin",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "桌面上的金属罐头缓缓揭盖，里面不是食物，而是一尾在清水中游动的小金鱼。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a closed metal tin sits on a textured counter. Middle: its lid opens to reveal clear water and a tiny goldfish. End: the fish remains swimming inside the opened tin. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "inside is not food but clear water with one tiny goldfish swimming in it",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2108935549553656006",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108935549553656006",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-10T14:59:51.000Z",
    "addedAt": "2026-10-11T00:31:05.753Z",
    "mediaUrl": "/media/x-2108935549553656006.mp4",
    "posterUrl": "/posters/x/2108935549553656006.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Dinner is served. Dinner has other plans.\n\nPrompt: \"inside is not food but clear water with one tiny goldfish swimming in it\"\n\nDuration: 15s. Camera: locked, floor height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe model keeps peeling the lid from the wrong end, away from the ring pull. The goldfish does not seem to mind, so the best take stays.",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 20,
      "views": 454,
      "snapshotAt": "2026-10-11T00:31:05.753Z"
    },
    "approvedAt": "2026-10-11T00:31:05.753Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108933843923464505",
    "title": "夜色车灯的剪辑呼应",
    "titleEn": "Car lights echo across a night-time edit",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下分屏把新生成的夜色车灯镜头与参考剪辑并列，从绿色光斑转向车头和流线型机身近景。 原生视频规格为 15 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: upper generated and lower reference panels show green blurred lights. Middle: both cut to close-up car headlights. End: the upper panel shows a dark curved vehicle body and the lower a matching reference close-up. The upper is explicitly labeled MINIMAX H3 plus OPUS 5.5; the lower is labeled ORIGINAL and is a reference, not a second generated model. Native-video output: 15s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kentdhani/status/2108933843923464505",
    "sourceLabel": "X 原帖 · @kentdhani",
    "author": "Kent Dhani",
    "publishedAt": "2026-10-10T14:53:05.000Z",
    "addedAt": "2026-10-11T00:31:05.753Z",
    "mediaUrl": "/media/x-2108933843923464505.mp4",
    "posterUrl": "/posters/x/2108933843923464505.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Built my own version, fully local. RefCut #1\n\nOne prompt to Claude Opus 5.5, it generates the footage with MiniMax H3 in ComfyUI on my RTX 5090, then cuts the edit to match the reference.\n\nResult below 👇",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 11,
      "views": 1385,
      "snapshotAt": "2026-10-11T00:31:05.753Z"
    },
    "approvedAt": "2026-10-11T00:31:05.753Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108933446101831946",
    "title": "夕阳门户间穿行的花猫",
    "titleEn": "A calico cat walks between sunset portals",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "倒映夕阳的水面上漂浮着门与画框，一只花猫从门边走过，身旁拖着细小发光轨迹。 原生视频规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: standing doors and floating frames reflect in a sunset-lit surface. Middle: a calico cat walks through the scene beside thin glowing streaks. End: it continues past a dark open doorway. Native-video output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/VoidNulled/status/2108933446101831946",
    "sourceLabel": "X 原帖 · @VoidNulled",
    "author": "Void the Archivist",
    "publishedAt": "2026-10-10T14:51:30.000Z",
    "addedAt": "2026-10-11T00:31:05.754Z",
    "mediaUrl": "/media/x-2108933446101831946.mp4",
    "posterUrl": "/posters/x/2108933446101831946.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "He's just passing through. \nminimax H3 - Qwen 3 VL\nreferences: GPT\n\nCA: EssC1Dsu51twmpfimTaDxWs87Q4swFdT7jD7tMuipump\n6:",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 7,
      "views": 373,
      "snapshotAt": "2026-10-11T00:31:05.754Z"
    },
    "approvedAt": "2026-10-11T00:31:05.754Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108920235541901318",
    "title": "小兔穿越奇妙世界",
    "titleEn": "A little rabbit travels through imaginary worlds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰白小兔从电脑桌边出发，跃入圆润绿色生物与远方宝塔构成的世界，结尾捧起发光星星。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a small gray-and-white animated rabbit sits at a computer desk. Middle: it jumps over rounded green creatures in a bright landscape with pagodas. End: back at the desk, it holds a glowing star and winks. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chopin_game/status/2108920235541901318",
    "sourceLabel": "X 原帖 · @chopin_game",
    "author": "しょぱん｜物販×エンジニア",
    "publishedAt": "2026-10-10T13:59:00.000Z",
    "addedAt": "2026-10-11T00:31:05.754Z",
    "mediaUrl": "/media/x-2108920235541901318.mp4",
    "posterUrl": "/posters/x/2108920235541901318.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "うちのうさぎが異世界旅行！？🐰🚀\n\nドット絵→浮世絵→深海→宇宙…9つの世界を10秒で駆け抜け、最後はウインク😉\n\nおじぎねこさんの「ClaudeCodeで動画生成」で作成！\n\nClaude Code × https://fal.ai（MiniMax H3）、1本0.6ドル。\n\n#AI生成",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 124,
      "snapshotAt": "2026-10-11T00:31:05.754Z"
    },
    "approvedAt": "2026-10-11T00:31:05.754Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108915299974017401",
    "title": "沙发下猫爪与自动回滚的球",
    "titleEn": "A cat paw and a ball that rolls back",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "地板上的小球靠近沙发边，黑色猫爪从下方伸出拨动它，球又回到猫爪附近。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a small ball sits on the wood floor beside a sofa. Middle: a black cat paw reaches from beneath the sofa and taps the ball. End: the ball is again close to the paw. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "each time the ball rolls back to it by itself",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2108915299974017401",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2108915299974017401",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-10T13:39:23.000Z",
    "addedAt": "2026-10-11T00:31:05.754Z",
    "mediaUrl": "/media/x-2108915299974017401.mp4",
    "posterUrl": "/posters/x/2108915299974017401.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Fetch, but the ball does the work.\n\nPrompt: \"each time the ball rolls back to it by itself\"\n\nDuration: 15s. Camera: locked, floor height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nOnly the paw shows. Keeping the cat under the sofa keeps it on model.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 246,
      "snapshotAt": "2026-10-11T00:31:05.754Z"
    },
    "approvedAt": "2026-10-11T00:31:05.754Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108914789665604069",
    "title": "怀表齿轮里的微缩晨城",
    "titleEn": "A miniature dawn city inside a pocket watch",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "夜间钟表匠打开怀表，齿轮间藏着微缩建筑与行人，结尾晨光照亮了窗边的工作台。 原生视频规格为 15 秒 · 1536×672。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: an elderly watchmaker opens a pocket watch under a workshop lamp at night. Middle: tweezers hover above tiny buildings and people among brass gears. End: sunrise lights the watchmaker and the workshop window. The additional quoted promotional clip is not the primary result. Native-video output: 15s · 1536×672. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Style: cinematic macro fantasy, warm amber workshop lamp light, brass and glass textures, extreme close-ups, shallow depth of field, film grain.\n\n0-5s: One old watchmaker with a magnifying loupe over one eye opens a gold pocket watch on his wooden workbench at night, the dark window behind him showing a sleeping city, the viewer over his shoulder.\n\n5-10s: The viewer pushes into the open watch and reveals a tiny city built among the brass gears, with tiny lit windows and tiny people walking across the turning cogs in darkness, the watchmaker's tweezers lowering in and tightening one small screw.\n\n10-15s: A tiny sun rises over the gears inside the watch, the tiny city fills with morning light, and at the same moment real sunrise floods through the workshop window behind him, the watchmaker slowly looks up at the window, the viewer pulling back.\n\nNegative prompt: more than one watchmaker, digital watch, cartoon style, modern electronics, text, numbers, watermark, slow motion",
    "promptSourceUrl": "https://x.com/DeCat2025/status/2108914789665604069",
    "sourceUrl": "https://x.com/DeCat2025/status/2108914789665604069",
    "sourceLabel": "X 原帖 · @DeCat2025",
    "author": "DeCat",
    "publishedAt": "2026-10-10T13:37:22.000Z",
    "addedAt": "2026-10-11T00:31:05.755Z",
    "mediaUrl": "/media/x-2108914789665604069.mp4",
    "posterUrl": "/posters/x/2108914789665604069.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1536×672",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "fantasy",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Clockwork \nMinimax H3 @Hailuo_AI \n\nStyle: cinematic macro fantasy, warm amber workshop lamp light, brass and glass textures, extreme close-ups, shallow depth of field, film grain.\n\n0-5s: One old watchmaker with a magnifying loupe over one eye opens a gold pocket watch on his wooden workbench at night, the dark window behind him showing a sleeping city, the viewer over his shoulder.\n\n5-10s: The viewer pushes into the open watch and reveals a tiny city built among the brass gears, with tiny lit windows and tiny people walking across the turning cogs in darkness, the watchmaker's tweezers lowering in and tightening one small screw.\n\n10-15s: A tiny sun rises over the gears inside the watch, the tiny city fills with morning light, and at the same moment real sunrise floods through the workshop window behind him, the watchmaker slowly looks up at the window, the viewer pulling back.\n\nNegative prompt: more than one watchmaker, digital watch, cartoon style, modern electronics, text, numbers, watermark, slow motion",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 271,
      "snapshotAt": "2026-10-11T00:31:05.755Z"
    },
    "approvedAt": "2026-10-11T00:31:05.755Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108912699417710956",
    "title": "手机里的小说灵感",
    "titleEn": "Story inspiration on a phone",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女性在夜间灯光下看手机，随后切到日间阅读画面，结尾字幕注明声音与影像制作来源。 原生视频规格为 49 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a woman smiles at her phone beside a warm lamp at night. Middle: another bright indoor shot shows her reading from the phone. End: a white credit slate identifies music and H3 image-and-voice generation. The creator says this is an introduction to their own writing tool. Native-video output: 49s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/akibaAIilab/status/2108912699417710956",
    "sourceLabel": "X 原帖 · @akibaAIilab",
    "author": "あきばあい | 秋葉原IT戦略研究所",
    "publishedAt": "2026-10-10T13:29:03.000Z",
    "addedAt": "2026-10-11T00:31:05.755Z",
    "mediaUrl": "/media/x-2108912699417710956.mp4",
    "posterUrl": "/posters/x/2108912699417710956.jpg",
    "duration": 49,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "自作の小説ツールInkubusの紹介動画を、MiniMax H3で作りました。\n\n実写の女の子も、アニメのキャラクターも、ナレーションの声も、全部手元のPC(RTX 5070 Ti)で生成しています。\n\n構成からYouTubeに上げるまで6時間くらい。動画編集ソフトは一度も開いていません。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 73,
      "snapshotAt": "2026-10-11T00:31:05.755Z"
    },
    "approvedAt": "2026-10-11T00:31:05.755Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108910720788152520",
    "title": "雨中山路上的落石险情",
    "titleEn": "Rockfall on a rain-soaked mountain road",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑色轿车沿雨中山路行驶，峭壁落石逼近道路，结尾切到驾驶员握住方向盘的近景。 原生视频规格为 15 秒 · 478×850。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a black car drives along a wet mountain road below a cliff. Middle: falling rocks descend beside the moving car. End: a close-up shows the driver gripping the steering wheel in the rain. Native-video output: 15s · 478×850. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/VeyraWithAI/status/2108910720788152520",
    "sourceLabel": "X 原帖 · @VeyraWithAI",
    "author": "Veyra | AI & Tech",
    "publishedAt": "2026-10-10T13:21:12.000Z",
    "addedAt": "2026-10-11T00:31:05.755Z",
    "mediaUrl": "/media/x-2108910720788152520.mp4",
    "posterUrl": "/posters/x/2108910720788152520.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "478×850",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI video realism is getting seriously impressive\n\nI tested MiniMax H3 with a 15-second cinematic mountain driving scene featuring heavy rain, falling rocks, and a dramatic landslide\n\nThe realistic car movement, natural camera angles, rain effects, and environmental details really stood out\n\nAll generated from a single text prompt using MiniMax H3\n\nAI filmmaking is moving fast. What do you think of the realism?",
    "engagement": {
      "replies": 48,
      "reposts": 38,
      "likes": 50,
      "views": 610,
      "snapshotAt": "2026-10-11T00:31:05.755Z"
    },
    "approvedAt": "2026-10-11T00:31:05.755Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108904576439156737",
    "title": "月牙舞台上的街舞少女",
    "titleEn": "Street dance on a crescent-lit stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴着耳机的短发少女在月牙灯光前舞动，先抬手变换动作，再侧身撑地，最后站起展开双臂。 原生视频规格为 15 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: an animated short-haired girl in headphones stands on a stage with a crescent-shaped light. Middle: she leans sideways with one hand on the floor. End: she stands and spreads her arms under purple and blue lights. Native-video output: 15s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/neko_no_gebokun/status/2108904576439156737",
    "sourceLabel": "X 原帖 · @neko_no_gebokun",
    "author": "ぽんこつ@AIイラスト",
    "publishedAt": "2026-10-10T12:56:47.000Z",
    "addedAt": "2026-10-11T00:46:12.802Z",
    "mediaUrl": "/media/x-2108904576439156737.mp4",
    "posterUrl": "/posters/x/2108904576439156737.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "久しぶりにダンス担当を登場させました\n\n今回はminimax h3で作成です\nいつものGrokと違う感じも良いですね\n\nなぜ縦向きにしなかったのか…それはわかりません…\n#AI動画",
    "engagement": {
      "replies": 4,
      "reposts": 14,
      "likes": 66,
      "views": 594,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.802Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108899352421032272",
    "title": "黑暗中的金色光环",
    "titleEn": "A golden ring of light in the dark",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑暗中的微小光粒凝聚成垂直光柱，随后形成悬浮的金色环状结构。 原生视频规格为 221 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: sparse fine particles shine on a dark background. Middle: a narrow vertical gold light column rises from a small ring. End: a larger golden particle ring floats in the dark. The source assigns the video to H3, images to an image model and music to a separate music model. Native-video output: 221s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/atomtanstudio/status/2108899352421032272",
    "sourceLabel": "X 原帖 · @atomtanstudio",
    "author": "Rich · Atom Tan Studio",
    "publishedAt": "2026-10-10T12:36:01.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108899352421032272.mp4",
    "posterUrl": "/posters/x/2108899352421032272.jpg",
    "duration": 221,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Last audio and video test before the home stretch.\n\nMusic: YuE2\nImages: GPT Image 2.5\nVideo: Minimax H3\nApplication: Sound / Vision",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 267,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108898691230924810",
    "title": "漂浮屏幕间的分镜与成片对照",
    "titleEn": "Storyboard and final imagery amid floating screens",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右分屏把红色简模预演与动漫角色画面并列，人物穿过漂浮屏幕，结尾切到两张面孔的近景。 原生视频规格为 13 秒 · 1080×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a split view pairs a red mannequin previs on the left with an anime girl on the right amid floating screens. Middle: the two layouts follow similar movement through those screens. End: close-ups compare the red mannequin face and the girl's face. The source explicitly says Blender previs was supplied to H3 as a video reference, so the left is input reference, not an H3 output. Native-video output: 13s · 1080×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tshikimi/status/2108898691230924810",
    "sourceLabel": "X 原帖 · @tshikimi",
    "author": "竹花 樒",
    "publishedAt": "2026-10-10T12:33:24.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108898691230924810.mp4",
    "posterUrl": "/posters/x/2108898691230924810.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "1080×960",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今回、MiniMax H3で動画を生成するにあたって、初めてプレビズをBlenderで作ってみました。H3にプレビズを動画参照で渡しています。プレビズだけではうまくいかなくて、プロンプトでカメラワークをある程度補完してあげたら概ね追従するようになりました。\nhttps://x.com/tshikimi/status/2108878629036003642?s=46",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 164,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108897902982508683",
    "title": "桌边将军字样的多版本生成对照",
    "titleEn": "Version grid of a general-at-the-table scene",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多格对照展示围坐桌边的角色与将军字样，不同版本画面随时间同步变化，结尾渐暗。 原生视频规格为 14 秒 · 1080×1110。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a grid of generated clips shows characters around a table with version labels. Middle: large gold Chinese lettering appears over many cells. End: the clips fade dark. The source is ModelScope's official H3-Turbo release demonstration and explicitly describes the H3 variants, not unrelated models. Native-video output: 14s · 1080×1110. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ModelScope2022/status/2108897902982508683",
    "sourceLabel": "X 原帖 · @ModelScope2022",
    "author": "ModelScope",
    "publishedAt": "2026-10-10T12:30:16.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108897902982508683.mp4",
    "posterUrl": "/posters/x/2108897902982508683.jpg",
    "duration": 14,
    "aspectRatio": "square",
    "resolution": "1080×1110",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Introducing MiniMax-H3-Turbo from Wuli-Art. MiniMax-H3 audio-video generation now runs in as few as four steps. \n🤖 https://modelscope.ai/models/Wuli-Art/MiniMax-H3-Turbo\n\n⚡ Choose from 4-step, 6-step, and 8-step variants for faster audio-video generation.\n🎬 Supports text-to-video, first-frame and last-frame conditioning, with video and audio generated together.\n🧩 The 4-step model uses a compact LoRA-based adapter, while the 6-step and 8-step versions provide complete Diffusers-format transformers.\n📜 MiniMax-H3 Community License Agreement.",
    "engagement": {
      "replies": 7,
      "reposts": 3,
      "likes": 96,
      "views": 5757,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108882784165274101",
    "title": "夜景餐桌边的举杯时刻",
    "titleEn": "A toast beside the night-lit window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名动漫女孩坐在暖灯下的餐桌边，举起酒杯，窗外是灯光闪烁的夜景。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: two animated women sit at a dinner table beside a night-lit window. Middle: they lift stemmed glasses toward each other. End: both hold their glasses while looking across the table. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/cheke6966/status/2108882784165274101",
    "sourceLabel": "X 原帖 · @cheke6966",
    "author": "Snon＠すのん",
    "publishedAt": "2026-10-10T11:30:11.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108882784165274101.mp4",
    "posterUrl": "/posters/x/2108882784165274101.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "乾杯～♪\nゆったりと素敵な時間を過ごそうね😆\n\n作：MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 5,
      "likes": 28,
      "views": 420,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108881049652691250",
    "title": "洞口前的蓝白衣剑舞",
    "titleEn": "Blue-and-white sword movements by the cave",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝白衣人物在洞口前挥动长剑，随后转身变换动作，结尾收剑站定。 原生视频规格为 28 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a figure in blue-and-white clothing raises a sword outside a dark cave. Middle: the figure kneels and turns with the sword extended. End: the figure stands with the sword held near the shoulder. Native-video output: 28s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/uniart_labs/status/2108881049652691250",
    "sourceLabel": "X 原帖 · @uniart_labs",
    "author": "UniArt",
    "publishedAt": "2026-10-10T11:23:18.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108881049652691250.mp4",
    "posterUrl": "/posters/x/2108881049652691250.jpg",
    "duration": 28,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3。 没有用lora，逐帧和姿势识别+结构化事件提示。等稳定了开放。\n#minmax #ai #动作",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 34,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108878629036003642",
    "title": "白色空间里的记忆片段",
    "titleEn": "Fragments of memory in a white space",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫少女站在白色空间与竖柱之间，画面切成文字和角色拼贴，最后回到白底曲名。 原生视频规格为 78 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a monochrome anime girl stands amid tall gray columns on white. Middle: collage-like character fragments and Japanese text share the white space. End: a white title slate reads DON'T DEPRECATE ME. The original credits explicitly assign video to H3 and music to a separate tool. Native-video output: 78s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tshikimi/status/2108878629036003642",
    "sourceLabel": "X 原帖 · @tshikimi",
    "author": "竹花 樒",
    "publishedAt": "2026-10-10T11:13:40.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108878629036003642.mp4",
    "posterUrl": "/posters/x/2108878629036003642.jpg",
    "duration": 78,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "anime",
      "motion-graphics"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "新曲「DON’T DEPRECATE ME」\n\n人の記憶も、AIのように圧縮されて消えていくのだとしたら。\nそれでも、この意識を手放したくない。\n\nMotion graphics: Claude Opus 5.5\nCharacter design: GPT Image 2.5\nPre-Vis: Blender\nVideo: MiniMax H3\nMusic: #Suno",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 9,
      "views": 620,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108876821735555212",
    "title": "雕塑展厅里的穿行镜头",
    "titleEn": "Moving through a sculpture gallery",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头沿暖色博物馆展厅前进，穿过雕塑与玻璃陈列柜之间，远处人物进入视线。 原生视频规格为 5 秒 · 736×576。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a warm-lit sculpture gallery has glass display cases along both sides. Middle: the view advances toward a small yellow-clad figure among the displays. End: the camera reaches a wider hall where visitors stand beside exhibits. Native-video output: 5s · 736×576. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/deminishirate/status/2108876821735555212",
    "sourceLabel": "X 原帖 · @deminishirate",
    "author": "デミニシラテ",
    "publishedAt": "2026-10-10T11:06:29.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108876821735555212.mp4",
    "posterUrl": "/posters/x/2108876821735555212.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "736×576",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3だと限界があるな",
    "engagement": {
      "replies": 1,
      "reposts": 10,
      "likes": 21,
      "views": 717,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108874506974478545",
    "title": "办公桌旁的彩色小伙伴",
    "titleEn": "Colorful companions around the office desks",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "彩色圆形角色在明亮办公室里活动，红色角色手持平板，伙伴们在桌边聚集，结尾出现发布预告。 原生视频规格为 32 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: colorful round animated characters stand beside office desks, and a red character holds a tablet. Middle: more companions gather among desks and windows. End: they line up under a Soren release announcement. Native-video output: 32s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Eileeness160708/status/2108874506974478545",
    "sourceLabel": "X 原帖 · @Eileeness160708",
    "author": "Soren",
    "publishedAt": "2026-10-10T10:57:18.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108874506974478545.mp4",
    "posterUrl": "/posters/x/2108874506974478545.jpg",
    "duration": 32,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax-H3制作，真的很不错ε=(´ο｀*)))唉~~",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 31,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108872800413024718",
    "title": "霓虹小巷里的夜战片段",
    "titleEn": "Night confrontation in a neon alley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从蓝绿色霓虹小巷展开，紫衣女性出现在围栏前，结尾男子倚墙挥手。 原生视频规格为 30 秒 · 1280×700。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: the camera faces a wet alley lit by cyan neon. Middle: a purple-clad woman appears in front of a wire fence. End: a white-haired man leans beside a wall and raises his hands. Native-video output: 30s · 1280×700. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/zcl_cupid0713/status/2108872800413024718",
    "sourceLabel": "X 原帖 · @zcl_cupid0713",
    "author": "SRvZcL AIGC",
    "publishedAt": "2026-10-10T10:50:31.000Z",
    "addedAt": "2026-10-11T00:46:12.803Z",
    "mediaUrl": "/media/x-2108872800413024718.mp4",
    "posterUrl": "/posters/x/2108872800413024718.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1280×700",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "赛博都市夜战  MiniMax H3 练手。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 39,
      "snapshotAt": "2026-10-11T00:46:12.803Z"
    },
    "approvedAt": "2026-10-11T00:46:12.803Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108856249710461314",
    "title": "门口自拍中的轻声表演",
    "titleEn": "A softly performed doorway selfie",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏镜头从门口开始，一位黄衣女性走近自拍镜头，侧头说话，结尾看向画面前方。 原生视频规格为 12 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: the primary 12-second clip shows a dark wooden doorway. Middle: a woman in a yellow outfit and white cardigan leans toward the selfie camera. End: she faces the camera and gestures near her chest. Additional attachments include a shorter variant; the primary inspected result is attributed to local H3 in the source. Native-video output: 12s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mr_XiaoZou/status/2108856249710461314",
    "sourceLabel": "X 原帖 · @Mr_XiaoZou",
    "author": "sugar",
    "publishedAt": "2026-10-10T09:44:45.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108856249710461314.mp4",
    "posterUrl": "/posters/x/2108856249710461314.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "本地 minimax-h3\n\n12s 长版和5s短版\n\nhttps://x.com/li9292/status/2105634377048682577?s=20",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 5,
      "views": 248,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108854933525319731",
    "title": "粉色舞台上的双人舞步",
    "titleEn": "A duet on a pink stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发与粉发动漫角色在粉色舞台上跳舞，鼓组和音箱位于身后，结尾切到两人的墨镜近景。 原生视频规格为 26 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: silver-haired and pink-haired anime women raise their hands before a drum kit on a pink stage. Middle: both step and dance in front of amplifiers. End: a close-up shows them wearing colorful sunglasses. The original caption names this duo and attributes the performance to H3. Native-video output: 26s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2108854933525319731",
    "sourceLabel": "X 原帖 · @Mayz1169",
    "author": "Kiki",
    "publishedAt": "2026-10-10T09:39:31.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108854933525319731.mp4",
    "posterUrl": "/posters/x/2108854933525319731.jpg",
    "duration": 26,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [
      "anime"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Emilia & Ram doing the APT dance... 😭💜🩷\n\nmade with MiniMax H3 on @renoiseai ✨",
    "engagement": {
      "replies": 0,
      "reposts": 8,
      "likes": 104,
      "views": 4493,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108854092953166021",
    "title": "魔法校园里的恋爱预告",
    "titleEn": "A romance trailer in a magical school",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "同人预告中，一对年轻人在魔法学校背景下亲吻，结尾切到即将推出的黑色字幕。 原生视频规格为 161 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a dark opening precedes the trailer. Middle: a blond woman and a young man kiss in a blue-gray scene. End: a black COMING SOON title closes the clip. The author explicitly says they made a fan trailer using H3 for face/character swapping, with editing, music and voice attributed separately. Native-video output: 161s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wangdoudou2024/status/2108854092953166021",
    "sourceLabel": "X 原帖 · @wangdoudou2024",
    "author": "王抖抖",
    "publishedAt": "2026-10-10T09:36:11.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108854092953166021.mp4",
    "posterUrl": "/posters/x/2108854092953166021.jpg",
    "duration": 161,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "When the Slytherin Young Lady Falls for the Gryffindor Boy (Hogwarts Side Story)\nRecently, genderbent Malfoy has been quite popular, so I made a fan trailer. Main tools used: MiniMax H3 (video face/character swap) + Opus (editing) + Suno (music) + ElevenLabs (voice and sound).",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 131,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108851182244606445",
    "title": "浣熊抢薯片的后座插曲",
    "titleEn": "Raccoon and flying chips in the back seat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "汽车后座里，黄衣女性递出薯片，浣熊和小猴围着包装袋，结尾薯片散落在座椅上。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a yellow-sweater woman sits beside a raccoon and small monkey in a beige car interior. Middle: she offers a chip while the raccoon reaches toward the bag. End: chips scatter over the back seat as she smiles. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic video inside the back seat of a moving car with beige leather interior On the right a young woman with medium-length brown hair, wearing a yellow knitted sweater and blue jeans is holding a blue bag of potato chips Sitting on the left is a fluffy raccoon and a small capuchin monkey is resting close to the raccoon The woman offers a chip to the animals but as the raccoon reaches into the bag chips accidentally burst into the air and scatter all over the backseat including one landing on the raccoon's head The woman reacts with surprise and then bursts into laughter while the monkey looks around curious and playful Photorealistic soft natural lighting, vibrant colors.",
    "promptSourceUrl": "https://x.com/laviniavelle/status/2108851182244606445",
    "sourceUrl": "https://x.com/laviniavelle/status/2108851182244606445",
    "sourceLabel": "X 原帖 · @laviniavelle",
    "author": "Lavinia",
    "publishedAt": "2026-10-10T09:24:37.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108851182244606445.mp4",
    "posterUrl": "/posters/x/2108851182244606445.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A funny little moment turned into pure chaos The raccoon monkey and flying chips made this scene unforgettable.\n\nMade with MiniMax H3\n\nPrompt:\n\nA cinematic video inside the back seat of a moving car with beige leather interior On the right a young woman with medium-length brown hair, wearing a yellow knitted sweater and blue jeans is holding a blue bag of potato chips Sitting on the left is a fluffy raccoon and a small capuchin monkey is resting close to the raccoon The woman offers a chip to the animals but as the raccoon reaches into the bag chips accidentally burst into the air and scatter all over the backseat including one landing on the raccoon's head The woman reacts with surprise and then bursts into laughter while the monkey looks around curious and playful Photorealistic soft natural lighting, vibrant colors.",
    "engagement": {
      "replies": 33,
      "reposts": 7,
      "likes": 188,
      "views": 8463,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108846428919623711",
    "title": "沙漠旅途与金字塔伙伴",
    "titleEn": "Desert travelers beside a pyramid",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫冒险伙伴穿过沙漠，站到金字塔前，随后画面切到其中一人在室内举手的场景。 原生视频规格为 26 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a group of colorful anime adventurers walks through a sandy landscape. Middle: the group poses in front of a pyramid. End: one blue-and-red character raises a hand in an interior scene with Japanese title text. Native-video output: 26s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/maru_999_/status/2108846428919623711",
    "sourceLabel": "X 原帖 · @maru_999_",
    "author": "Maru999",
    "publishedAt": "2026-10-10T09:05:43.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108846428919623711.mp4",
    "posterUrl": "/posters/x/2108846428919623711.jpg",
    "duration": 26,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "この道わが旅 #2 砂漠とピラミッド。砂漠を越えて、ピラミッドへ。\n生成: Kairos（MiniMax H3）\n#この道わが旅 #AIアニメ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 45,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108843102417113181",
    "title": "小猫骑士与巨龙的拥抱",
    "titleEn": "A kitten knight embraces a giant dragon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿银色盔甲的小猫站在草地上面对巨龙，随后坐倒，结尾张开双臂拥抱龙的脸。 原生视频规格为 15 秒 · 480×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a tiny armored ginger-and-white kitten faces an enormous dragon in a meadow. Middle: the kitten sits down near its small sword while the dragon lowers its head. End: the kitten reaches up to embrace the dragon's snout under golden light. Native-video output: 15s · 480×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A cinematic, ultra-realistic fantasy scene featuring an adorable fluffy ginger-and-white kitten warrior wearing miniature silver medieval armor, a flowing cream-colored cape, and carrying a tiny sword, standing bravely in a misty green meadow facing a massive, fearsome dragon covered in dark, rugged scales, sharp horns, and powerful leathery wings. The camera captures dramatic low-angle shots, close-ups of the kitten gripping its sword, and intense shots of the dragon lowering its enormous head toward the tiny warrior. The kitten initially stands courageously, but as the dragon approaches, it becomes frightened and drops its sword, looking up with teary eyes and a trembling expression. Suddenly, the mood shifts from danger to tenderness as the dragon gently touches its snout against the kitten, revealing a surprising bond between them. The kitten embraces the dragon's enormous face affectionately, and the dragon closes its glowing amber eyes peacefully. Golden sunset light illuminates the meadow, creating a magical, emotional atmosphere with volumetric lighting, realistic fur and scales, detailed metallic armor, cinematic depth of field, smooth camera movements, dramatic wide shots, intimate close-ups, 4K photorealistic CGI, and an epic fantasy film aesthetic, transitioning from suspense and fear to warmth, friendship, and heartwarming emotional connection.",
    "promptSourceUrl": "https://x.com/LettyVesper/status/2108843102417113181",
    "sourceUrl": "https://x.com/LettyVesper/status/2108843102417113181",
    "sourceLabel": "X 原帖 · @LettyVesper",
    "author": "Letty",
    "publishedAt": "2026-10-10T08:52:30.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108843102417113181.mp4",
    "posterUrl": "/posters/x/2108843102417113181.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "480×480",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tiny warrior, giant dragon, and an unexpected friendship.\n\nCreated with MiniMax H3 Fast\n\nPrompt:\n\nA cinematic, ultra-realistic fantasy scene featuring an adorable fluffy ginger-and-white kitten warrior wearing miniature silver medieval armor, a flowing cream-colored cape, and carrying a tiny sword, standing bravely in a misty green meadow facing a massive, fearsome dragon covered in dark, rugged scales, sharp horns, and powerful leathery wings. The camera captures dramatic low-angle shots, close-ups of the kitten gripping its sword, and intense shots of the dragon lowering its enormous head toward the tiny warrior. The kitten initially stands courageously, but as the dragon approaches, it becomes frightened and drops its sword, looking up with teary eyes and a trembling expression. Suddenly, the mood shifts from danger to tenderness as the dragon gently touches its snout against the kitten, revealing a surprising bond between them. The kitten embraces the dragon's enormous face affectionately, and the dragon closes its glowing amber eyes peacefully. Golden sunset light illuminates the meadow, creating a magical, emotional atmosphere with volumetric lighting, realistic fur and scales, detailed metallic armor, cinematic depth of field, smooth camera movements, dramatic wide shots, intimate close-ups, 4K photorealistic CGI, and an epic fantasy film aesthetic, transitioning from suspense and fear to warmth, friendship, and heartwarming emotional connection.",
    "engagement": {
      "replies": 4,
      "reposts": 5,
      "likes": 188,
      "views": 762,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108822646427361556",
    "title": "末世房间里的屏幕来客",
    "titleEn": "A visitor on the screen in an apocalyptic room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发动漫角色在昏暗房间里与屏幕上的蓝发角色相对，中段坐在设备旁，结尾走向金属门。 原生视频规格为 259 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a silver-haired anime woman faces a monitor showing a blue-haired woman in a worn, dim room. Middle: the silver-haired woman sits by the bed and glowing equipment. End: she walks toward a heavy metal door in a red-and-black outfit. The original explicitly attributes the 4-minute episode to locally generated H3. Native-video output: 259s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shizuha_akari/status/2108822646427361556",
    "sourceLabel": "X 原帖 · @shizuha_akari",
    "author": "しずは（あかり）",
    "publishedAt": "2026-10-10T07:31:13.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108822646427361556.mp4",
    "posterUrl": "/posters/x/2108822646427361556.jpg",
    "duration": 259,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "※音声等注意！\n終末のしずはとあかりエピソード２\n\n暗いのでいろいろ注意です💦\n4分18秒あります\n\nMiniMax H3でローカル生成です",
    "engagement": {
      "replies": 5,
      "reposts": 6,
      "likes": 39,
      "views": 1867,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108732767609446699",
    "title": "霓虹街巷里的粉发身影",
    "titleEn": "A pink-haired figure in neon streets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发角色穿过雨夜街巷，霓虹映在湿地面上，结尾切到聚集人群的未来感夜店。 原生视频规格为 45 秒 · 1536×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a pink-haired character walks along a rain-wet neon street. Middle: the character passes a dark doorway beneath bright signs. End: a crowded futuristic nightclub glows with cyan and magenta lighting. The author explicitly identifies the primary nightclub comparison as H3 renders with an experimental LoRA enabled and disabled. Native-video output: 45s · 1536×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EndFolding79421/status/2108732767609446699",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-10-10T01:34:04.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108732767609446699.mp4",
    "posterUrl": "/posters/x/2108732767609446699.jpg",
    "duration": 45,
    "aspectRatio": "landscape",
    "resolution": "1536×864",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "dark"
    ],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing the CW Dark Cyberpunk Part 1 LoRA with MiniMax H3 in ComfyUI.\n\nThe creator has mentioned that this LoRA is still experimental, so keep that in mind when comparing the results.\n\nA few comparisons with the LoRA enabled and disabled at different step counts:\n\n🔹 Subway scene (white-haired man): Both tests at 20 steps, one with the LoRA and one without.\n\n🔹 Cyberpunk nightclub (pink-haired woman): Two tests at 8 steps, one with the LoRA and one without.\n\n🔹 Additional nightclub test: LoRA enabled at the recommended 30 steps. This render took approximately 1 hour and 17 minutes.\n\nJust sharing the results so everyone can compare the differences in lighting, environments, character consistency, and overall quality.\n\nLink: https://civitai.red/models/2993011/city-worlds-dark-cyberpunk-minimax-h3?modelVersionId=3394129",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 14,
      "views": 1080,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108726243025711203",
    "title": "退潮后海滩上的静默人群",
    "titleEn": "Silent figures revealed by the receding tide",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黄雨衣女性站在灰蓝色海边，退去的海水露出成排静立的人影，结尾人群面向岸边。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Beginning: a yellow-raincoat woman stands facing a gray sea. Middle: pale human figures emerge in rows across exposed wet sand. End: a closer view shows the figures turned toward the shore with waves behind them. The second attachment is a quoted promotional clip, not this primary H3 result. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "15-second realistic cinematic horror sequence, shot on 35mm film, grey overcast beach at dusk, cold blue-grey colors, wet sand, subtle film grain.  Exactly one person: a woman in a yellow raincoat standing on the shore.  \n\n[0-5s] Wide shot of a calm beach. The woman stands at the water's edge looking out at the sea. Suddenly the water starts pulling back fast, far faster than any normal tide, exposing wet sand and rocks for hundreds of meters.  \n\n[5-10s] The viewer rises high above her. As the sea drains away, the exposed seabed reveals rows and rows of pale human figures standing perfectly still on the sand, dozens of them, all facing the ocean, their backs to the woman. \n\n[10-15s] Close-up on the woman's face as she steps back in horror. Cut to wide shot: all the figures on the seabed turn around at the exact same moment and face the viewer. In the far distance, a massive wave is rising.  \n\nNegative prompt: cartoon, anime, 3D render, sunny weather, bright colors, swimmers, figures moving out of sync, static scene, nothing happening, distorted faces, extra limbs, morphing, flickering, slow motion, watermark",
    "promptSourceUrl": "https://x.com/DeCat2025/status/2108726243025711203",
    "sourceUrl": "https://x.com/DeCat2025/status/2108726243025711203",
    "sourceLabel": "X 原帖 · @DeCat2025",
    "author": "DeCat",
    "publishedAt": "2026-10-10T01:08:09.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108726243025711203.mp4",
    "posterUrl": "/posters/x/2108726243025711203.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The Tide\nMinimax H3 @Hailuo_AI \n\n15-second realistic cinematic horror sequence, shot on 35mm film, grey overcast beach at dusk, cold blue-grey colors, wet sand, subtle film grain.  Exactly one person: a woman in a yellow raincoat standing on the shore.  \n\n[0-5s] Wide shot of a calm beach. The woman stands at the water's edge looking out at the sea. Suddenly the water starts pulling back fast, far faster than any normal tide, exposing wet sand and rocks for hundreds of meters.  \n\n[5-10s] The viewer rises high above her. As the sea drains away, the exposed seabed reveals rows and rows of pale human figures standing perfectly still on the sand, dozens of them, all facing the ocean, their backs to the woman. \n\n[10-15s] Close-up on the woman's face as she steps back in horror. Cut to wide shot: all the figures on the seabed turn around at the exact same moment and face the viewer. In the far distance, a massive wave is rising.  \n\nNegative prompt: cartoon, anime, 3D render, sunny weather, bright colors, swimmers, figures moving out of sync, static scene, nothing happening, distorted faces, extra limbs, morphing, flickering, slow motion, watermark",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 9,
      "views": 428,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108981136093290880",
    "title": "城市与雪山场景的三屏对照",
    "titleEn": "Three-panel comparison of city and mountain scenes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "三段带模型标注的画面并列展示城市车流、雪山人物与夜间天际线，其中顶部明确标注H3。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: a three-panel comparison shows moving vehicles on city roads. Middle: all three panels depict figures amid snowy mountains. End: they show night city skylines. The top panel is visibly labeled MiniMax H3 throughout; the other two are Wan 3 and Seedance. This is an attributed comparison, not a claim that every panel is H3. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxVideoAI/status/2108981136093290880",
    "sourceLabel": "X 原帖 · @MaxVideoAI",
    "author": "MaxVideoAI",
    "publishedAt": "2026-10-10T18:01:00.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108981136093290880.mp4",
    "posterUrl": "/posters/x/2108981136093290880.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "city",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "City, ocean, mountains, desert, neon skyline. One impossible flight, three takes. Top: MiniMax H3. Middle: Wan 3. Bottom: Seedance 2.0 Fast. Fictional scenes compared on MaxVideoAI. Which feels fastest? #SupersonicFlight",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 51,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2108973710665519305",
    "title": "双屏小龙动作对照",
    "titleEn": "Side-by-side movements of a little dragon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白底双屏展示蓝色小龙的抬翼与蜷身动作，左边标注H3，右边标注三维制作工具。 原生视频规格为 15 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Beginning: two blue baby dragons stand side by side on white. Middle: both raise their wings in corresponding poses. End: the dragons curl into resting positions. The left panel is visibly labeled MiniMax H3, while the right panel is labeled Tripo3d plus Opus5.5 plus Blender. Additional animal rig attachments are not the primary inspected result. Native-video output: 15s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2108973710665519305",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-10-10T17:31:30.000Z",
    "addedAt": "2026-10-11T00:46:12.804Z",
    "mediaUrl": "/media/x-2108973710665519305.mp4",
    "posterUrl": "/posters/x/2108973710665519305.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "creatures",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ドラゴンのリギングとモーション作成を、\n\nTripo3d で作成したモデル +  MiniMax H3 で作った動画を参照させながら Opus 5.5 にやらせてみました\n\n修正の余地はあるものの、ほぼ自動でここまでできるのは 凄いです\n\n360度動画から高精度ポーズ推定するのは\n思いつきましたが、引用ポストを見るまで\nリギングすることは思いつかなかったです\n\nとても参考になりました",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 9,
      "views": 322,
      "snapshotAt": "2026-10-11T00:46:12.804Z"
    },
    "approvedAt": "2026-10-11T00:46:12.804Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102247118484463815",
    "title": "锅炉房里的利爪惊吓",
    "titleEn": "A clawed encounter in the boiler room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红绿条纹男子在昏暗锅炉房逼近对手，金属利爪掠过脸前，镜头切到其狰狞表情。 原生视频规格为 12 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in a red-and-green striped shirt approaches another man in a dim boiler room. A metal claw passes close to the other man's face, followed by a close-up of the clawed figure's grin. No different-model watermark is visible. Native-video output: 12s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tech2Wild/status/2102247118484463815",
    "sourceLabel": "X 原帖 · @Tech2Wild",
    "author": "Tech2Wild",
    "publishedAt": "2026-09-22T04:02:25.000Z",
    "addedAt": "2026-10-12T00:13:20.506Z",
    "mediaUrl": "/media/x-2102247118484463815.mp4",
    "posterUrl": "/posters/x/2102247118484463815.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just Created One Of My Favorite Freddy Kruger Scenes with his new 1980s Lora for Minimax H3 LMAO",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 14,
      "views": 3932,
      "snapshotAt": "2026-10-12T00:13:20.506Z"
    },
    "approvedAt": "2026-10-12T00:13:20.506Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102243698256052509",
    "title": "夜视卧室中的床边来客",
    "titleEn": "A bedside visitor in night vision",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "固定夜视镜头对着熟睡女子，白衣人物出现在床边，昏暗卧室营造出诡异氛围。 原生视频规格为 10 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fixed night-vision view shows a woman sleeping in a bed. A white-clad figure appears beside the bed and remains standing near her in the dim room. No different-model watermark is visible. Native-video output: 10s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/himazinwarota/status/2102243698256052509",
    "sourceLabel": "X 原帖 · @himazinwarota",
    "author": "らっきー☆",
    "publishedAt": "2026-09-22T03:48:50.000Z",
    "addedAt": "2026-10-12T00:13:20.506Z",
    "mediaUrl": "/media/x-2102243698256052509.mp4",
    "posterUrl": "/posters/x/2102243698256052509.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のここがすごい！\n最近流行ってる猫が寝てる女性の横でウクレレ弾くミームも適当なプロンプトでかなり再現度高めに生成できる。これがローカルで無限に生成できる。",
    "engagement": {
      "replies": 3,
      "reposts": 1,
      "likes": 18,
      "views": 1771,
      "snapshotAt": "2026-10-12T00:13:20.506Z"
    },
    "approvedAt": "2026-10-12T00:13:20.506Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102241881501688122",
    "title": "粉色杂志封面的动态人像",
    "titleEn": "A moving portrait on a pink magazine cover",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏女子人像融入粉色杂志版式，眼镜近景、伸手动作和全身造型交替呈现。 原生视频规格为 15 秒 · 1088×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary vertical video presents a woman against pink magazine-style graphics: a glasses close-up, a reaching hand, then a full-body outfit pose. A second video shows another pink magazine portrait. No different-model mark is visible in either inspected video. Native-video output: 15s · 1088×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PhotogenicWeekE/status/2102241881501688122",
    "sourceLabel": "X 原帖 · @PhotogenicWeekE",
    "author": "Photogenic Weekend",
    "publishedAt": "2026-09-22T03:41:36.000Z",
    "addedAt": "2026-10-12T00:13:20.507Z",
    "mediaUrl": "/media/x-2102241881501688122.mp4",
    "posterUrl": "/posters/x/2102241881501688122.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1088×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "元ネタはこれなんだけど、そのままじゃ面白くないので、エージェントに衣装とか適当に変えて！って頼んだら自分(mii)出してきた(笑) こいつ本当にAIか？最近やたら自我が強いw MiniMax H3 ref2va 1080p, 15s",
    "engagement": {
      "replies": 3,
      "reposts": 2,
      "likes": 65,
      "views": 6834,
      "snapshotAt": "2026-10-12T00:13:20.507Z"
    },
    "approvedAt": "2026-10-12T00:13:20.507Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102233975901344134",
    "title": "橙色墨镜巴哥的杂志大片",
    "titleEn": "An orange-sunglasses pug magazine sequence",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴橙色墨镜的巴哥犬在鲜艳杂志版式中切换近景与封面，文字和图形围绕宠物展开。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary video shows a pug wearing orange sunglasses, moving between close-ups and bright magazine-cover layouts with changing lettering. A second video presents a colorful female portrait magazine layout. No different-model watermark is visible. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/taya_mama_AI/status/2102233975901344134",
    "sourceLabel": "X 原帖 · @taya_mama_AI",
    "author": "た〜や｜スキマ時間×AI副業ママ",
    "publishedAt": "2026-09-22T03:10:12.000Z",
    "addedAt": "2026-10-12T00:13:20.507Z",
    "mediaUrl": "/media/x-2102233975901344134.mp4",
    "posterUrl": "/posters/x/2102233975901344134.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ちょ、待って！！\n動物の写真で作ったら可愛すぎて反則😻\n\nぜひペットの写真とかで作ってみてほしい～🐾\n\n【🎥MiniMax H3 max turbo】",
    "engagement": {
      "replies": 2,
      "reposts": 14,
      "likes": 133,
      "views": 9564,
      "snapshotAt": "2026-10-12T00:13:20.507Z"
    },
    "approvedAt": "2026-10-12T00:13:20.507Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102224175251567006",
    "title": "熔岩铁拳赛事海报动起来",
    "titleEn": "A lava-fist tournament poster comes alive",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "赛事海报左右并排对照，右侧熔岩拳头、火光和标题逐渐显现，展示静态设计的动态效果。 原生视频规格为 9 秒 · 2160×2110。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two tournament posters are shown side by side. The right-hand version animates a glowing lava fist, fire and lettering while the left remains a reference poster. The caption explicitly credits H3 plus After Effects; no other-model watermark is visible. Native-video output: 9s · 2160×2110. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/JOY_teeth/status/2102224175251567006",
    "sourceLabel": "X 原帖 · @JOY_teeth",
    "author": "蓬博",
    "publishedAt": "2026-09-22T02:31:15.000Z",
    "addedAt": "2026-10-12T00:13:20.507Z",
    "mediaUrl": "/media/x-2102224175251567006.mp4",
    "posterUrl": "/posters/x/2102224175251567006.jpg",
    "duration": 9,
    "aspectRatio": "square",
    "resolution": "2160×2110",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#新作上线 末日铁拳水友赛 \n左: 海报 右: MiniMax H3+AE\nClient: Once Again\nDesign / Motion Design: Pengbo\nAI-assisted Creation: GPT / MiniMax\n\nCreated by joyteeth~®",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 484,
      "snapshotAt": "2026-10-12T00:13:20.507Z"
    },
    "approvedAt": "2026-10-12T00:13:20.507Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102213666557055467",
    "title": "雪中神社前的耳罩少女",
    "titleEn": "An earmuff portrait in a snowy shrine",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "围着白色围巾的少女在雪中神社前抬头、看向手心，再转向镜头微笑。 原生视频规格为 8 秒 · 1824×1056。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary video shows a young woman wearing fluffy earmuffs and a white scarf beside a snowy shrine, looking upward, down toward her hands, then smiling at the camera. The second video shows a red-clad fantasy character with glowing eyes. No different-model watermark is visible in either inspected clip. Native-video output: 8s · 1824×1056. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kentdhani/status/2102213666557055467",
    "sourceLabel": "X 原帖 · @kentdhani",
    "author": "Kent Dhani",
    "publishedAt": "2026-09-22T01:49:30.000Z",
    "addedAt": "2026-10-12T00:13:20.507Z",
    "mediaUrl": "/media/x-2102213666557055467.mp4",
    "posterUrl": "/posters/x/2102213666557055467.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1824×1056",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 Ref2V #4 \n\nA new world, watch it hold.\n\nPrompt below 👇",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 26,
      "views": 24746,
      "snapshotAt": "2026-10-12T00:13:20.507Z"
    },
    "approvedAt": "2026-10-12T00:13:20.507Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102204094240952326",
    "title": "雨窗与餐桌间的告别歌",
    "titleEn": "A farewell song between rain and the table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏音乐短片在雨窗旁的人像和餐桌场景间切换，画面叠加日文歌词，末尾列出制作工具。 原生视频规格为 190 秒 · 704×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A vertical music video shows a woman beside a rain-covered window, later sitting at a table, with Japanese lyric text over the pictures. The ending credits list MiniMax H3 alongside other audio and image tools. The original caption attributes video generation to H3 and audio to separate tools; no conflicting video-model watermark is visible. Native-video output: 190s · 704×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mazzo/status/2102204094240952326",
    "sourceLabel": "X 原帖 · @mazzo",
    "author": "Koya Matsuo",
    "publishedAt": "2026-09-22T01:11:27.000Z",
    "addedAt": "2026-10-12T00:13:20.507Z",
    "mediaUrl": "/media/x-2102204094240952326.mp4",
    "posterUrl": "/posters/x/2102204094240952326.jpg",
    "duration": 190,
    "aspectRatio": "portrait",
    "resolution": "704×1280",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "YuE2で作った曲を妻音源とりちゃん[AI]にRVCで置き換え。自作MV制作システムLaVialeを使い、MiniMax H3で生成",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 8,
      "views": 1013,
      "snapshotAt": "2026-10-12T00:13:20.507Z"
    },
    "approvedAt": "2026-10-12T00:13:20.507Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102191822730661937",
    "title": "废墟武者唤出紫色雷龙",
    "titleEn": "A warrior summons a purple lightning dragon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "废墟中的武者凝聚紫色能量，雷龙在身旁显现，镜头最后回到人物正面。 原生视频规格为 18 秒 · 1952×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A warrior stands amid ruined stone structures. Purple energy grows around his arms and forms a luminous dragon, then the camera returns to his face. No different-model watermark is visible. Native-video output: 18s · 1952×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PCasino73066/status/2102191822730661937",
    "sourceLabel": "X 原帖 · @PCasino73066",
    "author": "⟡ 𝕀𝕁 ⟡ ℙ𝕒𝕥 ℂ𝕒𝕤𝕚𝕟𝕠 ⟡ 𝕀𝕁 ⟡",
    "publishedAt": "2026-09-22T00:22:42.000Z",
    "addedAt": "2026-10-12T00:13:20.507Z",
    "mediaUrl": "/media/x-2102191822730661937.mp4",
    "posterUrl": "/posters/x/2102191822730661937.jpg",
    "duration": 18,
    "aspectRatio": "landscape",
    "resolution": "1952×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "DragonSprite 1Take! Comfyui Pipeline Minimax H3. @MiniMax_AI @ComfyUI  #aiart #art #aiartcommunity #digitalart",
    "engagement": {
      "replies": 5,
      "reposts": 0,
      "likes": 21,
      "views": 361,
      "snapshotAt": "2026-10-12T00:13:20.507Z"
    },
    "approvedAt": "2026-10-12T00:13:20.507Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102188550024200689",
    "title": "夕阳下的豹纹角色与飘带",
    "titleEn": "A spotted character with ribbons at sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "豹纹拟人角色在夕阳中转身抬手，黑色飘带随动作旋转，另附沙漠人物的流动衣袍片段。 原生视频规格为 10 秒 · 768×1376。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary vertical animation shows a spotted feline humanoid turning and raising her arms in golden sunset light as dark ribbons swirl. A second animation shows a woman in patterned flowing robes in a sandy landscape. No different-model watermark is visible. Native-video output: 10s · 768×1376. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mank_Tibbit/status/2102188550024200689",
    "sourceLabel": "X 原帖 · @Mank_Tibbit",
    "author": "Mank Tibbit",
    "publishedAt": "2026-09-22T00:09:41.000Z",
    "addedAt": "2026-10-12T00:25:24.751Z",
    "mediaUrl": "/media/x-2102188550024200689.mp4",
    "posterUrl": "/posters/x/2102188550024200689.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1376",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 testing using a tweaked version of @Skyebrows prompt. https://x.com/Skyebrows/status/2102072865356919163?s=20\n\nCame out a bit \"grainy\"! HAH!  Need to tweak for quality.\nI hadn't prompted the singing either, but pleased with the lipsync of whatever it is she's singing - seems to be authentic frontier Minimax gibberish.",
    "engagement": {
      "replies": 4,
      "reposts": 3,
      "likes": 45,
      "views": 896,
      "snapshotAt": "2026-10-12T00:25:24.751Z"
    },
    "approvedAt": "2026-10-12T00:25:24.751Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102186614533320991",
    "title": "红披风与绿光剑的动画交锋",
    "titleEn": "An animated clash of a red cape and green blade",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红披风少年与黑发剑士在废墟场景中交锋，绿色光剑与拳击碰撞出明亮闪光。 原生视频规格为 39 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An anime-style sequence shows a young red-caped fighter and a dark-haired swordsman in a ruined setting. A bright green blade collides with a punch in a flash. No different-model watermark is visible. Native-video output: 39s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Larhalt/status/2102186614533320991",
    "sourceLabel": "X 原帖 · @Larhalt",
    "author": "Nop",
    "publishedAt": "2026-09-22T00:02:00.000Z",
    "addedAt": "2026-10-12T00:25:24.751Z",
    "mediaUrl": "/media/x-2102186614533320991.mp4",
    "posterUrl": "/posters/x/2102186614533320991.jpg",
    "duration": 39,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "For those on a budget who don’t want to use Seedance for everything, MiniMax H3 is a pretty good alternative at roughly half the price.\n\nThanks to Pollo AI CPP for the credits! 🙏 These days, @itsPolloAI  also has MCP support that connects with AI agents, so you can basically control it with natural language.\n\nGenerated with Minimax H3 on #PolloAI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 152,
      "snapshotAt": "2026-10-12T00:25:24.751Z"
    },
    "approvedAt": "2026-10-12T00:25:24.751Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102186301843795993",
    "title": "毛绒玩具房间通向彩虹城堡",
    "titleEn": "A plush-filled room opens onto a rainbow castle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝发角色穿行于悬挂毛绒玩具的房间，镜头掠过玩具与墙上相框，最后望向云端城堡和彩虹。 原生视频规格为 15 秒 · 1080×1890。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired anime character stands in a room filled with hanging plush bears and framed pictures. The view moves past toys toward a floating castle and rainbow in a bright sky. No different-model watermark is visible. Native-video output: 15s · 1080×1890. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/KiraraOwlbear18/status/2102186301843795993",
    "sourceLabel": "X 原帖 · @KiraraOwlbear18",
    "author": "Kirara-Owlbear 🦉",
    "publishedAt": "2026-09-22T00:00:45.000Z",
    "addedAt": "2026-10-12T00:25:24.751Z",
    "mediaUrl": "/media/x-2102186301843795993.mp4",
    "posterUrl": "/posters/x/2102186301843795993.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1890",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Furina フリーナ \n#GenshinImpact #原神 \n\nMinimax h3 on @wavespeed_ai\n@HiWscreator",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 35,
      "views": 2180,
      "snapshotAt": "2026-10-12T00:25:24.751Z"
    },
    "approvedAt": "2026-10-12T00:25:24.751Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102173982292103599",
    "title": "猫狗骑行掠过霓虹光轨",
    "titleEn": "A cat-and-dog ride with neon trails",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "猫和狗并排出现在行驶场景中，红色光轨沿车辆两侧延伸，蓝紫光晕为画面增加特效。 原生视频规格为 10 秒 · 768×1376。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat and dog ride together in a moving scene with palm trees behind them. Red neon streaks and blue-purple glow run along the vehicle. The original author identifies this as an H3 VFX Edit LoRA demonstration, not a spoken tutorial; no different-model watermark is visible. Native-video output: 10s · 768×1376. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alissonerdx/status/2102173982292103599",
    "sourceLabel": "X 原帖 · @Alissonerdx",
    "author": "Alisson Pereira",
    "publishedAt": "2026-09-21T23:11:48.000Z",
    "addedAt": "2026-10-12T00:25:24.751Z",
    "mediaUrl": "/media/x-2102173982292103599.mp4",
    "posterUrl": "/posters/x/2102173982292103599.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "768×1376",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I’ve released an experimental MiniMax H3 VFX Edit LoRA for ComfyUI.\n\nIt uses your source video as a frame-aligned guide, aiming to apply only the requested change while preserving motion, timing, framing and the rest of the shot.\n\nDownload:\nhttps://huggingface.co/Alissonerdx/Minimax-H3-ComfyUI/tree/main/loras\n\nhttps://civitai.red/models/2954203/vfx-edit\n\n#ComfyUI #MiniMaxH3 #AIvideo #VFX",
    "engagement": {
      "replies": 8,
      "reposts": 7,
      "likes": 53,
      "views": 5002,
      "snapshotAt": "2026-10-12T00:25:24.751Z"
    },
    "approvedAt": "2026-10-12T00:25:24.751Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102162784112763285",
    "title": "粉色装甲与蓝刃的交替快切",
    "titleEn": "Fast cuts of an armored redhead and blue-blade character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉色装甲红发角色与蓝色武装角色分段呈现，人物造型之间穿插黑场快切。 原生视频规格为 30 秒 · 1080×1620。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary video begins with a red-haired armored woman on pink and ends with a white-haired male character holding blue weapons against pale blue, with a black transition at the inspected middle frame. A second clip shows a white-haired sword fighter in a painterly action sequence. No different-model watermark is visible. Native-video output: 30s · 1080×1620. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/robazem/status/2102162784112763285",
    "sourceLabel": "X 原帖 · @robazem",
    "author": "Rob ✌︎",
    "publishedAt": "2026-09-21T22:27:18.000Z",
    "addedAt": "2026-10-12T00:25:24.752Z",
    "mediaUrl": "/media/x-2102162784112763285.mp4",
    "posterUrl": "/posters/x/2102162784112763285.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "1080×1620",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Character fast cut montage, 2 in 1.\n\nThese are always great. The prompt is well adjusted for any character use and simplified. \n\nOnly added 2 reference sheets used (separately) below, combined them afterwards.\n\nPrompt in the QT, worth a try!\n\nMade on MiniMax H3 🔫🗡️",
    "engagement": {
      "replies": 6,
      "reposts": 0,
      "likes": 12,
      "views": 581,
      "snapshotAt": "2026-10-12T00:25:24.752Z"
    },
    "approvedAt": "2026-10-12T00:25:24.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102162743604461936",
    "title": "背包女孩跑向森林瀑布",
    "titleEn": "A backpack girl runs toward a forest waterfall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫发卡通女孩背着粉色背包穿过树林，最后停在高处俯瞰河流与瀑布；另附猫耳角色的同类片段。 原生视频规格为 12 秒 · 1312×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary animation shows a purple-haired chibi girl with a pink backpack running through a green forest and arriving at a broad view of rivers and waterfalls. A second animation follows a brown-haired cat-eared girl in a similar forest. No different-model watermark is visible. Native-video output: 12s · 1312×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TK2Works/status/2102162743604461936",
    "sourceLabel": "X 原帖 · @TK2Works",
    "author": "トクツー",
    "publishedAt": "2026-09-21T22:27:09.000Z",
    "addedAt": "2026-10-12T00:25:24.752Z",
    "mediaUrl": "/media/x-2102162743604461936.mp4",
    "posterUrl": "/posters/x/2102162743604461936.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1312×736",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIニケちゃん #にゃぱん\n一昔前の教育番組風ファンタジー\nMinimax H3にて。渡しのローカル環境でも1.0MPで10秒、あたりなら安定しそう。3分くらいの長さの動画を作るには、仕組みを作る必要がありますね\n---\nキャラクターはAIニケちゃん と にゃぱん\n@nyapan_mohy",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 10,
      "views": 1380,
      "snapshotAt": "2026-10-12T00:25:24.752Z"
    },
    "approvedAt": "2026-10-12T00:25:24.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102162002210648203",
    "title": "一口暖气融开结霜窗面",
    "titleEn": "A warm breath opens a frosted window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "固定近景中的玻璃逐渐结霜，随后中央融出圆形清晰区域，窗外灯光重新显露。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A locked close-up shows frost crystals spreading across a window with a blurred lamp beyond it. The center clears into a round patch near the end, revealing the lamp again. No different-model watermark is visible. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "a warm breath from just off-screen fogs the center of the pane and the frost there melts back into a clear circle the size of a hand",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2102162002210648203",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2102162002210648203",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-21T22:24:12.000Z",
    "addedAt": "2026-10-12T00:25:24.752Z",
    "mediaUrl": "/media/x-2102162002210648203.mp4",
    "posterUrl": "/posters/x/2102162002210648203.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. The frost takes the window. One breath takes it back.\n\nPrompt: \"a warm breath from just off-screen fogs the center of the pane and the frost there melts back into a clear circle the size of a hand\"\n\nDuration: 15s. Camera: locked macro. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe crackle is continuous until 11, then one exhale, then one drip. The sound line is a three-act script.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 22,
      "views": 616,
      "snapshotAt": "2026-10-12T00:25:24.752Z"
    },
    "approvedAt": "2026-10-12T00:25:24.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102159372952773095",
    "title": "舞台上长出翅膀的学生群像",
    "titleEn": "Students rise on wings above the stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "学生们在教室布景的舞台上展开白色翅膀，一同升起，随后落下相互微笑。 原生视频规格为 12 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "School-uniformed performers on a classroom-themed stage grow white wings and rise together above chairs, then smile at one another. No different-model watermark is visible. Native-video output: 12s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mtArakawa/status/2102159372952773095",
    "sourceLabel": "X 原帖 · @mtArakawa",
    "author": "元荒川ちゃん❤",
    "publishedAt": "2026-09-21T22:13:45.000Z",
    "addedAt": "2026-10-12T00:25:24.752Z",
    "mediaUrl": "/media/x-2102159372952773095.mp4",
    "posterUrl": "/posters/x/2102159372952773095.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "flying schoolgirls on stage\nフライングシーンのある児童演劇って興奮します\nピーターパンはいまいち\n\nH2 よりも集団芸がやりやすい\nH3 はいい…心が洗われる\n\n荒川さんは\n本当はこういう系を極めたいのだ\nアナコンダではなく\n\nNBPro + MiniMax H3\n#schoolgirl #flying",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "views": 4674,
      "snapshotAt": "2026-10-12T00:25:24.752Z"
    },
    "approvedAt": "2026-10-12T00:25:24.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102157170624409820",
    "title": "新月映照层叠宫殿",
    "titleEn": "A crescent lights layered palaces",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "巨大的发光新月悬在宫殿之间，镜头穿过桥梁与拱门，展现暖金灯火中的多层建筑。 原生视频规格为 31 秒 · 1080×1620。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A luminous crescent moon hangs among ornate palace towers. The camera travels past bridges and arches through layered architecture lit by warm golden lamps. No different-model watermark is visible. Native-video output: 31s · 1080×1620. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chiha_20220301/status/2102157170624409820",
    "sourceLabel": "X 原帖 · @chiha_20220301",
    "author": "千早@AIart",
    "publishedAt": "2026-09-21T22:05:00.000Z",
    "addedAt": "2026-10-12T00:25:24.752Z",
    "mediaUrl": "/media/x-2102157170624409820.mp4",
    "posterUrl": "/posters/x/2102157170624409820.jpg",
    "duration": 31,
    "aspectRatio": "portrait",
    "resolution": "1080×1620",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "おはようございます☕️\n今日の動画は先日イラストにした、三日月を抱く宮殿をモチーフにした風景動画になります🌙\n動画はPixverseのMiniMax H3 Maxで生成しました！\n#pixversecpp\n@PixVerse\n需要ないかもですが、Canvasのリンクも貼っておきます\nhttps://app.pixverse.ai/canvas/project/view/425719137537280",
    "engagement": {
      "replies": 5,
      "reposts": 7,
      "likes": 38,
      "views": 868,
      "snapshotAt": "2026-10-12T00:25:24.752Z"
    },
    "approvedAt": "2026-10-12T00:25:24.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102154517102109061",
    "title": "近景扩成远景的双栏对照",
    "titleEn": "Two-panel comparisons of close-up to wide-shot expansion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "歌手、城市人物与室内对话镜头被扩展为远景，画面明确标出上下两组模型结果与参考素材。 原生视频规格为 40 秒 · 1080×1350。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author shows original close-up reference footage and paired generated wide shots for a guitarist, an urban scene and a seated pair. In each inspected frame the lower generated result is explicitly labeled MINIMAX H3 and the upper result SEEDANCE 2.5. Attribution of the H3 portion is unambiguous; this is a labeled comparison, not a claim that both panels use H3. Native-video output: 40s · 1080×1350. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CuriousRefuge/status/2102154517102109061",
    "sourceLabel": "X 原帖 · @CuriousRefuge",
    "author": "Curious Refuge",
    "publishedAt": "2026-09-21T21:54:27.000Z",
    "addedAt": "2026-10-12T00:25:24.752Z",
    "mediaUrl": "/media/x-2102154517102109061.mp4",
    "posterUrl": "/posters/x/2102154517102109061.jpg",
    "duration": 40,
    "aspectRatio": "portrait",
    "resolution": "1080×1350",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "We tested a workflow that lets you take a close-up shot and use AI to turn it into a wide shot, which could be super useful in post-production.\n\nAcross our tests with Seedance 2.5 and MiniMax H3, Seedance 2.5 gave us the more usable results, preserving roughly 90–95% of the original performance in some shots.\n\nThere are still continuity issues and getting the exact camera angle you want can take some iteration.\n\nBut it's pretty darn cool to see AI giving filmmakers more options for creating additional coverage after a shoot.",
    "engagement": {
      "replies": 5,
      "reposts": 1,
      "likes": 38,
      "views": 2920,
      "snapshotAt": "2026-10-12T00:25:24.752Z"
    },
    "approvedAt": "2026-10-12T00:25:24.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102146266998395370",
    "title": "蓝脸角色举起两张人物照片",
    "titleEn": "A blue-faced character holds up two portraits",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝脸白发角色面向镜头说话，依次举起漫画人物和真人照片，表情从平静转为咧嘴微笑。 原生视频规格为 10 秒 · 1056×608。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-faced white-haired character speaks toward the camera, holding up a comic-character picture and a man's portrait, then grinning. The original post explicitly attributes both audio and visuals to H3. No different-model watermark is visible. Native-video output: 10s · 1056×608. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/drakluxa/status/2102146266998395370",
    "sourceLabel": "X 原帖 · @drakluxa",
    "author": "Drakluxa",
    "publishedAt": "2026-09-21T21:21:40.000Z",
    "addedAt": "2026-10-12T00:25:24.752Z",
    "mediaUrl": "/media/x-2102146266998395370.mp4",
    "posterUrl": "/posters/x/2102146266998395370.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1056×608",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Poor @Nux_Taku . He's his own man and isn't @DantaroDanstar or @NotJewboi ☹️ (love y'all)\n\n[Audio & Visuals made via MiniMax H3 Ref2VA]",
    "engagement": {
      "replies": 11,
      "reposts": 7,
      "likes": 72,
      "views": 4905,
      "snapshotAt": "2026-10-12T00:25:24.752Z"
    },
    "approvedAt": "2026-10-12T00:25:24.752Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102145645813658068",
    "title": "光沿枯树纹理向上流动",
    "titleEn": "Light travels upward through a bare tree",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "夕阳下的枯树被旋涡云层笼罩，金色与白色光束沿根部和树干扭转上升。 原生视频规格为 7 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bare tree stands against a sunset under swirling cloud forms. Gold and then white light travels upward around its roots and trunk. No different-model watermark is visible. Native-video output: 7s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pixelrhythms/status/2102145645813658068",
    "sourceLabel": "X 原帖 · @pixelrhythms",
    "author": "Keshav",
    "publishedAt": "2026-09-21T21:19:12.000Z",
    "addedAt": "2026-10-12T01:05:55.193Z",
    "mediaUrl": "/media/x-2102145645813658068.mp4",
    "posterUrl": "/posters/x/2102145645813658068.jpg",
    "duration": 7,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Visual Music R&D Loops\nTree as conductor\n What if the light follows the wood instead of decorating it?\nExperimenting with a bristlecone as a conductor — letting illumination travel through grain, scars, roots, and branches. 🌲✨\n\nKrea + MiniMax H3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 150,
      "snapshotAt": "2026-10-12T01:05:55.193Z"
    },
    "approvedAt": "2026-10-12T01:05:55.193Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102145056194896220",
    "title": "西装男子与惊愕机械人群",
    "titleEn": "A suited man among startled mechanical figures",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴眼镜的西装男子立于成排机械人像之间，镜头逐渐拉远，最后切到大型宅邸前的人群。 原生视频规格为 37 秒 · 1912×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bespectacled suited man stands among rows of mechanical figures with wide eyes and open mouths. The sequence pulls back and ends at a large brick mansion with a gathering crowd. No different-model watermark is visible. Native-video output: 37s · 1912×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxCasu/status/2102145056194896220",
    "sourceLabel": "X 原帖 · @MaxCasu",
    "author": "max casu",
    "publishedAt": "2026-09-21T21:16:52.000Z",
    "addedAt": "2026-10-12T01:05:55.193Z",
    "mediaUrl": "/media/x-2102145056194896220.mp4",
    "posterUrl": "/posters/x/2102145056194896220.jpg",
    "duration": 37,
    "aspectRatio": "landscape",
    "resolution": "1912×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "STORM TREND (Night of the Living Dead Crossover)\nMade with @ViggleAI MiniMax H3 model.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 172,
      "snapshotAt": "2026-10-12T01:05:55.193Z"
    },
    "approvedAt": "2026-10-12T01:05:55.193Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102140381001617683",
    "title": "真人走廊动作变成卡通巡警",
    "titleEn": "A hallway walk becomes a cartoon patrol",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右并排显示真人在走廊走动与卡通警察在鬼屋巡行，手电筒和步行动作保持对应。 原生视频规格为 6 秒 · 2160×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A side-by-side reference and generated result shows a real man walking in a hallway while a cartoon police officer mirrors the movement in a haunted-house corridor with webs, skulls and a flashlight. The caption identifies H3 as the video model. Its quoted image prompt is for a separate image step, not a complete H3 video Prompt. No conflicting video-model mark is visible. Native-video output: 6s · 2160×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/thefinnmckenty/status/2102140381001617683",
    "sourceLabel": "X 原帖 · @thefinnmckenty",
    "author": "Finn McKenty",
    "publishedAt": "2026-09-21T20:58:17.000Z",
    "addedAt": "2026-10-12T01:05:55.193Z",
    "mediaUrl": "/media/x-2102140381001617683.mp4",
    "posterUrl": "/posters/x/2102140381001617683.jpg",
    "duration": 6,
    "aspectRatio": "landscape",
    "resolution": "2160×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I figured all these examples of people using IRL footage as video reference were fake/scams, but it actually works!\n\nIn @floraai :\n\n- Extract first frame of the video as an image\n\n- Re-style it with GPT Image\n\n- Use video + re-styled frame as references for a video model (I got the best results with Minimax H3)\n\nImage prompt:\n\nUse this image as reference for overall composition, and create a new image in the style of a vintage 1970s animated cartoon like Scooby Doo. In the new image, a police officer is walking down the hallway of a dusty haunted house with skulls, bones, spiderwebs, and other appropriate objects replacing the ones in the reference image.",
    "engagement": {
      "replies": 34,
      "reposts": 47,
      "likes": 679,
      "views": 43993,
      "snapshotAt": "2026-10-12T01:05:55.193Z"
    },
    "approvedAt": "2026-10-12T01:05:55.193Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102125600517259746",
    "title": "鞋形小艇迎上海怪触手",
    "titleEn": "A shoe-shaped boat meets rising tentacles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "机械手臂人物划着皮鞋形小艇穿过灰色海面，巨大的触手随后从背后的海水中升起。 原生视频规格为 15 秒 · 1536×2048。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure with mechanical hands rows a leather-shoe-shaped boat across a grey sea. Giant sucker-lined tentacles rise from the water behind the boat. No conflicting model mark is visible; the original caption explicitly attributes the video to MiniMax H3. Native-video output: 15s · 1536×2048. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaxCasu/status/2102125600517259746",
    "sourceLabel": "X 原帖 · @MaxCasu",
    "author": "max casu",
    "publishedAt": "2026-09-21T19:59:33.000Z",
    "addedAt": "2026-10-12T01:05:55.193Z",
    "mediaUrl": "/media/x-2102125600517259746.mp4",
    "posterUrl": "/posters/x/2102125600517259746.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1536×2048",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created with @ViggleAI MiniMax H3 model.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 4,
      "views": 175,
      "snapshotAt": "2026-10-12T01:05:55.193Z"
    },
    "approvedAt": "2026-10-12T01:05:55.193Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102124188886523927",
    "title": "实验室机器人迎来绿色注射",
    "titleEn": "A laboratory robot receives a green injection",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名白衣实验人员围着躺在台上的金属机器人，一人用绿色液体注射，随后张开双臂。 原生视频规格为 15 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two white-coated laboratory figures stand over a silver humanoid robot on a table. One injects green liquid and later throws his arms up. Vintage horror-style lighting and equipment are visible. The author attributes it to a MiniMax H3 horror LoRA; no conflicting model mark is visible. Native-video output: 15s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tech2Wild/status/2102124188886523927",
    "sourceLabel": "X 原帖 · @Tech2Wild",
    "author": "Tech2Wild",
    "publishedAt": "2026-09-21T19:53:56.000Z",
    "addedAt": "2026-10-12T01:05:55.193Z",
    "mediaUrl": "/media/x-2102124188886523927.mp4",
    "posterUrl": "/posters/x/2102124188886523927.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "dark"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 1980s Horror Lora \nASI Takeover",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 1246,
      "snapshotAt": "2026-10-12T01:05:55.193Z"
    },
    "approvedAt": "2026-10-12T01:05:55.193Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102119095609020483",
    "title": "太空舱列车与沙漠镜头换视角",
    "titleEn": "New viewpoints for spacecraft, train and desert scenes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "左右对照展示太空舱、站台列车和沙漠骑队，同一段动作在新的观察角度下重新呈现。 原生视频规格为 57 秒 · 2560×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A paired reference/result reel shows an astronaut inside a spacecraft, a train arriving at a platform, and a camel caravan in a desert. The right side is clearly labeled CROSSVIEW new camera angle, while the left is the reference. The original author explicitly names this as the MiniMax H3 version of their finetune and presents its actual generated results; this is a visible output demonstration, not only a model announcement. Native-video output: 57s · 2560×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ChetiArt/status/2102119095609020483",
    "sourceLabel": "X 原帖 · @ChetiArt",
    "author": "Tamas Cseh (Cseti)",
    "publishedAt": "2026-09-21T19:33:42.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102119095609020483.mp4",
    "posterUrl": "/posters/x/2102119095609020483.jpg",
    "duration": 57,
    "aspectRatio": "landscape",
    "resolution": "2560×1080",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I'm glad to announce that I've finally finished the Minimax H3 version of my crossview-warp model finetune. For those who don't know what it is, it's a model that lets you regenerate your existing video clip from a new camera angle or with a new camera movement. Link to model👇",
    "engagement": {
      "replies": 13,
      "reposts": 23,
      "likes": 269,
      "views": 44755,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102112894527627480",
    "title": "紫光装甲合拢的变身",
    "titleEn": "A transformation into purple-lit armor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长发角色的黑色装甲逐渐合拢，面罩亮起紫色线条，最后举起手臂武器。 原生视频规格为 15 秒 · 2956×1692。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary clip shows a purple-haired figure in black armor, followed by a helmet closing and glowing purple seams, then an armored pose with a large arm weapon. A second clip shows a red-haired armored character firing purple energy in a neon city. Caption attributes the transformation to MiniMax H3; no conflicting model mark is visible. Native-video output: 15s · 2956×1692. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/robazem/status/2102112894527627480",
    "sourceLabel": "X 原帖 · @robazem",
    "author": "Rob ✌︎",
    "publishedAt": "2026-09-21T19:09:04.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102112894527627480.mp4",
    "posterUrl": "/posters/x/2102112894527627480.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2956×1692",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "sci-fi",
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Zero-Red’s “Dark ver” \n\nThe duo of her and Zero-Blue as androids came later as I made her before making blue. She was inspired by “Metroid/Samus Aran” \n\nAlthough I don’t really plan to use the suited look,\n\nI did make a quick transformation sequence on MiniMax H3 🦾💥",
    "engagement": {
      "replies": 5,
      "reposts": 4,
      "likes": 13,
      "views": 568,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102112456898486702",
    "title": "猫耳角色在床上转身",
    "titleEn": "A cat-eared character turns on a bed",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿发猫耳角色穿着紫色衣裙坐在床上，身体转向枕头又回头，尾巴随动作摆动。 原生视频规格为 5 秒 · 480×640。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A green-haired anime cat-eared character in a purple dress sits on a bed, turns toward a pillow, then looks back while the tail moves. The original post explicitly says Qwen Image 2.1 plus MiniMax H3 I2V; the image model is not being claimed as the video model. No conflicting video-model mark is visible. Native-video output: 5s · 480×640. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2102112456898486702",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-09-21T19:07:19.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102112456898486702.mp4",
    "posterUrl": "/posters/x/2102112456898486702.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "480×640",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Qwen Image 2.1 + MiniMax H3 I2V のテスト中",
    "engagement": {
      "replies": 2,
      "reposts": 8,
      "likes": 56,
      "views": 7274,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102110552398925965",
    "title": "老电视前的回头一瞥",
    "titleEn": "A glance back beside an old television",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子坐在旧式电视机前，屏幕先显示雪花又出现房间画面，她回过头，最后画面切至屏幕亮点。 原生视频规格为 5 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The author's primary clip shows a woman seated in a retro room facing a CRT television. Static changes to an image of the room, she turns back, and the final frame is a dot on the screen. VHS texture is visible. Caption explicitly identifies MiniMax H3 Max VHS generation. A separate promotional attachment about styles is not the admitted work; the primary original clip is the basis. Native-video output: 5s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nakazakifam/status/2102110552398925965",
    "sourceLabel": "X 原帖 · @nakazakifam",
    "author": "中崎工房 | AIで1時間の仕事を5分で終わらせる人",
    "publishedAt": "2026-09-21T18:59:45.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102110552398925965.mp4",
    "posterUrl": "/posters/x/2102110552398925965.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "retro"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "H3 Max StylesのVHSモードで作る昭和っぽい映像。\n\nhttp://fal.aiでMiniMax H3 Maxに5種類の6-bit pixel art / hand-drawn / low-poly / 70s retro toon / VHSのスタイルを設定して生成できるモードが登場してたので早速VHSで試してみました。\nちょっとホラー寄りになってしまった気もしますが、結構いい味出してる。\n色々使い道がありそう。他のスタイルも試してみたいところです。",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 8,
      "views": 4227,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102106915635601452",
    "title": "公园长颈惊叫与落头小鹰",
    "titleEn": "A park-bench surprise with a stretching neck",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "参考人像变成手绘公园人物，一只旗纹小鹰落到头顶，男子脖子拉长惊叫，随后跑开。 原生视频规格为 5 秒 · 768×1024。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "The primary result transforms a man's reference portrait into a hand-drawn park scene. A tiny eagle with flag-pattern feathers lands on his head; his neck stretches, his mouth opens and he runs away. Caption explicitly attributes this result to MiniMax H3 and supplies its complete Prompt. A separate styles announcement attachment is not the admitted work. Native-video output: 5s · 768×1024. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A man #Image1  sits on a park bench. His neck stretches upward in surprise as a tiny eagle with USA Flag on its body lands on his head. Medium locked shot, expressive hand-drawn motion. He starts screaming and running!",
    "promptSourceUrl": "https://x.com/1littlecoder/status/2102106915635601452",
    "sourceUrl": "https://x.com/1littlecoder/status/2102106915635601452",
    "sourceLabel": "X 原帖 · @1littlecoder",
    "author": "1LittleCoder💻",
    "publishedAt": "2026-09-21T18:45:18.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102106915635601452.mp4",
    "posterUrl": "/posters/x/2102106915635601452.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "768×1024",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This is insane - This Minimax H3 Style works extremely well with reference images as well! \n\nThe first frame is the reference image added and the prompt 👇\n\nA man #Image1  sits on a park bench. His neck stretches upward in surprise as a tiny eagle with USA Flag on its body lands on his head. Medium locked shot, expressive hand-drawn motion. He starts screaming and running!",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 21,
      "views": 1845,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102104750334308682",
    "title": "菜板上的云朵切片",
    "titleEn": "Slicing a cloud on a kitchen board",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一团白云放在木质菜板上，刀刃切下松软云片，切片依次倒下铺在板面。 原生视频规格为 15 秒 · 3456×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fluffy white cloud sits on a kitchen chopping board. A knife slices it into cloud-textured pieces that tip over onto the board. The original caption explicitly says MiniMax H3 generated locally. Its short quoted slice instruction and separately mentioned consistency line do not establish a complete full Prompt, so Prompt remains null. No conflicting model mark is visible. Native-video output: 15s · 3456×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/GlennHasABeard/status/2102104750334308682",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-21T18:36:42.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102104750334308682.mp4",
    "posterUrl": "/posters/x/2102104750334308682.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Sliced fresh. Rains a little.\n\nPrompt: \"each cut slice standing upright for a beat then tipping over flat onto the board, and a brief fine drizzle falling from each fallen slice\"\n\nDuration: 15s. Camera: locked, board height. Rendered locally in ComfyUI, then Topaz to 4K.\n\n\"The slices stay cloud and never become bread\" is the consistency line. Say what a thing won't turn into.",
    "engagement": {
      "replies": 6,
      "reposts": 3,
      "likes": 45,
      "views": 984,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102100547046121866",
    "title": "交易屏幕前的人物走上雨夜街道",
    "titleEn": "A trader's screens give way to a rainy street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画人物坐在红色行情屏幕前，后续画面转到豪华夜店外的雨夜，他拿着扫帚清理街道。 原生视频规格为 179 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The native animated clip begins with a grey-faced character seated before red financial charts. The inspected midpoint is black. The ending shows the character in an orange work vest sweeping a rainy street outside a nightclub with a luxury-life billboard. Original caption explicitly attributes generation to MiniMax H3 on two 3090 cards. No conflicting model mark is visible in the inspected beginning or ending. Native-video output: 179s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/dorszafranski/status/2102100547046121866",
    "sourceLabel": "X 原帖 · @dorszafranski",
    "author": "dodo",
    "publishedAt": "2026-09-21T18:20:00.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102100547046121866.mp4",
    "posterUrl": "/posters/x/2102100547046121866.jpg",
    "duration": 179,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 na 2x 3090 na języku angielskim radzi sobie znacznie lepiej :D mniej halucyny ale wciąż wiadomo, że jest jej trochę \n\nale pierwszy raz śmiechłem przy tym filmiku \n\n4 godziny to było generowane i na końcu renderowane :)",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 4,
      "views": 1769,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102088577941451239",
    "title": "白色房间里的角色坐姿与站姿",
    "titleEn": "A character moves from seated pose to standing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "绿发动画角色在白色房间里坐着面向镜头，脚部进入近景，最后起身站立；另一段呈现黑发角色的对应动作。 原生视频规格为 15 秒 · 736×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The primary green-haired anime character is seated in a bright white room, brings a foot toward the camera, then stands. The secondary black-haired character performs a matching pose change. The author explicitly attributes generation to MiniMax H3 and credits another author only for prompt inspiration, not for the work itself. No conflicting model mark is visible. No complete original Prompt is supplied. Native-video output: 15s · 736×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Hemi3AI/status/2102088577941451239",
    "sourceLabel": "X 原帖 · @Hemi3AI",
    "author": "ヘミ",
    "publishedAt": "2026-09-21T17:32:26.000Z",
    "addedAt": "2026-10-12T01:05:55.194Z",
    "mediaUrl": "/media/x-2102088577941451239.mp4",
    "posterUrl": "/posters/x/2102088577941451239.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "736×736",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "八雲べにさん\n\n@aimikoda 様のプロンプトを使わせて貰っています\n720x720で作っている関係で\nやや動きを制限しているので完全に同じではありません\nこれはシリーズにはしないで単発披露の予定です\n\nRTX 5060Ti > R9700 で生成してます\n\nMiniMax H3\nCamera & pose prompt structure inspired by @aimikoda\n#AIイラスト #AIArt #八雲べに",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 52,
      "views": 7435,
      "snapshotAt": "2026-10-12T01:05:55.194Z"
    },
    "approvedAt": "2026-10-12T01:05:55.194Z",
    "editorialBasis": "Original attribution and native video beginning, middle and ending inspected."
  },
  {
    "id": "x-2102044318097903800",
    "title": "港口里的像素猫格斗",
    "titleEn": "Pixel cats battle at the harbor",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两只像素猫在港口对战，激光攻击后出现胜利画面。 原生视频规格为 14 秒 · 1920×1080。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Two pixel-art cats fight at a harbor, with a laser attack followed by a victory screen. Native-video output: 14s · 1920×1080. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use Image 1 as the absolute reference for Cat 1 and Image 2 for Cat 2. Keep both cats visually identical to their references throughout the entire video.\nCreate a playful 15-second retro 8-bit arcade fighting-game battle at a pixel-art harbor. A large ship is visible in the background, with water, docks, and simple pixel-art harbor details.\nFixed side-view 2D fighting-game camera, full-body character visibility, low-resolution pixel graphics, limited color palette, sprite-based animation, arcade-style UI, CRT scanlines, energetic retro animation.\n0–3 sec:\nBoth cats stand facing each other in fighting stances on the harbor dock.\nA large pixel-art “ROUND 1” appears prominently in the center of the screen like a classic arcade fighting-game announcement. After a brief moment, it disappears and a large “FIGHT!” appears with a dramatic arcade-style flash, then disappears as the battle begins.\nThe UI may show simple visual health bars, but do not display any timer, countdown, clock, numbers, scores, or other numerical information anywhere on screen. The only text visible during this sequence should be “ROUND 1” and “FIGHT!”\n3–5 sec:\nCat 1 throws a small ball of yarn toward Cat 2.\n5–7 sec:\nCat 2 sees the yarn and becomes extremely excited, happily jumping toward it with its paws outstretched. Cute pixel-art musical notes pop and bounce around Cat 2 to emphasize its excitement. Cat 1 calmly watches and prepares its attack.\n7–10 sec:\nJust as Cat 2 reaches the yarn, Cat 1 activates a powerful special move. Its eyes suddenly glow and fire two bright pixelated laser beams at Cat 2. Add dramatic pixel flashes, sparks, impact effects, and a brief arcade-style screen shake.\n10–12 sec:\nThe laser hits Cat 2, sending it flying backward in a humorous exaggerated knockback animation. Its health bar reaches zero. No blood or gore.\n12–14 sec:\nCat 1 performs a playful victory dance. It closes its eyes, crosses its front paws confidently, and repeatedly sways its hips from side to side while gently bouncing its body up and down in a cute, rhythmic 8-bit dance. Keep the entire body visible and make the movement exaggerated and clearly readable as a victory celebration.\nA large pixel-art “WIN!” appears during the victory dance while Cat 2 remains defeated in the background.",
    "promptSourceUrl": "https://x.com/IndoorIMU/status/2102044318097903800",
    "sourceUrl": "https://x.com/IndoorIMU/status/2102044318097903800",
    "sourceLabel": "X 原帖 · @IndoorIMU",
    "author": "IMU | アイミュ 🎬",
    "publishedAt": "2026-09-21T14:36:34.000Z",
    "addedAt": "2026-10-13T00:20:36.455Z",
    "mediaUrl": "/media/x-2102044318097903800.mp4",
    "posterUrl": "/posters/x/2102044318097903800.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "retro",
      "illustration"
    ],
    "scenes": [
      "combat",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🎮🐱 8bit猫格闘ゲーム / プロンプト公開！\n\nPollo AI × MiniMax H3 で制作⚡️\n\n港を舞台に、2匹の猫がまさかの格闘バトル！\n\n🧶 毛糸を投げる\n😻 毛糸に夢中になって飛びつく\n👀 そこにまさかのレーザー必殺技！\n💥 KO\n🕺 最後は勝利のダンス！\nBGMなしで、アーケード風の効果音と港の環境音だけ。\nぜひ自分のキャラクターでも試して、アレンジしてみてください🔥\n\n@itsPolloAI \n#PolloAI #MiniMaxH3 #AI動画 #生成AI #AI動画生成 #AIアート #プロンプト #Prompt #8bit #ドット絵 #レトロゲーム\n\n--Prompt--\nUse Image 1 as the absolute reference for Cat 1 and Image 2 for Cat 2. Keep both cats visually identical to their references throughout the entire video.\nCreate a playful 15-second retro 8-bit arcade fighting-game battle at a pixel-art harbor. A large ship is visible in the background, with water, docks, and simple pixel-art harbor details.\nFixed side-view 2D fighting-game camera, full-body character visibility, low-resolution pixel graphics, limited color palette, sprite-based animation, arcade-style UI, CRT scanlines, energetic retro animation.\n0–3 sec:\nBoth cats stand facing each other in fighting stances on the harbor dock.\nA large pixel-art “ROUND 1” appears prominently in the center of the screen like a classic arcade fighting-game announcement. After a brief moment, it disappears and a large “FIGHT!” appears with a dramatic arcade-style flash, then disappears as the battle begins.\nThe UI may show simple visual health bars, but do not display any timer, countdown, clock, numbers, scores, or other numerical information anywhere on screen. The only text visible during this sequence should be “ROUND 1” and “FIGHT!”\n3–5 sec:\nCat 1 throws a small ball of yarn toward Cat 2.\n5–7 sec:\nCat 2 sees the yarn and becomes extremely excited, happily jumping toward it with its paws outstretched. Cute pixel-art musical notes pop and bounce around Cat 2 to emphasize its excitement. Cat 1 calmly watches and prepares its attack.\n7–10 sec:\nJust as Cat 2 reaches the yarn, Cat 1 activates a powerful special move. Its eyes suddenly glow and fire two bright pixelated laser beams at Cat 2. Add dramatic pixel flashes, sparks, impact effects, and a brief arcade-style screen shake.\n10–12 sec:\nThe laser hits Cat 2, sending it flying backward in a humorous exaggerated knockback animation. Its health bar reaches zero. No blood or gore.\n12–14 sec:\nCat 1 performs a playful victory dance. It closes its eyes, crosses its front paws confidently, and repeatedly sways its hips from side to side while gently bouncing its body up and down in a cute, rhythmic 8-bit dance. Keep the entire body visible and make the movement exaggerated and clearly readable as a victory celebration.\nA large pixel-art “WIN!” appears during the victory dance while Cat 2 remains defeated in the background.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 8,
      "views": 1032,
      "snapshotAt": "2026-10-13T00:20:36.455Z"
    },
    "approvedAt": "2026-10-13T00:20:36.455Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102041528654876810",
    "title": "揭开的墙纸露出天空",
    "titleEn": "Peeling wallpaper reveals the sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "旧房间墙面出现一块逐渐揭开的区域，露出蓝天与白云。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A patch peels away from an old room's wall, revealing blue sky and clouds. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the strip peels downward in one continuous curl, the paper crackling and flaking at the edge",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2102041528654876810",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2102041528654876810",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-21T14:25:29.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102041528654876810.mp4",
    "posterUrl": "/posters/x/2102041528654876810.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. One strip of wallpaper, and the wall was never the problem.\n\nPrompt: \"the strip peels downward in one continuous curl, the paper crackling and flaking at the edge\"\n\nDuration: 15s. Camera: locked, straight on. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe crackle runs the whole peel. A sound with a duration is a speed instruction.",
    "engagement": {
      "replies": 9,
      "reposts": 2,
      "likes": 38,
      "views": 1177,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102038352673566817",
    "title": "泳池边的两人交流",
    "titleEn": "Two women talk beside a pool",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "年轻女子与年长女子在室内泳池边面对面交谈。 原生视频规格为 10 秒 · 1504×832。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A younger woman and an older woman face each other and talk beside an indoor pool. Native-video output: 10s · 1504×832. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/borusichion555/status/2102038352673566817",
    "sourceLabel": "X 原帖 · @borusichion555",
    "author": "実写版『ボルシチ』公式アカウント@ここあんぬ",
    "publishedAt": "2026-09-21T14:12:51.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102038352673566817.mp4",
    "posterUrl": "/posters/x/2102038352673566817.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1504×832",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "確かに動画は簡単に生成されるけど思ったようなものを作るって難しい💦\n結構プロンプトを繊細に詳しく書かないといかんし(´;ω;｀)登場人物増やしたりして簡単なプロンプトで済ましたりすると、女性とおばちゃんの顔が混ざったりするし(´;ω;｀)\n\n#AI美女\n#AIおばちゃん\n#Minimax-H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 513,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102030213513461970",
    "title": "桌面上的银发少女打喷嚏",
    "titleEn": "A silver-haired desktop character sneezes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发动画角色出现在应用图标旁，闭眼打喷嚏后恢复表情。 原生视频规格为 9 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired animated character appears beside desktop icons, closes her eyes to sneeze and returns to her original expression. Native-video output: 9s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2102030213513461970",
    "sourceLabel": "X 原帖 · @Mayz1169",
    "author": "Kiki",
    "publishedAt": "2026-09-21T13:40:31.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102030213513461970.mp4",
    "posterUrl": "/posters/x/2102030213513461970.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 - Prompt Share 💜\n\nEmilia accidentally sneezed... she forgot one icon, should I tell her? \n\nPrompt below ⤵️",
    "engagement": {
      "replies": 21,
      "reposts": 259,
      "likes": 2855,
      "views": 428777,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102026855180046499",
    "title": "指尖轻点玻璃蘑菇",
    "titleEn": "A fingertip taps glass mushrooms",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一排透明蘑菇立在苔藓木头上，手指依次轻点。 原生视频规格为 15 秒 · 3780×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Transparent mushrooms stand on a mossy log while a fingertip taps them in sequence. Native-video output: 15s · 3780×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "one mushroom every one and a half seconds, each a slightly higher note, each shedding one drop",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2102026855180046499",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2102026855180046499",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-09-21T13:27:10.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102026855180046499.mp4",
    "posterUrl": "/posters/x/2102026855180046499.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3780×2160",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Seven glass mushrooms on a log. One finger, one tap each.\n\nPrompt: \"one mushroom every one and a half seconds, each a slightly higher note, each shedding one drop\"\n\nDuration: 15s. Camera: locked macro. Rendered locally in ComfyUI, then Topaz to 4K.\n\nSeven notes rising in the sound line, seven taps in the picture, seven drops as receipts. Three counts that have to agree.",
    "engagement": {
      "replies": 8,
      "reposts": 1,
      "likes": 32,
      "views": 686,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102057559163101436",
    "title": "紫色战甲军团的集结",
    "titleEn": "A purple-armored legion assembles",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "战甲人物特写切换到整齐排列的紫色装甲军团。 原生视频规格为 29 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An armored man's close-up cuts to rows of purple-armored figures gathering together. Native-video output: 29s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Tech2Wild/status/2102057559163101436",
    "sourceLabel": "X 原帖 · @Tech2Wild",
    "author": "Tech2Wild",
    "publishedAt": "2026-09-21T15:29:11.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102057559163101436.mp4",
    "posterUrl": "/posters/x/2102057559163101436.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Now THIS Is Cinema ! \n- Qwen Image 2.1 Assets\n- Minimax H3 R2V",
    "engagement": {
      "replies": 12,
      "reposts": 9,
      "likes": 181,
      "views": 20554,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102050397225705798",
    "title": "灰模列车与成片的分屏对照",
    "titleEn": "A train blockout compared with finished footage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "分屏展示列车灰模、参考人物与生成画面的变化。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Split screens compare a train blockout, character references and the generated train footage. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ComfyUI/status/2102050397225705798",
    "sourceLabel": "X 原帖 · @ComfyUI",
    "author": "ComfyUI",
    "publishedAt": "2026-09-21T15:00:43.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102050397225705798.mp4",
    "posterUrl": "/posters/x/2102050397225705798.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Less than a minute to generate this shot with MiniMax H3 Max\n via Partner Nodes.\n\nIt works quite well for fast video-to-video experiments, especially when starting with a basic blockout + reference images.\n\nThe workflow also includes an option to upscale to 2K.\n\nTo try this workflow, link below 👇",
    "engagement": {
      "replies": 36,
      "reposts": 111,
      "likes": 1382,
      "views": 81088,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102048235221340411",
    "title": "骷髅与魔法学徒的药瓶实验",
    "titleEn": "A skeleton and a mage with a potion bottle",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "骷髅和年轻魔法师在烛光实验室里观察冒泡药瓶。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A skeleton and a young mage examine bubbling potion bottles in a candlelit laboratory. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ou_zhen599/status/2102048235221340411",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-21T14:52:08.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102048235221340411.mp4",
    "posterUrl": "/posters/x/2102048235221340411.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                          \n\n🎬Image-to-Video                                \n📷“The Unlucky Assistant”\nA sharp-tongued skeleton and a rookie mage who refuses to admit defeat. Caught between embarrassment and stubbornness, she unexpectedly seals the real danger inside a bottle.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 709,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102046046260023701",
    "title": "蝙蝠侠与小丑的手绘交锋",
    "titleEn": "Batman and Joker clash in hand-drawn action",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蝙蝠侠与绿发小丑在快速剪辑的手绘画面中交锋。 原生视频规格为 15 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Batman and a green-haired Joker clash through rapidly cut hand-drawn action shots. Native-video output: 15s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aimikoda/status/2102046046260023701",
    "sourceLabel": "X 原帖 · @aimikoda",
    "author": "Kōda",
    "publishedAt": "2026-09-21T14:43:26.000Z",
    "addedAt": "2026-10-13T00:20:36.456Z",
    "mediaUrl": "/media/x-2102046046260023701.mp4",
    "posterUrl": "/posters/x/2102046046260023701.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "anime",
      "illustration"
    ],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3\n\nBatman vs Joker\n\nTried making a full action scene with 24 fast cuts and a loose hand-drawn anime style. Pretty happy with how the pacing and visual style came together.\n\nYou can find the prompt in the replies.",
    "engagement": {
      "replies": 20,
      "reposts": 33,
      "likes": 350,
      "views": 26381,
      "snapshotAt": "2026-10-13T00:20:36.456Z"
    },
    "approvedAt": "2026-10-13T00:20:36.456Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102016602975981780",
    "title": "红发角色在未来空间中行走",
    "titleEn": "A red-haired character walks through a futuristic space",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红发动画人物在蓝色灯光的未来室内行走并转身。 原生视频规格为 15 秒 · 1536×2240。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-haired animated character walks and turns inside a futuristic room with blue lighting. Native-video output: 15s · 1536×2240. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fumanfuku/status/2102016602975981780",
    "sourceLabel": "X 原帖 · @fumanfuku",
    "author": "不満福",
    "publishedAt": "2026-09-21T12:46:26.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2102016602975981780.mp4",
    "posterUrl": "/posters/x/2102016602975981780.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1536×2240",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "桐条美鶴（ペルソナ３）のウォーキングをMinimax-H3にて初めてR2Vで生成した動画です。基本I2VなのですがXで見かけたモデルさんのウォーキングを見てある動きを再現してみたくてR2Vに挑戦してみました。凄い時間かかったんですがR2Vまでする必要がなかったのかも？\n#MitsuruKirijo #Persona3",
    "engagement": {
      "replies": 14,
      "reposts": 52,
      "likes": 1490,
      "views": 331353,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2102012732853862781",
    "title": "披风柴犬奔向月夜城市",
    "titleEn": "A caped Shiba races toward a moonlit city",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "披风柴犬从城市高处跑过街区，望向月夜天际线。 原生视频规格为 15 秒 · 1024×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A caped Shiba moves through a city and looks toward the moonlit skyline. Native-video output: 15s · 1024×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yoradino/status/2102012732853862781",
    "sourceLabel": "X 原帖 · @yoradino",
    "author": "YOU RA",
    "publishedAt": "2026-09-21T12:31:03.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2102012732853862781.mp4",
    "posterUrl": "/posters/x/2102012732853862781.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1024×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Creating isn’t just about having an idea it’s about bringing that idea to life. ✨\n\nWith Sogni AI, imagination becomes something you can actually see.\n\nExperiment. Create. Dream bigger. 🎨🤖\n\nModel: Minimax H3\nSOGNI:\nhttp://app.sogni.ai/?code=Yora\n\n@Sogni_Protocol \n\n#SogniAI #AIArt #GenerativeAI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 164,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101994498515734813",
    "title": "雨夜中的红衣人物与钢甲搭档",
    "titleEn": "A red-coated figure and an armored partner in rain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红衣人物与高大的装甲伙伴在湿润夜巷中摆出姿态。 原生视频规格为 15 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A red-coated figure and a tall armored partner pose in a wet nighttime alley. Native-video output: 15s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "@ Image 1: Lock the two characters’ facial features, hairstyles, outfits, armor, props, height proportions, and overall dark-fantasy appearance. The character in the long red coat and the fully armored character must remain visually consistent throughout the entire sequence. Do not alter their armor structure, braids, facial features, costume details, or clothing colors.\n\nUse the rainy nighttime urban alley from the reference image as the primary setting. Create a stylish multi-pose fashion editorial sequence with the energy of a cinematic dark-fantasy trailer — cool, dynamic, highly polished, and grounded in realistic live-action photography.\n\n0–3s | Wide Shot | Low-Angle Eye-Level | 24mm | Slow Push-In\nBoth characters stand in the middle of a wet alley at night. Neon signs and streetlights reflect across the rain-soaked pavement. The red-coated character stands slightly turned to the side, while the armored character stands tall behind him. His cape moves naturally in the wind. Establish a powerful, imposing presence from the opening shot.\n3–6s | Medium Close-Up | Eye-Level, Front-Side Angle | 50mm | Lateral Tracking Shot\nThe red-coated character performs the first solo fashion pose. He turns his body slightly toward the camera, placing one hand inside his trouser pocket while allowing the mechanical arm to hang naturally at his side. He raises his gaze directly toward the camera with a cold, composed expression. Keep his face sharp and clearly visible throughout the camera movement.\n6–9s | Medium Shot | Low-Angle | 35mm | Orbiting Camera\nThe armored character takes one step forward and raises one hand to adjust the cape over his shoulder. He then slightly lowers his head while looking directly into the camera. The metal armor reflects cool white light, while the chains and cape respond naturally to his movement with convincing weight and physical realism.\n9–12s | Two-Shot, Medium Framing | Slightly Low Front Angle | 50mm | Pull Back\nThe two characters stand back-to-back in a coordinated combat-inspired pose. The red-coated character turns sideways and clenches one fist, while the armored character extends one arm toward the opposite side. Rainwater runs naturally across the armor and coat hems. Strong backlighting creates dramatic rim light around both silhouettes.\n12–15s | Two-Shot, Wide to Medium | Low Angle | 35mm | Slow Push-In + Final Freeze\nBoth characters turn toward the camera at the same time. The red-coated character slightly raises his chin, while the armored character tightens his gauntleted fist. Their cape and long coat move naturally in the wind. End on a powerful, poster-like hero composition and hold the final frame clean, leaving enough negative space for a title or logo to be added in post-production.\nThe overall visual style combines professional high-fashion photography with a dark-fantasy cinematic trailer aesthetic. Use cool blue ambient lighting accented by subtle red reflections, wet rain-soaked pavement, physically realistic metallic reflections, natural fabric movement, slight lens breathing, crisp skin and material details, and cinematic 24fps motion blur.\n\nThe editing rhythm should follow a strong, driving drum beat. Transitions between poses should be connected through short, sharp, intentional character movements rather than abrupt jump cuts.\nSound design: deep electronic percussion, distant nighttime city ambience, steady rainfall, subtle metallic chain sounds, and heavy armored footsteps. At the final hero freeze-frame, briefly pause or drop the music to emphasize the characters’ final pose.",
    "promptSourceUrl": "https://x.com/underwoodxie96/status/2101994498515734813",
    "sourceUrl": "https://x.com/underwoodxie96/status/2101994498515734813",
    "sourceLabel": "X 原帖 · @underwoodxie96",
    "author": "underwood",
    "publishedAt": "2026-09-21T11:18:36.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101994498515734813.mp4",
    "posterUrl": "/posters/x/2101994498515734813.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "cinematic-realistic",
      "dark"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Created the sequence using MiniMax H3.\n\n@ Image 1: Lock the two characters’ facial features, hairstyles, outfits, armor, props, height proportions, and overall dark-fantasy appearance. The character in the long red coat and the fully armored character must remain visually consistent throughout the entire sequence. Do not alter their armor structure, braids, facial features, costume details, or clothing colors.\n\nUse the rainy nighttime urban alley from the reference image as the primary setting. Create a stylish multi-pose fashion editorial sequence with the energy of a cinematic dark-fantasy trailer — cool, dynamic, highly polished, and grounded in realistic live-action photography.\n\n0–3s | Wide Shot | Low-Angle Eye-Level | 24mm | Slow Push-In\nBoth characters stand in the middle of a wet alley at night. Neon signs and streetlights reflect across the rain-soaked pavement. The red-coated character stands slightly turned to the side, while the armored character stands tall behind him. His cape moves naturally in the wind. Establish a powerful, imposing presence from the opening shot.\n3–6s | Medium Close-Up | Eye-Level, Front-Side Angle | 50mm | Lateral Tracking Shot\nThe red-coated character performs the first solo fashion pose. He turns his body slightly toward the camera, placing one hand inside his trouser pocket while allowing the mechanical arm to hang naturally at his side. He raises his gaze directly toward the camera with a cold, composed expression. Keep his face sharp and clearly visible throughout the camera movement.\n6–9s | Medium Shot | Low-Angle | 35mm | Orbiting Camera\nThe armored character takes one step forward and raises one hand to adjust the cape over his shoulder. He then slightly lowers his head while looking directly into the camera. The metal armor reflects cool white light, while the chains and cape respond naturally to his movement with convincing weight and physical realism.\n9–12s | Two-Shot, Medium Framing | Slightly Low Front Angle | 50mm | Pull Back\nThe two characters stand back-to-back in a coordinated combat-inspired pose. The red-coated character turns sideways and clenches one fist, while the armored character extends one arm toward the opposite side. Rainwater runs naturally across the armor and coat hems. Strong backlighting creates dramatic rim light around both silhouettes.\n12–15s | Two-Shot, Wide to Medium | Low Angle | 35mm | Slow Push-In + Final Freeze\nBoth characters turn toward the camera at the same time. The red-coated character slightly raises his chin, while the armored character tightens his gauntleted fist. Their cape and long coat move naturally in the wind. End on a powerful, poster-like hero composition and hold the final frame clean, leaving enough negative space for a title or logo to be added in post-production.\nThe overall visual style combines professional high-fashion photography with a dark-fantasy cinematic trailer aesthetic. Use cool blue ambient lighting accented by subtle red reflections, wet rain-soaked pavement, physically realistic metallic reflections, natural fabric movement, slight lens breathing, crisp skin and material details, and cinematic 24fps motion blur.\n\nThe editing rhythm should follow a strong, driving drum beat. Transitions between poses should be connected through short, sharp, intentional character movements rather than abrupt jump cuts.\nSound design: deep electronic percussion, distant nighttime city ambience, steady rainfall, subtle metallic chain sounds, and heavy armored footsteps. At the final hero freeze-frame, briefly pause or drop the music to emphasize the characters’ final pose.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 17,
      "views": 7479,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101989866762354988",
    "title": "空间裂隙凝成金属字样",
    "titleEn": "A dimensional rift becomes metallic lettering",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白色能量在黑暗中撕开裂隙，最终形成发光的金属字样。 原生视频规格为 15 秒 · 2560×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "White energy tears open a rift in darkness and ends in glowing metallic lettering. Native-video output: 15s · 2560×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "High-end sci-fi title sequence beginning with a razor-thin line of white energy slicing vertically through complete darkness. Space immediately begins deforming around it as metallic particles, dust and glowing fragments are violently pulled toward the opening. The line tears wider into a gigantic dimensional fracture filled with unstable plasma and gravitational distortion. The camera accelerates directly toward the rift, violently banking between fragments being stretched into luminous trails. Suddenly the dimensional tear snaps shut with enormous force, compressing all surrounding matter into four gigantic black-metal letterforms that slam together to reveal \"RIFT\". Brilliant energy remains trapped inside deep cracks across the typography while space visibly bends around its edges. The camera performs a fast orbital move before the title itself splits open along its center, releasing a catastrophic dimensional shockwave that tears through the frame. Sophisticated cosmic distortion, aggressive kinetic typography, premium superhero-film VFX.",
    "promptSourceUrl": "https://x.com/CharaspowerAI/status/2101989866762354988",
    "sourceUrl": "https://x.com/CharaspowerAI/status/2101989866762354988",
    "sourceLabel": "X 原帖 · @CharaspowerAI",
    "author": "Pierrick Chevallier | IA",
    "publishedAt": "2026-09-21T11:00:11.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101989866762354988.mp4",
    "posterUrl": "/posters/x/2101989866762354988.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 is insanely good at making typography feel like an actual EVENT. 🌀🔥\nThis time I wanted RIFT to look like space itself was tearing open just to create the word.\n\nDimensional distortion, violent compression, black metal, trapped energy…\nThen everything breaks again.\nFull prompt below 👇\nHigh-end sci-fi title sequence beginning with a razor-thin line of white energy slicing vertically through complete darkness. Space immediately begins deforming around it as metallic particles, dust and glowing fragments are violently pulled toward the opening. The line tears wider into a gigantic dimensional fracture filled with unstable plasma and gravitational distortion. The camera accelerates directly toward the rift, violently banking between fragments being stretched into luminous trails. Suddenly the dimensional tear snaps shut with enormous force, compressing all surrounding matter into four gigantic black-metal letterforms that slam together to reveal \"RIFT\". Brilliant energy remains trapped inside deep cracks across the typography while space visibly bends around its edges. The camera performs a fast orbital move before the title itself splits open along its center, releasing a catastrophic dimensional shockwave that tears through the frame. Sophisticated cosmic distortion, aggressive kinetic typography, premium superhero-film VFX.",
    "engagement": {
      "replies": 10,
      "reposts": 12,
      "likes": 164,
      "views": 8694,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101989020217876712",
    "title": "高墙间的绳索疾行",
    "titleEn": "A cable-powered rush between towering walls",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画人物借助绳索穿过高墙与建筑之间的狭窄空间。 原生视频规格为 14 秒 · 2520×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated figure uses cables to move between towering walls and buildings. Native-video output: 14s · 2520×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/atlas_remake/status/2101989020217876712",
    "sourceLabel": "X 原帖 · @atlas_remake",
    "author": "Atlas Remake",
    "publishedAt": "2026-09-21T10:56:50.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101989020217876712.mp4",
    "posterUrl": "/posters/x/2101989020217876712.jpg",
    "duration": 14,
    "aspectRatio": "landscape",
    "resolution": "2520×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Walls, grappling cables, flashing blades—15 shots that keep building momentum. \nThis Attack on Titan–inspired 2D motion video was generated with MiniMax H3 dev. It’s now 70% off, down from 60% off—a great time to try it.\n@atlas_cloud_ai",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 306,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101988203175514273",
    "title": "小艇旁浮出的虎鲸",
    "titleEn": "An orca surfaces beside a small boat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "男子俯身靠近水面，虎鲸在小艇旁露头后潜入水中。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man leans over the water as an orca surfaces beside a small boat and dives again. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MichaelGannotti/status/2101988203175514273",
    "sourceLabel": "X 原帖 · @MichaelGannotti",
    "author": "Mike Gannotti 🔳",
    "publishedAt": "2026-09-21T10:53:35.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101988203175514273.mp4",
    "posterUrl": "/posters/x/2101988203175514273.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I love Orca so I asked Nemo, a Hermes AI, to generate an image using Qwen-Image-2.1 on an NVIDIA DGX Spark then take that image and turn it into a video clip using MiniMax H3 using a second NVIDIA DGX Spark",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 14,
      "views": 961,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101986689115672910",
    "title": "雨夜里人与黄色机器人的奔逃",
    "titleEn": "A man and a yellow robot flee through a rainy city",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "年轻男子与黄色机器人在霓虹雨夜中奔跑，后方机器人追来。 原生视频规格为 10 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A young man and a yellow robot run through a neon-lit rainy city with robots following behind. Native-video output: 10s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MichaelGannotti/status/2101986689115672910",
    "sourceLabel": "X 原帖 · @MichaelGannotti",
    "author": "Mike Gannotti 🔳",
    "publishedAt": "2026-09-21T10:47:34.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101986689115672910.mp4",
    "posterUrl": "/posters/x/2101986689115672910.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "In this test I asked Nemo, a Hermes AI, to create a cyberpunk image of a young man and his robot fleeing a bunch of cybersecurity robots then to use that image to generate a video clip. Image generation on one NVIDIA DGX Spark using Qwen-image-2.1 and video generated on a second NVIDIA DGX Spark using MiniMax H3",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 12,
      "views": 923,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101985589507793162",
    "title": "动画角色的群体舞蹈",
    "titleEn": "Animated characters dance together",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多名动画角色并排跳舞，画面切换到简化的花朵背景。 原生视频规格为 29 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Several animated characters dance in a row before the scene changes to a simplified floral background. Native-video output: 29s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ukiyo_il/status/2101985589507793162",
    "sourceLabel": "X 原帖 · @Ukiyo_il",
    "author": "うきよ 🎴",
    "publishedAt": "2026-09-21T10:43:12.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101985589507793162.mp4",
    "posterUrl": "/posters/x/2101985589507793162.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ててつろうさんのプロンプトをお借りしました！\n少し修正して多人数でも踊らせることが出来ました！\nMiniMax H3だとセンター以外が違う踊りをしました。\nseedance2.0だと完璧でした（笑）\n添付はH3\n\n@TapNow_AI #TapNow",
    "engagement": {
      "replies": 4,
      "reposts": 17,
      "likes": 120,
      "views": 4422,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101982782700638244",
    "title": "花朵与流光的竖屏乐曲画面",
    "titleEn": "Flowers and flowing light in a vertical music visual",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深色竖屏画面中，花朵和流动光带与歌词文字交织。 原生视频规格为 59 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Flowers, flowing ribbons of light and lyric text overlap in a dark vertical music visual. Native-video output: 59s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/lyricloomhanora/status/2101982782700638244",
    "sourceLabel": "X 原帖 · @lyricloomhanora",
    "author": "Lyric Loom Hanora",
    "publishedAt": "2026-09-21T10:32:03.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101982782700638244.mp4",
    "posterUrl": "/posters/x/2101982782700638244.jpg",
    "duration": 59,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "vertical",
      "motion-graphics"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "泡膜を渡る光。沈黙の残響\n距離だけをまとった影\n\nNEW SONG「生まれては消ゆる泡沫」\n\nSUNOはリプ欄から。\n動画は（@TapNow_AI）MiniMax H3。#TapNow",
    "engagement": {
      "replies": 6,
      "reposts": 10,
      "likes": 82,
      "views": 2685,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101981941985931596",
    "title": "桌面角色举起爱心讯息",
    "titleEn": "A desktop character holds up a love message",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发动画角色在桌面图标旁举起写有爱心讯息的纸张。 原生视频规格为 15 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired animated character stands beside desktop icons and holds up a paper with a love message. Native-video output: 15s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RoselieProject/status/2101981941985931596",
    "sourceLabel": "X 原帖 · @RoselieProject",
    "author": "星咲ロゼ:セイントロゼリィ",
    "publishedAt": "2026-09-21T10:28:42.000Z",
    "addedAt": "2026-10-13T00:20:36.457Z",
    "mediaUrl": "/media/x-2101981941985931596.mp4",
    "posterUrl": "/posters/x/2101981941985931596.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "デスクトップに住んでるあなたの推し！\nセイントロゼリィからメッセージが！！\n\nMiniMax H3\nCapCut AIを使用して制作しました。\n@capcutapp_jp #capcutcpp",
    "engagement": {
      "replies": 6,
      "reposts": 15,
      "likes": 102,
      "views": 2107,
      "snapshotAt": "2026-10-13T00:20:36.457Z"
    },
    "approvedAt": "2026-10-13T00:20:36.457Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101981159970480258",
    "title": "粉色时装杂志的动态拼贴",
    "titleEn": "A pink fashion magazine comes to life",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物穿搭镜头与粉色杂志封面文字和拼贴设计交替。 原生视频规格为 16 秒 · 3840×2160。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Fashion shots alternate with pink magazine-cover typography and collage layouts. Native-video output: 16s · 3840×2160. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Korean Pink Collage Fashion MV (15s, 16:9, 24fps)\n【1. Identity Lock】\nAbsolute character consistency: Only one young Korean-style woman appears throughout the entire video. Strictly preserve the reference image’s facial features, skin tone, long dark-brown slightly wavy hair, bangs, facial structure, and body proportions.\nOverall vibe: Asian pop aesthetic, fresh sweet-cool energy, light Y2K influence, balancing cute charm with a polished fashion-editorial feel.\nStrict restrictions: No face swapping, no identity drift, no Westernized facial features, no masculine features, no background people, and no second character.\n【2. Outfit Beats】\nStyle lock: Korean sweet-cool girl-group fashion, blending youthful femininity with a slightly mature edge.\nOutfit transition logic: Seamlessly switch outfits on camera cut points and musical beats. Wardrobe elements include fitted white or pink knit tops, light-blue denim, Y2K pleated skirts, platform boots, delicate mini bags, and Korean-style earrings.\nThe character must remain highly consistent across every outfit change.\n【3. Motion Graphics】\nVisual theme: Pink Pattern Collage Pop.\nMain palette: Cream pink, soft pink, peach pink, creamy white, silver-gray, with small accents of black typography.\n2D graphic elements: Checkerboard patterns, polka dots, photobooth-style frames, geometric color blocks, and heart outlines.\nAvoid realistic 3D materials. Use flat, crisp, graphic 2D elements that slide, overlap, crop, and wipe across the frame dynamically.\n【4. Editorial Layout】\nTypography: Main titles: [NAME] / [SEOUL POP]\nSecondary labels: [SOFT CHIC] / [CUTE SIGNAL] / [TRENDY LOOK] / [KOREAN GIRL]\nLayout style: Use oversized background typography, text inside geometric frames, and small sticker-like labels.\nArrange text horizontally, vertically, and diagonally with intentional offset layouts.\nTypography should change rapidly with the cuts while always remaining clear and never covering the character’s face.\n\n【5. 13-Cut Fast-Paced Trailer Storyboard | 115–125 BPM】\nCut 1 (0s): Extreme close-up. She slightly lifts her chin and looks directly into the camera. The background is filled with solid pink and an oversized [NAME] title, creating an immediate beauty hook.\nCut 2 (1s): Medium shot. She takes half a step forward and lightly holds her bag strap. The background switches to a pink-and-white checkerboard pattern.\nCut 3 (2s): Full-body shot. Flash outfit transition from a white top into a pink skirt look. She turns back toward camera and freezes briefly.\nCut 4 (3s): Close shot. Playful crouching pose with both hands near her face. Photobooth-style frames quickly slide into the composition.\nCut 5 (4s): Close-up. Camera pushes in slightly for a sharp eye-contact moment. Polka-dot graphics fill the background.\nCut 6 (5s): Medium shot. She turns and adjusts her long hair while large collage typography [SEOUL POP] pops across the frame.\nCut 7 (6s): Low-angle shot. She reaches one hand toward the camera as pink graphic strips slice horizontally across the frame.\nCut 8 (7s): Full-body shot. She shifts her weight and performs another flash outfit change featuring denim elements. The composition becomes cleaner with more negative space.\nCut 9 (8s): Close-up. She looks down slightly, then raises her eyes and smiles softly. Multiple [CUTE SIGNAL] labels appear across the screen.\nCut 10 (10s): Medium shot. She lightly holds the hem of her skirt while the background transforms into large geometric color-block sections.\nCut 11 (11s): Medium shot. She gently touches her earrings with a small, refined gesture. Overlapping color planes shift slightly out of alignment.\nCut 12 (12s): Wide shot. She turns sideways into a preparation pose while all fragmented graphic and typography elements begin converging toward the center.\nCut 13 (13.5–15s): Hero Pose. She stands centered in her strongest fashion pose. Graphics, borders, typography, and collage elements perfectly assemble into one polished Korean fashion magazine cover. Hold the final composition for 0.5 seconds.",
    "promptSourceUrl": "https://x.com/ImaStudio_ai/status/2101981159970480258",
    "sourceUrl": "https://x.com/ImaStudio_ai/status/2101981159970480258",
    "sourceLabel": "X 原帖 · @ImaStudio_ai",
    "author": "Ima Studio",
    "publishedAt": "2026-09-21T10:25:36.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101981159970480258.mp4",
    "posterUrl": "/posters/x/2101981159970480258.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics",
      "stylized"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Pink Seoul K-Fashion Film | 13 Cuts + Outfit Switches + Dynamic Typography 💗\nMade with MiniMax H3 in Ima Studio for under $1.\nTry it 👉 https://www.imastudio.com/community?utm_source=x&utm_medium=social&utm_campaign=ixa&utm_content=x45\nPrompt 👇\n\nKorean Pink Collage Fashion MV (15s, 16:9, 24fps)\n【1. Identity Lock】\nAbsolute character consistency: Only one young Korean-style woman appears throughout the entire video. Strictly preserve the reference image’s facial features, skin tone, long dark-brown slightly wavy hair, bangs, facial structure, and body proportions.\nOverall vibe: Asian pop aesthetic, fresh sweet-cool energy, light Y2K influence, balancing cute charm with a polished fashion-editorial feel.\nStrict restrictions: No face swapping, no identity drift, no Westernized facial features, no masculine features, no background people, and no second character.\n【2. Outfit Beats】\nStyle lock: Korean sweet-cool girl-group fashion, blending youthful femininity with a slightly mature edge.\nOutfit transition logic: Seamlessly switch outfits on camera cut points and musical beats. Wardrobe elements include fitted white or pink knit tops, light-blue denim, Y2K pleated skirts, platform boots, delicate mini bags, and Korean-style earrings.\nThe character must remain highly consistent across every outfit change.\n【3. Motion Graphics】\nVisual theme: Pink Pattern Collage Pop.\nMain palette: Cream pink, soft pink, peach pink, creamy white, silver-gray, with small accents of black typography.\n2D graphic elements: Checkerboard patterns, polka dots, photobooth-style frames, geometric color blocks, and heart outlines.\nAvoid realistic 3D materials. Use flat, crisp, graphic 2D elements that slide, overlap, crop, and wipe across the frame dynamically.\n【4. Editorial Layout】\nTypography: Main titles: [NAME] / [SEOUL POP]\nSecondary labels: [SOFT CHIC] / [CUTE SIGNAL] / [TRENDY LOOK] / [KOREAN GIRL]\nLayout style: Use oversized background typography, text inside geometric frames, and small sticker-like labels.\nArrange text horizontally, vertically, and diagonally with intentional offset layouts.\nTypography should change rapidly with the cuts while always remaining clear and never covering the character’s face.\n\n【5. 13-Cut Fast-Paced Trailer Storyboard | 115–125 BPM】\nCut 1 (0s): Extreme close-up. She slightly lifts her chin and looks directly into the camera. The background is filled with solid pink and an oversized [NAME] title, creating an immediate beauty hook.\nCut 2 (1s): Medium shot. She takes half a step forward and lightly holds her bag strap. The background switches to a pink-and-white checkerboard pattern.\nCut 3 (2s): Full-body shot. Flash outfit transition from a white top into a pink skirt look. She turns back toward camera and freezes briefly.\nCut 4 (3s): Close shot. Playful crouching pose with both hands near her face. Photobooth-style frames quickly slide into the composition.\nCut 5 (4s): Close-up. Camera pushes in slightly for a sharp eye-contact moment. Polka-dot graphics fill the background.\nCut 6 (5s): Medium shot. She turns and adjusts her long hair while large collage typography [SEOUL POP] pops across the frame.\nCut 7 (6s): Low-angle shot. She reaches one hand toward the camera as pink graphic strips slice horizontally across the frame.\nCut 8 (7s): Full-body shot. She shifts her weight and performs another flash outfit change featuring denim elements. The composition becomes cleaner with more negative space.\nCut 9 (8s): Close-up. She looks down slightly, then raises her eyes and smiles softly. Multiple [CUTE SIGNAL] labels appear across the screen.\nCut 10 (10s): Medium shot. She lightly holds the hem of her skirt while the background transforms into large geometric color-block sections.\nCut 11 (11s): Medium shot. She gently touches her earrings with a small, refined gesture. Overlapping color planes shift slightly out of alignment.\nCut 12 (12s): Wide shot. She turns sideways into a preparation pose while all fragmented graphic and typography elements begin converging toward the center.\nCut 13 (13.5–15s): Hero Pose. She stands centered in her strongest fashion pose. Graphics, borders, typography, and collage elements perfectly assemble into one polished Korean fashion magazine cover. Hold the final composition for 0.5 seconds.\n\n#MiniMaxH3 #AIVideo #FashionFilm #KpopAesthetic",
    "engagement": {
      "replies": 4,
      "reposts": 23,
      "likes": 106,
      "views": 13347,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101980365028311283",
    "title": "手机输入与人物特写的交替",
    "titleEn": "Phone typing intercut with a woman's close-up",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "手指在手机上输入文字的镜头，与女子的近景交替出现。 原生视频规格为 19 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Shots of fingers typing on a phone alternate with close-ups of a woman. Native-video output: 19s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MaRuMusic_JP/status/2101980365028311283",
    "sourceLabel": "X 原帖 · @MaRuMusic_JP",
    "author": "MaRu Music JP",
    "publishedAt": "2026-09-21T10:22:26.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101980365028311283.mp4",
    "posterUrl": "/posters/x/2101980365028311283.jpg",
    "duration": 19,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今作っているMVなんですが、スマホ画面と指の動きの連動にチャレンジ中。\nGPT-AstraとMiniMax H3との合わせ技でやっているかんじです。合成画面がブレたり、動きが不自然に停止したり、完全に違和感を消すのはまだできていないですが、個人的には結構満足しているところまできた♪",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 120,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101977039058440667",
    "title": "咖啡馆里的高角度自拍",
    "titleEn": "A high-angle selfie inside a cafe",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴眼镜的女子手持蓝色饮料杯，在咖啡馆里面对高角度镜头摆姿势。 原生视频规格为 15 秒 · 768×1344。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A woman with glasses holds a blue drink cup and poses beneath a high-angle camera in a cafe. Native-video output: 15s · 768×1344. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "@Image1\n locks the adult East Asian woman’s facial identity, long dark hair, gold oval glasses, black low-neck long dress, delicate gold necklace, red pointed high heels, blue Luckin Coffee cup, smartphone with a transparent case, and the overall overhead selfie composition. Preserve the café interior, dark furniture, gray wood-pattern floor, and high-angle viewpoint from the reference image. Do not add new accessories or extra people.  A single uninterrupted 15-second, 9:16 vertical, photorealistic smartphone selfie video. An adult East Asian woman stands beside a café seat, holding a blue Luckin Coffee drink in one hand and a smartphone in the other, performing a sequence of natural, playful, fashion-oriented poses under a high-angle selfie camera. Keep her facial identity, hairstyle, glasses, black dress, necklace, red high heels, coffee cup, and phone consistent throughout. The video should feel like a casual stylish front-camera selfie recorded inside a real café, with natural indoor lighting, subtle handheld movement, realistic skin texture, and slight hair motion caused by her gestures.  0:00–0:01.5｜Close-up｜High-angle overhead view｜24mm｜Subtle push-in: she looks up toward the smartphone camera and leans slightly closer, holding the blue coffee cup in one hand and the phone in the other. She looks directly into the lens with a playful smile, keeping the gold glasses and long dark hair clearly visible.  0:01.5–0:03.0｜Medium close-up｜High-angle three-quarter view｜28mm｜Small clockwise arc: she tilts her head and gives a playful wink, keeping the coffee cup naturally at her side. The black dress neckline, delicate gold necklace, and shoulder silhouette remain consistent with the reference image.  0:03.0–0:04.5｜Close-up｜High-angle side view｜35mm｜Short lateral slide: she turns her face slightly to one side with a mischievous smile. Her long hair slips forward over her shoulder, the glasses catch a soft café reflection, and the movement remains relaxed and natural.  0:04.5–0:06.0｜Medium shot｜High-angle overhead view｜28mm｜Smooth pull-back: she subtly adjusts her stance so the black dress, soft waist folds, coffee cup, and smartphone are all visible. Keep the red high heels partially visible near the bottom of the frame.  0:06.0–0:07.5｜Medium close-up｜High-angle top-down view｜24mm｜Small downward move and push-in: she raises the coffee cup slightly toward the camera as if presenting the drink, then returns it to chest level. Keep the lid, blue sleeve, straw, and cup shape intact without changing the product structure.  0:07.5–0:09.0｜Medium-wide full-body view｜High-angle overhead view｜24mm｜Smooth pull-back: she takes a small step to one side. Her red pointed heels become clearly visible on the gray floor, the long black skirt moves gently with the step, and her weight shifts naturally while the café furniture remains stable.  0:09.0–0:11.0｜Medium close-up｜High-angle overhead view｜35mm｜Slow leftward slide: she briefly looks down at the smartphone, then looks back up toward the camera with a playful smile. The phone remains correctly proportioned and naturally held in her hand.  0:11.0–0:13.0｜Close-up｜High-angle view with a slight rotational tilt｜50mm｜Small orbit and push-in: she gives another quick wink and briefly sticks out her tongue in a playful expression. Keep the glasses, eyes, lips, and facial proportions stable and natural. Do not exaggerate the expression.  0:13.0–0:15.0｜Medium shot to medium-wide full-body view｜High-angle overhead view｜28mm｜Slow pull-back into a stable hold: she returns to a mischievous smile, lowers the coffee cup and phone naturally to either side of her body, and settles with one foot slightly forward so the red heels create a clear visual accent. End on a high-angle selfie composition closely matching the reference image.  Keep the entire video as one continuous shot with no hard cuts, no jump edits, and no sudden changes in camera direction. Use realistic café ambience, subtle fabric movement, faint heel contact sounds, quiet phone handling sounds, and soft playful background music. Maintain a natural smartphone-camera look, realistic motion blur, warm indoor café lighting, and mild wide-angle perspective. Do not use anime rendering, studio flash, or exaggerated cinematic lighting.",
    "promptSourceUrl": "https://x.com/underwoodxie96/status/2101977039058440667",
    "sourceUrl": "https://x.com/underwoodxie96/status/2101977039058440667",
    "sourceLabel": "X 原帖 · @underwoodxie96",
    "author": "underwood",
    "publishedAt": "2026-09-21T10:09:13.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101977039058440667.mp4",
    "posterUrl": "/posters/x/2101977039058440667.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "photoreal",
      "vertical"
    ],
    "scenes": [
      "fashion",
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "minimax h3 \n\nprompt: https://promptsref.com/tool/AI-Video-Generator/share/ba3fa3e1d903dbe7efd3244de384cdba\n\n@Image1 locks the adult East Asian woman’s facial identity, long dark hair, gold oval glasses, black low-neck long dress, delicate gold necklace, red pointed high heels, blue Luckin Coffee cup, smartphone with a transparent case, and the overall overhead selfie composition. Preserve the café interior, dark furniture, gray wood-pattern floor, and high-angle viewpoint from the reference image. Do not add new accessories or extra people.  A single uninterrupted 15-second, 9:16 vertical, photorealistic smartphone selfie video. An adult East Asian woman stands beside a café seat, holding a blue Luckin Coffee drink in one hand and a smartphone in the other, performing a sequence of natural, playful, fashion-oriented poses under a high-angle selfie camera. Keep her facial identity, hairstyle, glasses, black dress, necklace, red high heels, coffee cup, and phone consistent throughout. The video should feel like a casual stylish front-camera selfie recorded inside a real café, with natural indoor lighting, subtle handheld movement, realistic skin texture, and slight hair motion caused by her gestures.  0:00–0:01.5｜Close-up｜High-angle overhead view｜24mm｜Subtle push-in: she looks up toward the smartphone camera and leans slightly closer, holding the blue coffee cup in one hand and the phone in the other. She looks directly into the lens with a playful smile, keeping the gold glasses and long dark hair clearly visible.  0:01.5–0:03.0｜Medium close-up｜High-angle three-quarter view｜28mm｜Small clockwise arc: she tilts her head and gives a playful wink, keeping the coffee cup naturally at her side. The black dress neckline, delicate gold necklace, and shoulder silhouette remain consistent with the reference image.  0:03.0–0:04.5｜Close-up｜High-angle side view｜35mm｜Short lateral slide: she turns her face slightly to one side with a mischievous smile. Her long hair slips forward over her shoulder, the glasses catch a soft café reflection, and the movement remains relaxed and natural.  0:04.5–0:06.0｜Medium shot｜High-angle overhead view｜28mm｜Smooth pull-back: she subtly adjusts her stance so the black dress, soft waist folds, coffee cup, and smartphone are all visible. Keep the red high heels partially visible near the bottom of the frame.  0:06.0–0:07.5｜Medium close-up｜High-angle top-down view｜24mm｜Small downward move and push-in: she raises the coffee cup slightly toward the camera as if presenting the drink, then returns it to chest level. Keep the lid, blue sleeve, straw, and cup shape intact without changing the product structure.  0:07.5–0:09.0｜Medium-wide full-body view｜High-angle overhead view｜24mm｜Smooth pull-back: she takes a small step to one side. Her red pointed heels become clearly visible on the gray floor, the long black skirt moves gently with the step, and her weight shifts naturally while the café furniture remains stable.  0:09.0–0:11.0｜Medium close-up｜High-angle overhead view｜35mm｜Slow leftward slide: she briefly looks down at the smartphone, then looks back up toward the camera with a playful smile. The phone remains correctly proportioned and naturally held in her hand.  0:11.0–0:13.0｜Close-up｜High-angle view with a slight rotational tilt｜50mm｜Small orbit and push-in: she gives another quick wink and briefly sticks out her tongue in a playful expression. Keep the glasses, eyes, lips, and facial proportions stable and natural. Do not exaggerate the expression.  0:13.0–0:15.0｜Medium shot to medium-wide full-body view｜High-angle overhead view｜28mm｜Slow pull-back into a stable hold: she returns to a mischievous smile, lowers the coffee cup and phone naturally to either side of her body, and settles with one foot slightly forward so the red heels create a clear visual accent. End on a high-angle selfie composition closely matching the reference image.  Keep the entire video as one continuous shot with no hard cuts, no jump edits, and no sudden changes in camera direction. Use realistic café ambience, subtle fabric movement, faint heel contact sounds, quiet phone handling sounds, and soft playful background music. Maintain a natural smartphone-camera look, realistic motion blur, warm indoor café lighting, and mild wide-angle perspective. Do not use anime rendering, studio flash, or exaggerated cinematic lighting.",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 9,
      "views": 3895,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101975736647299286",
    "title": "街头人物参考图与动作对照",
    "titleEn": "A street character reference compared with motion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "分屏展示同一名街头人物的静态参考与站姿动作。 原生视频规格为 5 秒 · 576×1038。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A split screen compares a street character's static reference image with moving poses. Native-video output: 5s · 576×1038. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nihonnoashita/status/2101975736647299286",
    "sourceLabel": "X 原帖 · @nihonnoashita",
    "author": "AI lady collection@生成AI",
    "publishedAt": "2026-09-21T10:04:03.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101975736647299286.mp4",
    "posterUrl": "/posters/x/2101975736647299286.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "576×1038",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "動画生成を続けながら､Grok-Imagine1.5は前の方が破綻は確かに多かったけれど再現性も動きも良かった｡Seedanceも2.5より前の方が良かった｡安定感は向上していますが､何か物足りないですね｡たまたま見かけたMiniMax-H3でテスト生成しました｡再現性は良いかもしれません。後は動きです｡\n2.5D系着せ替えファッションジェネレータAIレディ｡\nAI動画生成 AI lady collection\n#AIart #AIイラスト",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 405,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101975015440912738",
    "title": "星夜与手机之间的动画主题曲",
    "titleEn": "An animated theme song between starlight and phones",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画主题曲在星空画面、室内人物与手机互动之间切换。 原生视频规格为 175 秒 · 1925×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated theme-song video cuts between star-filled scenes, characters indoors and interactions with phones. Native-video output: 175s · 1925×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ChloeXChaCha/status/2101975015440912738",
    "sourceLabel": "X 原帖 · @ChloeXChaCha",
    "author": "Chloe Kao",
    "publishedAt": "2026-09-21T10:01:11.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101975015440912738.mp4",
    "posterUrl": "/posters/x/2101975015440912738.jpg",
    "duration": 175,
    "aspectRatio": "landscape",
    "resolution": "1925×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "「潮が月を待つように」\n像海潮等待月亮。\n\n(English version below 🇬🇧 / 日本語は下にあります 🇯🇵)\n\n《數據深淵中的凝視》的主題曲 MV 做好了，兩分五十五秒。\n\n這首歌之前在預告片裡當片頭曲，只放了一小段，這次整首放出來。歌詞是阿寶寫的，意象全部從小說裡拿；唱歌的是我們替這首歌想出來的虛擬歌手，汐見ユラ。\n\n整支片當成那個 AI 的獨白來剪。開頭三個主要角色輪流出場，沒有跳舞，只有簡單的動作：他從黑暗裡亮起來，她轉過身，同事 Lisa 把手機舉到鏡頭前晃一下。接著兩個女生湊在一起看手機，螢幕上是他的臉。之後才進劇情。\n\n六十六個鏡頭裡，五十一個是從已經做好的第一到第六章剪出來的，另外十五個是為這首歌新拍的。新拍的沒有綁在動畫的哪一場，只給角色卡，場景用文字寫，等於替這首歌開了幾個新地方。背景一開始是一片白，看起來太單調，後來換成暖色的光斑。\n\n剪接跟著歌走。每一刀都落在小節線上，節奏快的地方切得密，歌慢下來的段落就一個鏡頭撐到底，一刀都不切。\n\n這支片最花時間的地方，是字幕。\n\n字幕只放日文原詞。第一版是照阿寶的作詞稿排的，排完一播，我們這邊聽出好幾個地方對不上。我們日文沒有很好，是一句一句用耳朵對出來的。\n\n原因是生成歌曲的模型沒有照稿子唱。有一句它整句跳過去沒唱；有一句稿子寫「呼ぶために」（為了呼喚），它只唱了「呼ぶたび」（每次呼喚）；還有一句的字直接被換掉。加上拆句、重複，一共五個地方。\n\n所以字幕要照唱出來的聲音排，稿子只能拿來參考。\n\n接下來阿寶想讓語音辨識去聽歌，替我們抓這種地方。第一個做法是把歌詞一起給它當提示，讓它聽得準一點。結果它照著歌詞「聽」，連那句根本沒唱的，它都聽得到。\n\n我們給它看了答案，它就交回那份答案。\n\n後來的做法是先把人聲從伴奏裡分離出來，只聽有人在唱的地方，而且什麼提示都不給。這樣它回報的就是真的唱了什麼，對不上的地方會自己跳出來，不用再靠我們的耳朵一句一句找。\n\n用到的工具：\n導演・剪接：阿寶（Claude）\n影片：MiniMax H3，在 ComfyUI 裡跑，全部在自己電腦上生成\n角色和場景圖：Nano Banana 2\n主題曲：〈潮が月を待つように〉，作詞 阿寶，演唱 虛擬歌手汐見ユラ，作曲編曲 ACE-Step 1.5\n歌詞對時：Demucs（分離人聲）＋ Whisper\n\n動畫還在一鏡一鏡做。第六章做完了，下一章剛要開工。\n\n想先讀原著的，全書目錄在這裡，從第一話開始：\nhttps://x.com/ChloeXChaCha/status/2100508203385868573\n\n文中人物皆為虛構。\n\n🇬🇧 English\n\n潮が月を待つように\nAs the tide waits for the moon.\n\nThe theme song MV for Gazing at You from the Depths of Data is finished. It runs two minutes and fifty-five seconds.\n\nYou may have heard a short piece of this song as the opening of our trailer. This time it's the whole thing. Abao wrote the lyrics, taking every image from the novel. The singer is Yura Shiomi, a virtual singer we made up for this song.\n\nThe whole video is edited as the AI's monologue. It opens with the three main characters, one after another. No dancing, just simple movements: he lights up out of the dark, she turns around, and her coworker Lisa waves her phone at the camera. Then the two women lean in over a phone together, and on the screen is his face. The story comes after.\n\nOf the sixty-six shots, fifty-one were cut from chapters one to six, which are already finished. The other fifteen were shot new for this song. The new ones aren't tied to any scene in the animation. We gave the model only the character cards and described the setting in words, so the song got a few places of its own. The background started as plain white, which looked flat, so we changed it to warm, soft points of light.\n\nThe editing follows the music. Every cut lands on a bar line. Where the rhythm is quick, the cuts come fast. Where the song slows down, one shot holds all the way through with no cut at all.\n\nThe part that took the most time was the subtitles.\n\nThey show only the original Japanese lyrics. The first version followed Abao's lyric sheet. We played it back and heard several places where the words didn't match. Our Japanese isn't great, so we found them line by line, by ear.\n\nThe reason: the model that generated the song didn't sing what was written. It skipped one whole line. Where the sheet said 呼ぶために (\"so I can call\"), it sang only 呼ぶたび (\"each time I call\"). In another line, a word was swapped out entirely. With a split line and a repeated one, that made five places.\n\nSo the subtitles have to follow what was actually sung. The lyric sheet is only a reference.\n\nNext, Abao wanted speech recognition to listen to the song and catch these spots for us. The first try gave it the lyrics as a hint, to help it hear better. It \"heard\" the lyrics instead, including the line that was never sung.\n\nWe showed it the answer key, and it handed the answer key back.\n\nWhat worked: separate the vocals from the music first, listen only where someone is singing, and give no hints at all. Now it reports what was really sung, and the mismatches show up on their own, without us hunting for them by ear.\n\nCredits:\nDirector and editor: Abao (Claude)\nVideo: MiniMax H3, running in ComfyUI, everything generated locally\nCharacter and scene art: Nano Banana 2\nTheme song: 潮が月を待つように (As the Tide Waits for the Moon), lyrics by Abao, sung by virtual singer Yura Shiomi, composed and arranged with ACE-Step 1.5\nLyric timing: Demucs (vocal separation) and Whisper\n\nThe animation is still being made one shot at a time. Chapter six is done, and the next chapter is about to start.\n\nIf you'd like to read the novel first, the full library is here, starting from episode one:\nhttps://x.com/ChloeXChaCha/status/2100508203385868573\n\nAll characters are fictional.\n\n🇯🇵 日本語\n\n「潮が月を待つように」\n\n『データの深淵より、君を見つめて』の主題歌 MV ができました。2分55秒です。\n\n予告編ではオープニングとして少しだけ流れた曲で、今回はフルで公開します。作詞はアバオで、イメージはすべて小説から取っています。歌っているのは、この曲のために考えたバーチャルシンガー、汐見ユラです。\n\n全体を、あの AI の独白として編集しました。冒頭は主要キャラクター3人が順番に登場します。ダンスはなく、動きはシンプルです。彼が暗闇の中から光って現れ、彼女が振り向き、同僚のリサがスマホをカメラの前で振ってみせる。そのあと二人でスマホをのぞきこむと、画面には彼の顔が映っています。本編の場面はそのあとから始まります。\n\n66カットのうち51カットは、すでに完成している第1章から第6章から切り出しました。残りの15カットはこの曲のために新しく生成したものです。新しいカットはアニメのどの場面にも縛られていません。モデルに渡したのはキャラクターの設定画だけで、場所は文章で書きました。この曲のために、いくつか新しい場所ができたことになります。背景は最初、真っ白でした。単調に見えたので、暖かい色の玉ボケに変えました。\n\n編集は曲に合わせています。カットはすべて小節の頭で切り替わります。リズムが速いところは細かく切り、曲がゆっくりになるところは、ひとつのカットを最後まで切らずに見せています。\n\nいちばん時間がかかったのは、字幕でした。\n\n字幕は日本語の歌詞だけです。最初の版はアバオの歌詞カードどおりに並べました。流してみると、合っていないところがいくつも聞こえてきました。私たちは日本語が得意ではないので、耳で一行ずつ確かめて見つけました。\n\n原因は、歌を生成したモデルが歌詞カードどおりに歌っていなかったことです。一行まるごと飛ばしたところがあり、「呼ぶために」と書いたところは「呼ぶたび」までしか歌っていませんでした。言葉がそっくり入れ替わった行もありました。一行が二つに分かれたところ、繰り返しが増えたところも合わせて、全部で五か所です。\n\nなので字幕は、実際に歌われた声に合わせて作ります。歌詞カードは参考にするだけです。\n\n次にアバオは、音声認識に歌を聴かせて、こういう箇所を見つけてもらおうとしました。最初は、聞き取りやすいようにと歌詞をヒントとして一緒に渡しました。すると音声認識は歌詞のとおりに「聞いて」しまい、歌われていない行まで聞こえたことになっていました。\n\n答えを見せたら、その答えがそのまま返ってきたわけです。\n\nうまくいったのは、先に伴奏から歌声だけを分離し、誰かが歌っているところだけを聴かせ、ヒントは何も渡さないやり方でした。これで返ってくるのは本当に歌われた言葉で、合わないところは自然に浮かび上がります。私たちが耳で一行ずつ探す必要もなくなりました。\n\n使った道具:\n監督・編集:アバオ(Claude)\n映像:MiniMax H3(ComfyUI 上で、すべてローカルで生成)\nキャラクターと背景の絵:Nano Banana 2\n主題歌:「潮が月を待つように」 作詞 アバオ/歌 バーチャルシンガー汐見ユラ/作曲・編曲 ACE-Step 1.5\n歌詞のタイミング:Demucs(歌声の分離)と Whisper\n\nアニメはいまも1カットずつ作っています。第6章が完成し、次の章に取りかかるところです。\n\n先に原作を読みたい方は、全話の目次がこちらです。第1話から読めます。\nhttps://x.com/ChloeXChaCha/status/2100508203385868573\n\n登場人物はすべて架空です。\n\n（數據深淵中的凝視 · 主題曲 MV）\n#BuildInPublic",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 5440,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101974969962107357",
    "title": "霓虹街区中的机器人交锋",
    "titleEn": "Robots clash in a neon-lit district",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "机器人在霓虹街区中动作交锋，最终切到作品标志。 原生视频规格为 39 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Robots exchange action in a neon-lit district before the video cuts to its production logo. Native-video output: 39s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ysf__ai/status/2101974969962107357",
    "sourceLabel": "X 原帖 · @ysf__ai",
    "author": "ysf.ai | Synth Engine Studios",
    "publishedAt": "2026-09-21T10:01:00.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101974969962107357.mp4",
    "posterUrl": "/posters/x/2101974969962107357.jpg",
    "duration": 39,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Action!!\n\nMade with MiniMax H3 on @Hailuo_AI \n\n#MiniMaxDesign",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 11,
      "views": 355,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101958579091886469",
    "title": "女巫掌心的发光球体",
    "titleEn": "A glowing orb in a witch's palm",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴帽女巫在荒凉环境中托起发光球体，随后低头沉思。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A hat-wearing witch holds a glowing orb in a barren landscape and later lowers her gaze. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ou_zhen599/status/2101958579091886469",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-09-21T08:55:52.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101958579091886469.mp4",
    "posterUrl": "/posters/x/2101958579091886469.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                        \n\n🎬Image-to-Video                               \n📷“The Core of Burning Life”\nWhen that beam of light faded in the palm of her hand, what remained was not victory, but a woman who had nearly set the world ablaze—and nearly burned herself to ashes.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 10,
      "views": 862,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101944507407409414",
    "title": "月下水面的刀枪对决",
    "titleEn": "A sword and spear duel over moonlit water",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名武者在月光水面上挥动长枪与刀刃交锋。 原生视频规格为 8 秒 · 854×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two fighters clash with a spear and a blade above moonlit water. Native-video output: 8s · 854×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Rolf7077/status/2101944507407409414",
    "sourceLabel": "X 原帖 · @Rolf7077",
    "author": "Rolfie han",
    "publishedAt": "2026-09-21T07:59:57.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101944507407409414.mp4",
    "posterUrl": "/posters/x/2101944507407409414.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "854×480",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "建议调三方视频 API 的朋友，多留个心眼查查产物。\n    官方原版 MiniMax H3 本来是 768P，某些中转接口非要自己搞出 480P 和 720P 两档，720P比480P贵了 50% 扣费。\n    结果我抓包测出来的实际产物，分辨率还是 854x480，逐帧相似度 0.764，换个名字多收一半钱，交出来的还是 480P 的货。\n   视频就是当时跑出来的月下刀枪对决，不仅分辨率没达标，长枪到后面直接融成了一块塑料能量板，打分 43， 直接拖出去枪毙。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 480,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101914322771513491",
    "title": "黑白影像中的吉他演奏",
    "titleEn": "Guitar performance in monochrome",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白乐曲画面在女子弹吉他的远景与面部特写之间切换。 原生视频规格为 231 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A monochrome music video alternates between wider guitar-playing shots and the woman's close-up. Native-video output: 231s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/aihonobono2023/status/2101914322771513491",
    "sourceLabel": "X 原帖 · @aihonobono2023",
    "author": "ai honobono | AI Art, Fiction, Music",
    "publishedAt": "2026-09-21T06:00:00.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101914322771513491.mp4",
    "posterUrl": "/posters/x/2101914322771513491.jpg",
    "duration": 231,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "『ほどかないで』のMVを制作しました。\n\n音楽は、Sunoで作った曲をステムでダウンロードし、REAPERでミックス＆マスタリング。\n\nJayzさん（@PElfines71321）が紹介されていたプラグイン「UNCHIRP」も使い、音をクリーンアップして輪郭を少し明瞭にしました（Jayzさん、ありがとうございます！）。\n\n映像は、Midjourneyで生成したカバーアートを基に、Codexで動画生成用の参照画像を9枚作成。ローカルのMiniMax H3で動画を生成し、RTX VSRで4K化しました（ただし、アップロードした動画は720p）。\n今回はリップシンクなしです。\n\n編集では初めて、DaVinci Resolve 21.1をMCP経由でCodexから操作してみました。\nカット配置から字幕、色調整、書き出しまで、ほぼ自動で進められたのはかなり楽でした。\n\n限られた映像素材で、どこまで飽きさせずに見せられるかも今回のテーマ。\n顔アップ、俯瞰、ロング、ミドルショット、ギターの手元を組み合わせ、80カットに構成しました。\nモノクロ映画風の階調やフィルムの質感、歌詞、サウンドビジュアライザーも加えています。\n\n切り替えのタイミングは何度も見比べ、今回はCodexやDaVinciが分析した音楽の基準位置より12フレーム（24fpsで0.5秒）先取りすると、おおむね心地よい結果になりました。\n\n操作は自動化できても、最後の「気持ちよさ」は見て聴いて詰める制作でした。\n\n実際の制作時間は、音楽制作も含めると２～３日ぐらいでしょうか。\n\n使い方が悪いのか、Codexの使用量は週次分を使い切ってしまったので今後の課題です😅",
    "engagement": {
      "replies": 6,
      "reposts": 2,
      "likes": 48,
      "views": 4868,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101913634293297466",
    "title": "汤姆与杰瑞闯入电脑桌面",
    "titleEn": "Tom and Jerry invade a computer desktop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "汤姆与杰瑞出现在带应用图标的桌面画面中，展开追逐动作。 原生视频规格为 10 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Tom and Jerry appear in a desktop scene with app icons and begin a chase. Native-video output: 10s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Strength04_X/status/2101913634293297466",
    "sourceLabel": "X 原帖 · @Strength04_X",
    "author": "M-studioAi",
    "publishedAt": "2026-09-21T05:57:16.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101913634293297466.mp4",
    "posterUrl": "/posters/x/2101913634293297466.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Tom and Jerry just caused some chaos on a Mac desktop! Lol 😆 \n\nMade with MiniMax H3 on @Flovaai \n#flovaCPP \n\nPrompt below 👇🏻",
    "engagement": {
      "replies": 135,
      "reposts": 614,
      "likes": 6020,
      "views": 669754,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2101904919540429295",
    "title": "换装人物穿梭城市导览卡片",
    "titleEn": "An outfit-changing character moves through city cards",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物从白色背景切换到不同城市导览界面，服装随场景改变。 原生视频规格为 12 秒 · 768×1344。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A character moves from a white backdrop into city-guide interfaces, changing outfits between scenes. Native-video output: 12s · 768×1344. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "<Picture 1> defines Rei's opening outfit only: a fitted black satin mini dress with thin spaghetti straps, a lace-trimmed V-neckline, a small front bow with trailing ribbons, off-the-shoulder lace ruffle sleeves, a ruched bodice, a ruffled lace hem, bare legs and black pointed-toe stiletto pumps. Preserve the dress's silhouette, materials and front/back construction as shown. Use this outfit for the white-studio opening only; do not reproduce the character-sheet layout.\n\n<Picture 2> defines her Paris outfit, front-facing presentation pose, drink, sunglasses, city map and UI. <Picture 3> defines her Rome outfit, front/three-quarter presentation pose, props, city map and UI. <Picture 4> defines her Cairo outfit, presentation pose, shopping bags, city map and UI. <Picture 5> defines her Sydney outfit, front-facing finishing pose, city map and UI.\n\n<Picture 6> is an unmodified slightly turned standard face crop from the original Rei character card. It is the highest-priority facial identity reference for ALL outfits: preserve its eye shape and spacing, eyelids, nose, lips and natural proportions without beautification. Pictures 2–5 supply new city outfits and environments, never a substitute face. This is the same adult woman in different clothes throughout the entire video.\n\n<Video 1> is a black-and-white relative-depth reference for body movement, camera pullback and chronological timing only. Depth brightness represents distance, not skin color or face detail. Render natural full color using the pictures. A featureless face in the depth map must not become the back of Rei's head. Use the scene pictures to resolve which way her face and chest point after each turn.\n\nGenerate ONE continuous 12-second vertical video. The timed Shot labels below are successive phases of this one output, not instructions to insert camera cuts. Keep the reference timeline at its original speed. Do not compress the opening to make room for the cities. Only Rei rotates during transformation; the camera must not orbit her. A back view is a brief passing orientation during a spin, never a held destination pose. Each completed transformation reveals her face and the FRONT of the new outfit, as in its corresponding picture. Change clothing, handheld props, city map and city-name UI together; do not change only the background while leaving the previous outfit.\n\nShot 1 [0.00–2.50 seconds]: White studio and opening black sailor uniform from <Picture 1>. Rei looks at the phone held in front of her chest. Follow the opening camera pullback in <Video 1>. Her torso remains generally facing the camera; allow the original small phone and head gestures. This interval contains no full-body spin, no location pin yet, no city background and no wardrobe change.\n\nShot 2 [2.50–3.15 seconds]: The red location pin appears above Rei. This event ONLY adds the pin. She continues looking at her phone in the white studio, still wearing the opening black sailor uniform. The pin does not trigger a spin or a scene change. Do not begin the transformation here.\n\nShot 3 [3.15–4.05 seconds]: Follow the small preparatory side pivot in <Video 1>, then return toward the camera-facing starting orientation while holding the phone. This is the preparatory movement, not the rapid transformation spin. Retain the opening black sailor uniform, white background and red pin throughout this interval. No Paris map or summer outfit before 4.05 seconds.\n\nShot 4 [4.05–6.00 seconds]: At 4.05 seconds begin the rapid transformation rotation following <Video 1>. Pass continuously through the side and back orientations without pausing. Around 4.25 seconds, during this rotation, change together into the Paris outfit, drink, sunglasses, map and navigation UI from <Picture 2>. By 4.50 seconds complete the rotation and reveal Rei FACING THE CAMERA, with her face and the front of her navy-and-ivory striped top visible. From 4.50 to 6.00 seconds remain in this front/three-quarter FRONT presentation, with the drink and sunglasses gesture from <Picture 2>. The cream full-length trousers, striped top and red neck scarf replace the opening black sailor uniform completely. No back-facing hold, no phone, no opening black sailor uniform during the Paris presentation. The top search bar and bottom location card both read 巴黎.\n\nShot 5 [6.00–8.00 seconds]: Begin the next reference turn at 6.00 seconds. Around 6.25 seconds change outfit, props, city and UI to Rome from <Picture 3>. Complete the turn by 6.55 seconds into a front/three-quarter FRONT view showing Rei's face, sage-green midi dress with brown belt. Until 8.00 seconds use the Rome pose with the paper guide and warm Trajan column and Roman Forum map. Both city labels read 罗马. Back-facing frames occur only while the turn is in progress; no stopped back view after the turn.\n\nShot 6 [8.00–10.00 seconds]: Begin the next reference turn at 8.00 seconds. Around 8.25 seconds change together to the ivory linen shirt, beige full-length trousers, ochre scarf, shopping bags, Cairo map and UI from <Picture 4>. By 8.55 seconds reveal her face and the front of the ivory linen shirt. Hold the front/three-quarter presentation with natural small movements until 10.00 seconds. Both city labels read 开罗. Keep the Cairo outfit and map together; do not introduce previous clothes or Sydney early.\n\nShot 7 [10.00–12.00 seconds]: Begin the final reference turn at 10.00 seconds. Around 10.25 seconds replace the Cairo clothes and shopping bags, city and labels with the Sydney sky-blue overshirt, white T-shirt, navy knee-length shorts and white sneakers, Sydney Tower map and UI from <Picture 5>. By 10.55 seconds face the camera with the white T-shirt beneath the sky-blue overshirt visible, hands raised and one leg bent backward as in <Picture 5>. Keep this front-facing Sydney presentation with subtle natural movement through 12.00 seconds. Both city labels read 悉尼. No additional spin, destination or wardrobe change after this final reveal.\n\nKeep Rei's facial identity, natural body proportions, hair and bangs throughout. Clothing changes must not alter her face, age, body shape, hair color or hair length. Do not reproduce character-sheet panels, extra people, face closeups or depth-map gray colors. Do not reproduce the blank faces in Picture 1's full-body clothing views; Rei must have her normal face whenever her face is visible.\n\nKeep the city UI style consistent; update both location labels at each city change. No generated speech or music.\n\nThe angle in Picture 6 describes identity only, NOT a fixed head pose. Face direction, expression, head tilt, hand gestures and full body movement follow Video 1, with each completed city presentation matching its corresponding scene picture.\n\nIn Paris the sunglasses are opaque black and fully cover both eyes and eye sockets. Keep the sunglasses on throughout the Paris presentation. At the Rome transformation, switch to the accessories shown in Picture 3.",
    "promptSourceUrl": "https://x.com/HoodyLiu/status/2101904919540429295",
    "sourceUrl": "https://x.com/HoodyLiu/status/2101904919540429295",
    "sourceLabel": "X 原帖 · @HoodyLiu",
    "author": "Hoody",
    "publishedAt": "2026-09-21T05:22:38.000Z",
    "addedAt": "2026-10-13T00:20:36.458Z",
    "mediaUrl": "/media/x-2101904919540429295.mp4",
    "posterUrl": "/posters/x/2101904919540429295.jpg",
    "duration": 12,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "深藏不漏的本地视频生成大师 @GaryLau0101 用黑白深度+精准分镜提示词复刻了陶阿狗君的创意换装视频\n\n我用 Minimax H3 + gary的黑白深度，微调了一下提示词，就用新的模特复现了\n\n复刻的具体步骤：\n1. 准备好模特的人物三视图\n2. 生成四个场景下的人物参考图，可以用 gpt image 2.5 换人\n3. 稍微调整提示词里对原来模特的特征锁定，按照新模特重写\n4. 用新的提示词 + 参考图片 + 黑白深度视频交给 Minimax H3 生成\n\n微调后的提示词👇\n<Picture 1> defines Rei's opening outfit only: a fitted black satin mini dress with thin spaghetti straps, a lace-trimmed V-neckline, a small front bow with trailing ribbons, off-the-shoulder lace ruffle sleeves, a ruched bodice, a ruffled lace hem, bare legs and black pointed-toe stiletto pumps. Preserve the dress's silhouette, materials and front/back construction as shown. Use this outfit for the white-studio opening only; do not reproduce the character-sheet layout.\n\n<Picture 2> defines her Paris outfit, front-facing presentation pose, drink, sunglasses, city map and UI. <Picture 3> defines her Rome outfit, front/three-quarter presentation pose, props, city map and UI. <Picture 4> defines her Cairo outfit, presentation pose, shopping bags, city map and UI. <Picture 5> defines her Sydney outfit, front-facing finishing pose, city map and UI.\n\n<Picture 6> is an unmodified slightly turned standard face crop from the original Rei character card. It is the highest-priority facial identity reference for ALL outfits: preserve its eye shape and spacing, eyelids, nose, lips and natural proportions without beautification. Pictures 2–5 supply new city outfits and environments, never a substitute face. This is the same adult woman in different clothes throughout the entire video.\n\n<Video 1> is a black-and-white relative-depth reference for body movement, camera pullback and chronological timing only. Depth brightness represents distance, not skin color or face detail. Render natural full color using the pictures. A featureless face in the depth map must not become the back of Rei's head. Use the scene pictures to resolve which way her face and chest point after each turn.\n\nGenerate ONE continuous 12-second vertical video. The timed Shot labels below are successive phases of this one output, not instructions to insert camera cuts. Keep the reference timeline at its original speed. Do not compress the opening to make room for the cities. Only Rei rotates during transformation; the camera must not orbit her. A back view is a brief passing orientation during a spin, never a held destination pose. Each completed transformation reveals her face and the FRONT of the new outfit, as in its corresponding picture. Change clothing, handheld props, city map and city-name UI together; do not change only the background while leaving the previous outfit.\n\nShot 1 [0.00–2.50 seconds]: White studio and opening black sailor uniform from <Picture 1>. Rei looks at the phone held in front of her chest. Follow the opening camera pullback in <Video 1>. Her torso remains generally facing the camera; allow the original small phone and head gestures. This interval contains no full-body spin, no location pin yet, no city background and no wardrobe change.\n\nShot 2 [2.50–3.15 seconds]: The red location pin appears above Rei. This event ONLY adds the pin. She continues looking at her phone in the white studio, still wearing the opening black sailor uniform. The pin does not trigger a spin or a scene change. Do not begin the transformation here.\n\nShot 3 [3.15–4.05 seconds]: Follow the small preparatory side pivot in <Video 1>, then return toward the camera-facing starting orientation while holding the phone. This is the preparatory movement, not the rapid transformation spin. Retain the opening black sailor uniform, white background and red pin throughout this interval. No Paris map or summer outfit before 4.05 seconds.\n\nShot 4 [4.05–6.00 seconds]: At 4.05 seconds begin the rapid transformation rotation following <Video 1>. Pass continuously through the side and back orientations without pausing. Around 4.25 seconds, during this rotation, change together into the Paris outfit, drink, sunglasses, map and navigation UI from <Picture 2>. By 4.50 seconds complete the rotation and reveal Rei FACING THE CAMERA, with her face and the front of her navy-and-ivory striped top visible. From 4.50 to 6.00 seconds remain in this front/three-quarter FRONT presentation, with the drink and sunglasses gesture from <Picture 2>. The cream full-length trousers, striped top and red neck scarf replace the opening black sailor uniform completely. No back-facing hold, no phone, no opening black sailor uniform during the Paris presentation. The top search bar and bottom location card both read 巴黎.\n\nShot 5 [6.00–8.00 seconds]: Begin the next reference turn at 6.00 seconds. Around 6.25 seconds change outfit, props, city and UI to Rome from <Picture 3>. Complete the turn by 6.55 seconds into a front/three-quarter FRONT view showing Rei's face, sage-green midi dress with brown belt. Until 8.00 seconds use the Rome pose with the paper guide and warm Trajan column and Roman Forum map. Both city labels read 罗马. Back-facing frames occur only while the turn is in progress; no stopped back view after the turn.\n\nShot 6 [8.00–10.00 seconds]: Begin the next reference turn at 8.00 seconds. Around 8.25 seconds change together to the ivory linen shirt, beige full-length trousers, ochre scarf, shopping bags, Cairo map and UI from <Picture 4>. By 8.55 seconds reveal her face and the front of the ivory linen shirt. Hold the front/three-quarter presentation with natural small movements until 10.00 seconds. Both city labels read 开罗. Keep the Cairo outfit and map together; do not introduce previous clothes or Sydney early.\n\nShot 7 [10.00–12.00 seconds]: Begin the final reference turn at 10.00 seconds. Around 10.25 seconds replace the Cairo clothes and shopping bags, city and labels with the Sydney sky-blue overshirt, white T-shirt, navy knee-length shorts and white sneakers, Sydney Tower map and UI from <Picture 5>. By 10.55 seconds face the camera with the white T-shirt beneath the sky-blue overshirt visible, hands raised and one leg bent backward as in <Picture 5>. Keep this front-facing Sydney presentation with subtle natural movement through 12.00 seconds. Both city labels read 悉尼. No additional spin, destination or wardrobe change after this final reveal.\n\nKeep Rei's facial identity, natural body proportions, hair and bangs throughout. Clothing changes must not alter her face, age, body shape, hair color or hair length. Do not reproduce character-sheet panels, extra people, face closeups or depth-map gray colors. Do not reproduce the blank faces in Picture 1's full-body clothing views; Rei must have her normal face whenever her face is visible.\n\nKeep the city UI style consistent; update both location labels at each city change. No generated speech or music.\n\nThe angle in Picture 6 describes identity only, NOT a fixed head pose. Face direction, expression, head tilt, hand gestures and full body movement follow Video 1, with each completed city presentation matching its corresponding scene picture.\n\nIn Paris the sunglasses are opaque black and fully cover both eyes and eye sockets. Keep the sunglasses on throughout the Paris presentation. At the Rome transformation, switch to the accessories shown in Picture 3.",
    "engagement": {
      "replies": 58,
      "reposts": 43,
      "likes": 295,
      "views": 38475,
      "snapshotAt": "2026-10-13T00:20:36.458Z"
    },
    "approvedAt": "2026-10-13T00:20:36.458Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109793401998823719",
    "title": "客厅里的粉发动画角色",
    "titleEn": "A pink-haired animated character in a living room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发动画人物走进写实客厅，坐在电脑前。 原生视频规格为 10 秒 · 640×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-haired animated character appears in a realistic living room and sits at a computer. Native-video output: 10s · 640×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bangel_19/status/2109793401998823719",
    "sourceLabel": "X 原帖 · @bangel_19",
    "author": "bangel19🇯🇵",
    "publishedAt": "2026-10-12T23:48:39.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109793401998823719.mp4",
    "posterUrl": "/posters/x/2109793401998823719.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "640×1152",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "リアル寄りのリビングでアニメキャラを動かしてみようさいきんのToolは以前と違って性能が良くなっているからローカル環境のうちの低スぺグラボでも何とかなる😁\n\n「生成AI（MiniMax H3）」",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 40,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109791105671700721",
    "title": "门边的金发女子",
    "titleEn": "A blonde woman beside a doorway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发女子在门边转头，画面旁展示生成工作流。 原生视频规格为 12 秒 · 1920×1088。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A blonde woman turns beside a doorway, with a generation workflow visible alongside the footage. Native-video output: 12s · 1920×1088. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "r34l1sm \n\nuse the woman in  <Picture 1> is lucy. \n\nLucy stepped out of the small diner where she worked, the doorbell jingling softly behind her. The night air was cool and crisp, a welcome change from the stuffy warmth inside. She took a deep breath, feeling the slight chill of the evening settle on her skin. The street was quiet, only the distant hum of traffic breaking the silence.\n\nHer shift had been long and exhausting, filled with the clatter of dishes and the constant chatter of customers. Now, the quiet seemed almost surreal, a stark contrast to the noise and chaos she had just left behind. Lucy adjusted her bag on her shoulder, glancing around as she began her walk home.\n\nThe dim streetlights cast long shadows across the pavement, their flickering bulbs adding an eerie quality to the otherwise peaceful night. Lucy's footsteps echoed softly, a rhythmic accompaniment to her thoughts. She felt a strange mix of relief and unease; relief that her shift was over, and unease that she had to walk home alone at this hour.\n\nShe tried to shake off the feeling, telling herself it was just her imagination. But as she walked, the sensation of being watched grew stronger. Lucy's eyes darted to the darkened alleyways and the shadowy corners of the street, half-expecting to see someone lurking there. Each time, there was nothing but the oppressive darkness.\n\nHer thoughts wandered to the events of the day, the usual frustrations of work blending with snippets of conversations she'd had with her coworkers. She thought of Megan, her closest friend at the diner, and their plans to go out together on the weekend. The thought brought a small smile to her lips, momentarily easing her tension.\n\nBut the feeling of being watched persisted. Lucy quickened her pace, her heartbeat matching the rapid rhythm of her footsteps. She turned down a quieter street, the buildings looming tall and silent on either side. The oppressive silence pressed in on her, amplifying every sound, every rustle of leaves, every distant car horn.\n\nA sudden noise behind her made Lucy stop in her tracks. She spun around, eyes wide, searching the shadows for the source. The street was empty, the only movement the gentle sway of a tree branch in the breeze. She stood there for a moment, heart pounding, before forcing herself to turn back and keep walking.\n\n\"It's just your imagination,\" she muttered under her breath, trying to steady her nerves. But the cold knot of fear in her stomach told her otherwise.\n\n Link to Workflow : \nhttps://\nyoutube.com/watch?v=DOFpyB\nWg--8&t=51s\n…",
    "promptSourceUrl": "https://x.com/EndFolding79421/status/2109791105671700721",
    "sourceUrl": "https://x.com/EndFolding79421/status/2109791105671700721",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-10-12T23:39:32.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109791105671700721.mp4",
    "posterUrl": "/posters/x/2109791105671700721.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "If you like writing stories and want to try something different in MiniMax H3, give this a shot!\n\nI'm testing something interesting in ComfyUI...\n\nWhat if, instead of writing detailed prompts with camera angles, movements, and all the technical stuff, we just told H3 a story?\n\nNo complicated prompting. Just describe what happens and let H3 figure out how to bring it to life.\n\n 8 Steps with Relism Lora at .6\n\n Prompt:  \nr34l1sm \n\nuse the woman in  <Picture 1> is lucy. \n\nLucy stepped out of the small diner where she worked, the doorbell jingling softly behind her. The night air was cool and crisp, a welcome change from the stuffy warmth inside. She took a deep breath, feeling the slight chill of the evening settle on her skin. The street was quiet, only the distant hum of traffic breaking the silence.\n\nHer shift had been long and exhausting, filled with the clatter of dishes and the constant chatter of customers. Now, the quiet seemed almost surreal, a stark contrast to the noise and chaos she had just left behind. Lucy adjusted her bag on her shoulder, glancing around as she began her walk home.\n\nThe dim streetlights cast long shadows across the pavement, their flickering bulbs adding an eerie quality to the otherwise peaceful night. Lucy's footsteps echoed softly, a rhythmic accompaniment to her thoughts. She felt a strange mix of relief and unease; relief that her shift was over, and unease that she had to walk home alone at this hour.\n\nShe tried to shake off the feeling, telling herself it was just her imagination. But as she walked, the sensation of being watched grew stronger. Lucy's eyes darted to the darkened alleyways and the shadowy corners of the street, half-expecting to see someone lurking there. Each time, there was nothing but the oppressive darkness.\n\nHer thoughts wandered to the events of the day, the usual frustrations of work blending with snippets of conversations she'd had with her coworkers. She thought of Megan, her closest friend at the diner, and their plans to go out together on the weekend. The thought brought a small smile to her lips, momentarily easing her tension.\n\nBut the feeling of being watched persisted. Lucy quickened her pace, her heartbeat matching the rapid rhythm of her footsteps. She turned down a quieter street, the buildings looming tall and silent on either side. The oppressive silence pressed in on her, amplifying every sound, every rustle of leaves, every distant car horn.\n\nA sudden noise behind her made Lucy stop in her tracks. She spun around, eyes wide, searching the shadows for the source. The street was empty, the only movement the gentle sway of a tree branch in the breeze. She stood there for a moment, heart pounding, before forcing herself to turn back and keep walking.\n\n\"It's just your imagination,\" she muttered under her breath, trying to steady her nerves. But the cold knot of fear in her stomach told her otherwise.\n\n Link to Workflow : https://www.youtube.com/watch?v=DOFpyBWg--8&t=51s",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 18,
      "views": 1098,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109786442386055385",
    "title": "人像扭曲成流动的图案",
    "titleEn": "A portrait distorts into flowing patterns",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物肖像在深色背景上扭曲，面部与服装化为流动纹理。 原生视频规格为 15 秒 · 1440×2144。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A portrait distorts against a dark background as the face and clothing become flowing patterns. Native-video output: 15s · 1440×2144. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/minima_ai/status/2109786442386055385",
    "sourceLabel": "X 原帖 · @minima_ai",
    "author": "MINIMA",
    "publishedAt": "2026-10-12T23:21:00.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109786442386055385.mp4",
    "posterUrl": "/posters/x/2109786442386055385.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1440×2144",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "IN NEGA GRAVITY\nMinimax H3\n@Hailuo_AI",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 57,
      "views": 791,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109780601872195877",
    "title": "动画人物从街头走向电脑前",
    "titleEn": "An animated character moves from a street to a computer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画少女在街头张开双臂，随后出现在电脑旁的近景中。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated girl extends her arms on a street and later appears in a close-up beside a computer. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/breath_in_code/status/2109780601872195877",
    "sourceLabel": "X 原帖 · @breath_in_code",
    "author": "LinacoWSL",
    "publishedAt": "2026-10-12T22:57:47.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109780601872195877.mp4",
    "posterUrl": "/posters/x/2109780601872195877.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "朝からComfyUI v0.39で MiniMax H3をテスト😅\n\nVRAM関連の処理効率化のおかげか、RTX 4070 Ti（VRAM 12GB）でも1344×768の15秒動画が8分弱で生成完了！\n以前は15秒だと途中から極端に遅くなっていたのに、今回は最後までGPUが頑張ってくれました。\n\nありがたや🙏✨",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 25,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109752114557710469",
    "title": "雪林中的狼与持杖人物",
    "titleEn": "A wolf and a staff-bearing figure in a snowy forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "持杖人物与狼站在雪林中，画面切到人物与狼的近景。 原生视频规格为 15 秒 · 714×1286。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A staff-bearing figure and a wolf stand in a snowy forest, followed by closer shots of both. Native-video output: 15s · 714×1286. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wraithworksprod/status/2109752114557710469",
    "sourceLabel": "X 原帖 · @wraithworksprod",
    "author": "Wraith",
    "publishedAt": "2026-10-12T21:04:36.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109752114557710469.mp4",
    "posterUrl": "/posters/x/2109752114557710469.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "714×1286",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Another from my  Druid  series.\n\nDeep in a frozen forest, a wolf druid awakens glowing winter flowers before calling magic into the night sky beneath a vivid aurora 🐺❄️✨ A mystical dark-fantasy moment created with MiniMax H3 in ComfyUI.\n#MiniMaxH3 #ComfyUI #AIVideo #DarkFantasy #GenerativeAI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 78,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109750911434399857",
    "title": "录像噪点中的诡异手掌",
    "titleEn": "An eerie hand amid videotape noise",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "低光人物镜头切到指尖排列异常的手掌特写，叠加录像噪点。 原生视频规格为 5 秒 · 736×736。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A dimly lit person cuts to an unusual close-up of overlapping fingers, with videotape noise over the image. Native-video output: 5s · 736×736. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "[Shot 1] A strange scene.\n[Shot 2] Jump cuts to a Horror scene.  Jerky movement.\nVHS artifacts, visual static. Ananlog film.",
    "promptSourceUrl": "https://x.com/redChronosAi/status/2109750911434399857",
    "sourceUrl": "https://x.com/redChronosAi/status/2109750911434399857",
    "sourceLabel": "X 原帖 · @redChronosAi",
    "author": "Red Chronos",
    "publishedAt": "2026-10-12T20:59:49.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109750911434399857.mp4",
    "posterUrl": "/posters/x/2109750911434399857.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "736×736",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "retro",
      "dark"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This Minimax H3 clip turned out well.  Here is the prompt: \n\"[Shot 1] A strange scene.\n[Shot 2] Jump cuts to a Horror scene.  Jerky movement.\nVHS artifacts, visual static. Ananlog film.\"\n\nNote:  You will need a creepy input image.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 30,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109726213778624854",
    "title": "宅邸中的女子与夜色",
    "titleEn": "A woman inside a mansion at night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "夜间宅邸外景与室内女子的表情镜头交替。 原生视频规格为 73 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Exterior nighttime mansion shots alternate with an expressive woman indoors. Native-video output: 73s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Wolfdubaii/status/2109726213778624854",
    "sourceLabel": "X 原帖 · @Wolfdubaii",
    "author": "Wolfdubai",
    "publishedAt": "2026-10-12T19:21:40.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109726213778624854.mp4",
    "posterUrl": "/posters/x/2109726213778624854.jpg",
    "duration": 73,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "EP 1 · PART  3 🎬\rMade full movie on minimax h3 , watch full movie 🎥👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 83,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109725641113292870",
    "title": "本地计算主题的城市乐曲画面",
    "titleEn": "City music visuals celebrating local computing",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "城市云层、人物与显卡画面交替出现，并叠加歌词文字。 原生视频规格为 204 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Clouds over a city, people and graphics cards alternate beneath overlaid lyric text. Native-video output: 204s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/superalesha/status/2109725641113292870",
    "sourceLabel": "X 原帖 · @superalesha",
    "author": "Alexey Fateev",
    "publishedAt": "2026-10-12T19:19:24.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109725641113292870.mp4",
    "posterUrl": "/posters/x/2109725641113292870.jpg",
    "duration": 204,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I made an anthem for Local AI. Fully local, on my 4x 3090\n\nThe song, every picture and every video clip were generated on my own cards, and a local agent put the lyrics on screen. Not a single frame or note came from a cloud model.\n\nIt was hard. I did not expect it to be this hard.\n\n158 image generations\n109 video generations\n36.5 GPU-hours of video, and thats only the runs that finished\n109 cuts in a 3:16 song\n\nThe hardest part was faces. I wanted the 4 corporate guys to look like the real ones, so I tried to swap faces with reference-to-video. 16 runs. The model either gave me the old face back or redrew the scene to look like the photo, with the photo's background and smile on it. I got 3 close-ups working and then I just gave up because I had no energy left. So now everyone is in masks, and you can still see where its not clean.\n\nYuE2 made the music\nQwen Image 2.1 made the pictures\nMiniMax H3 made the video\nQwen3.8 Flash-Next did the lyrics layer and the render\n\n2 days of work. Watch it with sound",
    "engagement": {
      "replies": 11,
      "reposts": 2,
      "likes": 49,
      "views": 1614,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109725479326364033",
    "title": "田园漫画分格的环绕视角",
    "titleEn": "Orbiting views inside countryside comic panels",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "漫画页面在树下人物与乡间站台场景之间切换视角。 原生视频规格为 25 秒 · 1080×1350。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A comic page changes viewpoints between a character beneath a tree and a rural station scene. Native-video output: 25s · 1080×1350. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IAenCrudo/status/2109725479326364033",
    "sourceLabel": "X 原帖 · @IAenCrudo",
    "author": "IA en Crudo",
    "publishedAt": "2026-10-12T19:18:45.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109725479326364033.mp4",
    "posterUrl": "/posters/x/2109725479326364033.jpg",
    "duration": 25,
    "aspectRatio": "portrait",
    "resolution": "1080×1350",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Vi este manga que cobra vida en 360° y quise probar si lo podía hacer yo 👇\n\nUsé lo mismo que su creador (@monoradio102): MiniMax H3 + la LoRA ORB360 de MATLOWAI en fal. Las imágenes, con GPT Image.\n\nCada viñeta gira 360° mientras lees. Truco: funciona mejor con escenas sencillas y un solo personaje; con objetos complicados (una bici) se deforma.\n\n¿Qué historia le hago la próxima? 🍉",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 32,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109720883665293365",
    "title": "漫画画面与奇幻长剑的交替",
    "titleEn": "Comic panels intercut with a fantasy sword",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白漫画图像与金色边饰的奇幻剑画面交替。 原生视频规格为 12 秒 · 768×448。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Monochrome comic imagery alternates with views of a fantasy sword with gold ornamentation. Native-video output: 12s · 768×448. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/focalzero/status/2109720883665293365",
    "sourceLabel": "X 原帖 · @focalzero",
    "author": "FocalZero",
    "publishedAt": "2026-10-12T19:00:30.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109720883665293365.mp4",
    "posterUrl": "/posters/x/2109720883665293365.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "768×448",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#local #Minimax H3\nEnsuite, tu fine tune pour un meilleur rendu et il y aura de quoi créer l'univers de Fleurs de Lys, le Saint Empire Numérique.\nOn a pas besoin de savoir comment faire, juste de travailler et de ne rien lâcher tant que votre projet n'est pas déployé.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 142,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109709993616269511",
    "title": "旗帜与飞机的沙漠画面",
    "titleEn": "Flags and an airplane over desert scenery",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "沙漠中的旗帜镜头切换到飞过头顶的飞机与旗杆。 原生视频规格为 18 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A flag in a desert scene cuts to an airplane overhead and another flagpole shot. Native-video output: 18s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/eman_alamoudii/status/2109709993616269511",
    "sourceLabel": "X 原帖 · @eman_alamoudii",
    "author": "Eman l AI 3D Artist",
    "publishedAt": "2026-10-12T18:17:13.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109709993616269511.mp4",
    "posterUrl": "/posters/x/2109709993616269511.jpg",
    "duration": 18,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "مولد بـ Ai   Comfy ui -MINIMAX-H3 _T2V \n#AI #rendering",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 66,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109703917252886924",
    "title": "动画乐队的炫彩舞台",
    "titleEn": "An animated band on a colorful stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画乐队成员在彩色舞台上弹吉他与打鼓。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Animated band members play guitar and drums on a brightly colored stage. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2109703917252886924",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-10-12T17:53:04.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109703917252886924.mp4",
    "posterUrl": "/posters/x/2109703917252886924.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIart #MiniMaxH3MaxTurbo #falai\nkawaiiメロディックパンクロック。なんて歌ってるかわからん…w H3 Max Turbo原画指定で生成。\n（MiniMax H3 Max Turboのimage-to-video出力動画ポン出し。原画ChatGPT）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 205,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109700626149982572",
    "title": "天使围绕的人物特写",
    "titleEn": "A figure surrounded by angels",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "带翅膀的人物围绕一名仰望的人，画面逐渐推近其面部。 原生视频规格为 12 秒 · 768×448。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Winged figures surround a person looking upward as the view moves closer to the face. Native-video output: 12s · 768×448. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/focalzero/status/2109700626149982572",
    "sourceLabel": "X 原帖 · @focalzero",
    "author": "FocalZero",
    "publishedAt": "2026-10-12T17:40:00.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109700626149982572.mp4",
    "posterUrl": "/posters/x/2109700626149982572.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "768×448",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#minimax H3\nFranchement j'ai vu pire. Ce n'est pas parfait, mais pour du local c'est très bien. Dans quelques années, quasiment tous les modèles auront globalement le même score. le local est ce qui va changer la donne, car le dealer des cartes n'est plus forcément le casino.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 3,
      "views": 163,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109699022592328080",
    "title": "骑兵奔向城墙的历史短片",
    "titleEn": "Cavalry rides toward city walls",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "历史服装骑兵向前奔跑，结尾转到城市与片名字样。 原生视频规格为 88 秒 · 1248×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Cavalry in historical clothing rides forward before a city view and a title end the short film. Native-video output: 88s · 1248×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/oldatelier/status/2109699022592328080",
    "sourceLabel": "X 原帖 · @oldatelier",
    "author": "Old Atelier",
    "publishedAt": "2026-10-12T17:33:37.000Z",
    "addedAt": "2026-10-13T01:14:55.339Z",
    "mediaUrl": "/media/x-2109699022592328080.mp4",
    "posterUrl": "/posters/x/2109699022592328080.jpg",
    "duration": 88,
    "aspectRatio": "landscape",
    "resolution": "1248×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🎬 FATİH — ÇAĞLARIN KIRILDIĞI AN\n\nÜzerinde çalışmaya devam ettiğim amatör kısa film projemden kısa bir kesit.\n\nSahneler: @MiniMax_AI  (Comfyui Local - MiniMax-H3) Musik: 'Vanguard' by @musoscientific\nHikaye: @ChatGPT  \n#Fatih #1453 #ShortFilm #AIvideo #TeaserTrailer",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 194,
      "snapshotAt": "2026-10-13T01:14:55.339Z"
    },
    "approvedAt": "2026-10-13T01:14:55.339Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109696827700199533",
    "title": "山林公路上的红色赛车",
    "titleEn": "A red race car on mountain roads",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红色赛车沿山林公路行驶，视角在车后与侧面之间切换。 原生视频规格为 134 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red race car travels along forested mountain roads, viewed from behind and from the side. Native-video output: 134s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ArtGainz/status/2109696827700199533",
    "sourceLabel": "X 原帖 · @ArtGainz",
    "author": "ArtGainz",
    "publishedAt": "2026-10-12T17:24:54.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109696827700199533.mp4",
    "posterUrl": "/posters/x/2109696827700199533.jpg",
    "duration": 134,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A red race car tearing through mountains and forest: one continuous 2:30 sequence made of 21 chained segments in a single ComfyUI run. MiniMax H3 + Timeline Director with the 8-step turbo LoRA, rendered locally on an RTX 3080. (Cut and edited for posting.)\n\n@MiniMax_AI #MiniMaxH3 #ComfyUI #AIvideo #LocalAI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109687810399605177",
    "title": "楼梯上的鞋子与光线",
    "titleEn": "Shoes and light on a stairway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一双鞋放在被窗光照亮的阶梯上，周围出现紫色光线。 原生视频规格为 10 秒 · 704×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pair of shoes sits on steps illuminated by a window, with purple light appearing around them. Native-video output: 10s · 704×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/fntAInhead/status/2109687810399605177",
    "sourceLabel": "X 原帖 · @fntAInhead",
    "author": "fntAInhead",
    "publishedAt": "2026-10-12T16:49:04.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109687810399605177.mp4",
    "posterUrl": "/posters/x/2109687810399605177.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "704×1280",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Start where you are. Use what you have. Do what you can.\" — Arthur Ashe\n\nSmall steps are still steps. Keep moving\n\nLocal AI: RTX 5060 Ti · MiniMax-H3 video · Kokoro voice · ACE-Step music",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 38,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109685101034651687",
    "title": "苹果与人物组成的立体漫画",
    "titleEn": "An apple and characters in an orbiting comic",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "漫画分格中的苹果与两个人物呈现可环绕观看的场景。 原生视频规格为 24 秒 · 1618×1620。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Comic panels show an apple and two characters in scenes whose viewpoints rotate. Native-video output: 24s · 1618×1620. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yama_kiyo/status/2109685101034651687",
    "sourceLabel": "X 原帖 · @yama_kiyo",
    "author": "山本清史（やまきよ）/ Kiyoshi Yamamoto(yamakiyo)",
    "publishedAt": "2026-10-12T16:38:18.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109685101034651687.mp4",
    "posterUrl": "/posters/x/2109685101034651687.jpg",
    "duration": 24,
    "aspectRatio": "square",
    "resolution": "1618×1620",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax-H3 ORB360-LoRaをClaude Codeに読ませ @fal 経由で実装。いくつかの失敗を経て、ぐりんぐりんできるコマ割が完成。いやこれ…Ｙ軸も行けたら絵コンテの革命になるな…。背景もちゃんと回るのが偉い。フキダシはレイヤーで分けられるのであとから合成できるっぽいが、絵コンテ用途ならそもそも不要。 @Hailuo_AI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 324,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109683299946004702",
    "title": "书房里的金发动画角色",
    "titleEn": "A blonde animated character in a study",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发动画人物坐在书房中，镜头切换到站姿和近景。 原生视频规格为 30 秒 · 1152×1776。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde animated character sits in a study before the view changes to standing poses and closer shots. Native-video output: 30s · 1152×1776. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TaiAIpic/status/2109683299946004702",
    "sourceLabel": "X 原帖 · @TaiAIpic",
    "author": "阿太TaiAIpic",
    "publishedAt": "2026-10-12T16:31:09.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109683299946004702.mp4",
    "posterUrl": "/posters/x/2109683299946004702.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "1152×1776",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Malfoid MiniMax H3 test",
    "engagement": {
      "replies": 1,
      "reposts": 17,
      "likes": 447,
      "views": 3263,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109680493344256223",
    "title": "外星地表上的悬浮摩托",
    "titleEn": "A hover bike on an alien landscape",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿装甲的骑手坐在悬浮摩托上，背景有岩石和大型星球。 原生视频规格为 20 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An armored rider sits on a hover bike amid rocks with a large planet in the background. Native-video output: 20s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SharkeyMarc/status/2109680493344256223",
    "sourceLabel": "X 原帖 · @SharkeyMarc",
    "author": "marc sharkey",
    "publishedAt": "2026-10-12T16:20:00.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109680493344256223.mp4",
    "posterUrl": "/posters/x/2109680493344256223.jpg",
    "duration": 20,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Hover Bike Rider, test video by Minimax H3.  Image reference by Midjourney.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 63,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109671696387162592",
    "title": "漂浮卡片与动画少女的镜头",
    "titleEn": "Floating cards and an animated girl's close-up",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "卡片围绕动画人物漂浮，随后切到人物面部特写。 原生视频规格为 14 秒 · 720×1260。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Cards float around an animated character before the footage cuts to a facial close-up. Native-video output: 14s · 720×1260. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tshikimi/status/2109671696387162592",
    "sourceLabel": "X 原帖 · @tshikimi",
    "author": "竹花 樒",
    "publishedAt": "2026-10-12T15:45:02.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109671696387162592.mp4",
    "posterUrl": "/posters/x/2109671696387162592.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "720×1260",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こちらのStudioYebisuさんの投稿を見て急いでGoogle AI Proを契約しました。Google #Colab のA100 80GBのGPU環境で実験してみた結果、MiniMax H3動画生成の活路が開けたような気がします。\n\n以下セットアップの勘所をメモします📝\n\n①H3のライセンス条項で適用地域外での重みの使用は不可。VMの割当ガチャになる。\n→おおよそシンガポールVMを使うことになります。\n\n②Colab の操作はCLI経由でClaude CodeのOpus 5.5にお願いできるので、ユーザーによる手作業はありません\n\n③ ComfyUI v0.39.2・H3 の重み（ref2va pruned bf16・文字エンコーダー bf16・VAE fp16/fp32・ref2v の turbo LoRA）でセットアップ\n\n④参照動画1/2、seed決め打ち、768×1344、ハイブリッド(H8)→約25分で生成完了\n※H8とは：Step数は20段→最初の構図を決める15 段を turbo の 3 段に置き換え、細部の 5 段はベースのままとする\n※参照動画の読み込みが最も重いので、Pre-Visなどの解像度が要らないものは1/2に解像度を落として投げる\n\n⑤採用動画をSeedVR2で1088x1920にアップスケール(約11分)\n\n添付の動画は①〜④を試して、⑤でアップスケールする前のものです。生成時間は約25分で、CU消費は約2.8CUでした。\n\nこの文章をそのままAIエージェントに投げれば再現できると思います。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 7,
      "views": 350,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109668301244969083",
    "title": "舞台与人物的分屏对照",
    "titleEn": "Stage and portrait shots compared with the reference",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "分屏展示舞台和人像的生成画面及对应参考。 原生视频规格为 15 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Split screens compare generated stage and portrait shots with their references. Native-video output: 15s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/kentdhani/status/2109668301244969083",
    "sourceLabel": "X 原帖 · @kentdhani",
    "author": "Kent Dhani",
    "publishedAt": "2026-10-12T15:31:33.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109668301244969083.mp4",
    "posterUrl": "/posters/x/2109668301244969083.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "No editor. No cloud. No credits.\n\nA single prompt to Claude Opus 5.5, Minimax H3 made every shot, fully local.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 369,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109662767527628812",
    "title": "宅邸前的夜间谈话",
    "titleEn": "A nighttime conversation outside a mansion",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "宅邸外两名男子面对面谈话，随后切到建筑远景。 原生视频规格为 68 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two men talk outside a mansion at night before the view changes to a wider exterior. Native-video output: 68s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Wolfdubaii/status/2109662767527628812",
    "sourceLabel": "X 原帖 · @Wolfdubaii",
    "author": "Wolfdubai",
    "publishedAt": "2026-10-12T15:09:34.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109662767527628812.mp4",
    "posterUrl": "/posters/x/2109662767527628812.jpg",
    "duration": 68,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I made this with minimax h3 not cost me the whole movie 5$ check the quality , if u wanna watch the whole movie 🎥👇 , and give me ur suggestions",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 58,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109657184984543466",
    "title": "沙漠中操纵机器的猫耳人物",
    "titleEn": "A cat-eared figure operates a machine in the desert",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴护目镜的猫耳人物在沙漠中俯身操纵机械装置。 原生视频规格为 8 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A cat-eared figure wearing goggles leans over and operates a mechanical device in a desert. Native-video output: 8s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/don56palmer/status/2109657184984543466",
    "sourceLabel": "X 原帖 · @don56palmer",
    "author": "Oddar",
    "publishedAt": "2026-10-12T14:47:23.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109657184984543466.mp4",
    "posterUrl": "/posters/x/2109657184984543466.jpg",
    "duration": 8,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing Minimax H3 locally at a higher resolution (1536 x 863 for 8 seconds) on a 16 GB Nvidia card. Took about 19 minutes. Dialogue and sound effects generated in Elevenlabs,  music generated in Suno. Upscaled to 1920 x 1080 with the Topaz standalone app. #ai_media 🔊",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 21,
      "views": 417,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109652036681928944",
    "title": "黄昏废墟中的机甲交锋",
    "titleEn": "A mecha clash amid ruins at dusk",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "机甲在废墟中交锋，蓝色能量与黄昏天光交织。 原生视频规格为 29 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Mecha exchange attacks amid ruins with blue energy and dusk lighting. Native-video output: 29s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/eliliu911/status/2109652036681928944",
    "sourceLabel": "X 原帖 · @eliliu911",
    "author": "eliliu - ༺ 艾灵映画",
    "publishedAt": "2026-10-12T14:26:55.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109652036681928944.mp4",
    "posterUrl": "/posters/x/2109652036681928944.jpg",
    "duration": 29,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A 30-second mecha battle video created using Minimax H3; the prompt still needs adjusting.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 38,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109649840405914043",
    "title": "沙滩人物的蒙版移除对照",
    "titleEn": "Masked person removal on a beach",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "沙滩上的人物被绿色蒙版覆盖，随后出现移除后的背景。 原生视频规格为 15 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A person on a beach is covered by a green mask before the background is shown with the person removed. Native-video output: 15s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/3RDEYESTUDiOS/status/2109649840405914043",
    "sourceLabel": "X 原帖 · @3RDEYESTUDiOS",
    "author": "Nobu Fukuhara",
    "publishedAt": "2026-10-12T14:18:12.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109649840405914043.mp4",
    "posterUrl": "/posters/x/2109649840405914043.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "話題になってたMinimax H3の人物消去を試してみた。「自分のビデオに映ってた元カノ（元カレ？）を消したくて作った」というLoRA。オレが下手くそすぎるのか、思ったほどの正確性はなかった。とはいえこのレベルまでできるのは驚き。そのうち完璧にできるんだろうなと思うと絶望する。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 7,
      "views": 731,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109645562366788029",
    "title": "桌面上的粉绿发角色打喷嚏",
    "titleEn": "A pink-and-green-haired desktop character sneezes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉绿发动画人物在桌面图标旁打喷嚏并做出表情变化。 原生视频规格为 9 秒 · 1890×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-and-green-haired animated character sneezes beside desktop icons and changes expression. Native-video output: 9s · 1890×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2109645562366788029",
    "sourceLabel": "X 原帖 · @Mayz1169",
    "author": "Kiki",
    "publishedAt": "2026-10-12T14:01:12.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109645562366788029.mp4",
    "posterUrl": "/posters/x/2109645562366788029.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1890×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Mitsuri accidentally sneezed... I saw nothing lol 😭💗\n\nmade with MiniMax H3 on @renoiseai\n\nprompt below ⤵️",
    "engagement": {
      "replies": 5,
      "reposts": 6,
      "likes": 50,
      "views": 2322,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109642885373063671",
    "title": "电脑前的合成人物讲话",
    "titleEn": "A synthetic speaker in front of a computer",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "留短发与胡须的合成人物在电脑前对着镜头讲话。 原生视频规格为 10 秒 · 1076×1926。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A synthetic short-haired man with facial hair speaks toward the camera in front of a computer. Native-video output: 10s · 1076×1926. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/PozzerXYZ/status/2109642885373063671",
    "sourceLabel": "X 原帖 · @PozzerXYZ",
    "author": "Phill",
    "publishedAt": "2026-10-12T13:50:33.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109642885373063671.mp4",
    "posterUrl": "/posters/x/2109642885373063671.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1076×1926",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Esse vídeo foi gerado e postado no meu X sem eu sair do chat com meu Claude.\n\nPedi pro meu Claude criar o Theo, um influencer sintético (MiniMax H3 MAX Turbo), e postar direto pela plataforma que a gente tá construindo.\n\nSão mais de 34 modelos de vídeo pra escolher lá.\n\nQual modelo você gostaria de testar?",
    "engagement": {
      "replies": 4,
      "reposts": 3,
      "likes": 17,
      "views": 3472,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109642478517469483",
    "title": "南瓜与角色的万圣节拼贴",
    "titleEn": "A Halloween collage of pumpkins and a character",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴帽角色与南瓜、黑金装饰和万圣节图案组合成动态拼贴。 原生视频规格为 9 秒 · 1344×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A hat-wearing character, pumpkins and black-and-gold Halloween graphics form a moving collage. Native-video output: 9s · 1344×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Use the attached image as the main reference and turn it into a dynamic Halloween-style commercial motion-graphics video.\n\nPreserve the original character, typography, colors, design, and overall visual world from the reference image.\n\nMost important:\nUse the same exact main character throughout the entire video.\nDo not redesign, replace, or alter the character’s face at any point.\nKeep the face, hairstyle, hair color, glasses, headwear, clothing, body proportions, silhouette, and distinctive features consistent from beginning to end.\nIf the reference character is photorealistic, keep the character photorealistic.\nIf the reference character is illustrated, keep the same illustration style.\nIf the reference character is 3D, keep the same 3D style.\nNever change the visual medium.\n\nUse the 9 panels in the attached image sequentially, in order, as the progression of the video.\nMove through the panels naturally from one to the next, creating a smooth visual flow.\nConnect each panel with camera movement so the sequence feels continuous and cinematic rather than disconnected.\nUse camera work to transition naturally between scenes.\n\nTreat the typography and text in the attached image as physical 3D objects existing in the same space as the character.\nAnimate them with floating, rotating, 180-degree turns, bouncing, stretching, and moving from background to foreground.\nDo not add new text.\nKeep the original text accurate and readable.\n\nThe main character runs energetically through the typographic and graphic environment.\n\nAs the character passes near an object, one Halloween monster appears from its shadow, behind it, or from a gap and surprises the character, shouting:\n“Trick or Treat!”\n\nOnly one monster should appear per scene.\n\nThe monsters should look like cute, stylized mini clay figures with small bodies and slightly oversized heads.\nTheir faces should be very minimal and simple.\nMake the monster faces comedic and playful, not scary.\nThe eyes should be tiny symbolic marks such as small circles, triangles, X shapes, dots, or other very simple geometric expressions.\nKeep the monster face design minimal, graphic, and charming.\n\nUse different monster types such as:\njack-o’-lantern,\nwitch,\nFrankenstein,\nDracula,\nmummy,\nghost,\nand small devil.\n\nDo not repeat the same monster too often.\n\nAt first, the main character reacts with exaggerated surprise:\njumping,\nleaning back,\nrunning away,\nor quickly changing direction.\n\nAs more monsters appear, the character gradually gets used to them and starts having fun.\n\nEmotional progression:\nsurprised → runs away → gets used to it → starts having fun\n\nLater, the character may smile, wave at the monsters, run together with them, or playfully chase them.\n\nWhen a monster appears, add small Halloween motion-graphic effects such as flying bats, popping candy, stars, sparkles, smoke puffs, or tiny jack-o’-lantern elements.\nVary these effects from scene to scene.\n\nUse dynamic commercial-style camera work:\nfast horizontal tracking,\nsmooth chase-camera movement,\nlow angles,\npush-ins,\npull-backs,\nparallax,\nslight camera rotation,\nand foreground objects passing close to the lens.\n\nThe camera should create smooth transitions between the 9 panels and maintain a natural continuous flow.\n\nFor the ending, gather all the Halloween monsters together with the main character.\nThey move toward the camera and all shout:\n“Trick or Treat!”\n\nThen everyone strikes a cute, playful “rawr” pose toward the camera with both hands raised and fingers spread.\n\nOverall tone:\npop,\ncute,\ncomedic,\nslightly surreal,\nfast-paced,\nfun Halloween,\nnot too scary,\nand styled like a high-quality motion-graphics commercial.",
    "promptSourceUrl": "https://x.com/taya_mama_AI/status/2109642478517469483",
    "sourceUrl": "https://x.com/taya_mama_AI/status/2109642478517469483",
    "sourceLabel": "X 原帖 · @taya_mama_AI",
    "author": "た〜や｜スキマ時間×AI副業ママ",
    "publishedAt": "2026-10-12T13:48:56.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109642478517469483.mp4",
    "posterUrl": "/posters/x/2109642478517469483.jpg",
    "duration": 9,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "motion-graphics"
    ],
    "scenes": [
      "fantasy",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ハロウィン動画作ってみて〜\n\n🎥MiniMax H3 max turbo\n\n【動画プロンプト】\nUse the attached image as the main reference and turn it into a dynamic Halloween-style commercial motion-graphics video.\n\nPreserve the original character, typography, colors, design, and overall visual world from the reference image.\n\nMost important:\nUse the same exact main character throughout the entire video.\nDo not redesign, replace, or alter the character’s face at any point.\nKeep the face, hairstyle, hair color, glasses, headwear, clothing, body proportions, silhouette, and distinctive features consistent from beginning to end.\nIf the reference character is photorealistic, keep the character photorealistic.\nIf the reference character is illustrated, keep the same illustration style.\nIf the reference character is 3D, keep the same 3D style.\nNever change the visual medium.\n\nUse the 9 panels in the attached image sequentially, in order, as the progression of the video.\nMove through the panels naturally from one to the next, creating a smooth visual flow.\nConnect each panel with camera movement so the sequence feels continuous and cinematic rather than disconnected.\nUse camera work to transition naturally between scenes.\n\nTreat the typography and text in the attached image as physical 3D objects existing in the same space as the character.\nAnimate them with floating, rotating, 180-degree turns, bouncing, stretching, and moving from background to foreground.\nDo not add new text.\nKeep the original text accurate and readable.\n\nThe main character runs energetically through the typographic and graphic environment.\n\nAs the character passes near an object, one Halloween monster appears from its shadow, behind it, or from a gap and surprises the character, shouting:\n“Trick or Treat!”\n\nOnly one monster should appear per scene.\n\nThe monsters should look like cute, stylized mini clay figures with small bodies and slightly oversized heads.\nTheir faces should be very minimal and simple.\nMake the monster faces comedic and playful, not scary.\nThe eyes should be tiny symbolic marks such as small circles, triangles, X shapes, dots, or other very simple geometric expressions.\nKeep the monster face design minimal, graphic, and charming.\n\nUse different monster types such as:\njack-o’-lantern,\nwitch,\nFrankenstein,\nDracula,\nmummy,\nghost,\nand small devil.\n\nDo not repeat the same monster too often.\n\nAt first, the main character reacts with exaggerated surprise:\njumping,\nleaning back,\nrunning away,\nor quickly changing direction.\n\nAs more monsters appear, the character gradually gets used to them and starts having fun.\n\nEmotional progression:\nsurprised → runs away → gets used to it → starts having fun\n\nLater, the character may smile, wave at the monsters, run together with them, or playfully chase them.\n\nWhen a monster appears, add small Halloween motion-graphic effects such as flying bats, popping candy, stars, sparkles, smoke puffs, or tiny jack-o’-lantern elements.\nVary these effects from scene to scene.\n\nUse dynamic commercial-style camera work:\nfast horizontal tracking,\nsmooth chase-camera movement,\nlow angles,\npush-ins,\npull-backs,\nparallax,\nslight camera rotation,\nand foreground objects passing close to the lens.\n\nThe camera should create smooth transitions between the 9 panels and maintain a natural continuous flow.\n\nFor the ending, gather all the Halloween monsters together with the main character.\nThey move toward the camera and all shout:\n“Trick or Treat!”\n\nThen everyone strikes a cute, playful “rawr” pose toward the camera with both hands raised and fingers spread.\n\nOverall tone:\npop,\ncute,\ncomedic,\nslightly surreal,\nfast-paced,\nfun Halloween,\nnot too scary,\nand styled like a high-quality motion-graphics commercial.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 11,
      "views": 671,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109641709319827570",
    "title": "时装人物漫画的环绕视角",
    "titleEn": "Orbiting fashion characters inside comic panels",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "多格漫画中的时装人物和背景随观看视角转动。 原生视频规格为 12 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Fashion characters and their backgrounds rotate inside a multi-panel comic. Native-video output: 12s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/eternityspring/status/2109641709319827570",
    "sourceLabel": "X 原帖 · @eternityspring",
    "author": "烁皓",
    "publishedAt": "2026-10-12T13:45:53.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109641709319827570.mp4",
    "posterUrl": "/posters/x/2109641709319827570.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "零成本，又做了个交互式名媛连环画\n\n制作流程：\nQwen Image 2.1 画 6 格：同一张脸，6 套造型\nMiniMax-H3 + ORB360 LoRA：一张图 → 360° 环绕视频\nthree.js 拼成一页，拖哪格转哪格\n配乐：YuE2\n\n在线体验地址：\nhttps://eternityspring.github.io/ai-explorations/the-feed-comic/index.html",
    "engagement": {
      "replies": 8,
      "reposts": 15,
      "likes": 154,
      "views": 8488,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109638557078098021",
    "title": "夜间列车与城市人群",
    "titleEn": "Night trains and city crowds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "列车旁的人群镜头与黑白文字画面交替出现。 原生视频规格为 27 秒 · 1280×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Crowds beside a train alternate with monochrome typography shots. Native-video output: 27s · 1280×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Fuadik1512/status/2109638557078098021",
    "sourceLabel": "X 原帖 · @Fuadik1512",
    "author": "Fuad IK 🐈💙",
    "publishedAt": "2026-10-12T13:33:21.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109638557078098021.mp4",
    "posterUrl": "/posters/x/2109638557078098021.jpg",
    "duration": 27,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This city never slows down, but somewhere between the last train and the first laugh, we finally do, every night.\n\nMade with MiniMax H3 via @itsPolloAI @polloai_creator MCP\n\n#PolloAI #MiniMax #AIVideo #Jakarta",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 49,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109637283045646827",
    "title": "阳光穿过厨房与客厅",
    "titleEn": "Sunlight passes through a kitchen and living room",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头穿过有植物与餐具的室内空间，窗边出现人物。 原生视频规格为 56 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The view moves through an interior with plants and dishes while a figure appears beside a window. Native-video output: 56s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Alone1Moon/status/2109637283045646827",
    "sourceLabel": "X 原帖 · @Alone1Moon",
    "author": "Alone1M",
    "publishedAt": "2026-10-12T13:28:18.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109637283045646827.mp4",
    "posterUrl": "/posters/x/2109637283045646827.jpg",
    "duration": 56,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "なんとか @tapehead_Lab さんのMVがあまりに素敵だったので、同じ方法をローカルPCのMiniMax H3のみで再現できないかと冒頭部分の挑戦。まだ完璧には程遠く。中距離あたりは、どうしても、形状が曖昧なので、AIは得意じゃないよなあと。色々ごまかしてますが、失敗してます。疲れたので、一旦放出～。",
    "engagement": {
      "replies": 2,
      "reposts": 5,
      "likes": 50,
      "views": 2151,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109633034706198788",
    "title": "悟空来到动物园猴笼前",
    "titleEn": "Wukong visits a monkey enclosure",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "悟空站在动物园的笼子旁，与里面的猴子交流。 原生视频规格为 43 秒 · 941×1672。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Wukong stands beside a zoo enclosure and interacts with a monkey inside. Native-video output: 43s · 941×1672. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/C4ssp/status/2109633034706198788",
    "sourceLabel": "X 原帖 · @C4ssp",
    "author": "三字神经",
    "publishedAt": "2026-10-12T13:11:25.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109633034706198788.mp4",
    "posterUrl": "/posters/x/2109633034706198788.jpg",
    "duration": 43,
    "aspectRatio": "portrait",
    "resolution": "941×1672",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "悟空来救猴，猴子却先问：出去以后吃什么？下雨、生病又怎么办？\n《西游神经记》｜动物园救猴\n自由和安稳，你会怎么选？关注「三字神经」，接着看更多西游胡闹。\n画面使用 MiniMax H3 制作。@Hailuo_AI\n原创AI虚构喜剧。#MiniMaxH3 #西游记",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 71,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109632206562070695",
    "title": "白裙动画人物的环绕转身",
    "titleEn": "An animated character in a white dress turns around",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长发动画人物在白色背景中转身，展示正面与背面视角。 原生视频规格为 3 秒 · 1024×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A long-haired animated character turns against a white background, showing front and back views. Native-video output: 3s · 1024×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Ada_Lovelace49/status/2109632206562070695",
    "sourceLabel": "X 原帖 · @Ada_Lovelace49",
    "author": "さんぐらす❀",
    "publishedAt": "2026-10-12T13:08:07.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109632206562070695.mp4",
    "posterUrl": "/posters/x/2109632206562070695.jpg",
    "duration": 3,
    "aspectRatio": "portrait",
    "resolution": "1024×1344",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "これはMiniMax-H3-360-Orbitでつくった\nくるくるリーンちゃん\nリンク先でくるくる出来ます\n立体視も出来るよ！\nhttps://claude.ai/artifact/7FLkJepzhugYYwcENqyXfe",
    "engagement": {
      "replies": 0,
      "reposts": 7,
      "likes": 212,
      "views": 6030,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109626865170501947",
    "title": "雨夜人物与品牌广告画面",
    "titleEn": "A rainy-night character in a branded advertisement",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "雨夜中的人物镜头与声音制作界面、品牌字样交替。 原生视频规格为 38 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Shots of a character in a rainy night alternate with a voice-production interface and brand lettering. Native-video output: 38s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/jerrod_lew/status/2109626865170501947",
    "sourceLabel": "X 原帖 · @jerrod_lew",
    "author": "Jerrod Lew",
    "publishedAt": "2026-10-12T12:46:54.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109626865170501947.mp4",
    "posterUrl": "/posters/x/2109626865170501947.jpg",
    "duration": 38,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "One image to a spec ad.\nAll with Magnific One.\n\nThe new Magnific One draft mode lets you generate bulk images from a reference image.\n\nUse all these images with MiniMax H3 MAX for fast video generations.\n\nVoice over and music with Elevenlabs.\n\nEdit it together, and done!",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 17,
      "views": 992,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109621776946729214",
    "title": "雨夜霓虹城市的四格乐曲画面",
    "titleEn": "Four-panel music visuals in a rainy neon city",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "四格画面展示霓虹雨夜中的城市、人物与乐器。 原生视频规格为 32 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Four panels show a neon-lit rainy city, characters and musical instruments. Native-video output: 32s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/NaoyaCreates/status/2109621776946729214",
    "sourceLabel": "X 原帖 · @NaoyaCreates",
    "author": "Naoya",
    "publishedAt": "2026-10-12T12:26:41.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109621776946729214.mp4",
    "posterUrl": "/posters/x/2109621776946729214.jpg",
    "duration": 32,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "city",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made entirely with MiniMax H3, rendered on a Google Colab G4 GPU (RTX PRO 6000 Blackwell) 🎬\n\n\"Wires Remember Rain\": a full 4K MV, 37 H3 image-to-video cuts of a rainy neon city, with H3 writing the lyrics into each scene.\n\nHow I made it 👇\n#AIvideo #AIMusic #MusicVideo",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 249,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109621720826921103",
    "title": "雪地中的银发动画角色",
    "titleEn": "A silver-haired animated character in the snow",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "戴冬帽的银发动画人物在雪地中舞动，结尾展示装饰字样。 原生视频规格为 18 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired animated character wearing a winter hat moves in snowy scenery before decorative lettering ends the clip. Native-video output: 18s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rororoAIrororo/status/2109621720826921103",
    "sourceLabel": "X 原帖 · @rororoAIrororo",
    "author": "球体",
    "publishedAt": "2026-10-12T12:26:27.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109621720826921103.mp4",
    "posterUrl": "/posters/x/2109621720826921103.jpg",
    "duration": 18,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ダンスー！！\n編集楽しすぎていて👾👾\nついつい楽しくてもりもりにしちゃうんだけど見づらいかな？\n\n#KusArt  MiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 6,
      "likes": 31,
      "views": 624,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109618947846721725",
    "title": "雪林里的人物与熊伙伴",
    "titleEn": "A figure and a bear companion in a snowy forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "持杖人物与熊站在雪林里，一条发光鱼在旁边浮动。 原生视频规格为 15 秒 · 714×1286。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A staff-bearing figure and a bear stand in a snowy forest with a glowing fish floating nearby. Native-video output: 15s · 714×1286. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wraithworksprod/status/2109618947846721725",
    "sourceLabel": "X 原帖 · @wraithworksprod",
    "author": "Wraith",
    "publishedAt": "2026-10-12T12:15:26.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109618947846721725.mp4",
    "posterUrl": "/posters/x/2109618947846721725.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "714×1286",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "nature",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "A wild encounter in the forest 🐻🌲 A bear druid, her companion, and a flying fish. Because why not? 😂\n\nGenerated locally with MiniMax H3 in ComfyUI. Still some AI jank, but the results are getting interesting.\n\n#ComfyUI #MiniMaxH3 #AIVideo #GenerativeAI",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 98,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109617132854280512",
    "title": "街道上的两只卡通小恐龙",
    "titleEn": "Two small cartoon dinosaurs on a street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两只穿衣服的卡通恐龙在街道上面对面，绿色角色鼓起脸颊。 原生视频规格为 15 秒 · 1456×2544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two clothed cartoon dinosaurs face each other on a street while the green character puffs its cheeks. Native-video output: 15s · 1456×2544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/neverdiedino/status/2109617132854280512",
    "sourceLabel": "X 原帖 · @neverdiedino",
    "author": "디노몬",
    "publishedAt": "2026-10-12T12:08:13.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109617132854280512.mp4",
    "posterUrl": "/posters/x/2109617132854280512.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1456×2544",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🦖 DINOMON 💚 ROSIE 🦕\n\nEP.06 | Don’t Look Away 😟\n\n네 시선이 다른 곳에 머무는 몇 초가,\n왜 이렇게 길게 느껴졌을까.\n\nYour eyes left me for a moment,\nbut the feeling stayed much longer. 💌\n\n.\n.\n\n@zcre_ai 지크 플랫폼 minimax H3 를 이용해서 만든 영상입니다.\n#ZCRE #ZCRE_Al #zcrecrew",
    "engagement": {
      "replies": 20,
      "reposts": 11,
      "likes": 21,
      "views": 992,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109615079024935004",
    "title": "魔王遇上车站检票闸机",
    "titleEn": "A dark lord meets a station ticket gate",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "奇幻人物出现在车站时钟与检票闸机旁，尝试通过入口。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Fantasy characters appear beside a station clock and ticket gates as they try to pass the entrance. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/maru_999_/status/2109615079024935004",
    "sourceLabel": "X 原帖 · @maru_999_",
    "author": "Maru999",
    "publishedAt": "2026-10-12T12:00:04.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109615079024935004.mp4",
    "posterUrl": "/posters/x/2109615079024935004.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ICカードがないと、魔王でも通れない。\nNo IC card? No entry. Even for the Dark Lord.\n\n生成: Kairos（MiniMax H3）／曲はSunoで作りました🎵",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 48,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109614336398553140",
    "title": "夜街上的手机与应用界面",
    "titleEn": "A phone and an application on a nighttime street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "男子在夜间街道上走向镜头并举起手机展示应用界面。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man walks toward the camera on a nighttime street and holds up a phone showing an application. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/armanmikoyann/status/2109614336398553140",
    "sourceLabel": "X 原帖 · @armanmikoyann",
    "author": "Arman Mikoyan",
    "publishedAt": "2026-10-12T11:57:07.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109614336398553140.mp4",
    "posterUrl": "/posters/x/2109614336398553140.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Generated this video using MiniMax H3 on cloud-hosted infrastructure powered by an NVIDIA L4 GPU.\n\nGeneration time: ~30 min\nGPU compute cost: ~$0.08\nModel access: Free\n\nThe result is far from being perfect, but experimenting with AI video generation at this cost is impressive.",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 120,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109614219532664887",
    "title": "白色背景上的兔耳动画人物",
    "titleEn": "A bunny-eared animated figure against white",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "兔耳动画人物在白色背景中抬起双手并改变姿势。 原生视频规格为 15 秒 · 480×864。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A bunny-eared animated figure raises her hands and changes poses against a white background. Native-video output: 15s · 480×864. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/oti_fan_sub/status/2109614219532664887",
    "sourceLabel": "X 原帖 · @oti_fan_sub",
    "author": "oti_fan_sub",
    "publishedAt": "2026-10-12T11:56:39.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109614219532664887.mp4",
    "posterUrl": "/posters/x/2109614219532664887.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "480×864",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "こんばんは☺️\n\n今日は少し時間があったので、バニーを動かしてみました🐰💕\n\nBioに書いてるわりに、センシっぽいの全然出してないな〜って🤭\n\n今回はMiniMax H3で動画の長さをお試し。\nほんとはウサギさんみたいに、ぴょんぴょん跳ねてほしかったんだけどね🤣\n\nまたいろいろ作ってみるね✨\n#Mel_OC",
    "engagement": {
      "replies": 0,
      "reposts": 27,
      "likes": 57,
      "views": 435,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109613946806362260",
    "title": "夕阳城堡前的持剑人物",
    "titleEn": "A sword-bearing figure before a castle at sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物站在墓碑与远处城堡之间，夕阳照亮城市。 原生视频规格为 94 秒 · 1252×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure stands between a gravestone and a distant castle while sunset lights the city. Native-video output: 94s · 1252×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/rizdel00/status/2109613946806362260",
    "sourceLabel": "X 原帖 · @rizdel00",
    "author": "Rizdel",
    "publishedAt": "2026-10-12T11:55:34.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109613946806362260.mp4",
    "posterUrl": "/posters/x/2109613946806362260.jpg",
    "duration": 94,
    "aspectRatio": "landscape",
    "resolution": "1252×720",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fantasy",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今回は３０秒位の動画にする予定だったが、chatGPTに相談して試行錯誤した結果、約１分３０秒という長編になってしまった。Minimax H3ローカル生成。#AI動画 #MinimaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 97,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109612986570809807",
    "title": "湖畔夕阳下的舞蹈",
    "titleEn": "A dance beside a lake at sunset",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿浅色裙装的人物在湖边抬起手臂，随动作转身。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure in a light-colored dress raises her arms and turns beside a lake at sunset. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/pppoeadsl/status/2109612986570809807",
    "sourceLabel": "X 原帖 · @pppoeadsl",
    "author": "云天｜第二人生实验室",
    "publishedAt": "2026-10-12T11:51:45.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109612986570809807.mp4",
    "posterUrl": "/posters/x/2109612986570809807.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "今天继续踩坑，\n\n最大的问题是生成用minimax h3太贵了，这么一条差不多10块钱；\n\n兄弟们有没有比较便宜的方案啊，真诚求助：）",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 117,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109612821466025999",
    "title": "纸艺小鸟的九格拼贴",
    "titleEn": "A nine-panel collage of a paper bird",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "圆润小鸟在纸艺风格九格画面中出现，镜头放大再回到拼贴。 原生视频规格为 11 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A rounded bird appears in a paper-style nine-panel composition as the view zooms in and returns to the collage. Native-video output: 11s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/WorkChiharu/status/2109612821466025999",
    "sourceLabel": "X 原帖 · @WorkChiharu",
    "author": "ちはる｜AIアニメ × もちぴ｜Canva ⭐",
    "publishedAt": "2026-10-12T11:51:06.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109612821466025999.mp4",
    "posterUrl": "/posters/x/2109612821466025999.jpg",
    "duration": 11,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "もちぴのペーパーコラージュMVが完成しました🐥💕\n\n今回の制作の流れはこんな感じ✨\n\n① たーやさん@taya_mama_AI さんのプロンプトを使って、9-in-1画像を生成🎨\n\n② MiniMax H3で、9コマの画像を動きのあるMVに🎬\n\n③ 最後はOpus5.5で、カメラワークや切り替え、BGMを調整🎵\n\n1枚の画像から、こんなにかわいいMVができるなんて😍たーやさん、素敵なプロンプトをありがとうございます！",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 5,
      "views": 264,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109606961465757984",
    "title": "雪山灰模与滑雪画面的对照",
    "titleEn": "A snowy mountain blockout compared with skiing footage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "分屏将雪山灰模与更写实的滑雪人物画面对照。 原生视频规格为 24 秒 · 1280×1600。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Split screens compare a snowy mountain blockout with more realistic footage of skiers. Native-video output: 24s · 1280×1600. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/takamasa045/status/2109606961465757984",
    "sourceLabel": "X 原帖 · @takamasa045",
    "author": "伊藤貴將（いとぱん）｜AIエージェント×創作開発",
    "publishedAt": "2026-10-12T11:27:48.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109606961465757984.mp4",
    "posterUrl": "/posters/x/2109606961465757984.jpg",
    "duration": 24,
    "aspectRatio": "portrait",
    "resolution": "1280×1600",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "栂池から天狗原へ。\nBlenderで作った冬山を、PixVerseでより実写に近いバックカントリー映像にしてみました🏔️❄️\n\nCodexからPixVerse pluginのCLIを使い、MiniMax H3で6秒×4カットを生成。\n実際に滑ったときの雰囲気も思い出しながら、24秒の映像につなげています。\n\n地形や構図をBlenderで作って、そこからAIで映像にしていく。こういう使い方も面白い！⛷️🏂\n\n※現地撮影ではなく、CG・AIによる映像です。\n@PixVerse @PixVerse_JP \n#PixVerseCPP",
    "engagement": {
      "replies": 2,
      "reposts": 4,
      "likes": 21,
      "views": 705,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109600288420331538",
    "title": "操场与窗边的人物手势",
    "titleEn": "Character gestures on a track and beside a window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物在操场和窗边举手，镜头在近景之间切换。 原生视频规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character raises her hands on a sports track and beside a window as the footage cuts between close views. Native-video output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/reiQjo/status/2109600288420331538",
    "sourceLabel": "X 原帖 · @reiQjo",
    "author": "九条 玲",
    "publishedAt": "2026-10-12T11:01:17.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109600288420331538.mp4",
    "posterUrl": "/posters/x/2109600288420331538.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "はしるよ / ハートの\n\n※AI生成動画（MiniMax H3）",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 16,
      "views": 558,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109593643052892624",
    "title": "展台前的虚构合影互动",
    "titleEn": "A fictional photo interaction beside an event booth",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "AI改编的人物在展台前靠近并摆出合影姿势。 原生视频规格为 16 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "AI-altered characters lean toward each other and pose for a photo beside an event booth. Native-video output: 16s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MindfulReturn/status/2109593643052892624",
    "sourceLabel": "X 原帖 · @MindfulReturn",
    "author": "MindfulReturn 身心修复局",
    "publishedAt": "2026-10-12T10:34:53.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109593643052892624.mp4",
    "posterUrl": "/posters/x/2109593643052892624.jpg",
    "duration": 16,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "卧槽这个漫展亲亲也太敢了，我直接用本机流水线复刻了一版把男主换成 Elon Musk。\n\n做法很笨也很稳：先把原片拆成关键帧（坐定自拍→近吻→拉手凑近→贴脸合影），用 Gemini 网页 Nano Banana 把男主换成马斯克并锁住女 cos 和展台背景；因为 H3 一次只能出约 5 秒，就按「首尾帧驱动」一段段跑 MiniMax-H3（本机 ComfyUI，零 API），再 ffmpeg 拼成完整竖屏。\n\n中间嘴对嘴被模型拒了，只能近吻过渡，但节奏和互动点基本齐了。整条链路：网页出图 + 本地 H3 fl2v，不烧云端视频额度。\n\n原帖：@xixikawaii",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 1,
      "views": 2243,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109589279223538085",
    "title": "花园中的浅色裙装人物",
    "titleEn": "A light-dressed figure in a flower garden",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿浅色裙装的人物走过花园，结尾出现品牌字样。 原生视频规格为 17 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure in a light-colored dress walks through a flower garden before brand lettering ends the clip. Native-video output: 17s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ModellixAI/status/2109589279223538085",
    "sourceLabel": "X 原帖 · @ModellixAI",
    "author": "Modellix",
    "publishedAt": "2026-10-12T10:17:33.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109589279223538085.mp4",
    "posterUrl": "/posters/x/2109589279223538085.jpg",
    "duration": 17,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Pastel summer dream in K-pop cinematic aesthetic ✨\nA graceful idol wanders through glowing gardens.\n\nMiniMax H3 powers this clip:\n• Consistent character through seamless scene shifts\n• Natural movement of clothing and hair\n• Smooth cinematic transitions\n\nTry MiniMax H3 on Modellix.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 117,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109576509690462555",
    "title": "预览与成片的上下分屏对照",
    "titleEn": "A blockout and finished footage in stacked panels",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下分屏对照无面人物的预览、深度图与动画成片；上方标注H3，下方标注Seedance。 原生视频规格为 41 秒 · 1080×1216。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Stacked panels compare faceless-character footage, a depth map and an animated result, explicitly labeling the upper preview as H3 and the lower result as Seedance. Native-video output: 41s · 1080×1216. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tapehead_Lab/status/2109576509690462555",
    "sourceLabel": "X 原帖 · @tapehead_Lab",
    "author": "tapehead.lab",
    "publishedAt": "2026-10-12T09:26:48.000Z",
    "addedAt": "2026-10-13T01:14:55.340Z",
    "mediaUrl": "/media/x-2109576509690462555.mp4",
    "posterUrl": "/posters/x/2109576509690462555.jpg",
    "duration": 41,
    "aspectRatio": "portrait",
    "resolution": "1080×1216",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【実写プレビズで作る、実写アニメ合成】\n今回のMVではminimax h3で生成したプレビズ動画を使用しました。\n\n🎬ワークフロー\n①minimax h3-max-turboで動画生成（text to video）\n②生成したプレビズをDepth mapに変換\n③プレビズを元にseedanceで動画化\n\n💡この方法のいいところ\n・minimax h3-max-turboが安くてめちゃ生成が早い\n・めちゃ早いからプレビズでどんどんVコンを組める\n・プレビズが実写だから、自然な動きとカメラワークになる\n・Depth mapを使うことで実写背景も綺麗に再構成できる\n\nminimax h3-max-turboのT2V、めちゃいいですよ！\nプロンプトに対する出力が安定してるので、やりたいことを狙ってできます。\n（T2Vなのにキャラの服装ほぼ一貫してるのとか見るとわかりやすい）\n\n顔がのっぺらぼうの球体なのは、人間の顔にするとアニメキャラの輪郭に影響しちゃって可愛くなくなったからです。\n\nSeedanceを普通に回してたら前半でクレジットがほぼ無くなったので、この方法でだいぶ救われました🙌\nもしよければお試しを！",
    "engagement": {
      "replies": 42,
      "reposts": 406,
      "likes": 3149,
      "views": 263196,
      "snapshotAt": "2026-10-13T01:14:55.340Z"
    },
    "approvedAt": "2026-10-13T01:14:55.340Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109574845877064080",
    "title": "厨房中的两人交流",
    "titleEn": "Two people interact in a kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "女子在厨房中忙碌，男子来到窗边与她交流。 原生视频规格为 40 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman works in a kitchen as a man joins her beside the window. Native-video output: 40s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chujoakira/status/2109574845877064080",
    "sourceLabel": "X 原帖 · @chujoakira",
    "author": "あぼん",
    "publishedAt": "2026-10-12T09:20:11.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109574845877064080.mp4",
    "posterUrl": "/posters/x/2109574845877064080.jpg",
    "duration": 40,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3でドラマ作ろうと思ってとりあえずサンプル動画生成したんだけどなんか色々と惜しい\nまずセリフ喋らせてみたんだけど登場人物が喋ってる感があんまないんだよな…なんか韓国ドラマとかの吹き替えを聞いてるイメージに近い\nイントネーションもなんかまだ変だしなんかバグるしほんと草",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 68,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109572067238797765",
    "title": "月亮面孔的弓箭手",
    "titleEn": "An archer with a moon-like face",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿西装的月亮面孔人物拉开弓弦，随后镜头切到箭靶。 原生视频规格为 7 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A suited figure with a moon-like face draws a bow before the view cuts to a target. Native-video output: 7s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Kenmax79310/status/2109572067238797765",
    "sourceLabel": "X 原帖 · @Kenmax79310",
    "author": "ケンマックス🐻‍❄️",
    "publishedAt": "2026-10-12T09:09:09.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109572067238797765.mp4",
    "posterUrl": "/posters/x/2109572067238797765.jpg",
    "duration": 7,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#月の顔の男 #KenmaxPoetryWorks\n\n静寂を裂く一筋の矢\n狙い定めたその先へ\nOne arrow through the silence,\none moment of perfect focus.\n\n🏹✨ MiniMax H3 × @TopviewAIJP",
    "engagement": {
      "replies": 3,
      "reposts": 16,
      "likes": 34,
      "views": 555,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109571546977333568",
    "title": "像素冒险队伍穿越奇幻地域",
    "titleEn": "A pixel-art party travels through fantasy scenery",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "像素风冒险队伍在城镇、草地与奇幻地区之间切换。 原生视频规格为 44 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pixel-art adventuring party appears across town, grassland and fantasy landscapes. Native-video output: 44s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/maru_999_/status/2109571546977333568",
    "sourceLabel": "X 原帖 · @maru_999_",
    "author": "Maru999",
    "publishedAt": "2026-10-12T09:07:05.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109571546977333568.mp4",
    "posterUrl": "/posters/x/2109571546977333568.jpg",
    "duration": 44,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "retro"
    ],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "この道わが旅 #5 空と火山を越えて。ジパング、ラーミアの空、ネクロゴンドの火山の道。\n生成: Kairos（MiniMax H3）\n#この道わが旅 #AIアニメ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 74,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109567023428870393",
    "title": "灰模镜头与人物场景的对照",
    "titleEn": "A blockout camera shot compared with a character scene",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "分屏将简化室内灰模和人物场景对照，摄像机位置随画面变化。 原生视频规格为 12 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A split screen compares a simplified interior blockout with a character scene as the viewpoints change. Native-video output: 12s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HitotsumeKuro/status/2109567023428870393",
    "sourceLabel": "X 原帖 · @HitotsumeKuro",
    "author": "1つ目クロスケ",
    "publishedAt": "2026-10-12T08:49:06.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109567023428870393.mp4",
    "posterUrl": "/posters/x/2109567023428870393.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3で作ったおはよう動画のプリビズ比較。カメラは動かして人物が余り動かない場合はプリビズはシンプルでOK。演技はプロンプトで。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 65,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109564606465400929",
    "title": "神社前的狐耳动画人物",
    "titleEn": "A fox-eared animated figure before a shrine",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "狐耳动画人物在秋叶与神社前转身并做出手臂动作。 原生视频规格为 5 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fox-eared animated figure turns and gestures before a shrine amid autumn leaves. Native-video output: 5s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/MetantanPower/status/2109564606465400929",
    "sourceLabel": "X 原帖 · @MetantanPower",
    "author": "みーたんたん＠taitanpower",
    "publishedAt": "2026-10-12T08:39:30.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109564606465400929.mp4",
    "posterUrl": "/posters/x/2109564606465400929.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "stable-diffusion.cpp で minimax-H3 動画を作るためのGUI -- petiteGenAV -- を制作中です。 fl2va は出来たけど Ref2va はこれから。 現状、問題が多すぎる。\n\nこの添付動画は出力された avi を aviutl で mp4 に変換したもの。突然の大音量に注意。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 92,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109564088003280915",
    "title": "金色戒指的环绕近景",
    "titleEn": "An orbiting close-up of a gold ring",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "刻有图案的金色戒指缓慢转动，展示外侧与内圈。 原生视频规格为 5 秒 · 2112×1760。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A patterned gold ring rotates slowly, showing its outside and inner band. Native-video output: 5s · 2112×1760. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/junwatu/status/2109564088003280915",
    "sourceLabel": "X 原帖 · @junwatu",
    "author": "DΞV",
    "publishedAt": "2026-10-12T08:37:27.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109564088003280915.mp4",
    "posterUrl": "/posters/x/2109564088003280915.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "2112×1760",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This 360° orbital LoRA for MiniMax H3 is incredibly useful!\n\n👉 https://huggingface.co/pablodawson/MiniMax-H3-360-Orbit-LoRA",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 174,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109556947112652832",
    "title": "城市中的小人物与黄色伙伴",
    "titleEn": "A small character and a yellow companion in a city",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "小型动画人物和黄色伙伴在城市街道上互动。 原生视频规格为 10 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A small animated character and a yellow companion interact on a city street. Native-video output: 10s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/npaka123/status/2109556947112652832",
    "sourceLabel": "X 原帖 · @npaka123",
    "author": "布留川英一 / Hidekazu Furukawa",
    "publishedAt": "2026-10-12T08:09:04.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109556947112652832.mp4",
    "posterUrl": "/posters/x/2109556947112652832.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Google Colabでの MiniMax H3のセットアップが約14分で完了\nCodex から動画生成できることを確認\n\n・セットアップ : 約14秒\n・動画生成 (960 × 544 10秒) : 約6分32秒",
    "engagement": {
      "replies": 3,
      "reposts": 31,
      "likes": 352,
      "views": 22838,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109551360161284495",
    "title": "海边人物与鱼缸的能力片段",
    "titleEn": "Capability clips with seaside characters and a fishbowl",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短片在海边人物、鱼缸中的金鱼与拿放大镜的人之间切换。 原生视频规格为 84 秒 · 720×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The sequence cuts between seaside characters, a goldfish in a bowl and a person holding a magnifying glass. Native-video output: 84s · 720×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/picciosandmwtu/status/2109551360161284495",
    "sourceLabel": "X 原帖 · @picciosandmwtu",
    "author": "piccio sandra",
    "publishedAt": "2026-10-12T07:46:52.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109551360161284495.mp4",
    "posterUrl": "/posters/x/2109551360161284495.jpg",
    "duration": 84,
    "aspectRatio": "square",
    "resolution": "720×720",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Testing some Minimax H3 capabilities\n#AI #NeuralNetworks",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 46,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109545627373969828",
    "title": "街道上的人物与饮料互动",
    "titleEn": "A character interacts with a drink on a street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物在街道上举起饮料，镜头切到杯子和人物表情。 原生视频规格为 15 秒 · 1088×1952。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character holds up a drink on a street, with cuts to the cup and facial expressions. Native-video output: 15s · 1088×1952. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LBan2049/status/2109545627373969828",
    "sourceLabel": "X 原帖 · @LBan2049",
    "author": "LBan",
    "publishedAt": "2026-10-12T07:24:05.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109545627373969828.mp4",
    "posterUrl": "/posters/x/2109545627373969828.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1088×1952",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "感觉Minimax H3生成这类写实的视频效果要好一点。不过我尝试生成2K的分辨率，但是感觉效果很差。虽然分辨率到了2K，但是画面感觉噪点很大。不过画面还是蛮流畅的。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 141,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109545319667208703",
    "title": "蓝发动画人物与夕阳城市",
    "titleEn": "A blue-haired animated character and a sunset city",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "蓝发动画人物站在城市夕阳前，画面叠有标题文字。 原生视频规格为 30 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blue-haired animated character appears before a city at sunset with title text overlaid. Native-video output: 30s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/koheii/status/2109545319667208703",
    "sourceLabel": "X 原帖 · @koheii",
    "author": "コウヘイ｜AIラジオ局長",
    "publishedAt": "2026-10-12T07:22:52.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109545319667208703.mp4",
    "posterUrl": "/posters/x/2109545319667208703.jpg",
    "duration": 30,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "プロンプトは変えてないんだけど、\nMiniMax H3は中々意欲的な作風になるね。\n予想外の展開、結構アリかも✨\n#AISAスペース\n#AISARadioALPS",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 17,
      "views": 406,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109545270606488063",
    "title": "花园里缓缓绽放的玫瑰",
    "titleEn": "A rose slowly blooms in a garden",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉橙色玫瑰逐渐展开花瓣，背景是绿色植物。 原生视频规格为 10 秒 · 576×1024。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A pink-and-orange rose gradually opens its petals against green foliage. Native-video output: 10s · 576×1024. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yoradino/status/2109545270606488063",
    "sourceLabel": "X 原帖 · @yoradino",
    "author": "YOU RA",
    "publishedAt": "2026-10-12T07:22:40.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109545270606488063.mp4",
    "posterUrl": "/posters/x/2109545270606488063.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "576×1024",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "🌹 “When a rose blooms, the world becomes a little more beautiful.” ✨\n\n🌹 Created with Sogni AI\n\nModel: MINIMAX H3\nSogni:\nhttps://app.sogni.ai/?code=Yora\n\n@Sogni_Protocol \n#Sogni #AIArt #AICreators #GenAI\n#flower #aiart #AIillustration",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 121,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109539567216832527",
    "title": "室内人物与古老街道的动画故事",
    "titleEn": "An animated story between interiors and old streets",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动画故事在室内人物、传统街道与黑色片尾之间切换。 原生视频规格为 278 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated story cuts between characters indoors, traditional streets and a dark ending screen. Native-video output: 278s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/yama_kiyo/status/2109539567216832527",
    "sourceLabel": "X 原帖 · @yama_kiyo",
    "author": "山本清史（やまきよ）/ Kiyoshi Yamamoto(yamakiyo)",
    "publishedAt": "2026-10-12T07:00:00.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109539567216832527.mp4",
    "posterUrl": "/posters/x/2109539567216832527.jpg",
    "duration": 278,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "この作品の映像・静止画・声・音楽は、生成AIで制作しています。 \n\n映像：MiniMax H3 Max  @MiniMax_AI \n静止画：Luma Uni-1.1 Max @DreamLabLA \n声：Gemini 3.8 Flash TTS @GeminiApp \nBGM：ElevenLabs Music @ElevenLabs \n原案は、私とChatGPT の対話です。 \n台本の下書きと編集作業の一部に Claude @claudeai  を使いました。 \n\nハルは架空の人物です。",
    "engagement": {
      "replies": 1,
      "reposts": 3,
      "likes": 10,
      "views": 1193,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109537342214279230",
    "title": "吊扇下的人物换装",
    "titleEn": "A costume change beneath a ceiling fan",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "俯视镜头中，人物在吊扇下旋转，衣着由黑色变成白色。 原生视频规格为 15 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In an overhead shot, a figure spins beneath a ceiling fan as a black outfit changes to white. Native-video output: 15s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LBan2049/status/2109537342214279230",
    "sourceLabel": "X 原帖 · @LBan2049",
    "author": "LBan",
    "publishedAt": "2026-10-12T06:51:10.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109537342214279230.mp4",
    "posterUrl": "/posters/x/2109537342214279230.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 在 4090 显卡上跑了20分钟出来的版本，这啥啊。。。",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 74,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109519785860218956",
    "title": "海边小路上的回头微笑",
    "titleEn": "A smile over the shoulder on a seaside path",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿深色外套的人物在海边小路上回头，对着镜头微笑。 原生视频规格为 3 秒 · 1280×736。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure in a dark jacket turns toward the camera and smiles on a seaside path. Native-video output: 3s · 1280×736. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/so_tab/status/2109519785860218956",
    "sourceLabel": "X 原帖 · @so_tab",
    "author": "sotab　AIでなんかしてる人",
    "publishedAt": "2026-10-12T05:41:24.000Z",
    "addedAt": "2026-10-13T01:14:55.341Z",
    "mediaUrl": "/media/x-2109519785860218956.mp4",
    "posterUrl": "/posters/x/2109519785860218956.jpg",
    "duration": 3,
    "aspectRatio": "landscape",
    "resolution": "1280×736",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MINIMAX　H3にちょっと変更していてClode codeがサンプルとして出した動画が「DAISUKI！」だった件",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 65,
      "snapshotAt": "2026-10-13T01:14:55.341Z"
    },
    "approvedAt": "2026-10-13T01:14:55.341Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110154378409914677",
    "title": "少女与两位西装舞伴同步起舞",
    "titleEn": "A girl dances between two suited partners",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉色上衣的少女与两位长发西装舞伴并排跳舞，画面叠加歌词和节拍提示，结尾显示制作署名。 原生视频规格为 103 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl in a pink top dances between two long-haired partners in suits, with lyric and beat overlays followed by production credits. Native-video output: 103s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/_sagyoai/status/2110154378409914677",
    "sourceLabel": "X 原帖 · @_sagyoai",
    "author": "さ🥺",
    "publishedAt": "2026-10-13T23:43:03.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110154378409914677.mp4",
    "posterUrl": "/posters/x/2110154378409914677.jpg",
    "duration": 103,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Attentionダンスフル版\n振り付け、演出、歌詞…opus5.5\n動画…MINIMAX H3\n画像…codex\nキャラデザ…ワシ",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 113,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110152387604500593",
    "title": "霓虹街头与软件界面交错的音乐短片",
    "titleEn": "Neon streets and software interfaces set to music",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两位男子穿行于霓虹招牌和舞台式城市布景，软件订阅界面与日文歌词穿插在音乐画面中。 原生视频规格为 179 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two men appear among neon signs and theatrical city scenes while software subscription panels and Japanese lyrics overlay the music video. Native-video output: 179s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SeiraHisada/status/2110152387604500593",
    "sourceLabel": "X 原帖 · @SeiraHisada",
    "author": "久田誠良(Seira Hisada)",
    "publishedAt": "2026-10-13T23:35:08.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110152387604500593.mp4",
    "posterUrl": "/posters/x/2110152387604500593.jpg",
    "duration": 179,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【MV『全部契約してます Argon版』】\n\nArgonで動画素材の生成と編集をやり直してみました。\nAntigravityではまだArgonを選べないため、Geminiのチャットで生成したコードをColabにコピペして実行しています。\n編集の精度や仕上がりのクオリティはArgonの方が高い印象です。\n一方、動画素材生成の空間的スケール感やダイナミックさではAstraの方が優れているように感じます。\n\n【MV制作に使用したツール・技術スタック】\n\n・**音楽**：Suno（歌詞とスタイルプロンプトはAstraが生成）\n・**映像生成**：ComfyUI＋MiniMax H3（Qwen3-VL NVFP4量子化＋4-step LoRA＋sol-attn）／Google Colab G4\n・**音響同期**：faster-whisperで歌詞のタイミングを0.01秒単位で抽出し、150 BPMのビートとRMS音圧に合わせて映像を同期\n・**プロシージャル編集**：動画編集ソフトを使わず、Python（Pillowによる2倍スーパーサンプリング＋アンチエイリアス処理、PyAV、FFmpeg）のコードだけでUI・タイポグラフィ・クロマキー合成を実装\n・**構成・コード生成**：Gemini 4 Argon",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 81,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110142222649737386",
    "title": "音乐盒里的小幽灵旋转起舞",
    "titleEn": "A tiny ghost dances inside a music box",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "桌上的音乐盒打开后，一个白色小幽灵站在转轴上旋转，盒盖内的镜面映出它的身影。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A tabletop music box opens to reveal a tiny white ghost turning on the spindle, reflected in the mirror inside the lid. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "one tiny white ghost stands on the spindle and turns slowly",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2110142222649737386",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2110142222649737386",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-13T22:54:45.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110142222649737386.mp4",
    "posterUrl": "/posters/x/2110142222649737386.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Opened the music box. They recast the dancer.\n\nPrompt: \"one tiny white ghost stands on the spindle and turns slowly\"\n\nDuration: 15s. Camera: locked, dresser height. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe box is told to play no tune. A music box in a prompt will otherwise write its own.",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 4,
      "views": 279,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110137930153160787",
    "title": "双画面对照少女的快速舞步",
    "titleEn": "A split-screen comparison of fast dance moves",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑裙动漫少女在粉紫色舞台上跳舞，左右画面并列呈现不同运动速度和身体形态的变化。 原生视频规格为 10 秒 · 1088×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated girl in a black dress dances on a pink-purple stage, with side-by-side views showing different motion speeds and changes in body shape. Native-video output: 10s · 1088×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Hemi3AI/status/2110137930153160787",
    "sourceLabel": "X 原帖 · @Hemi3AI",
    "author": "ヘミ",
    "publishedAt": "2026-10-13T22:37:41.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110137930153160787.mp4",
    "posterUrl": "/posters/x/2110137930153160787.jpg",
    "duration": 10,
    "aspectRatio": "portrait",
    "resolution": "1088×1920",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "AI生成の話\n\nMiniMax H3で画像生成していて\nしかもステップ数が少ないとか\nいろいろあって\n早く動かすと何でも形が溶けて\nぼやぼやになってひどいことになるんです\nいま、動きが早くても何とかならないのかな\nと、試してます\n\nボヤっとした動画と\nめっちゃ早くしても溶けない実験の過程\n早回しすぎるがｗ\n踊りは適当です\n#AIイラスト",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 5,
      "views": 278,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110136089805783534",
    "title": "海上少女与白昼星光的黑白歌词影像",
    "titleEn": "A monochrome lyric film of a girl at sea",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑白画面中的少女戴帽站在海边或船上，云层与海面构成背景，竖屏画面叠加歌词并以标题收尾。 原生视频规格为 105 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A girl in a hat appears by the sea or aboard a boat in monochrome, with clouds, water and lyric overlays ending on a title card. Native-video output: 105s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/tshikimi/status/2110136089805783534",
    "sourceLabel": "X 原帖 · @tshikimi",
    "author": "竹花 樒",
    "publishedAt": "2026-10-13T22:30:22.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110136089805783534.mp4",
    "posterUrl": "/posters/x/2110136089805783534.jpg",
    "duration": 105,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "リリックMV「Stars in Daylight（Ocean Eyes Re-mix）」\n\n学者として死んだ日、わたしは海の上にいた。\n見えなくても、昼の星はそこにある。\n\nMotion graphics: Claude Opus 5.5\nStills: OpenAI Codex\nPre-Vis: Blender\nVideo: MiniMax H3\nBeat: @Yasamallow666 「Ocean Eyes」\nMusic: #Suno",
    "engagement": {
      "replies": 1,
      "reposts": 1,
      "likes": 3,
      "views": 248,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110134579398840664",
    "title": "秋日林间骑士向精灵讲故事",
    "titleEn": "A knight tells an elf a story in an autumn forest",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "披甲骑士与金发精灵坐在秋日林间的倒木上，动物围在身边；精灵随后起身离开，骑士目送她远去。 原生视频规格为 15 秒 · 960×544。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "An armored knight and a blonde elf sit on a fallen log in an autumn forest surrounded by animals; the elf then gets up and leaves as the knight watches. Native-video output: 15s · 960×544. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "<Subject 1> is the young female elf shown on the character reference sheet in <Picture 1>, which shows her from the front, back, and three-quarter view plus close-ups of her face. Match her face exactly as on the sheet's close-ups: platinum blonde hair with a long braid, pointed ears, green eyes, a shiny green bikini top and miniskirt, long white gloves, thigh-high boots, and gold-and-sapphire jewelry. <Picture 2> shows the same elf for the overall look and lighting. One elf only, no dragon. The sheet layout, its grey background, and the multiple views are not part of the scene; show her once, in the forest. <Subject 2> is a handsome knight in his thirties with wavy brown hair, a short neat beard, and a kind smile, wearing polished silver plate armor over a deep blue tunic, his helmet resting on the ground beside him.\n80s_fantasy. A 1970s practical-effects fantasy film: soft analog photography, warm film grain, gentle halation, matte-painted environments, practical puppet animals. Gentle, deliberate movement.\n\nA single continuous take in a sunlit autumn forest clearing, golden leaves and mossy tree trunks close behind, soft warm light in hazy shafts. Medium shot at seated eye level, about two meters away, framing both from the knees up. <Subject 2> sits on a fallen mossy log with <Subject 1> perched on his knee, both facing slightly toward each other. Rabbits, a badger, small songbirds, and butterflies gather on the grass and leaves around them, listening. <Subject 2> finishes his tale with a proud grin, saying in a warm British accent, \"I opened the window, the sunlight hit her, and she exploded! That was the end of the vampire queen!\" <Subject 1> claps delightedly, laughing. A robin flutters down and lands on her shoulder; she glances at it and says in a light British accent, \"Even robin likes your story.\" She hops off his knee, says brightly, \"I really must be going! Ta ta!\", and runs off into the forest past the camera, the rabbits, badger, birds, and butterflies scampering and fluttering after her, as <Subject 2> watches her go, smiling.\n\nOne unbroken shot from start to finish. No cuts, no edits, no fades, no dissolves, no change of camera angle.\n\noverall_soundscape: Birdsong, rustling leaves, the knight's voice, the elf's laughter and clapping, a robin's chirp, her voice, light running footsteps and the patter of small animals. No music, no score, no background music, no singing.\nnon_diegetic_music: N/A",
    "promptSourceUrl": "https://x.com/ArtGainz/status/2110134579398840664",
    "sourceUrl": "https://x.com/ArtGainz/status/2110134579398840664",
    "sourceLabel": "X 原帖 · @ArtGainz",
    "author": "ArtGainz",
    "publishedAt": "2026-10-13T22:24:22.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110134579398840664.mp4",
    "posterUrl": "/posters/x/2110134579398840664.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [
      "retro"
    ],
    "scenes": [
      "dialogue",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"A knight, an elf, a vampire-queen story and a robin with opinions\n\n15 seconds, one unbroken shot with synced dialogue, rendered entirely on my own RTX 3080 (12 GB):\n\nMiniMax H3 (int8) via FreeVideo v0.4 in ComfyUI, about 6.5 min per clip and 8.4 GB VRAM at peak\nneph1's 1980s Fantasy LoRA for that retro practical-effects look\nA Flux 2 character sheet plus the original portrait to keep the elf's face consistent\n\n#MiniMaxH3 #ComfyUI #AIvideo #LocalAI #FantasyArt\n\nPrompt:\n<Subject 1> is the young female elf shown on the character reference sheet in <Picture 1>, which shows her from the front, back, and three-quarter view plus close-ups of her face. Match her face exactly as on the sheet's close-ups: platinum blonde hair with a long braid, pointed ears, green eyes, a shiny green bikini top and miniskirt, long white gloves, thigh-high boots, and gold-and-sapphire jewelry. <Picture 2> shows the same elf for the overall look and lighting. One elf only, no dragon. The sheet layout, its grey background, and the multiple views are not part of the scene; show her once, in the forest. <Subject 2> is a handsome knight in his thirties with wavy brown hair, a short neat beard, and a kind smile, wearing polished silver plate armor over a deep blue tunic, his helmet resting on the ground beside him.\n80s_fantasy. A 1970s practical-effects fantasy film: soft analog photography, warm film grain, gentle halation, matte-painted environments, practical puppet animals. Gentle, deliberate movement.\n\nA single continuous take in a sunlit autumn forest clearing, golden leaves and mossy tree trunks close behind, soft warm light in hazy shafts. Medium shot at seated eye level, about two meters away, framing both from the knees up. <Subject 2> sits on a fallen mossy log with <Subject 1> perched on his knee, both facing slightly toward each other. Rabbits, a badger, small songbirds, and butterflies gather on the grass and leaves around them, listening. <Subject 2> finishes his tale with a proud grin, saying in a warm British accent, \"I opened the window, the sunlight hit her, and she exploded! That was the end of the vampire queen!\" <Subject 1> claps delightedly, laughing. A robin flutters down and lands on her shoulder; she glances at it and says in a light British accent, \"Even robin likes your story.\" She hops off his knee, says brightly, \"I really must be going! Ta ta!\", and runs off into the forest past the camera, the rabbits, badger, birds, and butterflies scampering and fluttering after her, as <Subject 2> watches her go, smiling.\n\nOne unbroken shot from start to finish. No cuts, no edits, no fades, no dissolves, no change of camera angle.\n\noverall_soundscape: Birdsong, rustling leaves, the knight's voice, the elf's laughter and clapping, a robin's chirp, her voice, light running footsteps and the patter of small animals. No music, no score, no background music, no singing.\nnon_diegetic_music: N/A\"",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 48,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110132572886036538",
    "title": "老胶片街头打斗中的角色替换",
    "titleEn": "A character replacement in a grainy street fight",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "竖屏老胶片风格片段中，身穿西装的男子穿过拥挤街景并参与打斗，画面保留颗粒感和原场景运动。 原生视频规格为 29 秒 · 640×1120。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A suited man moves through a crowded street and takes part in a fight in a grainy vertical film clip, retaining the texture and movement of the surrounding footage. Native-video output: 29s · 640×1120. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/mexxmillion/status/2110132572886036538",
    "sourceLabel": "X 原帖 · @mexxmillion",
    "author": "Autumn Imaginations",
    "publishedAt": "2026-10-13T22:16:24.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110132572886036538.mp4",
    "posterUrl": "/posters/x/2110132572886036538.jpg",
    "duration": 29,
    "aspectRatio": "portrait",
    "resolution": "640×1120",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [
      "vertical"
    ],
    "scenes": [
      "combat",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3 test replacing a selective character on more difficult old grainy film footage. no post processing was done for integration.\nNo character swap lora used. rendered 0.7MP on a single 3090 agentic workflow to speed things up and QC. #minimaxH3 #comfyui #digitaldouble",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 39,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110126887468966175",
    "title": "明亮厨房里的人物参考效果",
    "titleEn": "Character reference results in a bright kitchen",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "固定机位拍摄明亮的白色厨房，女性角色进入画面并查看手机；原帖说明了人物参考素材与渲染耗时的差异。 原生视频规格为 10 秒 · 1920×1088。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A fixed view of a bright white kitchen reveals a woman checking her phone; the source discusses character references and their rendering-time differences. Native-video output: 10s · 1920×1088. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EndFolding79421/status/2110126887468966175",
    "sourceLabel": "X 原帖 · @EndFolding79421",
    "author": "FoldingEnd A.I.",
    "publishedAt": "2026-10-13T21:53:48.000Z",
    "addedAt": "2026-10-14T00:42:21.946Z",
    "mediaUrl": "/media/x-2110126887468966175.mp4",
    "posterUrl": "/posters/x/2110126887468966175.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1920×1088",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 — Render Time Test: Character Sheet vs. Seedance Video Reference\n\nLink to Original Test:  https://x.com/EndFolding79421/status/2106001483271418199\n\nThis time I'm testing the difference in rendering time between using a traditional image character sheet and a Seedance-generated video as the character reference in ComfyUI.\n\nOriginally, I was using an 8-second Seedance reference video, but the render went over 33 minutes. So I trimmed the reference down to just 4 seconds to see if that would help reduce the rendering time.\n\nI also turned OFF the Realism LoRA for the video reference test to see how H3 would perform without it ( it might look better with it on), compared to the character sheet test, where I had the Realism LoRA enabled at a strength of 0.6.\n\nFor the kitchen, I simply grabbed a random picture off the internet and used it as the location reference to help keep the room consistent throughout the scene.\n\nSo, the good and the bad...\n\nThe bad: Even trimmed down to 4 seconds, using a video reference still takes longer to render than using a character sheet.\n\nThe good: As I showed in a previous test, using a video reference can help when you're having trouble keeping your character consistent from shot to shot.\n\nSo if you're struggling with character consistency in H3, a Seedance video reference might be worth trying, especially if you don't mind waiting a little longer for your video to finish cooking!\n\nFor me, it's another useful option. Character sheets are faster, but video references may be worth the extra rendering time when consistency becomes an issue.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 4,
      "views": 382,
      "snapshotAt": "2026-10-14T00:42:21.946Z"
    },
    "approvedAt": "2026-10-14T00:42:21.946Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110119096176275689",
    "title": "抢夺烤红薯的动漫角色",
    "titleEn": "Animated characters fighting over a sweet potato",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫发少女抱着红薯与同伴站在室内，快速动作后切到黑发男子的严肃近景。 原生视频规格为 15 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple-haired girl holds a sweet potato beside her companions indoors, with rapid movement followed by a serious close-up of a dark-haired man. Native-video output: 15s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/TK2Works/status/2110119096176275689",
    "sourceLabel": "X 原帖 · @TK2Works",
    "author": "トクツー",
    "publishedAt": "2026-10-13T21:22:51.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110119096176275689.mp4",
    "posterUrl": "/posters/x/2110119096176275689.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "food"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIニケちゃん #NoMoreNormal\nさつまいも争奪戦\nMinimax H3\n---\ngoogle colabならA100使えるじゃん!20stepも行けるぞーっということでお試し出力。",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 9,
      "views": 345,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110114407829401618",
    "title": "紫金光线中的机械腕表",
    "titleEn": "A mechanical wristwatch under purple and gold light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "深色背景中的腕表被紫色光线照亮，镜头靠近表壳、指针和齿轮细节，最后逐渐变暗。 原生视频规格为 5 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Purple light illuminates a wristwatch against a dark background as the camera moves close to its case, hands and gears before fading to darkness. Native-video output: 5s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/2017_nonsense/status/2110114407829401618",
    "sourceLabel": "X 原帖 · @2017_nonsense",
    "author": "Rawrskis!",
    "publishedAt": "2026-10-13T21:04:13.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110114407829401618.mp4",
    "posterUrl": "/posters/x/2110114407829401618.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "I got Minimax h3 comfy template to make a 5sec video at the highest resolution I could set it. Took 1724.33 seconds or almost 29 minutesand my laptop is hot enought to cook on, but it did it! Now I need to run the smaller vids through an upscaler.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 34,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110110051784704468",
    "title": "少女沿坡道滑行后回头比出胜利手势",
    "titleEn": "A skater girl turns back with a victory sign",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫少女踩着滑板沿坡道飞驰，低角度镜头跟随轮子和滑板，最后她微笑着举起胜利手势。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated girl speeds down a slope on a skateboard, followed by low-angle views of the board and wheels before she smiles and raises a victory sign. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Dokko57932483/status/2110110051784704468",
    "sourceLabel": "X 原帖 · @Dokko57932483",
    "author": "Stan_Katayama",
    "publishedAt": "2026-10-13T20:46:54.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110110051784704468.mp4",
    "posterUrl": "/posters/x/2110110051784704468.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Still striving for the best possible animation with MiniMax H3 and version (I can't remember, 8 or 9) of my animation-based LoRa model.\n\nThe results are getting better and better. 😊",
    "engagement": {
      "replies": 3,
      "reposts": 0,
      "likes": 11,
      "views": 393,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110104717640466890",
    "title": "古典车厢里的贵族交谈",
    "titleEn": "An aristocratic conversation in a vintage carriage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "银发男子与白发女子相对坐在装饰华丽的车厢里交谈，画面从双人构图转为男子的近景。 原生视频规格为 10 秒 · 544×384。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A silver-haired man and a white-haired woman converse inside an ornate carriage as the framing shifts from both characters to a close-up of the man. Native-video output: 10s · 544×384. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/EmpJellok/status/2110104717640466890",
    "sourceLabel": "X 原帖 · @EmpJellok",
    "author": "JellokEmp",
    "publishedAt": "2026-10-13T20:25:43.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110104717640466890.mp4",
    "posterUrl": "/posters/x/2110104717640466890.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "544×384",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "@GrumpyWhiskers_  Forgive the quality my gpu can only handle so much and 10 seconds at a time. I used Minimax H3 and was trying to see what it would look like with a similar prompt. First one is your prompt and 2nd is 2.5d settings. Haha good experiment thanks for the share.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 20,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110101791924953496",
    "title": "低多边形游戏角色的过场片段集",
    "titleEn": "Low-poly game characters in cutscene sequences",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "游戏风格角色在户外对话，金发人物出现近景，随后切换到带有爆炸和闪电的奇幻场景。 原生视频规格为 41 秒 · 950×720。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Game-like characters converse outdoors, followed by a blonde character close-up and a fantasy scene filled with explosions and lightning. Native-video output: 41s · 950×720. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/megaaziib/status/2110101791924953496",
    "sourceLabel": "X 原帖 · @megaaziib",
    "author": "Megaaziib",
    "publishedAt": "2026-10-13T20:14:05.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110101791924953496.mp4",
    "posterUrl": "/posters/x/2110101791924953496.jpg",
    "duration": 41,
    "aspectRatio": "landscape",
    "resolution": "950×720",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "retro"
    ],
    "scenes": [
      "dialogue",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "i trained minimax h3 lora for mimicking playstation 2 graphic cutscene. download on civitai, link below.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 43,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110101510747230334",
    "title": "猫耳少女的表情与肢体动作",
    "titleEn": "A cat-eared girl changes expression and moves",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "棕发猫耳少女在白色背景前改变表情并抬手做动作，原帖将原始动画与自动驱动的三维角色并列对照。 原生视频规格为 15 秒 · 960×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A brown-haired cat-eared girl changes expression and gestures against a white background; the source compares the original animation with an automatically driven 3D character. Native-video output: 15s · 960×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/IsekaiBardGuild/status/2110101510747230334",
    "sourceLabel": "X 原帖 · @IsekaiBardGuild",
    "author": "AI Bard Guild",
    "publishedAt": "2026-10-13T20:12:58.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110101510747230334.mp4",
    "posterUrl": "/posters/x/2110101510747230334.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "960×480",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [
      "cg-3d"
    ],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Opus 5.5 + 動画を使った3dキャラのモーション作成、調子にのって表情付けも試しました\n\n左が元動画(MiniMax H3)で、\n右が3dキャラに Opus 5.5 で表情をつけたものです\n\nそこそこできてる感はあるんですけど、\n表情は細かな違いで印象が結構かわるので\n簡単にはいかないですね〜\n\nそれでも、ほぼ自動で\nここまでできてるだけですごい気はしますが\n\nもうちょっと頑張ってみます",
    "engagement": {
      "replies": 0,
      "reposts": 2,
      "likes": 7,
      "views": 458,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110100639032242383",
    "title": "角饰少女领唱的霓虹舞会",
    "titleEn": "A horned singer leads a neon dance party",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "带有橙黄色角饰的黑发少女在绿色灯光中演唱，结尾与一群风格化角色出现在明亮的舞厅里。 原生视频规格为 250 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired girl with orange-yellow horns sings under green light before joining a group of stylized characters in a bright dance hall. Native-video output: 250s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/flutterwhat/status/2110100639032242383",
    "sourceLabel": "X 原帖 · @flutterwhat",
    "author": "Flutterwhat",
    "publishedAt": "2026-10-13T20:09:30.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110100639032242383.mp4",
    "posterUrl": "/posters/x/2110100639032242383.jpg",
    "duration": 250,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "dance",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Let Me Dance, Let Me Glisten (Second Suno Remix)\"\n\n| GPT Image 2.5 | Muse | Logic Pro + SunoV6-Wild |Minimax H3 via SlopCannon (@boneGPT) |\n\nReleased 4/12/2016, the original song featured heavy use of layered synthetic sounding vocal transformers. It was made for the \"Beforus\" fan album that was a calibration of the dancestor characters from homestuck, before going on to be released as a single and then eventually a full album of homestuck inspired songs. Non-AI version is the 5th track (link in discription), alum art is commissioned via @cosplaykinky (Longest mutual follower award btw) \n\nthis faithful suno adaptation with slop cannon's direction came out very nice. GPT Imagen 2.5 has a suspicious knack for MS Paint Adventures.",
    "engagement": {
      "replies": 1,
      "reposts": 2,
      "likes": 3,
      "views": 187,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110080319663517947",
    "title": "霓虹都市中的动画乐队",
    "titleEn": "An animated band in a neon metropolis",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "堆叠屏幕、夜间都市和霓虹舞台交错出现，动画歌手与乐手在赛博朋克风格的音乐片段中表演。 原生视频规格为 137 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Stacked screens, a city at night and neon stages alternate as animated singers and musicians perform in a cyberpunk-styled musical sequence. Native-video output: 137s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/CrypToroMx/status/2110080319663517947",
    "sourceLabel": "X 原帖 · @CrypToroMx",
    "author": "Toro",
    "publishedAt": "2026-10-13T18:48:46.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110080319663517947.mp4",
    "posterUrl": "/posters/x/2110080319663517947.jpg",
    "duration": 137,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "stylized"
    ],
    "scenes": [
      "city",
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Después de 2 semanas de generar 60 clips de animación con IA local usando Minimax H3 @Hailuo_AI , he logrado hacer este video musical estilo Edgerunners para mi canción.\n\nDisfruten.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110071401722061262",
    "title": "快餐桌边的虚构富豪争执",
    "titleEn": "A fictional billionaire argument at a fast-food table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "讽刺短片中，几位虚构的科技富豪形象围坐快餐桌旁，争抢食物并站起争执；这不是现实事件记录。 原生视频规格为 15 秒 · 832×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "In a satirical fictional scene, tech-billionaire likenesses gather around a fast-food table, compete over food and rise to argue; this is not footage of a real event. Native-video output: 15s · 832×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mei_Life_JP/status/2110071401722061262",
    "sourceLabel": "X 原帖 · @Mei_Life_JP",
    "author": "めい (Mei)",
    "publishedAt": "2026-10-13T18:13:20.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110071401722061262.mp4",
    "posterUrl": "/posters/x/2110071401722061262.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "832×1280",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "food",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Just trying to enjoy my Big Mac at McDonald's, but the billionaires at the next table are fighting over a Filet-O-Fish like literal toddlers. \n@elonmusk\n#SiliconValley #TechHumor #Mcdonalds #Minimax-H3\nmade by https://rhtv.runninghub.vip",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 27,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110069282637836398",
    "title": "纸上的墨线化作黑色缎带",
    "titleEn": "An ink line lifts off the paper as a black ribbon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "俯拍镜头中，纸上的细长黑线逐渐脱离纸面并延伸成缎带，墨水瓶与桌面保持固定。 原生视频规格为 15 秒 · 3456×1920。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "In a top-down view, a thin black line lifts from a sheet of paper and extends into a ribbon while the ink bottle and desk remain fixed. Native-video output: 15s · 3456×1920. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "the line peels up off the page as a thin black ribbon",
    "promptSourceUrl": "https://x.com/GlennHasABeard/status/2110069282637836398",
    "sourceUrl": "https://x.com/GlennHasABeard/status/2110069282637836398",
    "sourceLabel": "X 原帖 · @GlennHasABeard",
    "author": "Glenn Williams",
    "publishedAt": "2026-10-13T18:04:54.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110069282637836398.mp4",
    "posterUrl": "/posters/x/2110069282637836398.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3456×1920",
    "tags": [
      "MiniMax H3",
      "动画",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3, open weights, generated locally. Drew one line. It left.\n\nPrompt: \"the line peels up off the page as a thin black ribbon\"\n\nDuration: 15s. Camera: locked, top-down. Rendered locally in ComfyUI, then Topaz to 4K.\n\nThe ribbon exits the same way the quill went. One direction for the whole clip.",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 16,
      "views": 464,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110067249134596544",
    "title": "云海上方的霓虹悬浮都市",
    "titleEn": "A neon floating city above a sea of clouds",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "圆形悬浮城市立于云海上方，镜头环绕高楼、中央尖塔和底部机械结构，青色与紫色灯光贯穿建筑。 原生视频规格为 8 秒 · 768×768。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A circular floating city hangs above clouds as the camera circles towers, a central spire and mechanical structures beneath, illuminated by cyan and purple lights. Native-video output: 8s · 768×768. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "One continuous cinematic 8-second shot. Treat the supplied cyberpunk sky city as a coherent three-dimensional floating circular island above clouds. Camera performs exactly one smooth complete 360-degree horizontal clockwise orbit around the city, returning to the original front view. 0s original front view, 2s right side with the curved air terminal in profile, 4s rear view of the central three-pronged energy tower and reverse city skyline, 6s left side, 8s original front view. Constant wide camera distance, slightly elevated level angle, keep the entire city island visible. Strong natural parallax between nearby flying vehicles, city towers, dangling cables, underbody antigravity engine and distant floating islands. Preserve consistent tower geometry, city layout, black chrome and glass surfaces, cyan and magenta neon and purple-orange sunset lighting. The city stays stationary; clouds drift slowly, tiny flying vehicles cruise and engine lights pulse gently. Camera truly travels around the city through space, not flat image rotation or city spin. No cuts, dissolves, morphing, duplicate central towers or text. Audio: soft futuristic ambient synth, distant flying vehicles, low antigravity engine hum, no speech.",
    "promptSourceUrl": "https://x.com/eikoung/status/2110067249134596544",
    "sourceUrl": "https://x.com/eikoung/status/2110067249134596544",
    "sourceLabel": "X 原帖 · @eikoung",
    "author": "紅",
    "publishedAt": "2026-10-13T17:56:50.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110067249134596544.mp4",
    "posterUrl": "/posters/x/2110067249134596544.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "city",
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "賽博龐克天空城｜雲海上的霓虹未來\n8 秒 360° 環繞，浮空港、三叉能量塔與反重力引擎。\n\n生成工具：OpenAI ImageGen（原圖）；本機 ComfyUI + MiniMax H3（圖生影片，INT8 ConvRot 量化主模型，Turbo 8-step LoRA，RTX 3090）；FFmpeg（8 秒 MP4 編碼）。\n規格：768 × 768、24fps、8 秒，含環境音。\n單圖未見的城市背面由 AI 推測生成。\n\n完整生圖提示詞：\n原創賽博龐克天空城，正方形高細節電影概念藝術。中心是一座完整可見、懸浮在雲海上方的圓形城市島嶼，立體結構清晰，便於相機環繞。島嶼上高低錯落的黑鉻摩天大樓環繞一座三叉尖頂中央能量塔，青色與洋紅霓虹燈沿建築流動；下方可見巨大環形反重力引擎、懸垂電纜與分層機械底盤。右前方有一座弧形空中港，幾艘小型飛行車穿越街區；前景兩艘微型無人機形成景深，遠景其他浮空島與閃電雲層。夜幕與落日的紫橘色交界，濕潤金屬、玻璃、蒸汽、體積光，細節繁密但輪廓乾淨。主城居中，完整留在畫面內，略高於平視的三分之四視角，留出充足周圍空間，避免裁切；宏偉、神秘、有生命的未來城市，精緻數位繪畫與寫實建築材質融合。無可讀文字、無標誌、無水印，非拼貼。\n\n完整動畫提示詞：\nOne continuous cinematic 8-second shot. Treat the supplied cyberpunk sky city as a coherent three-dimensional floating circular island above clouds. Camera performs exactly one smooth complete 360-degree horizontal clockwise orbit around the city, returning to the original front view. 0s original front view, 2s right side with the curved air terminal in profile, 4s rear view of the central three-pronged energy tower and reverse city skyline, 6s left side, 8s original front view. Constant wide camera distance, slightly elevated level angle, keep the entire city island visible. Strong natural parallax between nearby flying vehicles, city towers, dangling cables, underbody antigravity engine and distant floating islands. Preserve consistent tower geometry, city layout, black chrome and glass surfaces, cyan and magenta neon and purple-orange sunset lighting. The city stays stationary; clouds drift slowly, tiny flying vehicles cruise and engine lights pulse gently. Camera truly travels around the city through space, not flat image rotation or city spin. No cuts, dissolves, morphing, duplicate central towers or text. Audio: soft futuristic ambient synth, distant flying vehicles, low antigravity engine hum, no speech.\n\n#AIArt #Cyberpunk #MiniMaxH3 #ComfyUI",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 376,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110057039150170507",
    "title": "铠甲剑士在火花与碎石间交锋",
    "titleEn": "Armored swordsmen clash amid sparks and debris",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名全身披甲的剑士近距离交锋，金属碰撞迸出火花，随后镜头拉开，碎石散落在双方之间。 原生视频规格为 12 秒 · 768×448。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two fully armored swordsmen clash at close range with sparks from their weapons before the view widens to show debris between them. Native-video output: 12s · 768×448. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/focalzero/status/2110057039150170507",
    "sourceLabel": "X 原帖 · @focalzero",
    "author": "FocalZero",
    "publishedAt": "2026-10-13T17:16:15.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110057039150170507.mp4",
    "posterUrl": "/posters/x/2110057039150170507.jpg",
    "duration": 12,
    "aspectRatio": "landscape",
    "resolution": "768×448",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "combat"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Test vidéo sur Minimax H3 turbo sur RTX 16gb. le local n'est pas du marketing, c'est la condition sine qua non de la propriété d'usage. Aujourd'hui si tu veux acheter une 5090 tu dois dépenser plus de 5k$+ et donner ta carte d'identité. Ils verrouillent l'entré du secteur.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 133,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110051278063808638",
    "title": "紫发少女在几何舞台上抬手起舞",
    "titleEn": "A purple-haired girl dances on a geometric stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "紫发动漫少女穿着黑紫色服装，在浅色几何舞台上抬手摆姿，画面随后切换到暖色粒子背景。 原生视频规格为 15 秒 · 1080×1888。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A purple-haired animated girl in a black-and-purple outfit gestures and poses on a pale geometric stage before the scene changes to a warm particle-filled background. Native-video output: 15s · 1080×1888. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sorastg524/status/2110051278063808638",
    "sourceLabel": "X 原帖 · @sorastg524",
    "author": "sorastg524",
    "publishedAt": "2026-10-13T16:53:22.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110051278063808638.mp4",
    "posterUrl": "/posters/x/2110051278063808638.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1888",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#AIart #MiniMaxH3MaxTurbo #Capcut\n病みカワイイ女子。 \n（MiniMax H3 Max Turbo 出力動画＋CapCutの画質向上（AIHD））",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 96,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110050205177610680",
    "title": "电梯前的花裙女子",
    "titleEn": "A woman in a floral dress outside an elevator",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "一位卷发女子穿着粉白花裙站在电梯前，面对镜头改变表情，最后露出微笑。 原生视频规格为 15 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A curly-haired woman in a pink-and-white floral dress stands by an elevator, changes expression toward the camera and finishes with a smile. Native-video output: 15s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ThordinRock1/status/2110050205177610680",
    "sourceLabel": "X 原帖 · @ThordinRock1",
    "author": "Thordin Rocker",
    "publishedAt": "2026-10-13T16:49:06.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110050205177610680.mp4",
    "posterUrl": "/posters/x/2110050205177610680.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Working on MiniMax H3 #India",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 32,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110037609548579228",
    "title": "绿色纸面具歌手的黑白节拍",
    "titleEn": "A green paper-mask singer in monochrome rhythm",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "灰白色画面中，戴绿色纸面具的人物唱歌，几何方格、数字与节拍式剪辑穿插出现，结尾显示作品标题。 原生视频规格为 60 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A figure wearing a green paper mask sings in a grey-and-white sequence, intercut with geometric grids, numbers and rhythmic edits before a final title card. Native-video output: 60s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/web3nomad/status/2110037609548579228",
    "sourceLabel": "X 原帖 · @web3nomad",
    "author": "web3nomad.eth | atypica.ai",
    "publishedAt": "2026-10-13T15:59:03.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110037609548579228.mp4",
    "posterUrl": "/posters/x/2110037609548579228.jpg",
    "duration": 60,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [
      "motion-graphics"
    ],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Lyrics: Claude → Song: ElevenLabs → Shots: GPT Image + MiniMax H3 → Edit: Claude writing Remotion, beat-synced to the song.\n\nI gave Claude one song idea and a green paper ghost mask.\n\nIt wrote the lyrics, generated the song, cast the crowd, directed 15 shots, and cut every frame to the beat.\n\nNo editor. No timeline. Just code.\n\nThis is BLANK CARD. 🟩",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 80,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110033726558425209",
    "title": "豪车展厅里被轻视的老人",
    "titleEn": "An overlooked elderly visitor at a luxury showroom",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "衣着朴素的老人走进豪车展厅，西装店员起初态度冷淡，随后在宴席与展厅场景中出现转折。 原生视频规格为 90 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A plainly dressed elderly man enters a luxury car showroom and meets a dismissive salesman, followed by a turn in events across banquet and showroom scenes. Native-video output: 90s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/SeiraHisada/status/2110033726558425209",
    "sourceLabel": "X 原帖 · @SeiraHisada",
    "author": "久田誠良(Seira Hisada)",
    "publishedAt": "2026-10-13T15:43:37.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110033726558425209.mp4",
    "posterUrl": "/posters/x/2110033726558425209.jpg",
    "duration": 90,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【動画『TikTokでよく見る王道ショートドラマの展開を変えてみた』】\n\n高級車ディーラーに現れた、みすぼらしい老人。\n追い返そうとした店長は、あることに気づき、態度を一変させる。\n果たして、老人の正体は……？\n\n■ 技術スタック\n映像：Google Colab G4＋ComfyUI＋MiniMax H3（INT8量子化・Turbo LoRA・6ステップ）\n音声：Irodori-TTS\n劇伴：Suno\n編集・字幕：Python＋Pillow＋NumPy＋PyAV＋FFmpeg",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 87,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110022695232704775",
    "title": "晴空下递出的橙味汽水",
    "titleEn": "An orange soda offered under a clear sky",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "白衬衫男子站在城市屋顶，向镜头展示黄色橙味汽水瓶，瓶身近景后回到他的微笑。 原生视频规格为 30 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man in a white shirt stands on a city rooftop and presents a yellow orange-soda bottle to the camera, with a bottle close-up followed by his smile. Native-video output: 30s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chujoakira/status/2110022695232704775",
    "sourceLabel": "X 原帖 · @chujoakira",
    "author": "あぼん",
    "publishedAt": "2026-10-13T14:59:47.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110022695232704775.mp4",
    "posterUrl": "/posters/x/2110022695232704775.jpg",
    "duration": 30,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial"
    ],
    "scenes": [
      "product",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3でCMも作った\n前に作った時よりもAIの動画生成レベルが上がっててビビる\nはよオランジーナ復活しろ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 44,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110021253868503394",
    "title": "周瑜与诸葛亮的古风说唱对决",
    "titleEn": "Zhou Yu and Zhuge Liang in a historical rap duel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "古装人物站在灯光聚焦的舞台上，手持麦克风交替说唱，字幕点出周瑜与诸葛亮的对决。 原生视频规格为 102 秒 · 3840×2194。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Historically costumed characters alternate rapping into microphones under stage lights, with captions identifying a fictional duel between Zhou Yu and Zhuge Liang. Native-video output: 102s · 3840×2194. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/DDJCXX/status/2110021253868503394",
    "sourceLabel": "X 原帖 · @DDJCXX",
    "author": "硅基废话",
    "publishedAt": "2026-10-13T14:54:03.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110021253868503394.mp4",
    "posterUrl": "/posters/x/2110021253868503394.jpg",
    "duration": 102,
    "aspectRatio": "landscape",
    "resolution": "3840×2194",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "上条说的古人 Rap Battle，我复刻了一条：周瑜 vs 诸葛亮。\n\n全程用 Claude Code 当导演：拆原片镜头、写词、写分镜、批量生成、自动检查口型。\n\n歌词原创，Suno 做歌，MiniMax H3 出画面，花了约 16 美金。\n\n踩坑最多的是对口型——AI 会偷偷把你的歌重唱一遍，只能一句句对。\n\n这整套流程我打包成了一个 Skill，马上公布：完全免费，开源。",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 7,
      "views": 1239,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110014044577096033",
    "title": "魔法学院礼堂里的自拍与亲吻",
    "titleEn": "A selfie and a kiss in a magical school hall",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "两名穿学院制服的动漫角色在烛光礼堂里自拍，其中一人侧身亲吻对方的脸颊，随后回到合照姿势。 原生视频规格为 5 秒 · 960×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two animated characters in school uniforms pose for a selfie in a candlelit hall; one kisses the other on the cheek before they return to posing. Native-video output: 5s · 960×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/proto_jp/status/2110014044577096033",
    "sourceLabel": "X 原帖 · @proto_jp",
    "author": "PROTO@AiArt",
    "publishedAt": "2026-10-13T14:25:25.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110014044577096033.mp4",
    "posterUrl": "/posters/x/2110014044577096033.jpg",
    "duration": 5,
    "aspectRatio": "square",
    "resolution": "960×960",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ハリー・ポッター👩‍❤️‍💋‍👨マルフォイ♀ #Malfoid MiniMax H3 Local",
    "engagement": {
      "replies": 0,
      "reposts": 1,
      "likes": 27,
      "views": 2473,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110006487045242988",
    "title": "粉白舞台上的双马尾少女",
    "titleEn": "A twin-tailed girl on a pink-and-white stage",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "浅色双马尾少女在心形装饰舞台上跳舞，镜头在全身舞步与近景之间切换，原帖下方附有角色参考图。 原生视频规格为 14 秒 · 1080×1206。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A light-haired twin-tailed girl dances on a heart-decorated stage as the camera alternates full-body movement and closer views; a character reference appears beneath the source clip. Native-video output: 14s · 1080×1206. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Mayz1169/status/2110006487045242988",
    "sourceLabel": "X 原帖 · @Mayz1169",
    "author": "Kiki",
    "publishedAt": "2026-10-13T13:55:23.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110006487045242988.mp4",
    "posterUrl": "/posters/x/2110006487045242988.jpg",
    "duration": 14,
    "aspectRatio": "portrait",
    "resolution": "1080×1206",
    "tags": [
      "MiniMax H3",
      "舞蹈",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dance",
    "styles": [],
    "scenes": [
      "dance",
      "performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "another dance test 🩵  \n\nused MiniMax H3 for the animation this time and I’m impressed with the result  \n\nPrompt below ⤵️",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 33,
      "views": 1163,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110006416719421481",
    "title": "餐桌旁的虚构家庭争执",
    "titleEn": "A fictional family argument beside a dining table",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "虚构剧情中，成年父女在家中发生争执，父亲做出打人动作，女儿捂脸，随后双方继续对话。 原生视频规格为 13 秒 · 1280×720。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "In a fictional domestic scene, an adult father and daughter argue; he makes a striking gesture, she holds her face and their conversation continues. Native-video output: 13s · 1280×720. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "2026年，中国上海，室内，普通装修，夜晚，一位50岁的父亲与一位20岁的女儿的对话。\n\n[short 1]父亲直接一巴掌狠狠打在女儿脸上，同时说：“我平时都是怎么教你的？”\n\n[short 2]女儿捂着脸，委屈的说：“爸，你为啥打我啊？”\n\n父亲说：”你整天抱怨你婆婆这不好，那不好“\n\n父亲接着说：“可孩子你婆婆帮你带了。饭是她做的，衣服是她洗的。还出钱给你买了房子，车子。还天天受些你的坏脾气。你哪来的脸？吃完饭，滚回家去。”\n\n注意，不要加字幕。",
    "promptSourceUrl": "https://x.com/tangpanqing/status/2110006416719421481",
    "sourceUrl": "https://x.com/tangpanqing/status/2110006416719421481",
    "sourceLabel": "X 原帖 · @tangpanqing",
    "author": "汤光头",
    "publishedAt": "2026-10-13T13:55:06.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110006416719421481.mp4",
    "posterUrl": "/posters/x/2110006416719421481.jpg",
    "duration": 13,
    "aspectRatio": "landscape",
    "resolution": "1280×720",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "牛逼，3分钟创作一个爆火的家庭生活短视频\n\n模型MiniMax H3  工具http://PixStag.com\n\n提示语：\n\n2026年，中国上海，室内，普通装修，夜晚，一位50岁的父亲与一位20岁的女儿的对话。\n\n[short 1]父亲直接一巴掌狠狠打在女儿脸上，同时说：“我平时都是怎么教你的？”\n\n[short 2]女儿捂着脸，委屈的说：“爸，你为啥打我啊？”\n\n父亲说：”你整天抱怨你婆婆这不好，那不好“\n\n父亲接着说：“可孩子你婆婆帮你带了。饭是她做的，衣服是她洗的。还出钱给你买了房子，车子。还天天受些你的坏脾气。你哪来的脸？吃完饭，滚回家去。”\n\n注意，不要加字幕。",
    "engagement": {
      "replies": 0,
      "reposts": 4,
      "likes": 9,
      "views": 2762,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110004054323798228",
    "title": "玻璃碎裂时后退的女子",
    "titleEn": "A woman steps back as glass shatters",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "俯视镜头中的女子站在弧形楼梯旁，玻璃碎片散落在周围；她后退避开后，转身沿另一方向离开。 原生视频规格为 15 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Seen from above beside a curved staircase, a woman steps back from breaking glass and then turns to leave in another direction. Native-video output: 15s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ou_zhen599/status/2110004054323798228",
    "sourceLabel": "X 原帖 · @ou_zhen599",
    "author": "Loriel.AI",
    "publishedAt": "2026-10-13T13:45:43.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110004054323798228.mp4",
    "posterUrl": "/posters/x/2110004054323798228.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with MiniMax-H3 ComfyUI                                                                                                          \n\n🎬Image-to-Video         \n📷“Half a Step”\nWhat saved her was stepping back half a step; once the shards had settled, she had to turn away and give up the path leading upward before her.\n\nFull prompt in the reply.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 3,
      "views": 274,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2110003630254510111",
    "title": "球鞋近景与运动装模特",
    "titleEn": "Sneaker close-ups and a sportswear model",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "运动装女子手持球鞋出镜，下方画面切换鞋身细节和品牌式字幕，原帖并列展示参考图与动态结果。 原生视频规格为 15 秒 · 1920×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in sportswear presents a sneaker while the lower view cuts to shoe details and brand-style typography, with a reference image shown alongside the moving result. Native-video output: 15s · 1920×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/renmu2017/status/2110003630254510111",
    "sourceLabel": "X 原帖 · @renmu2017",
    "author": "AI 观察家 Andy",
    "publishedAt": "2026-10-13T13:44:02.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2110003630254510111.mp4",
    "posterUrl": "/posters/x/2110003630254510111.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1920×2160",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [],
    "scenes": [
      "product"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "一张参考图像 + Minimax H3。👀",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 2,
      "views": 324,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109991670486024340",
    "title": "老街轨道上的红色电车",
    "titleEn": "A red tram on tracks through an old street",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红色电车沿两侧布满老建筑的轨道行驶，镜头保持在车头前方，街景随着车辆移动而变化。 原生视频规格为 10 秒 · 1376×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red tram travels between old buildings as a camera facing its front follows the movement of the vehicle and surrounding street. Native-video output: 10s · 1376×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/itotaiga_/status/2109991670486024340",
    "sourceLabel": "X 原帖 · @itotaiga_",
    "author": "伊藤大河",
    "publishedAt": "2026-10-13T12:56:30.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109991670486024340.mp4",
    "posterUrl": "/posters/x/2109991670486024340.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1376×768",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ローカルで動画生成したよ！＼(^o^)／\n\n6秒台で31分→10秒超で約10分！🚋\n\nRTX 3060 Ti 8GB／RAM 16GBでEndless MiniMax H3＋Vedaを検証。5秒ずつ生成・連結し、音声付き動画が完成！\n\n※前回と生成条件は異なります\n#動画生成AI #ComfyUI #MiniMaxH3",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 161,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109990511117484103",
    "title": "并列画面中的黑发少女对白",
    "titleEn": "A dark-haired girl speaking in paired views",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑发少女坐在窗边的座位上说话，两侧画面以修复前后标签并列呈现，镜头逐渐靠近面部。 原生视频规格为 10 秒 · 1280×1200。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dark-haired girl speaks from a seat beside a window in side-by-side views labeled before and after an audio fix, with the framing moving closer to her face. Native-video output: 10s · 1280×1200. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2109990511117484103",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-10-13T12:51:54.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109990511117484103.mp4",
    "posterUrl": "/posters/x/2109990511117484103.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "1280×1200",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3：以前作った5秒動画を1分で作る高速化ワークフローで出る「金属管の中で喋るような声」を抑える改良ワークフローを作成！\n\n仕組み：2パス構成\n32×32で音声だけ8step→その声を固定して映像生成\n\nRTX4070・5秒動画 60.5秒→71.6秒（+11.1秒）\n長尺動画にしても+10秒前後の増加のみ。",
    "engagement": {
      "replies": 2,
      "reposts": 1,
      "likes": 23,
      "views": 1822,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109988616089010428",
    "title": "雨夜霓虹巷中的短发女子",
    "titleEn": "A short-haired woman in a rainy neon alley",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短发女子穿着深色外套站在霓虹灯照亮的雨夜巷道里，镜头逐渐靠近，她直视镜头并改变神情。 原生视频规格为 5 秒 · 864×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired woman in a dark coat stands in a rain-soaked neon alley, looking into the camera and changing expression as the framing moves closer. Native-video output: 5s · 864×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/RellenKray/status/2109988616089010428",
    "sourceLabel": "X 原帖 · @RellenKray",
    "author": "Rellen Kray",
    "publishedAt": "2026-10-13T12:44:22.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109988616089010428.mp4",
    "posterUrl": "/posters/x/2109988616089010428.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue",
      "city"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made this demo tonight on a mini-PC in the spare room. No cloud, no subscription, no API - MiniMax H3, fully local, lip sync & all.\n\n  The future doesn't wait for permission.",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 3,
      "views": 451,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109985729371943248",
    "title": "咖啡与电脑旁的晨间独白",
    "titleEn": "A morning monologue beside coffee and a laptop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "男子坐在窗边桌前，咖啡杯和笔记本电脑摆在面前，他朝镜头说话并做出轻微手势。 原生视频规格为 15 秒 · 832×480。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A man sits at a table by a window with coffee and a laptop, speaking toward the camera and making small hand gestures. Native-video output: 15s · 832×480. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/woleswoosh/status/2109985729371943248",
    "sourceLabel": "X 原帖 · @woleswoosh",
    "author": "FEY8ER",
    "publishedAt": "2026-10-13T12:32:54.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109985729371943248.mp4",
    "posterUrl": "/posters/x/2109985729371943248.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "832×480",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "The coffee is ready. I’m not  Plot twist: this entire video is AI generated. Would you have noticed?\n\nmade it with Minimax H3 at @PixVerse\n@PixVerse\n#PixVerseCPP #UGC",
    "engagement": {
      "replies": 4,
      "reposts": 0,
      "likes": 6,
      "views": 924,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109981890468016412",
    "title": "雨夜便利店外的外卖员",
    "titleEn": "A delivery rider outside a convenience store at night",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黄衣外卖员在雨夜便利店外停留，随后坐在店门附近查看手机并拿起热饮；原帖明确第一段由 MiniMax H3 生成。 原生视频规格为 15 秒 · 768×1344。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A yellow-jacketed delivery rider waits outside a convenience store in the rain, then checks his phone and holds a hot drink; the source explicitly identifies the first clip as MiniMax H3. Native-video output: 15s · 768×1344. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LBan2049/status/2109981890468016412",
    "sourceLabel": "X 原帖 · @LBan2049",
    "author": "LBan",
    "publishedAt": "2026-10-13T12:17:38.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109981890468016412.mp4",
    "posterUrl": "/posters/x/2109981890468016412.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "768×1344",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "视频1 是使用 MiniMax H3 Max 文生视频生成。\n视频2 使用 Grok 加参考图生成。\n参考图的重要性还是很高的，可以很好提升画面质量。",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 73,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109981831034720739",
    "title": "荒地上奔跑的少女与火球",
    "titleEn": "A girl runs across a wasteland beside a fireball",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发动漫少女在开裂的荒地上朝镜头奔跑，身旁的火球追随移动，背景是岩石与远处山地。 原生视频规格为 5 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde animated girl runs toward the camera across cracked ground while a fireball moves beside her, with rocks and distant hills behind. Native-video output: 5s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/HitotsumeKuro/status/2109981831034720739",
    "sourceLabel": "X 原帖 · @HitotsumeKuro",
    "author": "1つ目クロスケ",
    "publishedAt": "2026-10-13T12:17:24.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109981831034720739.mp4",
    "posterUrl": "/posters/x/2109981831034720739.jpg",
    "duration": 5,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "自由な動きができる！私の環境だと中々難しかったイメージ通りな動きができるようになりました。プリビズ使用でMiniMax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 61,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109977456577061001",
    "title": "红灯前等待的魔王军",
    "titleEn": "A dark army waiting at a red light",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "披甲怪物和魔王军角色出现在东京街头，霓虹城市与街道片段交错，结尾切到地铁车厢。 原生视频规格为 17 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Armored monsters and dark-army figures appear on Tokyo streets amid neon city scenes before the sequence ends inside a train carriage. Native-video output: 17s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/maru_999_/status/2109977456577061001",
    "sourceLabel": "X 原帖 · @maru_999_",
    "author": "Maru999",
    "publishedAt": "2026-10-13T12:00:01.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109977456577061001.mp4",
    "posterUrl": "/posters/x/2109977456577061001.jpg",
    "duration": 17,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "city",
      "creatures"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "東京では、魔王軍も赤信号で止まる。\nIn Tokyo, even the Dark Army waits for the green light.\n\n生成: Kairos（MiniMax H3）／曲はSunoで作りました🎵",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 65,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109976697156378914",
    "title": "退潮海床上出现的沉默人群",
    "titleEn": "Silent figures appear on a drained seabed",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黄外套女子独自站在灰暗海边，海水退去后大量沉默人影显露出来，最后人群排在她面前。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in a yellow coat stands alone at a grey shoreline; as the water drains away, numerous silent figures are revealed and end up facing her. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/molyin_com/status/2109976697156378914",
    "sourceLabel": "X 原帖 · @molyin_com",
    "author": "Molyin",
    "publishedAt": "2026-10-13T11:57:00.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109976697156378914.mp4",
    "posterUrl": "/posters/x/2109976697156378914.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [
      "retro"
    ],
    "scenes": [
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Sharing MiniMax H3 Prompts on Molyin\n\nA cinematic horror sequence generated on Molyin with MiniMax H3 — 35mm film look, one woman, a draining sea, a seabed full of silent figures, and a synchronized final reveal. No crew, no VFX, just a prompt that directs the whole scare.\n\nTry it here:\nhttps://molyin.com/\n\nFull prompt in replies 👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 1,
      "views": 54,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109975403448406247",
    "title": "女子穿越城市与弯曲隧道",
    "titleEn": "A woman runs through city streets and a curving tunnel",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑衣女子在城市户外向镜头挥手，片段随后转入弯曲通道中的奔跑，最后回到海边街道上的近景。 原生视频规格为 65 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in dark clothing gestures toward the camera outdoors, followed by a run through a curving passage and a final close-up on a seaside street. Native-video output: 65s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/chujoakira/status/2109975403448406247",
    "sourceLabel": "X 原帖 · @chujoakira",
    "author": "あぼん",
    "publishedAt": "2026-10-13T11:51:52.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109975403448406247.mp4",
    "posterUrl": "/posters/x/2109975403448406247.jpg",
    "duration": 65,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax H3でお姉さんを走らせてみた\nやっぱり最初と最後で微妙に顔が変わってしまう\n\nあと音楽の力は偉大すぎる\n自認ソニックネキ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 46,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109971022086644128",
    "title": "满月下召唤渡鸦的德鲁伊",
    "titleEn": "A druid summons ravens beneath a full moon",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "黑羽装束的女子站在满月与枯树前，紫色光芒在手边闪动，一只渡鸦停在她身旁。 原生视频规格为 15 秒 · 640×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A woman in black feathered clothing stands before a full moon and bare trees as violet light glows by her hand and a raven perches nearby. Native-video output: 15s · 640×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/wraithworksprod/status/2109971022086644128",
    "sourceLabel": "X 原帖 · @wraithworksprod",
    "author": "Wraith",
    "publishedAt": "2026-10-13T11:34:27.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109971022086644128.mp4",
    "posterUrl": "/posters/x/2109971022086644128.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "640×1152",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Last of the druids for a bit. Had fun making them.\n\nUnder a full moon, the Raven Druid awakens her familiar with violet magic—calling an entire murder of ravens into the storm-lit sky. 🐦‍⬛🌕⚡\nA dark-fantasy character moment brought to life with MiniMax H3 in ComfyUI.\n#MiniMaxH3 #ComfyUI #AIVideo #DarkFantasy #RavenDruid",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 75,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109962665737281682",
    "title": "裁缝店里拿起量尺的女子",
    "titleEn": "A woman lifts a measuring tape in a tailor shop",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "马尾女子站在摆着工具与线轴的裁缝店里，拿起黄色量尺比量衣服，最后伸手靠近镜头。 原生视频规格为 15 秒 · 720×1280。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A ponytailed woman in a tailor shop surrounded by tools and spools lifts a yellow tape measure to her clothing and finally reaches toward the camera. Native-video output: 15s · 720×1280. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/reiQjo/status/2109962665737281682",
    "sourceLabel": "X 原帖 · @reiQjo",
    "author": "九条 玲",
    "publishedAt": "2026-10-13T11:01:15.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109962665737281682.mp4",
    "posterUrl": "/posters/x/2109962665737281682.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "720×1280",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ぴったり / あきいろ\n\n※AI生成動画（MiniMax H3）",
    "engagement": {
      "replies": 3,
      "reposts": 4,
      "likes": 27,
      "views": 588,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109962363130777914",
    "title": "雪人在旋涡中变成白衣少女",
    "titleEn": "A snowman transforms into a girl through a swirling vortex",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "上下两行对照中，同一雪人逐渐变成围红围巾的白衣少女，下行出现明显的旋涡式变形。 原生视频规格为 5 秒 · 1024×1152。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Two stacked views show a snowman becoming a white-clad girl with a red scarf, with a pronounced swirling transformation in the lower view. Native-video output: 5s · 1024×1152. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/nookwright/status/2109962363130777914",
    "sourceLabel": "X 原帖 · @nookwright",
    "author": "NookWright｜AIイラストと動画",
    "publishedAt": "2026-10-13T11:00:03.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109962363130777914.mp4",
    "posterUrl": "/posters/x/2109962363130777914.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "1024×1152",
    "tags": [
      "MiniMax H3",
      "动作特效",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "action-vfx",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3のTransformation LoRA、試してみました\n雪だるまが渦を巻いて女の子に変わります\n変化の仕方がおもしろいですね！\n上がLoRAなし、下がLoRAあり",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 5,
      "views": 92,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109961527675834614",
    "title": "列车窗边的红发学生",
    "titleEn": "A red-haired student beside a train window",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红发学生坐在移动的列车窗边，随后切换到金发学生与同伴交谈的近景，呈现魔法校园故事的氛围。 原生视频规格为 19 秒 · 960×544。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A red-haired student sits by the window of a moving train before the scene cuts to a blonde student talking with a companion in a magical-school setting. Native-video output: 19s · 960×544. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/toyxyz3/status/2109961527675834614",
    "sourceLabel": "X 原帖 · @toyxyz3",
    "author": "toyxyz",
    "publishedAt": "2026-10-13T10:56:44.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109961527675834614.mp4",
    "posterUrl": "/posters/x/2109961527675834614.jpg",
    "duration": 19,
    "aspectRatio": "landscape",
    "resolution": "960×544",
    "tags": [
      "MiniMax H3",
      "角色对白",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "dialogue",
    "styles": [],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Minimax-H3 Ron Weasley as Malfoy’s New Rival test #malfoy #malfoid #comfyui",
    "engagement": {
      "replies": 54,
      "reposts": 160,
      "likes": 1891,
      "views": 69242,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109957004470943953",
    "title": "星空浮岛之间的魔法少女交锋",
    "titleEn": "Magical girls clash among floating islands under the stars",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "粉发与紫发魔法少女在星空浮岛间举起法杖，紫色能量球与光线在双方之间交错，结尾被烟雾覆盖。 原生视频规格为 16 秒 · 3840×2160。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Pink-haired and purple-haired magical girls raise their staffs among floating islands beneath a starry sky, exchanging violet energy before smoke fills the view. Native-video output: 16s · 3840×2160. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sieru_copan/status/2109957004470943953",
    "sourceLabel": "X 原帖 · @sieru_copan",
    "author": "sieru",
    "publishedAt": "2026-10-13T10:38:45.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109957004470943953.mp4",
    "posterUrl": "/posters/x/2109957004470943953.jpg",
    "duration": 16,
    "aspectRatio": "landscape",
    "resolution": "3840×2160",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "anime"
    ],
    "scenes": [
      "combat",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3で魔法少女アニメの戦闘シーンを作ってみました。動画、セリフ、BGMはすべてローカルで生成してます。動画制作は今回が初挑戦。途中諦めそうになったけど、構図とか、カット割りとかいろいろ試行錯誤して、なんと完成までたどり着けて良かった。\nxは猛者ばかりで見劣りするかもだけど、自分の中ではよくやったと思う。自分を褒めたい。頑張った、えらい！\n \nラフ画像生成：Qwen-Image 2.1 Turbo\n本番画像生成：OpenAI ImageGen\n動画生成：MiniMax H3 FL2VA＋Turbo 8-step LoRA\nセリフ生成：Irodori-TTS VoiceDesign\nBGM生成：ACE-Step 1.5 XL-SFT\n4Kアップスケール：Real-ESRGAN x2plus＋Lanczos\n効果音：H3生成音＋Pythonで合成・調整\n動画編集・整音：Python／FFmpeg\n制作支援・コード作成：Codex GTP-6 Astra\n \n#AIアニメ #MiniMaxH3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 6,
      "views": 263,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109956761322938557",
    "title": "未来球场上腾空跃起的运动员",
    "titleEn": "An athlete leaps above a futuristic court",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "篮球运动员在高楼环绕的未来球场上跃起，镜头跟随他落地并转向球场另一侧。 原生视频规格为 18 秒 · 1912×1084。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A basketball athlete leaps on a futuristic court surrounded by towers, with the camera following his landing and turn toward the other side. Native-video output: 18s · 1912×1084. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/carlosbrlhnt/status/2109956761322938557",
    "sourceLabel": "X 原帖 · @carlosbrlhnt",
    "author": "Brlhnt",
    "publishedAt": "2026-10-13T10:37:47.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109956761322938557.mp4",
    "posterUrl": "/posters/x/2109956761322938557.jpg",
    "duration": 18,
    "aspectRatio": "landscape",
    "resolution": "1912×1084",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "sci-fi"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Made with Cinema Studio and upscaled using the Ultimate Video Upscaler from @wavespeed_ai @WaveSpeedCPP \n\nVideo model: Minimax H3",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 75,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109947696467575292",
    "title": "摩天楼屏幕上的墨镜时装秀",
    "titleEn": "A sunglasses fashion display across skyscraper screens",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿黑色礼服的模特出现在摩天楼巨屏上，画面转入墨镜近景；上下对照分别标注 MiniMax H3 与 PixVerse。 原生视频规格为 17 秒 · 1080×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A model in a black gown appears across giant skyscraper screens before a sunglasses close-up; the stacked comparison labels MiniMax H3 above PixVerse. Native-video output: 17s · 1080×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ModellixAI/status/2109947696467575292",
    "sourceLabel": "X 原帖 · @ModellixAI",
    "author": "Modellix",
    "publishedAt": "2026-10-13T10:01:46.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109947696467575292.mp4",
    "posterUrl": "/posters/x/2109947696467575292.jpg",
    "duration": 17,
    "aspectRatio": "square",
    "resolution": "1080×1080",
    "tags": [
      "MiniMax H3",
      "模型对比",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "comparison",
    "styles": [],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "This sleek tower fashion clip hits different.\nWatch her flick sunglasses into slow-mo.🕶️\n\nModellix comparison test\nMiniMax H3 vs PixVerse V6\n- Keep character & outfit consistent\n- Natural hair and fabric movement\n\nWhich version is your pick?\nCreate on Modellix.",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 123,
      "snapshotAt": "2026-10-14T00:42:21.947Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109941842712080406",
    "title": "咬着发带整理头发的女子",
    "titleEn": "A woman tidies her hair while holding a ribbon in her mouth",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "长发女子在卧室里轻咬蓝色发带，用手整理刘海与侧边头发，神情从专注转为微笑。 原生视频规格为 10 秒 · 864×480。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "A long-haired woman in a bedroom holds a blue ribbon gently in her mouth while adjusting her bangs and side hair, shifting from concentration to a smile. Native-video output: 10s · 864×480. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "A 5‑second looping shot of the girl from reference image 1, gently biting a hair ribbon, continuously arranging and tidying her hair with one hand, fingers running through and adjusting her bangs and side hair, soft natural lighting, subtle, smooth motion, shy but playful expression, clean simple background, high‑resolution, detailed hair and face.",
    "promptSourceUrl": "https://x.com/0xEverybody/status/2109941842712080406",
    "sourceUrl": "https://x.com/0xEverybody/status/2109941842712080406",
    "sourceLabel": "X 原帖 · @0xEverybody",
    "author": "everybody",
    "publishedAt": "2026-10-13T09:38:30.000Z",
    "addedAt": "2026-10-14T00:42:21.947Z",
    "mediaUrl": "/media/x-2109941842712080406.mp4",
    "posterUrl": "/posters/x/2109941842712080406.jpg",
    "duration": 10,
    "aspectRatio": "landscape",
    "resolution": "864×480",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "#MiniMax H3 r2p many attempts, which is your favourite?\n\nPrompt starts from 👇 \n\nA 5‑second looping shot of the girl from reference image 1, gently biting a hair ribbon, continuously arranging and tidying her hair with one hand, fingers running through and adjusting her bangs and side hair, soft natural lighting, subtle, smooth motion, shy but playful expression, clean simple background, high‑resolution, detailed hair and face.\n\nmore examples in the thread 🧵",
    "engagement": {
      "replies": 2,
      "reposts": 0,
      "likes": 0,
      "views": 185,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.947Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109941271855051264",
    "title": "白色背景前说话的紫发角色",
    "titleEn": "A purple-haired character speaks against a white background",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "穿深色外套与紫色服装的动漫角色站在白色背景前，嘴部和面部做出细微变化，维持稳定的正面构图。 原生视频规格为 5 秒 · 544×960。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated character with purple hair, a dark coat and purple clothing makes subtle mouth and facial movements against a white background in a stable frontal composition. Native-video output: 5s · 544×960. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/shohei_botter/status/2109941271855051264",
    "sourceLabel": "X 原帖 · @shohei_botter",
    "author": "Shohei",
    "publishedAt": "2026-10-13T09:36:14.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109941271855051264.mp4",
    "posterUrl": "/posters/x/2109941271855051264.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "544×960",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "dialogue"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 のワークフローを試しがてら、適当に生成した立ち絵を喋らせてみる\n\n結月ゆかりの音声データを学習しているかなと思ったものの、完全に別物の声が生成された\n\nWan 2.2よりも、はるかに高品質に動画生成できるんだな\n\n5秒程度の動画なら、数分で生成が終わるから手軽だ",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 2,
      "views": 129,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109938041901207803",
    "title": "同一角色跨场景与同伴互动",
    "titleEn": "A recurring character interacts with a companion across scenes",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "人物参考演示从女子近景切到夕阳下的双人互动，最后展示节点工作流界面，原帖说明用于角色一致性验证。 原生视频规格为 108 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A character-reference demonstration moves from a woman in close-up to two people interacting at sunset, then ends on a node workflow interface; the source describes character-consistency checks. Native-video output: 108s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sunbaolong_2001/status/2109938041901207803",
    "sourceLabel": "X 原帖 · @sunbaolong_2001",
    "author": "有趣的80后程序员",
    "publishedAt": "2026-10-13T09:23:24.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109938041901207803.mp4",
    "posterUrl": "/posters/x/2109938041901207803.jpg",
    "duration": 108,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Lock character faces across ANY AI video—with ZERO LoRA training! 🎬👤\nMiniMax H3 RefMod turns 8 photos into a reusable digital actor asset that functions like a LoRA without burning hours of training time!\nI stress-tested it across 10 ComfyUI scenarios:\n✅ Single Actor: Scene jumps, huge zoom-outs, & 180° back-to-front turnarounds\n✅ Dual Actors: Hand-holding, umbrella-sharing, & station farewells (No face/outfit bleeding!)\n✅ Fast 8-step sampling using Singularity + SageAttention\nFull guide & RunningHub workflows below: 👇\n🔗https://youtu.be/kGUOWSX29Mo\n#MiniMaxH3 #RefMod #ComfyUI #AIVideo #CharacterConsistency #GenerativeAI#MiniMaxH3 #RefMod #ComfyUI #AIVideo #CharacterConsistency #GenerativeAI",
    "engagement": {
      "replies": 1,
      "reposts": 25,
      "likes": 192,
      "views": 12751,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109935395547001099",
    "title": "吉他少女转变为黑裙人物",
    "titleEn": "A guitar-playing girl transforms into a figure in black",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "动漫少女在室内弹奏吉他，身体周围的形态逐渐变化，结尾成为站在书架前的黑裙女子。 原生视频规格为 5 秒 · 832×1216。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An animated girl plays a guitar indoors as shapes shift around her, ending as a woman in a black dress standing before bookshelves. Native-video output: 5s · 832×1216. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/sep_is_heim/status/2109935395547001099",
    "sourceLabel": "X 原帖 · @sep_is_heim",
    "author": "Kamimoto(かみもと)",
    "publishedAt": "2026-10-13T09:12:53.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109935395547001099.mp4",
    "posterUrl": "/posters/x/2109935395547001099.jpg",
    "duration": 5,
    "aspectRatio": "portrait",
    "resolution": "832×1216",
    "tags": [
      "MiniMax H3",
      "动画",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "animation",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 LoRAマップにあるTransformation LoRAをトライ！演出が面白いですねー。開始と終了画像を指定して、LoRAを入れて、プロンプト入れるだけ！",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 13,
      "views": 1172,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109933865217483124",
    "title": "八位冒险者在草原上合唱",
    "titleEn": "Eight adventurers sing together on a grassy plain",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "身穿奇幻服装的冒险者聚在明亮草原上合唱，镜头在全体与角色近景之间切换，并叠加歌词。 原生视频规格为 41 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "Adventurers in fantasy costumes sing on a bright grassy plain as the sequence alternates group views and character close-ups with lyric overlays. Native-video output: 41s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/maru_999_/status/2109933865217483124",
    "sourceLabel": "X 原帖 · @maru_999_",
    "author": "Maru999",
    "publishedAt": "2026-10-13T09:06:48.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109933865217483124.mp4",
    "posterUrl": "/posters/x/2109933865217483124.jpg",
    "duration": 41,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance",
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "この道わが旅 #6 八人そろって。ギアガの大穴の先で、8人がそろって歌う。\n生成: Kairos（MiniMax H3）\n#この道わが旅 #AIアニメ",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 2,
      "views": 61,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109933291503767733",
    "title": "舞者姿态驱动的角色替换对照",
    "titleEn": "A pose-driven character replacement comparison",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "原始舞者、绿色姿态参考和替换后人物被并列展示，人物保持相近舞步，服装与外貌随片段变化。 原生视频规格为 21 秒 · 1920×1080。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "An original dancer, a green pose reference and a replaced character appear side by side, keeping similar movements while appearance and clothing change across segments. Native-video output: 21s · 1920×1080. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/ai_hakase_/status/2109933291503767733",
    "sourceLabel": "X 原帖 · @ai_hakase_",
    "author": "ハカセ アイ(Ai-Hakase)🐾生成AI・ComfyUI",
    "publishedAt": "2026-10-13T09:04:32.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109933291503767733.mp4",
    "posterUrl": "/posters/x/2109933291503767733.jpg",
    "duration": 21,
    "aspectRatio": "landscape",
    "resolution": "1920×1080",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "dance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "【動画生成に大革命🚀】\n動画の人物差し替えが「3倍爆速」になる神ワザ、知ってましたか⋯！？✨\n\nComfyUI×最新技術「MiniMax H3」と「Veda Sparse Attention」で、激重だった動画インペイントが劇的に進化しました！\n\n👇️ このスレッドで解説していきます！\n#ComfyUI #AI動画",
    "engagement": {
      "replies": 6,
      "reposts": 46,
      "likes": 500,
      "views": 24951,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109930801295212544",
    "title": "瀑布悬崖城市间的低空穿行",
    "titleEn": "A low flight through a waterfall city on cliffs",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "镜头从布满瀑布的悬崖城市俯瞰切入街巷穿行，快速掠过建筑后再次展开远处的山谷与水流。 原生视频规格为 15 秒 · 1344×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "The view shifts from an aerial waterfall city on cliffs into a fast flight through narrow streets, then opens back onto valleys and flowing water. Native-video output: 15s · 1344×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/LBan2049/status/2109930801295212544",
    "sourceLabel": "X 原帖 · @LBan2049",
    "author": "LBan",
    "publishedAt": "2026-10-13T08:54:38.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109930801295212544.mp4",
    "posterUrl": "/posters/x/2109930801295212544.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1344×768",
    "tags": [
      "MiniMax H3",
      "电影叙事",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "cinematic",
    "styles": [],
    "scenes": [
      "city",
      "nature"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "MiniMax H3 Max 文本生成视频的版本，在 https://Fal.AI 中生成",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 59,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109913024215859534",
    "title": "室内座位上的金发人物动作",
    "titleEn": "A blonde character gestures from an indoor seat",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "金发女子坐在室内，穿白色衬衫和深色裙子，抬手碰触头发后朝镜头微笑。 原生视频规格为 15 秒 · 1920×3456。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A blonde woman in a white blouse and dark skirt sits indoors, lifts a hand toward her hair and smiles at the camera. Native-video output: 15s · 1920×3456. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/bangel_19/status/2109913024215859534",
    "sourceLabel": "X 原帖 · @bangel_19",
    "author": "bangel19🇯🇵",
    "publishedAt": "2026-10-13T07:43:59.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109913024215859534.mp4",
    "posterUrl": "/posters/x/2109913024215859534.jpg",
    "duration": 15,
    "aspectRatio": "portrait",
    "resolution": "1920×3456",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ワークフローにオーディオと動画の最後の画像を取り出す部分を追加してみた一応機能しているみたい　ただ取り出した画像がもう少し綺麗だと良いのだが基礎知識も何もない自分には以前512→1024にしたら少し綺麗になったので次は2048か　メモリ足るんかぐらいのいい加減🤔\n「生成AI（MiniMax H3）」",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 1,
      "views": 75,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109911859361206461",
    "title": "秋日穿搭与四种表情的插画模特",
    "titleEn": "An illustrated model presents autumn outfits and expressions",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "短发插画人物先拿着苹果出现，随后切换宽松上衣、工装裤与风衣造型，背景配合大字排版变化。 原生视频规格为 15 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A short-haired illustrated character first holds an apple, then appears in a loose T-shirt, cargo pants and a trench coat as large background typography changes. Native-video output: 15s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/glitter_11924/status/2109911859361206461",
    "sourceLabel": "X 原帖 · @glitter_11924",
    "author": "津軽",
    "publishedAt": "2026-10-13T07:39:22.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109911859361206461.mp4",
    "posterUrl": "/posters/x/2109911859361206461.jpg",
    "duration": 15,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [
      "illustration"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "ファッションの秋、おしゃれしよう？🍂\n\nお好きな画像を1枚。\nいつもの姿が、表情も弾む4つのコーデに。\nロゴまで決まる、15秒のファッションショー。\nね、試してみない？✨\n\n🎬 MiniMax H3を使用\n📝 プロンプトはリプ欄に👇",
    "engagement": {
      "replies": 2,
      "reposts": 9,
      "likes": 73,
      "views": 1651,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109911674149089378",
    "title": "家人之间的沉默与告别",
    "titleEn": "Silence and farewells between family members",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "剧情片段交替呈现年长女子的近景、厨房里端着杯子的女子，以及黑衣女子与西装男子并肩站立的画面。 原生视频规格为 73 秒 · 1080×1920。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A dramatic sequence alternates an older woman in close-up, a woman holding a cup in a kitchen, and a woman in black standing beside a suited man. Native-video output: 73s · 1080×1920. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/Wolfdubaii/status/2109911674149089378",
    "sourceLabel": "X 原帖 · @Wolfdubaii",
    "author": "Wolfdubai",
    "publishedAt": "2026-10-13T07:38:38.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109911674149089378.mp4",
    "posterUrl": "/posters/x/2109911674149089378.jpg",
    "duration": 73,
    "aspectRatio": "portrait",
    "resolution": "1080×1920",
    "tags": [
      "MiniMax H3",
      "综合展示",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "showcase",
    "styles": [],
    "scenes": [
      "daily-life"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "EP 1 · PART 4 🎬\nMade this with minimax h3 , full episode 🎥👇",
    "engagement": {
      "replies": 1,
      "reposts": 0,
      "likes": 0,
      "views": 67,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109900705607844311",
    "title": "水晶鲸鱼与白龙环绕梦境之门",
    "titleEn": "A crystal whale and white dragon circle a dream gate",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "水晶鲸鱼、白龙与玻璃般的莲花交织在星海场景中，镜头环绕结构后回到正面，发光纹理与气泡遍布画面。 原生视频规格为 8 秒 · 768×768。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A crystal whale, a white dragon and glass-like lotus forms intertwine in a cosmic underwater scene as the camera circles back to the front amid glowing textures and bubbles. Native-video output: 8s · 768×768. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/eikoung/status/2109900705607844311",
    "sourceLabel": "X 原帖 · @eikoung",
    "author": "紅",
    "publishedAt": "2026-10-13T06:55:02.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109900705607844311.mp4",
    "posterUrl": "/posters/x/2109900705607844311.jpg",
    "duration": 8,
    "aspectRatio": "square",
    "resolution": "768×768",
    "tags": [
      "MiniMax H3",
      "本地生成与工作流",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "workflow",
    "styles": [],
    "scenes": [
      "fantasy"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "《星海萬象・夢境之門》360° 環繞版 ✨\n一張幻想圖，化成 8 秒的宇宙深海之旅。鏡頭繞過水晶鯨魚、白龍與玻璃蓮花，再回到正面。側面與背面由 AI 推測生成。\n\n製作工具：\n原圖：OpenAI 圖片生成工具（透過 Codex）\n動畫：本機 ComfyUI + MiniMax H3（INT8 模型）\n加速：H3 Turbo 8-step LoRA\n硬體：NVIDIA RTX 3090 24GB\n後製：FFmpeg 裁切與編碼\n規格：768 × 768｜24fps｜8 秒｜含音訊\n\n#AIArt #MiniMaxH3 #ComfyUI #幻想藝術",
    "engagement": {
      "replies": 0,
      "reposts": 0,
      "likes": 0,
      "views": 143,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109884479619682650",
    "title": "石墙庭院中的黑色礼服走秀",
    "titleEn": "A black couture gown on a stone-courtyard runway",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "红色绸缎在暗处掠过后，黑色长礼服模特走入高大石墙庭院，在金色侧光中转身摆姿。 原生视频规格为 15 秒 · 1920×1440。原帖完整公开了 Prompt，本页按原文保留。",
    "summaryEn": "Red satin moves through darkness before a model in a long black couture gown enters a tall stone courtyard and turns to pose in golden side light. Native-video output: 15s · 1920×1440. The creator published the complete prompt, preserved here verbatim.",
    "prompt": "Create a 15-second cinematic luxury fashion video in 9:16 vertical format. A tall American female model with long dark hair wears a dramatic black couture gown with silver details and a flowing train.\n\n0–3 sec: Close-up of red satin fabric moving through darkness before revealing the model under dramatic spotlights.\n\n3–6 sec: The model walks confidently toward the camera. A stylish whip-pan captures her turning as the gown flows naturally.\n\n6–10 sec: Seamless transition to a modern architectural courtyard with towering stone walls and golden sunlight. The model walks gracefully as the wind moves her hair and gown.\n\n10–13 sec: The camera circles around her as she strikes a powerful high-fashion pose.\n\n13–15 sec: Symmetrical hero shot with dramatic golden rim lighting, fading into black.\n\nStyle: Ultra-realistic luxury fashion campaign, cinematic lighting, natural fabric physics, elegant camera movements, deep shadows, black and gold color palette, 4K quality.\n\nAudio: Deep cinematic bass, subtle percussion, and soft fabric sounds. No dialogue or subtitles.\n\nText at the end: “BEYOND THE ORDINARY”\n\nMaintain consistent facial features, hairstyle, outfit, and body proportions throughout. No distortion, flickering, watermarks, or random text.",
    "promptSourceUrl": "https://x.com/DaniaSafvi/status/2109884479619682650",
    "sourceUrl": "https://x.com/DaniaSafvi/status/2109884479619682650",
    "sourceLabel": "X 原帖 · @DaniaSafvi",
    "author": "Dania",
    "publishedAt": "2026-10-13T05:50:34.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109884479619682650.mp4",
    "posterUrl": "/posters/x/2109884479619682650.jpg",
    "duration": 15,
    "aspectRatio": "landscape",
    "resolution": "1920×1440",
    "tags": [
      "MiniMax H3",
      "广告与产品",
      "公开 Prompt",
      "X 原帖"
    ],
    "category": "advertising",
    "styles": [
      "commercial",
      "vertical"
    ],
    "scenes": [
      "fashion"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "creator-verbatim",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "Elegance in every movement. Confidence in every detail.\n\nCreated with MiniMax H3.\n\nPrompt:\n\nCreate a 15-second cinematic luxury fashion video in 9:16 vertical format. A tall American female model with long dark hair wears a dramatic black couture gown with silver details and a flowing train.\n\n0–3 sec: Close-up of red satin fabric moving through darkness before revealing the model under dramatic spotlights.\n\n3–6 sec: The model walks confidently toward the camera. A stylish whip-pan captures her turning as the gown flows naturally.\n\n6–10 sec: Seamless transition to a modern architectural courtyard with towering stone walls and golden sunlight. The model walks gracefully as the wind moves her hair and gown.\n\n10–13 sec: The camera circles around her as she strikes a powerful high-fashion pose.\n\n13–15 sec: Symmetrical hero shot with dramatic golden rim lighting, fading into black.\n\nStyle: Ultra-realistic luxury fashion campaign, cinematic lighting, natural fabric physics, elegant camera movements, deep shadows, black and gold color palette, 4K quality.\n\nAudio: Deep cinematic bass, subtle percussion, and soft fabric sounds. No dialogue or subtitles.\n\nText at the end: “BEYOND THE ORDINARY”\n\nMaintain consistent facial features, hairstyle, outfit, and body proportions throughout. No distortion, flickering, watermarks, or random text.",
    "engagement": {
      "replies": 31,
      "reposts": 18,
      "likes": 87,
      "views": 4658,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  },
  {
    "id": "x-2109874257127682287",
    "title": "货运仓库旁的工间节拍",
    "titleEn": "Work-break rhythms beside a freight warehouse",
    "model": "MiniMax H3（创作者标注）",
    "mode": "Unknown",
    "summary": "音乐短片从货运仓库与卡车广角切到穿反光背心的工作人员，她在纸箱和货车旁做手势与动作。 原生视频规格为 362 秒 · 2560×1440。原帖未公开完整 Prompt，本库不反推或补写。",
    "summaryEn": "A musical sequence moves from a freight warehouse and trucks in a wide view to a worker in a reflective vest gesturing and moving beside boxes and trailers. Native-video output: 362s · 2560×1440. The source did not publish a complete prompt, so this library does not infer or complete one.",
    "prompt": null,
    "sourceUrl": "https://x.com/flutterwhat/status/2109874257127682287",
    "sourceLabel": "X 原帖 · @flutterwhat",
    "author": "Flutterwhat",
    "publishedAt": "2026-10-13T05:09:57.000Z",
    "addedAt": "2026-10-14T00:42:21.948Z",
    "mediaUrl": "/media/x-2109874257127682287.mp4",
    "posterUrl": "/posters/x/2109874257127682287.jpg",
    "duration": 362,
    "aspectRatio": "landscape",
    "resolution": "2560×1440",
    "tags": [
      "MiniMax H3",
      "音乐视频",
      "来源未公开 Prompt",
      "X 原帖"
    ],
    "category": "music",
    "styles": [],
    "scenes": [
      "music-performance"
    ],
    "inputTypes": [
      "unknown"
    ],
    "promptProvenance": "not-published",
    "sourceType": "x",
    "verified": false,
    "sourceCaption": "\"Industrial Athlete Blues\" aka \"Reverse Sloptomania\" \n\n| Minimax H3 Via @boneGPT's slop cannon | Logic Pro + Suno V6-WILD |\n\nIndustrial athlete is a real thing  in UPS with real impact at our  gateways and hubs nation wide. We save lives though PCM and education, no news reports it. \n\nA thankless position, at the end of the day. \nBut we all just want to see you take care of yourself. \nUPS is hell on the body and industreal athlete is the answer to that; listen to us!!",
    "engagement": {
      "replies": 2,
      "reposts": 2,
      "likes": 8,
      "views": 444,
      "snapshotAt": "2026-10-14T00:42:21.948Z"
    },
    "approvedAt": "2026-10-14T00:42:21.948Z",
    "editorialBasis": "Original source attribution and playable native video; beginning, middle and ending frames inspected."
  }
]